Population and Development Review, Volume 27, Number 3

POPULATION AND DEVELOPMENT REVIEW

James R. Carey and Debra S.

VOLUME 27 NUMBER 3 Judge Life span extension in humans is S E P T E M B E R 2 0 0 1 self-reinforcing: A general theory of longevity Alexander A. Weinreb First politics, then culture: Accounting for ethnic differences in demographic behavior in Kenya Ulrich Mueller Is there a stabilizing selection around average fertility in modern human populations? Zai Liang The age of migration in China Notes and Commentary G. P. Freeman and R. Birrell on divergent paths of immigration politics in the United States and Australia Data and Perspectives R. Schoen and N. Standish on the retrenchment of marriage in the US Archives William Farr on the economic value of the population Book Reviews Review essays by J. E. Cohen and J. A. Goldstone; reviews by J. Mokyr, D. Hodgson, P. Heuveline, E. D. Carlson, J. T. Boerma, and others Documents The socioeconomic impact of the HIV/AIDS epidemic Population and Development Review seeks to advance knowledge of the interrelationships between population and socioeconomic development and provides a forum for discussion of related issues of public policy.

EDITOR Paul Demeny

MANAGING EDITOR Ethel P. Churchill

EDITORIAL COMMITTEE Paul Demeny, Chair Geoffrey McNicoll Ethel P. Churchill Michael P. Todaro Susan Greenhalgh

EDITORIAL STAFF Robert Heidel, Production Editor Y. Christina Tse, Production/Design Margaret A. Knoll, Circulation Sura Rosenthal / Mike Vosika, Production

ADVISORY BOARD Gustavo Cabrera Milos˘ Macura John C. Caldwell Carmen A. Miró Mercedes B. Concepción Asok Mitra Richard A. Easterlin Samuel H. Preston Akin L. Mabogunje

Signed articles are the responsibility of the authors. Views expressed in the Review do not necessarily reflect the views of the Population Council. Direct manuscripts, comments on articles, and correspondence to: Population and Development Review Population Council One Dag Hammarskjold Plaza New York, New York 10017 USA Subscription information appears on the inside back cover. Support from the United Nations Population Fund is gratefully acknowledged. Volumes are available on microfilm from University Microfilms, Inc., 300 North Zeeb Road, Ann Arbor, Michigan 48106. The website for Population and Development Review is www.popcouncil.org/pdr. Subscribers have access to the electronic edition of the journal, which is published simultaneously with the printed issue. The full contents of Volumes 1–24 (1975–98) are available through participating libraries from JSTOR at www.jstor.org/journals/00987921.html Population and Development Review (ISSN 0098-7921) is published quarterly in March, June, September, and December by the Population Council, One Dag Hammarskjold Plaza, New York, NY 10017 USA. Periodicals postage paid at New York, NY, and additional mailing offices. POSTMASTER: Send address changes to Population and Development Review, One Dag Hammarskjold Plaza, New York, NY 10017. © 2001 by The Population Council, Inc. All rights reserved. Printed in the United States of America. Library of Congress Catalog Card Number: 75-27311 ISSN 0098-7921 Information for Subscribers Population and Development Review is available on a paid subscription basis at the following rates: One year (4 issues) US$40.00 Two years (8 issues) US$64.00 To enter a subscription, send payment by check or money order drawn on a US bank, payable to the Population Council, or by Visa or MasterCard (provide card number and expiration date), together with mailing address to: Population and Development Review Population Council One Dag Hammarskjold Plaza New York, New York 10017 USA Credit card orders may also be placed by phone to (212) 339-0514, by fax to (212) 755-6052, or by e-mail to [email protected]

Automatic renewal For credit card payment only, standing orders are available at US$32.00 per year, with automatic renewal each year until subscription is cancelled by subscriber. Subscriber’s credit card will automatically be charged annually in December for the next year’s volume.

Complimentary subscriptions Complimentary subscriptions are available to qualified applicants residing in developing countries. Individuals and institutions wishing to apply for complimentary subscriptions should send requests on letterhead stationery to the above address stating the nature of professional involvement in development- and population-related issues.

Back issues Casebound (hardcover) volumes are available at US$45.00 each. Single back issues are available at US$10.00 per issue. Please designate copies desired by volume and issue number. Orders may be placed by mail, phone, fax, or e-mail as indicated above. Complimentary copies are available to qualified individuals residing in developing countries. A twenty-year cumulative index to Volumes 1–20 (1975–94) is available at no charge to subscribers to the Review.

Printed on recycled paper in the USA. POPULATION AND DEVELOPMENT REVIEW VOLUME 27 NUMBER 3 S E P T E M B E R 2 0 0 1

ARTICLES

Life Span Extension in Humans Is Self-Reinforcing: A General Theory of Longevity 411

JAMES R. CAREY

DEBRA S. JUDGE

First Politics, Then Culture: Accounting for Ethnic Differences in Demographic Behavior in Kenya 437

ALEXANDER A. WEINREB

Is There a Stabilizing Selection Around Average Fertility in Modern Human Populations? 469

ULRICH MUELLER

The Age of Migration in China 499

ZAI LIANG

NOTES AND COMMENTARY

Divergent Paths of Immigration Politics in the United States and Australia 525

GARY P. FREEMAN

BOB BIRRELL DATA AND PERSPECTIVES

The Retrenchment of Marriage: Results from Marital Status Life Tables for the United States, 1995 553

ROBERT SCHOEN

NICOLA STANDISH

ARCHIVES

William Farr on the Economic Value of the Population 565

BOOK REVIEWS

Linking Human and Natural History: A Review Essay

JOEL E. COHEN 573

Population and Progress in the Middle Ages: A Review Essay

JACK A. GOLDSTONE 585

Julian L. Simon, The Great Breakthrough and Its Cause, edited by Timur Kuran

JOEL MOKYR 596

Angela Hattery, Women, Work, and Family: Balancing and Weaving

DENNIS HODGSON 599

John I. Clarke, The Human Dichotomy: The Changing Numbers of Males and Females

PATRICK HEUVELINE 600

Giovanni Andrea Cornia and Renato Paniccià (eds.), The Mortality Crisis in Transitional Economies

ELWOOD D. CARLSON 602 Caleb E. Finch, James W. Vaupel, and Kevin Kinsella (eds.), Cells and Surveys: Should Biological Measures Be Included in Social Science Research?

J. TIES BOERMA 605

Short Reviews 608

DOCUMENTS

On the Socioeconomic Impact of the HIV/AIDS Epidemic 619

ABSTRACTS 625

AUTHORS FOR THIS ISSUE 632

Life Span Extension in Humans Is Self-Reinforcing: A General Theory of Longevity

JAMES R. CAREY DEBRA S. JUDGE

MORE THAN TWO decades ago gerontologist George Sacher (1978) noted that the approach to one basic question in the biology of aging, “Why do we grow old?,” is guided largely by a research paradigm involving model systems designed to compare biological functions—those of individual organisms—in old and young animals. One of his primary concerns with this aging-oriented approach to understanding longevity was that the far-reaching physiological, cellular, and molecular correspondence between model systems (mice) and humans concealed a paradox: if model species are so similar to humans in molecular makeup that they can serve as models, why do mice age as much in 2 years as humans do in 70 years? This paradox motivated him to ask his second question: “Why do we live as long as we do?” This question cannot be answered within the framework of developmental research on aging, but rather requires the formulation of an evolutionary- comparative paradigm concerned with longevity. We use longevity as a general term for three separate but interrelated measures having to do with length of life: maximum observed life span defined as the highest verified age at death, maximum potential life span defined as the theoretical highest attainable age, and mean life span defined as the expected average number of years of life, or life expectancy.1 Despite the arguments by Sacher and others (e.g., Hayflick 2000) in support of developing a longevity-oriented theory of the finitude of life, no such theory has ever been published. Our objective in this article is to describe a general theory of longevity extension in social species, particularly as it applies to humans. The central concepts of such a theory derive from patterns identified in life span data for vertebrates (Carey and Judge 2000), insects (Carey 2001b), and humans (Kannisto et al. 1994; Tuljapurkar, Li, and Boe 2000; Vaupel et al. 1998; Wilmoth et al. 2000); from environmental factors associated with the evolution of extended longevity in insects

POPULATION AND DEVELOPMENT REVIEW 27(3):411–436 (SEPTEMBER 2001) 411

Click here to print article Click to return to Table of Contents 412 L IFE SPAN EXTENSION IN HUMANS

(Carey 2001a, 2001b); and from the preliminary idea that longevity extension in humans is self-reinforcing2 as first proposed by us (Carey and Judge 2001). Our theory also builds on and extends work by Fogel (1993, 1994, 1997) on what he and his colleagues refer to as “technophysio evolution”—a theory explaining the decline in morbidity and mortality since 1700. A theory of longevity that extends beyond the classical evolutionary theory of aging is important for several reasons. First, whereas senescence is a byproduct of evolution (Medawar 1955), life span is an evolved life-history trait that results from positive natural selection. Second, unlike the evolutionary theory of senescence, which is based solely on individual natural selection (Williams 1957), this theory includes processes of sexual selection and kin selection, bringing life-history theory more fully to bear on questions concerned with the latter portion of the life cycle. Third, longevity-oriented theory allows consideration of behaviors that are characteristic of older individuals, including divisions of labor and intergenerational transfers (Beshers and Fewell 2001). Fourth, it allows for consideration of mortality factors not directly related to aging (acci- dents, acute diseases, socioeconomic circumstances), hence a longevity-oriented theory fosters a greater integration of perspectives from demography (Hauser and Duncan 1959) and gerontology (Finch 1990).

Comparative demography of longevity

Our purpose in this section is to situate human longevity in a broad biological context. We first describe a general classification system for factors that favor the evolution of extended longevity and then provide a brief sketch of longevity extension in social wasps, including a discussion of its relevance to humans.

A general classification scheme

The literature on aging and longevity includes descriptions of a small number of life span correlates, including (a) well-known relationships between longevity and both body mass and relative brain size (Austad 1997; Comfort 1961; Hakeem et al. 1996) and (b) the observation that animals that possess armor (e.g., beetles; turtles) or capability of flight (e.g., birds; bats) are often long- lived (Austad 1997; Kirkwood 1992). The evolutionary theory of senescence suggests that animals better able to escape sources of stochastic mortality such as predators (e.g., via armored defenses or effective escape mechanisms) live longer, and thus the force of natural selection at older ages is increased and the evolution of longer pre-senescent life span is possible. But major inconsisten- cies exist within even this small set of correlates. For example, there are several exceptions regarding the relationship of extended longevity to large body size (e.g., bats are generally small but most species are long-lived), and this positive relationship may be either absent or reversed within taxonomic orders.3 Like-

Click to return to Table of Contents J AMES R. CAREY / DEBRA S. JUDGE 413 wise, the observation that flight ability and extended longevity are correlated does not provide any insight into why within-group differences in life span exist (e.g., among birds), nor does it account for the variation in longevity in insects where adults of the majority of species can fly. An alternative approach for identifying broad correlates of longevity emerged from an examination of several large-scale databases containing the maximum recorded life spans of both vertebrate and invertebrate species (Carey 2001a, 2001b; Carey and Judge 2000). Across a wide taxonomic spectrum, many long-lived species appeared to cluster within one of two general ecological and/or life-history criteria: (1) species that live in either unpredictable environments (e.g., deserts) or where food resources are scarce (e.g., caves; deep water); or (2) species that exhibit extended parental care and/or live in groups with complex or advanced social behavior. This led us to a two-part classification system regarding the life span determinants of species with extended longevity, shown in Table 1, that we believe is general and applies to a wide range of invertebrate and vertebrate species: (1) Environmentally selected life spans. This category includes animals whose life histories evolved under conditions in which food is scarce and where resource availability is uncertain or environmental conditions are predictably adverse part of the time.4 The extended longevity of animals in this category evolved through natural selection. (2) Socially selected life spans. This category includes species that exhibit extensive parental care and eusociality.5 The extended longevity of animals in this category results from natural, sexual, and kin selection.6 This classification system allows us to move beyond the overly simplistic, bivariate approaches common to discussions of the natural history of longevity. The classic correlates of longevity such as metabolic rate, flight, relative brain mass, and defensive morphology can be combined with varying life-history trajectories to yield a more complex but also more complete suite of explanations for the variation in (and therefore evolution of) longevity across various arrays of organisms. And intensive parental care is linked to flight capability in birds and bats,7 which, in turn, is also linked to extended life span. No system of classification is perfect and the one presented in Table 1 is no exception—the categories are not mutually exclusive and therefore some species could be placed in either or both categories.8 Nevertheless, this classification serves as a practical and heuristic tool for considering the evolution of animal life spans. In particular, it provides a general background for closer examination of specific human attributes and sets the stage for addressing questions about the process of longevity extension.

Evolution of extended longevity in wasps: The interactive role of sociality

Classifying long-lived species by life-history characteristics is apposite because it suggests that the evolution of extended longevity is fundamentally

Click to return to Table of Contents TABLE 1 Categories of factors that favor the evolution of extended life span in insects, arachnids, and vertebrates: Selected examples of species or groups, with maximum observed life span (years) in parentheses Selection factorsa and taxonomic group Examplesb

Environmentally selected life spans (conditions of resource uncertainty and/or scarcity) Insects/acarina Heliconius butterflies (0.9); cave beetles (4); orchard bees (1); paper wasp queens (1); soft ticks (20); flour beetles (4); monarch butterfly (1); African locusts (2); tarantulas (10); bed bugs (1) Mammals Dromedary camel (40); horseshoe bat (30); little brown bat (30); flying fox (31); African wild ass (47); white rhinoceros (45); short-nosed echidna (50); wombat (26); gray seal (46); crabeater seal (39); ringed seal (46); Caspian seal (50); kangaroo rat (10); Greater Egyptian gerbil (8); hairy armadillo (20); rock hyrax (14) Birds Great blue heron (23); marabou (41); southern ground hornbill (70); sandhill crane (22); whooping crane (40); Cheriway caracara (26) Reptiles and amphibians American toad (30); Japanese giant salamander (55); Eastern hellbender (25); greater siren (25); common boa constrictor (39); American alligator (56); Chinese alligator (38); West African dwarf crocodile (42); salt water crocodile (42); slowworm (54); Mexican burrowing python (33); loggerhead turtle (33); common snapping turtle (47); alligator snapping turtle (59); stinkpot (55); Eastern indigo snake (25); San Diego gopher snake (20); Aldabra tortoise (152); tuatara (77) Fish White sturgeon (100); Beluga sturgeon (118); European eel (88); carp (38); wrasse (53); sole (45); rockfish (140); Bunnylake brook trout (24); arctic char (41); spur dogfish (60) Socially selected life spans (kinship, cooperation, parental care, monogamy, and helpers) Insects/acarina Tsetse fly (1); dung beetles (2); Bembix spp—progressive provisioning wasp (1); Polistes spp (1); aposematic Saturnid butterflies (1); bumblebees (1); ant queens (30); termite queens (30) Mammals Bowhead whale (40); blue whale (65); pilot whale (63); sperm whale (70); golden jackal (16); spotted hyena (40); naked mole rat (21); black-legged mongoose (16); humans (122); white-faced capuchin (47); squirrel monkey (27); gray- cheeked mangabey (33); yellow baboon (45); chimpanzee (60); orangutan (59) Birds African grey parrot (73); golden-naped parrot (49); red and blue macaw (64); trumpeter swan (33); mute swan (27); ring-billed gull (32); scrub jay (16); Royal albatross (58); northern fulmar (48); Manx shearwater (30) Reptiles and amphibiansc Rosenberg’s tree frog (3.5) Fishd Basking shark (32); dusky shark (30); hound shark (28); nurse shark (25); whale shark (70) aThe two general categories—“Environmentally selected” and “Socially selected”—are not mutually exclusive. Environmentally selected life span extension may allow overlapping generations and sociality; sociality may allow expansion into otherwise inhospitable environments. bScientific names given in Carey 2001b; Carey and Judge 2000. cCare of the mass of developing eggs and/or the hatchlings is exhibited in some members of all three major groups of amphibians (Wake 1998); however, no species of either amphibian or reptile is truly social and only a small number of reptilian species such as crocodilians and pythons exhibit any level of parental care (protection). dThe shark species listed here all gestate internally (live-bearing). SOURCES: Modified from Carey 2001b; Carey and Judge 2000; unpublished database.

Click to return to Table of Contents J AMES R. CAREY / DEBRA S. JUDGE 415 different in the social versus solitary species (Table 1): the extended longevity of species in these two categories was a response to different types of environmental and socio-ecological circumstances. In this section, we consider the coevolution of longevity extension and sociality in social wasps in a sequence of grades (Carey 2001a; Carey and Gruenfelder 1997; Evans 1958) that can be seen logically as evolutionary steps (Hölldobler and Wilson 1990; Wilson 1971). Understanding this evolutionary progression from short-lived, solitary species to long-lived, eusocial species facilitates understanding of the process of longevity extension in a broad array of social species. This schema is summarized in Table 2. Nesting behavior and sociality result in improved micro-environmental conditions, which foster greater survival; and greater survival improves conditions for increased provisioning and more-intensive social organization, creating a positive feedback and longevity self-reinforcement. Incipient sociality creates conditions for the evolution of incremental increases in longevity that, in turn, create conditions for the evolution of more complex and innovative social structure.

TABLE 2 Evolutionary stages in wasp life span and social complexity Ranges of Stage Description Key concepts and/or events life spans (days)

Ia Solitary Parasitoids; host as incubator 10–60 of parasitoid larvae IIb Nest as locus Nest provides a protected micro- 30–90 environment for female and brood; thus different selective factors operate; selection reduces senescence rates and increases life span IIIc Extensive parental care Extensive parental care increases life 60–365 span up to 1 year; larval and pupal mortality reduced to near zero; this fosters reduction in fertility; allows females to invest more resources for their own maintenance and for rearing their offspring IVd Colony and queen concepts Female’s life further prolonged and 180–1,000 overlaps with progeny, which remain in nest to form an extended family (helpers); the colony rather than the individuals within it becomes the unit of selection, with the fate of the queen inextricably tied to fate of the colony; new level of individualism emerges (colony) and thus life of queen becomes adaptive to long- term colony needs a Example: spider wasp, Natocyphus b Example: sand nesting wasps, Haploneurion; Amophila c Example: progressive provisioning Gorytes spp.; and primitive paper wasps, Stenogaster spp. d Example: advanced eusocial wasps including yellow jacket, Vespula spp. SOURCES: Carey 2001b; Evans 1958; Hölldobler and Wilson 1990; Wilson 1971, 1977.

Click to return to Table of Contents 416 L IFE SPAN EXTENSION IN HUMANS

This positive feedback is based on the demographic relationship between increased offspring survival and reduced fertility, ceteris paribus. Because of the reduced costs of reproduction and parental care, female wasps with fewer offspring remained healthier longer and raised healthier offspring with higher survival rates, which, in turn, fostered additional reductions in mortality and reproduction. Greater longevity of parents also increased the likelihood that they could contribute as grandparents to the fitness of both their offspring and their grand-offspring. Inasmuch as the principles of social evolution are general (Wilson 1977), it follows that the pattern of evolution of longevity extension in social species will also be general. The main point is that the ongoing process of longevity extension in social species adds age classes that increase the fitness of other age classes. This creates a dynamic that changes the rhythm and synchrony of life cycle events and thus alters the qualitative properties of the life history.

Foundational principles: Longevity extension in social systems

This section describes three basic principles concerning longevity that emerge from the wasp example and will serve as the foundation for our theory that longevity extension is self-reinforcing in social animals. The underlying concept is that sociality initially evolved via natural selection to increase survival and/or reproduction through sharing and helping (Bonner 1980; Wil- son 1977); that is, the proximate rewards of cooperation (including status, power, sex, access, comfort, and health) are reflected in the universal bottom line of fitness, including greater longevity and reproduction (Wilson 1998). However, these conditions then became both cause and consequence of extended adult longevity as an outcome of natural and kin selection. The evolutionary theory of aging (Principle 1) serves as the basis by which adult longevity is initially extended, which then sets the stage for intergenerational transfer (Principle 2) and division of labor (Principle 3), both of which underlie the self-reinforcing dynamic of prolonged life span.

Principle 1: Evolutionary theory of aging

Medawar (1957) proposed that if deleterious hereditary factors are expressed at some intermediate age of an individual, and if the age of this expression is both variable and heritable, then selection will weed out earlier expressions more effectively than later expressions, delaying the average age of expression and increasing longevity. As the force of selection is reduced by the declining reproductive value9 of increasingly older individuals, those deleterious traits will accumulate—resulting in a mosaic and variable pattern of age-specific infirmity and thus senescence (Kirkwood 1997). This argument requires that reproductive value (Frank 1998) decrease with age,

Click to return to Table of Contents J AMES R. CAREY / DEBRA S. JUDGE 417 and this will be generally true owing to stochastic sources of mortality even in the absence of physiological deterioration. As more individuals live longer (especially because of reductions in early life mortality), the force of selection increases at later ages, weeding out later-expressing deleterious alleles; selection for somatic maintenance is prolonged; and senescence is delayed (Charlesworth 1994; Medawar 1957; Roff 1992). The result is extended life spans.10 Williams elaborated this line of thinking by noting that pleiotropic genes may have both beneficial and deleterious effects, and, to the extent that beneficial effects precede deleterious effects, the genes may be selected for even in the face of their positive effect on mortality later in life (Williams 1957).

Principle 2: Intergenerational transfers

Investment of resources in reproduction can be extended to investment of resources in offspring and other relatives. Increased per capita investment in offspring decreases juvenile mortality, increases the health and well-being of offspring, and thus improves adult health and survival. At the extreme, older parents can “bankroll” reproductive offspring and thus increase their production of grandchildren (Hill and Kaplan 1999; Kaplan et al. 2000). In societies where different age classes interact, natural selection will favor net transfers from old to young because the reproductive value of young individuals is greater (Frank 1998). However, intergenerational transfers can flow up as well as down generations (Lee 1997). If investment from older to younger individuals decreases mortality, then selection for increased longevity results and that investment can flow for a longer period. Selec- tion for longevity also increases the return on higher levels or prolonged periods of investment. Intergenerational transfers are most prolonged in highly social species (Wilson 1971, 1977). While transfers of genetic information and resources from parents to offspring contribute to the evolution of longevity, additional forms and pathways of intergenerational transfer can elaborate and increase the rate of impact on longevity extension. Social learning and community-level organization increase (a) the pathways and volume of intergenerational transfers of information and (b) the benefits that accrue from innovation (Boyd and Richerson 1985); and these increases in turn should be associated with more-extreme longevity extension than occurs in related species with fewer transmission paths. In humans the transmission of longevity-promoting genes and resources is augmented by culturally transmitted information (or “memes”: see Dawkins 1976).

Principle 3: Division of labor

Division of labor distributes related tasks among members of a population in a coherent system of symbiotic relationships (Petersen 1986; Smith 2000 [1776]). Such division is based on specialization of function: individuals use

Click to return to Table of Contents 418 L IFE SPAN EXTENSION IN HUMANS their differences in skills and resources to best advantage (Abramovitz 1989), and the energy returns per unit of labor thereby increase.11 The concept, fundamental to economics and business, as well as to biology and anthropology, is the basis for increasing efficiency at all levels of organization— from enzymes and cells to organisms and societies. In addition to increased efficiency, division of labor: (1) accentuates individual differences; (2) produces new specialization through finer and finer task division, and (3) increases interdependencies—the entire colony will die if the queen dies; families are at great risk if a parent is lost (Wilson 1971). All of these outcomes have a direct bearing on longevity extension because they reduce innovation costs, increase the resource-use efficiency and thus the net resource return to the (kin) group, and increase the behavioral repertoire of the colony or community as a whole. The concept of a complex, integrated society is inextricably linked to the dual phenomena of division of labor and individual continuity, and, by extension, to individual specialization and longevity.

A model of longevity extension

Our objective is to describe a general model of longevity extension in social species that builds on the three fundamental principles described above and on basic demographic and life-history tenets. A primary tenet is the cost of reproduction (Bell and Koufopanou 1986; Partridge and Harvey 1988; Reznick 1985): allocation of energy to reproduction removes its availability for somatic growth and repair and thus is reflected in increased mortality. The second tenet is that demographic/epidemiologic transition theory is based on declines in mortality that precede declines in fertility (Caldwell 1976; Demeny 1968; Keyfitz 1985; Notestein 1945). Changes in patterns of disease result in shifts in ages of deaths as well as changes in the age-sex structure of the population (Omran 1971; Preston, Keyfitz, and Schoen 1978; Ross 1982). The three principles presented in the previous section serve as the foundation for demographic shifts, including mortality reduction in infants, reduction in numbers of births, improvement in parental health, and increase in offspring quality. All of these factors both contribute to and interact with longevity. Although we draw many of the concepts from the demographic literature, we believe that the model is general and applies to a wide range of both vertebrate and invertebrate species.

I. Reduced infant mortality

All else equal, increased survival from birth to sexual maturity improves energetic efficiency to the parent by decreasing reproductive waste (offspring who die prior to adulthood but after parental investment) and allows parents either to produce more young or to invest more per capita in existing

Click to return to Table of Contents J AMES R. CAREY / DEBRA S. JUDGE 419 young (Clutton-Brock 1991; Preston and Haines 1991; Szirmai 1997). Inas- much as early reproduction in growing populations will have a greater effect on fitness than later reproduction (Futuyma 1998), the former populations will outcompete the latter. But, even in growing populations, increased per capita investment can be selected for if the age-specific increase in fitness of the offspring more than offsets losses owing to numbers of young forgone. Increased resources available for reproduction can have any of the following effects depending on which age classes can access them: (1) increase the number of young, (2) increase the survival of the young, (3) delay maturity of the young by allowing the parental generation to support overlapping sets of offspring, (4) reduce the cost of reproduction to mothers through “grandmothering” (Hawkes et al. 1998). The first outcome will have no necessary effect on longevity while the other three outcomes are conducive to longevity reinforcement through increased size, improved health, and better-developed capacity for resource acquisition skills.

II. Demographic transition

A general observation from studies on the comparative life histories of both invertebrate and vertebrate species is that an increase in offspring survival is followed by a decrease in fertility (Clutton-Brock 1991; Ricklefs 1979). Most demographers and sociologists agree that one of the basic causes of a general decline in fertility is a reduction in mortality that reduces the number of births necessary to have any “desired” number of children (Mont- gomery and Cohen 1998; Preston 1978; Becker 1991: 135–154). Parents (especially mothers) with fewer offspring remain healthier because of reduced energy allocation to early-stage parental effort (especially pregnancy and lactation) in contrast to effort expended on older children, and the result is healthier young offspring with even higher survival rates at both young and old ages (van de Walle 1986). The decline in fertility may also be un- derstood as a balance between growing family size and declining economic advantage of the family (Buikstra and Konigsberg 1985). The transition from simple agriculture to an industrial economy devalues children as early pro- ducers. A young person on a farm becomes a useful member of the labor force at an early age (Cain 1977; Kaplan 1996), while a young person in industrial or postindustrial society often requires an extensive and expen- sive education (Ryder 1959). Kaplan et al.’s model of embodied human capital offers insights into the tradeoffs inherent in providing necessary skills to offspring versus producing additional offspring (Kaplan et al. 2000) and suggests that the inherent mechanisms of demographic transition characterized pretransitional human societies as well (Kaplan et al. 2000). There is some empirical evidence that mothers in traditional horticultural/pastoral societies produce optimal numbers of offspring to maximize their subse-

Click to return to Table of Contents 420 L IFE SPAN EXTENSION IN HUMANS quent reproductive success (Borgerhoff Mulder 2000). Where heritable resources influence reproductive success, there appears to be a tradeoff between numbers of offspring produced and the per capita wealth that is passed on to heirs (Borgerhoff Mulder 1998).

III. Improvement in parental health and survival

Parents experiencing fewer births but unchanged levels of reproduction because of higher rates of child survival remain healthier. Consequently parents themselves experience higher survival and can invest more of their resources in a smaller number of “high-quality” offspring—healthier, larger, more competitive (Clutton-Brock 1994), skilled, and innovative (Preston and Haines 1991; Richard 2001). That healthier mothers produce healthier infants (Wood 1994) and that excessive maternity demands deplete maternal health are well-documented assertions (Hrdy 1999). In addition to intergenerational effects, reducing numbers of reproductive attempts without sacrificing reproductive output will improve the health and survivorship of first-generation mothers. The fitness benefits of this trajectory are even greater in species with long and costly post-infancy dependency. For example, improving the health and nutrition of girls and women and reducing maternal depletion improves the health of subsequent generations because healthy, better-educated mothers can provide high-quality care over an extended period, including such investments as prolonged breastfeeding, increased vigilance, and longer education. Improved survivorship and competitive ability increase the number of productive adult (parent and grandparent) years relative to the number of juvenile/sub-adult years as well as the per-year return; years of additional productivity can further expand intergenerational transfers (Hawkes et al. 1998; Kaplan et al. 2000).

IV. Increase in offspring quality

A second result of a decrease in the number of offspring can be an increase in the average quality of a smaller number of offspring because the per capita amount of both depreciable care (e.g., food; nutrients) and nondepreciable care (e.g., vigilance; teaching) will increase (Clutton-Brock 1991). For example, in humans a decrease in the number of children will increase the average quality of a smaller number of offspring in two respects. First, the nourishment an infant receives from its mother mediates its early mortality risk from infectious diseases; and nutrition and exposure to infection after birth determine an infant’s susceptibility to disease in later life (Barker 1994; Elo and Preston 1992). Thus healthy mothers may prevent the early onset in their offspring of the degenerative diseases of old age that are linked to inadequate cellular development early in life. The increase in childhood sur-

Click to return to Table of Contents J AMES R. CAREY / DEBRA S. JUDGE 421 vival then fosters even further reductions in fertility and improvements in offspring quality. Second, parents can invest heavily in one or a small number of offspring for a longer period. Development of the brain, of language as a means of transmitting social learning, and of group size are characteristic of hominid evolution (Hammer and Foley 1996). Language as a mechanism for social learning allowed humans to communicate about solutions to ecological problems (e.g., hunting strategies). Cultural transmission increased the production of innovation and the rate at which innovations dispersed, and allowed innovations to be directly targeted toward hazards of the environment that threatened survival. Contemporary studies demonstrated that family size is negatively related to completion of high school (Blake 1989a, 1989b), but this effect is mediated through resources (socioeconomic status). These are modern in- carnations of processes ongoing since the first information revolution in Homo within the last 2 million years. Education is a measure of offspring “quality” that affects longevity in two respects. At the individual level, the relationship between educational attainment and longevity is well documented (Hoyert, Kochanek, and Murphy 1999). The correlation between these two is mediated by intelligence (which is susceptible to poor prenatal and post- natal environments). At the societal level, economic growth can be attributed to advances in science and technology made possible through education and training of highly skilled specialists.12 The focus of contemporary specialists on biomedical research (Cruse 1999; Nutton 1996; Szirmai 1997; Thomas 1977) is a natural outcome of the same processes as was the use of fire to cook, to keep warm, and to deter predators by Homo erectus.

Incremental increase in longevity as cause and consequence

The culmination of the effects of the four factors discussed above is an overall increase in longevity that is related to the extent and speed of changes in resources, quality, and productivity. Lovejoy (1981) noted that the very social bonds, intelligence and learning, and intensive parenting that contributed to reducing mortality levels in primates both required and contributed to longevity in higher primates. Reduced numbers of offspring (i.e., litter size) and reduced mortality and increased parental investment are characteristic of a broad array of organisms with prolonged life spans. As generational overlap extends, healthy adults live longer and thus become healthy grandparents who can help rear offspring and further increase the subsidies to development discussed above. In humans this process is exaggerated through the ability to transmit symbolic information across and within generations and to do so at a tempo substantially faster than generation times. Solutions to environmental or physiological hazards are transmitted at high

Click to return to Table of Contents 422 L IFE SPAN EXTENSION IN HUMANS speed across all age classes of the population rather than solely to younger age classes through genetic modes. Incremental increases in longevity increase survival at early ages through increased subsidization of development. This new and expanded way of viewing the “contract between the generations” is invoked for economic transactions across age classes (Szirmai 1997). When intergenerational transfers occur at the pace of generation times (via natural selection), the process is self-promoting but proceeds more slowly than when multiple means and directions of information and resource flows are possible (Boyd and Richerson 1985). In humans the cognitive ability to target innovation to solve specific problems can result in technologies that further improve health and do so much more rapidly than could be accomplished by natural selection on random genetic variation (Bonner 1980; Durham 1991). Better-educated, more-innovative children then increase the rate of those technological advances (Chant 1989) that contribute to health and longevity (Cruse 1999; Nutton 1996) and to higher adult survival to older and older ages. Innovation is spurred by the increased return through prolonged later life13 to the increased number of surviving older adults who can benefit. Furthermore, expectations of increased benefits accrued through longer life may promote longevity itself (Philipson and Becker 1998).

An application of the model

Human longevity in a biodemographic context

A graphical perspective on the primate life span relative to life spans of other mammalian orders reveals two relationships that provide necessary context. First, the life spans of nearly all primate species are greater than those predicted by body mass alone. Chiroptera (bats) are the only other major order that uniformly exhibits greater life spans than would be predicted from body mass alone. Figure 1 depicts the relative life spans for species in 62 families from nine orders of mammals. The family means were incorporated into a linear regression of life span on body mass. Case deletion regression resulted in a predicted life span based on mass for each mammalian family. The relative life spans in this figure are the simple differences between observed and expected values. Primates are long-lived mammals, with life spans of most species ranging from 1.25-fold to 1.5-fold greater than that predicted from body mass alone (Judge and Carey 2000). Hu- mans, in turn, are long-lived primates (Figure 2) with a life span over 2- fold greater than that predicted from body mass. Including primate brain mass in predictive regressions increased the predicted human life span into the region of eight to nine decades depending on the comparison group (Judge and Carey 2000). Anthropoid primates are relatively large-brained, social creatures with complex forms of communication. Language, as the

Click to return to Table of Contents FIGURE 1 Relative life span for nine mammalian orders 0.6

0.4 B B 0.2 B B

0 B B B –0.2 B

–0.4 Relative life span B –0.6

–0.8

–1.0 Sirenia Proboscidea Chiroptera Primates Cetacea Carnivora Hyracoidea Artiodactyla Rodentia Order NOTE: Relative life span is the record life span less the life span value predicted from mammalian regression of life span on body mass. Predictions were calculated using ordinary least squares regression with log transformed values and case deletion. Sample comprises average logged values of species for 62 families of mammals in 9 orders. Means are indicated within boxes including 25th to 75th percentiles of life span values for each order; vertical lines indicate range. SOURCE: Carey and Judge 2000 and unpublished data.

FIGURE 2 Record life spans by body size for primate genera

Homo sapiens (contemporary) 2.0 100

80 Homo sapiens (hunter-gatherer) 1.8

1.6 40

1.4 Record life span (years)

Record life span, log (years) 20

1.2 Old World monkeys and apes New World monkeys Prosimians

1.0 10 0.0 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 Female body mass, log (kg+1) NOTE: Mean record captive life spans (years) by female body mass (kg+1) for 53 genera of primates. The left axis and abscissa are the log transformed values used in ordinary least squares regressions yielding a positive relationship with r = .817, p < 0.001, R2 adjusted = .660); the right axis provides values in years. The statistical outlier in the midrange of body mass is the New World monkey genus, Cebus, with a mean life span record of 46 years. SOURCE: Adapted from data presented in Judge and Carey 2000. 424 L IFE SPAN EXTENSION IN HUMANS co-evolutionary product of social and biological processes, changed the nature of information transfer across and within generations (Deacon 1997: 349–365). Potential life span appears to have increased rapidly in Homo with predicted life spans into seven decades for Homo erectus, between eight and ten decades for Homo sapiens, to a current record of 122 years (Allard, Lèbre, and Robine 1998)—over twice the life span of any other documented primate. Evidence of extensive intergenerational subsidization is widespread, giving rise to hypotheses regarding the role of grandmothers (Hawkes 1997; Hawkes, O’Connell, and Blurton Jones 1998) for which corroborative data are mixed (Hill and Hurtado 1991) and regarding the role of parental investment more generally (Kaplan 1996). We believe that our model helps to explain the role of intergenerational transfers in the evolution of longevity and suggests that the specific transfer pathways may differ under varying ecological circumstances.

Primate origins of longevity extension

Elsewhere we (Judge and Carey 2000) combined the primate life span and body mass data presented in Figure 1 with brain size to develop a regression equation for predicting the evolved life span of both modern humans and ancestral hominids (Table 3). The analysis revealed that humans, both archaic and modern, have a primate morphology that predicts a life span of 72 to 90 years—well beyond the age of menopause in modern females.14 These predictions also exceed other estimates (Angel 1969; Buikstra and Konigsberg 1985; Washburn 1981). With a primate longevity advantage, reduction of litter size to one, and shifts to a higher-quality diet that required more skill to obtain (Kaplan 1996), hominids appear to have entered into a longevity feedback loop. Fossil evidence indicates that Homo erectus demonstrated larger stature (McHenry 1994) and slower dental development, suggesting longer inter-birth intervals than earlier hominoids. This suggests overall increases in per-offspring parental investment rather

TABLE 3 Estimates of longevity for hominids Incremental Life span change in Hominid species (years) life span (years)

Australopithecus afarensis 46.6 Homo habilis 55.0 8.4 H. erectus 62.0 7.0 H. sapiens (prehistorical) 72.9 10.9 H. sapiens (contemporary) 122.0 49.1

SOURCES: Judge and Carey 2000; Hammer and Foley 1996.

Click to return to Table of Contents J AMES R. CAREY / DEBRA S. JUDGE 425 than translation of higher-quality diet into more offspring. Greater adult life expectancy and generational overlap are required for the intergenerational transfers of resources, knowledge, and skills in later Homo. Reduc- tions in juvenile and sub-adult mortality, through adult subsidization of juvenile diet, probably increased life expectancy at birth as well as age at sexual maturity. Thus, additional investment in individual young by parents was more likely to have a fitness payoff (Kaplan et al. 2000).

Prehistoric and historical patterns of longevity change

Human life spans that approach and sometimes exceed the “design capacity” indicated from primate allometry are not unusual in contemporary Western societies and may have been more common in early hunter-gatherer societies than is generally realized.15 Studies of extant hunter-gather- ers find that a significant proportion of those reaching adulthood live into their 60s (Hill and Kaplan 1999). As more people reached older ages, investments in health and well-being were more likely to yield returns, in terms of prolonged working life span, to more individuals. Historical developments substantially reduced early mortality and more recently allowed technological attenuation of early-senescing systems (e.g., through eye glasses; dentures) (Fogel 1997; Landes 1998). Only in the last quarter-century have humans pressed beyond the “design capacity for longevity,” extending that capacity with large-scale systemic innovations such as joint replacement and organ transplants (Nutton 1996; Smith 1993; Thomas 1977). The self-reinforcing process of increasing life expectancy, increasing age-specific productivity, and thus increasing demand and resources for health improvements should occur only to the extent that productive life span is increased along with absolute life span.

Health, wealth, and longevity

It is intuitively obvious that wealth is conducive to longevity—wealthier countries have higher life expectancies and larger proportions of very old citizens (Bloom and Williamson 1998). At the population level, moreover, increases in longevity also are conducive to improvements in wealth and thus create a virtuous spiral. Four main categories of relationships may account for this spiral (Bloom and Canning 2000): (1) Productivity. Longer- lived populations tend to be healthier and thus have higher labor productivity because workers are physically more energetic and mentally more robust for longer. Societies in which a large fraction of the population at- tains older ages can have a more extensive division of labor. (2) Teaching, learning, and education. Long-lived people are repositories of knowledge, wisdom, and technical expertise and therefore are teachers and educators.

Click to return to Table of Contents 426 L IFE SPAN EXTENSION IN HUMANS

Healthier people who live longer have stronger incentives to invest in developing their skills because they expect to reap those benefits over a longer period. Increased technical training and schooling promotes greater productivity and higher-quality output and thus produces higher income. (3) In- vestment in physical capital. Improvements in health and longevity create an incentive to plan for the future, including the need for savings, which, in turn, promotes greater investment. Investment promotes employment, and workers will thus have access to more resources and their incomes rise. A healthy, long-lived, and educated workforce acts as a strong magnet for investment by outsiders. (4) Demographic dividend. The transition from high to low mortality rates, particularly in the young, typically triggers declines in fertility rates. This transition, in turn, temporarily increases the proportion of the population of working age whose contributions lead to higher per capita productivity. In addition, populations with large cohorts of older (retired) persons actually stimulate economic growth, presumably because they either work themselves or enable others to work by minding their children, and many continue to impart their accumulated knowledge to others.16

Implications of a longevity-oriented theory

The theory we advance in this article has several implications for speculation on the human future. First, expansion of the theoretical foundations of the finitude of life provides a new framework for discussing the future of human life span (Hayflick 2000; Manton and Stallard 1996; Vaupel et al. 1998; Wilmoth et al. 2000). In particular, it shifts discussion from the issue of mortality reduction based on putative limits imposed by the evolutionary theory of aging toward a more inclusive process of longevity extension of which reduction in the aging rate is one component. Second, a theory of longevity raises questions about the interpretation of studies on the biological mechanisms of aging that rely on model species such as fruit flies, laboratory rodents, and nematode worms, all of which are solitary (nonsocial) species. Inasmuch as our theory suggests that extended longevity in solitary species evolved under different ecological contexts than in social species, the underlying mechanisms of aging in animals in these two broad categories may be different. For example, the aging response to caloric restriction (Sohal and Weindruch 1996; Weindruch 1996) in solitary species, which must survive independent of a social group, may be fundamentally different from the mechanisms in social species with evolved behaviors for helping, sharing, and food storage. Whereas aging- oriented research necessarily focuses on lower levels of biological organization such as the molecule and cell, longevity-oriented theory focuses on the level of organization considered by many biologists to be the quintes- sence of biological relevance—the whole organism. All discoveries related

Click to return to Table of Contents J AMES R. CAREY / DEBRA S. JUDGE 427 to the rate of aging at lower levels of biological organization must ultimately be tested at the level of the individual. Third, our theory suggests that life span should be considered more central to life-history theory than has been the case. The concept of life span either is absent from essentially all of the mainstream texts in ecology (Begon, Harper, and Townsend 1996), evolution (Futuyma 1998), and population biology, or is discussed as an outcome variable of body size and mortality (Roff 1992: 120–121). Classic life-history theory treats life span as an outcome of the relative mortality of juvenile and adult stages and of tradeoffs between early and late reproduction (Stearns 1992). Inasmuch as life-history theory is concerned with tradeoffs, questions concerning tradeoffs between extraordinary longevity and other traits are central to this area of inquiry.17 Fourth, a longevity-oriented theory concerns human longevity in the broader context of life history (Hill and Kaplan 1999) and culture (Durham 1991). Whereas most anthropologists consider many traits such as covert ovulation, menopause, and speech as uniquely human, and bipedalism, large relative brain size, and the use of tools as exceptional, extended longevity remains unappreciated. Human life span should be considered nearly as extraordinary as is human brain size.

Human life span extension: A framework for the future

The extension of human longevity occurs through the same biological processes as in other organisms. The difference is not in the relationships of reproductive value, intergenerational transfer, and specialization, but in the mechanisms by which those characteristics are manifested. Intergenerational transfers are expanded beyond genetic and immediate resources to include the information necessary to increase the production of resources, and through information about ecological problems that allows targeted adaptive responses. Specialization is developed beyond what is found in other species through these same unusual mechanisms. The theory presented here extends Fogel’s concept of technophysio evolution (Fogel 1994, 1997; Fogel and Costa 1997), which states that mortality reductions in human populations occur because our species has a level of control over its environment that is unprecedented among animals. The model demonstrates the link between the Fogel model and Becker’s “demand for children,” human capital, and specialization (Becker 1991: 135– 154; Becker and Murphy 1992). We argue that advances in human longevity are the result of certain special conditions within a broader evolutionary model applicable to animals more generally. This self-reinforcement model of longevity extension in humans integrates evolutionary (sociobiological)

Click to return to Table of Contents 428 L IFE SPAN EXTENSION IN HUMANS and economic processes that collectively apply to longevity extension in many social species. In other words, sociality evolved in many groups in response to the perceived advantages of controlling the physical and biological environment. Advantages include protection from predators, control over temperature and humidity, and food acquisition, storage, and production (Hill and Kaplan 1999; Turner 2000). Generalizing the concept reveals that humans are simply one of a number of social species in which longevity extension has occurred and is continuing to occur as a result of basic principles linking sociality and longevity extension. Our theory expands Fogel’s time horizon from 300 years of modern technological development to the entire history of human technological development—on the order of some 2 million years. Early technological developments influenced longevity through environmental modifications including the use of stone tools, control of fire, shelter construction, invention of farming, and the emergence of bureaucracies (Foley 1992; Zvelebil 1994). Subsequent technological changes in communication and organizational efficiency included the invention of writing and improvements in information transfer; improvements in transportation, industry, and sanitation (Cruse 1999; DeGregor 1985); and continuing scientific specialization such as cloning, genomics, and molecular techniques in biomedicine (Collins 1998; Gurdon and Colman 1999; Silver 1997). The declining costs of coordination associated with the advances in information technology over the past century may have contributed to increased specialization, to increased rates of scientific progress, and to the major increases in observed human life spans over the same period (see Becker and Murphy 1992). We believe that one of the most important contributions of a longevity-oriented theory is to re-frame the way scientists consider the future of human life expectancy. The more conventional aging-oriented question is, “How low do mortality rates have to fall to substantially increase longevity in the future?” (Fries 1980; Olshansky, Carnes, and Desesquelles 2001), whereas the longevity-oriented question is, “To what extent will future gains in longevity (or improvements in mortality) extend life expectancy further?” Whereas the aging-oriented question assumes that age-specific improvements in survival exert their impact only at older ages, the self-reinforcing model argues that age-specific decrements in late-life mortality can also have a positive effect on early survival and productivity through selection, intergenerational transfers of resources and information, and increasing innovation from specialization. The longevity-oriented question recognizes a built-in dynamic whereby the effects of reducing mortality at one age are spread across multiple ages and thus amplify the effects of longevity extension. The longevity-oriented approach also suggests that the present trajectory of science and technology, including organ cloning, xenotransplantation, and molecular medicine, will lead to sweeping transformations in health and to what Wilson (1998: 299) refers to as “volitional evolution”—the de-

Click to return to Table of Contents J AMES R. CAREY / DEBRA S. JUDGE 429 commissioning of natural selection so that individuals can make choices conducive to engineering a long life for themselves and their children (Bal- timore 2001). As de Grey (2000) notes, human scientific knowledge and consequent technological prowess will continue to advance at a non-negli- gible rate for as long as civilization survives. It is thus likely that human longevity, too, will continue to advance and to have lasting and transfor- mational effects on society.18

Conclusions

To understand human longevity we must look not only to our ability to control our environment, but also to our phylogenetic and historical legacy (Coe 1990; Foley 1995). Parental care is part of our mammalian heritage, as is the tendency for females to be more risk-averse and males to be more risk-prone. Our anthropoid heritage produced long gestation times, single infant births, and long inter-birth intervals consonant with high levels of maternal effort; and sociality is a firmly established core heritage. Brain expansion and use of manufactured tools were linked to increased longevity in the genus Homo. With the fully modern brain, anatomy, and behavioral repertoire of Homo sapiens came an awareness of our own mortality and directed attempts to escape it. Our modern human heritage includes living in large groups through the development of agriculture and subsequent sophisticated systems of production (Diamond 1998). Humans were gradually able to order their experience more consciously in accordance with a plan (Southern 1953: 222). They replaced the concept of life as an exercise in endurance with the thought of life as a seeking and journeying. We are now in the era of emerging ability to control our actuarial destiny in response to the desire in humans throughout history to live comfortably and to delay death.

Notes

The concept for this article was initially pre- viding guides to a diverse literature, discus- sented at a research workshop sponsored by sion, and/or constructive comments on the the International Union for the Scientific manuscript. The research was supported by Study of Population (IUSSP) held in Mont- Grant AG-08761 from the National Institute pellier, France, 21–25 October 2000. The au- on Aging. thors are grateful to the IUSSP Committee 1 These distinctions are made by Gold- for their encouragement and thank Steven wasser (2001). Life span is a general term that Austad, Shelley Cargill, Ed Caswell-Chen, needs an adjective for specific definitions. Lon- Carl Chen, James Curtsinger, John Edman, gevity is more general and not related specifi- Leonard Hayflick, Mark Hayward, Shiro cally to time units. Horiuchi, Hillard Kaplan, Peter Lindert, Marc Mangel, George Martin, Linda Martin, Jean- 2Self-reinforcing is synonymous with the Marie Robine, Thomas Scott, Linda Styer, terms “positive feedback” used in systems en- Jacques Vallin, and John Wilmoth for pro- gineering and “autocatalytic processes” used in

Click to return to Table of Contents 430 L IFE SPAN EXTENSION IN HUMANS biochemistry; it refers to a process that accel- explain the spectrum of sociality seen in the erates at a rate which increases with time be- family Bathyergidae (Faulkes 1998). Con- cause the process catalyzes (or feeds on) itself. straints imposed by ecological factors (unpre- 3 For example, body mass and life span dictable rainfall) causing widely distributed are negatively correlated within the Pinnipedia edible roots led to living in large social groups (seals and walruses) and are uncorrelated with cooperation and reproductive suppression within the Chiroptera (bats). to better exploit an ecological niche where solitary animals would be unlikely to survive. The 4 Some of the longest-lived small and me- extended longevity of mole rats is first a re- dium-sized mammals, including gerbils, rock sponse to environment that is then augmented hyrax, and feral asses, live in deserts where by the mole rats’ sociality. rainfall and thus reproduction are episodic and unpredictable (Bronson 1998). 9 Reproductive value is the average expected reproduction remaining at a given age. 5 Eusocial insects (ants, bees, wasps, ter- mites) have overlapping generations, are co- 10 Evolutionary theory of aging postulates lonial, and show cooperative care of young and that the force of natural selection will decrease a reproductive division of labor. The naked with age when populations are steady or in- mole rat, Heterocephalus glaber, is the only mam- creasing in size. The theory provides the con- mal to display true eusociality. This sub-Sa- ceptual foundation for two population genetic haran species lives in colonies of 10 to 100 in- hypotheses to explain aging: (1) negative pleio- dividuals with a single dominant “queen” that tropy—alleles that have beneficial effects on fit- selects and mates with one to three breeding ness early in life can also have deleterious ef- males. Longevity in captivity can exceed 21 fects on fitness later in life (Williams 1957), years (Faulkes 1998). and (2) mutation accumulation—a decrease in 6 Evolutionary theory considers three the force of natural selection with age results types of selection, with variations within each in little selection on deleterious mutations of the types (Futuyma 1998). Natural selection whose effects are confined to late life. is the direct reproductive outcome of differ- 11 Three principles emerged from Adam ential survival and/or mortality of individuals. Smith’s Wealth of Nations. Division of labor is In sexual selection traits are selected as a func- worthwhile when there is increased efficiency tion of their contribution to acquiring more or of production; it pays off only if the market is better mates. In kin selection traits are repro- sufficiently large; and the extent of division of duced differentially through their beneficial ef- labor depends on the ease of communication fects on the reproduction of other relatives. within a market of a given size (Smith 2000 7 In mammals, flight ability and large lit- [1776]). ter size are incompatible life-history traits. In- 12 Just as highly trained industrial work- sectivorous female bats could not forage effi- ers in the nineteenth and twentieth centuries ciently while carrying multiple fetuses. replaced untrained agrarian workers, highly Producing singleton births over a long pe- educated technologists (those who work with riod—i.e., the production of replacement off- their hands and use theoretical knowledge, spring over the course of a long life span rather such as X-ray technicians and computer op- than the production of large litters—allows erators) of the twenty-first century “knowledge higher per capita investment per pup. society” will replace industrial workers While flight may reduce adult mortality in (Drucker 1994). birds, development of eggs in a nest rather than 13 Long life expectancy is one of the most fetuses in a uterus also selects for increased important explanatory factors proposed by his- paternal investment and lower clutch and nest- torians to explain why technological develop- ling mortality rates; variation in paternal care ments accelerate. Long life gives prospective should contribute to explaining variation in inventors the years necessary to accumulate avian longevity. technical knowledge, as well as the security to 8 The sociality and extreme longevity of embark on development programs yielding African mole rats have led comparative biolo- delayed rewards (Diamond 1998). Thus, in- gists to develop a “food-aridity” hypothesis to creases in life expectancy brought by modern

Click to return to Table of Contents J AMES R. CAREY / DEBRA S. JUDGE 431 medicine may have contributed to the recently took place until the emergence of agriculture accelerating pace of biomedical innovation and some 12,000 years ago at end of last ice age. It invention. is not at all clear that human health and lon- 14 We find it useful to deconstruct con- gevity improved with the invention of agricul- temporary human life span (with a recorded ture, since the human settlements associated maximum of 122 years; Robine and Allard with farming also cultivated disease (Larsen 1998) into two segments: (1) the Darwinian 1995). or “evolved” segment of 72–90 years; and 16 But as Hillman (1999) notes, common (2) the post-Darwinian segment, the compo- definitions of productivity are too narrow a nent that emerged as a result of improved liv- measure of usefulness. Older parents may pro- ing conditions of modern society. The argu- vide social support, information, and the like ment that human life span has not changed that are not captured in economic definitions. in 100,000 years (Fries 1980; Hayflick 2000) 17 For example, Moyle and Herbold can be considered substantially correct when the (1987) found that a slightly higher percent- “evolved” life span is considered and is likely age of fishes either hide their brood or are live- to include a post-reproductive period in fe- bearers in western North America, where males even before the origin of Homo sapiens stream and lake levels are less predictable, than (Judge and Carey 2000). The argument is in- is the case in eastern North America or Eu- correct when the nonevolved segment of hu- rope. More parental care is associated with man life span is considered (Wilmoth et al. greater longevity. 2000). 18 To borrow a metaphor from Peter 15 The advent of sophisticated tool culture Drucker in reference to long-term social around 40,000 years ago led to efficiencies in change, like “ocean currents deep below the hunting and food preparation, but no major hurricane-tormented surface of the sea” change in nomadic/hunter-gatherer life style (Drucker 1994).

References

Abramovitz, M. 1989. “Manpower, capital, and technology,” in M. Abramovitz (ed.), Thinking About Growth: And Other Essays on Economic Growth and Welfare. Cambridge: Cambridge University Press, pp. 173–186. Allard, M., V. Lèbre, and J.-M. Robine. 1998. Jeanne Calment: From Van Gogh’s Time to Ours, 122 Extraordinary Years. New York: W. H. Freeman. Angel, J. L. 1969. “The bases of paleodemography,” American Journal of Physical Anthropol- ogy 30: 427–438. Austad, S. N. 1997. Why We Age. New York: John Wiley & Sons. Baltimore, D. 2001. “Our genome unveiled,” Nature 409: 814–816. Barker, D. J. P. 1994. Mothers, Babies, and Disease in Later Life. London: BMJ Publishing Group. Becker, G. S. 1991. A Treatise on the Family. Enlarged Edition. Cambridge, MA: Harvard Uni- versity Press. Becker, G. S. and K. M. Murphy. 1992. “The division of labor, coordination costs, and knowledge,” Quarterly Journal of Economics 107: 1137–1160. Begon, M., J. L. Harper, and C. R. Townsend. 1996. Ecology. London: Blackwell Science. Bell, G. and V. Koufopanou. 1986. “The cost of reproduction,” in R. Dawkins and M. Ridley (eds.), Oxford Surveys in Evolutionary Biology. Oxford: Oxford University Press, pp. 83– 131. Beshers, S. N. and J. H. Fewell. 2001. “Models of division of labor in social insects,” Annual Review of Entomology 46: 413–440. Blake, J. 1989a. Family Size and Achievement. Berkeley: University of California Press. ———. 1989b. “Number of siblings and educational attainment,” Science 245: 32–36. Bloom, D. E. and D. Canning. 2000. “The health and wealth of nations,” Science 287: 1207– 1209.

Click to return to Table of Contents 432 L IFE SPAN EXTENSION IN HUMANS

Bloom, D. E. and J. G. Williamson. 1998. “Demographic transitions and economic miracles in emerging Asia,” World Bank Economic Review 12: 419–455. Bonner, J. T. 1980. The Evolution of Culture in Animals. Princeton: Princeton University Press. Borgerhoff Mulder, M. 1998. “The demographic transition: Are we any closer to an evolutionary explanation?” Trends in Ecology and Evolution 13: 266–270. ———. 2000. “Optimizing offspring: The quantity-quality tradeoff in agropastoral Kipsigis,” Evolution and Human Behavior 21: 391–410. Boyd, R. and P. J. Richerson. 1985. Culture and the Evolutionary Process. Chicago: University of Chicago Press. Bronson, F. H. 1998. “Global zones and reproduction,” in E. Knobil and J. D. Neill (eds.), Encyclopedia of Reproduction. San Diego: Academic Press, pp. 452–460. Buikstra, J. E. and L. W. Konigsberg. 1985. “Paleodemography: Critiques and controversies,” American Anthropologist 87: 316–333. Cain, M. 1977. “The economic activities of children in a village in Bangladesh,” Population and Development Review 3: 201–227. Caldwell, J. C. 1976. “Toward a restatement of demographic transition theory,” Population and Development Review 2: 321–366. Carey, J. R. 2001a. “Demographic mechanisms for the evolution of long life in social insects,” Experimental Gerontology 36: 713–722. ———. 2001b. “Insect biodemography,” Annual Review of Entomology 43: 79–110. Carey, J. R. and C. A. Gruenfelder. 1997. “Population biology of the elderly,” in K. W. Wachter and C. E. Finch (eds.), Between Zeus and the Salmon: The Biodemography of Lon- gevity. Washington, DC: National Academy Press, pp. 127–160. Carey, J. R. and D. S. Judge. 2000. Longevity Records: Life Spans of Mammals, Birds, Amphib- ians, Reptiles, and Fish. Odense, Denmark: Odense University Press. ———. 2001. “Principles of biodemography with special reference to human longevity,” Population: An English Selection 13(1): 9–40. Chant, C. (ed.). 1989. Science, Technology, and Everyday Life 1870–1950. New York: Routledge in association with the Open University. Charlesworth, B. 1994. Evolution in Age-Structured Populations. Cambridge: Cambridge Uni- versity Press. Clutton-Brock, T. H. 1991. The Evolution of Parental Care. Princeton: Princeton University Press. ———. 1994. “The costs of sex,” in R. V. Short and E. Balaban (eds.), The Differences Between the Sexes. Cambridge: Cambridge University Press, pp. 347–362. Coe, C. L. 1990. “Psychobiology of maternal behavior in nonhuman primates,” in N. A. Krasnegor and R. S. Bridges (eds.), Mammalian Parenting. New York: Oxford Univer- sity Press, pp. 157–183. Collins, F. S. 1998. “New goals for the U.S. human genome project: 1998–2003,” Science 282: 682–689. Comfort, A. 1961. “The life span of animals,” Scientific American 205: 108–119. Cruse, J. M. 1999. “History of medicine: The metamorphosis of scientific medicine in the ever-present past,” American Journal of the Medical Sciences 318: 171–180. Dawkins, R. 1976. The Selfish Gene. Oxford: Oxford University Press. Deacon, T. 1997. The Symbolic Species: The Co-Evolution of Language and the Brain. New York: W. W. Norton. DeGregor, T. R. 1985. A Theory of Technology. Ames: Iowa State University Press. deGrey, A. D. N. J. 2000. “Gerontologists and the media: The dangers of over-pessimism,” Biogerontology 1: 369–370. Demeny, P. 1968. “Early fertility decline in Austria-Hungary: A lesson in demographic transition,” Daedalus 97: 502–522. Diamond, J. M. 1992. The Third Chimpanzee: The Evolution and Future of the Human Animal. New York: HarperCollins. ———. 1998. Guns, Germs, and Steel: The Fates of Human Societies. New York: W. W. Norton. Drucker, P. F. 1994. “The age of social transformation,” Atlantic Monthly 274: 53–80.

Click to return to Table of Contents J AMES R. CAREY / DEBRA S. JUDGE 433

Dunbar, R. I. M. 1992. “Neocortex size as a constraint on group size in primates,” Journal of Human Evolution 20: 469–493. Durham, W. H. 1991. Coevolution. Stanford: Stanford University Press. Elo, I. T. and S. H. Preston. 1992. “Effects of early-life conditions on adult mortality: A review,” Population Index 58: 186–212. Evans, H. E. 1958. “The evolution of social life in wasps,” Proceedings of the 10th International Congress of Entomology 2: 449–457. Faulkes, C. G. 1998. “Naked mole-rats,” in E. Knobil and J. D. Neill (eds.), Encyclopedia of Reproduction. San Diego: Academic Press, pp. 321–326. Finch, C. E. 1990. Longevity, Senescence, and the Genome. Chicago: University of Chicago Press. Fogel, R. W. 1993. “New sources and new techniques for the study of secular trends in nutritional status, health, mortality and the process of aging,” Historical Methods 26: 5–43. ———. 1994. “Economic growth, population theory, and physiology: The bearing of long-term processes on the making of economic policy,” American Economic Review 84: 369–395. ———. 1997. “Economic and social structure for an ageing population,” Philosophical Trans- actions of the Royal Society of London, Series B 352: 1905–1917. Fogel, R. W. and D. L. Costa. 1997. “A theory of technophysio evolution, with some implications for forecasting population, health care costs, and pension costs,” Demography 34: 49–66. Foley, R. 1992. “Evolutionary ecology of fossil hominids,” in E. Alden Smith and B. Winterhalder (eds.), Evolutionary Ecology and Human Behavior. New York: Aldine de Gruyter, pp. 131–164. ———. 1995. Humans Before Humanity: An Evolutionary Perspective. Cambridge, MA: Blackwell Publishers. Frank, S. A. 1998. Foundations of Social Evolution. Princeton: Princeton University Press. Fries, J. F. 1980. “Aging, natural death, and the compression of morbidity,” The New Eng- land Journal of Medicine 303: 130–135. Futuyma, D. J. 1998. Evolutionary Biology. 3rd ed. Sunderland, MA: Sinauer Associates. Gill, F. B. 1995. Ornithology. New York: W. H. Freeman. Goldwasser, L. 2001. “The biodemography of life span: Resources, allocation and metabolism,” Trends in Ecology and Evolution 16(10): 536–538. Gurdon, J. B. and A. Colman. 1999. “The future of cloning,” Nature 402: 743–746. Hakeem, A., R. Sandoval, M. Jones, and J. Allman. 1996. “Brain and life span in primates,” in J. Birren (ed.), Handbook of the Psychology of Aging. New York: Academic Press, pp. 78–104. Hammer, M. and R. Foley. 1996. “Longevity, life history and allometry: How long did hominids live?” Human Evolution 11: 61–66. Hauser, P. M. and O. D. Duncan. 1959. The Study of Population: An Inventory and Appraisal. Chicago: University of Chicago Press. Hawkes, K., J. F. O’Connell, and N. G. Blurton Jones. 1997. “Hadza women’s time allocation, offspring provisioning, and the evolution of long postmenopausal life spans,” Current Anthropology 38: 551–577 (with commentary). Hawkes, K., J. F. O’Connell, N. G. Blurton Jones, H. Alvarez, and E. L. Charnov. 1998. “Grandmothering, menopause, and the evolution of human life histories,” Proceedings of the National Academy of Sciences 95: 1336–1339. Hayflick, L. 2000. “The future of ageing,” Nature 408: 267–269. Hill, K. and A. M. Hurtado. 1991. “The evolution of premature reproductive senescence and menopause in human females: An evaluation of the ‘Grandmother hypothesis,’” Human Nature 2: 313–350. Hill, K. and H. Kaplan. 1999. “Life history traits in humans: Theory and empirical studies,” Annual Review of Anthropology 28: 397–430. Hillman, J. 1999. The Force of Character. New York: Ballantine Books. Hölldobler, B. and E. O. Wilson. 1990. The Ants. Cambridge, MA: Harvard University Press. Holliday, R. 2001. “Human ageing and the origins of religion,” Biogerontology 2: 73–77.

Click to return to Table of Contents 434 L IFE SPAN EXTENSION IN HUMANS

Hoyert, D. L., K. D. Kochanek, and S. L. Murphy. 1999. “Deaths: Final data for 1997,” Atlanta: Centers for Disease Control and Prevention. Hrdy, S. B. 1999. Mother Nature: A History of Mothers, Infants, and Natural Selection. New York: Pantheon Books. Judge, D. S. and J. R. Carey. 2000. “Postreproductive life predicted by primate patterns,” Journal of Gerontology: Biological Sciences 55A: B201–B209. Kannisto, V., J. Lauritsen, A. R. Thatcher, and J. W. Vaupel. 1994. “Reductions in mortality at advanced ages: Several decades of evidence from 27 countries,” Population and De- velopment Review 20: 793–810. Kaplan, H. 1996. “A theory of fertility and parental investment in traditional and modern human societies,” Yearbook of Physical Anthropology 39: 91–135. Kaplan, H., K. Hill, J. Lancaster, and A. M. Hurtado. 2000. “A theory of human life history evolution: Diet, intelligence, and longevity,” Evolutionary Anthropology 9: 156–185. Keyfitz, N. 1985. Applied Mathematical Demography. 2nd ed. New York: Springer-Verlag. Kirkwood, T. B. L. 1992. “Comparative life spans of species: Why do species have the life spans they do?” American Journal of Clinical Nutrition 55: 1191S–1195S. ———. 1997. “The origins of human ageing,” Philosophical Transactions of the Royal Society of London, Series B 352: 1765–1772. Landes, D. S. 1998. The Wealth and Poverty of Nations: Why Some Are So Rich and Some So Poor. New York: W. W. Norton. Larsen, C. S. 1995. “Biological changes in human populations with agriculture,” Annual Review of Anthropology 24: 185–213. Lee, R. D. 1997. “Intergenerational relations and the elderly,” in K. W. Wachter and C. E. Finch (eds.), Between Zeus and the Salmon: The Biodemography of Longevity. Washington, DC: National Academy Press, pp. 213–233. Lovejoy, C. O. 1981. “The origin of man,” Science 2111: 341–350. Manton, K. G. and E. Stallard. 1996. “Longevity in the United States: Age and sex-specific evidence on life span limits from mortality patterns 1960–1990,” Journal of Gerontology 51A: B362–B375. Maynard Smith, J. M. and E. Szathmáry. 1997. The Major Transitions in Evolution. Oxford: Oxford University Press. McHenry, H. M. 1994. “Behavioral ecological implications of early hominid body size,” Jour- nal of Human Evolution 27: 77–87. Medawar, P. B. 1955. “The definition and measurement of senescence,” in G. E. W. Wolstenholme and M. P. Cameron (eds.), Ciba Foundation Colloquia on Ageing. Boston: Little, Brown. ———. 1957. The Uniqueness of the Individual. New York: Basic Books. Michod, R. E. 1999. “Individuality, immortality, and sex,” in L. Keller (ed.), Levels of Selec- tion in Evolution. Princeton: Princeton University Press, pp. 53–74. Montgomery, M. R. and B. Cohen (eds.). 1998. From Death to Birth: Mortality Decline and Reproductive Change. Washington, DC: National Academy Press. Moyle, P. B. and B. Herbold. 1987. “Life-history patterns and community structure in stream fishes of Western North America: Comparisons with Eastern North America and Eu- rope,” in W. J. Matthews and D. C. Heins (eds.), Community and Evolutionary Ecology of North American Stream Fishes. Norman: University of Oklahoma Press, pp. 25–34. Notestein, F. W. 1945. “Population—The long view,” in T. W. Schultz (ed.), Food for the World. Chicago: University of Chicago Press, pp. 36–57. Nutton, V. 1996. “The rise of medicine,” in R. Porter (ed.), The Cambridge Illustrated History of Medicine. Cambridge: Cambridge University Press, pp. 52–81. Olshansky, S. J., B. A. Carnes, and A. Desesquelles. 2001. “Prospects for human longevity,” Science 291: 1491–1492. Omran, A. R. 1971. “The epidemiologic transition: A theory of the epidemiology of population change,” Milbank Memorial Fund Quarterly 49: 509–538.

Click to return to Table of Contents J AMES R. CAREY / DEBRA S. JUDGE 435

Partridge, L. and P. H. Harvey. 1988. “The ecological context of life history evolution,” Sci- ence 241: 1449–1455. Petersen, W. 1986. Dictionary of Demography. Westport: Greenwood Press. Philipson, T. J. and G. S. Becker. 1998. “Old-age longevity and mortality-contingent claims,” Journal of Political Economy 106: 551–573. Preston, S. H. 1978. The Effects of Infant and Child Mortality on Fertility. New York: Academic Press. Preston, S. H. and M. R. Haines. 1991. Fatal Years: Child Mortality in Late Nineteenth-Century America. Princeton: Princeton University Press. Preston, S. H., N. Keyfitz, and R. Schoen. 1978. Causes of Death: Life Tables for National Popu- lations. New York: Seminar Press. Reznick, D. 1985. “Costs of reproduction: An evaluation of the empirical evidence,” Oikos 44: 257–267. Richard, M. 2001. “New centre to investigate global health threats,” British Medical Journal 322(7290): 813. Ricklefs, R. E. 1979. Ecology. 2nd ed. New York: Chiron Press. Robine, J.-M. and M. Allard. 1998. “The oldest human,” Science 279: 1834–1835. Roff, D. A. 1992. The Evolution of Life Histories: Theory and Analysis. New York: Chapman & Hall. Ross, J. A. (ed.). 1982. International Encyclopedia of Population. New York: Free Press. Ryder, N. B. 1959. “Fertility,” in P. M. Hauser and O. D. Duncan (eds.), The Study of Popula- tion: An Inventory and Appraisal. Chicago: University of Chicago Press, pp. 400–426. Sacher, G. A. 1978. “Longevity and aging in vertebrate evolution,” Bioscience 28: 497–501. Silver, L. 1997. Remaking Eden: Cloning and Beyond in a Brave New World. New York: Avon Books. Smith, A. 2000 [1776]. The Wealth of Nations. New York: Modern Library. Smith, D. W. E. 1993. Human Longevity. New York: Oxford University Press. Sohal, R. S. and R. Weindruch. 1996. “Oxidative stress, caloric restriction, and aging,” Sci- ence 273: 59–63. Southern, R. W. 1953. The Making of the Middle Ages. New Haven: Yale University Press. Stearns, S. C. 1992. The Evolution of Life Histories. Oxford: Oxford University Press. Szirmai, A. 1997. Economic and Social Development: Trends, Problems, Policies. New York: Prentice Hall. Thomas, L. 1977. “On the science and technology of medicine,” Daedalus 106: 35–46. Tuljapurkar, S., N. Li, and C. Boe. 2000. “A universal pattern of mortality decline in the G7 countries,” Nature 405: 789–792. Turner, J. S. 2000. The Extended Organism: The Physiology of Animal-built Structures. Cambridge, MA: Harvard University Press. van de Walle, F. 1986. “Infant mortality and the European demographic transition,” in A. J. Coale and S. C. Watkins (eds.), The Decline of Fertility in Europe. Princeton: Princeton University Press, pp. 201–233. Vaupel, J. W. et al. 1998. “Biodemographic trajectories of longevity,” Science 280: 855–860. Wake, M. H. 1998. “Amphibian reproduction, overview,” in E. Knobil and J. D. Neill (eds.), Encyclopedia of Reproduction. San Diego: Academic Press, pp. 161–166. Washburn, S. L. 1981. “Longevity in primates,” in J. L. McGaugh, S. B. Kiesler, and J. G. March (eds.), Aging—Biology and Behavior. New York: Academic Press, pp. 11–29. Weindruch, R. 1996. “Caloric restriction and aging,” Scientific American 274: 46–52. Williams, G. C. 1957. “Pleiotropy, natural selection, and the evolution of senescence,” Evo- lution 11: 398–411. Wilmoth, J. R., L. J. Deegan, H. Lundstrom, and S. Horiuchi. 2000. “Increase of maximum life-span in Sweden, 1861–1999,” Science 289: 2366–2368. Wilson, D. E. 1997. Bats in Question. Washington, DC: Smithsonian Institution Press. Wilson, E. O. 1971. The Insect Societies. Cambridge, MA: Harvard University Press.

Click to return to Table of Contents 436 L IFE SPAN EXTENSION IN HUMANS

———. 1977. Sociobiology: The New Synthesis. Cambridge, MA: Harvard University Press. ———. 1998. Consilience: The Unity of Knowledge. New York: Alfred A. Knopf. Wood, J. W. 1994. Dynamics of Human Reproduction: Biology, Biometry, Demography. New York: Aldine de Gruyter. Zvelebil, M. 1994. “Plant use in the Mesolithic and the transition to farming,” Proceedings of the Prehistoric Society 60: 35–74.

Click to return to Table of Contents First Politics, Then Culture: Accounting for Ethnic Differences in Demographic Behavior in Kenya

ALEXANDER A. WEINREB

ETHNIC DIFFERENCES IN demographic behavior are often noted in social research, especially research on less developed societies. Yet such differences are rarely explored analytically. Either they are completely ignored, usually by disguising them behind the opaque terms “district” or “region”—the favored approach in the National Research Council volumes on sub-Saharan Africa (e.g., National Research Council 1993; Brass and Jolly 1993). Or else they become grist for the dominant cultural explanatory paradigm, referred to as the “culture as identifier” model by Hammel (1990). That is, differences in unobserved tastes and values are inferred post hoc from observed differentials in the outcome measures (e.g., Bulatao and Lee 1983a: 15; 1983b: 260). In this article I offer an alternative explanatory model for ethnic differences that draws on new political economy, sociological theories of the state, as well as institutional approaches. Its aim is to ascertain to what extent access to political power and, through power, access to a state’s resources, is a key factor underlying ethnic differences. The reason an alternative explanatory model is needed is simply that, as Hammel (1990) notes, the culture as identifier model is flawed. It ignores the processual nature of ethnicity, in particular the interaction between, on the one hand, historically rooted and temporally specific political factors and, on the other, ethnic identity and behavior (Chazan et al. 1988; Nash 1989). The principal argument in this article derives from this recognition. It can be summarized as follows: to retreat to speculative culture as identifier explanations is only legitimate once we have eliminated the influence of political factors.1 This argument is particularly important in heterogeneous societies where political elites “employ ethnic security maps in the inner corridors of power” (Young 1994) since in these types of political cultures the distribution of state resources—including resources devoted to clinics, schools, labor opportunities, and other determinants of demographic

POPULATION AND DEVELOPMENT REVIEW 27(3):437–467 (SEPTEMBER 2001) 437

Click here to print article Click to return to Table of Contents 438 E THNIC DIFFERENCES IN DEMOGRAPHIC BEHAVIOR behavior—can be directed less at serving the general public interest than at the narrower goal of maintaining political support within the decisionmaker’s core constituency. Moreover, since governments in these types of states often remain in power for long periods, the compounded effect of these types of distributive patterns is to increase inequality. In short, ethnic differences in demographic outcomes can be literally manufactured, perhaps unwit- tingly, perhaps not, by the political actors charged with distributing scarce development resources. If we are to have a theoretically and contextually valid explanation of population trends, we need to incorporate indicators of political interaction since only thus can we examine the distribution of determinants of individuals’ demographic decisions. These themes are explored in detail in two main sections, the first theoretical, the second empirical. The theoretical section presents a partial theory of politics for demography. Its principal aim is to describe political systems in which ethnic-specific patron–client relationships mediate individuals’ access to those in power and, consequently, to a state’s resources. It highlights: (1) problems in existing conceptions of intra-state political competition in demography; (2) the multiple demands on, and competing interests of, political actors; (3) more specific to sub-Saharan Africa, the augmented distributive roles of political actors in African states; and (4) the special interaction between ethnicity and politics in sub-Saharan Africa. In the second section I relate these issues to demographic decisionmaking empirically, with reference to Kenya. First I introduce district-level estimates of “political capital,” an analytical indicator of access to government resources. I merge these estimates with Demographic and Health Sur- vey data for Kenya and explore their effect on models of contraceptive use using regression analysis. In particular, I explore their effect on residual ethnic differences.

A theory of politics for demography

Problems in existing conceptions of intra-state political competition in demography and in apolitical empirical models

Notwithstanding what Freedman (1987) refers to as the “longstanding sociological assumptions” that the general social environment—including political institutions—affects fertility and other demographic outcomes, notions of the state, government, or more general political constructs play a very limited role in explanations of demographic patterns. This conceptual weakness has been recognized, although, with the exception of McNicoll (1980, 2001), more commonly by non-demographers (e.g., Rouyer 1987; Greenhalgh 1995; Kertzer 1995).

Click to return to Table of Contents A LEXANDER A. WEINREB 439

This is not to say that population-related discussions have completely ignored the role of government and the state. The state’s ability to constrain individuals’ demographic choices either directly—for example, changing the range and cost of available health procedures and contraceptive choices—or indirectly—for example, through economic policies that affect labor market opportunities—is long-established. Elsewhere I have reviewed more overtly political discussions about population, including “conflict” and “institutional” approaches (Weinreb 2000a). In terms of assessing the impact of political competition on demographic behavior, however, each contribution to the existing literature exhibits at least one of two shortcomings. First, most of the literature on political effects is discursive rather than empirical, meaning that the more sophisticated theoretical propositions on “institutional effects” and “governance” (e.g., McNicoll 1980, 2001; McNicoll and Cain 1989; Hyden 1989) remain untested. Similarly, those contributions that include some level of testing have been limited to macro-level, mostly cross-country analyses (e.g., Kugler et al. 1983; Organski et al. 1984; Lapham and Mauldin 1984; Dahlgren 1994; Rouyer 1987). This is far from the micro-level arena that is the focus of most social demographic theory and research. Second, the literature tends to present a monolithic conception of the political environment, although this is less true of the general theoretical literature than of the empirical. Lapham and Mauldin’s (1984) indexes of political support for family planning, for example, ignore variation in support among different branches of government, or among different interests within the state. Most macro-theorists of the state, in contrast, describe a more pluralistic and competitive arena (e.g., foundational theorists like Marx and Weber, and contemporary theorists like Skocpol 1979 and Mann 1986). That is, the state is composed of multiple institutions; diverse interests compete to exercise an element of control over some or all of these institutions (especially government, the primary instrument of state control); and individual political actors engaged in these activities have both individual and corporate goals, which are liable to clash. I return to some of these themes below. The primary consequence of these shortcomings in the normal practice of demography is that standard models of demographic behavior are necessarily apolitical, since theoretical developments have not yet been translated into usable empirical applications. This has some undesirable and un- acknowledged analytic consequences. Specifically, by ignoring conflicts among the state’s existing interests, models in micro-level research inevitably make a couple of implicit, simplifying assumptions: (1) population-related policy is designed to be implemented uniformly across all sectors, groups, and areas within national boundaries—notwithstanding the intense political competition for control of political institutions by representatives of those same sectors and groups; and (2) population-related policy is in

Click to return to Table of Contents 440 E THNIC DIFFERENCES IN DEMOGRAPHIC BEHAVIOR fact implemented uniformly, notwithstanding institutionalized biases in administration. I show below that these assumptions defy both classical and modern social research on the nature of the political process. More relevant to the goals of this article, they also hinder the explanation of group-level differentials in behavior by not allowing for the possibility of structural bias in development spending. The most dynamic analytic notion of the state in empirical demographic literature exists, ironically, in one of the least-theorized areas, the family planning/health program evaluation literature. By “dynamic notion of the state” I refer to the following. Evaluations of program effects typically control for the endogeneity of program placement. Implicit in this is the idea that government, the institution charged with establishing clinics and programs, is a responsive institution. In other words, the scarcity of resources means that services can only be established in a limited number of places. Therefore decisions need to be made about where to direct services. As a result, some bias is inherent in development spending. The attempt to control for the endogeneity of program placement is an analytic response to this economic-political problem. It assumes that, ceteris paribus, political decisionmakers are more likely to place services in communities that display the strongest demand for those services. Therefore community-level differences in the availability and quality of services are, at least in part, a result of differences in the strength of communities’ demand for those services. To the extent that such approaches recognize that causal inference about program effects is sensitive to the embedded bias in the pattern of development spending, attempts to deal with the endogeneity of program placement are laudable. But because they ignore underlying relational characteristics, their implicit representation of a state’s government as a responsive and politically disinterested institution is antithetical to all but the simplest representations of the policymaking process. This is because distributive decisions that we associate with government are made by individual political actors whose authority to make the decision is legitimated in state law. But whose interests do these corporate actors serve? Clearly there can be no fixed answer to this question across all societies. But what is certain is the basic axiom of sociological and new political economic theories of the state: political actors have their own individual, class, and other status-group interests, and social change on both a small and a large scale reflects tensions between these and other organized interests in the state.

The multiple roles and goals of political actors

The fundamental assertion of the “new political economy” is that classical economic approaches to government are flawed insofar as they assume that “government is composed of Platonic guardians and that the state acts be- nevolently in seeking the public interest” (Meier 1991: 6). Rather, the new

Click to return to Table of Contents A LEXANDER A. WEINREB 441 political economy asserts, political decisions are determined by factors other than the objective merits of the proposed policy. In particular, they are determined by private and potentially “perverse” political interests (Birdsall and James 1993; Mbaku 1999), whether “society-centered” or “state-centered” in approach (for a review, see Grindle and Thomas 1991). These alternative interests can lead to greater inefficiency and inequity in the use of a state’s resources (Bates 1989; Buchanan 1991; Grindle and Thomas 1991; Meier 1991; Birdsall and James 1993; Hoogvelt 1997; Kimenyi and Mbaku 1999; Mbaku 1999; Rowley 1999). The assumption that political decisions are determined by things other than the objective merits of the proposed policy is an old theme in mainstream sociological and political theory. I have already mentioned the por- trayal of the state as an arena of competing interests by the foundational social theorists. More recently, attempts have been made to refine these models by, for example, combining general features of the Marxist and Weberian frameworks. In most such models the state plays an independent role in the economy, essentially paralleling new political economy assumptions. O’Connor (1973), for example, distinguishes three economic sectors that constitute a modern capitalist society: the “monopoly” sector, the “competitive” sector, and the “state” sector. Class conflict occurs between these three groups, not between capitalists and workers, because “the three sectors…have distinct economic interests to defend” (Collins 1988: 111). Skocpol (1979) makes a similar point about the competing interests of the “state class,” the “property owning class” (indistinguishable from the “state class” in certain societies), and the producing class. Equivalent examples are presented in Mann (1986) and Collins (1988). Theoretical innovations in micro-sociology allow us to link the fundamental observation of these accounts—the diversity of political interests that constitute the state—with more behaviorally oriented social science. Chief among these innovations is Coleman’s (1990) distinction between “corporate” and “natural” actors. The former represent at least two sets of interests, their own and those of a corporate body for whom they act as an agent and over whose resources they have partial control. The latter are individual social actors. Legislative representatives are, by their very nature, corporate actors since in their corporate roles they are the agents of: (1) a given constituency in the legislature; (2) a given party; and (3) the state in the wider polity. In addition, they may also be (4) the agents of the government if, in a parliamentary system, they are coopted from the legislature into the cabinet. And (5) they necessarily also have individual interests. The question, as posed above, is: whose interests do they serve? This is an important question for demographers since the answer may signal the main beneficiary of a given political actor’s decisions. A reasonable answer, I think, is represented by a simple “accountability hypothesis.” This is con-

Click to return to Table of Contents 442 E THNIC DIFFERENCES IN DEMOGRAPHIC BEHAVIOR sistent with the skepticism of new political economy and sociological approaches to the state and can be defined as follows: political actors seek to maintain their positions, hence their decisions favor the expected preferences of their most important patrons. In democratic electoral systems, this suggests that political actors need to play a “distributive role” to maintain their position, that is, maximize the amount of government funds they direct to their constituents (although, unless they are independents, these demands may at times conflict with those of their party). In more authori- tarian systems, by contrast, a political actor’s position may be more dependent on the whim of the ruler. Fealty to the ruler in such cases is, therefore, probably a more important determinant of political actors’ behavior than fear of general electoral accountability.2 Empirical evidence bolsters these hypotheses. In the United States, for example, the longevity of congressional representatives’ careers is partly a function of the extent to which they can direct funds to their home constituencies. Specifically, each additional $100 per capita federal spending allocated to congressional districts, net of federal spending in other districts in the same state, is worth up to 2 percent of the popular vote for those districts’ incumbent representatives in a subsequent congressional election (Levitt and Snyder 1997). Similarly, states represented by senior Democratic congressmen in the 1953–90 period—for almost the whole period, congressional committee chairpersons were Democrats—have enjoyed faster economic growth in that period than states with more junior congressional representation (Levitt and Poterba 1999). Barkan (1979) identifies the same pattern in Kenya. Although Kenya was a de facto one-party state in the 1970s, he shows a positive correlation between an elected representative’s success in bringing development funds to his home constituency and his longevity in elected office. In other words, underperforming representatives were replaced in within-party interfactional competition for legislative seats (rather than being rejected from their parliamentary seats, as they might have been in open, multiparty competition).

How the structural characteristics of African states augment political actors’ roles

In typical sub-Saharan African states, politicians appear much more likely to act as state-sponsored patrons—that is, to have a powerful distributive role—than is the norm in Euro-American liberal democracies. According to Dietz and Houtkamp (1998: 101), this expansive role is a result of the structural characteristics of African economies: “In the virtual absence of private capital stock and of indigenous private capital tradition, governments in independent Africa have long seen themselves as the engines of growth, the design and implementation machinery for development.” Indeed, notwith-

Click to return to Table of Contents A LEXANDER A. WEINREB 443 standing assertions that African states tend to be administratively ineffective (Callaghan 1987), they are also characterized as having “a virtual monopoly in development” (Hyden 1989: 193). Not only are African states the primary providers of basic infrastructure such as schools, health care, and roads, but, equally important in an environment in which employment opportunities are extremely limited, they are also the largest formal-sector employers. This refers not only to employment opportunities in the civil service, but also to opportunities in parastatal enterprises. Throughout the 1970s and 1980s, Kenya was often cited as operating a relatively free market in comparison to most other sub- Saharan states. The Kenyan government’s monopoly on development was therefore less pronounced than in other countries (e.g., Hyden 1979; Khadiagala and Schatzberg 1987; Bates 1989). Yet even here, according to the United Nations Centre on Transnational Corporations (1989: 1), the 177 “organizations with some level of equity participation by government or associated institutions and enterprises...constitute about one third of the Kenyan economy in terms of employment and wages paid.” This is in addition to direct government employment through the civil service and military forces. The key theoretical point, then, is that the dearth of economic opportunity (and political accountability) outside the state apparatus makes members of African governments much more powerful, in a relative sense, than their Euro-American counterparts. Moreover, public figures at all levels are much more open about this “patron” role and about the extent to which political support translates into government expenditure. This is one of the first lessons readers can draw from African newspapers. When politicians— both incumbents and opposition leaders—campaign or give speeches in un- developed areas, they openly blame the lack of infrastructure in those areas on the limited local support for the governing party or coalition. Incum- bents typically pledge to increase the share of the state’s resources devoted to a particular constituency if its residents support the government in a subsequent election. This can be done without legislating increased funds simply by creating a new administrative district in their new supporters’ area, whose new district capital qualifies for a district hospital, increased bus services, and so on. And incumbents threaten to continue to withhold development spending if the level of support does not increase. Similarly, opposition leaders cite the lack of local development spending and hence infrastructure as an example of the current leaders’ lack of interest in the area’s residents and argue that it shows the leaders should be replaced.3 A second symptom of this augmented and more explicit political patron role is the apparent popular expectation that politicians use their positions to enrich themselves. There have been few formal studies of this process—most likely because of the difficulty of collecting data—but it is one of

Click to return to Table of Contents 444 E THNIC DIFFERENCES IN DEMOGRAPHIC BEHAVIOR the fundamental popular beliefs about political life, as seen in the array of journalistic and fictional portrayals of such interactions.4 Yet the real significance of this pattern of enrichment is that “informal taxation” systems—Harden (1991) calls the extreme type a “kleptocracy”— work at all levels of state organization. The state is a vehicle for upward social mobility insofar as it provides opportunities for any given actor to impose his own informal tax on potential clients. Toward the top of the hierarchy, employment as a legislator is, in this context, an opportunity to amass wealth. Thus, political leaders become what Ngugi wa Thiong’o (1981: 131), in the Kenyan context, calls “brothers-in-the-eating of plunder.” They constitute a socioeconomic elite that combines characteristics of Skocpol’s (1979) state class and property-owning class. From the electorate’s perspective, the popular indifference to politicians’ enrichment is not unconditional. That is, although there are com- plaints about politicians (and civil servants) who use their positional authority to line their own pockets, the enrichment appears not to constitute political misconduct unless it omits all gain for the politician’s own clients. This is a simple extension of the accountability hypothesis: it is the corporate duty of a political patron to direct wealth to his clients. Indeed, implicit in Barkan’s (1979) findings that legislative representatives’ careers are enhanced by successfully bringing home development resources is the notion that community leaders often try to find political representatives who they think can fill this role effectively. This underlies Holmquist’s (1979: 136) assertion that “Communities searched for patrons and offered themselves as clients.” A fictional representation of this process is the starting point for Namale’s (1994) novel about a Kenyan election, transparently titled Honourable Criminals. In short, it is generally acceptable for a legislative representative to use his position to pursue narrow self-interest. But in exchange for the opportunity for personal enrichment, representatives are expected to direct some of the state’s resources to their area. Their clients expect to see development funds.

The role of ethnicity

Thus far, this rendering of the nature of politics in sub-Saharan Africa is a simple story about the accumulation of resources. Applying it to demographic trends might allow us to explain some class-based differentials, but the standard socioeconomic indicators of those class differences—such measures as educational status and familial income—would probably explain any variance. In the sub-Saharan African context in general, however, accumulation of resources is only secondarily a class-based phenomenon. Ethnicity, often referred to as “tribalism” in this African political guise, is more impor-

Click to return to Table of Contents A LEXANDER A. WEINREB 445 tant. For many reasons—among them continued poverty and scarcity of development resources, manipulative and hate-mongering leadership, histories of ethnic domination and brutality, the attractions of group identity— ethnicity remains the primary status-group affiliation for many people in most areas of sub-Saharan Africa (Bangura 1994; Berkeley 2001). This is in defiance of the hopes and expectations of many independence-era African intellectuals (e.g., Achebe 1983).5 The political consequences of this flourishing of ethnicity, and part of the reason for ethnicity’s continued salience, are twofold. Moreover, each of the two reinforces the other: first, an ethnic constituency remains a necessary springboard for participation in national politics; and second, ethnic constituencies are used to validate the informal networks and patron–client relationships through which the state’s resources are distributed. Taken together, these two characteristics sustain the perpetual crucial issue in Afri- can politics: ethnic competition for state power and resources (Holmquist 1979; Hyden 1979; Shaw 1986; Khadiagala and Schatzberg 1987; Kitching 1982; Nyangira 1987; Throup 1987; Chazan et al. 1988; Bates 1987, 1989; Bradshaw 1990; Stamp 1991; Schraeder 1993; Bangura 1994; Himbara 1994; Holmquist and Ford 1994; Young 1994; Gordon 1995). This is a matter of public record. At its most extreme, ethnic competition for state power has led to recurrent genocide, as seen in Rwanda and Burundi since the 1960s, though most seriously in the 1990s (e.g., Gourevitch 1998); to large-scale ethnic conflicts such as the those that resulted in Igbo secession from Nigeria in Biafra (during which more than 2 million people are said to have died); and in southern Sudan to rebellions against the Mus- lim north after the imposition of sharia. Ethnic competition has also led to multiple systematic massacres, events that now constitute “ethnic cleans- ing.” These include massacres of the Acholi and Lango under Amin in Uganda, and of the inhabitants of Amin’s home region in northwestern Uganda under Obote; of Frelimo-affiliated groups by Rhodesian and South African supporters of Renamo; of Ndebele by the Shona-dominated government in Zimbabwe; of African National Congress supporters by Inkatha in South Africa’s Natal Province and vice versa; of Gio and Mano by Sam- uel Doe’s ethnic Krahn militias in Liberia, and subsequent revenge attacks on Krahn (and anyone who looked wealthy) by Charles Taylor’s Gio militias; and more recent (and ongoing) conflicts in Sierra Leone, the Demo- cratic Republic of Congo, and Sudan. In most of these conflicts, victims are said to have numbered in the tens of thousands, and displaced persons in the hundreds of thousands. Ethnic competition has also led to short-term ethnic rebellions and insurgencies. These include those of the Tuareg in Mali and Niger, and coups—often repeated, some successful, others failures—against ethnic oli- garchies in Cameroon (against Biya’s southern Christian coalition), Chad

Click to return to Table of Contents 446 E THNIC DIFFERENCES IN DEMOGRAPHIC BEHAVIOR

(against southerners by Muslim “gorane”), Comoros (repeated coup attempts finally ending in secession by Anjouan and Moheli), Kenya (against Moi in 1982), and Madagascar (by mixed-heritage Cotiers against the Malayo-In- donesian Merina elite). We can also include tensions between smaller client groups in this category: for example, about 5,000 people were killed in 1992 conflicts between Tiv and Jukun in Nigeria’s Middle Belt; a few thousand were killed in Kenya’s Rift Valley prior to the 1992 multiparty elections. Even where ethnic competition for control of the state does not result in significant ethnic clashes, the competition itself is an open and acknowledged feature of the political landscape. For example, in sub-Saharan Afri- can states with competitive party systems, political parties have often been associated with ethnic blocs, allowing the largest such bloc to control the political process. Examples are Shona-associated ZANU’s domination in Zim- babwe (until the most recent election) and domination by the Hutu-associated National Revolutionary Movement for Development during the 1980s and early 1990s in Rwanda. At various times in each of these states, ethnic majorities have taken power in elections or referendums and have maintained that power through elections and referendums. But ethnic competition also occurs in multiparty systems with multiple ethnic blocs, even in states that have been traditionally considered relatively stable in terms of ethnic conflict. The 1992 elections in Kenya, for example, resulted in KANU dominance in Kalenjin-dominated Rift Valley Province (helped by the dis- placement of non-Kalenjin in the months prior to the election), Ford-Asili dominance in Kikuyu-dominated Central Province, and Ford-Kenya dominance in Luo areas of Nyanza Province. In each of these provinces, virtually every elected Member of Parliament was from the dominant local party. KANU retained a plurality because it also did better among smaller ethnic blocs in other regions (I return to the Kenyan case in the following section). These general distributions were repeated in 1997. Similarly, the 1999 elections in Malawi resulted in victory for Bakili Muluzi’s UDF, largely because he was from the more populous Yao-dominated Southern Region, while the other two parties, AFORD and MCP, which respectively dominated the voting among their Tumbuka and Chewa-led blocs in Northern and Central Region, were unable to present a unified front. In both Kenya and Malawi rioting took place in areas associated with losing candidates, and there were multiple allegations of voter fraud. In short, the absolute centrality of ethnicity in African political life places “tribalistic” politics—ethnic competition for control of state resources—at the center of any analysis of demographic trends. The two-way political relationship enshrined in the notion of ethnicity—the upward flow of political support and the downward flow of resources within the loose ethnic- specific network (Barkan 1979; Smoke 1993)—means that the severity of communities’ needs has less effect on the distribution of development goods

Click to return to Table of Contents A LEXANDER A. WEINREB 447 than communities’ access to empowered political decisionmakers. But this is itself largely a function of ethnicity, since to provide funds to his constituents an elected representative must satisfy two necessary conditions: first, he needs to be elected, unlikely if he belongs to a different ethnic group than his constituents; and second, he needs to find favor with the executive, also a function of ethnicity since the executive tends to be dominated by a particular ethnic bloc.

Summarizing the arguments

One of the principal reasons that people want to acquire political power, in sub-Saharan Africa and elsewhere, is that the state is the most important avenue to upward social mobility. The means that they use to achieve power is to construct, then use, a range of affective ties, with patron–client relations at their core. In sub-Saharan Africa these affective ties are limited, in large part, to members of the same ethnic group or part of the same ethnic coalition. Once in power, the state’s resources—including employment opportunities—are used to maintain one’s clients (as well as, presumably, to try to recruit new members). Kimenyi characterizes this structure as inherently conflictual. He writes that “ethnic groups are trapped in a Pareto-inferior ‘steal-steal’ strategy” (1999: 360). That is, leading politicians in sub-Saharan Africa have important distributive roles associated with their position in the state hierarchy. They determine overall and emergent patterns of social stratification by regu- lating access to an array of government-funded institutions such as clinics and other public health infrastructure, schools, and employment opportunities. But these are precisely the institutions that also shape individuals’ demographic decisions, allowing us to draw a conceptual link between the micro-level context within which demographic preferences and decisions are made and the broader political struggle between ethnic/political groups. Moreover, to the extent that the ethnic-specific networks of political support and politically motivated distribution of resources are stable over a long period, we can expect to see a concentration of development goods and of “the nation’s wealth in the hands of a few individuals and groups, further contributing to the severe inequalities in income distribution” (Mbaku 1999: 327–328). To the extent that contemporary demographic analysis and theory ignore such issues, and therefore limit the interpretive range of notions of “government” and the “state,” they are flawed. Government decisions about how to distribute development goods, many of which shape the risk environment in which demographic decisions are made, are an outcome of internal political rivalries within the state. These rivalries are, in turn, affected less by the actual strength of local demand for development goods than by

Click to return to Table of Contents 448 E THNIC DIFFERENCES IN DEMOGRAPHIC BEHAVIOR the access-mediated strength of local demand. In the sub-Saharan African context, “access” refers to one’s specific patron–client relationship, which is largely a function of ethnic identity.6 In sum, then, my argument is based on two simple assertions. First, the accountability principle makes it a general rule that political actors distribute a disproportionate share of the state’s resources to their constituents if they wish to remain in power. Second, in the African context the effects of this stance are compounded by the multiethnic structure of societies, meaning that ethnic differences in demographic outcomes function both as indicators of past political competition for state power and as more singular indicators of different cultural-normative preferences. Traditional demographic analysis has all but ignored this political component. Instead, it has tended toward the cultural explanation. I now show how this propensity weakens sociodemographic analysis.

Empirical tests

In the empirical tests that follow, I explore ethnic differences in the current use of modern contraception by women in Kenya. There are four reasons for the focus on contraceptive use. The first is that it is a variable of interest in relation to the demography of sub-Saharan Africa. The speedy reductions in fertility throughout East and Southern Af- rica—in Kenya, total period fertility rates have fallen from 8.1 births per woman in 1977 to 4.7 in 2000 (Population Reference Bureau 2000)—largely reflect women’s contraceptive behavior. Second, it is a common focus of social demographic analysis, making it easy to employ standard regression models. Third, unlike child mortality, another contender for the dependent variable role, contraceptive-use decisions are fully endogenous. That is, they are not affected by environmental-ecological factors—for example, differential disease burden—that complicate attempts to explain ethnic or regional variation in child mortality. Finally, contraceptive use refers to current behavior, not to an event that may have occurred either yesterday or ten years beforehand. While many macro-level causes are, as implied above, embedded in past political events, it is easier to identify those causal links by focusing on current decisions. Figure 1 presents one representation, net of feedback loops, of how ethnicity might affect an individual’s decision to use contraception. Eth- nicity affects contraceptive use through both political and cultural mechanisms. The political mechanism—referred to above as a manifestation of “tribalistic” politics—affects the distribution of supply-side determinants of contraceptive use such as the availability and quality of family planning clinics. It also affects the distribution of demand-side covariates, at least those that are themselves a function of supply-side factors. The level of educa-

Click to return to Table of Contents A LEXANDER A. WEINREB 449

FIGURE 1 Schematic representation of political and cultural effects of ethnicity on contraceptive use

Ethnicity Macro-level Micro-level Access to political power

Supply-side Normative characteristics: constraints Availability Quality Socioeconomic status: Education Employment Children’s survival

Contraceptive use tion, for example, is an outcome of past access to education, common measures of which are the supply and quality of schools and teachers. Employ- ment opportunities may also be an outcome of supply-side factors, especially where the civil service or parastatal organizations are among the main employers. There are a number of reasons for choosing Kenya as the exclusive focus of the empirical analysis. The first is related to its reputation as “poster child” of demographic change in Africa, especially with respect to fertility. For example, notwithstanding a number of skeptical forecasts throughout the 1980s, in particular by Frank and McNicoll (1987), Kenyan contraceptive prevalence has increased dramatically over the last decade. According to the 1993 DHS, for example, 30.6 percent of currently married, nonpreg- nant women in rural areas were using modern contraception at the time of the survey (my calculation). This high average provides appropriately high levels of variance, allowing for considerable analytic leverage given our choice of contraceptive use as the main dependent variable. The next two reasons are related to the country’s more general characteristics. Kenya is structurally a modal African state in that it has multiple ethnic groups and large ethnic and regional variations in development characteristics. It is also typical insofar as “tribalism” has been a constant in its party and factional politics. Ethnicity is one of the most important determinants of access to political power and to the subsequent downward flow of patronage (Barkan 1979, 1992; Kitching 1982; Nyangira 1987; Holmquist and Ford 1994).

Click to return to Table of Contents 450 E THNIC DIFFERENCES IN DEMOGRAPHIC BEHAVIOR

Finally, there are two uniquely Kenyan reasons. The first is that Kenya’s relative political stability has made it one of the most widely researched and data-rich countries in sub-Saharan Africa. As I show below, this has allowed me to collect “full” data on political representation. The second is that Kenya is one of only a few countries in sub-Saharan Africa in which two ethnic coalitions have taken turns dominating the state apparatus, and in which the transition from one to the other was peaceful. In this context a brief sketch of Kenya’s political history would be informative. Since before independence from Britain in 1963, Kenyan politics has been dominated by tensions between three main ethnic blocs: the Kikuyu-led alliance of Central and Eastern Province, the Luo-led alliance of Nyanza Province, and the Kalenjin-led alliance of Rift Valley Province. The Kikuyu, Luo, and Kalenjin are, in descending order, the three largest ethnic groups in Kenya (though according to 1989 census figures, Kalenjin may now outnumber Luo).7 In the preindependence era, both the Kikuyu, members of whom were at the forefront of the Mau Mau rebellion in the 1950s, and the Luo, who were also disproportionately represented among the leaders of the independence movement, favored a unitary constitution. The Kalenjin and other smaller groups, in contrast, favored a federalist constitution (Majimboism) since they feared domination by the Kikuyu–Luo majority. Upon independence, a unitary constitution was adopted, and Jomo Kenyatta became president, leading a Kikuyu-dominated coalition whose other client groups included related ethnic-linguistic groups such as the Meru, the Embu, and, to a lesser extent, the Kamba, but also increasingly concentrating power among ministers from his home district of Kiambu. Wary of competition from Luo leaders, Kenyatta replaced his Luo vice president, Oginga Odinga, with Daniel Arap Moi, leader of the Kalenjin bloc who had previously been associated with the opposition KADU (Majimbo) bloc. When Kenyatta died in 1978, Moi inherited the office. He discarded Kenyatta’s old Kiambu–Kikuyu elite and in their place, especially after an attempted coup in 1982, forged alliances with smaller ethnic groups with whom he had originally been allied as part of the Majimbo movement. These included other groups in Rift Valley such as the Maasai, as well as various groups in Coastal Province and the Luhya-speaking groups in Western Prov- ince (Bates 1989; Barkan 1992). Moi has successfully remained in power through the end of one-party rule. Under his leadership, and with help from an opposition divided along ethnic lines (primarily Kikuyu versus Luo), the ruling KANU party won a plurality of votes in multiparty elections in 1992 and 1997, albeit with the regional concentrations of votes described earlier and some level of voting fraud (especially in 1992: Holmquist and Ford 1994). These key events are worth bearing in mind because they lend themselves to a slightly more complicated analytic picture. Specifically, they suggest that from the time of the Kenyatta-to-Moi transition in the late 1970s,

Click to return to Table of Contents A LEXANDER A. WEINREB 451 disproportionate levels of government spending and services should have moved from areas associated with the Kikuyu-led coalition to areas associated with the Kalenjin-led coalition. The events also suggest that there should be little effect of this ethnic transition in the Luo areas, since although Luos were at times represented in both the Kenyatta and Moi administrations, they tended to be underrepresented, and as a bloc were associated with the opposition. I describe some supporting evidence for these expectations below.

Measuring access to political power

Since its independence Kenya has had a Parliamentary system with a uni- cameral legislature. Thus, Presidents Kenyatta and Moi have also been sit- ting Members of Parliament (MPs) representing a single constituency. So, too, have been most of their cabinet members. Consistent with the discussion of the nature of politics in sub-Saharan Africa, and with the structure of Kenya’s Parliamentary system and unitary state, it is possible to think of a measure of an area’s political power as being equivalent to the seniority associated with its political representative in the central government. The higher on the national political ladder an area’s representative has climbed, the greater the amount of state resources he should be able to direct to his home area. With this principle in mind I collected a full history of political representation in Kenya. That is, in any year from independence until 1996 I have a measure of the highest-level governmental position filled by a particular constituency’s Parliamentary representative. Out of these data I constructed an area-specific political score consistent with the top tier in Barkan’s depiction of the “patron–client linkage structure” (1979: 73) in Kenya. These are presented in Figure 2. The left-hand panel depicts the patron–client structure. The chief patron is the President, also a Parliamentary representative. Under him are the chief ethnic/regional leaders, who serve as cabinet ministers. The lowest members in this top tier are the emerging ethnic/regional leaders, who serve as assistant ministers. Under them, not shown, are the constituency patrons—the backbench members of Parliament—other constituency leaders, clan leaders, and finally the public. The right-hand panel shows the constituency-specific political score into which I translated each of these positions for each year of observation, yielding a simple measure of political points per person-year in office. In the cross-section this score should be interpreted as the level of seniority in government, an indicator of the political actor’s access to state resources. As a cumulative score it incorporates a measure of the duration of a politician’s access. Although the range of scores that I assign, from 1 to 5 per person- year, is arbitrary, the specific values—positively associated with the seniority of the member—fit with the overall framework, since the greater a po-

Click to return to Table of Contents 452 E THNIC DIFFERENCES IN DEMOGRAPHIC BEHAVIOR

FIGURE 2 Construction of the measure of an ethnic-specific political score in Kenya (right panel), patterned on Barkan’s patron–client linkage structure (left panel)

Top tier in Barkan’s “patron–client Political score linkage” Position Points per year in office Chief patron (president) President 5

Ethnic/regional leaders (cabinet ministers) Vice president 4 Cabinet ministers 3

Future ethnic/regional leaders (assistant ministers)

Assistant ministers 2 Downward flow of state resources Parliamentary secretaries 1

Upward flow of political support

litical actor’s seniority, the greater his access to resources, including those of the government. These political scores were designed to be merged with data from the Kenya Demographic and Health Survey (KDHS), discussed below. Because the latter do not identify respondents’ residence at agglomerations below the district level, I aggregated the constituency-level political scores into a single district-level measure and then assigned the scores as a global characteristic—that is, as a characteristic that measures “the context directly, instead of using aggregates of variables that are measured over individuals” (Kreft and de Leeuw 1998: 9)—to all residents in that district. As a property of each individual in a given district, these political scores are interpreted as indicators of a type of social capital that can be called “political capital.” I discuss this in more detail elsewhere (Weinreb 2001). In particular, I show that the notion of political capital is conceptually sound insofar as it contains the three key characteristics that Portes (1998) identifies as the essence of social capital.8 I also argue that it has a high level of construct validity. First, ethnic variations in political capital follow expected paths—Kikuyu and Luo dominating in early postindependence years, Kalenjin by the 1980s. Second, when merged with KDHS data, political capital is associated with various socioeconomic and health characteristics in

Click to return to Table of Contents A LEXANDER A. WEINREB 453 ways that are consistent with the ethnic-political story of distribution described in the theoretical section above. For example, residents in areas with lower levels of political capital tend to have lower educational status, fewer economic opportunities, and reduced access to birth assistance by trained medical personnel. In addition, they are less likely to have vaccinated their children and to have access to either clean piped water or electricity.9

Data

The primary data are from the 1989 and 1993 rounds of the KDHS. Exclud- ing seven northern districts that contain what is estimated to be less than 4 percent of the population, these are nationally representative samples of women aged 15–49 years, representing all major ethnic groups (NCPD 1994: 5). I eliminated all urban women from the 1993 data, and residents of the three largest and most ethnically heterogeneous urban areas in the 1989 data (i.e., Nairobi, Mombasa, and Nakuru), on the grounds that the absence of constituency-level data makes it impossible to disaggregate these ethnically heterogeneous areas into smaller, ethnically homogeneous ones. Dis- trict-level measures of political capital would be difficult to interpret in such cases since they would be unlikely to capture local differences in access to development goods. Such goods may be widely available in a neighboring area, for example, even if they were primarily intended to privilege a rival political-ethnic group. I also eliminated women who reported themselves as currently preg- nant, on the grounds that such women are not at risk of pregnancy so the issue of choosing whether to use contraception does not arise. Together, these two omissions yielded a total sample of 5,425 women in 30 districts in 1989 and 5,370 women in 15 districts in 1993, equivalent to 76 percent and 75 percent, respectively, of the original samples.10

The models and results

I now examine the primary empirical question of the article: do the measures of political capital account for the ethnic differences in contraceptive use and, if so, how much of the ethnic difference? Tables 1 and 2 present results from different types of models, using the same methodological approach. Both present robust maximum-likelihood estimates, that is, with added controls for the nonindependence of observations at the cluster level. The aim of the first three models is to explore simple associations between the political capital variables and the other covariates of contraceptive use. Model 1 presents a standard model with controls for demographic and socioeconomic characteristics and for the availability of contraception, and dummy variables for those ethnic groups whose

Click to return to Table of Contents 454 E THNIC DIFFERENCES IN DEMOGRAPHIC BEHAVIOR differences with respect to contraceptive use are not explained by the other variables in the model and can therefore be attributed to some notion of culture, in keeping with the “culture as identifier” paradigm, described above.11 Model 2 uses the same control variables but substitutes the measures of political capital for the ethnicity variables. Model 3 includes all variables— the basic controls, the ethnicity dummies, and the measures of political capital. Model 4 is the full model, intended to deal with the causal inadequacy of model 3. Specifically, both the theory described in the opening section and the empirical tests described in relation to the validity of the political construct (and reported in detail in Weinreb 2001) describe how some of the covariates of contraceptive use in model 3 are endogenous with respect to the political process. They are the result of political actors’ decisions. For example, women who grew up in areas with low levels of political capital during the 1960s, and who were of school age at that time, were less likely to go to school. They also had fewer current economic opportunities. Both of these circumstances affect fertility decisions. Model 4 addresses this problem by cleaning a set of these covariates of the effects of preceding political action. Thus, in model 4 both the socioeconomic measures—years of education, current work status, and household wealth—and the indicator of the availability of family planning are the predicted residuals of preliminary equations in which each was regressed on measures of political capital.12 Model 4 should therefore be treated as a more robust test of associations between political capital and contraceptive use.13 As described above, model 1 identifies the standard covariates of contraceptive use. In both the 1989 and 1993 data, current female users of contraception are likely to be older, have more surviving children, be more educated, be currently employed, live in wealthier households, and live closer to a family planning provider. Net of these basic associations, significant “unexplained” ethnic differences are picked up by dummy coefficients on ethnic identity. In both 1989 and 1993, the Kikuyu, other ethnic groups associated with them during the Kenyatta years (the Kamba, Meru, and Embu), and the Kisii had significantly higher use of contraception than the omitted groups—the Kalenjin, Luo, Luhya, Mijikenda, and Taita. There are only two substantive differences between the two surveys in terms of these general results. First, in the 1993 data there is no significant net difference between the contraceptive use of married and nonmarried women. This is likely indicative of the increased use of contraception by unmarried women in the four years between surveys. Second, the control for parity—that is, net of the number of surviving children—is more weakly associated with contraceptive use in 1993 than in 1989. Again, this may reflect the increasing use of contraception across different subpopulations, in this case among low-parity women. The general effect of the measures of political capital on nonethnic covariates in the model is minimal with respect to both the 1989 and 1993

Click to return to Table of Contents A LEXANDER A. WEINREB 455 data. None of the point estimates on respondents’ demographic and socioeconomic characteristics, and on the availability of contraception, changes significantly between models 1, 2, and 3. Nor do the significance levels. This implies that the measures of political capital are truly exogenous because they are not capturing the respondents’ basic sociodemographic characteristics. In contrast, significant changes in the estimated ethnicity parameters occur when measures of political capital are introduced into the same model (model 3). This allows us to answer the primary research questions with a qualified “yes,” that is, measures of political capital partly explain the ethnic differences in contraceptive use. Specifically, model 3 in Table 1 shows that political capital appears to explain the Meru-Embu difference in the 1989 data. The estimated logit on the Meru-Embu drops from 1.48 to 0.16 when we add the measure of political capital. The significance of the Kisii difference falls from the 1 percent to the 10 percent level. The effects are smaller in the analysis of the 1993 data (Table 2). The measures of political capital explain the Kisii difference—the point estimate falls from 0.61 to 0.44; they reduce the Kamba and Kikuyu difference by roughly 20 percent; and they double the size of the standard errors for these three groups. The more robust test of these associations presented in model 4 shows that model 3 underestimates the overall effects of political capital on the ethnic differences. Model 4 results imply that a large part of the ethnic differences in contraceptive use is caused by historical differences in political capital, which operate through current differences in educational and employment opportunities, in household wealth, and in differential access to family planning services. In the 1989 data, for example, controlling for the endogeneity of years of education, current work status, household wealth, and availability of family planning—all of which remain stable and highly significant—eliminates the Kamba and Kisii difference in contraceptive use. This leaves the Kikuyu as the sole ethnic group with a significantly different—in this case, higher—likelihood of using contraception once the effects of political capital are taken into account. But even with the Kikuyu, the point estimate in model 4 is half the size of that in model 3, and only 2.4 rather than 6.4 standard errors away from zero. Slightly different results can be seen in relation to the 1993 data. First, controlling for the endogeneity of respondents’ socioeconomic characteristics completely undermines the relationship between current work status and contraceptive use. The point estimate falls from 0.43 in model 3 to –0.004 in model 4. At the same time, the significance of the Kamba difference falls from the 5 percent to the 10 percent level, and the standard error around the Kisii difference increases still further. In the most general terms, therefore, the results presented in Tables 1 and 2 point to a positive answer to our primary research question. Ethnic

Click to return to Table of Contents TABLE 1 Regression coefficients showing the effects of district-level political power and other covariates on women’s current use of modern contraception in rural Kenya (N = 5,425), 1989 Independent variables Model 1 Model 2 Model 3 Model 4

Demographic characteristics Age 0.029 0.030 0.028 0.034 (3.61)*** (3.82)*** (3.47)*** (4.21)*** Currently married 0.803 0.732 0.809 0.740 (6.68)*** (6.16)*** (6.71)*** (5.47)*** Number of children ever born –0.130 –0.193 –0.128 –0.116 (2.24)** (3.34)** (2.17)** (1.86)* Number of surviving children 0.317 0.392 0.320 0.302 (5.18)*** (6.43)*** (5.15)*** (4.68)*** Socioeconomic characteristics Years of educationen 0.125 0.133 0.131 0.149 (8.67)*** (9.16)*** (8.92)*** (10.43)*** Currently workingen 0.635 0.608 0.642 0.112 (5.12)*** (4.89)*** (5.14)*** (3.97)*** Household economic measure1,en 0.325 0.258 0.311 0.339 (6.06)*** (4.83)*** (5.69)*** (5.14)*** Availability of contraception Average time to source of FPen –0.003 –0.004 –0.004 –0.004 (2.94)*** (3.58)*** (3.05)*** (3.01)*** Ethnicity Kamba 1.06 0.89 0.62 (2.54)*** (2.08)** (1.19) Kikuyu 1.18 0.99 0.48 (12.23)*** (6.40)*** (2.35)** Kisii 0.50 0.57 0.44 (2.94)*** (1.77)* (1.48) Meru-Embu 1.48 0.16 0.44 (9.68)*** (0.30) (0.93) District population per unit of political power/10,000 1963–67ex 0.00 0.02 0.01 (0.53) (2.41)** (0.84) 1968–72ex 0.06 0.07 0.05 (8.03)*** (2.84)*** (2.11)** 1973–77ex –0.14 –0.17 –0.12 (7.57)*** (2.74)*** (2.11)** 1978–82ex 0.01 0.01 0.01 (6.43)*** (2.70)*** (1.78)* 1983–87ex 0.22 0.19 0.19 (9.84)*** (2.61)*** (2.83)*** Constant –5.145 –7.000 –7.200 –6.627 (19.40)*** (16.75)*** (8.59)*** (8.94)*** Chi2 –1833.92 –1858.26 –1828.38 –1833.82

NOTE: Coefficients [eβ(z-score)] were estimated using robust logit regression on 1989 Kenya DHS data. Significant at the *** 1% level; ** 5% level; *10% level. enTreated as endogenous in model 4. FP = family planning. exTreated as exogenous in model 4. 1Household economic measure is constructed as a composite scale: has electricity + flush toilet + access to piped water.

Click to return to Table of Contents TABLE 2 Regression coefficients showing the effects of district-level political power and other covariates on women’s current use of modern contraception in rural Kenya (N = 5,370), 1993 Independent variables Model 1 Model 2 Model 3 Model 4

Demographic characteristics Age 0.016 0.020 0.016 0.021 (2.24)** (2.72)** (2.10)** (2.84)*** Number of children ever born –0.093 –0.137 –0.086 –0.080 (1.76)* (2.62)*** (1.66)* (1.53) Number of surviving children 0.359 0.394 0.355 0.350 (6.46)*** (7.13)*** (6.45)*** (6.32)*** Socioeconomic characteristics Years of educationen 0.171 0.174 0.172 0.177 (12.90)*** (13.21)*** (12.79)*** (13.16)*** Currently workingen 0.410 0.514 0.429 –0.004 (5.28)*** (6.49)*** (5.35)*** (0.31) Household economic measure1,en 0.169 0.235 0.143 0.139 (2.13)** (2.72)** (1.76)* (1.78)* Availability of contraception Average time to source of FPen –0.256 –0.167 –0.243 –0.249 (5.85)*** (3.47)*** (5.58)*** (5.64)*** Ethnicity Kamba 0.62 0.50 0.47 (4.05)*** (2.01)** (1.89)* Kikuyu 0.87 0.71 0.74 (8.12)*** (4.34)*** (3.64)** Kisii 0.61 0.44 0.49 (4.18)*** (1.36) (0.33) Meru-Embu 0.93 1.04 1.09 (6.59)*** (6.81)*** (5.67)*** District population per unit of political power/10,000 1963–67ex 0.013 0.019 0.025 (1.39) (1.84)* (2.36)** 1968–72ex 0.008 –0.004 –0.011 (1.08) (0.45) (1.08)* 1973–77ex 0.012 0.006 0.010 (3.62)*** (1.18) (1.83)* 1978–82ex 0.008 0.005 0.004 (2.71)*** (0.71) (0.57) 1983–87ex –0.130 –0.072 –0.010 (3.68)*** (1.63) (2.23)** 1988–92ex –0.009 0.006 0.006 (1.60) (1.60) (1.08)** Constant –3.158 –3.505 –3.250 –3.138 (10.79)*** (10.89)*** (3.70)*** (16.60)*** Log likelihood –2480.00 –2510.63 –2474.21 –2489.96

NOTE: Coefficients [eβ (z-score)] were estimated using robust logit regression on 1993 Kenya DHS data. Significant at the *** 1% level; ** 5% level; *10% level. enTreated as endogenous in model 4. FP = family planning. exTreated as exogenous in model 4. 1A composite scale: has electricity + flush toilet + access to piped water + bicycle.

Click to return to Table of Contents 458 E THNIC DIFFERENCES IN DEMOGRAPHIC BEHAVIOR differences in contraceptive use are largely explained in the 1989 data, and partly explained in the 1993 data, by differences in the level of political representation in the top tier of government.14

Conclusion

How valid are these results? I have argued that the indicator of political capital on which the results depend appears to be a valid construct for two reasons. First, the positive relationship that it posits between the amount of influence associated with a given position and that position’s seniority is consistent with the general theory of patron–client networks. And second, ethnic-specific fluctuations in the indicator of political capital correspond to political transitions in Kenya from the 1960s to the early 1990s insofar as, with the intriguing exception of family planning services, Kenya’s favored ethnic groups appear to have received more services. This, in turn, is consistent with the depiction of the structure of political relationships described in the theoretical section. Nevertheless, there are some weaknesses. The first is related to the level of analysis. The hypotheses would be better tested at the level of political constituencies rather than the district level since (1) they are based on a depiction of the patron–client network, that is, the relational structure that underlies an individual’s political capital, and (2) the locus of the patron–client network is the constituency. This is the level where the largest effects of unequal development and, consequently, political effects on demographic behavior can be expected. I was unable to conduct a constituency-level analysis because DHS data do not include the census enumeration areas from which the DHS sample is derived. Therefore sampled clusters cannot be identified with specific constituencies. The analysis has assumed, then, that the effects of patron–client networks are powerful enough to have had some spillover into neighboring constituencies, a reasonable assumption given the relative ethnic ho- mogeneity of administrative districts in Kenya, and an assumption that appears justified given the strength of the observed empirical effects. A second potential weakness is related to the construction of the measure of political capital. As mentioned above, the values assigned were arbitrary in at least one sense since it is not clear, for example, that a Parlia- mentary secretary has one-fifth the political capital of the president; he may have one-tenth the level, or one-fiftieth. On the other hand, there is clearly no a priori way to determine the appropriate relative weights of political capital within the top tier of the government. I have simply assumed, therefore, that there is a positive linear relationship between the amount of influence associated with a given position and that position’s seniority. Pre- sumably other specifications—in particular, nonlinear relationships of various

Click to return to Table of Contents A LEXANDER A. WEINREB 459 types—could be determined. To remain consistent with the political science and related literature reviewed above, however, they would all have to make the same key assumption: that political influence increases with political seniority. A third potential weakness is more substantive. It is possible that I have underestimated other sources of capital investment that may affect the distribution of determinants of individuals’ demographic decisions. This is because since the 1970s the Kenyan government has emphasized the importance of community self-help, known as Harambee, particularly in relation to the building of schools and clinics (Seeley 1987). Similarly, Kenya has received large amounts of donor funds. Some of these funds have been spent by nongovernmental organizations (NGOs) and churches. Other funds, we can assume, have been used to build or fund schools and clinics. I know of no data on how this supplementary nongovernmental spending occurs, which makes it difficult to directly assess the extent to which the focus on political capital misses this part of the relationship between development spending and subsequent demographic decisionmaking. But there are some indirect indications to suggest that it would have minimal effects on the analysis. For example, even though under Harambee rules the Kenyan government is supposed to provide a teacher or nurse for any school or clinic that a community builds, by the early 1990s only 60 percent of the clinics in Kenya had nurses (Schwarz 1996), and many of these clinics had no refrigeration, which limits the range of medications that a nurse can prescribe and the procedures that he or she can conduct (Ombura 1997, personal communication).15 In short, in relation to both medical staff and medical equipment, government provision of services does not automatically follow from community self-help. The danger of endogeneity in program placement, as I mentioned earlier, operates at a higher analytic level, allowing for more systematic biases in development spending than result from simple differences in communities’ demand for services. Private and NGO hospitals do not seem to have filled the gaps that have arisen from differential government spending by region. Ministry of Health data on numbers of doctors and hospital beds in the 1980s, including their affiliation (state, private, or NGO), are provided in Bloom and Segall (1993). Elsewhere I have weighted these by district population size, showing that per capita, the largest numbers of state, private, and NGO doctors are found in Kikuyu-dominated Central Province, slightly fewer in Kalenjin Rift Valley, and the fewest in Luo areas (Weinreb 2001). In short, I think it safe to claim that the analyses support the proposition that incorporating measures of political capital can account for the bulk of observed ethnic differences in contraceptive use. This is an important result for a number of reasons. First, it demonstrates the limited utility of ascribing ethnic differences, net of individual-level controls, to a residual

Click to return to Table of Contents 460 E THNIC DIFFERENCES IN DEMOGRAPHIC BEHAVIOR notion of culture. In the Kenyan context, for example, only Kikuyu and, to a more limited extent, Meru-Embu contraceptive use remains significantly different from that of other groups. With the exception of these two groups, there are no unexplained cultural differences at the level of ethnicity, although they presumably operate at the more local level.16 Second, showing that political relationships have micro-level consequences provides strong empirical support for the institutional arguments of McNicoll (1980) and others. Specifically, it shows that political (corporate) actors are at the forefront of constructing the risk environments in which demographic decisions are made. This has been long believed but rarely established. Third, the argument is also more generally useful. I presented a partial list of countries in sub-Saharan Africa in which ethnic competition for state power has been manifest—in terms of large-scale genocide and violence, less destructive (in terms of mortality) rebellions and insurgencies, and relatively peaceful competition within one- or multi-party states. In each of those cases, and in the many others that I ignored, I would expect differential access to development goods to follow from the ethnic competition. Dif- ferentials in demographic behavior would inevitably follow as well. The overall argument of this article, therefore, can be easily applied to observed differences in those societies. More generally still, the argument could be recast and applied to other residual group differences since the key conceptual focus is on the relationship between an individual’s status-group affiliation—which includes ethnicity, race, tribe, clan, religion, and class—and the individual’s access to political power and resources. Wherever these status groups are differentially associated with political power, these factors are likely to come into play. Nevertheless, I would expect the impact of such processes on demographic patterns to be much greater in low-income societies, since that is where the state tends to be a primary source of patronage and since relatively small investments in health and educational infrastructure have relatively large and observable effects, whether on demographic measures themselves or on the distribution of determinants of those measures. Finally, the results may have implications for the design and implementation of development policy. Some observers have suggested that the best way to minimize inefficient and inequitable distribution of development goods in sub-Saharan Africa is to design policies and strategies that take into account the multiethnic nature of most countries in the region (e.g., Kimenyi 1999; Mbaku 1999; Rowley 1999). This analysis strengthens the claims of these commentators by showing that development funds that shape demographic patterns support a “Pareto-inferior ‘steal-steal’ strategy” (Kimenyi 1999: 360) among ethnic groups. Administrators who design or fund development projects should be aware of how this can occur; of how, at the

Click to return to Table of Contents A LEXANDER A. WEINREB 461 national and local level, resources are used to promote narrower interests, irrespective of the accompanying rhetoric. If donors aim to maximize the efficiency and equity of their capital inputs, they may need to incorporate checks on distributive patterns. Demographers have a role to play in this process because, for better and worse, their role is necessarily political: they are at the forefront of those who measure non–wage-related inequalities— inequalities that result from unequal investments in health, social, and economic infrastructure.

Notes

The archival political data used in this article ernment representatives, including civil ser- were collected with the support of a Mellon vants, that while they may take, they should Foundation Grant to the Population Studies not to take “too much” (e.g., Harden 1991; Center at the University of Pennsylvania. Ear- Berkeley 2001). The crucial distinction, how- lier versions were presented at the 1999 An- ever, is that in this case, the accountability is nual Meeting of the Population Association of indirect: barring rebellion, extreme exploita- America and at the Policy Research Division tion could only be ended through Mobutu. of the Population Council in January 2000. The 3 According to Kenya’s The Daily Nation, author acknowledges comments from both sets members of the government and opposition of audience members. He is also indebted to made pronouncements about the link between Ruth Waswa for research assistance in Kenya political support and development funds al- and to Randall Collins, Herbert Smith, Susan most daily in the months preceding the 1997 Watkins, and Amelia Weinreb for comments election; current and past editions of this news- on an earlier draft. paper can be read at «http://www.nationaudio. 1 Even if one retreats to a “culture as iden- com/News/DailyNation/Today». Similarly, ex- tifier” model after eliminating the political fac- amples of district creation in Kenya during the tor, this does not rid the former of a more fun- past decade include the breaking off of new damental epistemological problem, also Kuria and Suba Districts from South Nyanza identified by Hammel (1990). Namely, such District in order, as one Suba informant told explanations are flawed insofar as they do not me, to “bring us closer to the President.” Luo identify the specific values and tastes that are informants were less sanguine. Both the Kuria responsible for behavioral differences. This may and Suba had been considered client groups be less problematic in some populations than of the Luo, so Luo leaders thought the district in others. For example, it seems more valid to creation smacked of “tribalism.” I discuss this ascribe stubbornly low contraceptive preva- theme below. lence in one group to an embedded pronatal- Additionally, strong cultural markers of the ism than, say, to ascribe lower rates of child structural characteristics that favor augmented vaccination in that same group to its disregard roles for political leaders can be found through- for children’s well-being. But in both cases the out African fiction. Leaders are invariably pic- lack of specificity makes these speculative tured as powerful, corrupt, and largely unac- rather than analytic solutions. countable. They are also commonly nameless. For example, Chinua Achebe’s (1966) term 2I am purposefully simplifying this pic- “Big Man,” since popularized in the journalis- ture since even in non-democracies there are tic literature (e.g., Harden 1991; Berkeley often limitations on the degree to which rep- 2001), first appears in A Man of the People. resentatives appointed by an autocrat can ig- Francis Imbuga’s (1976) Betrayal in the City and nore, exploit, and therefore remain unaccount- John Ruganda’s (1980) The Floods both refer able to local residents. In Zaire, for example, to “the Boss.” Kyeyune, one of Ruganda’s char- Mobutu famously (and openly) warned gov- acters, also refers to “the ogre.” Leaders as

Click to return to Table of Contents 462 E THNIC DIFFERENCES IN DEMOGRAPHIC BEHAVIOR

“Gamblers” appear in Opinya Okoth-Ogendo’s fective leaders can attempt to weaken opposi- (1971) poem of the same name and in Ambay- tion blocs by appealing to the smaller sub- isi Namale’s (1994) novel Honourable Criminals. factions to differentiate themselves. Several A more totalitarian (or perhaps theocratic) Kenyan informants described to me the re- term, “Father,” is also used (e.g., in Nevanji emergent Suba ethnicity in terms of Moi’s at- Madanhire’s 1996 novel Goatsmell). tempts to divide the Luo-led political bloc in 4 For journalistic accounts, see Harden Nyanza Province using “divide and rule” tactics. (1991), Buckoke (1991), and Berkeley (2001); It is also important not to discount the fact for a more focused study of corruption in that these hypotheses are more suitable for cer- Kenya, see Kibwana et al. (1996); fictional rep- tain African states than for others. Nyerere, for resentations appear throughout Chinua example, recognized the potential for a clash Achebe’s work as well as works of the authors of interests between legislators’ personal and mentioned in note 3. corporate interests and therefore prohibited politicians in Tanzania from running private 5 Achebe’s comments are worth quoting businesses. He also attempted to minimize eth- at length: “Nothing in Nigeria’s political his- nic competition, for example by enforcing the tory captures her problem of national integra- use of Kiswahili as a national language. Both tion more graphically than the chequered for- of these moves limited the potential for the tune of the word tribe.... Perhaps it was an emergence of tribalistic politics, though some unrealistic dream at the best of times [i.e., to tensions remained and may be reemerging dream of a Nigeria without tribalism], but some (e.g., Zanzibar versus mainland Tanzania). young, educated men and women of my gen- Finally, the distribution of the state’s re- eration did dream it.... But all this self-con- sources may be affected by other types of in- scious wish to banish tribe has proved largely terest-group conflict. Kitching (1982), Thiong’o futile because a word will stay around as long (1981), Bates (1989), and Himbara (1994), for as there is work for it to do. In Nigeria, in spite example, have argued that class is also a fac- of our protestations, there is plenty of work tor in distributive decisions in Kenya since re- for tribe.... A Nigerian child seeking admission sources have sometimes been directed toward into a federal school, a student wishing to en- members of the economic elite irrespective of ter a College or University, a graduate seeking their ethnicity. In general, however, while al- employment in the public service, a business- lowing for the effects of class on the distribu- man tendering for a contract, a citizen apply- tion of these development goods, I think such ing for a passport, filing a report with the po- effects are limited. While it is important to lice or seeking access to any of the hundred avoid reifying ethnicity, ethnic boundaries are thousand avenues controlled by the state, will relatively stable (Nash 1989). Moreover, the sooner or later fill out a form which requires accumulated evidence in support of the “eth- him to confess his tribe (or less crudely, and nic hypothesis” is compelling. more hypocritically, his state of origin)” (1983: 5–7; emphasis in original). 7 The Kalenjin is actually a coalition of 6I am not asserting maximalist hypoth- several “tribes” (e.g., the Nandi, Kipsiki, Pokot) eses in which (a) ethnicity is a “total” deter- who all speak the same “Kalenjin” (literally “I minant of political support and (b) the state’s told you”) language. They began to coalesce resources are distributed solely along ethnic into a more unitary political force in the 1940s. lines. Good leadership strategists—Presidents Similarly, the Luhya ethnic group is a coali- Kenyatta and Moi in Kenya are examples— tion of closely related groups such as the are careful to employ a range of strategies. Bukusu, Tadjoni, Wanga, Nyala, Kabras, These can include coopting members of “op- Maragoli, and Samia of Western Province. position” ethnic groups into the government Their political unification began slightly ear- (though some of the most able politicians in lier, in the 1930s. (Luhya, which is short for Kenya—for example, Pio Gama Pinto, Kung’u Abaluhya, loosely means “those of the same Karumba, Tom Mboya, J. M. Kariuki, and Rob- hearth.”) ert Ouko—have been murdered under mys- 8 In brief, the three characteristics of so- terious circumstances, and their murderers cial capital are: (1) it “inheres in the structure have never been identified). Alternatively, ef- of...relationships. To possess social capital, a per-

Click to return to Table of Contents A LEXANDER A. WEINREB 463 son must be related to others” (Portes 1998: 7); cedures make it impossible to disaggregate (2) it facilitates “certain action of individuals who these two groups, which in turn means that are within the structure” (Coleman 1990: 302), we cannot estimate how much of an effect but specifies “those others, not himself” as the the change in the sample’s composition be- “actual source of his or her advantage” (Portes tween 1989 and 1993 has on the estimates. 1998: 7); (3) it is a public good—that is, a resource (2) Small urban areas in 1989 can be identi- that can benefit all members of the relatively fied with a district; those sampled in 1993 closed social network. cannot. The latter were simply lumped into 9 By contrast, the effect of political capi- a general “small urban area” category. With- tal on the availability of contraception appears out such district-level identifiers, aggregated to operate in a completely different direction. political capital scores cannot be assigned, so Women with lower levels of capital tend to these data must be dropped. This likely re- have better access to sources of contraception. duces the analytic leverage of the 1993 data I discuss competing explanations for this in- by limiting the overall heterogeneity of the triguing relationship in Weinreb (2000b). The sample. (3) Because political capital scores most promising explanation, I think, com- are assigned at the district level, the smaller bines several factors. Specifically, the rela- number of districts sampled in 1993 also tionship is the product of (1) NGO attempts somewhat reduces analytic leverage. In short, to direct contraceptive services to parts of the while I expect the same general pattern of country that are underserved in other ways; political capital effects in both the 1989 and (2) a large Luo effect insofar as Luo men tend 1993 data, I also expect a more attenuated to be among the most educated in Kenya and pattern in the latter. are therefore heavily involved in NGO work 11 The only difference between the two (especially as there are constraints on their tables is that Table 1 includes a control for employment in the public sector), hence marital status. The control was not included would likely direct such resources back to in Table 2 since it had no significant effect in their home areas; and (3) measures of ac- 1993 and its exclusion did not substantively cessibility are likely sensitive to population change any of the other relationships. density in that, ceteris paribus, more densely 12 The results of these preliminary equa- populated areas need more clinics and pro- tions are available from the author upon re- viders, and they also tend to have more fre- quest. quent transport opportunities. Because the 13 Cleaning these independent variables most densely populated areas of the coun- in this way is conceptually equivalent to in- try include Luo and Kikuyu areas, both of strumenting the endogenous variables with the which had relatively low levels of political measures of political capital. This has the added capital during the 1980s and early 1990s, this advantage of allowing me to compare the di- likely affects the relationship between politi- rection, significance, and point estimates from cal capital and availability of contraception. model 4 to those generated in models 1 to 3. 10 The analysis is replicated on two 14 The independent effects of political datasets from the same setting in order to capital on contraceptive use—in particular, the validate the notion of political capital with extent to which, with the exception of the more confidence. That said, it is important Kikuyu, the initial users of contraception ap- to note some limitations with respect to the pear to have been women from areas with little comparability of the data. We can identify representation in the top tier of government— three concerns, all of which stem from are discussed elsewhere (Weinreb 2000). changes in DHS sampling and coding procedures between the two surveys. (1) Embu 15 Dr. Ombura was the Chief Medical Of- District in Eastern Province was sampled in ficer in the Homa Bay Hospital when I spoke 1989, but not in 1993, yet the ethnic coding with him in January 1997. “Meru-Embu” remained the same. While the 16 This, of course, begs the question: what Meru and Embu are closely related—cultur- makes the Kikuyu and Meru different? (Be- ally and linguistically—and were both allied cause Embu district was not sampled in 1993, with the dominant Kikuyu, these coding pro- observed 1993 differences primarily reflect

Click to return to Table of Contents 464 E THNIC DIFFERENCES IN DEMOGRAPHIC BEHAVIOR

Meru, rather than joint Meru and Embu, dif- proudly told me that the Kikuyus are the “Jews ferences, even though the DHS continued to of East Africa” because they practice circum- use joint Meru-Embu coding. I restrict my cision and are the best businessmen “and comments accordingly.) The Kikuyu and Meru criminals” in the region (obviously this says are concentrated in Central Province, the epi- as much about stereotypes of Jews as of center of British colonial settlement, land ap- Kikuyus, another interesting case of ideational propriation, and Mau Mau resistance, and the diffusion). Similarly, Kikuyu women are com- area closest to Nairobi. Largely because of this, monly described as counting their pennies at Central Province had the best general infra- the same time as they are having sex. In short, structure—roads, clinics, schools, etc.—and the it is quite possible that Central Province resi- most monetized economy in all of Kenya at dents—Kikuyu and Meru—adopted contra- independence (Bates 1989). The extended ceptive behavior earlier because their colonial family was also said to be relatively unimpor- experience undermined a range of traditional tant for these groups in comparison to other social institutions to a much greater extent groups. Indeed, popular stereotypes about the than was the case in other areas of the coun- Kikuyu reflect their perceived materialism. try, freeing them to make more unilateral de- One informant, a Kikuyu insurance executive, cisions with respect to reproductive behavior.

References

Achebe, Chinua. 1966. A Man of the People. Nairobi: East African Educational Publishing House. ———. 1983. The Trouble With Nigeria. Nairobi: Heinemann Kenya. Bangura, Yusuf. 1994. “Economic restructuring, coping strategies and social change: Impli- cations for institutional development in Africa,” Development and Change 25: 785–827. Barkan, J. D. 1979. “Legislators, elections, and political linkage,” in Joel D. Barkan and John Okumu (eds.), Politics and Public Policy in Kenya and Tanzania. New York: Praeger, pp. 64–92. ———. 1992. “The rise and fall of governance in Kenya,” in G. Hyden and M. Bratton (eds.), Governance and Politics in Africa. Boulder, CO: Lynne Rienner, pp. 167–192. ———. 1994. “Resurrecting modernization theory and the emergence of civil society in Kenya and Nigeria,” in David E. Apter and Carl G. Rosberg (eds.), Political Development and the New Realism in Sub-Saharan Africa. Charlottesville: University Press of Virginia, pp. 87–116. Bates, Robert H. 1987. “The politics of food crises in Kenya,” in Michael G. Schatzberg (ed.), The Political Economy of Kenya. New York: Praeger, pp. 75–92. ———. 1989. Beyond the Miracle of the Market: The Political Economy of Agrarian Development in Kenya. Cambridge: Cambridge University Press. Berkeley, Bill. 2001. The Graves Are Not Yet Full: Race, Tribe and Power in the Heart of Africa. New York: Basic Books. Birdsall, Nancy and Estelle James. 1993. “Efficiency and equity in social spending: How and why governments misbehave,” in Michael Lipton and Jacques van der Gaag (eds.), Including the Poor: Proceedings of a Symposium Organized by the World Bank and the Interna- tional Food Policy Research Institute. Washington, DC: The World Bank, pp. 335–358. Bloom, Gerald, and Malcolm Segall. 1993. Expenditure and Financing of the Health Sector in Kenya. IDS Commissioned Study no. 9. Brighton, UK: Institute of Development Stud- ies at the University of Sussex. Bradshaw, York W. 1990. “Perpetuating underdevelopment in Kenya: The link between agriculture, class, and state,” African Studies Review 33(1): 1–28. Brass, William and Carol L. Jolly (eds.). 1993. Population Dynamics of Kenya. Washington, DC: National Academy Press. Buchanan, James M. 1991. Constitutional Economics. Oxford: Basil Blackwell. Buckoke, Andrew. 1991. Fishing in Africa: A Guide to War and Corruption. London: Picador.

Click to return to Table of Contents A LEXANDER A. WEINREB 465

Bulatao, Rodolfo A. and Ronald D. Lee. 1983a. “A framework for the study of fertility determinants,” in Rodolfo A. Bulatao and Ronald D. Lee (eds.), Determinants of Fertility in Developing Countries: Volume 1: Supply and Demand for Children. New York: Academic Press, pp. 1–26. ———. 1983b. “The demand for children: A critical essay,” in Rodolfo A. Bulatao and Ronald D. Lee (eds.), Determinants of Fertility in Developing Countries: Volume 1: Supply and De- mand for Children. New York: Academic Press, pp. 233–287. Callaghan, Thomas M. 1987. “The state as lame leviathan: The patrimonial administrative state in Africa,” in Zaki Ergas (ed.), The African State in Transition. New York: St. Martin’s Press, pp. 87–116. Chazan, Naomi, Robert Mortimer, John Ravenhill, and Donald Rothschild. 1988. “Ethnicity, class, and the state,” in Naomi Chazan et al. (eds.), Politics and Society in Contemporary Africa. Boulder, CO: Lynne Rienner, pp. 101–125. Coleman, James S. 1990. Foundations of Social Theory. Cambridge, MA: Harvard University Press. Collins, Randall. 1988. Theoretical Sociology. San Diego, CA: Harcourt Brace Jovanovich. Dahlgren, Goran. 1994. “The political economy of health financing strategies in Kenya,” in Lincoln C. Chen, Arthur Kleinman, and Norma C. Ware (eds.), Health and Social Change in International Perspective. Boston: Harvard School of Public Health, pp. 453–470. Dietz, Tom and John Houtkamp. 1998. “Foreign aid to Africa: The rise and fall of structural development aid,” in Richard Grant and Jan Nijman (eds.), The Global Crisis in Foreign Aid. Syracuse, NY: Syracuse University Press, pp. 89–102. Frank, Odile and Geoffrey McNicoll. 1987. “An interpretation of fertility and population policy in Kenya,” Population and Development Review 13(2): 209–243. Freedman, Ronald. 1987. “The social and political environment, fertility, and family planning effectiveness,” in Robert J. Lapham and George B. Simmons (eds.), Organizing for Effective Family Planning Programs. Washington, DC: National Academy Press, pp. 37–57. Gordon, April. 1995. “Gender, ethnicity, and class in Kenya: ‘Burying Otieno’ revisited,” Signs: Journal of Women in Culture and Society 20(4): 883–912. Gourevitch, Philip. 1998. We Wish to Inform You That Tomorrow We Will Be Killed with Our Families: Stories from Rwanda. New York: Farrar, Straus, and Giroux. Greenhalgh, Susan. 1995. “Anthropology theorizes reproduction: Integrating practice, political economic, and feminist perspectives,” in Susan Greenhalgh (ed.), Situating Fer- tility: Anthropology and Demographic Inquiry. Cambridge: Cambridge University Press, pp. 3–28. Grindle, Merilee S. and John W. Thomas. 1991. Public Choices and Policy Change: The Political Economy of Reform in Developing Countries. Baltimore: Johns Hopkins University Press. Hammel, E. A. 1990. “A theory of culture for demography,” Population and Development Re- view 16(3): 455–485. Harden, Blaine. 1991. Africa: Dispatches from a Fragile Continent. London: HarperCollins. Himbara, David. 1994. “The failed Africanization of commerce and industry in Kenya,” World Development 22(3): 469–482. Holmquist, Frank. 1979. “Class structure, peasant participation, and rural self-help,” in Joel D. Barkan and John Okumu (eds.), Politics and Public Policy in Kenya and Tanzania. New York: Praeger, pp. 129–153. Holmquist, Frank and Michael Ford. 1994. “Kenya: State and civil society the first year after the election,” Africa Today 41(4): 5–25. Hoogvelt, Ankie M. M. 1997. Globalization and the Postcolonial World: The New Political Economy of Development. Baltimore: Johns Hopkins University Press. Hyden, Goran. 1979. “Administration and public policy,” in Joel D. Barkan and John Okumu (eds.), Politics and Public Policy in Kenya and Tanzania. New York: Praeger, pp. 93–113. ———. 1989. “Local governance and economic-demographic transition in rural Africa,” in Geoffrey McNicoll and Mead Cain (eds.), Rural Development and Population: Institutions and Policy. Population and Development Review 15(Supp): 193–211.

Click to return to Table of Contents 466 E THNIC DIFFERENCES IN DEMOGRAPHIC BEHAVIOR

Imbuga, F. D. 1976. Betrayal in the City: A Play. Nairobi: Heinemann Kenya. Kertzer, David I. 1995. “Political-economic and cultural explanations of demographic behavior,” in Susan Greenhalgh (ed.), Situating Fertility: Anthropology and Demographic In- quiry. Cambridge: Cambridge University Press, pp. 29–52. Khadiagala, Gilbert M. and Michael G. Schatzberg. 1987. “The Kenya bourgeoisie, external capital, and the state: An introduction,” in Michael G. Schatzberg (ed.), The Political Economy of Kenya. New York: Praeger, pp. 1–13. Kimenyi, Mwangi S. 1999. “Spatial competition, ethnicity, and the optimal size and composition of units of collective choice,” in Mwangi D. Kimenyi and John Mukum Mbaku (eds.), Insititutions and Collective Choice in Developing Countries: Applications of the Theory of Public Choice. Aldershot: Ashgate, pp. 355–380. Kimenyi, Mwangi S. and John Mukum Mbaku (eds.). 1999. Institutions and Collective Choice in Developing Countries: Applications of the Theory of Public Choice. Aldershot: Ashgate. Kitching, Gavin. 1982. Class and Economic Change in Kenya: The Making of an African Petite- Bourgeoisie, 1905–1970. New Haven: Yale University Press. Kreft, Ita and Jan de Leeuw. 1998. Introducing Multilevel Modeling. London: Sage Publica- tions. Kugler, Jacek, A. F. K. Organski, J. Timothy Johnson, and Youssef Cohen. 1983. “Political determinants of population dynamics,” Comparative Political Studies 16: 3–36. Lapham, Robert and W. Parker Mauldin. 1984. “Family planning program effort and birth rate decline in developing countries,” International Family Planning Perspectives 10: 109– 118. Levitt, Steven D. and James M. Poterba. 1999. “Congressional distributive politics and state economic performance,” Public Choice 99: 185–216. Levitt, Steven D. and James M. Snyder, Jr. 1997. “The impact of federal spending on house election outcomes,” Journal of Political Economy 105: 30–53. Madanhire, Nevanji. 1996. Goatsmell. Nairobi: East African Educational Publishing House. Mann, Michael. 1986. The Sources of Social Power. Cambridge: Cambridge University Press. Mbaku, John Mukum. 1999. “Democratization and the crisis of policy reform in developing countries,” in Mwangi D. Kimenyi and John Mukum Mbaku (eds.), Insititutions and Collective Choice in Developing Countries: Applications of the Theory of Public Choice. Aldershot: Ashgate, pp. 327–354. McChesney, Fred S. 1997. Money for Nothing: Politicians, Rent Extraction, and Extortion. Cam- bridge, MA: Harvard University Press. McNicoll, Geoffrey. 1980. “Institutional determinants of fertility change,” Population and Development Review 6(3): 441–462. ———. 2001. “Government and fertility in transitional and post-transitional societies,” in Rodolfo A. Bulatao and John B. Casterline (eds.), Global Fertility Transition. Population and Development Review 27(Supp): 129–159. McNicoll, Geoffrey and Mead Cain. 1989. “Institutional effects on rural economic and demographic change,” in Geoffrey McNicoll and Mead Cain (eds.), Rural Development and Population: Institutions and Policy. Population and Development Review 15(Supp): 3– 42. Meier, Gerald M. (ed.). 1991. Politics and Policy Making in Developing Countries: Perspectives on the New Political Economy. San Francisco: ICS Press. Namale, Ambayisi. 1994. Honourable Criminals. Nairobi: East African Educational Publish- ing House. Nash, Manning. 1989. The Cauldron of Ethnicity in the Modern World. Chicago: University of Chicago Press. National Council for Population and Development (NCPD), Central Bureau of Statistics, and Macro International. 1994. Kenya Demographic and Health Survey 1993. Calverton, MD: NCPD, CBS, and MI. National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: National Academy Press.

Click to return to Table of Contents A LEXANDER A. WEINREB 467

Nyangira, Nicholas. 1987. “Ethnicity, class, and politics in Kenya,” in Michael G. Schatzberg (ed.), The Political Economy of Kenya. New York: Praeger, pp. 15–31. O’Connor, James. 1973. The Fiscal Crisis of the State. New York: St. Martin’s Press. Okoth-Ogendo, Opinya H. W. 1971. “The gambler,” in David Cook and David Rubadiri (eds.), Poems from East Africa. Nairobi: East African Educational Publishing House, pp. 132–133. Organski, A. F. K. et al. 1984. Births, Deaths, and Taxes: The Demographic and Political Transi- tions. Chicago: University of Chicago Press. Population Reference Bureau. 2000. World Population Data Sheet. «www.prb.org». Portes, Alejandro. 1998. “Social capital: Its origin and applications in modern sociology,” Annual Review of Sociology 24: 1–24. Rouyer, Alwyn R. 1987. “Political capacity and the decline of fertility in India,” The Ameri- can Political Science Review 81(2): 453–470. Rowley, Charles K. 1999. “Rent seeking and rent extraction from the perspective of Af- rica,” in Mwangi D. Kimenyi and John Mukum Mbaku (eds.), Insititutions and Collec- tive Choice in Developing Countries: Applications of the Theory of Public Choice. Aldershot: Ashgate, pp. 223–254. Ruganda, John. 1980. The Floods: A Play. Nairobi: East African Educational Publishing House. Schraeder, Peter J. 1993. “Ethnic politics in Djibouti: From ‘eye of the hurricane’ to ‘boiling cauldron,’” African Affairs 92: 203–221. Schwarz, Ronald A. 1996. The Health Sector in Kenya: Health Personnel, Facilities, Education and Training (2nd Edition). Nairobi: Development Solutions for Africa. Seeley, Janet. 1987. “Social welfare in a Kenyan town: Policy and practice, 1902–1985,” African Affairs 86: 541–566. Shaw, Timothy M. 1986. “Ethnicity as the resilient paradigm for Africa: From the 1960s to the 1980s,” Development and Change 17: 587–605. Skocpol, Theda. 1979. States and Social Revolutions: A Comparative Analysis of France, Russia, and China. New York: Cambridge University Press. Smoke, Paul. 1993. “Local government fiscal reform in developing countries: Lessons from Kenya,” World Development 21(6): 901–923. Stamp, Patricia. 1991. “Burying Otieno: The politics of gender and ethnicity in Kenya,” Signs: Journal of Women in Culture and Society 16(4): 808–845. Thiong’o, Ngugi wa. 1981. Detained: A Writer’s Prison Diary. Nairobi: Heinemann. Throup, David W. 1987. “The construction and destruction of the Kenyatta State,” in Michael G. Schatzberg (ed.), The Political Economy of Kenya. New York: Praeger, pp. 33–74. United Nations Centre on Transnational Corporations. 1989. “Volume Eleven: Appendix IX: Evolution of Parastatal Joint Ventures in Kenya.” Development Through Partnerships: Business Relations Between State and Foreign Enterprises in Kenya. New York. Weinreb, Alexander A. 2000a. “Integrating respondents, community, and the state in the analysis of contraceptive use in Kenya,” Ph.D. dissertation, University of Pennsylvania. ———. 2000b. “Crisis or conspiracy? The effects of political capital on contraceptive use in Kenya.” Mimeo: NORC/University of Chicago. ———. 2001. “A state-centered, political capital approach to the explanation of demographic differences: With special reference to regional/ethnic inequality in sub-Saharan Af- rica,” in Proceedings of the General Conference of the International Union for the Scientific Study of Population (IUSSP), 18–24 August, Salvador, Brazil. Young, Crawford. 1994. “Evolving modes of consciousness and ideology: Nationalism and ethnicity,” in David E. Apter and Carl G. Rosberg (eds.), Political Development and the New Realism in Sub-Saharan Africa. Charlottesville: University Press of Virginia, pp. 61–86.

Click here to print article ClickClick toto returnreturn toto TableTable ofof ContentsContents Click to return to Table of Contents Is There a Stabilizing Selection Around Average Fertility in Modern Human Populations?

ULRICH MUELLER

IN MOST PLACES and at most times throughout history, the human species has displayed a fertility level well below the physiological maximum. In modern industrialized societies—with well-developed health care, infant mortality at an all-time low, and completed fertility per woman barely at the replacement level—the difference between the physiological maximum and the actual number of births may be greater than at any previous time. It seems that in all human societies the distribution of completed fertility by number of children born to the average women is unimodal: it displays a single most frequent value. In industrialized societies, furthermore, the most frequent number of children is often the median number, around which the distribution is relatively steep: about 60–70 percent of all women have a completed fertility within one birth above or below the modal value. This can be observed in various industrialized societies and seems to be a fairly stable feature over time (see Figure 1 depicting fertility distribution for US female birth cohorts 1891–1958). One might speculate that many of these women and their husbands with one, two, or three children could easily bring up one or two additional children, but they prefer not to. Empirically well-founded sociological explanations for this behavior abound: the increasing economic cost of rearing children, changing value orientations, alternative sources of satisfac- tion, both sexes spending long years in the educational system, the special burdens of working mothers, and so on. From an evolutionary perspective, however, these are only proximate factors. The economic and technological conditions in industrialized societies may well enforce a low mean fertility, but that still would leave room for an individual competition for maximal fertility.

POPULATION AND DEVELOPMENT REVIEW 27(3):469–498 (SEPTEMBER 2001) 469

Click here to print article Click to return to Table of Contents 470 S TABILIZING SELECTION IN MODERN HUMAN POPULATIONS

FIGURE 1 Distribution of US women aged 50–54 by number of children ever born: Selected cohorts 1891–1958a Birth cohorts 1891–1900 Birth cohorts 1901–10 40 40 35 35 30 30 25 25 20 20 15 15 10 10 Percent of all women 5 5 0 0 01234567+ 0 1234567+ Number of children ever born

Birth cohorts 1911–20 Birth cohorts 1921–30 40 40 35 35 30 30 25 25 20 20 15 15 10 10 5 5 0 0 01234567+ 0 1234567+

Birth cohorts 1940–44 Birth cohorts 1954–58 40 40 35 35 30 30 25 25 20 20 15 15 10 10 5 5 0 0 01234567+ 0 1234567+

aBirth cohorts 1940–44 and 1954–58: women aged 40–44. SOURCES: Birth cohorts up to 1930: Demographic Yearbook of the United Nations; Birth cohorts 1940–44 and 1954–58: U.S. Census Bureau, Current Population Surveys 1980 and 1998.

Click to return to Table of Contents U LRICH MUELLER 471

In her review Borgerhoff Mulder (1998) lists four evolutionary explanations of the demographic transition: (1) Stabilizing selection: some intermediate, rather than the largest possible number of children may maximize long-term fitness (Hill and Hurtado 1996; Kaplan et al. 1995a, 1995b). (2) Darwinian, but nongenetic cultural mechanisms of inheritance may spread fitness-depressing behavior (Boyd and Richerson 1985). (3) Maladaptive adaptation: low fertility may be a byproduct of the requirements of modern labor markets to which human fertility behavior has not adapted (Vining 1986). (4) Evolutionary economics: people maximize the sum of energetic endowments of their descendants, which means, in modern economies, that they maximize the total income of their descendants (Kaplan 1996). Borgerhoff Mulder did not mention a fifth explanation with a distinguished ancestry in demography: group selection (as described by Carr- Saunders 1922) is a prime candidate for explaining voluntary reproductive restraint using the Darwinian principles of genetic mutation and natural selection. While individuals not restraining their reproduction (whether by genetic or cultural predisposition) will be fitter than those who do, unre- strained reproduction will deplete the natural resources of the group down to mere subsistence level, with greater risk of extinction of whole family lineages. Groups of individuals who cooperate for joint reproductive restraint do not deplete their resources and, therefore, may be fitter than groups of non-cooperators. An unsuccessful strategy in competition between individuals is successful in competition between groups. Stabilizing selection is the only other of the five approaches to explaining the demographic transition to tackle the problem within the context of natural selection without making far-reaching additional assumptions about human reproductive decisionmaking. The argument is as follows. In industrialized societies after demographic transition, the relatively narrow distribution of completed family size around the population average could be the result of a stabilizing selection around that mean. One might further hypothesize that demographic transition is the result of first a directional, then a stabilizing selection toward a decreasing optimal family size, later fixed at some low level. Stabilizing selection means that some intermediate value of the trait in question (thickness of fur, body size, number of children) is optimal with respect to long-term reproductive success. Consequently, in a stable environment, most members of the population will have trait values close to this optimum. Alternatives would be (1) directional selection where the optimum is at the margin, or maybe even beyond the range of observed trait values; or (2) disruptive selection where there is more than one optimum.1 To argue in favor of a stabilizing selection around average fertility in a given population requires empirical evidence to test two predictions.

Click to return to Table of Contents 472 S TABILIZING SELECTION IN MODERN HUMAN POPULATIONS

(1) There exists a parents’ reproductive fitness curve conveniently measured, with a maximum at the most frequent number of live births (for a conceptual representation see Figure 2A). Strictly speaking, this holds true only if parents are in full control of allocating investments to rear children. To the degree that children may gain some control over investment allocation by parents, a lower number of children may be selected than the one— still an intermediate number—that maximizes parents’ fitness. In any fitness curve, no children at all, of course, corresponds to zero fitness, and so does a family size beyond some threshold when parental investment per child becomes too small for survival. (2) There is a decreasing marginal fitness gain for each additional child in the broad neighborhood of the mean (see Figure 2B). We need not show that total parental investment is fixed or varies independently of the number of children. Parents may even be able to increase their total investment (direct investment plus opportunity costs) with additional children infinitely, as long as the investment grows less than proportionally. Then the marginal fitness gain—if any—for each additional child will decrease, and there will be a tradeoff between fitness and total number of children. Initiated by Lack’s studies of optimal clutch size in birds (Lack 1947, 1948, 1954, 1966, 1967, 1968), a long tradition of theoretical and empirical research has sought to determine the optimal number that maximizes long- term reproductive success. Results from numerous empirical studies of animals seem to support Lack’s hypothesis that the modal number of offspring is the median and that it maximizes parental reproductive fitness. But these results still leave important questions unanswered. Most of this research has been conducted on offspring and clutch size in insects, fish, and birds. Various criteria of fitness were used. Surveys of the literature (Klomp 1970; Nur 1984; Finke, Milinkovick, and Thompson 1987; Roff 1992: 292 ff; Stearns 1992: 155 ff.; Sinervo and Svensson 1998) demonstrate that, in fish and birds, Lack’s central prediction does not always hold. Some studies showed lower fitness in above-average-sized broods; others found the average number of offspring to be below the most productive number. A possible explanation for the latter could be parent–offspring conflict: from a child’s perspective optimal family size generally is smaller than from a parent’s perspective. A recent study (Einum and Fleming 2000) on the tradeoff between offspring size and number in a fish species with high juvenile mortality showed, however, that mothers have the upper hand in this conflict: optimal egg size maximizes mother’s, not offspring’s fitness. Whether this is so in species—like ours—that rear their children remains an open question. Also, experimentally enlarged clutch sizes decrease survival of parents, and thus reduce parents’ lifetime number of surviving children (overview in Roff 1992: 157–178). A crucial shortcoming of many studies in this field is that the criteria used (e.g., number of offspring fledged, average weight of fledglings, sur-

Click to return to Table of Contents U LRICH MUELLER 473

FIGURE 2A Total fitness as a function of number of children as implied in stabilizing selection theory otal fitness T

Number of children

FIGURE 2B Marginal fitness as a function of number of children as derived from the total fitness function

0 Marginal fitness

Number of children

Click to return to Table of Contents 474 S TABILIZING SELECTION IN MODERN HUMAN POPULATIONS vival of fledglings or parents to next season) are imperfect indicators of future reproductive performance of offspring, the true criterion of long-term fitness of parents and species. In the few cases where offspring lifetime performance was available, the predicted negative effect of enlarged brood size could be demonstrated. Humans usually have a “litter size” of only one. In natural contexts twins, not to speak of triplets or quadruplets, have a greatly increased mortality. On the other hand, humans make lifelong direct investments in the form of resource transfers to their children, even after they reach adulthood and have their own children (Low 2000: chapter 6 with many references). This rare trait is one we share with only a few species—none of them primates—such as beavers (Ryden 1986). Reproductive success of children crucially depends on these transfers. Therefore a competition for parental investment occurs between children born at different times, just as competition takes place between fledglings in the same brood. By this reasoning we may extend Lack’s model of optimal brood size to optimal lifetime number of offspring in humans. But there are few studies of this kind in humans. In studies on optimal interbirth intervals in hunter-gatherer societies, total number of children surviving to adolescence is maximized at birth intervals of four years (Hobcraft, McDonald, and Rutstein 1983; Blurton Jones 1989). In modern societies, with low infant mortality even at very short interbirth intervals, the only meaningful parameter of fitness must be reproductive performance of offspring. Testing the two aforementioned predictions is not easy, since birth registration records and census data do not contain information about a person’s number of children and grandchildren. One approach would be to document an adult person’s number of siblings and nieces and nephews, together with own children, thus in effect measuring the reproductive performance of the parents of the respondent. The few empirical findings available on modern societies do not support the stabilizing selection hypothesis. For example, Essock-Vitale and McGuire (1985a, 1985b) found that never-married middle-class women (a very small subsample N=17/300) in the Greater Los Angeles area had not only fewer children but also fewer siblings and fewer nieces and nephews than ever-married women. The authors suggest that low numbers of nieces and nephews may indicate that “helping-at-the-nest” does not play a role in enhancing reproductive fitness in industrial societies—although it does do so in traditional societies as shown in Vernier’s (1977) study of unmarried younger sisters in peasant families on the Greek island of Kárpathos. Since in modern societies unmarried females usually do not cohabit with their married siblings, this finding should not come as a surprise. Lopreato and Yu (1988), in a 1984 US national sample of ever-married women, found for the age group 15–49 years a positive but low correlation (r = .14, p <.001) between number of siblings and number of live births. They found no such correlation in the

Click to return to Table of Contents U LRICH MUELLER 475 age group 50 years and older. Unfortunately, the authors did not control for birth cohort effects. This considerably limits the relevance of their findings, since this age group contains women and their siblings born before, during, and after the baby boom. Lancaster and Kaplan (1991) and Kaplan et al. (1995a, 1995b), in a large sample of men in Albuquerque, New Mexico, detected a robust positive correlation between number of children and number of grandchildren. Similar findings were reported by Murphy and Knudsen (2000) using register data from Statistics Denmark. Little is known about the effect of parental investment on fitness as measured by number of grandchildren. Parental investment may take the form of investment in a child’s education or of direct transfer payments to young couples with small children. Opportunity costs of investment in children are even less tangible and more difficult to measure. In a traditional society of pastoralists (Mace 1996, 1998, 2000) and in nineteenth-century Sweden (Low 1991) younger sons had fewer surviving children than older sons, presumably because they inherited fewer resources from their parents. No such differential was observed among daughters, presumably because they were treated equally in the inheritance regime. For the Ache foragers in Paraguay in the 1990s, Hill and Hurtado (1996) found a positive correlation between number of children and number of grandchildren for men, but not for women. For the Kipsigis, agropastoralists in Kenya in the 1980s, Borgerhoff Mulder (2001) showed that an intermediate number of children may be optimal for maximizing women’s number of grandchildren. For men, the results were inconclusive. In some historical European societies a positive relation between birth order and child mortality has been observed (Cohen 1975; Wrigley and Schofield 1981; Voland and Dunbar 1995; Voland et al. 1997). To explain such a relationship, we need not directly measure parental investment. We need only take for granted that total parental investment, as far as it has fertility-enhancing effects, keeps increasing, but does so less than proportionally to the total number of children. For our two predictions to make sense requires a measure of reproductive fitness other than completed family size (F1), the simplest such measure. For measuring long-term fitness effects, we can use either a deterministic measure—number of grandchildren (F2) or great-grandchildren (F3); or a stochastic measure—long-term lineage survival probabilities, as projected from present reproductive patterns within a family (Tuljapurkar 1990; Yoshimura and Clarke 1991, 1993; Mueller 1991). There is no universally accepted definition of fitness (Stearns 1992: 31–33), and none is suggested here. For the purpose of this article, we adopt a pragmatic stance, namely that long-term reproductive fitness is a positive function of both a deterministic and a stochastic measure. Neither sort of fitness measure necessarily gives unequivocal results. Con- sider a simple example. Assume nonoverlapping generations of equal length among two groups of families, A and B. Among families in group A, one-third

Click to return to Table of Contents 476 S TABILIZING SELECTION IN MODERN HUMAN POPULATIONS remain childless and the rest have four children. Among families in group B, one-quarter remain childless and the rest have three children. With sexual reproduction, cumulative extinction probabilities converge fast; the majority of all extinctions will occur during the first two generations. Then, as a simple calculation will show, the average individual in group A can expect to have 18.96 great-grandchildren, and the average individual in group B can expect to have 11.39. But the group A individual faces a .36 chance of having no great-grandchildren, while B’s chances of having no great-grandchildren is only .26. Which family has greater long-term reproductive fitness? Since all populations are finite, every lineage has an extinction probability greater than zero. This would imply that evolution selects reproductive strategies that maximize lineage survival probability rather than expected number of descendants. An alternative formulation would be: in a finite population, a reproductive strategy that increases the expected number of descendants will become fixed in the respective lineage if and only if this strategy increases the lineage survival probability. Consequently, appropriately defined stochastic reproductive fitness measures carry more information than deterministic ones. The question is whether a stabilizing selection around average— that is, replacement—fertility has some relevance for the sociological understanding of industrialized societies as well. Evidence for a stabilizing selection, together with low childlessness rates, would support Daly and Wilson’s conclusion (1983: 337) that in such societies selection pressure and speed of evolution have decreased considerably, in comparison to all previous societies. This view has become almost a commonplace among evolutionary demographers. Evidence for a stabilizing selection, together with low childlessness rates, would also mean that the present population in a society consists of direct descendants of a large proportion of those living, say, ten generations ago. The alternative would be that we all are descendants of only a small proportion of those living ten generations ago, mostly large families. Such families are known to differ from other families in many respects: they are more frequent in rural than in urban areas; they are more frequent in certain status and occupation groups; they may migrate less, be more religious, and have different values and behavior patterns. If the second alternative holds, then anyone interested in the long waves of social change should pay much more attention to the influence of large families than is usually done.

New datasets for testing stabilizing selection in modern populations

In two mailed surveys conducted in 1990 and 1991 (for details see Mueller and Mazur 1998), 437 of the 540 surviving members of the class of 1950 of

Click to return to Table of Contents U LRICH MUELLER 477 the US Military Academy at West Point (born between 1923 and 1929, median year 1927) and 163 out of a sample of 408 retired noncommissioned officers of the US Army (born between 1913 and 1937, median year 1923) were questioned about their professional and family biographies. The two groups of men had rather different social backgrounds: more middle class in the West Point group, more lower class in the noncommissioned officers sample. For instance, while 47 percent of the West Point graduates had at least one parent and 27 percent at least one grandparent with a college degree, the corresponding proportions for the noncommissioned officers were 4 percent and 2 percent. A larger proportion of the parents in both samples than in the general American population lived in rural areas. About 10 percent of respondents in each group had a professional soldier as father. The West Point group was distinguished by low childlessness (6 percent) and rather large families: the average number of children born was 3.16—both of these being characteristics of a cultural submilieu with strong pronatalist values (Atkinson 1989; Fleming 1981). The corresponding figures for the retired noncommissioned officers were 15 percent childless and 2.33 children born. Two similar surveys were conducted by the author in 1992 and 1993 on a sample of 480 western German physicians (29 percent female) from the state of Hessen, and a sample of 493 eastern German physicians (40 percent female) from the state of Saxony (both groups born between 1930 and 1935). Questionnaires had been sent to all registered active physicians in the two states, yielding a return rate of approximately 45 percent in each. Respondents in both samples had a strong middle-class background and up- bringing. Sixteen percent of the respondents were childless and the average family size was 2.16 in the Hessian sample; 13 percent were childless and the average family size was 1.96 in the Saxonian sample (for details see Mueller 2001a). From the four surveys we use three variables for the present study: number of respondent’s own children; number of respondent’s brothers and sisters; and number of respondent’s nieces and nephews, by sibling. Thus, all children and grandchildren of respondents’ parents are known. The questionnaires did not ask for number of children and grandchildren of respondents’ aunts and uncles, since a pretest showed that the answers were not reliable. Thus, the total number of great-grandchildren of respondents’ grandparents is not available. Because of the sampling method, average family size in the generation of the respondents’ parents is overestimated: childless couples cannot appear, and children from large families have a greater chance to be sampled than would be expected from the relative number of these large families among all families. However, we are interested in the shape of the distribution of offspring over generations, not in estimating their true parameters

Click to return to Table of Contents 478 S TABILIZING SELECTION IN MODERN HUMAN POPULATIONS in the population. We want to know whether there is a fitness-optimizing number of children close to the population average; or whether such an optimal number, while substantially larger than the average, is still some intermediate value beyond which fitness will decline again; or whether there is a monotonically increasing relation between number of children and long- term reproductive fitness secured by them. Finally, from public sources (McNaughton 1973; for dataset details see Mueller 2001b) in a register-based total documentation, vital data were obtained for 1,545 members of European royalty (from the dynasties of Bernadotte, Bourbon-Parma, Guelph-Hannover-Windsor, Habsburg, Hesse, Hohenlohe, Hohenzollern, Holstein-Gottorp-Romanov, Mecklenburg, Nassau, Oldenburg- Holstein, Reuss, Wettin, Wittelsbach, Württemberg) born between 1 January 1790 and 31 December 1939, including dynastic status and exact relationship to each other (a highly inbreeding group). This dataset has the special feature that number of children, grandchildren, great-grandchildren, and great-great- grandchildren could be identified for about 100 males. Of all members of the royalty sample surviving to age 50, 29 percent were childless; the average number of children born was 2.87, with little difference in average childlessness and family size between sexes (Mueller 1993). The demographic transition took place in this special population around the time of the French Revolution. Sta- tistical tests show that birth cohort had no effect on childlessness but had an effect on number of children: earlier cohorts had more children. The latter has no effect on the relation between number of children and number of grandchildren—the focus of this article—as long as the cohort effect affects families of all sizes equally. This was indeed the case. The five dynasties with the lowest overall fertility (Hesse, Mecklenburg, Oldenburg-Holstein, Holstein-Gottorp- Romanov, Württemberg) as well as the five dynasties with the highest overall fertility (Bourbon-Parma, Guelph-Hannover-Windsor, Habsburg, Nassau, Wittelsbach) displayed associations of the same strength between birth cohort and childlessness or birth cohort and number of children. While these five samples are in no way representative of the general population of their respective countries, they are independent of one another; sampling selectivities with regard to the target variables of this study are not plausible. Therefore, similar results in all five samples could not be explained as peculiarities of one special subpopulation.

Measuring fitness

We chose two measures of fitness: one deterministic, the other stochastic. As demonstrated in the earlier hypothetical example of groups A and B, the stochastic measure is the less obviously relevant one, albeit the more important measure. Since, in the end, all populations become extinct, extinction probability has become the measure of choice for comparing the differential survival chances

Click to return to Table of Contents U LRICH MUELLER 479 of populations and lineages (for reviews of measures and models, see Ludwig 1996; Halley and Iwasa 1998). Here we model different lineages within populations as small, asexually reproducing populations, which we can justify as modeling the reproduction of Y-chromosomes (from father only) or of mitochondrial DNA (from mother only). Applying this to real human lineages is acceptable once we show that the sex ratio among offspring is not related to number of offspring. Comprehensive models of the dynamics of small populations consider demographic stochasticity (random fluctuations caused by the discreteness of natural numbers) and environmental stochasticity (typically incorporating random fluctuations of demographic parameters as white noise, and rare catastrophic population crashes, often modeled as a Poisson process, that may reflect a limited carrying capacity). Here, we consider only demographic stochasticity, since all strata and lineages in these particular populations, irrespective of family size throughout the last centuries, have been equally well shielded against all kinds of environmental stochasticity and will continue to be so in the foreseeable future. Still, a variety of measures of lineage survival are available, such as the probability of extinction occurring within a specified number of generations; an asymptotically instant population extinction rate; or any measure of the distribution of times to extinction (Halley and Iwasa 1998). Here, instead, we propose the expected number of generations to lineage extinction, as projected from two observed variables: the proportions of childless children among F1 (the first generation of offspring) and the total fertility of all children in F1. Projections of cumulative event probability from a set of cross- sectional data are standard practice in demography. Also, the two parameters capture all we know about the reproductive regimes of individual family lineages. By considering only demographic stochasticity, we will certainly overestimate the true expected number of generations to lineage extinction, which, however, is not problematic for comparing fitness differentials. How are we to model the childlessness rates in our projections of the expected number of generations to lineage extinction? On the one hand, in our datasets fertility of parents did not predict the fertility or mortality of children, but this does not exclude various direct or indirect influences of parents’ reproductive behavior on children’s reproductive behavior (Kaplan et al. 1995a and 1995b report such influences). A non-negative effect of family size on childlessness rates among descendants, on the other hand, directly follows from the stabilizing selection hypothesis investigated here. Therefore we chose to model future childlessness in the lineage as being in part determined by average childlessness rates among offspring by family size category and in part by childlessness rates among the children of the individual’s family. Then, the expected number of generations to lineage extinction can be calculated as follows:

Let pindividual (childlessness among children of i) be the proportion childless among the children of individual i, and let there be a pcollective (average childlessness, F1=j)—the proportion of childless individuals in the population, by fam-

Click to return to Table of Contents 480 S TABILIZING SELECTION IN MODERN HUMAN POPULATIONS ily size classes j, that is, the proportion childless if F1 = 1, F1 = 2, F1 = 3, and so on. Then we assume that the risk of childlessness is half determined by influences of particular families and half by the population average in families of a given size. Thus, for an individual i with a family size of F1 = j, the estimated proportion childless in his or her family lineage is

p (childless) = mean(p (average childlessness, F1 = j), i collective (1) pindividual(childlessness among children of i)). Then the survival chances of the whole family lineage through F3, given present levels of childlessness and fertility, can be projected as

F2 αi = 1 – pi(childless) . (2) Since an extinction in generation k occurs with probability

k–1 (1 – αi)αi , (3) the expected number of generations to lineage extinction is:

k k−1 kk1− αα with = 123 , ,→∞ , ∑ ()i i (4) 1 which, after algebraic manipulation, gives:

1 . (5) 1− αi

If F2 correlates with F1, we have to show that the proportion of childless individuals in F1 in our datasets is independent of the size of F1, as equation (2) presupposes. We also have to show that the sex ratio in F1 is independent of its size, since equations (1) – (4) model a one-sex population. We present the results of these calculations in a later section. Operationalizing our general hypothesis as in our first prediction from the introductory section would lead to the result that, if there is a stabilizing selection in modern populations, the average number of F1 will maximize F2, or the average number of F1 will maximize the expected number of generations before lineage extinction. We may relax this assumption as in our second prediction above and still be able at least partly to verify our general hypothesis. There may be various forms of unobserved heterogeneity with regard to the optimal size of F1. A necessary condition of any stabilizing selection is that there be a tradeoff between possible parental investment per child—resulting in an increased fitness gain to the parent from this child—and number of children. Fitness gains achieved by adding children to already existing ones beyond average F2 size—the marginal fitness gain—will be possible only at a

Click to return to Table of Contents U LRICH MUELLER 481 decreasing rate. An overall gain may still be possible, but the margins get smaller and smaller, approaching zero at some finite number of children. Operationalizing this more relaxed version of our general hypothesis is equivalent to the prediction that, in deterministic fitness, number of grandchildren obtained per child is a decreasing function of F1, the number of children. Alternatively, in stochastic fitness, the expected number of generations to extinction gained per child is again predicted to be a decreasing function of F1. We end this section with a note of caution. Stochastic fitness measures, which try to capture cumulative lineage extinction probabilities over several generations, inevitably are more arbitrarily defined than the obvious measure of number of grandchildren (or great-grandchildren, or even great-great-grandchildren). The stochastic fitness measure chosen here, which may be mathematically sounder than some alternatives, may also easily yield very high numbers. Note that we have not considered any stochastic fluctuations in childlessness rates nor any environmental stochasticity. Every step toward more realistic models inevitably will considerably lower the expected number of generations to lineage extinction. However, as long as we cannot demonstrate a relation between (a) stochastic fluctuations in childlessness rates among descendants or any future environmental stochasticity and (b) family size in the first generation, we need not worry about these high numbers. Their true meaning lies in their quantitative relationship to each other. That is, do they monotonically go up (linearly, exponentially, logarithmically?), or do they first go up, and then go down as number of children increases?

Predictions

The predictions with regard to the strict and the relaxed versions of the general hypothesis, using a deterministic and a stochastic fitness measure, are as follows: 1. Average size of F1 a) maximizes F2 (and F3 and F4); b) maximizes the expected number of generations to lineage extinction. We make no prediction of the complete distribution of F2 with regard to F1, nor do we need one. From our general assumption—that total parental investment, insofar as it has fertility-enhancing effects in children, may grow infinitely but less than proportionally with total number of children—we may predict: 2. Marginal fitness gain per child decreases monotonically from the first child onward, which means: a) grandchildren per child (as well as great-grandchildren per child; great-great-grandchildren per child) is a decreasing function of F1; b) the expected number of generations to lineage extinction per child is a decreasing function of F1.

Click to return to Table of Contents 482 S TABILIZING SELECTION IN MODERN HUMAN POPULATIONS

Results

First, we must demonstrate the independence of the proportion childless in F1 from the size of F1, that is, number of children in the first generation, essential for our modeling as in equation (2). In four of our five datasets no significant relationship of this sort could be found (see Table 1). Among the West Point sample, there is a positive relationship caused by the very low proportion of childlessness among the West Point graduates themselves, irrespective of their own birth-order position among their siblings. Once we consider their siblings only, the positive relationship between the size of F1 and the proportion childless in F1 disappears. Thus, we can assume that childlessness in F1 is indeed independent of the size of F1, and our modeling as in equation (2) is feasible. Also, we found no relationship between number of children and children’s sex ratio at birth. Thus, the asexual reproduction model as in equations (1) – (4) is admissible. The data were also scrutinized for any relationship between infant mortality in F2 and the size of F1. Only for the royalty sample could infant mortality among grandchildren be measured: no such relationship was found. For all other samples, only infant mortality among the children of the target individuals (one of the children in F1) could be measured: no relationship was found for the Hessian and Saxonian physicians sample or the West Point sample, but a negative relationship (r = –.170; p = .046) was found for the sample of noncommissioned officers—meaning that the more children a family had, the lower was infant mortality among them. Again, no explanation could be found for this observation, and any such relationship would be incompatible with stabilizing selection theory. Second, the number of grandchildren and the expected number of generations to lineage extinction, as well as these parameters per child, by parents’ number of children, are displayed in Table 2 and Figure 3. Table 2

TABLE 1 Association between number of siblings—children of parents—and proportion childless among them (ordinary least square (OLS) beta values; for European royalty partial correlation coefficient, with cohort controlled) OLS beta values / partial correlation coefficient

US Army: Retired noncommissioned officers –.109 (p= .296) US Military Academy, West Point, class of 1950 Siblings including target person (see text) .230 (p= .0001) Siblings of target person only .053 (p= .319) Western German physicians (Hessen) .025 (p= .583) Eastern German physicians (Saxony) .042 (p= .352) European royalty, birth cohorts 1790–1939 .039 (p= .536)

NOTE: Significance level in parentheses.

Click to return to Table of Contents U LRICH MUELLER 483

FIGURE 3 Relationship between number of children of parents and (a) number of grandchildren and (b) number of grandchildren per child in five populations

US Army noncommissioned officers 50

■ 40

■

■ ■ 30

■ ■ ■ ■

■ ■ ■ 20 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 10 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ Number of grandchildren ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 0 ■ ■ ■ ■

–10 024681012 Number of children

5 ■ ■ ■ ■

■ ■ ■ ■ ■ ■ 4 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 3 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 2 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 1 ■ ■ ■ ■ ■ ■ ■ ■ ■

Number of grandchildren per child ■ ■ ■ ■ ■ ■ 0 ■ ■ ■ ■

–1 024681012 Number of children

Click to return to Table of Contents 484 S TABILIZING SELECTION IN MODERN HUMAN POPULATIONS

FIGURE 3 (continued)

US Military Academy, West Point, class of 1950 40 ■

■ 30 ■

■

■ ■ ■ ■ ■ ■ 20 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 10 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ Number of grandchildren ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 0 ■ ■

–10 02 46 8 10 Number of children

8 ■

■

6 ■ ■ ■

■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 4 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 2 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■

Number of grandchildren per child 0 ■ ■

–2 02 46 8 10 Number of children

Click to return to Table of Contents U LRICH MUELLER 485

FIGURE 3 (continued)

Western German physicians (Hessen)

60 ■

■

30 ■ ■

■ ■ ■ ■ ■ ■ ■ 20 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■

Number of grandchildren 10 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 0 ■ ■ ■ ■

–10 02 46 8 10 Number of children

■

■ 6

■ ■

■

■ ■ ■ ■ 4 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 2 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ Number of grandchildren per child 0 ■ ■ ■ ■

–2 02 46 8 10 Number of children

Click to return to Table of Contents 486 S TABILIZING SELECTION IN MODERN HUMAN POPULATIONS

FIGURE 3 (continued)

Eastern German physicians (Saxony)

20 ■

■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 10 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■■■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 0 ■ ■ ■ ■ ■ Number of grandchildren

–10 0 135624 78 Number of children

5 ■

■

4 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 3 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 2 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 1 ■ ■ ■ ■ ■ ■ Number of grandchildren per child ■ ■ ■ ■ 0 ■ ■ ■ ■ ■

–1 0 135624 78 Number of children

Click to return to Table of Contents U LRICH MUELLER 487

FIGURE 3 (continued)

European royalty: Birth cohorts 1790–1939

■

■ ■ 40 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 30 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 20 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 10 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ Number of grandchildren ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 0 ■ ■ ■ ■ ■ ■ ■

–10 0 10 20 30 Number of children

12 ■

■ 10 ■ 8

■ ■ 6 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 4 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 2 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ Number of grandchildren per child ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 0 ■ ■ ■ ■ ■ ■ ■ ■

–2 0 10 20 30 Number of children

Click to return to Table of Contents 488 S TABILIZING SELECTION IN MODERN HUMAN POPULATIONS

TABLE 2 Indicators of reproductive fitness by number of children Expected number Mean number Expected number of generations to Number of Mean number of grandchildren of generations to lineage extinction children of grandchildren per child lineage extinction per child

US Army: Noncommissioned officers—parents 1 (n=14) 2.43 2.43 8.5E+04 8.5E+04 2 (n=16) 4.19 2.09 2.1E+07 1.0E+07 3 (n=37) 6.08 2.03 1.5E+08 5.0E+07 4 (n=27) 8.11 2.03 6.7E+14 1.7E+14 5 (n=25) 10.04 2.01 8.8E+12 1.8E+12 6 (n=14) 10.07 1.68 3.9E+11 6.4E+10 7 (n=12) 10.08 1.44 9.6E+07 1.4E+07 8 (n=10) 18.90 2.36 5.0E+16 6.2E+15 9 (n=6) 18.83 2.09 2.5E+13 2.7E+09 10 (n=1) 32.00 3.20 1.0E+17 1.0E+16 11 (n=1) 42.00 42.00 1.2E+12 1.1E+12 US Military Academy, West Point, class of 1950—parents 1 (n=73) 3.06 3.05 3.6E+09 3.6E+09 2 (n=148) 5.37 2.68 8.8E+11 4.4E+11 3 (n=112) 7.89 2.63 9.2E+12 3.1E+12 4 (n=60) 10.37 2.59 5.2E+15 1.3E+15 5 (n=16) 12.38 2.48 7.1E+15 1.4E+15 6 (n=11) 12.64 2.10 1.8E+15 3.0E+14 7 (n=4) 13.50 1.93 3.2E+09 4.6E+08 8 (n=5) 24.40 3.05 6.1E+16 7.6E+15 9 (n=5) 19.80 2.20 2.0E+16 2.2E+15 Western German physicians (Hessen)—parents 1 (n=68) 1.88 1.88 9.4E+04 9.4E+04 2 (n=129) 3.75 1.88 1.5E+07 7.6E+06 3 (n=108) 5.63 1.88 2.1E+13 6.9E+12 4 (n=76) 7.34 1.84 1.6E+11 3.9E+10 5 (n=41) 11.10 2.22 7.3E+19 1.5E+19 6 (n=22) 11.68 1.95 9.1E+19 1.5E+19 7 (n=25) 15.76 2.25 3.2E+20 4.6E+19 8 (n=7) 15.00 1.88 1.4E+20 1.8E+19 9 (n=4) 33.00 3.67 1.0E+21 1.1E+20 Eastern German physicians (Saxony)—parents 1 (n=131) 1.75 1.75 3.0E+03 3.0E+03 2 (n=155) 3.48 1.74 1.9E+07 9.7E+06 3 (n=102) 5.45 1.82 2.7E+10 9.1E+09 4 (n=53) 6.83 1.71 1.1E+11 2.7E+10 5 (n=32) 9.69 1.94 1.3E+14 2.6E+13 6 (n=17) 12.82 2.14 5.9E+19 9.8E+18 7 (n=3) 8.00 1.14 3.9E+08 5.5E+07

Click to return to Table of Contents U LRICH MUELLER 489

TABLE 2 (continued) Expected number Mean number Expected number of generations to Number of Mean number of grandchildren of generations to lineage extinction children of grandchildren per child lineage extinction per child

European royalty: Birth cohorts 1790–1939 1 (n=49) 2.27 2.27 4.2E+09 4.2E+09 2 (n=72) 3.79 1.90 1.1E+12 5.4E+11 3 (n=64) 5.67 1.89 5.3E+09 1.8E+09 4 (n=59) 6.81 1.70 3.7E+09 9.2E+08 5 (n=52) 8.52 1.70 2.1E+12 4.3E+11 6 (n=42) 13.29 2.21 6.8E+16 1.1E+16 7 (n=21) 12.38 1.77 5.4E+08 7.7E+07 8 (n=15) 13.60 1.70 7.8E+11 9.8E+10 9 (n=11) 21.45 2.38 9.2E+16 1.0E+16 10 (n=5) 23.80 2.38 4.9E+11 4.0E+12 11 (n=6) 24.33 2.21 3.3E+17 3.0E+16 12 (n=5) 25.80 2.15 2.0E+17 1.6E+16 13 (n=3) 21.00 1.62 4.5E+08 3.4E+07 14 (n=3) 28.67 2.05 3.3E+17 2.4E+16 15 (n=1) 11.00 0.73 1.0E+03 6.7E+01

… 24 (n=1) 25.00 1.04 6.4E+04 2.6E+03 indicates, for example, that average number of grandchildren among parents of noncommissioned officers with two children is 4.19 (grandchildren per child: 2.09); the expected number of generations to lineage extinction as calculated by equation (5) is 2.1E+07 or 21,000,000 (or calculated per child: 1.0E+07). Figure 3, as a scattergram, depicts the same association by giving the distribution of (a) number of grandchildren and (b) number of grandchildren per child, for number of children in ascending order. A straight line is fitted by the OLS method through these distributions, marking the tendency of their association with an increasing number of children. Table 3 indicates effects of number of children on fitness, estimated by OLS regression. For the first four samples standardized beta values are shown. Because of the cohort effect mentioned in our discussion of the sample of European royalty, partial correlation coefficients with cohort controlled are shown for this sample. In other words, Table 3 gives the strength and non- randomness of the statistical associations between number of children of parents as the independent variable, and various indicators of total and marginal fitness, measured in number of grandchildren and expected number of generations to lineage extinction, as the dependent variable. The inverse relationship between number of children and number of grandchildren per child in the West Point sample is the result of the above-

Click to return to Table of Contents 490 S TABILIZING SELECTION IN MODERN HUMAN POPULATIONS

TABLE 3 Association between number of siblings—children of parents—and the four fitness indicators (OLS beta values; for European royalty partial correlation coefficients, with cohort controlled) Expected Expected number of number of generations generations to lineage F2 per to lineage extinction F2 child extinction per child

US Army: Noncommissioned +.659 –.036 +.351 +.243 officers (p=.0004) (p=.646) (p=.0004) (p=.0004) US Military Academy, West Point, +.713 –.130 +.352 +.283 class of 1950 (p=.0004) (p=.007) (p=.0004) (p=.0004) Western German physicians +.774 +.125 +.427 +.392 (Hessen) (p=.0004) (p=.006) (p=.0004) (p=.0004) Eastern German physicians +.751 +.047 +.124 +.125 (Saxony) (p=.0004) (p=.299) (p=.006) (p=.006) European royalty: Birth cohorts +.652 –.049 +.232 +.183 1790–1939 (p=.0004) (p=.747) (p=.0004) (p=.0004)

average family size of the West Point graduates themselves, in comparison to the family size of their siblings; when only their siblings are considered, there is no negative relationship between parents’ number of children and number of grandchildren per child, as is also true in the other samples. We encountered a similar peculiarity above with regard to the low childlessness among the West Point graduates (Table 1). We could find no explanation for the slight positive relationship between the number of children and the number of grandchildren per child in the Hessian physicians sample; such a relationship would in any case be incompatible with the predictions of a stabilizing selection model. Third, size of F3 (great-grandchildren) and F4 (great-great-grandchildren) were regressed on size of F1 in the European royalty dataset. Simi- larly, number of F3 and number of F3 per child were regressed on size of F1. Again, cohort was controlled (see Table 4 and Figures 4 and 5). Table 4

TABLE 4 Association between number of siblings—children of parents—and number of great-grandchildren (F3) and number of great-great-grandchildren (F4) (OLS beta values; for European royalty partial correlation coefficient, with cohort controlled) F3 F4 F3 per child F4 per child European royalty: +.639 +.006 +.568 –.012 Birth cohorts 1790–1939 (p=.0004) (p=.972) (p=.0004) (p=.891)

Click to return to Table of Contents U LRICH MUELLER 491

FIGURE 4 Relationship between number of children of parents and (a) number of great-grandchildren and (b) number of great-great-grandchildren in the European royalty

European royalty: Birth cohorts 1790–1939 50

■ ■

40 ■ ■

■ ■

■ ■ ■ ■ 20 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ Number of great-grandchildren ■ ■ ■ ■ ■ ■ ■ 10 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 0 ■ 0123456 Number of children

■ ■ ■

■ 60 ■

■ ■ ■ ■ ■ 40 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 20 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 0 ■ ■ Number of great-great-grandchildren

–20 0123456 Number of children

Click to return to Table of Contents 492 S TABILIZING SELECTION IN MODERN HUMAN POPULATIONS

FIGURE 5 Relationship between (a) number of grandchildren of parents and number of great-grandchildren and between (b) number of great-grandchildren and number of great-great-grandchildren in the European royalty

European royalty: Birth cohorts 1790–1939 50

■ ■

40 ■ ■

■ ■

■ ■ ■ 20 ■ ■ ■ ■ ■ ■ ■ ■ ■ Number of great-grandchildren ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 10 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 0 ■ 024681012 Number of grandchildren

■ ■ ■

■ 60 ■

■ ■ ■ ■ ■ 40 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 20 ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ 0 ■ ■ ■ ■ ■ Number of great-great-grandchildren

–20 01020304050 Number of great-grandchildren

Click to return to Table of Contents U LRICH MUELLER 493 gives the strength and non-randomness of the statistical associations between number of children of parents as the independent variable, and two more indicators of total and marginal fitness, measured in number of great- grandchildren and number of great-great-grandchildren among the Euro- pean royalty. Figure 4, analogous to Figure 3, depicts the association between the distribution of a) number of great-grandchildren and b) number of great- great-grandchildren for number of children in ascending order. Figure 5 depicts the association between the distribution of a) number of great-grandchildren for number of grandchildren, and b) number of great-great-grandchildren for number of great-grandchildren in ascending order. Again, a straight line is fitted by the OLS method through these distributions, marking the tendency of their association with an increasing number of children, grandchildren, or great- grandchildren, respectively. The results all point in one direction: 1. There is a monotonic positive effect of number of children on fitness: a) the more children, the more grandchildren (shown in Tables 2 and 3; Figure 3), the more great-grandchildren, and the more great-great-grandchildren (shown in Table 4; Figures 4 and 5); b) the more children, the greater the expected number of generations to lineage extinction (shown in Tables 2 and 3). 2. There is no decreasing marginal fitness gain by number of children: a) increasing the number of children leaves number of grandchildren per child (and great-grandchildren per child; great-great-grandchildren per child) as well as their mortality unaffected; it does not entail a decreasing number of grandchildren and later descendants (shown in Tables 2, 3, and 4 and Figure 3), nor an increased infant mortality risk per child; b) increasing the number of children comes with an increasing, not a decreasing expected number of generations to extinction per child (shown in Tables 2 and 3), which we may cautiously see as indicating a marginal stochastic fitness that definitely is not inversely related to family size.

Discussion

In each of our five datasets, we found, over the full range of parents’ fertility, a monotonic positive association with long-term fitness as measured by total offspring fertility and lineage survival: the more children, the more grandchildren; and the more children, the higher the lineage survival probability. No observed intermediate number of children is optimal for long-term fitness. It is possible, furthermore, that children have the upper hand in the unavoidable parent–offspring conflict, decreasing parents’ fertility. Also, there may be unobserved heterogeneity in these populations. Thus, it might be argued that the optimal number of children is higher than the observed average, or that there may be different values for optimal number of children in different subpopulations.

Click to return to Table of Contents 494 S TABILIZING SELECTION IN MODERN HUMAN POPULATIONS

The point, in any event, is not that the observed mean fertility level may not be optimal, as measured in reproductive fitness. Rather, the observations provide no hint of any stabilizing selection and, therefore, of any intermediate optimum. Even as seen from the perspective of a family of twice or thrice the size of the population average, additional children mean additional fitness gains at a nondecreasing marginal rate. So any speculation about some higher-than- average intermediate optimum number of children is unfounded. The findings from our five samples leave no other conclusion than that the maximal number of children is the optimal number of children for reproductive fitness. If there is a number of children beyond which long-term reproductive fitness of parents decreases, it must be in a range that only a tiny fraction of all individuals in a population ever approach in their reproductive biographies. Our findings parallel the observations of Kaplan et al. (1995a, 1995b) in their sample of Albuquerque men and of Murphy and Knudsen (2000) in their Danish register data (although the Danish study uses only deterministic fitness measures). Four of our five samples are from high-status populations. This need not be a disadvantage for our purposes. Because the demographic transition originated in the upper classes, those classes would be the most likely place to find traces of stabilizing selection, if this were the appropriate mechanism for explaining the transition. The two groups of German physicians and the West Point graduates for most of their lives would have enjoyed a higher and more secure standard of living than the population at large, but their parents’, their siblings’, and their own resources for parental investment would still have been limited; and if there were any tradeoff between number of children and total fitness, it would have emerged in our data. For the European royalty with their virtually un- limited material resources, a special situation exists. But then, this poses the question with even greater force, namely why the majority of these individuals, who could have had a much greater number of children than they had, preferred not to. As a consequence of their decision, as was also the case in other populations, descendants of individuals with average fertility were being replaced by the descendants of small numbers of large families. The size of the fitness differentials—particularly evident in the stochastic measures—makes it probable that even in today’s societies the majority of people may be the descendants not of 30–60 percent, but perhaps of just 1–3 percent of the people living ten generations earlier. Usually a proportion much higher than, say, 1–3 percent of past populations is thought to have contributed to the present gene pool (Wachter 1980). Since our data show that parental fertility is not associated with individual offspring fertility, one might argue that the initial effect of having a large family ten generations ago would be quite diluted. To show that this argument is wrong, imagine a stationary one-sex population with 20 percent mortality before reproduction, 93.75 percent of the survivors having one child, and 6.25 percent having five children. Assume that parental fertility is associated neither with

Click to return to Table of Contents U LRICH MUELLER 495 offspring fertility nor with offspring mortality and that the population is large enough so that random extinction of the whole population will not occur in any arbitrarily chosen finite observation time. The average number of generations to lineage extinction calculated for a randomly chosen survivor before onset of own reproduction would be 5.33; for a lucky survivor with a completed family of five it would be 4,312.82 generations, for a survivor lucky enough to have five children and to have one of them bear five grandchildren, the number would be 654,062.49. Thus, from any point t0 in time onward, the stationary population will increasingly be composed of descendants of the few individuals with five children—and ultimately of descendants of just one of them—living at time t0. In a real population, a substantial part of any fitness differentials within a given generation is at least slightly associated with the distribution of some alleles—that is, some heritable traits—in that population. The enormous fitness advantages obtained by these few individuals with large families, which we observed, suggest that evolution by natural selection and not just by random drift may also in modern large populations be as fast and as vigorous as ever in the history of mankind.2 One might object that high-fertility minority populations have rarely (or never) come to dominate the majority populations in which they live. Catho- lics in the United States, for example, have always had higher fertility than non-Catholics, but over many decades the proportion of US Catholics has barely risen. But this objection is not valid. What really counts in the evolutionary competition between groups with partial endogamy is the prevalence of very large families and the distribution of lineage extinction probabilities within these minorities as compared with the majority population. High-fertility minorities typically are low-status groups. It is imaginable—and would be compatible with a higher mean fertility—that, among them, very large families may be even rarer, and lineage extinction probabilities may be generally higher than in the majority population. At first glance, our conclusions also seem to contradict findings from evolutionary genetics that (a) the most recent common ancestor of the Y-chromosome (“Adam”), which is transmitted exclusively by fathers to sons, lived 2,000–4,000 generations ago (Gibbons 1997; Underhill et al. 2000) and (b) the most recent common ancestor of human mitochondrial DNA (“Eve”), which is transmitted exclusively by mothers, lived 4,000–8,000 generations ago (Ayala 1995). Our findings would be consistent with an even lower number of generations back to the most recent common ancestor—the so-called coales- cence time. An explanation of this inconsistency could lie in the spatial, linguistic (Cavalli-Sforza, Menozzi, and Piazza 1994), and social segregation between marriage markets, characteristic of all human populations. Different lineages can coexist for long time periods, whereas when people live and breed together, chances are that even in large populations, seen from any point in time, the descendants of just a few lineages—and ultimately one—will replace the de-

Click to return to Table of Contents 496 S TABILIZING SELECTION IN MODERN HUMAN POPULATIONS scendants of all others. Such a dynamic of segregated and then fused marriage markets might explain why the vast majority of all presently living Europeans may be descendants of ten paternal lineages, which separated from each other 1,000–1,600 generations ago (Semino et al. 2000). If we want to understand the demographic dynamics of our societies, we should devote much more attention to the study of large families, rare as they are. It is well known that they differ in many potentially important features from the population average: in political attitudes, in lifetime number of sex partners, in religiosity, in urbanity, in educational characteristics. If studied more closely, additional differences from average-sized families may surface. Large families are also potentially very important subjects of genetic studies.

Notes

The author thanks Juha M. Ahlo, Noël havioral sciences in Low (2000), Mealey Bonneuil, Adam Lomnicki, and Jose A. O. (2000), and Ridley (1996). Osana for helpful comments. 2 For a review of the controversy over se- 1 The reader is referred to the excellent lection versus random drift see Ridley (1996: glossaries of evolutionary concepts for the be- ch. 7).

References

Atkinson, R. 1989. The Long Gray Line. Boston: Houghton Mifflin. Ayala, F. J. 1995. “The myth of Eve: Molecular biology and human origins,” Science 22: 1930–1936. Blurton Jones, N. G. 1989. “The costs of children and the adaptive scheduling of births: Towards a socio-biological perspective on demography,” in A. E. Rasa, C. Vogel, and E. Voland (eds.), The Sociobiology of Sexual and Reproductive Strategies. London and New York: Chapman & Hall. Borgerhoff Mulder, M. 1998. “Demographic transition: Are we any closer to an evolutionary explanation?” Trends in Ecology and Evolution 13: 266–270. ———. 2001. “Optimizing offspring: The quantity–quality tradeoff in agropastoral Kipsigis,” Evolution and Human Behavior (in press). Boyd, R. and P. J. Richerson. 1985. Culture and the Evolutionary Process. Chicago: University of Chicago Press. Carr-Saunders, A. M. 1922. The Population Problem: A Study in Human Evolution. Oxford: Clarendon Press. Cavalli-Sforza, L. L., P. Menozzi, and A. Piazza. 1994. The History and Geography of Human Genes. Princeton: Princeton University Press. Cohen, J. E. 1975. “Childhood mortality, family size and birth order in pre-industrial Eu- rope,” Demography 12: 35–55. Daly, M. and M. Wilson. 1983. Sex, Evolution and Behavior. 2nd ed. Boston: PWS Publishers. Einum, S. and I. A. Fleming. 2000. “Highly fecund mothers sacrifice offspring survival to maximize fitness,” Nature 405: 565–567. Essock-Vitale, S. M. and M. T. McGuire. 1985a. “Women’s lives viewed from an evolutionary perspective. I. Sexual histories, reproductive success, and demographic characteristics of a random sample of American women,” Ethology and Sociobiology 6: 137–154. ———. 1985b. “Women’s lives viewed from an evolutionary perspective. II. Patterns of helping,” Ethology and Sociobiology 6: 155–173.

Click to return to Table of Contents U LRICH MUELLER 497

Finke, M. A., D. J. Milinkovick, and C. F. Thompson. 1987. “Evolution of clutch size: An experimental test in the house wren (Troglodytes aedon),” Journal of Animal Ecology 56: 99–114. Fleming, T. 1981. The Officers’ Wives. New York: Warner Books. Gibbons, A. 1997. “Y chromosome shows that Adam was African,” Science 278: 804–805. Halley, J. M. and Y. Iwasa. 1998. “Extinction rate of a population under both demographic and environmental stochasticity,” Theoretical Population Biology 53: 1–15. Hill, K. and A. M. Hurtado. 1996. Aché Life History: The Ecology and Demography of a Foraging People. New York: Aldine de Gruyter. Hobcraft, J., J. W. McDonald, and S. Rutstein. 1983. “Child-spacing effects on infant and early child mortality,” Population Index 49: 585–618. Howell, N. 1979. Demography of the Dobe !Kung. New York: Academic Press. Kaplan, H. 1996. “A theory of fertility and parental investment in traditional and modern human societies,” Yearbook of Physical Anthropology 39: 91–135. Kaplan, H. et al. 1995a. “Fertility and fitness among Albuquerque men: A competitive labor market theory,” in R. I. M. Dunbar (ed.), Human Reproductive Decisions. New York: St. Martin’s Press. ———. 1995b. “Does observed fertility maximize fitness among New Mexican men?” Hu- man Nature 6: 325–360. Klomp, H. 1970. “The determination of clutch size in birds: A review,” Ardea 58: 1–124. Lack, D. 1947. “The significance of clutch size,” Ibis 89: 302–352. ———. 1948. “The significance of litter size,” Journal of Animal Ecology 17: 45–50. ———. 1954. Natural Regulation of Animal Numbers. Oxford: Clarendon Press. ———. 1966. Population Studies of Birds. Oxford: Oxford University Press. ———. 1967. The Natural Regulation of Animal Numbers. Oxford: Oxford University Press. ———. 1968. Ecological Adaptations of Breeding in Birds. London: Methuen. Lancaster, J. B. and H. S. Kaplan. 1991. “Male fertility and parenting: Tests of models with a sample of Albuquerque men,” paper presented at the 3rd Annual Meeting of the Human Behavior and Evolution Society, 22–25 August, Hamilton, Ontario. Lopreato, J. and M. Yu. 1988. “Human fertility and fitness optimization,” Ethology and Socio- biology 9: 269–289. Low, B. S. 1991. “Reproductive life in nineteenth century Sweden: An evolutionary perspective on demographic phenomena,” Ethology and Sociobiology 12: 411–448. ———. 2000. Why Sex Matters: A Darwinian Look at Human Behavior. Princeton: Princeton University Press. Ludwig, D. 1996. “The distribution of population survival times,” American Naturalist 147: 506–526. Mace, R. 1996. “Biased parental investment and reproductive success in Gabbra pastoralists,” Behavior Ecology and Sociobiology 38: 75–81. ———. 1998. “The co-evolution of human fertility and wealth inheritance,” Philosophical Transactions of the Royal Society of London, Series B 353: 389–397. ———. 2000. “An adaptive model of human reproductive rate: Why people have small families,” in L. Cronk, W. Irons, and N. Chagnon (eds.), Human Behavior and Adapta- tion: An Anthropological Perspective. New York: Aldine. McNaughton, A. 1973. The Book of Kings: A Royal Genealogy. 3 vols. London: Garnstone Press. Mealey, L. J. 2000. Sex Differences: Developmental and Evolutionary Strategies. San Diego: Aca- demic Press. Mueller, U. 1991. “Differential reproduction of the rich and powerful reconsidered: The case of Germany, Great Britain, Japan, and the USA during the 19th and 20th century,” ZUMA working paper 14/91. ———. 1993. “Social status and sex,” Nature 363: 490. ———. 2001a. “Stichprobe, Instrument und Variablen der ZUMA Aerzteerhebungen in Hessen und Sachsen” (Sample, instrument, and variables of the ZUMA physicians survey in Hessen and Saxony), WP 01/4. Center for Methodology and Health Research. Medical School. Philipps University Marburg.

Click to return to Table of Contents 498 S TABILIZING SELECTION IN MODERN HUMAN POPULATIONS

———. 2001b. “Testing group selection theory with applications to demography: Selection, but no group selection in the closed marriage market of European royalty 1790–1939,” WP 01/1. Center for Methodology and Health Research. Medical School. Philipps Uni- versity Marburg. Mueller, U. and A. Mazur. 1998. “Reproductive constraints on dominance competition in male Homo sapiens,” Evolution and Human Behavior 19: 387–396. Murphy, M. and L. B. Knudsen. 2000. “The relationship of fathers’, mothers’, brothers’ and sisters’ fertility histories with the onset of childbearing in Denmark,” paper presented at the Annual Meeting of the Population Association of America, Los Angeles, March. Nur, N. 1984. “The consequences of broodsize for breeding blue tits. II. Nestling weight, offspring survival and optimal brood size,” Journal of Animal Ecology 53: 497–517. Ridley, M. 1996. Evolution. 2nd Edition. Cambridge, MA: Blackwell Science. Roff, D. A. 1992. The Evolution of Life Histories. London: Chapman & Hall. Ryden, H. 1986. The Beaver. New York: Lyons & Bunford. Semino, O. et al. 2000. “The genetic legacy of paleolithic Homo sapiens sapiens in extant Europeans: A Y chromosome perspective,” Science 290: 1155–1159. Sinervo, B. and E. Svensson. 1998. “Mechanistic and selective causes of life history tradeoffs and plasticity,” Oikos 83: 432–442. Stearns, S. C. 1992. The Evolution of Life Histories. Oxford: Oxford University Press. Tuljapurkar, S. D. 1990. Population Dynamics in Variable Environments. New York: Springer. Underhill, P. A. et al. 2000. “Y chromosome sequence variation and the history of human populations,” Nature Genetics 26: 358–361. Vernier, B. 1977. “Emigration et dereglement du marche matrimonial,” Actes Recherches Sci- ences Sociales 1977: 31–58. Vining, D. R. 1986. “Social versus reproductive success: The central problem of human sociobiology,” The Behavioral and Brain Sciences 9: 167–216. Voland, E. and R. I. M. Dunbar. 1995. “Resource competition and reproduction—The relationship between economic and parental strategies in the Krummhörn population (1720–1874),” Human Nature 6: 33–49. Voland, E., R. I. M. Dunbar, C. Engel, and P. Stephan. 1997. “Population increase and sex- biased parental investment: Evidence from 18th and 19th century Germany,” Current Anthropology 38: 129–135. Voland, E. and C. Engel. 1989. “Women’s reproduction and longevity in a premodern population (Ostfriesland, Germany, 18th century),” in A. E. Rasa, C. Vogel, and E. Voland (eds.), The Sociobiology of Sexual and Reproductive Strategies. London and New York: Chapman & Hall. Wachter, K. 1980. “Ancestors at the Norman Conquest,” in B. Dyke and W. T. Morrill (eds.), Genealogical Demography. New York: Academic Press, pp. 85–93. Wrigley, E. A. and R. S. Schofield. 1981. The Population History of England 1541–1871: A Re- construction. London: Edward Arnold. Yoshimura, J. and C. W. Clark. 1991. “Individual adaptations in stochastic environments,” Evolutionary Ecology 5: 173–192. ———. 1993. Adaptation in Stochastic Environments. Berlin: Springer.

Click here to print article Click to return to Table of Contents Click to return to Table of Contents The Age of Migration in China

ZAI LIANG

IN THE PAST half-century the People’s Republic of China has experienced a number of noteworthy demographic changes. During the 1950s and 1960s life expectancy at birth substantially increased. The 1970s and early 1980s were marked by a sharp decline in fertility. The late 1980s and the 1990s may be characterized as a period of great migration. Today in every major Chinese city migrants are found in many occupations—as taxi drivers, construction workers, restaurant workers, and petty traders in free-market enterprises. Migrants are changing the face of urban China. For a long time, research on migration and urbanization in China was hampered by a lack of data. Questions on migration were not included in the first three population censuses of the People’s Republic: in 1952, 1964, and 1982 (Goldstein and Goldstein 1990; Liang and White 1996). Since 1987, China has conducted several national surveys and censuses that request information on migration. Both the 1987 and 1995 China One Percent Popu- lation Sample Surveys included questions on migration, as did the 1988 China 2 per 1,000 Fertility and Birth Control Survey. These data sets allow researchers to analyze Chinese migration patterns using measures that are widely employed in the study of migration in other countries. Data from the various surveys, when used together, can illustrate the changes that occurred during 1982–95, a period when China was moving steadily toward a market-oriented economy. This article uses data from the 1987 and 1995 China One Percent Popu- lation Sample Surveys to describe the patterns of migration during 1982– 95. Evidence presented highlights the increasing importance of temporary migration since the late 1980s, as a direct response to the market reforms initiated in the late 1970s. Major patterns emerge in temporary migration; short-distance versus long-distance migration; migration flows among cities, towns, and rural areas; migrant destination choices; and reasons for migration. Finally, the effectiveness of the Chinese government’s policy promoting development of small towns—referred to as ”urbanization from below”—is assessed.

POPULATION AND DEVELOPMENT REVIEW 27(3):499–524 (SEPTEMBER 2001) 499

Click here to print article Click to return to Table of Contents 500 T HE AGE OF MIGRATION IN CHINA

Market reforms and the erosion of China’s household registration system

Since its implementation in the late 1950s, China’s household registration system (hukou) has been a major determinant of an individual’s life chances in China (Cheng and Selden 1994; Wang 1997). Hukou governs where one can live and the public benefits to which one is entitled. Until recently, an individual without an urban household registration status could not buy food or get a job in a city. Any individual who wanted to migrate had to obtain permission from local authorities in the place of origin and the place of destination. Under such conditions, it was difficult for an unauthorized migrant to survive in a city. As a result, China’s household registration system was a major deterrent to rural-to-urban migration. Even before Hong Kong’s return to China in 1997, the expression “one country and two systems” was apt (Wang 1997). An urban household registration status is highly desirable because of the benefits and subsidies it grants (e.g., related to housing, medical services, and access to foods that are difficult to obtain). Rural areas lack such amenities. Similar urban biases are found in other developing countries (Lipton 1977). China’s household registration system was very effective in stemming the tide of migration from the countryside and keeping the country’s urbanization level relatively low. Figure 1 shows the level and rate of urbanization in China during 1951–95. The history of urbanization in the country can be divided into four stages. The first stage (1951–60) was a period of extensive rural-to-urban migration. This period coincides with the first Five- Year Plan (1953–57) and the Great Leap Forward that began in 1958. Fol- lowing the Soviet model, a disproportionately large investment in heavy industries was made during this time. Peasants rushed to cities to work in booming industries. As a consequence, 20 percent of the population lived in cities by 1960. During much of this period a laissez-faire migration policy facilitated the movement of millions of people into urban centers. The second period (1961–65) was one of sharp urban retrenchment. By this time, China’s leaders had realized that the country’s agricultural output would be unable to supply enough grain to sustain the urban population even at the level of 20 percent of the total (130 million in 1960). To reduce the urban population, they sent some 24 million workers, many of whom had originally migrated from rural areas, to the countryside (Chan 1994). In the third period (1966–77) urbanization levels declined slightly as the rustication movement sent cadres, intellectuals, and young people to the countryside. The proportion of the population living in China’s urban areas stabilized at around 17 percent. The fourth period (1978–95 and apparently beyond) has been a time of substantial and sustained rural-to-urban migration and increasing urbanization. The transformation from former production teams to household

Click to return to Table of Contents Z AI LIANG 501

FIGURE 1 Urbanization in China, 1951–95

30 Percent of population urban 25

10 Rate of growth of urban population Percent 5

–5

–10

–15 1950 1960 1970 1980 1990 2000 Year SOURCES: SSB (1994, 1997).

farming (or the household production responsibility system) has greatly increased agricultural productivity, thus decreasing the need for rural laborers (Li 1996). Other significant changes have also made migration much easier than before. Grain rationing coupons were eliminated and replaced by access to the free market. At the same time, the transition toward a market economy transformed the urban industrial structure and gradually created more employment opportunities in nonstate sectors of the economy. The mushrooming of joint-venture enterprises and other privately owned businesses has created a great demand for migrant workers. By 1991, 39 percent of employees in urban areas were working for non–state-owned work units (SSB 1997). These nonstate enterprises can provide migrant workers with temporary urban resident cards. Even without such cards, migrant workers can live in cities as long as they have the means to support themselves. As a result, migration, often temporary in nature, has increased significantly since the mid-1980s (Li 1996 discusses other factors conducive to migration). This article examines the major features of migration in the 1980s and 1990s.

Data and definitions

As noted above, data used in this article are primarily from the 1987 and 1995 China One Percent Population Sample Surveys. The 1987 survey, con-

Click to return to Table of Contents 502 T HE AGE OF MIGRATION IN CHINA ducted with the reference date of 1 July, was the first national sample survey to collect information on migration.1 All individuals in the sample household were asked whether they had moved any time during the five years prior to the survey (SSB 1988). For interprovincial migrants, the provinces of origin and destination were identified; furthermore, origin and destination were grouped into three categories: city, town, and county (the last referring to rural areas). The reasons for migration were also recorded. Designed as a mini-census between the 1990 and 2000 censuses, the 1995 China One Percent Population Sample Survey was conducted in Oc- tober by the State Statistical Bureau of China (CPSSO 1997). With respect to migration, the 1995 survey collected the same information as in 1987, with one exception: information on reasons for migration was not recorded. For this question, I include data from the 1990 China Population Census. One of the unique features of migration in China is the distinction between temporary and permanent migrants. Permanent migrants are defined as migrants who have obtained permanent household registration status at their place of destination; temporary migrants (often referred to as the floating population) do not have such status.2 One can further divide temporary migrants into those who hold temporary registration cards and those who do not. This article makes no distinction between the two categories of temporary migrants. The 1987 and 1995 surveys both define temporary migrants as individuals who do not have permanent household registration status at their place of destination and who have stayed there for at least six months.3 Of course, the use of a six-month criterion to define a temporary migrant will miss migrants who stayed for a shorter period of time. But the six-month criterion captures those temporary migrants who stayed at their place of destination long enough to have had a social and economic impact. The similarity of definition of temporary migrants in the two surveys permits us to assess the extent of changes over time. One of the sources of confusion in migration studies in China is the different ways of defining and operationalizing the concept of temporary migration. For example, the 1990 Chinese census defines temporary migrants as individuals who do not have permanent household registration status at the place of destination and who had stayed at the place of destination for at least a year. Other surveys (such as the 1994 Beijing Floating Population Sample Survey) define temporary migrants simply as individuals who are present but not registered at the place of destination (Zou 1996).

The rise of temporary migration

Although scholars agree that the number of temporary migrants in China is on the rise, there is little consensus about its magnitude, with estimates varying from 50 million to 90 million. Shen and Tong (1992) used results

Click to return to Table of Contents Z AI LIANG 503 from a survey of seven cities in the late 1980s to derive a distribution for duration of residence at place of destination.4 According to their calculation, 28.7 percent of temporary migrants had stayed at their place of destination for more than a year. Shen and Tong also used 1990 Chinese census data to calculate the number of temporary migrants who had stayed at their place of destination for at least one year. They further assumed that the distribution by duration for all temporary migrants in China would follow the distribution of migrants from the seven-city survey (which presumably covers temporary migration of any duration). On the basis of these assumptions and using data from the 1990 Chinese census, Shen and Tong (1992) estimated China’s temporary migrant population to be around 70 million in 1990. Other researchers (quoted in Roberts 1997) used data from sample surveys in Beijing and Shanghai to derive percentages for the temporary migrant population in 1994 and then applied them to China’s 1990 urban population of 290 million. According to these estimates, China’s temporary migrant population in 1990 was about 70–90 million, depending on whether the percentage for Beijing or Shanghai is used. Following a similar approach, Gu and Jian (1994) estimated the number of China’s temporary migrants to be around 50–60 million in the late 1980s. These estimates rest on some potentially problematic assumptions. For example, Shen and Tong’s seven-city survey was conducted in three separate years (Guangzhou in 1990, Shanghai in 1988, and the other cities in 1989). One should note that the distribution of duration of residence for temporary migrants throughout China may not follow the distribution of duration of residence for temporary migrants derived from the seven-city survey. One can also question the extent to which these seven cities represent the several hundreds of Chinese cities. From several nationally representative surveys and censuses, Figure 2 shows the stock of the temporary migrant population in 1982, 1989, 1990, 1995, and 1996.5 In 1982, only 11 million temporary migrants were recorded. In the seven-year period from 1982 to 1989 the temporary migrant population increased by 7 million, or roughly one million per year. Then, in just one year (from 1989 to 1990), China’s temporary migrant population increased by 11 million (almost 60 percent). The momentum accelerated in the 1990s. Results from the 1995 China One Percent Sample Sur- vey show that there were 56 million temporary migrants in that year.6 In 1996, the temporary migrant population increased to 68 million. Table 1 lists the total number of temporary migrants in each province or municipality as of 1995. Beijing had almost 2 million temporary migrants and Shanghai had 1.7 million. The coastal provinces also had large numbers of temporary migrants (Jiangsu, Zhejiang, Shandong, and Guangdong). Guangdong has the largest temporary migrant population (7 million) of all provinces, accounting for nearly 13 percent of the total temporary migrant population. During the recent period of economic transition, Guangdong

Click to return to Table of Contents 504 T HE AGE OF MIGRATION IN CHINA

FIGURE 2 Estimated size of the temporary migrant population, China, 1982–96

40 Millions

0 1982 1989 1990 1995 1996

SOURCES: SSB (1991, 1997); CPSSO (1997). has been China’s flagship province for economic development: three of the first four special economic zones (Shenzhen, Zhuhai, and Shantou) are located in Guangdong. Since 1978, the province has enjoyed an economic boom associated with its proximity to Hong Kong and the influx of foreign investors (Liang 1999). Examination of China’s 1990 Population Census also showed that two of the largest interprovincial migration streams during 1985–90 headed to Guangdong (Poston and Mao 1998). The extent to which temporary migrants have affected the local economy can be seen in the proportion of the population of a province composed of temporary migrants (see third column in Table 1). Beijing’s temporary migrants accounted for about 15 percent of its total population in 1995. Shen and Tong (1992) reported that temporary migrants accounted for 13 percent of Beijing’s resident population in 1988. Not surprisingly, Shanghai had the second highest percentage, 12 percent. As the largest metropolis in China, and particularly because of development in the Pudong area (east of the Huangpu river) since the early 1990s, Shanghai provides significant opportunities for migrant workers. Ironically, Beijing and Shang- hai are two of the Chinese cities in which the government has sought to exert the tightest control seeking to moderate the influx of temporary migrants.

Click to return to Table of Contents Z AI LIANG 505

TABLE 1 Size and distribution of the temporary migrant population by province, China, 1995 Number of Percent Percent of all Total temporary of total temporary population migrants population migrants Province (000) (000) in province in China

North Beijing 12,901 1,891 14.7 3.4 Tianjin 9,704 1,033 10.6 1.9 Hebei 66,216 1,866 2.8 3.3 Shanxi 31,649 1,482 4.7 2.7 Inner Mongolia 23,492 1,746 7.4 3.1 Northeast Liaoning 42,137 2,241 5.3 4.0 Jilin 26,689 1,494 5.6 2.7 Heilongjiang 38,094 2,411 6.3 4.3 East Shanghai 14,582 1,704 11.7 3.1 Jiangsu 72,733 5,180 7.1 9.3 Zhejiang 44,467 2,323 5.2 4.2 Anhui 61,844 1,497 2.4 2.7 Fujian 33,289 1,925 5.8 3.5 Jiangxi 41,734 1,723 4.1 3.1 Shandong 89,625 3,191 3.6 5.7 Central and south Henan 93,544 1,742 1.9 3.1 Hubei 59,358 2,206 3.7 4.0 Hunan 65,865 1,892 2.9 3.4 Guangdong 70,516 7,075 10.0 12.7 Guangxi 46,684 993 2.1 1.8 Hainan 7,425 332 4.5 0.6 Southwest Sichuan 116,334 4,084 3.5 7.3 Guizhou 36,042 969 2.7 1.7 Yunnan 40,951 1,364 3.3 2.4 Tibet 2,465 70 2.8 0.1 Northwest Shannxi 36,093 1,003 2.8 1.8 Gansu 25,018 832 3.3 1.5 Qinghai 4,940 315 6.4 0.6 Ningxia 5,249 279 5.3 0.5 Xinjiang 17,057 920 5.4 1.6 China 1,236,697 55,785 4.5 100.0

SOURCES: CPSSO (1997): Table 7-1, pp. 538–539.

Click to return to Table of Contents 506 T HE AGE OF MIGRATION IN CHINA

Several of China’s coastal provinces (Liaoning, Jiangsu, Zhejiang, Fujian, Guangdong) have high proportions of temporary migrants. Guangdong again stands out: in 1995 10 percent of its total population was composed of temporary migrants. As many observers have noted, China’s coastal provinces are the most dynamic economic regions in the country (Yang 1991). The flow of the temporary migrant population follows China’s economic development strategy. In 1984, five years after the opening of four special economic zones, 14 coastal cities were designated as open development zones and granted the favor of special policies, such as tax incentives for joint-venture enterprises. Although the spatial concentration of temporary migrants in large cities and coastal provinces has long been recognized, often overlooked are the substantial numbers of temporary migrants in nearly every province of China. For example, Sichuan, China’s most populous province, had 4 million temporary migrants as of 1995. The northeast provinces of Liaoning and Heilongjiang both had over 2 million such migrants. Even some of China’s remote provinces are attracting temporary migrants. In the northwest provinces of Qinghai, Ningxia, and Xinjiang, the latter two largely Muslim, temporary migrants constituted more than 5 percent of the resident population in 1995. In percentage terms this is higher than in some coastal provinces, such as Shandong, which has a much larger absolute number of temporary migrants. The last column of Table 1 gives the share of temporary migrants for each province as a proportion of the total for the country as a whole. Again, in 1995 Guangdong had the largest share of China’s temporary migrants (13 percent) and Jiangsu ranked second (9 percent). Although the proportions of temporary migrants in Beijing and Shanghai were large relative to their resident populations, they did not account for a high proportional share of the national temporary migrant population.

Short-distance versus long-distance migration

Some idea of the distances covered by migrants can be obtained by distin- guishing migrants by origin—that is, whether their migration was interprovincial or intraprovincial, since the former is likely to have involved longer distances. In the rest of this article our analysis of migration patterns covers only migrants who moved anytime during the five-year period prior to either survey. This also means that we include both temporary and permanent migrants as they are defined in the 1987 and 1995 surveys. Table 2 gives the estimated size of the migrant population in 1987 and 1995, whether temporary or not, and the percentages within the total.7 The modest increase in the size of the migrant population (both interprovincial and intraprovincial) from 30.5 million in 1987 to 33.2 million in 1995 masks the changes that have taken place, given the large increase in the number

Click to return to Table of Contents Z AI LIANG 507

TABLE 2 Size of the migrant population and distribution of intraprovincial and interprovincial migrants by province, China, 1987 and 1995 1987 1995 Percent Percent Percent Percent Total intra- inter- Total intra- inter- Province (000) provincial provincial (000) provincial provincial

North Beijing 866 62.1 37.9 1,340 48.2 51.8 Tianjin 174 23.3 76.7 490 54.4 45.6 Hebei 1,536 61.3 38.7 1,339 62.4 37.6 Shanxi 946 82.6 17.4 478 66.9 33.1 Inner Mongolia 737 77.2 22.8 932 70.5 29.5 Northeast Liaoning 1,286 75.6 24.4 1,601 72.8 27.2 Jilin 1,093 84.5 15.5 770 80.6 19.4 Heilongjiang 1,060 82.0 18.0 1,288 82.6 17.4 East Shanghai 649 42.1 57.9 1,689 57.0 43.0 Jiangsu 1,823 73.8 26.2 3,038 68.1 31.9 Zhejiang 914 86.4 13.6 1,218 61.8 38.2 Anhui 1,022 83.8 16.2 717 78.3 21.7 Fujian 544 83.4 16.6 996 65.4 34.6 Jiangxi 632 84.1 15.9 607 79.3 20.7 Shandong 2,061 73.4 26.6 1,941 72.8 27.2 Central and south Henan 1,159 77.2 22.8 1,030 73.8 26.2 Hubei 1,890 85.5 14.5 1,070 74.7 25.3 Hunan 1,438 84.9 15.1 1,213 82.3 17.7 Guangdong 2,536 88.0 12.0 4,090 52.4 47.6 Guangxi 730 91.9 8.1 925 87.0 13.0 Hainan NA NA NA 232 55.2 44.8 Southwest Sichuan 3,677 89.4 10.6 2,367 83.3 17.0 Guizhou 667 82.9 17.1 564 73.0 27.0 Yunnan 747 85.9 14.1 836 75.3 24.7 Tibet NA NA NA 70 48.8 51.2 Northwest Shannxi 1,010 77.7 22.3 694 76.5 23.5 Gansu 469 81.3 18.7 547 74.5 25.5 Qinghai 90 68.1 31.9 188 72.7 27.3 Ningxia 182 49.8 50.2 118 58.5 41.5 Xinjiang 557 63.7 36.3 845 33.0 67.0 China 30,533 79.3 20.7 33,230 67.9 32.1

NOTE: Data for Hainan and Tibet in 1987 are not available. SOURCES: 1987 and 1995 China One Percent Population Sample Surveys.

Click to return to Table of Contents 508 T HE AGE OF MIGRATION IN CHINA of temporary migrants between 1982 and 1995. One can deduce from the data that there were many more intraprovincial than interprovincial migrants during both periods. By definition, interprovincial migration involves crossing province boundaries and often traveling long distances; therefore the cost of migration is likely to be higher than that involved in intraprovincial migration. The most notable change between the two surveys is the 68 percent increase in the numbers of interprovincial migrants, from 6.3 million in 1987 to 10.7 million in 1995. The number of intraprovincial migrants changed only slightly over the period, declining from 24.2 million to 22.6 million. For China as a whole, the number of interprovincial migrants as a proportion of all migrants increased from 21 percent in 1987 to 32 percent in 1995. This increase is partly the result of increasing interregional inequality in economic opportunities during this period. Regions with greater access to markets developed faster than other regions (Nee 1989). Indeed, on the basis of household surveys in China in 1988 and 1995, Khan and Riskin (1998) showed that interregional income inequality had increased markedly. Of the 28 provinces (which account for 99 percent of the population of China) for which we have data,8 22 provinces (or municipalities) experienced increases in the percentage share of interprovincial migrants between the two surveys.9 Among these provinces, Beijing, Shanxi, Zhejiang, Fujian, Guangdong, and Xinjiang exhibit particularly striking patterns of change. In Guangdong and Xinjiang, the proportions of interprovincial migrants as a percent of all migrants increased by 36 and 31 percentage points, respectively, as shown by comparing the 1995 survey with the survey of 1987. In the other four provinces (municipality in the case of Beijing), the corresponding increases ranged from 16 percentage points (Shanxi) to 25 percentage points (Zhejiang). That these coastal provinces saw large increases in the numbers of interprovincial migrants conforms to expectations, but the fact that more interprovincial migrants were registered in the northwest province of Xinjiang (565,800 in 1995 compared with 202,000 in earlier period) may be considered surprising. The largest numbers of migrants in Xinjiang originate from Sichuan province in the southwest. A recent report suggests that as many as 400,000 migrant workers, most of them from Sichuan province, travel to Xinjiang each year to pick cotton during the harvest season (Mer 1998). Many migrants also work in the mining industry in Xinjiang. Answers to the questions of who participated in intraprovincial and interprovincial migration and whether there were any changes between the two time periods can be given through a two-step process. First, the distribution of places of origin by type of migrants (top panel in Table 3) suggests who contributed to the intraprovincial and interprovincial migration streams and whether the patterns have changed over time. Second, the distribution of types of migrants by place of origin (bottom panel in Table 3) shows

Click to return to Table of Contents Z AI LIANG 509

TABLE 3 Percent distribution of the intraprovincial and interprovincial migrant populations by place of origin and percent distribution of the total migrant population classified by place of origin between intra- and interprovincial migrants, China, 1987 and 1995 1987 1995 Intra- Inter- Intra- Inter- provincial provincial provincial provincial Place of origin migrants migrants migrants migrants

City 13.5 35.4 36.5 19.0 Town 16.4 4.9 10.0 7.8 Rural 70.1 59.8 53.4 73.2 Total (percent) 100.0 100.0 100.0 100.0 1987 1995 Intra- Inter- Intra- Inter- provincial provincial provincial provincial Place of origin migrants migrants Total migrants migrants Total

City 59.4 40.6 100.0 80.3 19.7 100.0 Town 92.8 7.2 100.0 73.4 26.7 100.0 Rural 81.8 18.2 100.0 60.7 39.3 100.0 Total (000) 24,221 6,312 22,569 10,661

NOTE: Percent totals may not add to 100 due to rounding. SOURCES: 1987 and 1995 China One Percent Population Sample Surveys. changes in the share of type of migrants (intraprovincial vs. interprovincial) for city, town, and rural residents. Although the top panel suggests that rural-origin migrants play the most important role in both interprovincial and intraprovincial migration flows, an interesting pattern emerges over time. In 1987, 70 percent of the intraprovincial migrant population was found to be of rural origin; however, this dominance dissipated over time, with only a little more than half being of rural origin by 1995. Mi- grants of city origin were more likely to join the wave of intraprovincial migration during the early 1990s than they were during the mid-1980s. A different pattern characterizes interprovincial migration. According to the 1987 survey, interprovincial migrants disproportionately came from urban areas (cities and towns). Although the urban population was 25 percent of China’s total population in 1987, persons from urban areas (cities and towns) accounted for some 40 percent of interprovincial migrants in 1987. The high share of migrants of urban origin reflects China’s continuing rural/urban divide. The urban Chinese not only enjoy special privileges associated with urban household registration status; they also have a higher degree of freedom to move. It is usually easier for migrants of urban origin to obtain household registration status at a place of destination than for

Click to return to Table of Contents 510 T HE AGE OF MIGRATION IN CHINA migrants of rural origin (Zhang 1994: 242). Given the higher costs associated with migration across province boundaries and the fact that per capita income in urban China is much higher than in rural China, migrants of urban origin are more likely to be able to afford migration from one province to another. By 1995, nevertheless, migrants of rural origin were more heavily represented among interprovincial migrants. The decrease in the share of intraprovincial migrants of rural origin seems to be compensated in part by their greater participation in interprovincial migration (signaled by an increase from about 60 percent of the total interprovincial migrants in 1987 to 73 percent in 1995). The growing importance of rural populations in the stream of interprovincial migration is corroborated by findings in the bottom panel of Table 3. For example, in 1987 only 18 percent of migrants of rural origin were interprovincial migrants. By contrast, in 1995 the corresponding figure has more than doubled, to 39 percent. This change suggests that China’s household registration system presented less of a hurdle for rural migrants in the early 1990s than it did in the mid-1980s. Although official regulations concerning the household registration system remain in place, the authorities have found it almost impossible to strictly enforce the system. For example, it has become very easy to obtain local temporary registration cards, especially for employees of emerging urban enterprises. Migrant workers can work for joint-venture enterprises for as long as five years while in temporary resident status. Even when their temporary registration card expires, individuals can readily find new employment, thereby maintaining continuous temporary residence status. Additionally, food is widely available in the free market, and housing, if not provided by employers, can be arranged. In short, migrants can survive at their places of destination without much difficulty. The social nature of the migration process also facilitates interprovincial migration and sustains its momentum (Massey and Espinosa 1997). During the late 1980s and early 1990s, migrants from certain provinces began to establish “ethnic enclaves” in particular labor markets (Ma and Xiang 1998).10 This creation of a niche in many ways secures the continuing flow of migrants from certain provinces and makes it increasingly difficult to enforce household registration. Thus, even if some individuals fail to obtain temporary local household registration cards, they can manage to get a job through migrant networks—perhaps working as a household servant (as do many migrant women from Anhui province) or working in a restaurant or clothing shop for someone who comes from the same province. Table 3 also implies that migrants from urban areas were more numerous than in previous years, especially for intraprovincial migrants. The mass media present the misleading impression that migrants come only from rural areas. The active participation of China’s urban population in the wave

Click to return to Table of Contents Z AI LIANG 511 of migration, both interprovincial and intraprovincial, has gone largely un- noticed. The occupational distributions for rural and urban migrants may well be different, but urban dwellers actually have a higher propensity to migrate than rural dwellers.11

Intersectoral migration flows and “urbanization from below”?

Despite the continuing flow of migrants both interprovincially and intrapro- vincially, China’s official urbanization policy remains “to strictly control the growth of the large cities, to rationally develop the medium-size cities, and to vigorously promote the growth of the small cities and towns” (Renmin Ribao, 16 October 1986; quoted in Ma and Lin 1993). Implicit in this policy are the assumptions that cities, especially large cities, have a limited capacity to receive migrants and that a large increase in the migrant population will result in crowded housing conditions, congestion, and perhaps social un- rest, as has been the case for cities in many other developing countries. Since its implementation in 1979, however, the household responsibility system in rural China has improved the efficiency of agricultural production and resulted in millions of surplus workers in rural areas (Johnson 1988; Lin 1988). Recent estimates indicate that there may be as many as 130 million surplus rural workers. Based on the rural population enumer- ated in the 1990 Population Census, this represents about 15 percent of China’s rural labor force (SSB 1991). The estimated number of surplus rural workers was expected to climb as high as 200 million by the year 2000 (Renmin Ribao, 12 July 1995). To avoid the urban problems experienced by some rapidly developing countries, Chinese policymakers and some scholars have advocated a general policy of “leaving the land, but not the rural areas.” One concrete step toward implementing this policy is to encourage the establishment of rural enterprises in small towns so that rural migrants can work there instead of moving to cities (Duan, Wei, and Shan 1997). Using Guangdong as a case study, Ma and Lin documented major growth in the number of towns and their populations in the mid-1980s. Further, they argued that “[i]f this policy and the current pattern of town growth are allowed to continue for an extended time, the towns may form a second track of ‘urbanization from below’ to complement the dominant city-based track of ‘urbanization from above’” (Ma and Lin 1993: 603). The prospect of “urbanization from below” has both theoretical and policy implications. Because the survey used in Ma and Lin’s study was completed in 1987, we have an opportunity to examine the extent to which urbanization from below occurred in China during 1982–95. If their thesis holds, the importance of towns as migrant destinations should increase over

Click to return to Table of Contents 512 T HE AGE OF MIGRATION IN CHINA

time. Table 4 shows the percentages of the intersectoral migrant population according to origin and destination, by city, town, and rural location. The top panel (for cities) shows the distribution of the migrant population by these criteria. For example, among migrants in cities in 1987, 29 percent originated from cities and 62 percent from rural areas. Among migrants who originated from cities, as of 1987 60 percent were residing in cities and 14 percent in rural areas. Several patterns emerge from an examination of Table 4. First, there has been a major drop in migration to towns whether from cities, towns, or rural areas. The share of city-to-town migrants declined from 26 percent in 1987 to about 9 percent in 1995 (right-hand side of the top panel), and the share of town-to-town migrants declined from 50 percent in 1987 to 19 percent in 1995. The largest drop is in the share of rural- to-town migrants. According to the 1987 survey, 41 percent of migrants originating from rural areas were living in towns; by 1995, however, the corresponding figure had declined to 9 percent. This finding is corroborated by migration patterns to towns. In 1995, 56 percent of the migrant population living in towns were from rural areas, compared to 71 percent in 1987 (left-hand side of middle panel). Where did rural migrants go in the 1990s? The lower panel of Table 4 shows that they moved increasingly to cities and other rural areas. For instance, according to the 1987 survey, 33 percent of migrants originating from rural areas were living in cities; that proportion increased to about 51 percent in 1995.

TABLE 4 Percent distribution of the intersectoral migrant population according to place of origin and destination, China, 1987 and 1995 Migrants in cities originating from Migrants originating from cities living in Period Cities Towns Rural areas Total Cities Towns Rural areas Total

1987 29.4 9.0 61.6 100.0 59.9 26.1 14.0 100.0 1995 40.6 10.0 49.5 100.0 80.6 8.7 10.7 100.0 Migrants in towns originating from Migrants originating from towns living in Cities Towns Rural areas Total Cities Towns Rural areas Total

1987 11.8 17.6 70.6 100.0 23.6 49.8 26.6 100.0 1995 26.8 17.6 55.6 100.0 65.6 18.8 15.6 100.0 Migrants originating from rural areas Migrants in rural areas originating from living in Cities Towns Rural areas Total Cities Towns Rural areas Total

1987 10.7 15.8 73.7 100.0 33.2 41.2 25.6 100.0 1995 11.6 5.1 83.3 100.0 50.8 9.3 39.8 100.0

NOTE: Percent totals may not add to 100 due to rounding. SOURCES: 1987 and 1995 China One Percent Population Sample Surveys.

Click to return to Table of Contents Z AI LIANG 513

What are the net results of this exchange of in-migration and outmigration for cities and towns? To answer this question, I conducted an analysis of changes in origin and destination between 1982 and 1995 and results are summarized in the Appendix Table. This exercise was done separately for the two time periods: 1982–87 (from Goldstein 1990) and 1990– 95. For each time period, the top panel shows the estimated size of the migrant population according to origin and current place of residence. Net migration is derived from the top panel. For example, for the time period of 1990– 95, there were 10.096 million migrants who moved from rural areas to cities and there were 1.098 million migrants who moved from cities to rural areas. Thus the net migration from rural areas to cities is 8.998 (10.096 – 1.098) million (see lower part of the panel for period 1990–95). According to the Appendix Table, cities showed a net gain of 5.8 million migrants and towns had a net gain of 7.2 million migrants during 1982–87. In contrast, during 1990–95 cities had net gain of 10.1 million migrants; towns had a net gain of less than a quarter-million migrants. Towns have been clearly losing the contest of attracting rural migrants. In sum, although migration to towns continued in the 1990s, their importance as migrant destinations has declined over time, a circumstance that undermines the thesis of “urbanization from below”.12 A second pattern that emerges in Table 4 is the increase in city-to-city and town-to-city migration. According to the 1987 survey, 24 percent of migrants originating from towns were living in cities; by 1995, the figure jumped to 66 percent. Similarly, according to the 1995 survey, 81 percent of migrants originating from cities were living in other cities, as compared to 60 percent according to the 1987 survey. Overall, people in the early 1990s, regardless of their place of origin, were more likely than in the mid-1980s to select cities as their destination. Among migrants to cities in 1995, 41 percent were from other cities; the corresponding figure for 1987 was 29 percent. A significant increase is also found in the share of migrants moving from cities to towns. Among migrants to towns in 1995, 27 percent were from cities, as compared to only 12 percent in 1987. The share of the city-origin population also increased slightly among the migrants to rural areas. Other evidence indicates that migrants from cities to towns and rural areas are often temporary. This suggests that migrants from cities to towns and rural areas are likely to be technical personnel, hired by industries in those places (Yang 1994). Finally, the data in Table 4 indicate that during the periods considered, rural areas became less attractive as destinations for urban migrants, while they became more attractive for migrants from other rural locations. According to the 1987 survey, 14 percent of migrants originating from cities were living in rural areas; in 1995 the figure declined to 11 percent. The

Click to return to Table of Contents 514 T HE AGE OF MIGRATION IN CHINA share of migrants to rural areas from towns experienced an even sharper decline from 27 percent to 16 percent. Somewhat surprisingly, in 1995, 40 percent of migrants in rural areas originated from other rural areas, representing about a 14 percentage point increase compared to the corresponding share in 1987. This phenomenon, relatively new in China, suggests a novel type of migration pattern. Rural migrants from certain provinces are migrating to cities but holding onto their land in case their expectations are not realized.13 As they attempt to establish themselves in the cities, they may hire migrants from other rural areas to plant and harvest certain crops on the rural land they have retained (Du and Bai 1997: 177; Huang 1997).14 The main choices of destinations for migrants from rural areas seem to be cities or other rural areas, but not towns. The uneven distribution of population and economic development in China is well known. The most densely populated provinces, with population densities reaching as high as 372 persons per square km in 1982, lie in the coastal regions in the east (Zhu 1994). In the rest of this section, I examine changes in destination patterns of interprovincial migrants during 1982–95. Table 5 shows the migrants in each province as a percentage of the interprovincial migrant population in 1987 and 1995. In 1987, the top three destination provinces for interprovincial migrants, all located in the coastal region, are Hebei (9.4 percent), Shandong (8.7 percent), and Jiangsu (7.6 percent). Capitalizing on the advantaged geographic position of a shared boundary with Shanghai, Jiangsu had a head start in the initial stages of economic reforms. All three provinces enjoyed very strong economic growth during this period. For instance, from 1978 to 1988 per capita income in rural areas increased by almost 200 percent in Jiangsu province and by 252 percent in Hebei.15 By 1995 the most important destination province was Guangdong. With only 6 percent of China’s population, Guangdong was home to about 18 percent of China’s interprovincial migrants.16 Table 6 aggregates China’s provinces into six regions to illustrate the broad geographic distribution of interprovincial migrants in 1987 and 1995. Among China’s six administrative regions, two gained interprovincial migrants in 1995 as compared to 1987: the East and Central and south. The Central and south region’s gain is primarily attributable to Guangdong. An- other way to classify migrants is by location in the coastal and the noncoastal regions.17 This classification is meaningful because of the dynamic economic development in the coastal regions, which contain about 40 percent of China’s population. In 1987, 53 percent of China’s interprovincial migrants were found in coastal provinces. This figure had increased more than 10 percentage points by 1995, when 66 percent were so located.18

Click to return to Table of Contents Z AI LIANG 515

TABLE 5 Interprovincial migrants in a province as percent of all interprovincial migrants in China, 1987 and 1995 1987 1995 Province Number (000) Percent Number (000) Percent

North Beijing 328 5.2 695 6.5 Tianjin 134 2.1 223 2.1 Hebei 594 9.4 503 4.7 Shanxi 165 2.6 158 1.5 Inner Mongolia 168 2.7 275 2.6 Northeast Liaoning 314 5.0 435 4.1 Jilin 169 2.7 150 1.4 Heilongjiang 191 3.0 224 2.1 East Shanghai 376 6.0 727 6.8 Jiangsu 477 7.6 969 9.1 Zhejiang 124 2.0 465 4.4 Anhui 166 2.6 156 1.5 Fujian 90 1.4 344 3.2 Jiangxi 101 1.6 126 1.2 Shandong 548 8.7 527 4.9 Central and south Henan 264 4.2 270 2.5 Hubei 274 4.3 271 2.5 Hunan 217 3.4 215 2.0 Guangdong 305 4.8 1,947 18.3 Guangxi 59 0.9 120 1.1 Hainan NA NA 104 1.0 Southwest Sichuan 388 6.1 395 3.7 Guizhou 114 1.8 152 1.4 Yunnan 106 1.7 207 1.9 Tibet NA NA 36 0.3 Northwest Shannxi 225 3.6 163 1.5 Gansu 93 1.5 140 1.3 Qinghai 29 0.5 52 0.5 Ningxia 92 1.4 49 0.5 Xinjiang 202 3.2 566 5.3 China 6,312 100.0 10,661 100.0

NOTE: Data for Hainan and Tibet in 1987 are not available. SOURCES: 1987 data from SSB (1988): Table 1-22, pp. 23–24; 1995 data from SSB (1997): Table 7-2, pp. 540– 541.

Click to return to Table of Contents 516 T HE AGE OF MIGRATION IN CHINA

TABLE 6 Interprovincial migrants in a geographic region as percent of all interprovincial migrants in China, 1987 and 1995 1987 1995 Region Number (000) Percent Number (000) Percent

North 1,389 22.0 1,854 17.4 Northeast 674 10.7 810 7.6 East 1,882 29.8 3,313 31.1 Central and south 1,119 17.7 2,926 27.4 Southwest 608 9.6 789 7.4 Northwest 640 10.1 969 9.1 China 6,312 100.0 10,661 100.0

NOTE: Percent totals may not add to 100 due to rounding. SOURCES: 1987 data from SSB (1988): Table 1-22, pp. 23–24; 1995 data from SSB (1997): Table 7-2, pp. 540–541.

Reasons for migration

The 1987 China One Percent Population Sample Survey includes information on reasons for migration. Although the question was dropped from the 1995 sample survey, China’s 1990 Population Census also asked migrants about reasons for moving. In this section, I use data from the 1987 survey and the 1990 census to indicate changes over time in reasons for migration. The identical response categories used in both data sets make this comparison possible. Table 7 gives reasons for migration as reported in 1987 and 1990, dis- tinguishing whether migration was intraprovincial or interprovincial. In 1987 28 percent of migrants stated that they moved because of marriage and 16 percent because they were dependents of migrants. In other words, family reunification accounted for 44 percent of China’s migrant population at this date. This is consistent with steps taken in 1980 to relax policies regarding movement of couples who have long lived in separate locations (Wang 1997; Zhang 1994: 248). The pattern, however, differs by type of migration. Mar- riage-related migration occurred much more frequently among intraprovincial migrants. This is especially true for women, a finding that is common in other countries (Skeldon 1986). For interprovincial migrants, the most common reason was job transfer (20 percent), which often results in a permanent move with secured household registration status at the place of destination. Fewer than 10 percent of all migrants reported business or factory work as reasons for migration. Given the fact that people who engage in such work are most likely to be temporary migrants, the small percentage suggests that temporary migration was not yet at center stage in 1987. In the 1990 census, a new pattern emerges. Migrants who reported having moved because of business or factory work became the largest group (24

Click to return to Table of Contents Z AI LIANG 517

TABLE 7 Reasons for migration by type of migration, reported by migrants, 1987 and 1990 1987 1990 Intra- Inter- Intra- Inter- provincial provincial provincial provincial Reason migration migration Total migration migration Total

Job transfer 11.4 19.9 13.2 10.0 14.6 11.5 Job assignment 6.6 5.3 6.3 6.5 4.7 5.9 Business or factory work 9.3 9.6 9.3 22.0 29.4 24.4 Education or job training 8.0 9.0 8.2 15.3 7.8 12.9 Live with friends/relative 8.8 13.4 9.8 9.5 10.6 9.9 Retirement 2.2 2.5 2.3 1.6 1.5 1.6 Dependents of migrants 15.0 18.7 15.8 10.2 10.8 10.4 Marriage 30.8 15.5 27.7 14.3 14.2 14.3 Other 7.9 6.2 7.6 10.6 6.5 9.3 Total 100.0 100.0 100.0 100.0 100.0 100.0

NOTE: Percent totals may not add to 100 due to rounding. SOURCES: 1987 China One Percent Population Sample Survey and 1990 China Population Census. percent); among interprovincial migrants, the figure approached 30 percent. Nearly half of interprovincial migrants cited a work-related reason for moving (job transfer, job assignment, and business or factory work). This change clearly indicates that migration was more likely to be a response to economic opportunities in the late 1980s and early 1990s. Although we do not have identical information for the 1995 survey, the fact that a large volume of temporary migration was recorded in that survey suggests that this job-related pattern of migration should hold for the mid-1990s as well.

Discussion and conclusions

Clearly, temporary migration in China is on the rise. Even by conservative estimates there were 68 million temporary migrants in China in 1996. That figure represents only those temporary migrants who have resided at the place of destination for six months or more; with a shorter time reference the number would of course be higher. The rise of temporary migration is, in large measure, the result of China’s transition to a market-oriented economy and the continuing legacy of the household registration system. Although the magnitude of China’s temporary migrant population may fluc- tuate over time, policymakers must realize that temporary migration (and migration in general) will be an important feature of China’s ongoing transition to a market economy. In fact, the size of the migrant population is likely to continue rising—in light of projections that the surplus agricultural labor force will approach 150–200 million early in the twenty-first

Click to return to Table of Contents 518 T HE AGE OF MIGRATION IN CHINA century and in response to the formation of migrant networks linking origin and destination communities. Findings from a national survey place the total number of temporary migrants at 69.3 million in 1997, an increase of more than one million since 1996 (SSB 1998).19 Other distinct migration patterns are evident in this analysis. Over time, migration increasingly involves long distances and the crossing of province boundaries. Migrants from rural origins increasingly choose cities as destinations; there is no sign that this trend will abate in the near future. Inter- provincial migrants continue to move to coastal provinces in large numbers. Migration is more and more driven by individuals seeking to engage in business and factory work. It is not an exaggeration to call this the age of migration in China. The massive scope of migration in the 1990s is unprecedented in the 50-year history of the People’s Republic. In fact, migration in the 1990s was one of the most prominent demographic events in twentieth-century China. The volume of migration is helping to transform China from an agrarian society into a modern industrialized society, a change that is being felt worldwide in this era of globalization. In many places in China, migrants form the backbone of several industries—working in construction, joint- venture manufacturing enterprises, and the service industry; they also serve in managerial and professional capacities. Migration has created opportunities for social mobility for millions of Chinese people, especially peasants, and has begun to challenge the traditional hierarchy of social and economic stratification stemming from the rural/urban household registration status. The migration of peasants helps to reduce rural/urban inequality because rural migrants earn more money in cities and send remittances back to the countryside (World Bank 1997). Migration has become part of a strategy for survival for an increasing number of rural households: a 1994 survey suggests that as much as 18 percent of China’s rural household income is derived from migrant remittances (Cai 1997). Return migration, although difficult to measure precisely, seems to play an increasingly important role in rural transformation. Findings from a recent nine-province survey of return migration show that “labor migration can not only enhance the transformation of rural farmers to commercial and industrial entrepreneurs, but also facilitate agricultural development” (Ma 1999: 29). Behind the backdrop of China’s large volume of migration is the rigid household registration system that prevents most migrants from becoming full citizens at their places of destination, especially in urban areas. Without urban household registration status, migrants still face many difficulties when they seek to marry, secure housing, or enroll their children in neighborhood schools (Solinger 1999). Yet now may not be the best time to relin- quish the household registration system. After all, the differentials between rural and urban China are still pronounced. Meanwhile, recent changes in

Click to return to Table of Contents Z AI LIANG 519 state-owned enterprises have increased uncertainties and anxieties, fueled in part by reports that more than 10 million urban workers have been laid off as a result (Du and Bai 1997; Pan 1997).20 The number of unemployed workers is likely to increase as state-owned enterprises take more drastic steps to increase efficiency and to compete in the global economy. These factors argue against rushing to a quick solution to the issue of the household registration system.21 In the long run, however, concerted efforts must be made to accommodate China’s newest urban citizens.

APPENDIX TABLE Size of the migrant population according to place of origin and current residence, 1987 and 1995 and estimated net migration flows between cities, towns, and rural areas 1982–87 and 1990–95 Place of origin in 1982 City Town Rural Total

1987 residence Number of migrants (millions) City 3.296 1.017 6.943 11.256 Town 1.403 2.001 8.053 11.457 Rural 0.796 1.260 5.764 7.820 Total 5.495 4.278 20.760 30.533

Place of destination Net migration (millions) between 1982 and 1987 City — 0.386 –6.147 –5.761 Town –0.386 — –6.793 –7.179 Rural 6.147 6.793 — 12.940 Total 5.761 7.179 –12.940 — Place of origin in 1990 City Town Rural Total

1995 residence Number of migrants (millions) City 8.278 2.038 10.096 20.412 Town 0.893 0.584 1.851 3.328 Rural 1.098 0.483 7.908 9.489 Total 10.269 3.105 19.855 33.229

Place of destination Net migration (millions) between 1990 and 1995 City — –1.145 –8.998 –10.143 Town 1.145 — –1.368 –0.223 Rural 8.998 1.368 — 10.366 Total 10.143 0.223 –10.366 — SOURCES: Data for top panel from Goldstein (1990); data for bottom panel are author’s calculations based on CPSSO (1997): Table 7-5, p. 558.

Notes An earlier version of this article was pre- by grants from the National Institute of Child sented at the Annual Meeting of the Ameri- Health and Human Development (1R55HD/ can Sociological Association, San Francisco, OD3487801A1 and 1R29HD34878-01A2), the CA, August 1998. This research is supported PSC-CUNY Research Award Program, and the

Click to return to Table of Contents 520 T HE AGE OF MIGRATION IN CHINA

Queens College Presidential Research Award. rary migrants are applied to the 1995 and 1996 The author gratefully acknowledges their sup- data (a six-month instead of one-year crite- port. He thanks Kam Wing Chan, Sidney rion was used). Because of these differences Goldstein, Denise Hare, Cheng Li, Lawrence in data definitions, we need to note that: (1) Ma, Kenneth Roberts, and Dorothy Solinger the changes in the size of the temporary mi- for their insightful comments on earlier ver- grant population during 1982–90 can be as- sions. Katerina Sergi and Leah Weiss provided sessed unambiguously and the same can be able research assistance. said for the period 1995–96; (2) the magni- 1 In 1986 the Chinese Academy of Social tude of change in the size of the temporary Sciences conducted a survey of 74 cities and migrant population between 1990 and 1995, towns (Day and Ma 1994). Although the sur- however, is overestimated. There are at least vey was not a random sample of the Chinese two advantages of using these data. One is that population, it contains some important infor- they are nationally representative samples, mation on migration (e.g., income and edu- hence we are more confident about making cation at the time of migration) that is not generalizations to the Chinese population as a available from other surveys or censuses. whole. In addition, even in the case of differ- 2 The term “temporary migrant” signals ences in definitions across data sets, we know the fact that these migrants do not have per- the likely direction of the bias of the estimate. manent household registration status. Strictly 6 The estimate of 56 million temporary speaking, it is a misnomer because many tem- migrants in China in 1995 covers only tempo- porary migrants stay at their place of destina- rary migrants who have stayed at the place of tion for months or years. Recently some re- destination for six months or more. A survey searchers began to use alternative terms, such of temporary migrants in Beijing conducted in as “unofficial migrants,” that more closely cap- 1994 shows that some 37 percent of them had ture the meaning of the type of migrants be- stayed in Beijing for less than six months at ing described (e.g., Goldstein, Liang, and the time of the survey (Zou 1996). Assuming Goldstein 2000). To be consistent with most that temporary migrants in China as a whole previous work (Goldstein 1990; Goldstein and follow the same distribution of duration of resi- Goldstein 1991; Gu and Jian 1994; Yang 1994), dence, the estimated total number of tempo- I retain the use of “temporary migrants.” rary migrants in China in 1995 was about 88.5 3 To be precise, temporary migrants in the million. 1987 and 1995 surveys include: individuals 7 The 1987 and 1995 surveys defined mi- who have resided at their place of destination grants in slightly different ways. In 1987 two for six months or longer but have permanent questions were asked about migrants: place of household registration status elsewhere; indi- origin of individuals who have lived at the cur- viduals who have resided at the place of desti- rent location for no more than five years and nation for less than six months and left the duration of residence at the current location place of permanent household registration (SSB 1988: 819). The 1995 survey has (at least) more than six months ago; individuals who are two ways of defining migrants (CPSSO 1997: without permanent household registration sta- 644). One is the question on the location of tus anywhere (SSB 1988; CPSSO 1997). residence five years ago. If an individual re- 4 The cities surveyed are Chengdu, sided in a different location five years ago, this Guangzhou, Harbin, Jilin, Shanghai, Taiyuan, individual was defined as a migrant. Another and Zhengzhou (Shen and Tong 1992). way is to use the information on the time of 5 Identical definitions of temporary mi- arrival at the place of destination: anyone who grants were employed in data collection for arrived at a destination after 30 September 1990 1982, 1989, and 1990. This means that to be was considered a migrant. Table 2 is based on counted as a temporary migrant, an individual the first definition for both the 1987 and 1995 must have resided at the place of destination surveys. for at least a year or have been absent from 8 There was no information about the the place of household registration for more number of migrants in Tibet in the 1987 data; than a year. Identical definitions for tempo- and Hainan did not become a separate prov-

Click to return to Table of Contents Z AI LIANG 521 ince until 1988. Thus we cannot compare 17 The 12 coastal provinces and munici- changes for these provinces. palities are Beijing, Fujian, Guangdong, 9 Hebei, Heilongjiang, Ningxia, Qinghai, Guangxi, Hainan, Hebei, Jiangsu, Liaoning, Shanghai, and Tianjin are the provinces (mu- Shandong, Shanghai, Tianjin, and Zhejiang. nicipalities) where the proportional share of in- 18 Since my measure of migration to terprovincial migrants declined between 1987 coastal regions refers only to interprovincial and 1995. migrants, it is quite possible that intraprovincial 10 In Beijing, for example, migrants from migration to coastal cities may have increased Zhejiang (such as residents of the city of much more during the same period. Wenzhou) own many clothing shops and have 19 The figure for 1997 is taken from the a strong presence in the cloth retailing indus- 1997 China Survey of Population Dynamics try. In Xinjiang village in Beijing, migrants from (SSB 1998). The definition of a temporary mi- Xinjiang province congregate to do business. grant in this survey is the same as that of the Southern China has experienced a similar phe- 1995 China One Percent Population Sample nomenon. Major streams of migrants from Survey and the 1996 China Survey of Popula- Sichuan province and from northeast prov- tion Dynamics (CPSSO 1997; SSB 1997). inces are influential in Guangdong’s economy. 20 Other writers have reported much 11 Urban-origin (city and town) individu- larger estimates of workers laid off from state- als accounted for almost half of the intraprovin- owned enterprises in the late 1990s, on the cial migrant population in 1995, but note that order of 30–40 million (Wright 1998). China’s urban population was only 29 percent 21 In several cities such as Beijing, Shang- in the same year. Using data from the 1990 hai, Shenzhen, and Xiamen, a proposal for a China Population Census, I have also shown “blue chip household registration system,” (Liang 1998) that, controlling for individual- which grants certain migrants a special house- level characteristics such as education and age, hold registration status, has been tested. In urban-origin people are more likely to make Shanghai, for example, migrants who are able permanent moves than rural-origin people. to invest large amounts of capital or who pos- 12 My results here may be affected by the sess certain professional skills are eligible to rise of the number of cities in China during apply for blue chip household registration sta- 1982–95. To the extent that towns are pro- tus. After five years in this status, migrants can moted to city status, they are likely to be apply for permanent household registration smaller cities. A more detailed analysis would (Wong and Huen 1998). need to examine the size of cities to which mi- Some scholars have proposed gradually grants moved during 1982–87 and 1990–95. reducing the privileges attached to urban I did not have access to such information at household registration (instead of granting full the time of writing this article. residential rights to all migrants). In October 13 For example, in her anthropological 1998, a new set of flexible guidelines regard- research in Zhejiang village in Beijing, Zhang ing household registration system is reportedly (2000) found that most of the migrant work- being implemented (Lu 1998). The major pro- ers retained their land in their home province. posed changes are: (1) parents can choose for their newborn babies the household registra- 14 Charles Tilly (private communication, tion status of either parent (as compared to that 1994) made similar observations during his trip of the household head before); (2) permanent to China in the early 1990s. household registration status can be granted 15 Growth in per capita income was 146 to a spouse after co-residing for a certain time percent for China as a whole during the same with a spouse who has permanent household period: 1978–88. registration status; (3) elderly people over 60 16 Hainan became a separate province years old can be granted permanent household from Guangdong in 1988; otherwise the per- registration status at places where their chil- centage for Guangdong would be even higher. dren reside.

Click to return to Table of Contents 522 T HE AGE OF MIGRATION IN CHINA

References

Cai, Fang. 1997. “On the issue of floating population,” in Min Xu (ed.), Critical Moment: 27 Critical Problems in Contemporary China. Beijing: Today Press, pp. 423–435. Chan, Kam Wing. 1994. Cities with Invisible Walls. New York: Oxford University Press. Cheng, Tiejun and Mark Selden. 1994. “The origins and social consequences of China’s hukou system,” The China Quarterly 139: 644–668. China Population Sample Survey Office (CPSSO). 1997. Tabulations of the 1995 China 1% Population Sample Survey. Beijing: China Statistics Publishing House. Day, Lincoln and Ma Xia. 1994. Migration and Urbanization in China. Armonk, NY: M. E. Sharpe. Du, Ying and Bai Nansheng. 1997. Out of the Countryside: An Empirical Study of Labor Mobility in China. Beijing: Economic Science Publishing House. Duan, Benluo, Heng Wei, and Qiang Shan. 1997. “The issue of small towns,” in Min Xu (ed.), Critical Moment: 27 Critical Problems in Contemporary China. Beijing: China Today Press, pp. 408–422. Fei, Xiaotong. 1985. Small Towns in China: Functions, Problems, and Prospects. Beijing: New World Press. Gao, Gang. 1997. “The wandering of 200 million peasants at the end of the century,” China News Digest 312: 2–4. Goldstein, Sidney. 1990. “Urbanization in China, 1982–87: Effects of migration and reclas- sification,” Population and Development Review 16: 673–701. Goldstein, Sidney and Alice Goldstein. 1990. “China,” in Charles B. Nam, William Serow, and David F. Sly (eds.), International Handbook on Internal Migration. New York: Green- wood Press, pp. 63–84. ———. 1991. ”Permanent and temporary migration differentials in China,” Papers of the East-West Population Institute no. 117. East-West Center, Honolulu, Hawaii. Goldstein, Sidney, Zai Liang, and Alice Goldstein. 2000. “Migration, gender, and labor force in Hubei Province, 1985–1990,” in Barbara Entwisle and Gail Henderson (eds.), Re- Drawing Boundaries: Work, Households, and Gender in China. Berkeley: University of Cali- fornia Press, pp. 214–230. Gu, Shengzu and Jian Xinhua. 1994. Contemporary Population Migration and Urbanization in China (dangdai zhongguo renkou liudong yu chengshehua). Wuhan: Wuhan University Press. Huang, Ping (ed.). 1997. The Search for Survival: A Sociological Study of Rural-Urban Migration. Kunming: Yunnan People’s Press. Johnson, D. Gale. 1988. “Economic reforms in the People’s Republic of China,” Economic Development and Cultural Change 36: S225–S245. Khan, Azizur Rahman and Carl Riskin. 1998. “Income and inequality in China: Composi- tion, growth and distribution of household income, 1988 to 1995,” The China Quar- terly 154: 222–232. Li, Cheng. 1996. “Surplus rural laborers and internal migration in China: Current status and future prospects,” Asian Survey 36: 1122–1145. Liang, Zai. 1998. “Market reforms and changing patterns of migration and occupational attainment,” paper presented at the Annual Meeting of the Population Association of America, San Francisco, CA, March. ———. 1999. “Foreign investment, economic growth, and temporary migration: The case of Shenzhen Special Economic Zone, China,” Development and Society 28(1): 115–137. Liang, Zai and Michael J. White. 1996. “Internal migration in China, 1950–1988,” Demogra- phy 33: 375–384. Lin, Justin Yifu. 1988. “The household responsibility system in China’s agricultural reform: A theoretical and empirical study,” Economic Development and Cultural Change 36: S199– S224. Lipton, Michael. 1977. Why Poor People Stay Poor: Urban Bias in World Development. Cam- bridge, MA: Harvard University Press.

Click to return to Table of Contents Z AI LIANG 523

Liu, Zheng (ed.). 1990. Economic Development and Population Migration in Small Towns Along China’s Coastal Region. Beijing: Zhanwang Press. Lu, Yang. 1998. “Major changes in China’s household registration system,” Huasheng Monthly 43: 78. Ma, Lawrence J. C. and Chusheng Lin. 1993. “Development of towns in China: A case study of Guangdong Province,” Population and Development Review 19: 583–606. Ma, Lawrence J. C. and Biao Xiang. 1998. “Native place, migration, and the emergence of peasant enclaves in Beijing,” The China Quarterly 155: 546–581. Ma, Zhongdong. 1999. “Labor migration and return transformation in China,” unpublished manuscript, Hong Kong University of Science and Technology. Massey, Douglas S. and Kristin E. Espinosa. 1997. “What’s driving Mexican–U.S. migration: A theoretical, empirical, and policy analysis,” American Journal of Sociology 102: 939–999. Mer, Fu. 1998. “Maiden workers in China’s west,” People’s Daily, overseas edition, 27 Feb- ruary. Nee, Victor. 1989. “A theory of market transition: From redistribution to markets in state socialism,” American Sociological Review 54: 663–681. Pan, Qi. 1997. “Reform of state enterprises and social security,” Huasheng Monthly 33: 40– 43. Poston, Dudley L. and Michael Xinxiang Mao. 1998. “An ecological analysis of interprovincial migration in China,” in Harry K. Schwarzweller and Brendan P. Mullan (eds.), Research in Rural Sociology and Development, Volume 7. Stanford, CT: JAI Press, pp. 227–250. Renmin Ribao (People’s Daily, overseas edition). 1995. “Strengthening the control of floating population,” 12 July. Roberts, Kenneth. 1997. “China’s tidal wave of migrant labor: What can we learn from Mexican undocumented migration to the United States?” International Migration Re- view 31: 249–293. Shen Yimin and Tong Chengzhu. 1992. Population Migration in China, Historical and Contem- porary Perspectives. Beijing: China Statistics Publishing House. Skeldon, Ronald. 1986. “On migration patterns in India during the 1970s,” Population and Development Review 12: 759–779. Solinger, Dorothy J. 1999. Contesting Citizenship in Urban China: Peasant Migrants, the State, and the Logic of the Market. Berkeley: University of California Press. State Statistical Bureau (SSB). 1988. Tabulations of China 1987 1% Population Sample Survey. Beijing: China Statistics Publishing House. ———. 1991. China Population Statistics (1990). Beijing: Science and Technology Press. ———. 1994. China Population Statistics. Beijing: China Statistical Publishing House. ———. 1997. China Statistical Yearbook 1997. Beijing: China Statistics Publishing House. ———. 1998. China Statistical Yearbook 1998. Beijing: China Statistics Publishing House. World Bank. 1997. Sharing the Rising Income: Disparities in China. Washington, DC. Wang, Feng. 1997. “The breakdown of a Great Wall: Recent changes in household registration system in China,” in Thomas Scharping (ed.), Floating Population and Migration in China: The Impact of Economic Reforms. Hamburg: Institute of Asian Studies, pp. 149– 165. Wong, Linda and Huen Wai-Po. 1998. “Reforming the household registration system: A preliminary glimpse of the blue chip household registration system in Shanghai and Shenzhen,” International Migration Review 32: 974–996. Wright, Daniel B. 1998. “One in fifty million: Traveling with China’s migrant labor,” Insti- tute of Current World Affairs Letters. DBW-5. February. Yang, Dali. 1991. “China adjusts to world economy: The political economy of China’s coastal development strategy,” Pacific Affairs 64: 42–64. Yang, Xiushi. 1994. “Urban temporary out-migration under economic reforms: Who moves and for what reasons?” Population Research and Policy Review 13: 83–100.

Click to return to Table of Contents 524 T HE AGE OF MIGRATION IN CHINA

Zhang, Li. 2000. “The interplay of gender, space, and work in China’s floating population,” in Barbara Entwisle and Gail E. Henderson (eds.), Re-Drawing Boundaries: Work, House- holds, and Gender in China. Berkeley: University of California Press, pp. 171–196. Zhang, Qingwu. 1994. Hukou Migration and Floating Population. Beijing: Institute of Popula- tion Management of China People’s Security University. Zhu, Nong. 1994. “Population distribution and urbanization,” in Sun Jingxin (ed.), The Popu- lation of China Towards the 21st Century (Comprehensive Volume). Beijing: China Statistics Publishing House, pp. 203–237. Zou, Lanchun (ed.). 1996. The Floating Population in Beijing. Beijing: China Population Pub- lishing House.

Click here to print article

ClickClick toto returnreturn toto TableTable ofof ContentsContents NOTES AND COMMENTARY

Divergent Paths of Immigration Politics in the United States and Australia

GARY P. FREEMAN BOB BIRRELL

SETTLER SOCIETIES with long traditions of immigration, Australia, Canada, and the United States have in recent decades typically adopted liberal and expansive policies toward migrants. All three countries appeared to be con- verging by the late 1980s in ways consistent with a model of immigration policymaking that might be called client politics. In this model organized interest groups benefiting from migration tend to prevail over less-well-organized or less-intense opposition. The main political parties sought to avoid open conflict over migration issues, disagreeing only at the margins and creating a rough consensus that supported gradual expansion despite evidence that public opinion was split between reducing admissions or leaving them where they were (Freeman 1995). In the 1990s, however, a variety of populist uprisings challenged policies permitting high levels of immigration. Only in Australia did this lead to tighter control over legal and illegal entries, declining annual numbers, and significant retrenchment with regard to policies supporting multiculturalism. Canada appeared for a time to be heading cautiously in the direction taken by Australia. After the 1993 election, in which the anti-immigration Re- form Party made a decisive showing, Canada’s new Liberal government cut back annual admissions figures and a departmental advisory group issued a report highly critical of aspects of immigration policy (Legislative Review Advisory Group 1998). Upon retaining office in 1998, however, the Liber- als rejected many of the group’s proposals and reaffirmed their commitment to the immigration status quo. In the United States a grassroots cam-

POPULATION AND DEVELOPMENT REVIEW 27(3):525–551 (SEPTEMBER 2001) 525

Click here to print article Click to return to Table of Contents 526 D IVERGENT PATHS OF IMMIGRATION POLITICS paign aimed at illegal immigrants in California appeared to herald a period of skepticism about the benefits of mass immigration. The new Republican Party majority in the 1994 Congress took a hard line against illegal movements across the southern border with Mexico and denied access to public benefits for many legal immigrants. In addition, presidential candidate Patrick J. Buchanan, who proposed a “moratorium” on new legal admissions, enjoyed some short-lived success in the 1996 election campaign. But by the end of the decade American policy was once again firmly on the liberal track. In this context Australia presents an analytical puzzle and a theoretical challenge. Why did Australia take a divergent path when similar im- pulses in the United States and Canada were defeated? We answer this question through a close analysis of Australia and the United States, setting aside Canada so as to narrow our focus. There are three broad perspectives in the literature on the comparative politics of immigration. Some scholars work primarily out of a framework built on the logic of markets and explore the way this logic shapes interest groups active in immigration politics. Others employ concepts deriving from the logic of rights and their expansion in democracies. Finally, there is work that focuses on the state and its capacity to manage population flows. Each of these is useful and, indeed, most authors at least occasionally slip from one to the other point of view. We begin by laying out the three theoretical perspectives. We turn then to a review of immigration policy in Australia and the United States, concentrating on the period from the late 1980s to the turn of the century. Having established the divergent paths taken by Australia and the United States, we evaluate the relative explanatory power of interests, rights, and states.

Interests, rights, and states in immigration policymaking

Several reviews of international migration theories published in the 1990s are notable for having nothing to say about the politics of immigration or the sources and consequences of migration policy (Massey et al. 1993, 1994, 1998). Scholars outside political science have tended to neglect the political dimensions of migration; and political scientists, for their part, have largely ignored migration as a topic for analysis. Nevertheless, state actions have a major impact on the size, direction, and effects of migration flows (Zolberg 1999; Portes 1999). Theories of migration that neglect the state role are necessarily incomplete. At long last a growing body of self-consciously theoretical work on migration has emerged from within political science (see Hollifield 2000). The concept of interest, long central to modern political analysis, would seem a natural starting point for the study of immigration politics in West- ern societies. Despite the claims of some advocates that immigration makes

Click to return to Table of Contents G ARY P. FREEMAN / BOB BIRRELL 527 everyone better off, in the real world there are likely to be winners and losers as the consequences of mass migration manifest themselves. Political economists believe that the operation of the market, broadly conceived, generates the most politically relevant interests. They assume as well that individuals seek to maximize their interests as they understand them and that expressed political preferences and most overt political conflicts reflect this inclination. Such a perspective leads the analyst to consider what is at stake in immigration—how it affects the jobs, incomes, and living standards of immigrants and members of the host society alike. Put simply, who benefits and who loses? The pattern of winners and losers in immigration should be reflected, given the above assumptions, in the lines of cleavage and patterns of interest-group formation around immigration as a political issue. Freeman (1995) has applied a framework devised by Wilson (1980) to derive four types of immigration politics depending on whether the benefits and costs of policies are concentrated or diffuse. If both costs and benefits are concentrated (i.e., fall on identifiable and relatively discrete groups in the population), then interest groups should organize to represent both sides. Those who are absorbing the costs should seek tougher immigration rules, while those accruing benefits should favor continuation or expansion of current policies. Conflict should be sharp because a great deal is at stake on both sides, and policies should shift significantly if one side or the other is influential in the rise and fall of incumbent political parties. Client politics, alternatively, develops when benefits are concentrated but costs are not. Conflict should be relatively inconsequential because those in the majority, while paying the costs of immigration, are not severely affected, display little interest in the issue, and are hindered by obstacles to collective action. On the other hand, if it is costs that are concentrated, conditions are ripe for the emergence of an entrepreneurial leader or party to mobilize discontent and seek redress for grievances. If, finally, both costs and benefits are diffuse, policies can be sustained only by building consensual majorities for or against changes in the status quo. Freeman argues that the most common mode of immigration politics in the liberal democracies is client politics, in which policymakers interact intensively and often out of public view with groups directly standing to gain from immigration. While not denying that interests are involved in disputes over immigration, an alternative scholarly perspective stresses the extent to which the rights that are central to the character of democratic states can override interests and constrain the actions of states in conflicts over immigration (Soysal 1994; Jacobson 1996; DeLaet 2000). Hollifield (1992, 1994) has developed this approach most persuasively. He suggests that the migration policies of Western states must be seen from the viewpoint of two institutional features of modern democratic politics. At the international level they

Click to return to Table of Contents 528 D IVERGENT PATHS OF IMMIGRATION POLITICS reflect the “embedded liberalism” that embodies the principles of the hegemonic states in the postwar international order: support for free trade, democratic governance, anticommunism, and military cooperation under Ameri- can leadership (Ruggie 1992). Beyond that, the rights perspective argues that national policymakers are constrained by the universal human rights norms embodied in United Nations conventions and other treaties and agreements and by the pressure and oversight that emanate from international agencies, both public and private. At the level of national states, domestic politics assures that liberal regimes place a premium on individual rights. Migrants, as well as citizens, embrace the language and privileges of rights- based discourses and policies. In this view the liberalism of Western states, and the international order they create, protect migrants against arbitrary treatment and constrain the ability of states to regulate their behavior in ways dictated by the outcome of interest-group conflicts on the issue. While this rights-based model seeks to generalize across Western democracies, it allows for the possibility that some countries are more fully integrated into liberal international structures than others and that some adopt more thoroughly liberal, rights-based domestic institutions. In particular, as Joppke (1997, 1998, 1999) has most cogently argued, whether the judiciary plays an independent role in protecting the rights of persons can be a decisive factor in a state’s immigration policy choices. The interests and rights perspectives provide a challenging agenda for comparative research on immigration politics. However, both tend to concentrate on societal pressures and claims without bringing states directly into the analysis as independent entities. They concentrate on the demand for policy rather than looking directly at the dynamics of state responses. We treat states both as actors and as frameworks. As actors, state authorities have agendas of their own, which they may pursue in the face of conflicting demands from private interests and claims for the recognition and protection of rights. As frameworks, states constitute the institutions and processes through which authoritative decisions are made. The United States and Australia diverge significantly in terms of the legislative, judicial, and administrative bodies involved in policy as well as with respect to electoral arrangements and party systems. These institutional variations affect the capacity of the two states to develop and pursue strategic interests and to resist or mold societal demands. Our analysis, therefore, treats immigration policy outcomes as a result of the interplay of interests, rights, and agendas in specific national institutional contexts.

Recent immigration policy in the United States and Australia

In absolute terms, migration to the United States is far more substantial than that to Australia. In relative terms, however, the scale of migration to

Click to return to Table of Contents G ARY P. FREEMAN / BOB BIRRELL 529 the two countries is similar. Legal admissions to the United States have approached historic highs since the mid-1990s. In the latest year for which official data are available, FY 1998, 660,477 persons were granted permanent residency visas. This was down somewhat from recent years; average annual admissions from 1991 to 1998 were 845,000. It is generally estimated that there are an additional net 300,000–400,000 illegal entries per year, with the total unauthorized population between 6 and 10 million. Ac- cording to the 2000 U.S. Census, foreign-born residents jumped by 13.3 million in the 1990s to 11.1 percent of the total population of 275 million. Australia’s formal program has been running at about 80,000–90,000 per annum in recent years, plus an additional 15,000–20,000 New Zealand set- tlers. The projected total for the formal program in 2000–01 is 91,000. Ille- gal migration to Australia is less than to the United States. Unauthorized arrivals by boat and air currently number about 6,000 per year. Total population is 19.2 million. Thus the annual inflows (legal and illegal) to the two countries expressed as a percent of their populations (between 0.4 and 0.5 percent) are similar. However, the foreign born as a proportion of the total population in Australia (at 23 percent) is twice as high as in the United States.

United States

After legislation in 1965 ended the national origins quota system and made family connections the central criterion for entry for settlement, annual admissions to the United States grew steadily and source countries shifted from Europe to Asia and Latin America. The legal immigration policy became highly institutionalized and mostly noncontroversial, even as opinion polls showed the absence of majorities supporting expansion and with significant minorities favoring retrenchment (Espenshade and Hempstead 1996; Espenshade and Belanger 1997; Simon and Lynch 1999). The most politically salient immigration problem became dealing with the flows of undocumented migrants, mostly from Mexico. Legislation in 1986 sought a middle ground by imposing sanctions on employers who “knowingly” hire unauthorized workers and by granting amnesty to around 3 million persons who had been residing illegally in the country since 1982. In 1990 new legislation dealing with legal immigration raised admissions by about one-third and, in conjunction with the influx of the dependents of those legalized by the 1986 law, pushed annual admissions to near historic highs by the mid-1990s. Meanwhile, employer sanctions failed to deter illegal entries, which were estimated to be producing a net increase of 275,000 entrants per year. One source put the total population of undocumented migrants (“illegal aliens”) at 5.4 million in 1995 (Center for Immigration Studies 1995). Several developments shook this context as the decade wore on. In 1994 voters in California, the state with the largest share of foreign born

Click to return to Table of Contents 530 D IVERGENT PATHS OF IMMIGRATION POLITICS and reportedly the largest illegal population, overwhelmingly passed a referendum item (Proposition 187) excluding illegal migrants from receipt of all public benefits except emergency health care. At the same time the Re- publican Party received control of both Houses of Congress in the 1994 mid- term elections, giving key leadership positions to individuals critical of US immigration policy. The United States Commission on Immigration Reform, set up in 1990, issued its findings in the mid-1990s calling for tougher measures on illegal immigration, restraint on family reunification, and gradual reduction in annual numbers (US Commission on Immigration Reform 1994, 1995). Finally, in 1996 the nationally syndicated columnist and television personality Patrick J. Buchanan made a surprisingly strong bid for the Re- publican nomination for president with anti-immigration proposals part of his platform. The stage appeared set for comprehensive changes in American immigration policy. Because the period since 1994 has been frequently portrayed as a time of backlash against immigrants, it is important to consider what actually happened. Many restrictive measures were proposed and a number were included in bills being considered by the House of Representatives or the Senate. Most measures were directed at illegal aliens but some would have slashed the numbers of legal immigrants granted visas each year and have changed the criteria of admission. Only a handful of these proposals survived the tortuous American legislative process. Nevertheless, by the end of 1996 a number of laws with significant restrictive implications had been adopted.1 Most of the new provisions pertained to controlling the border and containing illegal immigration. These included a massive fiscal commitment to expand and equip the Border Patrol. Over 1,000 new Border Patrol agents per year for five years were authorized, doubling its size. New workplace inspectors from the Immigration and Naturalization Service (INS) were funded. Foreigners convicted of entering the United States illegally, or over- staying a previous visa, could be denied new visas for the next three to ten years. After a long legislative struggle, limited “summary exclusion” provisions were enacted. INS officials were thus given the power to decide whether persons seeking asylum at the US border had a credible case and, if not, to deny them entry. A negative decision must be appealed within seven days, after which further appeals are impossible. Asylum officers were given the power to exclude aliens seeking asylum if they arrived in the United States without documents or entered illegally. Deportation procedures were expedited, deportable offenses were expanded, and judicial review of decisions was limited or denied. The only measures that dealt with legal admissions tended to be related to budgetary considerations. Although Congress did not alter the categories under which citizens and permanent residents could sponsor the

Click to return to Table of Contents G ARY P. FREEMAN / BOB BIRRELL 531 immigration of their extended family members, the financial arrangements for sponsorship were tightened. Sponsors of immigrants will hence- forth be held financially accountable for the support of those they sponsor, and only persons with incomes 125 percent of the federal poverty level can sponsor immigrants. Restrictions were also imposed on legal immigrant access to public benefits. Beginning in August 1997 most immigrants were to be excluded from receipt of means-tested old-age pensions, food stamps, public welfare, Medicaid, and other federally funded, means- tested benefits until they became US citizens. In addition, states were given the option to bar both current residents and new immigrants from nonemergency Medicaid, welfare, and other entirely state-funded benefit programs. These provisions represented a major reversal in policy: under prior law states had been prohibited from discriminating against legal immigrants in the provision of assistance. Although then-President Clinton signed these bills into law, he opposed many of their provisions and vowed to work for their repeal. After winning reelection in 1996 he made good on many of these promises. Dur- ing 1997 and 1998 many of the restrictive provisions enacted in 1996 were either delayed, watered down, or rescinded by the Congress, the president, or the federal courts, or their effects were weakened by compensatory actions taken by the states. Eventually the ban on receipt of pensions, Medic- aid, and food stamps was lifted for those already on the rolls before 22 Au- gust 1996. In addition, most state governments stepped in to make up at least a substantial portion of the benefits the welfare legislation had denied to legal immigrants. One overview concluded that “almost every state has decided, sometimes using its own money, to keep immigrant benefits in- tact” (Associated Press, 19 October 1997). Finally, the federal courts intervened on a number of fronts to prevent officials from pursuing deporta- tions under the new provisions. In addition to these reversals of policy, a number of other measures signaled that United States immigration policy was still in an expansionist mode. These include more-liberal entry of temporary agricultural workers, action to allow Central Americans and Haitians in “temporary protected status” to regularize their situations, wider scope for undocumented persons seeking to change their status to permanent resident, and expansion of the temporary skilled worker (H1-B) visa category2 (Freeman 1998, 1999). Moreover, the federal government launched a concerted effort to facilitate the naturalization to citizenship of the millions of permanent residents who had lived in the United States for at least five years. Citizenship would re- store eligibility for the full range of public benefits and give immigrants access to the voting booth. Naturalization applications swelled to record numbers even as the INS was roundly criticized for the administrative backlogs that developed as a result.

Click to return to Table of Contents 532 D IVERGENT PATHS OF IMMIGRATION POLITICS

Australia

Through the 1950s and 1960s successive Australian governments made migration a national priority. This stance was partly based on calls to “popu- late or perish” resulting from concerns that Australia lacked the people to defend itself against more populous northern neighbors (Betts 1988). How- ever, during these decades powerful new interests, including the housing and development industries and manufacturers reliant on the domestic market, emerged as proponents for continued high migration. In the early 1970s there was a brief break in the pattern. The new Labor government led by Gough Whitlam included a cut in migration target numbers as part of its response to some of the urban problems resulting from rapid population growth. Further cuts ensued in reaction to a sudden increase in unemployment with the economic downturn of 1974. With the reelection of the conservative Coalition government in late 1975, led by Malcolm Fraser, the migration program was restored as an important national priority. Population growth fitted snugly within the Coalition’s nation-building agenda. But there were political implications as well. Three influential political constituencies were targeted. The first was the development industries cited above, which saw population growth as crucial to their well-being; the second was employers, who were becoming increasingly anxious about shortages of skilled labor as the economy revived; and the third was the ethnic communities. These communities were seen as the key to a potentially large group of voters. A breakthrough had occurred at this time in the legitimacy of claims about the disadvantages experienced by non-English-speaking background (NESB) migrants and about their rights to celebrate their cultures in Australia. Ethnic leaders managed to convince the political parties that their communities could be reached via policies responding to these concerns. Both major political parties embraced the challenge. The Fraser government took the initiative by opening up the family reunion program to allow sponsorship of brothers and sisters and parents. It also embraced a new multicultural definition of Australia that in principle accorded ethnic cultures equal dignity to the traditions of mainstream Australia and (in sharp contrast to previous assimilationist policies) encouraged the right of these cultures to be maintained and reproduced in Australia. The Labor government led by Bob Hawke that came to power in 1983 pursued similar immigration policies, albeit after some initial hesitancy owing to trade union concerns about migrant worker competition in the labor market. Further significant family reunion concessions were made. Spon- sored parents had virtually unrestricted entry regardless of their age or the number of their children living in Australia. Labor also embraced multiculturalism. It responded to concerns about migrant disadvantage through

Click to return to Table of Contents G ARY P. FREEMAN / BOB BIRRELL 533 a range of measures including substantial funds for English-language training. By the end of the 1980s it could be said that the welcome mat was out for migrants. In addition to programs targeted to migrants, almost all of Australia’s welfare programs (which are far more extensive than those in the United States) were made accessible to migrants immediately on arrival in Australia, as was access to the universal public health and education systems. Skilled migration was also increased during the 1980s. In the wake of a contentious debate about the challenge posed by Asian migration for traditional views on the nature of the Australian community and identity (Blainey 1984), the Hawke government sought to legitimize the migration program on the basis of the economic contributions of skilled and entrepreneurial migrants. Opportunities for business migrants were expanded; and a category was created for young applicants with work experience, English skills, and trade or university education. By 1990–91 this category peaked at 35,000 visas to become the largest component of the migration program. By the end of the 1980s the migration program appeared to have been consolidated around the interests of its major beneficiaries. At the same time, opinion polls showed substantial majorities of voters did not support the increased outreach to migrants or the commitment to multiculturalism (Betts 1999: 120–131). Australia appeared to be a classic case of client politics. While voters may have been troubled, with both major political parties focusing on the organized migrant constituencies, there seemed little prospect of policy change. In 1976, 20 percent of the population had been born overseas. By 1991 this figure had grown to almost 23 percent, of whom 58 percent were born in non-English-speaking countries. The migrant constituency was thus getting bigger. Not surprisingly, one group of leading commentators observed at the time that “there was a certain unstoppable momentum” behind the government program (Castles et al. 1992: 169). The “rights” factor also entered the equation in the 1980s. Persons living in Australia temporarily (including illegals) were increasingly able to appeal through the courts to gain permanent residence, usually on humanitarian grounds. In making their judgments, the courts proved responsive to appeals based on rights drawn in part from international agreements, including the International Covenant on Civil and Political Rights (of which Australia was a signatory). In the process, the Immigration Department found that it had lost to the courts much of its control over selection criteria (Birrell 1992: 27–31). This momentum notwithstanding, during the next decade this policy trajectory was first slowed and then reversed. Surprisingly, both the Labor government of Paul Keating and the Liberal-National Coalition that took over in 1996 participated in this effort. One factor in the general rethinking was a report by a blue-ribbon government committee released in 1988. Con- vened by the Labor government, the Committee to Advise on Australia’s

Click to return to Table of Contents 534 D IVERGENT PATHS OF IMMIGRATION POLITICS

Immigration Policy (CAAIP) was, nonetheless, critical of some aspects of the policy, especially multiculturalism (CAAIP 1988). Then the cross-party consensus that had buttressed expansive immigration policies began to un- ravel. In 1988, John Howard, Liberal leader of the Opposition, voiced concerns about the ethnic composition of recent migration, and partly as a result temporarily lost his leadership position. Although he retracted his comments, a precedent of sorts had been set. His successor, John Hewson, further breached bipartisanship by challenging a number of features of Labor’s migration policies. Furthermore, economic conditions became less favorable to migration, with a deep recession in the late 1980s and early 1990s suggesting to some that migration might threaten jobs and burden the treasury in a time of declining revenues. Policy consequences followed in rapid succession. An important administrative measure, which laid the groundwork for the feasibility of migration-control measures in the 1990s, had been put into place in 1989. To reduce the influence of the courts in overseeing its immigration policy decisions, the Immigration Department reorganized its regulatory structure so as to minimize the scope for applicants, their lawyers, and the courts to evade the policy’s intent. It removed as far as possible any scope for administrative discretion on the part of decisionmaking officers, or of the courts in interpreting these officers’ decisions, by laying out in detail the specifics of all departmental regulations. These regulations, in turn, were given the status of law by incorporating them into acts of Parliament (for a detailed account see Cronin 1993: 91–104). Overall admissions were cut substantially in the 1990s. The migration program covering all permanent entry visas (excluding New Zealanders, who have unrestricted entry) had reached a peak of 124,700 in 1988–89. The figures dropped sharply thereafter and stood at about 80,000 in 1998–99, although they were projected to rise to 91,000 in 2000–01. The smaller numbers reflected changes in all migrant categories. The family reunion program was delivering a substantial flow of low- skilled, non-English-speaking, and relatively capital-poor migrants who often needed welfare assistance. A number of studies indicated high rates of unemployment benefits and old-age pensions going to migrants (Birrell 1993; Healy 1994; Evans 1994). Concern about the impact of these trends on the federal budget was high. One 1997 poll showed that 70 percent of a national sample indicated they were “very concerned” about “immigrants getting welfare on arrival” (White 1997: 13). To assert control over the family reunion category, Labor passed a “balance of family” rule in 1989 requiring that at least half of a family’s number of siblings had to be residing in Aus- tralia before a child could sponsor the migration of his or her parents. The numbers of sponsorships of parents did decline but began to rise again in the mid-1990s. In 1997 the Coalition (with Labor support) placed a quota

Click to return to Table of Contents G ARY P. FREEMAN / BOB BIRRELL 535 on the number of parents permitted to enter each year (Birrell 1997: 24). The government then cut visas for parents from 7,580 in 1996–97 to a planned 500 in 2000–01. The numbers of spouses and fiancé(e)s entering annually has also been reduced sharply since 1995–96, mainly through measures designed to cull applications whose bona-fides appeared to be sus- pect. The Coalition failed in its attempt to gain legislative approval for a cap on visas for spouses, but did cap those for fiancé(e)s (down about 50 percent from their peak). Spouse visas are now limited to two years, pending subsequent proof that the relationship is “genuine and continuing.” The numbers of independent skilled migrants were also reduced, in part as a response to data indicating that many skilled professionals could not obtain certification or, if they did, could not find professional-level jobs in Australia (Hawthorne 1994). The total skilled category (including siblings, who are assessed primarily on skill-based criteria) was 72,300 in 1990– 91. The numbers fell as low as 27,700 in 1993–94 before gradually rising again to 35,000 at the end of the decade (DIMA 1999: Fig. 2.3, p. 14). Most of the recent rise has been in the skilled program. Despite opposition from ethnic communities, both the Labor and Coalition governments have contributed to a tightening of the selection criteria for brothers and sisters. The numbers obtaining visas declined by more than half since the late 1980s, to around 9,000 at the end of the 1990s. Since 1999 applicants have had to demonstrate functional English before they can even be considered for selection. Finally, the business migration program, which had been plagued by charges of corruption and inefficiency, was effectively scuttled and re- named the business skills category in 1991–92. Major changes with respect to asylum seekers and illegal migrants took place as well. The government narrowed the basis for making on-shore humanitarian claims to keep the courts out of the process. The 1992 Migra- tion Act, passed under the Keating Labor government, empowered the Im- migration Department to detain all persons residing in Australia without authorization. All those who arrive illegally and apply for asylum must be detained while their cases are heard. Rights of appeal of initial decisions were drastically reduced. The Coalition (with Labor support) has legislated that those who arrive illegally (such as boat people) and are granted asylum will initially receive only temporary visas valid for three years. This means that in contrast to those selected off-shore as part of the humanitarian program or those who make asylum claims while in Australia legally, they receive only limited access to welfare benefits, cannot sponsor family members for migration, and cannot reenter if they depart Australia in the three-year period. Successive governments in the 1990s pruned back access to welfare benefits. The Labor government began the process in 1992 when it imposed a two-year bond on the sponsors of parents (A$3,500 for the principal ap-

Click to return to Table of Contents 536 D IVERGENT PATHS OF IMMIGRATION POLITICS plicant), which was repayable if no social security benefits were paid out during the time of the bond. It also required a nonreturnable A$822 health charge on all visa recipients in the parent category (Birrell 1997: 8). In 1994 Labor implemented a moratorium on the payment of most social security benefits for the first six months of residence for all migrants except those entering under the humanitarian categories. In 1996 the Coalition government (with Labor support) extended this moratorium to two years. In 2000 the moratorium was applied to New Zealanders. Some other migrant services, such as Adult Migrant English training courses, were put on a user- pays basis. These measures were justified in fiscal terms and carried out by the Department of Social Security, but ethnic leaders saw them as part of the reform of the immigration process. Finally, the strong commitment to multicultural activities, a mark of the Labor government in the late 1980s, waned. The Coalition government abolished the Office of Multicultural Affairs, which had been established in 1987 by Labor “as part of the attempt to cater to the ethnic lobby” (Grattan 1993: 137), its activities being folded into the Department of Immigration. The Bureau of Immigration, Population, and Multicultural Research, which had played an important role under Labor in providing research in support of a large immigration program, was also shut down. Table 1 summarizes the main differences in immigration policy trends in the United States and Australia. In respect to policy outcomes the two countries appear to be headed in opposite directions on the issue of numbers and with respect to family, skilled, and on-shore asylum claimants. In particular, Australia’s retraction of a number of family reunion rights puts it sharply at odds with the recent experience in the United States and Canada. Policy toward illegal migrants has become considerably tougher in both countries (the Bush administration may reverse this course), but Australia’s stance is more severe. Both countries have taken steps to limit migrant use of public benefits, but the US initiative, to the extent it applied to migrants already resident, has been blocked or reversed. Both countries have also expanded access for temporary skilled migrants. Having no formal multicultural policy, the US government has taken fewer visible steps than Australia’s in this arena. The picture is, therefore, mixed and it is impossible to state how long lasting present regulations and policies will prove to be. Nevertheless, the evidence supports the conclusion that the two countries are on divergent paths. Why?

The interplay of interests, rights, and states

The immigration contexts of the United States and Australia, while exhibit- ing broad similarities, are in some respects distinct. This, in itself, may account for some of the disparity between the two. All rich countries experienced intensifying migration pressures in the 1980s and 1990s (OECD 1999).

Click to return to Table of Contents G ARY P. FREEMAN / BOB BIRRELL 537

TABLE 1 Trends in US and Australian immigration policy in the 1990s relative to the 1980s Policy United States Australia Annual admissions Up by one-third. Down around one-third. Family migration Sponsors financially accountable for Parent sponsors must post bond; “balance immigrants they sponsor, but all of family” rule; annual cap on parental family preferences remain. sponsorships and fiancé(e)s; spouse visas limited to two years initially. Rights to sponsor brothers and sisters sharply curtailed. Skilled migration Occupational category increased Independent category reduced, points in 1990. test tightened to require English test, preference to those educated in Australia. Refugee/humanitarian Off-shore No changes. No changes. On-shore Temporary protected status for Central Asylum applicants who arrive in Americans and Haitians extended and Australia illegally placed in detention then converted to permanent status; while claims assessed. Unlike those who asylum seekers can be excluded if they arrive legally, if granted asylum visas have no credible case or if they arrive limited initially to three years and limits with no documents or enter illegally. placed on access to welfare, family reunion rights, and travel overseas. Reduced scope for filing on-shore humanitarian claims; reduced grounds of appeal against a negative decision. Temporary work visas Temporary skilled worker program Temporary skilled worker program expanded. deregulated since August 1996. Employers may sponsor as many qualified workers as they wish. No welfare benefits are provided. Migrant welfare benefits Most federally funded means-tested Sponsors must post bond for parents benefits denied those entering after (see above); two-year moratorium on 1996 except for exempt categories welfare benefits for all migrants, except (e.g., refugees); income of sponsors of those entering under the humanitarian family members is “deemed” (see above). program. Illegal migrants Vast expansion of men and materiel to 1992 Migration Act gives power to control border; entry with false documents detain all persons illegally in country; debars entry for five years; pilot program detention mandatory for those entering for employer verification of work autho- Australia illegally. rization; expedited deportation procedures. Multiculturalism Dismantlement of government agencies supporting multiculturalism.

Nevertheless, some countries are more susceptible to these pressures than others. The United States is more prone to migration pressure than Austra- lia because of its land borders with and close proximity to sending countries, Mexico in particular, and the powerful attraction its economy and culture exert on the world’s potential migrants. The expanding American economy of the 1990s only added strength to preexisting “pull” factors. Moreover, its image as a land of opportunity for immigrants is central to its

Click to return to Table of Contents 538 D IVERGENT PATHS OF IMMIGRATION POLITICS national myth and, arguably, to its role as a global power. Australia, also an attractive destination, is situated in the vicinity of some of the world’s largest population centers; but it is often a second choice for migrants preferring to go to North America, and its island status gives it a decided advantage in controlling unwanted entries. On the other hand, Australia’s ability to run its immigration policy as it sees fit is constrained by the perceived necessity to take into account the reactions from its Asian neighbors with whom its hopes for economic growth are intertwined. Nevertheless, with respect to external considerations, Australia is probably more favorably situated than the United States to control migration if the government chooses to do so. The evidence suggests that Australia has reacted more decisively than the United States, despite being under less pressure, making the analytical puzzle all the more intriguing.

Interests

The principal interests associated with immigration policy are economic, ethnic, or status oriented (in the case of the intelligentsia), although these distinctions are somewhat arbitrary given that persons with ethnic or intellec- tual interests in immigration typically have clear economic stakes as well. Serious challenges to “client politics” emerged in both countries in the 1990s. First, entrepreneurs from outside the political establishments emerged. These sought to break up the prevailing consensus and to mobilize majorities against mass immigration and policies promoting multiculturalism or, in the case of some activists in the United States, against perceived government complacency toward illegal migration. Second, the terms in which immigration issues were debated changed. In both countries immigration became entangled with arguments over welfare use and fiscal burdens. In public discussions immigration moved from being seen primarily as a distributive policy with many winners and few if any discernible losers to a redistributive policy that entailed zero-sum choices in which the gains of some come at the expense of others. Redistributive policies tend to generate an interest-group form of political interaction that involves large coalitions arrayed on either side. These new pressures challenged client politics more seriously in Aus- tralia. There, the debate about immigration was also transformed by the extent to which issues concerning the alleged economic costs and benefits became overshadowed by the link between migration and multiculturalism; by the late 1980s the two were inextricably intertwined. To most people more immigration meant “more” multiculturalism. When the CAAIP reported in 1988, its major point was that if high immigration was to be sustained (which the Committee favored) it had to regain public support. According to the Committee, such support could be mustered only if immigration and multi-

Click to return to Table of Contents G ARY P. FREEMAN / BOB BIRRELL 539 culturalism were separated in the public mind. As the matter stood, “the program is not identified in the public mind with the national interest” (CAAIP 1988: xi)—by which the Committee meant that most people saw migration as benefiting exclusively the migrant community. John Howard, as Leader of the Opposition during the Hawke and Keating governments, became engaged in the debate when he sought to capitalize on public concerns that their identity as Australians was under challenge. Behind these concerns was a pronounced increase in the Asian component of the migration intake. This intervention brought to the surface a highly emotive issue with the potential to blast immigration politics out of the narrow mode of client politics. In the United States, many factors contributed to the eruption of legislation against illegal migration and the decision to strip legal migrants of most of their federally financed social benefits. Perhaps the most important was that key political actors began to see undocumented migrants and their receipt of welfare benefits as a serious drain on federal and state budgets. Rather than focusing discussion on the economic contributions of migrants, as was customary, the argument shifted to redistributive issues triggered by concern over immigrants’ use of public benefits but linked inextricably to the porousness of the Southwest border. This debate took place at both the state and federal levels, as is indicated by the California campaign for Propo- sition 187 and later initiatives against affirmative action and bilingual education and by Congressional passage of welfare reform and immigration legislation in 1996. The emergence of redistributive conflicts should, in the Wilson/Freeman model outlined above, produce interest-group politics in which coalitions organized around the benefits of particular policies challenge coalitions organized around the costs. For a while immigration politics took on some of the more contested character of interest-group politics, but only for a few issues. Both the revolt against illegal immigration in California and the Buchanan presidential campaign were successful in mobilizing discontent over immigration problems, but only temporarily. Indeed, they generated a backlash among Hispanics, leading to increased voter registration efforts. Many observers believed that this backlash contributed to the reelection of Bill Clinton in 1996 and to the losses suffered by Republicans in the 1998 Congressional elections. In any event, both the Democratic and Republican parties were eager to dissociate themselves from anti-immigrant rhetoric; and Buchanan, seeking the presidency once again in 2000 first as a Repub- lican and later as a Reform Party candidate, received less than one percent of the vote. The highly partisan immigration politics that characterized the 1994–98 period may well be a thing of the past. In the 2000 presidential campaign, both major party candidates endorsed expansive immigration policies (Krikorian 2000). In policy terms, the only areas seriously affected by

Click to return to Table of Contents 540 D IVERGENT PATHS OF IMMIGRATION POLITICS the 1996 legislation dealt with illegal migration, welfare benefits for legal immigrants, and the rights of asylum seekers to judicial appeal—all cast primarily in redistributive terms. Although asylum policy was not debated in terms of its monetary costs, it got conflated with the more general fiscal argument over illegal entry. Despite a myriad of proposals from various quar- ters, no significant constituency developed against the legal admissions policy, which continued to be seen as producing important benefits with few costs. Although the American client system withstood the assault from anti- immigration entrepreneurs, it nonetheless was changed through the introduction of new participants. When the dust had settled, the interest-group configurations based around immigration issues produced enhanced support for expansion rather than opposition. The most important new players represented the rapidly growing and highly valued technology sector. They lobbied successfully for expansion of temporary skilled-worker programs. More surprisingly, organized labor, traditionally the most influential and legitimate opponent of illegal migration and long-time critic of temporary- worker programs and mass immigration, in spring 2000 announced its support for a general amnesty for most illegal aliens in the country. This was widely seen as recognition by the AFL-CIO that the large illegal labor pool constitutes a rich target for labor unions seeking to reverse their decline in membership. Labor strategists seem to have concluded that skilled migrants compete for jobs against American workers (labor opposed expansion of the H1-B temporary visa program for skilled labor). Unskilled illegal migrants, on the other hand, are complementary and should be legalized and organized, a development that is seen in Europe as well (Haus 1995, 1999; Watts 2000). By the late 1980s Australia had established a bipartisan client politics of immigration built around the support of business, the ethnic communities, and, more generally, intellectuals. (See the articles by McAllister, Warhurst, Jupp, and Betts in Jupp and Kabala 1993; Betts 1988, 1999.) Events of the last decade demonstrate, however, that the system was surprisingly vulnerable to attack. The advent of the conservative Coalition government in 1996 and the emergence of the anti-immigration One Nation party (detailed below), while not the direct causes of policy shifts, nonetheless indicate how much room there is in Australia for attacks against the pro-immigration consensus, and how fragile that consensus is. The bipartisan nature of immigration policy leading up to the 1990s has been well established (McAllister 1993; Hardcastle et al. 1994). At first glance it would appear that with the election of the Coalition government, the era of bipartisanship had ended. But we believe this distorts the reality of the origins within the Australian Labor Party (ALP) of many recent policy trends and Labor support, however grudging, of major Coalition initiatives. Nevertheless, the limits of bipartisanship are being tested, and the central

Click to return to Table of Contents G ARY P. FREEMAN / BOB BIRRELL 541 analytical question is why the party consensus on immigration shifted markedly to the right in the 1990s. The main explanation is linked to the earlier discussion of the transformation of the ways immigration issues were perceived. Concern about the costs (particularly in the early 1990s when unemployment exceeded 10 percent and government budgets were being squeezed) was particularly intense in Australia because of the greater relative generosity of welfare benefits and because the welcome mat had been so much further extended than in the United States. The debate in the late 1980s about immigration and identity and accompanying opinion polls showing that most voters opposed the prevailing policies gave politicians new cause to pay attention to the issue. Both major parties feared the other might successfully exploit public concern. In this context, proposals to cut components of migration thought responsible for high welfare bills and to more tightly target the program to skills needed in Australia were looked on much more favorably by both Labor and Coalition governments. A further factor, which we explore later, is that once the policy tide turned, the incumbent governments were in a far more powerful position to implement restrictive measures than is the case in the United States. Client politics tends to prevail in Australia as long as public opinion against immigration is not mobilized, so that the governing party believes it runs no electoral risks in giving immigrant advocates what they want. When this favorable environment changes, the influence of immigration clients declines accordingly. As in the United States, ethnic groups in Australia had established closer working relationships with one of the major parties (the ALP). But they proved ineffective during the 1990s in blocking the aforementioned legislative changes. With Labor out of office and the public climate on immigration less favorable, pro-immigration groups were power- less to prevent major retrenchments. But why did the Coalition, with strong business leanings, favor policies contrary to the preferences of employers? Immigration to Australia is generally legitimized through economic arguments, particularly those that link migration to an economic strategy that seeks to tie Australian development to growth in the Asian region (Warhurst 1993; Healey 1997; Betts 1999). Even so, annual admissions have always been influenced by the unemployment rate. This legitimating rationale, therefore, is vulnerable if general unemployment rises, if migrants experience higher rates of unemployment than natives, if welfare costs of maintaining migrants expand, and if the government of the day is less convinced of the wisdom of tying Australia’s fate to that of its Asian neighbors. Short-term considerations like these can override the longer-term faith in the benefits of immigration. But for the Coalition, the overriding policy determinant was the legacy of the late-1980s transformation of the immigration debate. The new government’s main objective was to defuse public concerns about links between migration, multi-

Click to return to Table of Contents 542 D IVERGENT PATHS OF IMMIGRATION POLITICS culturalism, and challenges to Australians’ sense of identity. Its strategy was to closely target migration to Australia’s economic needs where possible, and to distance itself from past multicultural policies. This it did by abolish- ing agencies that were symbols of government support for multiculturalism (like the Office of Multicultural Affairs) and by simply ignoring the issue. If these concerns had not been high in the government’s priorities, the Coali- tion might well have responded to the demands of business interests for much higher intakes. On gaining office in March 1996, the Coalition government took immediate action in announcing its welfare proposals and restrictive immigration measures. It was already well down this path when in September 1996 Pauline Hanson, elected to the House of Representatives as an Inde- pendent, made her maiden speech in Parliament. Subsequent developments were interpreted within the government as vindicating its policy initiatives. Hanson’s speech contained the inflammatory line that Australia was in danger of “being swamped by Asians.” In the accompanying furor, Hanson achieved brief but spectacular success with her One Nation party (formed in April 1997) in the June 1998 Queensland elections, where one in every four voters supported it. In the October 1998 federal election, One Nation ran candidates for 139 of the 148 House of Representative seats, winning 8.4 percent of first-preference votes, but no seats. This experience proved that a significant minority of voters could be mobilized around protests against mainstream politicians. Much of Hanson’s support appears to have come from people whose concerns included migration and Aboriginal issues (Goot 1998: 68–70). The One Nation party’s challenge had collapsed by the end of the decade, mainly because of organizational deficiencies. But the Coalition’s moves to neutralize the migration and multicultural issue also appear to have diminished public concerns on immigration policy matters (Goot 2000). When business interests, led by the Business Council of Australia, be- latedly tried to counteract Coalition policy (Birrell 1998: 40), they made little headway. The government was willing to admit as many skilled migrants as met their rather stringent criteria—knowledge of English and credentials that would be accepted in Australia and were in short supply in the Austra- lian labor market—but it was not prepared to sanction a general increase in admissions without regard to skills. Our analysis of the interplay of interests suggests that public opinion in Australia is more easily mobilized around immigration controversies than in the United States and that Australian politicians, consequently, appear to be more consistently worried about a possible outbreak of anti-immigrant populism. Moreover, as we show below, the institutional framework and legislative process make it more difficult for pro-immigration interests in Australia to have their way if a governing party takes a restrictive stance.

Click to return to Table of Contents G ARY P. FREEMAN / BOB BIRRELL 543

This is just the opposite of the situation in the United States, where political elites routinely ignore unorganized popular concerns about immigration but are sharply attuned to the demands of organized advocates of immigration causes to whom the institutional framework gives ready access.

Rights

The rights perspective holds that national states are embroiled in an international system of organizations, regimes, treaties, and agreements increasingly devoted to universalistic human rights norms that limit their ability to pursue narrow national interests with regard to immigration. Quite apart from these external pressures, domestic arguments over immigration policy also tend to be carried out in the language of rights. This is more than mere rhetoric or convention. In this line of reasoning courts play a leading role in establishing and protecting immigrant rights against the policy preferences of governments. Altogether, rights sharply constrain the pursuit of market benefits through state immigration policies. We use the term rights to mean not only legal rights that are recognized in constitutions and by the courts and that might be seen to follow from principles of natural justice, but also entitlements that derive from a political understanding of rules of fair play and reciprocity. The evidence of the two countries suggests that the claims of this perspective have been overstated. Rights obviously matter, and more now than in the past, but they are not as important as interests. Moreover, democratic states vary considerably in the extent to which liberal norms are embedded in their institutions and practices (see Joppke 1999). Under the appropriate circumstances, states can enact immigration policies that deny rights to specific categories of migrants: legal immigrants, illegal immigrants, or asylum seekers. Governments in both Australia and the United States challenged the “rights” of migrants, but these efforts were more strenuously and successfully contested in the latter than the former. Even so, the story is sufficiently mixed to defy easy generalization. Both the United States and Australia are liberal democracies fully integrated into the international regimes that in the rights perspective are seen as markedly reducing the options of nations in responding to the claims of migrants. By most accounts, however, the United States is more firmly “embedded” in this system and is a more thoroughly liberal regime. The culture is strongly individualist if not as thoroughly liberal as commonly argued (Hartz 1955; Smith 1997). Individual rights are delineated and given primacy in the Bill of Rights of the US Constitution and are backed by a system of strong judicial review. The fourteenth amendment, which guar- antees due process of law, applies not only to the federal government, but increasingly to the states as well. The civil rights movement of the 1960s,

Click to return to Table of Contents 544 D IVERGENT PATHS OF IMMIGRATION POLITICS which focused on the legal situation of African-Americans, accelerated and institutionalized the national preoccupation with the politics of rights, spawn- ing in its wake movements for the protection of a host of additional minorities. Immigration was explicitly linked to the cause of civil rights when proponents of the 1965 amendments to the McCarran-Walter Act of 1952 argued that the national origins quota system, by discriminating on the basis of national origins, violated the spirit and letter of the Constitution. Af- firmative action programs and the Voting Rights Act of 1965, originally targeted toward historically oppressed minorities, now apply to many recent immigrant groups. A powerful immigration Bar, benefiting from a decid- edly litigious culture, peppers the administration with lawsuits on behalf of migrants and asylum seekers that challenge the constitutionality of those measures that the well-financed pro-immigration lobby is unable to defeat in the Congress. Australia, on the other hand, has been described as deriving from a working-class fragment of British society and producing a collectivist culture (Rosecrance 1964; see Freeman and Jupp 1992). Liberal individualism has been slower to develop in a country whose founding, in contrast to that of the United States, required the active efforts of strong states (first in the form of British colonial rule and, after federation, the Commonwealth government). Governments always sought to organize and control immigration to Australia, both because state action was necessary to promote migration and because in the collectivist Australian political culture the state shouldered the responsibility to ensure successful adjustment upon arrival. This meant that if immigrants could not make a go of it, the state budget bore the consequences (Jupp 1992). The Liberal Party was not based on individual rights in the European liberal tradition. The party saw itself as a representative of business but was not a proponent of laissez faire. The Liberal Party, not the Australian Labor Party, engineered and oversaw the protectionism, limited welfare, immigration restriction, and labor arbitration system that was central to Australia’s political economy until the early 1980s (F. G. Castles 1988; Birrell 1995). The Australian Constitution has no bill of rights, nor have the courts exer- cised the same extensive powers of judicial review as the US Supreme Court. Civil rights gained ground as a political organizing concept in recent years, with respect, first, to the Aborigines and, later, to ethnic minorities. But this movement was late in coming and is weak compared to that in the United States. Significantly, a major advocate for migrant rights in Austra- lia has been the Human Rights and Equal Opportunity Commission, itself a government agency. Its advocacy role has diminished since the Coalition came to power, and its more activist appointments have been replaced. We would anticipate, from all this, that migrants would be accorded more—and better-protected—rights in the United States than in Australia

Click to return to Table of Contents G ARY P. FREEMAN / BOB BIRRELL 545 and, as our case studies show, this is largely the case. The main inroads made by recent American policy against the rights of migrants (summarized in Table 1) involve deportation powers, the rights of asylum seekers, and the prohibition against legal immigrants’ receiving most public benefits for the first five years of residency. Attempts to limit family reunion rights of adult siblings in the United States were roundly defeated. Although most of these measures deal with persons who have not been admitted legally, several hit legal entrants hard. Despite the fact that the Constitution gives Congress “plenary” powers over immigration policy, the US federal courts have intervened repeatedly to overturn measures that they see as violating individual rights of migrants, legal and illegal. We noted above that the courts struck down a number of features of the 1996 federal legislation, as well as almost the en- tirety of the 1994 California referendum (Proposition 187). But the courts up- held as constitutional the denial of public benefits to legal immigrants. The American case tends to confirm, though imperfectly, the propositions that rights once given cannot be taken back and that rights extended to one immigrant group must be given to others. Section 245(i) of the immigration law was extended and one million unauthorized immigrants were “grandfathered” under its provisions. Temporary protected status deadlines and coverage were extended, and family sponsorship of siblings was maintained. While the welfare rights of prospective migrants have been removed, at least for the moment, those of migrants in the US at the time of passage have been restored. The Australian government took more aggressive steps than did its American counterpart to limit judicial oversight of immigration decisionmaking. The administrative reforms of 1989, which amounted to “judge- proofing” the immigration department, have “almost certainly limited the number and range of immigration decisions presently going to the Federal Court” (Cronin 1993: 103). This is reflected in the ability of the government to sharply reduce immigrant entitlements. Perhaps the most remark- able aspect of the Australian case is the large number of changes in the law respecting family admissions. Migrant access to state-paid benefits was also trimmed. Determined steps were taken to deter unauthorized entry by asylum seekers. Retrenchment in multiculturalism was not necessarily about rights but did represent a statement about recognition of the status of minorities and members of minority groups, and was a rollback of benefits that had come to be seen as entitlements. Rights-based liberalism is more fully embedded in the United States than in Australia, with the result that US policymakers were less able to limit the spread or reverse the recognition of immigrant rights. The rights perspective usefully accounts for some of the differences in the two countries, but it also suggests the need for a more nuanced statement of the thesis than is normally given.

Click to return to Table of Contents 546 D IVERGENT PATHS OF IMMIGRATION POLITICS

States

Our discussion of states deals both with policy preferences and strategic initiatives deriving from inside state agencies and with the impact of state institutions and procedures on the shape of immigration policy. With respect to the first, the Australian authorities were much more influential in the formulation, adoption, and implementation of policy changes than were their American counterparts. Numerous legislative and administrative initiatives were developed in the United States, but many were thwarted by Congress or were negated when the president reversed his previous stances on key issues. The extent to which outside interests, working through members of the legislature, dominated decisionmaking is seen in the spectacle of President Clinton signing legislation of which in large part he disapproved. Wanting to clamp down on undocumented immigration, he signed a bill that went much further than he wished. Hoping to amend the policy on legal immigration, he saw his proposals repudiated, and eventually he de- serted them himself. The INS, under commissioner Doris Meissner, enhanced its profile in participating in the planning of immigration policy, but the scorecard would have to show that federal agencies did not control agenda- setting on immigration, even less the details of the final legislation. The situation in Australia is strikingly different. The Department of Immi- gration played the central role in laying out the contours of immigration policy changes, providing the minister with ideas and data. The current minister, Phillip Ruddock, has held the immigration portfolio since the early 1990s when he was a member of the Opposition and has been minister for a lengthier period than any of his Labor predecessors, who moved in and out of office quickly.3 The result is a strong, knowledgable minister willing to use his departmental advisers to reshape policy. If the minister has had political or even ideological reasons for preferring a variety of restrictive measures, his department has been interested less in restriction, per se, than in maintaining the credibility of the immigration program. This has especially been the case given that the costs of migrants’ failure to thrive in the private employment market fell on the federal government budget. As indicated, the department has sought to insulate itself from political, ethnic, and judicial interference through the creation of a “rule-based decision-making regime” (Bruer and Power 1993: 122), grounded on the assumption that popular support for the program depends on the widespread belief that it is well managed. Added to this propi- tious mix, Prime Minister Howard was not a proponent of mass immigration, especially if it originated largely from Asia and seemed likely to alter the cultural character of Australia. The Coalition has been prepared to defy the interest groups that were the dominant clients in the 1980s because of its larger concern to quiet populist fears. It has not had to worry that the Labor opposition would try to take advantage of its neglect of this constituency because Labor, too, fears offending the larger electorate.

Click to return to Table of Contents G ARY P. FREEMAN / BOB BIRRELL 547

When an Australian government is intent on action, even if powerful and articulate interests are involved, it is in a far better position to implement its policies than is its American counterpart. Major pieces of new legislation in Australia often emerge from the bureaucracy without any inter- mediary hearings that would allow affected interests to mobilize opposition before the matters enter Parliament. In regard to immigration legislation, the Department of Immigration initiated legislation affecting family reunion and tightening the policy for skilled workers. The Department’s chief priority is to manage the immigration program with a minimum of political flak. As a result it is strongly attuned to controlling issues and to ensuring that it has the legislative powers to exercise this control. These priorities frequently clash with those of the ethnic and humanitarian constituencies. The hearings that precede legislation in the US Congress and the capacity of Con- gressional committees to delay floor votes have at most weak and intermit- tent parallels in the Australian Parliament. Hence restrictive Australian legislation is not subject to the same scrutiny or obstacles that are present in the United States. Although the Coalition government does not have a majority in the Senate and thus in principle its legislation can be denied, this usually does not happen. Furthermore, the government announces annual decisions about how many migrants will be granted visas in the various categories without reference to Parliament. Our attempt to separate the role of state authorities as actors and as institutional framework is at times necessarily artificial. The institutional configura- tion of states in part determines the ability of state elites to manage and dominate policy development. Australian institutions foster a strong state role in initiating policies, whereas those in the United States vitiate such a role. Gov- ernments in Australia can depend on the loyalty of their members in Parlia- ment and are relatively insulated from pressure groups in a way that governments in the United States are not. Former President Clinton not only was unsure of the support of his own party members in Congress, but his White House was itself open to intense lobbying from a multiplicity of interests. He at times supported and at other times abandoned the INS; he at first endorsed and later jettisoned the recommendations of the US Commission on Immigra- tion Reform. In contrast, the policy changes instituted by the Coalition in Aus- tralia, for all the clamor they have produced from the press and public, are still in place and the government is holding firm.

Conclusion

The chief puzzle facing scholars of comparative immigration politics is why public opinion on immigration issues, which in all the liberal democracies is relatively conservative and favors restrictive measures, is not more readily translated into policy decisions. The disjuncture between opinion and policy is greater in the former settler societies of Australia, Canada, and the United

Click to return to Table of Contents 548 D IVERGENT PATHS OF IMMIGRATION POLITICS

States than in Western European countries, where immigration is of more recent vintage. Seen from this angle, we might well restate our question as being not why Australia has deviated from a liberal immigration path, but why the United States and Canada have not done so as well. The principal reason we have adduced is that popular opinion in Aus- tralia is more readily mobilized and volatile than in the United States and more efficiently translated into pressures on governments. Australian politicians, therefore, need to attend to the public’s anxieties in a way that their American counterparts do not. The Australian public is susceptible to arguments about the threat posed by immigration and multiculturalism for national identity. Such fears are less palpable to Americans. In addition, the costs to the public treasury of sustaining the immigration policy appear to be more salient in Australia than in the United States, where immigration is widely believed to make a positive contribution to the country’s economy and future whatever the short-term costs. The greater concern of the Australian electorate about immigration issues is more easily brought to bear on politicians because the institutional channels are more direct than in the United States, where organized interest groups have significant advantages of access over those available to the general public. In Australia, under the right circumstances, the threat of a populist backlash can overwhelm the entrenched interests supporting immigration and multiculturalism and, at least for a time, retard or reverse policy trends. Scholarship on comparative immigration policy tends to be under-theorized and overly value-laden. We have sought to account for the experience of two main immigrant-receiving states by setting out competing analytical frameworks that can be tested against empirical data. Our three theoretical perspectives were only roughly delineated, and the evidence adduced is admittedly thin in places. Nevertheless, we believe the concepts of interests, rights, and states provide a useful starting place for understanding immigration politics that other researchers will no doubt be able to extend and refine.

Notes Support for this project came from the Public 2 After a well-funded and intense lobby- Policy Clinic of the Department of Govern- ing effort on the part of the technology sector, ment, University of Texas at Austin, and the in September 2000 Congress passed and Presi- Visiting Collaborative Scholar Program of dent Clinton signed a bill to expand the visas Monash University. available for skilled temporary migrants. 1 These were the Illegal Immigration Re- 3 For discussion of the role of immigra- form and Immigrant Responsibility Act, the tion ministers in the Hawke governments, see Anti-Terrorism and Effective Death Penalty Parkin and Hardcastle (1990). Act, and the Personal Responsibility and Work Opportunity Reconciliation Act, all passed in 1996.

Click to return to Table of Contents G ARY P. FREEMAN / BOB BIRRELL 549

References

Betts, Katharine. 1988. Ideology and Immigration: Australia 1976 to 1987. Melbourne: Melbourne University Press. ———. 1993. “Public discourse, immigration and the new class,” in James Jupp and Marie Kabala (eds.), The Politics of Australian Immigration. Canberra: Bureau of Immigration Research, pp. 222–239. ———. 1996. “Immigration and public opinion in Australia,” People and Place 4(3): 9–20. ———. 1999. The Great Divide. Sydney: Duffy and Snellgrove. Birrell, Robert. 1992. “Problems of immigration control in liberal democracies: The Austra- lian experience,” in Gary P. Freeman and James Jupp (eds.), Nations of Immigrants: Australia, the United States, and International Migration. New York and Melbourne: Ox- ford University Press, pp. 23–40. ———. 1993. “Unemployment benefit dependency amongst newly arrived migrants,” People and Place 1(1): 19–22. ———. 1995. A Nation of Our Own. Melbourne: Longman. ———. 1997. Immigration Reform in Australia: Coalition Government Proposals and Outcomes Since March 1996. Melbourne: Centre for Population and Urban Research, Monash University. ———. 1998. “Skilled migration policy under the coalition,” People and Place 6(4): 37–51. Blainey, Geoffrey. 1984. All for Australia. Sydney: Methuen Haynes. Bruer, Jeremy and John Power. 1993. “The changing role of the Department of Immigra- tion,” in James Jupp and Marie Kabala (eds.), The Politics of Australian Immigration. Canberra: Bureau of Immigration Research, pp. 105–123. CAAIP. 1988. Commitment to Australia. Report of the Committee to Advise on Australia’s Immigration Policy. Canberra: Australian Government Printing Service. Castles, Francis G. 1988. Australian Public Policy and Economic Vulnerability. Sydney: Allen & Unwin. Castles, Stephen et al. 1992. Mistaken Identity: Multiculturalism and the Demise of Nationalism in Australia (3rd ed.). London: Pluto Press. Center for Immigration Studies. 1995. Immigration-Related Statistics. Washington, DC. Cronin, Kathryn. 1993. “A culture of control: An overview of immigration policy-making,” in James Jupp and Marie Kabala (eds.), The Politics of Australian Immigration. Canberra: Bureau of Immigration Research, pp. 83–104. DeLaet, Debra L. 2000. U.S. Immigration Policy in an Age of Rights. Westport, CT: Praeger. Department of Immigration and Multicultural Affairs (DIMA). 1999. Population Flows: Im- migration Aspects. Canberra: Australian Government Printing Service. Espenshade, Thomas J. and Maryann Belanger. 1997. “US public perceptions and reactions to Mexican migration,” in Frank D. Bean et al. (eds.), At the Crossroads: Mexican Migra- tion and US Immigration Policy. Lanham, MD: Rowman and Littlefield, pp. 227–262. Espenshade, Thomas J. and Katherine Hempstead. 1996. “Contemporary American attitudes toward U.S. immigration,” International Migration Review 30(2): 535–570. Evans, Samantha. 1994. “The assurance of support and welfare dependency among newly arrived migrants,” People and Place 2(4): 57–59. Freeman, Gary P. 1995. “Modes of immigration politics in liberal democratic states,” Inter- national Migration Review 29(4): 881–902. ———. 1998. “Reform and retreat in United States immigration policy,” People and Place 6(4): 1–11. ———. 1999. “From selection to regularisation in U.S. immigration policy,” Growth 47 (No- vember): 100–111. Freeman, Gary P. and James Jupp. 1992. “Comparing immigration policy in Australia and the United States,” in Gary P. Freeman and James Jupp (eds.), Nations of Immigrants: Australia, the United States, and International Migration. New York and Melbourne: Ox- ford University Press, pp. 1–20.

Click to return to Table of Contents 550 D IVERGENT PATHS OF IMMIGRATION POLITICS

Garnaut, Ross. 1990. Australia and the Northeast Asian Ascendancy. Report to the Prime Min- ister and the Minister of Foreign Affairs and Trade. Canberra: Australian Government Publishing Service. Goot, Murray. 1998. “Hanson’s heartland: Who’s for one nation and why,” in Tony Abbott et al. (eds.), Two Nations: The Causes and Effects of the Rise of the One Nation Party in Aus- tralia. Melbourne: Bookman. ———. 2000, “More ‘relaxed and comfortable’: Public opinion on immigration under Howard,” People and Place 8(3): 46–59. Grattan, Michelle. 1993. “Immigration and the Australian Labor Party,” in James Jupp and Marie Kabala (eds.), The Politics of Australian Immigration. Canberra: Bureau of Immi- gration Research, pp. 127–143. Hardcastle, Leonie et al. 1994. “The making of immigration and refugee policy: Politicians, bureaucrats, and citizens,” in Howard Adelman et al. (eds.), Immigration and Refugee Policy: Australia and Canada Compared. Toronto: University of Toronto Press, pp. 95–124. Hartz, Louis. 1955. The Liberal Tradition in America: An Interpretation of American Political Thought Since the Revolution. New York: Harcourt, Brace. Haus, Leah. 1995. “Openings in the wall: Transnational migrants, labor unions, and U.S. immigration policy,” International Organization 49(2): 285–313. ———. 1999. “Labor unions and immigration policy in France,” International Migration Re- view 33(3): 683–716. Hawthorne, L. 1994. Labour Market Barriers for Immigrant Engineers in Australia. Canberra: Bureau of Immigration and Population Research. Healey, Ernest. 1994. “The management of the immigration portfolio under Senator Bolkus,” People and Place 2(4): 37–44. ———. 1997. “The political economy of immigration and multicultural policies under La- bor in the 1980s and early 1990s,” Doctoral thesis, Department of Anthropology and Sociology, Monash University, Australia. Hollifield, James F. 1992. Immigrants, Markets, and States: The Political Economy of Postwar Eu- rope. Cambridge, MA: Harvard University Press. ———. 1994. “Immigration and republicanism in France: The hidden consensus,” in Wayne A. Cornelius, Philip L. Martin, and James F. Hollifield (eds.), Controlling Immigration: A Global Perspective. Stanford, CA: Stanford University Press, pp. 143–175. ———. 2000. “The politics of international migration: How can we ‘bring the state back in’?” in Caroline B. Brettell and James F. Hollifield (eds.), Migration Theory: Talking Across Disciplines. New York: Routledge. Jacobson, David. 1996. Rights Across Borders: Immigration and the Decline of Citizenship. Balti- more: Johns Hopkins University Press. Joppke, Christian. 1997. “Asylum and state sovereignty,” Comparative Political Studies 30(3): 259–298. ———. 1998. “Why liberal states accept unwanted immigration,” World Politics 50(2): 266– 293. ———. 1999. Immigration and the Nation-State: The United States, Germany, and Great Britain. Oxford: Oxford University Press. Jupp, James. 1992. “Immigrant settlement policy in Australia,” in Gary P. Freeman and James Jupp (eds.), Nations of Immigrants: Australia, the United States, and International Migration. New York and Melbourne: Oxford University Press, pp. 130–144. ———. 1993. “The ethnic lobby and immigration policy,” in James Jupp and Marie Kabala (eds.), The Politics of Australian Immigration. Canberra: Bureau of Immigration Research, pp. 204–221. Jupp, James and Marie Kabala (eds.). 1993. The Politics of Australian Immigration. Canberra: Bureau of Immigration Research. Krikorian, Mark. 2000. “‘Me too’ on immigration: The Republican Party’s kinder, gentler approach to immigration continues,” National Review Online (1 August).

Click to return to Table of Contents G ARY P. FREEMAN / BOB BIRRELL 551

Legislative Review Advisory Group. 1998. Not Just Numbers. Ottawa: Citizenship and Immi- gration Canada. Massey, Douglas S., et al. 1993. “Theories of international migration: A review and appraisal,” Population and Development Review 19(3): 431–465. ———. 1994. “An evaluation of international migration theory: The North American case,” Population and Development Review 20(4): 699–751. ———. 1998. Worlds in Motion: Understanding International Migration at the End of the Millen- nium. Oxford: Clarendon Press. McAllister, Ian. 1993. “Immigration, bipartisanship, and public opinion,” in James Jupp and Marie Kabala (eds.), The Politics of Australian Immigration. Canberra: Bureau of Im- migration Research, pp. 161–178. OECD. 1999. Trends in International Migration. Paris: SOPEMI. Parkin, Andrew and Leonie Hardcastle. 1990. “Immigration policy,” in Christine Jennett and Randal G. Stewart (eds.), Hawke and Australian Public Policy: Consensus and Restruc- turing. Melbourne: Macmillan. Portes, Alejandro. 1999. “Immigration theory for a new century: Some problems and opportunities,” in Charles Hirschman, Philip Kasinitz, and Josh DeWind (eds.), The Hand- book of International Migration: The American Experience. New York: Russell Sage Foun- dation, pp. 21–33. Rosecrance, Richard. 1964. “The radical culture of Australia,” in Louis Hartz (ed.), The Found- ing of New Societies: Studies in the History of the United States, Latin America, South Africa, Canada, and Australia. New York: Harcourt, Brace, and World, pp. 275–318. Ruggie, John. 1992. “International regimes, transactions and change: Embedded liberalism in the postwar economic order,” International Organization 36: 379–415. Simon, Rita J. and James P. Lynch. 1999. “A comparative assessment of public opinion toward immigrants and immigration policies,” International Migration Review 33(2): 455– 467. Smith, Rogers M. 1997. Civic Ideals: Conflicting Visions of Citizenship in U.S. History. New Ha- ven, CT: Yale University Press. Soysal, Yasemin. 1994. Limits of Citizenship: Migrants and Postnational Membership in Europe. Chicago: University of Chicago Press. US Commission on Immigration Reform. 1994. U.S. Immigration Policy: Restoring Credibility: A Report to Congress. Washington, DC: US Commission on Immigration Reform. ———. 1995. Legal Immigration: Setting Priorities: A Report to Congress. Washington, DC: US Commission on Immigration Reform. Warhurst, John. 1993. “The growth lobby and its opponents: Business, unions, environ- mentalists and other interest groups,” in James Jupp and Marie Kabala (eds.), The Politics of Australian Immigration. Canberra: Bureau of Immigration Research, pp. 181– 203. Watts, Julie R. 2000. An Unconventional Brotherhood: Union Support for Liberalized Immigration in Europe. San Diego: Center for Comparative Immigration Studies. White, Sally. 1997. The Silent Majority III: The Everyday Problems of the Average Australian. Melbourne: A Clemenger Report. Wilson, James Q. (ed.). 1980. The Politics of Regulation. New York: Harper. Zolberg, Aristide R. 1999. “Matters of state: Theorizing immigration policy,” in Charles Hirschman, Philip Kasinitz, and Josh DeWind (eds.), The Handbook of International Mi- gration: The American Experience. New York: Russell Sage Foundation, pp. 71–93.

Click here to print article

Click to return to Table of Contents Click to return to Table of Contents DATA AND PERSPECTIVES

The Retrenchment of Marriage: Results from Marital Status Life Tables for the United States, 1995

ROBERT SCHOEN NICOLA STANDISH

GIVEN THE PROFOUND yet incomplete transformations in family life that have occurred over the past 30 years, changes in patterns of marriage and divorce continue to be of great interest. Historically, the United States has had higher rates of marriage than Western Europe and has witnessed perhaps the highest divorce rates in the world (see Schoen and Baj 1984). Vital statistics counts, crude marriage and divorce rates, and some survey data suggest that in the 1990s American trends in family formation and dissolution continued along the lines seen in the 1980s, that is, cohabitation increased, marriage declined, and divorce remained relatively constant. However, the US National Center for Health Statistics has not published an analysis of marriage and divorce data for years after 1989–90, and actually suspended the collection of detailed marriage and divorce data in January 1996. While Goldstein (1999) found that the plateau in crude divorce rates was not an artifact of compositional changes in such factors as age structure, age at marriage, and marriage order, no study has specifically addressed recent US marriage and divorce patterns or quantified current levels. Here we employ marital status life tables to depict the experience of US males and females using rates for the year 1995, examine recent patterns, and relate them to past experience.

Data and methods

Derived from data on age/marital status–specific rates1 of marriage, divorce, widowhood, and death, a marital status life table can reflect the experience

POPULATION AND DEVELOPMENT REVIEW 27(3):553–563 (SEPTEMBER 2001) 553

Click here to print article Click to return to Table of Contents 554 T HE RETRENCHMENT OF MARRIAGE of a hypothetical cohort of persons who experience those rates over the course of their lives. Here we focus on marriage and divorce, examining males and females separately and recognizing the marital statuses of never married, presently married, widowed, and divorced. We use rates for the year 1995 to depict the implications of recently observed behavior and compare the results with figures for previous periods. (Calculations were carried out using a computer program similar to the one in Schoen 1988: Ap- pendix D.2.) The basic data on population, mortality, marriage, and divorce for 1995 are found in tabulations released by the U.S. Census Bureau and by the National Center for Health Statistics (NCHS). Shortcomings related to data availability and consistency prevent examination by race/ethnicity. The 1995 US population, by age, sex, and marital status can be obtained for each state from the March 1995 Current Population Survey (U.S. Bureau of the Cen- sus 1997). Deaths by age, sex, and marital status are available in NCHS (1997a). To calculate death rates for single years of age up to age 30, five- year age groups for ages 30–34 through 80–84, and ages 85+, we employed indirect standardization2 as necessary, with 1988 rates (from Schoen and Weinick 1993) as the standard. We calculated widowhood rates from deaths of married persons of the other sex, assuming husbands’ ages exceeded wives’ ages by 2.5 years (for five-year age groups) or by 3 years (for single-year age groups). Deaths to persons of unknown marital status were allocated in proportion to deaths of persons with known marital status. In 1995, the Marriage Registration Area (MRA) included 41 states and the District of Columbia. (The missing states are Arizona, Arkansas, Maine, Nevada, New Mexico, North Dakota, Oklahoma, Texas, and Washington, comprising some 14.9 percent of the total US population.) For the MRA, the number of marriages by age, sex, and marital status can be found using NCHS (1997b). Marriages of persons of unknown marital status were allocated in proportion to marriages of persons with known marital status. The NCHS (1997b) data for California in 1995 excluded 51,889 nonlicensed marriages that occurred in that state. Those marriages were apportioned by age and marital status using indirect standardization, where the standard was the relevant California marriage rate for persons included in the NCHS data. Age/sex/marital status–specific marriage rates were first calculated using population data for the Marriage Registration Area only. Using the known total number of 1995 US marriages and the total US population by age, sex, and marital status, we increased the male and female MRA-based rates proportionately to yield the correct total number of marriages. This adjustment, while not exact, is quite reasonable as areal differences in marriage are modest and the procedure assumes only that the pattern of marriage is the same in MRA and non-MRA states. Finally, because the single-year age- specific marriage rates were somewhat ragged, a 5-term (maximum reduc-

Click to return to Table of Contents R OBERT SCHOEN / NICOLA STANDISH 555 tion in mean square error) moving average was used to smooth the data for persons aged 17–29 years. The Divorce Registration Area (DRA) in 1995 consisted of 31 states and the District of Columbia. (The missing states are those missing from the 1995 MRA plus California, Colorado, Florida, Indiana, Louisiana, Minnesota, Mis- sissippi, New Jersey, North Carolina, and West Virginia, comprising some 46.8 percent of the total US population.) The number of 1995 divorces in the DRA, by age, was obtained from NCHS (1997b). The DRA is much less complete than the MRA, but again areal differences are not large, and it is reasonable to assume that DRA and non-DRA states have the same age pattern of divorce. Paralleling the approach used for marriages, we first calculated age/sex–specific divorce rates using population figures for the Divorce Registration Area only. We then increased those male and female rates proportionately to yield the known total number of 1995 divorces. A 5-term moving average was again used to smooth the divorce rates of persons aged 17–29.

Marital status life table summary measures

The results of the male and female marital status life tables for 1995 are summarized in Table 1, along with comparable values for selected years from 1970 through 1988.3 The most noteworthy occurrence between 1988 and 1995 is the continued increase in the average age at first marriage. In 1995, the mean first marriage ages—28.6 years for males and 26.6 years for females (Table 1, line 4)—exceed the comparable values for the earlier periods shown and also exceed the mean ages at first marriage of all US cohorts back to those born 1888–92 (Schoen et al. 1985). The delay in first marriage was not accompanied by a decline of similar size in the likelihood of marriage. For persons surviving to age 15, comparison of the 1988 table to the 1995 table shows that the percent of males ever marrying (line 3) fell by 0.4 percentage points to 83.1, while the percent of females ever marrying rose 0.8 to 88.7. (In Table 1, male and female marriage behavior differs for at least three reasons. First, there is no expectation of equality because both sexes marry persons predominantly from annual birth cohorts other than their own. Second, the actual sex ratio at birth is approximately 105 males per 100 females, rather than unity as implied by the table. Thus, abstracting from sex-differential mortality, higher-order marriage, and intercohort marriage, a female percent marrying of 88.7 implies 887 of 1,050 men marry, or a male percent marrying of 84.5. Third, men who marry are more likely to remarry. Line 10 of the table indicates that in 1995, men have 1.56 marriages per person marrying versus 1.46 for women.) Between 1988 and 1995 sharp declines occurred in first marriage rates for ages through the late 20s, but male first marriage rates rose at ages 29–

Click to return to Table of Contents 1995

(1987); figures for 1975 and 1980 are from Schoen et al. (1985);

Males Females

7.4 6.6 7.7 7.9 8.4 6.9 14.8 14.3 15.0 15.1 15.3 13.2 4.3 4.2 6.0 6.0 7.9 8.4 9.4 9.0 10.8 11.8 13.4 14.5

3.1 2.5 2.8 2.8 2.8 1.7 12.6 11.8 10.9 10.4 10.0 9.9

3.6 4.2 5.6 5.3 6.2 6.4 6.3 7.4 8.5 9.2 9.6 9.2

1.59 1.74 1.67 1.69 1.58 1.56 1.50 1.63 1.58 1.57 1.51 1.46

37.3 38.4 38.5 40.2 40.1 42.5 35.5 37.0 34.7 36.7 36.1 39.7 26.5 23.6 23.8 23.7 24.5 25.7 26.8 23.9 24.4 24.2 24.8 25.7

35.0 38.4 41.8 44.1 46.2 46.5 30.6 33.1 36.4 37.4 39.1 39.8 67.2 68.0 69.7 70.4 70.7 72.4 74.8 75.8 77.6 78.2 78.558.8 79.7 60.4 61.9 62.2 62.9 65.658.4 54.8 53.9 49.9 54.9 47.8 56.0 44.8 55.8 45.3 56.5 56.9 50.5 47.6 44.3 43.0 41.2 41.2 92.9 90.7 87.4 84.2 82.0 82.1 94.1 92.7 89.3 88.3 86.6 87.8

96.0 93.1 89.2 85.9 83.5 83.1 96.5 94.5 90.8 89.7 87.9 88.7 23.4 25.0 26.3 26.8 27.5 28.6 21.8 23.1 24.1 24.5 25.1 26.6 37.3 43.0 44.4 43.9 42.7 43.786.0 88.3 35.7 82.7 41.6 85.0 42.9 78.3 44.1 78.1 43.2 42.5 80.2 83.2 78.3 76.2 72.3 68.7 18.7 16.5 17.1 17.5 18.3 14.1 45.3 41.6 40.0 38.9 39.3 43.4 44.0 40.5 38.5 38.5 39.0 42.2 19.0 16.8 17.1 17.0 17.6 14.1 67.8 68.8 71.1 72.0 72.3 77.8 65.2 66.4 67.9 68.6 68.9 72.0 36.4 36.9 37.1 37.8 37.7 39.7 33.9 34.4 33.1 34.8 34.4 37.3

27.0 28.6 20.8 19.3 17.3 12.3 10.3 9.6 7.6 7.1 6.3 4.8

ersons remarrying

for divorced persons

for widowed persons

those surviving to age 15

2. Percent ever marrying 3. Percent ever marrying among

4. Average age at first marriage 5. Percent of marriages ending in divorce

9. Average age at divorce

1. Expectation of life at birth

6. Percent of marriages ending in widowhood 7. Percent of marriages ending in death 8. Average age at widowhood

TABLE 1 Summary measures from marital status life tables, United States, males and females, 1970, 1975, 1980, 1983, 1988, and 15. Average duration of marriage 16. Average duration of widowhood Line 1970 1975 1980 198317. Average duration of divorce 1988 1995 1970 1975 1980 1983 1988 1995 18. Percent of life spent never married 19. Percent of life spent presently married 14. Average age at remarriage 20. Percent of life spent widowed

13. Average age at remarriage

10. Marriages per person marrying 11. Percent of widowed persons remarrying 12. Percent of divorced p

21. Percent of life spent divorced

SOURCES: Figures for 1995 are taken from marital status life tables as described in the text; figures for 1983 are from Schoen and figures for 1988 are from Schoen and Weinick (1993).

Click to return to Table of Contents R OBERT SCHOEN / NICOLA STANDISH 557

39 and female first marriage rates rose at ages 28–44. Although first marriage comes later in the life course, marriage itself remains highly normative. Under 1995 rates, some five-sixths of men and seven-eighths of women eventually marry. Levels of divorce remain high but lower than the peaks set in the early 1980s. For males, between 1988 and 1995 the percent of marriages ending in divorce rose 1.0 percentage point to 43.7, while for females the figure fell 0.7 to 42.5 (line 5). In effect, our results are consistent with those of Goldstein (1999), as we find little change in the overall risk of divorce. The likelihood of remarriage for divorced persons fell slightly (line 12). For males, the proportion fell 0.2 percentage points to 78.1, while for females it fell 3.6 points to 68.7. Still, over three-fourths of men and two- thirds of women remarry after a divorce. The net result of the changes in divorce and remarriage is a slightly lower average number of marriages per person marrying (line 10). For the first time since 1970, the figure for females is below 1.5, the level that may be said to suggest “serial monogamy.” Simultaneously, the average duration of marriage increased (line 15). For males, it rose 1.2 years to 25.7, and for females it rose 0.9 year to 25.7. The proportion of life spent never married increased with the rise in the average age at first marriage (line 18). For both males and females, it reached levels higher than seen in any period since 1970 and higher than for any cohort born since 1888–92. Females lived two-fifths of their lives never married, and males nearly half (46.5 percent). Nonetheless, the proportion of life lived in the presently married state did not fall between the two most recent years (line 19). It remained steady at 41.2 percent for males and increased 0.5 percentage point to 45.3 for females.

Patterns in the rates of marriage and divorce

Age/sex–specific rates of first marriage, by birth cohort, are shown in Figure 1.4 There is a clear pattern of decline over time at all ages—more pronounced for men than women and particularly strong at younger ages. Figure 2 shows age/sex–specific divorce rates by period, as cohort patterns are harder to interpret. Although some fluctuation is seen at younger ages, considerable stability over time is evident. There is an increase in divorce over time to about 1980/83, with an uneven plateau since then. For all ages below 40, the divorce rate in 1995 was less than the maximum seen for that age in prior years. Divorce rates clearly decline with age; hence, other things equal, a rise in the age at marriage should lead to a decline in the likelihood that a marriage will end in divorce. It has frequently been stated in the press, and more than once in the demographic literature (Bumpass 1990; Castro Martin and Bumpass 1989), that half or more of American marriages end in divorce or separation. With regard to legal divorce, that level has never been attained in any period

Click to return to Table of Contents 558 T HE RETRENCHMENT OF MARRIAGE

FIGURE 1 Age/sex–specific first marriage rates, United States, by 5-year age groups and 5-year birth cohorts, 1936–73

Men 250 F 1946–50

[1941–45 200

1951–55 B 150 F B 3 1936–40 1956–60 C C I [ J Rate per 1000 100 I F J I 1964–68 B D 1961–65 C 1969–73 3 [ 50 CF B B 3 C [BF I J 3F D [ [3 3 0 15 20 25 30 35 40 45 50 55 Age

Women 300

F1946–50 250

1951–55 200 B

1956–60 [1941–45 150 C I CIJ B

Rate per 1000 J F 1964–68 100 B D 1936–40 1969–73 [3I C F CB 3 I C 50 J 1961–65 [BF D B [3F 3F [ [3 3 0 15 20 25 30 35 40 45 50 55 Age

NOTE: Age-specific first marriage rates are calculated by dividing the number of first marriages to males or females in the indicated age group by the never-married population of that sex in that age group. SOURCE: See references in text.

Click to return to Table of Contents R OBERT SCHOEN / NICOLA STANDISH 559

FIGURE 2 Age/sex–specific divorce rates, United States, by 5-year age groups, 1970–95 Men 80

70 3

60 20–24 years 3

3 50 25–29 years [ 3 [ [3 [ [ 40 30–34 years F F F F [3 Rate per 1000 35–39 years B F 30 B B B B F C C C 40–44 years C 20 B C 45–49 years I I I C I I I 10

0 1970 1975 1980 1983 1988 1995 Year

Women 80

70 20–24 years 3 3 3

50 3 25–29 years 3 [ 40 [ [ [ 3 [F F Rate per 1000 30 F 30–34 years [ F B F B B 35–39 years B C F B C 20 C 40–44 years B C C I 45–49I years C I I I 10 I

0 1970 1975 1980 1983 1988 1995 Year

NOTE: Age-specific divorce rates are calculated by dividing the number of divorces to males or females in the indicated age group by the married population of that sex in that age group. SOURCE: See references in text.

Click to return to Table of Contents 560 T HE RETRENCHMENT OF MARRIAGE marital status life table, although the patterns of recent cohorts have yet to unfold. Although age-specific divorce rates have shown relatively little consistency by cohort, the maximum divorce rates were experienced by the male cohort aged 15–19 in 1983 and by the female cohort aged 15–19 in 1980. To find the probability of a marriage ending in divorce for those cohorts, we calculated additional marital status life tables from marriage, divorce, and mortality data by cohort. Cohort experience was completed using 1995 rates (which show quite high rates of divorce for ages over 35). Those tables reflect a plausible maximum cohort probability of divorce of .457 for men and .472 for women. However, not all permanent separations result in legal divorce. Castro Martin and Bumpass (1989: 40) suggest that perhaps 5 percent of a marriage cohort could separate without divorcing. Thus while US experience never attained a level where half of all marriages end in legal divorce, the cohorts with the highest rates of divorce may have half of their marriages end in either divorce or separation.

Marriage metabolism

A different perspective on marital formation and dissolution is provided by a measure of “metabolism” (Ryder 1975; Schoen and Weinick 1993). Mar- riage metabolism (MM) is defined as

FIGURE 3 Marriage metabolism, US males and females, 1910–95 0.035

F 0.033

0.031 Men F F F 0.029 F Women B

0.027 F B B F F B B F 0.025 F F B B B Metabolism F F 0.023 B B B B F F F B 0.021 B B B F F 0.019 B B

0.017

0.015 1910 1920 1930 1940 1950 1960 1970 1980 1990 Year

SOURCE: See references in text.

Click to return to Table of Contents R OBERT SCHOEN / NICOLA STANDISH 561

MM = (All life table marriages and divorces) / (Total life table population).

Marriage metabolism thus reflects the total rate of marriage turnover in the population, and it increases whenever either marriage or divorce increases. Figure 3 shows period marriage metabolism values for US males and females for the years 1910 through 1995. Both sexes reached their peak metabolism in 1975, when both marriage and divorce rates were quite high (though neither was at its peak). Marriage metabolism dropped sharply between 1988 and 1995, especially for women. The continuing fall in marriage rates at the peak marriage ages and the plateau in divorce rates give 1995 the lowest marital turnover rate since the early 1960s.

Cohabitation

Although the vital statistics data exclude cohabitation, its footprints are evident. Rising levels of cohabitation can be seen behind the steep rise in the average age at first marriage and in declines in the proportion of divorced persons who remarry. Still, the major effect of cohabitation has been to delay, not replace, marriage. Marriage rates at ages 29–39 rose between 1988 and 1995, keeping the 1995 proportion of persons who ever marry near, or above, the 1988 level. The proportion of life lived never married continues to rise, while the proportion of life lived presently married appears to have reached a plateau. Rates of divorce have also remained fairly level, despite later ages at marriage. The role of cohabitation in that trend in divorce is unclear, although a number of studies have shown that marriages preceded by a cohabitation are more likely to dissolve, and most marriages now start as co- habitations (Cherlin 1992; Bachrach, Hindin, and Thomson 2000). The figures suggest an emerging pattern in which marriage consoli- dates its position at the older ages while cohabitation characterizes most unions between younger persons. It is unclear what implications that pattern might have for either marital stability or childrearing.

Notes

Helpful comments from James A. Weed and curate measures of movement between states from members of the Penn State Family and (see Schoen and Woodrow 1980). Children Discussion Group are gratefully ac- A number of analyses of divorce have knowledged. tracked marriage cohorts rather than birth co- 1 Such multistate life tables can also be cal- horts (e.g., Weed 1980), and in fact there is culated from “prevalence rates,” i.e., the pro- good reason to believe that marriage duration portions of persons in each state at every age. may be more predictive of divorce than is age Prevalence rates, however, confound current (see Schoen 1977). Unfortunately, the use of and previous behavior and do not provide ac- marriage cohorts would make it impossible to

Click to return to Table of Contents 562 T HE RETRENCHMENT OF MARRIAGE

capture the lifetime experience of a cohort or Percent marrying Average Percent of to combine the analysis of divorce with that among age at marriages of first marriage, remarriage, and widowhood survivors first ending in in a single, integrated framework. to age 15 marriage divorce Whole US (Table 1) 2 Indirect standardization assumes that a Males .83 28.6 .44 given population has the same age pattern of Females .89 26.6 .42 demographic rates as a chosen standard, ad- justing the level in the standard rates to re- Registration Areas produce the level of behavior observed in the Males .80 29.0 .42 given population (see Shryock, Siegel, and as- Females .86 27.0 .40 sociates 1973: 421–422, 482–484). 3 The 1995 data indicate that there are dif- 4 The values graphed in Figures 1–3 draw ferences in the level of marriage and divorce on previous work, particularly Schoen et al. between the Registration Areas and the United (1985), Schoen (1987), and Schoen and Wei- States as a whole, with the national level nick (1993). Values for 1975 and earlier years higher in both cases. To reproduce the known have been archived by the National Technical total number of US marriages and divorces, we Information Service (NTIS). The full citation increased the Marriage Registration Area mar- is “United States Marital Status Life Tables for riage rates by 8.2 percent, and the Divorce Reg- Periods 1910–1975 and Cohorts Born 1888– istration Area divorce rates by 8.8 percent. Re- 1945,” NTIS Accession No. PB87-222485/AS sults from marital status life tables calculated (Report No. NICHD/CPR/DBSB/87-2). on the basis of MRA/DRA data alone thus differ from those shown in Table 1, but the differences are not very large. For example:

References

Bachrach, Christine, Michelle J. Hindin, and Elizabeth Thomson. 2000. “The changing shape of ties that bind: An overview and synthesis,” in Linda J. Waite (ed.), The Ties That Bind: Perspectives on Marriage and Cohabitation. New York: Aldine de Gruyter, pp. 3–16. Bumpass, Larry L. 1990. “What’s happening to the family? Interactions between demographic and institutional change,” Demography 27: 483–498. Castro Martin, Teresa and Larry L. Bumpass. 1989. “Recent trends in marital disruption,” Demography 26: 37–51. Cherlin, Andrew J. 1992. Marriage, Divorce, Remarriage (Revised and enlarged ed.). Cam- bridge, MA: Harvard University Press. Goldstein, Joshua R. 1999. “The leveling of divorce in the United States,” Demography 36: 409–414. National Center for Health Statistics (US DHHS). 1997a. “Report of final mortality statistics, 1995,” Monthly Vital Statistics Report, vol. 45, no. 11, by R. N. Anderson, K. D. Kocharek, and S. L. Murphy. Washington, DC: US Government Printing Office. ———. 1997b. Marriage and Divorce Data, 1989–1995, CD-ROM series 21, no. 6. Hyattsville, MD: DHHS. Ryder, Norman B. 1975. “Notes on stationary populations,” Population Index 41: 3–28. Schoen, Robert. 1977. “On choosing an indexing variable in demographic analysis,” Social Science Research 6: 246–256. ———. 1987. “The continuing retreat from marriage: Figures from 1983 U.S. marital status life tables,” Sociology and Social Research 71: 108–109. ———. 1988. Modeling Multigroup Populations. New York: Plenum. Schoen, Robert and John Baj. 1984. “Cohort marriage and divorce in five Western countries,” Comparative Social Research 7: 197–229.

Click to return to Table of Contents R OBERT SCHOEN / NICOLA STANDISH 563

Schoen, Robert, William Urton, Karen Woodrow, and John Baj. 1985. “Marriage and divorce in twentieth century American cohorts,” Demography 22: 101–114. Schoen, Robert and Robin M. Weinick. 1993. “The slowing metabolism of marriage: Fig- ures from 1988 U.S. marital status life tables,” Demography 30: 737–746. Schoen, Robert and Karen Woodrow. 1980. “Labor force status life tables for the United States, 1972,” Demography 17: 297–322. Shryock, Henry S., Jacob S. Siegel, and associates. 1973. The Methods and Materials of Demog- raphy (rev. ed.) Washington, DC: US Government Printing Office. U.S. Census Bureau. 1997. Current Population Survey: Annual Demographic File, 1995 [Com- puter file]. ICPSR version. Washington, DC: U.S. Census Bureau [producer], Ann Ar- bor, MI: Interuniversity Consortium for Political and Social Research [distributor]. Weed, James A. 1980. “National estimates of marriage dissolution and survivorship,” US National Center for Health Statistics, Vital and Health Statistics, Series 3, No. 19. Hyattsville, MD: DHHS pub. No. (PHS) 81-1403.

Click to return to Table of Contents Click here to print article Click to return to Table of Contents ARCHIVES

William Farr on the Economic Value of the Population

The modern theory of human capital came to prominence in economics only in recent decades, but its antecedents can be traced back to the earliest economic writings. The notion that human skills represent economic value comparable to that of capital was clearly articulated by the classical economists, notably by William Petty and Adam Smith. That improvement of skills and increase in the number of persons in whom skills are embodied are sources of economic progress follows from their conceptual clarification. Attempts to quantify the economic value of the population and assessment of the effect of mortality improvements and population growth were, however, later developments. Among the earliest contributions to such calculations, one by William Farr, published in 1877 and reprinted below, is particularly notable. Defining the economic value of a person as the discounted sum of expected future earnings, Farr arrives at a figure of £5,250 million (for 1876) “as an approximation to the value which is inherent in the people [of the United Kingdom], and may be fairly added to the capital in land, houses, cattle or stock, and other investments.” In addition to providing insightful commentary on the rationale and weaknesses of his calculations, he estimates the addition to that amount from population growth in the preceding four decades, discusses the impact of outmigration to the colonies and the United States during that time, and notes the dependence of the economic value of the population on the level of education, on the state of health of the population, and on people’s longevity. William Farr (1807–1883) was perhaps the most influential British statisti- cian of the nineteenth century. Although trained as a physician, in 1839 he accepted a post in the General Register Office and from 1842 to 1880 he served as Statistical Superintendent. During his long tenure he was the main force in the development and analysis of British vital statistics and in setting the foundations of modern epidemiology. He constructed the first British life table (based on deaths in 1841) and carried out a wide range of creative analyses of British mortality statis-

POPULATION AND DEVELOPMENT REVIEW 27(3):565–571 (SEPTEMBER 2001) 565

Click here to print article Click to return to Table of Contents 566 A RCHIVES tics, especially on mortality differentials, with an aim of promoting social reform. The most accessible route to his written output for modern readers is a posthumously published collection: Vital Statistics: A Memorial Volume of Selections from the Reports and Writings of William Farr (London: Offices of the Sanitary Institute, 1885). A reprint edition of this work was published under the auspices of the Li- brary of the New York Academy of Medicine (Metuchen, N. J.: The Scarecrow Press, 1975). The passage reproduced below (pp. 59–64) is an excerpt from the Registrar- General’s 39th Annual Report (1877), titled The Economic Value of Population. The topic was earlier treated in Farr’s paper “The income and property tax,” Quar- terly Journal of the Statistical Society of London (March 1853), excerpted on pp. 531–550 of the book under the title Cost, and the Present and Future Economic Value of Man.

Various attempts have been made to estimate the amount and the increase of the capital of the United Kingdom. The most recent attempt of the kind has been made by the chief of the statistical department of the Board of Trade. The value of the most important part of the capital of the United Kingdom and the increase have yet to be determined; I mean the economic value of the population itself. To this I propose to call attention briefly. As lands, houses, railways, and the other categories in the income tax schedules are of value, because they yield annual returns; so, for the same reason, and on the same principle, the income of the population derived from pay of every kind for professional or other services and wages can be capitalized; not precisely, it is true, unless the income of every person living were returned at least as nearly as the incomes subject to income tax; but sufficiently near to the true value to show that the value of the population itself is the most important factor in the wealth of the country. It will be sufficient to state here that the capitalization of personal incomes always proceeds upon the determination of the present value at any age of the future annual earnings at that and all future ages; hence the value of future wages rises from the date of birth, when it is a notable quantity; is highest in the labouring classes at the age of 25; and declines as age advances, until in extreme age, when no wages are earned, it disappears. The living by the Life Table are most numerous in childhood, and gradually fall off till they are all extinct; and so in the population enumer- ated at the Census the numbers decline from the first year to the ultimate year of age. While the rates of wages rise rapidly from birth to the age of manhood, and afterwards decline, the numbers living constantly decline. Taking a series of observations on the wages of agricultural labourers1 some years ago at different ages; determining their value by a Life Table at five

1See Journal of the Statistical Society, Vol. xvi., pp. 42–43. Extracts from this paper will be found on pp. 531–7 of this volume.

Click to return to Table of Contents W ILLIAM FARR 567 per cent. rate of interest for each age; and multiplying the numbers living by these values, it is found that the mean gross value at all ages is 349l. But the mean value of the subsistence of the labourer as child and man, determined by the same method, is about 199l.; and deducting this sum from 349l., there remain 150l. as the mean net value of the male population, estimated by this standard of the agricultural labourer. To extend the value to the whole population, including females, the standard might be lowered from 150l. to 110l. a head. Then multiplying the population of the United Kingdom by 110 we have as the aggregate value £3,640 million; this including only as much of the income as approximates in annual amount to the wages of agricultural labourers. Only a small part of it is subject to assessment under the income tax schedules. The gross assessment under the income tax affords the means of estimating the value of incomes exceeding 100l. a year under Schedules D. and E.; excluding companies, mines, and works, these profits and salaries amount to £214 million a year, to which about £92 million a year may be added for incomes above 30l. and below 100l. a year; thus making the aggregate of such incomes £306 million a year; which when the assessments of B. (farmers’) are added becomes £373 million a year. Deduct the half of this revenue as due to external capital, and as required for the necessary sustenance of farmers, tradesmen, and professional men and there 1 remain £1862 million a year as pure profit; which cannot be capitalized as a perpetuity inasmuch as the interest is limited by the lives of the produc- ers, but taking life contingencies into account may be capitalized at ten years’ purchase. This makes the value of these incomes £1,865 million. Allowing £255 million for the part of the incomes of about a million people paying the income tax previously valued in the £3,640 million, and for other deductions, £1,610 million remain, which, added to the £3,640 million already obtained, make £5,250 million.2 Thus by capitalizing the earnings, fees, salaries, wages of the professional, mercantile, trading, and working classes, £5,250 million are obtained as an approximation to the value which is inherent in the people, and may be fairly added to the capital in land, houses, cattle or stock, and other investments. The amount would be increased by taking into account the rise of wages, and the income omitted in the returns of Schedule D. With an industrial Census an accurate estimate can be made of this most important part of the capital of the country. The minimum value of the population of the United Kingdom, men, women, and children, is 159l. a head; that is the value inherent in them as a productive, money-earning race. The incomes chiefly under schedules D., E., and B., raise the mean value from 110l. to 150l. (see above).

2Mr. Giffen makes the value of the capital in other forms £8,500 million; making with the value of the population itself, £13,750 million. See Journal of the Statistical Society, Vol. xli., pp. 1–31.

Click to return to Table of Contents 568 A RCHIVES

Again, it must be borne in mind that the value under Schedule A. is dependent upon the population; where there is little population land itself is of little value. The increase of the value of house property is directly due to the increased numbers and earnings of the inhabitants. The railways yield no profit where there is no population. The profits of quarries, mines, ironworks (Schedule D.), and other concerns are mainly due to the skill and industry of the masters and men who work them. Upon the other hand the products of human industry are multiplied a hundredfold by the tools, machinery, steam power, and all the appliances which capital com- mands and represents. Should the population of a country decay, the value of its capital might sink to the vanishing point. 1 What I wish further to point out is that during the 392 years this office has existed there have been added to the population of the United Kingdom 7,619,759 people who, valued as land is valued by the annual yield of net profit, constitute an addition of £1,212 million to the wealth of the nation. The value of labour—that is of working men—varies, and is greatest where there is the greatest facility for profitable use, and where it is in greatest demand. Thus a large stream of the population of England flows to the Metropolis; and England is to the United Kingdom what the Me- tropolis is to England. So the populations of Ireland and Scotland flow into England, where they find more profitable employment, and are of more value than they are at home. For the same and other reasons large armies of the population of the United Kingdom passed into the colonies and the United States; during the thirty-nine and a half years (1837–76) the excess of births over deaths was nearly 16 millions, of which nearly 8 millions augmented the ranks of the population at home, and more than 8 millions settled in other lands; chiefly in the midst of the old English stock of the United States and in the Colo- nies extending from Canada in America, to Africa and to Australasia. Of the 8,013,267 people who must have left the country, only about 6,580,000 are accounted for by the Emigration Commissioners, whose returns were imperfect in two ways; they neither included the whole of the emigrants nor recognised emigrants returning recently in large numbers.3 The emigrants are chiefly adults married and unmarried; the men greatly exceeding the women in number. A few infants accompany their parents. Valuing the emigrants as the agricultural labourers have been valued at home—taking age and service into account—the value of emigrants in 1876 was 175l. per head. If we may venture to apply this standard to the whole period it will follow that the money value of the 8,000,000 people that left England,

3See Census Report, 1871, where this was first demonstrated; and Emigration Report, 1877.

Click to return to Table of Contents W ILLIAM FARR 569

Scotland, and Ireland in the years 1837–76 was £1400 million, or on an average about 35,000,000l. a year. In round numbers taking into account their aptitude to earn wages in future years at the home rates the annual industrial army that went out was worth at starting 35,000,000l. Many of the emigrants are skilled artizans, and considerable numbers are returned as farmers, gentlemen, professional men, and merchants; some of whom no doubt carried away a certain amount of capital which is not here brought into account. The policy of the people of this country has thus been a policy of progress; instead of resting as they were in 1837, they have added since that year on an average of 192,873 souls annually to the population at home, and sent 202,868 sons and daughters to seek their fortune abroad in other fields of labour. The women, instead of to 644,214 children, who would just replace the population removed by deaths, have given birth to 1,039,987 annually, at a certain loss of their own lives with intermingled sorrows and joys such as befall mothers in rearing children: while the men instead of expending the whole of their gains on themselves have devoted a large share to their wives and families; besides that, as we have seen, the external wealth of the country has increased, as the nation has, without conquering territory or levying heavy contributions on its European neighbours. The value of men varies with their earnings, which differ considerably in the colonies from the earnings of agricultural labourers at home; and on the whole before the civil war the emigrants to the United States got higher wages, and at the same time gave a higher value to the territory. It may be contended that emigration is a loss to the mother country. It seems so. It is like the export of precious goods for which there is no return. But experience proves that simultaneously with this emigration there has been a prodigious increase of the capital of the country, especially in recent years. Wages have risen, and the value of the labourer has risen in proportion. In Norfolk, where wages are intermediate between the rates in the north and south, the rise has apparently been about 20 per cent.; so a fifth may be added to the estimated value of the workman. When the man leaves the village where he was born and bred, he leaves the market open to his fellows; he removes to a field where his work is in demand, and carries his fortune with him. It is the same when he emigrates to the colonies. His parents in rearing him have expended their gains in the way most agreeable to themselves. They have on an average five children, instead of two or three, or none. Taking a wider view, the emigrants create articles of primary use with which in exchange they supply the mother country; they have sent to England in the 39 years wheat, cotton, wool, gold to the value of hundreds of millions. What is of still more vital importance, they grow into new nations; they multiply discoveries; by confed-

Click to return to Table of Contents 570 A RCHIVES eration they will be to the Anglo-Saxon race outposts of strength across the Atlantic, in the Pacific, in South Africa, and in Australasia on the flank of India. And, moreover, to all it is an advantage to speak a wide spread language, and thus to be in social, literary, and scientific communion with millions of the same race. The increasing numbers enable them, advanced as they are in the arts, in the sciences, and in civil government, to do more for the good of kindred races; and to endow them with advantages which could not be attained in other ways for centuries. They govern India. The economic value of the population depends very much on their command over the powers of nature; which they acquire by education. Put barbarians in possession of the land, the mines, the manufactures, the machines, the ships, the triumphant position of these islands on the sea between two continents, and what would be the result? Another Asia Mi- nor, Egypt, or Syria? The better educated the English people become, the more skilful they will become, and the more valuable in an economic sense they will be. The clever artisan is worth more than the rude labourer. Now the art of reading and writing their own language is by no means proof of complete education, or of any technical training, but it is a proof that men in possession of it are preparing to enter on the heritage of thought, and knowledge, and sentiment, which men of all ages have bequeathed to mankind, and which is enshrined in the writers of an admirable language. In 1837 not more than 58 in 100 men and women possessed this art; but there has been progress, and I have year by year assiduously noted the increase of their numbers in the 39 years, so that I am now able to report, that instead of 58, eighty-one in a hundred write their names in the marriage registers. It is evident that there are other elements on which the economic value of the working population depends; and foremost among them stand health and long life. The longer men live, and the stronger they are, the more work they can do. Epidemic diseases in rendering life, render wages, insecure. These diseases are most fatal in cities whither the population—to secure all the advantages of the division of labour—have been congregat- ing every year in increased numbers: villages have become populous or have grown into towns; so the population has been growing denser. And that by a definite law, other things being equal, tends to increase weakness, sickness, and mortality. There have been counteracting agencies in operation in the thirty-nine years. Asiatic cholera was epidemic in England in 1831–2; influenza followed at intervals in 1833, 1837, and 1847; and laid thousands of the population low; in 1848–9 the cholera epidemic in England and Wales alone was fatal to fifty-three thousand people;4 its ravages in every corner of the kingdom were described; the conditions of its diffu-

4Registrar General’s Report on Mortality of Cholera in 1848–9.

Click to return to Table of Contents W ILLIAM FARR 571 sion and fatality were brought to light, and the further investigations of the slighter epidemics of 1854 and 1866 prove that this plague is under the control of science. Other epidemics have since been fatal especially to children, and fever has struck at prince and peers as well as peasants; but upon the whole the great zymotic diseases have been quelled. Plague in its various forms has been kept at bay by a series of defences based upon minute precautions. In some epidemics I found it necessary to publish daily par- ticulars respecting deaths in the Metropolis. By pursuing such inquiries, year after year, not only many of the causes that induce sickness and de- stroy life have been discovered, but observations of the same kind have shown that their removal has been followed by health and longer, more vigorous life. The economic value of the population of several towns has been increased by sanitary measures. The truths established, the facts as- certained, the remedies discovered in the thirty-nine years past await their full administrative applications in the years to follow; and the savings of time wasted in sickness, as well as of precious lives prematurely lost in youth and manhood, will enhance the value of the population to an incal- culable extent. The famines so fatal in Ireland are not likely to recur; part of the population has emigrated to England or to America, and the intelli- gent landowners of Ireland, through the extension of the Poor Law, now insure their countrymen against death by starvation. The same beneficient law has in the thirty-nine years been extended to the Highlands of Scot- land. Every improvement in health recorded makes it clearer and clearer that the gloom of sickness and premature death flies away before sanitary measures; and when the qualified health officers whom the Universities are offering to examine, are in suitable positions under enlightened local authorities all over the country they will no doubt prove as efficient in preventing as their medical brethren are in treating sickness. The result on human happiness cannot be calculated; but a future Industrial Census will show in a very definite shape its effect in raising the economic value of the population. The mean lifetime by the English Life Table is 40.86 years; by the Healthy Life Table it is 49.0 years, which is attainable in every well organized State. It is fair to assume that if a fifth part be added to the mean lifetime, at least a fifth part will be added to the worth of a living and labouring population. Upon this estimate £1,050 million will be added to the economic value of the population of the kingdom. Its value will increase with its numbers, and so will the value of its emigrating thousands.

Click here to print article ClickClick toto returnreturn toto TableTable ofof ContentsContents Click to return to Table of Contents BOOK REVIEWS

Linking Human and Natural History: A Review Essay*

JOEL E. COHEN

John R. McNeill, professor of history at Georgetown University, Washing- ton, DC, has written an excellent environmental history of the twentieth- century world. The first two-thirds of his book are extensive and engaging reviews of changes in the Earth’s soil and crust, atmosphere, hydrosphere, and biosphere. The last third attempts to identify the driving forces behind these unprecedented environmental changes. McNeill’s account is broad, balanced, insightful, scholarly, and enjoyable to read. It will be valuable to anyone interested in how our environments have come to be as they are, whether as an educated layperson, student, or scholar. His many concrete descriptions suggest a richer picture of the engines of environmental change than is sometimes advertised elsewhere. This enriched picture will be useful and important to those seeking guidance on how to manage, understand, or simply live with environmental changes in the twenty-first century. Environmental changes in the twentieth century were dramatic. Table 1, based on McNeill’s Table 12.1 (pp. 360–361), estimates the magnitude of some of them. I have added some estimates from other sources. McNeill’s note, appended to the table, that “some of the numbers are more trustworthy than others” illustrates his characteristic and laudable caution, and applies to the estimates I have added as much as to his. Not one of the environmental, economic, and demographic variables listed in Table 1 was the subject of an objective, reliable, global statistical system in 1900, and most still were not in 2000. (Now is the time to begin writing the environmental history of the twenty-first century, by getting global environmental and social data

*Review of J. R. McNeill, Something New Under the Sun: An Environmental History of the Twen- tieth-Century World. New York: W. W. Norton, 2000. xxvi + 421 p. $29.95; US pbk. ed. 2001, $15.95.

POPULATION AND DEVELOPMENT REVIEW 27(3) (SEPTEMBER 2001) 573

Click here to print article Click to return to Table of Contents TABLE 1 Changes in selected environmental, economic, and demographic indicators during the twentieth century Increase Factor McNeill’s estimate, My estimate, Item 1890s–1990s 1900s–2000

World population 4 3.7 Annual increment to world population 7.8 Life expectancy at birth 2.2 United States population 3.7 China population 3.3 Urban proportion of world population 3 3.6 Total world urban population 13 13.7 Population of single largest metropolitan area 4.3c World economy 14 16 Gross domestic product per person 4.1 Industrial output 40 Energy use 16a 10c Coal production 7 Air pollution ≈5

Nitrogen released in NOx from fossil fuels 21 Carbon dioxide emissions 17 14.6 Carbon dioxide emissions, 1900–1996b 12.2 United States carbon dioxide emissions, 1900–1996b 8

CO2 partial pressure in the atmosphere 1.2–1.3 Sulfur dioxide emissions 13 Lead emissions to the atmosphere ≈8 United States nonfuel minerals consumption, 1900–1995c 29 United States nonfuel wood products consumption, 1900–1995c 3 United States metals consumption, 1900–1995c 14 United States fossil-fuel-based synthetics consumption, 1900–1995c 82 Water use 9 7–9 Marine fish catch 35 Cattle population 4 Pig population 9 Horse population 1.1 Blue whale population (Southern Ocean only) 0.0025 (99.75% decrease) Fin whale population 0.03 (97% decrease) Bird and mammal species 0.99 (1% decrease) Irrigated area 5 6c Forest area 0.8 (20% decrease) Cropland 2 World grain harvest 4.8c

NOTE: Some of the numbers are more trustworthy than others. Comments on their reliability appear in the text, from which they are all drawn. Estimates by McNeill are from pp. 360–361. Unless here attributed to another source, my estimates are derived from Cohen (2002, pp. 30–32), where citations to original sources are given. aRevised to 13 in the 2001 British Penguin paperback edition. bHunter 1999, p. 13. cStarke 1999.

Click to return to Table of Contents B OOK REVIEWS PDR 27(3) 575 systems in place.) McNeill notes (p. 361): “This table ignores expansions that took place entirely after 1900, such as CFC releases or tractor numbers. It ignores expansions for which the baseline of 1900 would produce astronomical coefficients of increase, for example, the world’s total of auto- mobiles, the quantities of chemical fertilizer applied, or the tonnage of synthetic chemicals produced. The table is an imperfect measure, but it gives the right impression.” McNeill titled his version of Table 1 “The measure of the twentieth century.” That is a slight overstatement, considering the twentieth century’s changes in science, medical, chemical, and information technology, women’s rights, life expectancy, literacy, democratic governance, trade, and material intensity of production, for example. Nevertheless, McNeill is right: the table is both revealing and suggestive. Accepting McNeill’s round numbers, total population grew nearly 4 fold. The only other indicator listed that also grew 4 fold was the size of the cattle population. Many other indicators of human impact on the environment grew by far more than a factor of 4: carbon dioxide emissions (17 fold), sulfur dioxide emissions (13), lead emissions to the atmosphere (8), water withdrawals (9), and marine fish catch (35). Some indicators that might have been expected (in 1900) to grow in proportion to human population size grew by less than a factor of 4: the horse population (1.1) and the area of cropland (2). Clearly human numbers were not the sole determinant of environmental change in the twentieth century. From the past lack of proportionality between human population growth and the magnitude of environmental changes, it seems reasonable to anticipate that if global human population growth ends in the twenty-first century, human impacts on the environment will not necessarily level off in parallel. What else was involved in past changes? Table 1 suggests some of the answers: economic growth (the increase of capital, production, and income) and technological change. Economic growth, illustrated by the reported 40-fold increase in industrial production and 16-fold increase in energy use, was partly responsible for increased carbon and sulfur emissions to the atmosphere, as well as for a roughly 4-fold rise in global average income per person during the twentieth century. McNeill’s text, but not Table 1, points out that economic growth has also enabled environmental clean-ups in wealthy countries that chose to clean up. Technological changes displaced the horse. Technological changes also increased industrial water usage. An increase in irrigated area (up 5 fold) contributed to the 9-fold increase in water withdrawals. Irrigated cropland can be much more productive than rain-fed cropland. The spread of irrigated cropland since 1960 contributed importantly to keeping the total area of cropland under cultivation more or less steady since world population

Click to return to Table of Contents 576 PDR 27(3) B OOK REVIEWS reached 3 billion in 1960; prior to 1960, cultivated cropland had increased nearly in direct proportion to population (Evans 1998, p. 205). Other factors besides increased irrigation that enabled the same amount of land that fed 3 billion people in 1960 to feed more than 6 billion in 2000 were new genetic varieties of major cereals, pesticides, artificial nitrogenous fertilizers, credit and market institutions, improved transport, and farmer education. Thus technological and other changes brought environmental tradeoffs in agriculture. More nitrogen (and methane) into the atmosphere, more nitrogen into groundwater, more water use, and massive global genetic change (evolution under human guidance) in cultivars of wheat, rice, and maize accompanied a leveling off of the area under cultivation and a savings by 2000 of cultivated cropland roughly equal to all currently cultivated cropland. During the twentieth century, other environmentally influential technological changes included the invention and worldwide diffusion of stain- less steel, the tractor, new varieties of other agricultural plants and animals, modern contraceptives, radio, television, films, computers, the Internet, an- tibiotics, vaccines, the chainsaw, the airplane, and physical, chemical, and biological weapons of unprecedented destructive power. Obviously technology affected the environment. That the collective environmental impact of humans depended not only on how many people there were, but also on how wealthy those people were and on the technology those people used to generate their wealth was evident long ago to Barry Commoner (1971) and John Holdren and Paul Ehrlich (1974). They proposed and subsequently promoted a simple, very influential formula, widely taught, used, and criticized, which is commonly referred to as the IPAT formula: I=PAT, or

Impact on the environment = Population × Affluence × Technology.

Because McNeill’s book is not explicitly theoretical, he never directly addresses the IPAT formula, for or against. But empirical description and interpretation reflect a theory, possibly implicit, about what factors are perti- nent to understanding change. McNeill’s interpretive section called “Engines of change,” the last third of the book, has three main chapters: “More people, bigger cities” (here is P, in a sophisticated form that recognizes environmental impacts of migration and urbanization); “Fuels, tools, and economics” (here are A and T, but also observations about economic systems covered by neither Affluence nor Technology); and “Ideas and politics.” The IPAT formula reflects some of the contributions of wealth and technology to environmental impact and makes it possible to organize several of the indicators of change in Table 1 as contributors. McNeill builds on, and goes beyond, the IPAT formula by adding to its right side (PAT) three important classes of factors, at least equally important but perhaps not so tidily quantified:

Click to return to Table of Contents B OOK REVIEWS PDR 27(3) 577

—Culture, including “big ideas,” environmental ideas, domestic and international politics, war, and environmental policies; —Market size and structure; and —Autonomous or exogenous environmental change not caused by humans. In this sense, McNeill’s book is not a “PAT answer” to the challenge of understanding environmental change in the twentieth century. Let me illustrate.

Culture, especially politics

Agricultural changes in the twentieth century depended on culture as well as on technology. For example (p. 226), while the “biological, chemical, and mechanical transformations of modern agriculture took hold” in much of the West and much of Asia, the “Soviet Union partly missed out because modern genetics rankled the socialist sensibilities of Stalin and Khrushchev, delaying progress in plant breeding until the mid-1960s. Before 1960 the USSR partook of mechanization and irrigation, but only after 1965 did it undertake conversion to the doctrine of genetic manipulation and heavy use of nitrogen. Thus the total geopolitical effect of modern changes in agriculture improved the relative position of the West and Japan slightly, that of China, the Asian tigers (South Korea, Taiwan, Malaysia), and Latin America even more slightly, while contributing to the relative decline of Soviet status, and to the weakness of Africa.” Culture kept modern genetics out of the Soviet Union for a time, and that lag weakened the Soviet Union geopolitically. (Could there be a lesson here for the United States in its approach to continued genetic research?) Like this example, many of McNeill’s historical cases, and nearly all the topics in his chapter on “Ideas and politics,” involve culture, though the word “culture” does not appear in the table of contents or index of McNeill’s book. To see why culture is a relevant concept, it is useful to recall some definitions. Richard Shweder, an anthropologist at the University of Chi- cago, writes (2000, p. 7): “By ‘culture’ I mean community-specific ideas about what is true, good, beautiful and efficient. To be ‘cultural’ those ideas about truth, goodness, beauty and efficiency must be socially inherited and customary; and they must actually be constitutive of different ways of life. Alternatively stated, ‘culture’ refers to what Isaiah Berlin (1976) called ‘goals, values and pictures of the world’ that are made manifest in the speech, laws and routine practices of some self-monitoring group.” Shweder quotes with approval Robert Redfield’s 1941 definition of culture as “shared un- derstandings made manifest in act and artifact.” All these definitions serve here, and all include politics. Some of the great political changes of the twentieth century include: the end of colonialism; the end of legally sanctioned racial segregation; the

Click to return to Table of Contents 578 PDR 27(3) B OOK REVIEWS establishment of political, social, and economic rights for women; the rise and fall of Communism; the spread of democracy; and the establishment of economic growth as what McNeill calls a kind of “state religion” in authori- tarian and democratic nations alike. McNeill illustrates the environmental impact of these political changes. In North America, politics affected the environment before the exist- ence of the United States. The tall trees of New England were felled in part to provide masts for the ships of the British Navy. Politics influenced defor- estation in the twentieth century, for example through government subsidies to logging, land clearing, and road building in forests. Politics also promoted the recovery of forests through subsidies for carbon sequestration and conservation. The military draft of men to fight in the American Civil War in the 1860s caused a labor shortage, which accelerated the spread of horse-drawn thresh- ers and reapers. These machines prepared the way for the American conversion to the tractor and other mechanized agriculture between 1920 and 1955. Wherever mechanization was powered by fossil fuels, fewer people and fewer draft animals were required to work the fields and bigger farm fields became economical. Mechanization released a vast agricultural labor force to go to cities. In the former Soviet Union, an ideological commitment to big enterprises as demonstrations of the power of the state led to state farms that aver- aged, by 1977, 40,000 hectares, three times the size of Washington, DC. Often these vast treeless fields suffered intense soil erosion from wind. In the Po Valley in northern Italy, in the Tennessee Valley Authority and the Colorado River system in the United States, along the Volga River in the Soviet Union, in India, China, and Egypt and elsewhere, dams to extend irrigation, control floods, and generate electricity always also served larger political purposes. They demonstrated to the people the power of the state to control nature for the social good. This demonstration was helpful to whoever was in power at the time, be it Roosevelt, Stalin, Nehru, Nasser, Nkrumah, or many other leaders. In the 1960s, more than one large dam at least 15 meters high was completed every day on average. By 1990, two- thirds of all the world’s streamflow passed over or through dams. The diver- sions of water from natural courses reduced the Aral Sea from the world’s fourth- largest lake to its eighth-largest lake. The full environmental consequences of damming the rivers that drain into the Mediterranean Sea have yet to be felt. Only the politics of distrust could account for Nasser’s placement of the Aswan dam in southern Egypt, rather than further upstream in existing lakes in Ethiopia and Uganda. In southern Egypt, a region with one of the highest rates of evaporation on Earth, the reservoir increased the surface area from which Nile water evaporated. Politics governed societies’ responses to air pollution in Mexico City, Athens, London, Pittsburgh, Peru, Ontario, the Ruhr of Germany, and the

Click to return to Table of Contents B OOK REVIEWS PDR 27(3) 579 industrial complex between Dresden, Prague, and Krakow. From the beginning of Japan’s Meiji restoration in 1868, the central government’s intense commitment to industrialization, militarization and imperialism sac- rificed local livelihoods and overran local resistance. After World War II, economic power replaced military power as the state religion of Japan. In 1961 the Yawata steelworks dumped 27 tons of soot and dust from its smoke- stacks every day. A fisherman whose livelihood was destroyed by the steelworks lamented: “With the development of the Japanese nation, and the development of this region, it is us fishermen who have become the victims” (quoted on p. 95). Until about 1965 Japanese polluters were protected in the national interest. Remarkably, between 1965 and 1985 Japan completely reversed its pollution policies and practices. By 1985 its citizens had nearly the cleanest air of any industrialized country. According to McNeill (p. 116), “women played a conspicuous role” in successful efforts toward smoke abatement in the United States, the United Kingdom, and Japan. The political, economic, and social status of women changed in most countries around the world during the twentieth century. In 1900 women had full suffrage in New Zealand, and in no other nation (Kerber 2001, p. 20), although a few American states allowed women to vote. The Constitu- tion of the United States was amended to guarantee women the right to vote only in 1920, 42 years after the amendment was first introduced into Congress. Between 1970 and 1990 the worldwide labor force participation of women relative to men nearly doubled. The number of economically active women rose from 37 to 62 for every 100 men engaged in the cash economy. These dramatic changes in the cultural and economic roles of women coincided with the equally dramatic fall in global fertility since about 1970. The direction of causation, if the link is causal, is not obvious, but the importance of culture in the change of fertility is clear. Political and economic changes in the status of women affect and are affected by changes in the number of children they have. Changes in the number, size, and composition of individuals and households have environmental consequences, for example through the number of refrigerators, housing units, and fuel consumption per person (Hunter 2000).

Market size and structure

That increases in industrial production and increases in income or purchas- ing power can (but need not necessarily) have environmental consequences is obvious and familiar. Economic integration has had environmental consequences that go beyond those captured by Affluence and Technology in the IPAT formula. As McNeill observes (p. 320), “Economic integration focused the dispersed demand of millions upon limited zones of supply.” When the object of purchase was biological, such as elephant ivory, rhinoceros

Click to return to Table of Contents 580 PDR 27(3) B OOK REVIEWS horn, giant panda skin, or tortoise shell, the rate of production typically could not be accelerated to keep pace with the increased demand. In such cases, instead of taking only a sustainable yield, extraction drew down the productive stock, pushing many exploited species toward extinction. Tech- nology interacted with economic integration. Information technology spread news about what could be bought. The invention of canning factories changed many salmon runs on the American West Coast from resources for local consumption to resources for global food markets. In many regions, traditional regimes for managing common property prevented the environmental damage that results from overexploitation. “The buffeting winds of globalization brought new shocks to these small- scale social systems. In fishing, for example, bigger operators tapping dis- tant markets [for both sales and capital investments] introduced trawlers and overwhelmed artisanal fisherman, whose common property regimes often collapsed. Free-for-alls ensued, and the fisheries collapsed too” (p. 321). Where economic integration, war, and political uncertainty weakened local people’s security in holding local productive resources (as in Ethiopia, China, and Russia), the incentives for speedy exploitation, even if unsus- tainable, increased. After World War II, “development banks” (such as the World Bank, the Inter-American Development Bank, the Asian Development Bank, and a few others) were charged with helping poor countries escape poverty. Using the funds generated by global economic integration, including funds from the oil trade, the development banks invested in infrastructure such as roads, power plants, dams, and extraction facilities for fuels and minerals. According to McNeill (p. 323), until 1987 the banks, with the consent and cooperation of recipient governments such as Brazil, India, China, and Indo- nesia, paid little attention to the ecological effects of their loans. Only after 1987, in response to American pressure, did the World Bank begin to require environmental assessments. Other banks changed their practices slowly.

Exogenous environmental change

Exogenous environmental change is part of the story of environmental change in the twentieth century. When drought hit Saskatchewan in the 1930s, the farmers were practicing cultivation techniques more appropriate to humid lands of eastern North America and Europe. Dust storms destroyed 3 to 4 million hectares of prairie wheat lands. By 1934 the dust blew as far east as the Atlantic. Farmers in Oklahoma and Kansas experienced similar distress in the Dust Bowl of 1931–38. The drought-stricken regions of western Canada and the United States saw farmers leave en masse. According to McNeill (p. 47), “two chief factors explain modern increases in erosion: migration or growth of population, and the intensification of market links.”

Click to return to Table of Contents B OOK REVIEWS PDR 27(3) 581

True, if farmers had not brought European agricultural practices into the prairies of United States and Canada, the massive soil erosion experienced in the 1930s could probably have been avoided. However, if there had not also been sustained drought, there would not have been dust bowls. Saline seeps, not caused by irrigation, retired about one million hectares of farmland in North America’s high plains between 1945 and 1990— an area equal to Lebanon’s (p. 47). Before the twentieth century, the end of the three-century-long European Little Ice Age around 1850 “may also have had a minor role in permitting the great modern expansions” (p. 17) of human numbers, consumption of energy, and economic production. I find it irresistible to corre- late the warmer and wetter climate of the European Medieval Warming of 1100–1300 (Roberts 1998, p. 215) with the onset and spread of European urbanization at the same time (Cipolla 1994, pp. 183–198). According to Roberts (1998, p. 214), “it is hard not to see a causal link between the medieval cooling of the fourteenth and fifteenth centuries and the extinction of the Norse settlement in Greenland around AD 1500.” Just as a single year is an insufficient basis to estimate the environmental variability of a century, so a century is an insufficient basis to estimate the environmental variability of the last hundred centuries. A longer perspective (Roberts 1998) is required to complement McNeill’s story of the environmental changes in the last century. Geologists call the period of the last 11,500 years the Holocene. The 115 centuries of the Holocene have seen enormous changes in landforms and landscapes. Some coastal regions rebounded upward when released from the weight of glacial ice, while others were submerged as the seas filled with the melted ice and seawater expanded in the warmth. Forests spread over vast regions that had been bare tundra in the previous ice age, which lasted nearly 120,000 years. Soils were formed and transformed. Until almost the time when the pyramids were built in Egypt about 4,600 years ago, what is today called the Sahara Desert had been a land of lakes and streams for more than 4 millennia. Humans were greatly affected by all these changes, but began to have major effects on landforms and landscapes only in the latter half of the Holocene, as agricultural implements and agricultural fields were invented and fixed human settlements were established. (Human mass-hunting and burning had important biotic effects much earlier.) A still longer perspective is required to see that even the dynamic Ho- locene has enjoyed an exceptionally placid and benign climate, compared with the rest of human prehistory. Multiplying 100 centuries by another factor of 100 reaches back a million years to the middle of the Pleistocene, when tool-making hominids already existed. In the last 800,000 years, there have been about eight major glacial-interglacial cycles, each lasting about 100,000 years, and for most of that time much more of the Earth than at

Click to return to Table of Contents 582 PDR 27(3) B OOK REVIEWS present was covered with ice. For example, about 18,000 years ago, the Wisconsinan ice sheet reached northern Pennsylvania, Ohio, Indiana and Illinois. Modern humans (Homo sapiens sapiens) first appeared about 120,000 years ago, at or shortly after the onset of the last glacial stage, which ended with the beginning of the Holocene. While many genera and species of Pleis- tocene plants and animals still survive, many large land mammals and birds have vanished, including mammoths, mastodons, long-horned bison, sa- ber-toothed cats, giant ground sloths, and great predatory birds with 25- foot wingspans. Native horses (now extinct) and camels crossed the plains of North America. Ice cores from Greenland and Antarctica demonstrate high levels of climatic variability on millennial and longer timescales (Turekian 1996, pp. 126–127). Abrupt transitions on the scale of decades may have been associated with major changes in oceanic circulation (Broecker 1997). Climatic variability has been hypothesized to have played a crucial role in the evolution of modern humans during the Pleistocene, although the details of this process remain controversial (Vrba et al. 1995; Richerson and Boyd 2000). The Holocene and Pleistocene perspectives reveal that the environments of humans have always been dynamic on the grand scale and that the thousand-fold growth of the human population since the beginning of the Holocene has been fostered and sheltered by exceptionally favorable climate. Recent biological, chemical, and physical perturbations by humans of oceans, continents, and atmosphere amount to a global experiment with unpredictable outcome. If the global cooling that has been the dominant trend of climate during the last 50 or 60 million years continues, and if humans continue to exist a few million years more, it seems plausible to expect humans to evolve and adapt further in response to climatic change. Apart from a few minor errors of detail, McNeill’s book is a trustworthy guide to the most recent century of humanity’s global experiment. (The statement on page 31 that coal extraction grew tenfold in the nineteenth century is contradicted by the data in the table on the same page, which show a 76-fold increase. The percent increases on page 264 should all be reduced by 100. The statement on page 205 that “All tropical diseases are waterborne in a loose sense” would not seem to consider African trypanosomiasis (sleeping sickness) or American trypanosomiasis (Chagas’ disease). The statement on page 212 that the land surface of the globe not covered by ice or sand is 133 million hectares should be 133 million km2. Small errors like these are inevitable in a work of this scope and complexity, but suggest that it would be worthwhile for readers to check details with independent sources.) These small errors do not undermine the strength and value of this book. Nature is variable. The Biblical story of the seven fat years followed by the seven lean years illustrates that the consequences for humans of natu-

Click to return to Table of Contents B OOK REVIEWS PDR 27(3) 583 ral variability in rainfall depend crucially on human preparations for and responses to such variability. Exogenous environmental change deserves recognition because it will be part of environmental change in the twenty- first century and thereafter.

Concluding remarks

In its report on development and the environment, the World Bank (1992, p. 39, Fig. 1.4) displayed a more general equation, based on the IPAT formula, to relate economic activity to its environmental impact. According to the World Bank’s equation, the quality of the environment equals the product of four factors: the scale of the economy (itself a product of income per person × population) × output structure (what people demand as final products, such as private cars or public transport, beef or beans) × input–output efficiency (the amount of various resources consumed to produce each final product) × environmental damage per unit of input. Here Population, Affluence and Technology all appear explicitly. Culture appears implicitly in the output structure and input–output efficiency (what do people want the economy to produce? how important are formal education and economically productive work in the culture? how well do workers apply knowledge to their economic activities?) (Cohen 1995, p. 388). Moreover, culture affects policy: the World Bank recognized that economic policies affect productivity and the composition of output (the first three of these four factors) while environmental policies change incentives governing the use of environmental resources (the last three of these four factors). The World Bank’s implicit recognition of culture gives no representation to McNeill’s accounts of the roles of high politics and global market structure in shaping the environmental impacts of economic activity. The World Bank’s equation certainly does not take account of exogenous environmental change. The empirical riches of McNeill’s book pose a conceptual challenge to general understanding that has not yet been met. Appreciation of the complexities he describes should save his readers from an unjustified faith in the promise of simple remedies to environmental challenges. McNeill writes in closing (p. 362): “Modern history written as if the life-support systems of the planet were stable, present only in the background of human affairs, is not only incomplete but is misleading. Ecology that neglects the complexity of social forces and dynamics of historical change is equally limited. Both history and ecology are, as fields of knowledge go, supremely integrative. They merely need to integrate with one another.” The science that integrates the human and natural history of the Earth remains to be created.

Click to return to Table of Contents 584 PDR 27(3) B OOK REVIEWS

Note

I thank Henry C. Harpending, John R. and development: Culture matters,” and McNeill, and Peter J. Richerson for helpful were posted on the Internet at http:// comments on previous drafts of this review. www.un.org/esa/population/cpdcohen.pdf. I Portions of this review were presented orally am grateful for the partial support of NSF to the United Nations Commission on Popu- grant DEB 9981552 and for the hospitality lation and Development, 2 April 2001, of Mr. and Mrs. William T. Golden during under the title “Population, environment the preparation of this review.

References

Broecker, W. S. 1997. “Thermohaline circulation, the Achilles heel of our climate system: Will

man-made CO2 upset the current balance?” Science 278(5343): 1582–1588, 28 November. Cipolla, Carlo M. 1994. Before the Industrial Revolution: European Society and Economy, 1000–1700. 3d ed. New York: W. W. Norton. Cohen, Joel E. 1995. How Many People Can the Earth Support? New York: W. W. Norton. ———. 2002. “The future of population,” in What the Future Holds: Insights from Social Science, ed. Richard N. Cooper and Richard Layard, pp. 29–75. Cambridge, MA: MIT Press. Commoner, Barry. 1971. The Closing Circle; Nature, Man, and Technology. New York: Alfred Knopf. Evans, L. T. 1998. Feeding the Ten Billion: Plants and Population Growth. Cambridge UK and New York: Cambridge University Press. Holdren, John P. and Paul R. Ehrlich. 1974. “Human population and the global environment,” American Scientist 62(3): 282–292, May–June. Hunter, David. 1999. “Global environmental protection in the 21st century,” Foreign Policy in Focus, Special Report #4, September. Albuquerque, NM: Interhemispheric Resource Cen- ter. Hunter, Lori M. 2000. The Environmental Implications of Population Dynamics. Population Matters project: A RAND Program of Policy-Relevant Research Communication. Santa Monica, CA: RAND. Kerber, Linda K. 2001. “Gender and inequality: Old questions, new answers,” Bulletin of the American Academy of Arts and Sciences 54(4): 11–31, Summer. Richerson, Peter J. and Robert Boyd. 2000. “The Pleistocene and the origins of human culture: Built for speed,” in Perspectives in Ethology: Evolution, Culture, Behavior, Vol. 13, pp. 1–45, ed. François Tonneau and Nicholas S. Thompson. «http://www.des.ucdavis.edu/faculty/ Richerson/Speed.htm» Roberts, Neil. 1998. The Holocene: An Environmental History. 2nd ed. Oxford; Malden, MA: Blackwell Publishers. Shweder, Richard A. 2000. “Rethinking the object of anthropology (and ending up where Kroeber and Kluckhohn began),” Social Science Research Council, Items and Issues 1(2): 7–9, Summer. Starke, Linda (ed.). 1999. State of the World 1999. New York: W. W. Norton. Turekian, Karl K. 1996. Global Environmental Change: Past, Present, and Future. Upper Saddle River, NJ: Prentice Hall. Vrba, E. S. et al. (eds.). 1995. Paleoclimate and Evolution, With Emphasis on Human Origins. New Haven: Yale University Press. World Bank. 1992. World Development Report 1992: Development and the Environment. New York: Oxford University Press.

Click to return to Table of Contents

B OOK REVIEWS PDR 27(3) 585

Population and Progress in the Middle Ages: A Review Essay*

JACK A. GOLDSTONE

David Levine’s At the Dawn of Modernity is like a magnificent medieval tapes- try: the richness of the detail sometimes overpowers the overall design. Amidst a sweeping and intricate depiction of medieval life in Western Eu- rope—including demography, family structure, technology, military techniques, urbanization, secular and sacred construction, religious organization, and political struggles—there lies a tale. In Levine’s view, that tale is nothing less than the origins of the modern world, here retroactively back- dated well beyond what historians have become accustomed to think of as “early modernity.” Where most historians see the origins of our modern world in the decisive break between the feudal, manorial patterns that prevailed before the Black Death and the more capitalist, individualist, secular societies that grew out of the Renaissance and Reformation, Levine claims that the decisive steps toward modernity were taken by the distinctive high medieval culture that developed between 1000 and 1300. In this view, Levine joins a growing chorus that includes Alfred Crosby (1997) and David Landes (1998), who also have argued that leaps in technological inventiveness during the Middle Ages propelled Europe on the path to modernity. Levine outdoes these authors, however, in the degree to which he weaves such advances into an encompassing story of social, political, and economic change.

*Review of David Levine, At the Dawn of Modernity: Biology, Culture, and Material Life in Eu- rope after the Year 1000. Berkeley: University of California Press, 2001. vii + 431 p. $50.00.

Click to print article Click to return to Table of Contents 586 PDR 27(3) B OOK REVIEWS

The argument for the medieval origins of modernity assumes particular pertinence these days, when another group of historians—whom I have referred to as the “California school” and with whom I confess to identify (Goldstone 2000)—has argued for precisely the opposite change in our dating of the origins of modernity. These scholars (Wong 1997; Frank 1998; Pomeranz 2000a) argue that as late as 1750 Europe held no major economic advantages over other leading civilizations; rather with regard to urbanization, standard of living, and involvement in long-distance and global trade, the leading areas of Europe were not ahead of the leading areas of East Asia. In particular, I have argued (Goldstone 1998) that the critical organizational elements that distinguish the modern West—secular, non- hereditary, and egalitarian politics; religion as a personal choice and not a state-backed official orthodoxy; and an industrial economy powered by fossil fuels—are not generally visible before 1800. Levine and the California school thus find themselves in agreement against the old historical mainstream, both arguing that no decisive change or break toward modernity occurred circa 1400–1500. Instead they see a strong continuity across the period from 1000 to 1800. The main difference—and a critical difference indeed!—is that according to Levine, the continuity is there because modernity is already evident by 1000; while according to the California school, the continuity is there because the key economic and organizational aspects of modernity do not arise until nearly 1800. Of course, it is difficult to argue that secular and egalitarian government, pluralism and free choice as the normal state of religion, and an industrialized economy were apparent much before 1800 anywhere in the world, not even in England or elsewhere in Europe. Thus to say that the year 1000 was the dawn of modernity must imply, first, that modernity does not lie mainly in these three characteristics of modern Western societies, rather these characteristics are secondary and necessary products of a more fundamental change in which modernity lies; and second, that whatever that more fundamental change is, it has clearly begun by 1000 in the northern and western regions of the European peninsula. Levine thus has two hurdles to pass in order to make his argument. First, he has to demonstrate that some critical breakthrough in Europe around 1000 set society on a distinctly modern path; second, he has to demonstrate that that path leads with very high likelihood to the destruction of traditional regimes, to religious pluralism, and to industrial society within the following millennium. Levine, in fact, does not even try to make the second and, in my view, more crucial point, as he ends his story in 1348 with the onset of the Black Death and the temporary demographic eclipse of growth in Europe. Rather, he assumes that if he can demonstrate the first point—that a characteristically modern social framework had developed in the centuries just after 1000, distinct from the Ancient/Classical

Click to return to Table of Contents B OOK REVIEWS PDR 27(3) 587 social framework that prevailed before 800—then the second point would follow without much difficulty. Here I would have to say Levine should know better. Levine’s greatest expertise is in social demography, and it may be true that recognizably modern family structures and demographic patterns can be seen in Europe after 1000. Yet traditional monarchies and lineage-based nobilities reigned virtually unchallenged until 1789. Religious wars and nationally exclusive churches prevailed everywhere until the seventeenth century (and even then the Anglican church restricted civil rights of Catholics in England until the nineteenth century, and the Dutch Re- form Church banned Jews, Catholics, and even Lutherans in many Dutch territories until the nineteenth century as well). Traditional power sources— wind, water, and muscle—had no alternatives in sight until the eighteenth century, and no real rivals until the nineteenth (Wrigley 1988). Why it took seven centuries after modern family structures were established for the other major props of traditional society to fall remains a significant problem for Levine’s story. Be that as it may, let us take Levine’s argument in the manner in which he frames it and concentrate on the first issue. Does he make the case for a distinctively modern social framework emerging from the high medieval culture that arises in the centuries immediately after 1000? We can easily concede two points: high medieval society was quite distinct in its social, technical, military, and economic organization from the Dark Age societies that dominated Europe from the collapse of the West- ern Roman Empire to the time the Domesday Book was compiled in the late eleventh century. High medieval society was also quite distinct from the more bureaucratic, politically unified, urban-based citizen-empire established by the Romans. So we have change, and lots of it—no doubt about it. That’s the easy part. But change in social formations is an old and universal story. The Buddhist-influenced Tang empire in China was quite different from the warring kingdoms that had immediately preceded it, and even more so from the Han empire. An even greater gulf of social change separates the emergent citizen-based urban polis-societies of Classical Greece from the dispersed agrarian society of the Greek “Dark Ages” or the even earlier Mycenean aristocratic kingdoms. The Ottoman empire was far more bureaucratic, militarily advanced, and more powerful than the fragmented medieval Islamic caliphates. Tokugawa Japan was a remarkably distinct political, social, and technological formation—far more decentralized, more urbanized, more demographically stable, and broadly prosperous—than Ja- pan in the previous century of civil wars or in the Ashikaga period that came before. Unless we call any change that results in more powerful, richer, or more technically advanced states “modernizing change,” in which case modernity starts with the first stone age settlements that replaced foraging societies, Levine cannot make his case simply by demonstrating, as he does

Click to return to Table of Contents 588 PDR 27(3) B OOK REVIEWS with unquestionable verve and skill, the extent to which high medieval society involves change from what went before. So the question must be: What are the specific changes that Levine points to after 1000, and why are those changes modern? Here we get to the crux of the argument. Levine does a praiseworthy job of showing how extensive were the changes between 1000 and 1300 in many fields. Let us start with power—wind and water mills grew by thousands to supplement muscle power. Muscle power was also improved by substituting horses for oxen and by harnessing the former to new deep-turning moldboard plows and to carts with the new and superbly efficient shoulder-harness horse- collar. Power was also refined and transmitted by iron; the growth of towns and the proliferation of village smithies making horseshoes and steel-edged tools greatly increased the capacity of medieval craftsmen. These energy sources allowed a vast tapping of new natural resources in heavy-soiled valleys and swiftly flowing streams, greatly increasing agricultural productivity and mechanical power for grinding, fulling, sawing, and other processes. Then there is the state. By 1300, the medieval monarchies of Britain, France, and Germany, though still far from the bureaucratized absolute monarchies of the seventeenth and eighteenth centuries that ruled over domesticated nobilities, had begun to take a shape that was recognizably different both from the old Imperium of Rome and the Carolingians and from the feudal vassalage of basically independent lords and knights that dominated the period from 800 to 1000. In social organization, Levine does a brilliant job presenting the latest theories of the interrelationships among family and household structure, manorial organization, and agriculture. The nuclear family of primogeni- ture appears and takes root in this period. Such family structures were far from the norm—in both England and France, counties where the eldest son inherited all lands were found next to other counties where inheritances were divided among heirs. Levine is deftly sensitive to local and cross- national variations in England, Germany, and France. Such variation makes it still unclear whether this family structure was the result of manorial control or of local custom. What is clear, however, is that a recognizably modern family life-course of relatively late, postpubertal marriage for women who entered simple nuclear family households had become established in northern and western Europe by the eleventh century—distinct from patterns both in Antiquity and elsewhere in the Old World. This late-marriage pattern itself has controversial origins. Levine notes that it might derive from ancient Germanic tribal marriages, in which women were expected to be full partners rather than junior dependents; or it might derive from the particular age distribution left north of the Alps by the Jus- tinian pandemics of the sixth century; or it might derive from practices of early Christianity with its negative attitude toward sexuality. Or it might

Click to return to Table of Contents B OOK REVIEWS PDR 27(3) 589 have arisen from an amalgamation of all these influences. What is certain is that this marriage pattern was distinctive, that it emerged sometime around 1000, that it suppressed potential fertility in the immediate postpubertal years, and that it remains characteristic of modern Western society. Finally, Levine notes that all of these changes together created a positive-feedback system from 1000 to 1300. Better tools and energy sources raised agricultural output; later marriage and undivided inheritances allowed accumulation of capital; these together raised the productive potential of society, allowing the growth of towns, higher population densities across the land, the growth of wealthier noble and middling classes, and the basis for more commerce and market-oriented behavior. These last in turn provided incentives for greater investment and production, and so on in a virtuous circle. This huge release of energy allowed a new “mastery of nature” by Europeans that in Levine’s view, aligning with Herlihy (1985) and other optimists about the high levels of medieval population, fueled a demographic expansion taking England and France to population numbers circa 1250 that were not surpassed for another five centuries (Levine, pp. 241–242). Levine thus argues that Europe was well on the path to modernity when the Black Death struck with unprecedented ferocity in the 1340s. Levine grants that the plague turned medieval relationships upside down, leading villeins to contest serfdom and win freedom, and leading to the replacement of manorial power by landlord power levied through market rents and dues. Levine likens the plague to the meteorite that wiped out the di- nosaurs: a sudden event that shifts the course of evolution. Yet Levine also argues that the plague was not necessary for subsequent modernization, for in the absence of the Black Death the energy unleashed by the high medieval burst of progress in so many fields would have found another way to continue forward. Many paths are possible, Levine notes; we just happen to trace our path through the fourteenth-century demographic debacle. This is a powerful argument. And I have not been able to do justice to the richness of the evidence that Levine brings forth, touching on every- thing from medieval carpentry to religious disputes, from Germanic settlement of Slavic lands to English inheritance practices. Levine’s knowledge of the period is encyclopedic, drawing on vast reading and masterfully presented. The book brings this period to life with a richness that has few rivals outside of classics by Huizinga (1954), Duby (1968), and Herlihy (1985). And yet in this very richness, contradictions abound. While extolling the enormous productive energy of the period 1000–1300, the virtues of the late-marriage family structure, and the inventiveness and uniqueness of this period, Levine still is forced to note that “the secular boom ran out of steam. By 1300, the first cycle of early modernization, which had originated with the positive feedback mechanisms that congealed after the year 1000, was over. Rural Europe had become a low-level equilibrium trap in

Click to return to Table of Contents 590 PDR 27(3) B OOK REVIEWS which demographic pressure seemed to forestall economic advance” (p. 287). Europe may have burst with productive and reproductive energies, but after three centuries the latter had come to overwhelm the former. By 1300, poverty apparently was widespread, a worsening climate produced sharply diminishing returns to recently claimed hill and coastal lands, forests and pasture were overcut and overgrazed, and “the benefits of commercializa- tion were largely thwarted” (p. 161). So what of progress? Had medieval Europe gone as far as it could go without industrialization? Perhaps. But Levine apparently thinks that Eu- rope had made such great leaps during 1000–1300 that it could emerge from the testing centuries of the Black Death with momentum to continue toward full modernization, while the rest of the world merely recovered. Yet these assumptions depend, in my view, on some very peculiar elisions and comparisons. The elisions are related to progress after 1500, where obstacles and detours on the road to full modernization are hardly visible from a medieval vantage point. Russia’s population and territories grew faster than those of any country in Europe to 1850; yet it lagged in every aspect of technological and political progress. Holland, by contrast, was in the lead in every technical and political aspect by 1670; yet it then stagnated, losing its religious pluralism, its thrust toward democracy, and its technological leadership. England by 1650 was in similar straits to England circa 1350: densely populated, with widespread poverty and inequality, plunging real wages, and widespread landlessness. Where and when did the leap to full political and economic modernization become inevitable? Not clearly in any of these cases, centuries after Levine’s dawn of modernization circa 1000. Levine at times seems more specific, arguing that progress in urbanization and agricultural productivity (including use of legumes and fertilizer to raise soil nitrogen) moved Europe to a uniquely high level that could not be gainsaid or fundamentally reversed. Thus Levine argues: “A new kind of society was emerging in northwestern Europe; this novelty is apparent when we contrast it with late Manchu China, where under 2 percent of the 400 million (i.e., eight million) people were nonproducers [i.e., nonagricultural households.] Perhaps 15 percent of the 40 million people in northwestern Europe were not peasants. These 6 million nonproducers were supported by the agricultural surplus” (pp. 238–239). This comparison makes thirteenth-century Europe (actually northwest Europe) appear able to support a nonagrarian percentage of its population over seven times greater than was the case in nineteenth-century China. Yet this comparison is skewed in many ways. By late Manchu times, China was in a period of great distress, and possibly more impoverished than it had been for many centuries, as Kenneth Pomeranz (2000b) has argued. The combination of massive civil wars in the mid-nineteenth century, the earlier Opium Wars and foreign incursions, the environmental

Click to return to Table of Contents B OOK REVIEWS PDR 27(3) 591 crisis of the Yellow River flooding and shifting course, and the decay of irrigation works and canals put China in a bad way. Plus, most of China’s population increase of the prior century had been in the poorer frontier regions, the European equivalent of the Russian steppes and Balkan moun- tain valleys. Europe’s nonpeasant proportion would shrink massively if the eastern European and Russian peasant populations are factored in; China’s nonpeasant proportion looks much larger if we compare an equivalent demographic region at a time before the crises of the nineteenth century. If we consider Jiangnan province in the lower Yangzi valley circa 1750, whose approximately 35 million people make it roughly equivalent to Levine’s 40- million-strong northwestern Europe, we find that in every material respect it compares favorably with the most advanced regions of northwestern Eu- rope. Pomeranz (2000a) argues that in the late eighteenth century, Chinese farming fixed more nitrogen (using soybeans and beancake fertilizers in addition to manures) than comparable English farms. This productivity supported a huge commercial trade: Pomeranz reports that over half the cultivated land in most Jiangnan prefectures was given over to nongrain crops, mostly silk and cotton for textile production and export. He notes that at least 10 percent of the rural population circa 1750 was in nonfarming households, and that another 15 percent of Jiangnan’s population was urban. Thus over 25 percent of Jiangnan’s 35 million people were “nonproducers” in Levine’s terms; this was almost twice the percentage of Levine’s northwestern Europe. Nor was this exceptional—“It has been estimated that 22 percent of Japan’s eighteenth-century population lived in cities, versus 10– 15 percent for western Europe” (Pomeranz 2000a: 35). The growth and extent of protoindustrial livelihoods in late Imperial China seems quite similar to that in late preindustrial Europe (Wong 1997: 38–42). This level of productivity was also accompanied by a mastery over nature in China— from the Grand Canal that crossed mountains and rivers to vast hydraulic works for harbors and irrigation systems (Elvin and Liu 1998). Levine also seems to feel that medieval Europe made real breakthroughs in power, noting the huge expansion of windmills and watermills. Yet even here there are odd comparisons. Vertical watermills, dating from a first- century design, were widely used in late Roman times, when Christianity ordered slaves to be freed. The basic design—a wooden undershot mill with straight blades—remained in use from well before Domesday to the eighteenth century. It was only in the later eighteenth century, when careful measurements by Smeaton and others demonstrated the superiority of over- shot wheels and wheels made of cast iron, and the nineteenth-century development of breastwheels and wheels with curved blades, that efficiency radically increased (Mokyr 1990). Moreover, the increase in the number of mills in England, from some 5,000 recorded in Domesday to perhaps 15,000

Click to return to Table of Contents 592 PDR 27(3) B OOK REVIEWS in the fourteenth century, is not striking when combined with Levine’s assertion that England in 1300 held “many times more people than in the year 1000” (p. 239), and that in the north population had increased by five to ten times (p. 160). Even allowing for improvements in efficiency resulting from the use of larger wheels, water chutes, and better gears, the overall ratio of mills to population showed no dramatic increase. In fact, as Levine himself points out, lordly monopolies, providing profits to lords from mill- ing fees, may have limited the spread of millpower and imposed heavy costs on the peasantry. Windmills were not, as Levine asserts, “an English invention” (p. 168). Horizontal windmills are attested in the Middle East from the seventh century, and the vertical mill design may have come to northern Europe from the Arab lands after the Crusades. What is clear is that the familiar vertical mills do not appear in Europe until the very late twelfth century, by which time the population had already grown considerably; nor do the basic designs change much from medieval times to the 1700s. Again, given the huge growth in population from 1000 to 1300 and the relatively late introduction of vertical tower mills, it is not clear that mills were producing huge increases in net power per capita. Given Levine’s assessment that “[b]y 1300, ... [r]ural Europe had become a low-level equilibrium trap in which demographic pressure seemed to forestall economic advance” (p. 287), it seems hard to argue that these were major power breakthroughs. Levine’s late-marriage system distinguished Europe from other regions. However, if James Lee and Wang Feng (1999: 38–47) are correct, China’s system of female marriage at puberty did not result in any greater demographic pressures than northwestern Europe’s late marriage system, because the Chinese used various means—including breastfeeding, infanticide, male sojourning, and prohibitions on widow remarriage—to limit fertility within marriage. Even though marriage in China took place shortly after puberty, this was not true of childbearing, for there was apparently an average in- terval of three years between age at first marriage and age at first birth. In addition, Chinese women had longer intervals between births and evidently stopped childbearing earlier than Europeans. Thus while northwestern Eu- rope had limited marriage access but high marital fertility, China’s family system had high marriage access but low marital fertility. In fact, in the eighteenth century, it appears that married women in France, England, Scandinavia, and Germany bore roughly twice as many children in their 20s as did married women in China. Although controlling access to marriage and controlling fertility within marriage were different means of limiting total fertility, both were evidently successful. The net result for key demographic variables in both Europe and China—in terms of completed household size, net reproductive rates, and long-term population growth from 1500 to 1750— was roughly the same. Medieval Europe thus broke through no preindustrial barriers: its levels of soil improvement, productivity, urbanization, and crafts-employed

Click to return to Table of Contents B OOK REVIEWS PDR 27(3) 593 population were not unique in the preindustrial world. What is clear is that in all of these respects, Europe was breaking through Classical levels of productivity, at least in northern Europe (Levine admits that southern Europe remained at or near Roman levels and technologies of agriculture until modern times). Yet this one-time breakthrough seems more like one of the great efflorescences or bursts of creative energy and technological and social organization that have periodically allowed expansion of preindustrial wealth and population, such as those of imperial Rome, Song China, Renaissance Italy, seventeenth-century Holland, and eighteenth-century Japan. None of the preceding, however, led to modernization in the sense of creating anything like the late-nineteenth-century European political, religious, and economic world. No doubt the medieval European advances contributed foundations for later change; in that sense they may have been necessary to later modernity, as were the achievements of Greece and Rome, the gun- powder of China, the astronomy and medicine of the Islamic world, and the “arabic” numerals of India. But as the recitation of this list makes clear, concatenation is not causation, and we are still uncertain as to the sufficient conditions, apparently arising in seventeenth- and eighteenth-century Eu- rope but at no other time and place in world history, that ignited the potential of all prior advances and produced democratic, pluralist, industrial societies. I have argued elsewhere that the major constraint on premodern societies with regard to innovation was the tendency of orthodox churches and states to combine to assert a conformity of practice and ideas, creating high risks for innovators (Goldstone 1987; see also Mokyr 1990). I have also suggested that a major bottleneck for preindustrial societies was not merely land or energy per se, but the ability to concentrate and cumulate motive power, a bottleneck that was only breached with the invention of practical steam engines that could convert fossil-fuel energy to motion on demand (Goldstone 1998). Levine does not touch upon issues of orthodoxy/pluralism or engines and modern science. Without attending to these issues, no amount of emphasis on the sheer exuberant growth that is apparent in high medieval society can bridge the gap that still looms to reach the modern world. I fear that the term modernization is losing its meaning when Levine can confidently describe the Gregorian remodeling of the papacy, with its effort to subordinate temporal authority and secular life to the authority of the Church, as “one aspect of the early modernization of Europe” (p. 189). At one time, the term modern clearly stood for an industrial, secular, indi- vidualistic mode of social organization; all preindustrial, religion-steeped, and communal societies were lumped together as traditional. This oversim- plified bifurcation of modernization theory was clearly unable to cope with the great variety and dynamic changes within traditional societies. Yet we now appear to be going too far in the opposite direction, by seeing any ten-

Click to return to Table of Contents 594 PDR 27(3) B OOK REVIEWS dencies within premodern societies that seem to have modern counterparts— urbanization, growth in local and long-distance trade, a rise in per capita incomes, technological improvements—as evidence of precocious modernization. What Levine’s book makes absolutely clear is that we need to extri- cate ourselves from the dilemma of seeing world history in dualistic terms, with modern societies characterized by growth, innovation, and the production of rising standards of living, and nonmodern or premodern societies characterized by slow growth, stagnant technologies, and subsistence-level production. In fact, periods of dynamic growth have occurred throughout history and civilizations (Jones 1988). The modern world, I believe, is distinguished by the mainly voluntary and pluralist character of religion, the secular and egalitarian nature of its governing institutions, and the carefully engineered application of intense and concentrated energy resources to all aspects of its economy. However, it is not distinguished by innovation or growth per se. The sooner we learn to accept that premodern societies—Chinese, Greek, Roman, Japanese, Hindu, Islamic, medieval Europe—were capable of re- markable bursts of creativity, invention, and growth, many of which in some way provided foundations for our late-eighteenth- and early-nineteenth- century development of modern societies, without necessarily entailing that last step, the sooner we will be able to properly approach the origins of modernization in all of its contingency and complexity. Levine shows with breathtaking depth and clarity that high medieval civilization was a kind of golden age, or efflorescence of growth. What he fails to show is that this constituted a clear breakthrough to modernity. What he provides, without any argument, is a panoramically rich, thoughtful, and challenging analysis of medieval economic and demographic growth.

References

Crosby, Alfred W. 1997. The Measure of Reality: Quantification and Western Society, 1250–1600. Cam- bridge: Cambridge University Press. Duby, Georges. 1968. Rural Economy and Country Life in the Medieval West. Trans. by Cynthia Postan. London: Edward Arnold. Elvin, Mark, and Ts’ui-jung Liu. 1998. Sediments of Time: Environment and Society in Chinese His- tory. New York: Cambridge University Press. Frank, Andre Gunder. 1998. ReOrient: Global Economy in the Asian Age. Berkeley and Los Ange- les: University of California Press. Goldstone, Jack A. 1987. “Cultural orthodoxy, risk, and innovation: The divergence of East and West in the early modern world,” Sociological Theory 5(2): 119–135. ———. 1998. “The problem of the ‘early modern’ world,” Journal of the Economic and Social His- tory of the Orient 41(3): 249–284. ———. 2000. “The rise of the West—or not? A revision to socio-economic history,” Sociological Theory 18(2): 157–194. Herlihy, David. 1985. Medieval Households. Cambridge, MA: Harvard University Press. Huizinga, Johan. 1954. The Waning of the Middle Ages. Garden City, NY: Doubleday.

Click to return to Table of Contents B OOK REVIEWS PDR 27(3) 595

Jones, E. L. 1988. Growth Recurring: Economic Change in World History. Oxford: Clarendon Press. Landes, David S. 1998. The Wealth and Poverty of Nations: Why Some Are So Rich and Some So Poor. New York: W. W. Norton. Lee, James and Wang Feng. 1999. “Malthusian models and Chinese realities: The Chinese demographic system 1700–2000,” Population and Development Review 25(1): 33–65. Mokyr, Joel. 1990. The Lever of Riches: Technological Creativity and Economic Progress. New York: Oxford University Press. Pomeranz, Kenneth. 2000a. The Great Divergence: China, Europe, and the Making of the Modern World Economy. Princeton: Princeton University Press. ———. 2000b. “Re-thinking the late imperial Chinese economy: Development, disaggregation, and decline, circa 1730–1930,” Itinerario 24(3/4): 29–74. Wong, Roy Bin. 1997. China Transformed: Historical Change and the Limits of European Experience. Ithaca, NY: Cornell University Press. Wrigley, E. A. 1988. Continuity, Chance and Change: The Character of the Industrial Revolution in England. Cambridge: Cambridge University Press.

Click to return to Table of Contents

596 PDR 27(3) B OOK REVIEWS

JULIAN L. SIMON The Great Breakthrough and Its Cause, edited by Timur Kuran Ann Arbor: University of Michigan Press, 2000. xi + 214 p. $39.50. Julian Simon, a maverick and controversial economist, died in February 1998. This book, unfinished at the time of his death, was edited by Timur Kuran and published posthumously. Although clearly not as polished as Simon would have wanted, it is nevertheless an effective restatement of his well-known position on the causes of modern economic growth. The essence of this position is that economic growth is caused by an increase in useful knowledge, which in turn is a function of population size. Large, dense populations are necessary to produce the critical mass that ignites the fires of technological progress and productivity growth; and thus, in Simon’s view, the more the merrier and the faster the better. He rejected with some contempt the various Malthusian doomsayers and wholeheart- edly rejected the contention that, as population grows, diminishing returns will set in. Instead, income will grow faster than population, because more people will create more knowledge, generate more major inventions, and thus increase productivity at a rate fast enough to guarantee increasing returns. The cause of the “great breakthrough,” as Simon liked to call the Industrial Revolution after 1760, is thus simple: it is the growth in population experienced everywhere in Europe after the middle of the eighteenth century. Economists and economically literate historians will scratch their heads. Simon is nothing if not enthusiastic, and he produces endless tables and graphs that seem at first glance to back up his argument. One of the least controversial observations in economic history is that predictions of doom and gloom about resource scarcity owing to population pressure, from Malthus to Paul Ehrlich, have not been borne out. Yet, scholars trained to believe in diminishing returns will have difficulty in shedding their prejudices altogether. Does a large population necessarily increase technical knowledge to the point where the increased pressure on natural resources is more than offset? To make his case convincing, Simon would have to produce a detailed model of how population growth leads to an increase in knowledge and how the expansion of such knowledge has the capacity to fend off population pressure on resources. Simon works, for all intents and purposes, with a sample of one: the experience of the Western world after 1750, what he calls “Sudden Modern Progress.” The stylized facts are familiar: in the middle of the eighteenth century population in the West started to grow rapidly, and this growth was followed by radical changes in science and technology. Simon refuses to entertain the notion that the causation may have been the reverse or that both the growth of knowledge and demographic change may have been related to other factors. For him, the arrow of causation that counts is that numbers drive knowledge. Scientists produce a roughly constant amount of new knowledge per capita, he believes, and so if there are more of them, the overall base of knowledge increases. And since knowledge is a public good, such an increase augments the amount of knowledge available to all of society. Why is it that Simon, despite his eloquence and deeply felt belief, has failed to persuade most of his peers? (Even Richard Easterlin, who contributed a respectful

Click to print article Click to return to Table of Contents B OOK REVIEWS PDR 27(3) 597 foreword to this book, adds that he disagrees with Simon’s interpretation.) There are two main reasons for doubt, one empirical, the other theoretical. The historical record is far less supportive of Simon’s case than he imagined. For instance, he argues, on rather weak grounds, that the decades following the sharp decline of population in Europe in the wake of the Black Death were a time of technological retrenchment. But the facts he produces (that land reclamation and polder-making ceased in the Low Countries in those decades, pp. 99–100) prove only that, not surprisingly, a decline in population reduced the marginal product of land. In fact the decades following the Black Death, during which European population was considerably lower than at the end of the thirteenth century, witnessed the introduction of blast furnaces, three-masted ships, carvel construction of ships, and in the end movable type printing. Whether these advances represented a de- celeration compared to the previous century is hard to say, but some doubt is cast on Simon’s views. But even progress in the period after 1750 is far more ambiguous than Simon gives it credit for. It is true that Britain, the cradle of the Industrial Revolution, experienced major population growth roughly at the same time, and that if we take Western Europe as our unit of analysis, the same pattern holds. But there are troubling outliers. Ireland up to the Famine experienced even faster population growth than Britain. France, the cradle of some of the most important scientific and technological breakthroughs of the nineteenth century, experienced a decline in birth rates that by that century’s end had brought it close to zero population growth. Twentieth-century population growth in much of Asia, from Karachi to Manila, seems to have led to little of what Simon hoped population growth will do. These historical facts are, of course, commonplace, and despite a final chapter to deal with these paradoxes, it is hard to see how he intended to dispose of them. One might also question Simon’s assertion that the mortality decline in the late nineteenth century was primarily a result of rising incomes as opposed to increases in knowledge (so that, in his view, the causal link goes from population to knowledge but not the reverse). He feels that the knowledge that dirt caused diseases was already widespread and that improvements in sanitation were income- induced. Perhaps so, but the “great sanitary awakening” took many decades to be widely accepted, and many of its implications were more complex and nuanced than the “filth kills” simplicities: answers such as how to produce and supply safe drinking water and milk, which diseases were caused by bacteria and which had other causes, and the modes of transmission of many germs were not really nailed down until the first decades of the twentieth century. Yet the deeper issue that ultimately renders Simon’s interpretation unsatisfy- ing is that his model of knowledge-creation is too simplistic. Knowledge is a tricky concept for economists to handle. Part of the problem is that the propensity of a society to produce knowledge is not just a function of its size: the number of scientists and engineers per capita, the willingness and ability of these people to engage in knowledge-creation, and the kind of topics that appear on their agendas can hardly be regarded as constant. Another part of the problem is that different societies tended to create different kinds of knowledge, some of which has technological significance and some of which does not. The Jewish sages who created the

Click to return to Table of Contents 598 PDR 27(3) B OOK REVIEWS

Mishnah in the second century AD were surely creating knowledge, but little of it led to technological advances. The same must be true for the medieval Islamic madrasahs and even for many of the universities of Europe, including Oxford and Cambridge, during much of their long histories. What Simon must have in mind is what we might call useful knowledge, knowledge that can lead to technological advances. Even then, however, there are hard problems he does not address. Who knew exactly what was known? How widely diffused was this knowledge? How easy was it to persuade others to act on this knowledge? How accessible was knowledge to the technicians and mechanics in their workshops and the farmers in the fields? How much contact was there between the savants and the fabricants? Equally perplexing, how much underlying knowledge was necessary to develop the new techniques that shifted the product possibility frontier outward? Much of the growth in agricultural productivity before 1850 had little scientific support, based as it was on serendipitous invention, trial and error, and custom. To be sure, the generation of trial-and-error knowledge may well have been a function of population size. But if it works everywhere as a historical rule, why do we observe the massively overpopulated impoverished regions of Java and the Nile Delta? Finally, most readers will be dissatisfied with the way Simon treats institutions and culture, which, by all accounts, played some role in the sudden takeoff of the Western world even if nobody can quite agree what that role was. Institu- tions, Simon argues, were important, but they were by and large endogenous to population growth. Perhaps, but population growth led to various institutional outcomes. Between 1850 and 1914 population in India, Germany, and Russia grew rapidly, but the institutions that emerged were quite different from one another. Simon’s explanation of such differences is that perhaps we should regard Eurasia as one big unit in which population was growing and thus was producing the conditions for the great breakthrough, and that different national “teams” were working on the technological advances that would create the modern world. “Team China” and “team India” simply were not quite as good as “team Western Eu- rope.” This line of arguing may take us back to institutions as an explanation. It is, of course, not quite fair to criticize an unfinished book. Given Simon’s erudition and intelligence, I am convinced that, had he been allowed more time, he would have filled in many of the gaps in this book. Since the death of Julian Simon, a number of major contributions such as Jared Diamond’s Guns, Germs, and Steel and Kenneth Pomeranz’s The Great Divergence have raised a number of new issues that Simon does not address. As it is, the University of Michigan Press has published a credo by a scholar with a vision. It is left to the rest of us to pick up where Simon left off.

Departments of Economics and History JOEL MOKYR Northwestern University

Click to return to Table of Contents B OOK REVIEWS PDR 27(3) 599

ANGELA HATTERY Women, Work, and Family: Balancing and Weaving Thousand Oaks, CA: Sage Publications, 2001. xiii + 233 p. $29.95. From January through March of 1994, 1,000 women gave birth in Dane County, Wisconsin—home of the city of Madison. A year later Angela Hattery sent 450 of these women a survey on childcare and employment. Half responded, of whom about a third volunteered to be interviewed. Hattery selected 30 married white mothers to interview: 10 who worked full-time, 10 who worked part-time, and 10 who stayed at home. Hattery uses her interview findings to question the ad- equacy of various theoretical approaches commonly used to study maternal labor force participation: structural functionalism, rational choice theory, neoclassical economics, feminism, and race/class/gender theory. She faults all these approaches for not recognizing how greatly maternal employment decisions are influenced by “motherhood ideology,” the set of beliefs a woman holds about appropriate behavior for mothers. Hattery contends that “the ideology of intensive motherhood” that arose in the 1950s “continues hegemonic rule” in American society. This set of beliefs holds that women are naturally better able to care for children than men and that mothers are better able than anyone else to care for their own children (p. 40). Five of the women Hattery interviewed, whom she labels “conformists,” so totally accept the precepts of this ideology that they willingly stay home even when their families live in near poverty. Fourteen others are “pragmatists” who make their employment decisions only after weighing the benefits of taking care of their own children against the improvement in living standards that their earnings might allow. Pragmatists tend to face the greatest number of conflicts. Those who choose to work worry of shortchanging their children, and those who choose to stay home worry about not providing adequately for their families. Hattery found eight “innovators” who sidestep this conundrum by consciously devising work options that allow them to simultaneously care for their children. Finally, three “nonconform- ists” believe none of the basic tenets of the intensive motherhood ideology, have full-time careers, and do not think that their working harms their children. As the catchy category labels indicate, Hattery makes an effort to attract a general audience. In fact, numbers and tables are hard to find in this volume. She makes scant mention of twentieth-century trends in women’s labor force participation, and even relegates the reporting of her own quantitative survey findings to three pages of commentary and four tables in an appendix. Even so, there is a reason for quantitative researchers to read this book: to hear mothers of 13-month- old-infants tell their stories about deciding whether to go back to work or stay home. Their explanations might be the simple rationalizations of individuals who find themselves in certain social circumstances with particular opportunity structures. They might also be the nuanced accounts of what truly motivates mothers’ employment decisions, choices that even the most sophisticated analyses would never be able to tease from survey data. Paying attention to such accounts can only add depth to our understanding of a dramatically changing aspect of Ameri- can family life. The experience of being a working mother with a young child, still uncommon in the 1950s, has now become the norm. If the intensive motherhood

Click to return to Table of Contents 600 PDR 27(3) B OOK REVIEWS ideology nevertheless rules supreme, then more American mothers are likely to feel distraught as they leave their small children in the care of others, and real strain should be evident in the American social structure. But Hattery’s research is not convincing concerning the societal significance of motherhood ideology. Her categories (conformist, nonconformist, pragmatist, and innovator) make sense only if the intensive motherhood ideology actually is hegemonic. But interview findings from 30 mothers can hardly be used to document this hegemony. Her survey produced a mean score of 23.4 on an eight-item “intensive mother ideology” scale, where a score of 8 indicates strong disagree- ment and a score of 40 indicates strong agreement (pp. 208–209). Apparently, mothers of young children in Dane County tend to have neutral feelings about the precepts of this ideology, a finding not commented on by Hattery. She relies on anecdotes to illustrate hegemony—the occasional court case, magazine article, or quote from a presidential candidate. Yet better evidence exists. For decades the General Social Survey has tracked American attitudes about mothers’ working and its effects on children (see GSS data at: www.icpsr.umich.edu/GSS/). These data show Americans becoming much more approving of working mothers, even those with young children. If beliefs and behavior are changing in such a coordinated fashion, then fewer problems might be in store for individuals and society. Such coordinated change also raises theoretical questions about the relationship between beliefs and behavior. Hattery wants us to see a woman’s employment decision as flowing from her beliefs about mothering. But might her beliefs about appropriate mothering in fact be changing as she finds herself moving into the labor force?

Department of Sociology DENNIS HODGSON Fairfield University

JOHN I. CLARKE The Human Dichotomy: The Changing Numbers of Males and Females Oxford: Elsevier, 2000. xiv + 146 p. $78.50. The population sex ratio—the number of males per 100 females—and its determinants are the topics of this slim volume. This close look at a demographic variable that has received limited attention quickly betrays one likely reason for that neglect: across populations, the sex ratio varies mostly within a very narrow band centered just below unity. Probably by necessity, most of the estimates presented throughout the volume are national statistics and, with the exceptions of Arab countries with high immigration, national sex ratios range from 85 to 110 males per 100 females (p. 18). At some point, the reader might thus be reminded of Ronald Rindfuss’s (1991) observation that demographers “have had the tough job of making a straight line interesting. But demographers are adept with boring data; we disaggregate” (p. 506). Indeed, John Clarke presents a number of disaggrega- tions of population sex ratios by residential area and by age group. But with the exception of areas of high and sex-selective migration and of the oldest age groups,

Click to return to Table of Contents B OOK REVIEWS PDR 27(3) 601 most of the hundreds of sex ratio values he provides remain stubbornly within the central range of 85–110. Perhaps demographers are supposed to take as an article of faith that understanding the determinants of any demographic variable is a worthwhile enterprise, but Clarke’s short introductory chapter on sex ratios and gender provides little justification, focusing instead on definitions and data. While the author never offers any theoretical framework for considering some values of the sex ratio more socially desirable than others, the dominant perspective one gathers from remarks scattered throughout the volume seems to be that a shortage of females leads to competition among males. For example, among males it generates violence (the Wild West in nineteenth-century America, p. 36, and the “many social problems” of Qatar and Oman, p. 121); and among females it leads to sexual exploitation (the “rising tide of prostitution in China,” p. 52, or large-scale polygyny, “only possible where there are unbalanced sex ratios,” p. 23). While such phenomena are probably mentioned to raise the level of interest in sex ratios, the author himself seems at times to doubt the validity of the causal links implied and seldom provides convincing evidence in this regard. Another theoretical perspective implicit in Clarke’s thesis is gender power im- balance. Even if the intention is laudable, his description of deviations from a balanced sex ratio, in either direction, as reflecting the lower status of women appears too systematic. The problem of too few (“missing”) females is traced to the “scandal of female neglect” (p. 83) or the “scourge of maternal mortality” (p. 80), but a shortage of men is also described as detrimental to women consequently exposed to a marriage squeeze (p. 23) or widowhood (p. 92). Without dismissing the negative consequences of outliving their male counterparts, one would think that those are still—as the joke about aging goes—preferable to the alternative. To keep an essentially statistical volume entertaining is no doubt a difficult enterprise, and the superficiality of many assertions probably reflects the author’s intention to keep the discussion light. Still, all but the most quantitatively inclined readers would likely need to find at the outset a compelling reason to ingest any high dose of statistics. The thinness and occasional imprecision of the argument are likely to disappoint others. The closing chapter, for instance, is a meager three- page discussion of the future. The list of factors “greatly affecting local, national and/or international population dynamics, and consequently sex ratios” (p. 122) includes the AIDS pandemic, movements of millions of refugees, and the effects of natural disasters, all factors that would seem to affect men and women in roughly equal numbers. While the book is hardly a compelling treatise, the author overall achieves his stated aim to describe “how births, deaths and migrations affect the numbers and balances of males and females in residential populations of all sizes” (p. 4). A chapter on sex ratios at birth discusses son preference and the surge in sex-selective abor- tions in East and South Asia. The resulting increase in a few national sex ratios at birth—for example, to 1.14 in China in 1989, and 1.16 in South Korea in 1993—is perhaps too recent to fully assess the social consequences. The analysis of mortality differentials by sex skillfully summarizes data that should be familiar to most demographers. Perhaps most engaging is Clarke’s discussion of sex-selective mobility, given

Click to return to Table of Contents 602 PDR 27(3) B OOK REVIEWS that migration data are less frequently analyzed and migration is really the only factor able to account for sex ratios substantially different from 100. For the wealth of data marshaled by the author, this book constitutes a comprehensive reference on population sex ratios, sex ratios at birth, and sex differentials in mortality and migration.

Department of Sociology PATRICK HEUVELINE University of Chicago

Reference

Rindfuss, Ronald R. 1991. “The young adult years: Diversity, structural change, and fertility,” Demography 28(4): 493–512.

GIOVANNI ANDREA CORNIA AND RENATO PANICCIÀ (EDS.) The Mortality Crisis in Transitional Economies Oxford: Oxford University Press, 2000. xxiv + 456 p. $95.00. This edited volume explores mortality trends in the chaotic environment of eastern Europe and the former Soviet Union during the 1990s. Recent years have generated a fashion for looking at this first post-communist decade in the region, often with insufficient grounding of such analysis in the societal contexts produced by the communist era itself (and by even earlier histories). Contributions in this book do pay some attention to that earlier context. However, the reference to “crisis” in the title is symptomatic of this broader problem in the field. While some countries in the region experienced rising mortality in the early 1990s, the magnitude of longer-term increases in death rates during the communist period itself in most cases dwarfed these short-term increases. The levels to which mortality returned by the second half of the 1990s also remained disturbingly higher than normal in a larger international context. The opening article by Cornia and Paniccià suggests that mortality trends in transitional economies in the 1990s represented a statistical deviation from preceding decades, referring specifically to short-term mortality spikes. As a matter of quantitative description this is true. However, underlying factors responsible for the slow degradation of survival chances for working-age adults (particularly men) before 1990 could also have contributed to the elevated mortality of the early 1990s. While the surface-level statistical trend may not be continuous, the underlying explanatory model very well may be. Rejecting famine-related and direct environmental effects along with erosion of health services as implausible and mis- placed explanations, the authors advance psychosocial stress as the key underlying mechanism responsible for the crisis, operationalized in terms of unemployment, labor turnover, and shifts away from marriage to other marital statuses. A chapter by Livi Bacci reviews mortality crises of the past eight centuries, finding tempo-

Click to return to Table of Contents B OOK REVIEWS PDR 27(3) 603 rary dramatic increases in mortality (particularly working-age mortality) that exhibit striking parallels to the contemporary situation under study. In the second section of the book, Moskalewicz, Wojtyniak, and Rabczenko take a refreshingly skeptical look at official statistics showing a decline in alcohol consumption in the 1990s, and instead estimate alcohol consumption from cir- rhosis mortality levels. They suggest a continuing increase in consumption during the post–state socialist period, but of course cannot address changes in the composition of alcohol consumed (whether beer, wine, or spirits). Using nutritional and household budget surveys, Paniccià documents dietary diversity in the region, concluding that empirical grounds exist for considering potential health and even mortality impacts of nutritional changes, particularly in Romania, Bulgaria, and Lithuania. He also suggests that impoverishment (particularly among middle-aged men) through loss of employment, breakup of marriage, homelessness, and other changes could exacerbate the effect of dietary factors. In their speculative essay, Marmot and Bobak focus on psychological and physiological pathways by which stress may affect cardiovascular disease mortality. The articles in the third section are less directly related to mortality trends. Tchernina’s case study from Novosibirsk contrasts entrepreneurial coping strategies of the intelligentsia with the passive despair of less-educated, less-adaptable segments of the population. Entrepreneurial behavior, while it may be uncom- fortable for the intelligentsia involved, is shown elsewhere in the volume to intensify mortality less than the passivity found in other population groups. In this sense, changes in life conditions perceived as painful, disorienting, and threaten- ing may actually be protective, compared to the alternative of passive resignation. Davis’s article on health production and medical system effectiveness during the economic transition compares health service performance across countries. The relevance of this line of inquiry to elevated mortality, of course, is weakened by the observations, found in several other places in the book, that most of the mortality increase did not occur among population groups most affected by quality of health care. In fact, not many data are available to suggest that these most-vulnerable persons (infants, children, disabled and chronically ill adults, the elderly) have experienced a mortality crisis at all. Finally, Nesporova addresses labor market policies by which various countries in the region attempted to fight unemployment. The effectiveness of such policies is relevant because she links unemployment to stress, and stress to mortality, based in part on research presented elsewhere in the book. In the fourth and final section of the book, Riphahn and Zimmermann find that psychosocial stress explains the small mortality spike observed in 1990–91 in eastern Germany. Shkolnikov and Cornia suggest that at least part of the upswing in Russian mortality in the 1990s may reflect postponed effects of long-term alcohol abuse. The article on Latvia by Krumins and Usackis finds further support for psychosocial stress as a causal factor. Blazek and Dzúrová present what they call a “counterfactual” case study detailing a decline of mortality in the Czech Republic, and here if anywhere we are presented with a legitimate argument that better delivery of medical care had an impact on health and survival during the 1990s. One cannot help observing that in market-oriented health care, better care is pos-

Click to return to Table of Contents 604 PDR 27(3) B OOK REVIEWS sible but available differentially according to ability to pay for it. This applies to countries as well as to individuals, and the less traumatic transition experienced by the Czech Republic (including low levels of both unemployment and income inequality) may have provided an opportunity for better medical care to make a difference. The final four articles consider countries outside the former state socialist zone. Are there links between economic dislocations and mortality also in situations where sudden transformation of state socialism to market economies is not a factor? Abdala, Geldstein, and Mychaszula show that in Argentina, rapid industrialization and societal transformation beginning in the mid-1970s were accompanied by rising working-age male mortality, including intriguing parallels with the cause-of-death structure observed in eastern Europe. Jäntti, Martikainen, and Valkonen conclude that unemployment insurance provisions and other features of Finland’s welfare state prevented any increase in mortality despite rapid economic transformation and increased unemployment in the 1990s. Thus, the analysis of Finland and Argentina broadens the applicability of the psychosocial stress argument. Eriksson correlates an increase in unemployment in Denmark in the 1980s with admissions to psychiatric hospitals, reinforcing the unemployment–psychosocial stress linkage. McDonough, Duncan, Williams, and House explore deterio- rating black male mortality relative to that of whites in the United States, interpreting it as an income effect. They find correlation between income, income trends (particularly income loss), and income differences, on the one hand, and mortality risk, mortality trends, and mortality differences on the other. This correlation persists when circumstances that might jointly determine both outcomes are controlled. What is missing, as usual, from this income-based explanation is identification of the specific mechanism by which income “buys” increased years of life (or prevents death). Such a reduced-form model absolutely requires the additional insights about specific linkage mechanisms offered by the rest of this excellent volume. In particular, the underlying theme of the book is the identification of psychosocial stress as an outcome of complex social-structural processes, and as a key risk factor affecting the chances of survival. The concept of stress, however, still needs refinement. For example, many psychologists have concluded that stress, like cholesterol, may come in “good” and “bad” forms (sometimes called eustress and distress). While bad stress may constitute a mortality risk, good stress may be as necessary for healthy life as are essential fatty acids. Future research that moves in the direction of conceptual refinements will be forced to delineate the conditions of stress, systematically studying the context of institutionalized social forces that shape our lives at the psychosomatic level. As one step along that path, this volume is essential reading, not only for those interested in demographic changes in eastern Europe, but for anyone seeking better understanding of the interface between social structure and well-being and survival in human societies.

Max Planck Institute for Demographic Research ELWOOD D. CARLSON Rostock, Germany

Click to return to Table of Contents B OOK REVIEWS PDR 27(3) 605

CALEB E. FINCH, JAMES W. VAUPEL, AND KEVIN KINSELLA (EDS.) Cells and Surveys: Should Biological Measures Be Included in Social Science Research? Washington, DC: National Academy Press, 2001. xii + 374 p. $49.00 (pbk.). Bioindicators, biomarkers, demogenes, allostatic load, and biodemography: are we about to experience a revolution in the demography of health and mortality? Should surveys become an important vehicle to reveal how cells influence health? This interesting book addresses these and other questions based on 14 contributions made at a workshop organized by the Committee on Population of the Na- tional Research Council, as a follow-up to a publication of the same committee in 1997 that focused on the demography and biology of human longevity (Wachter and Finch 1997). Nearly half of the chapters in the current volume treat issues related to aging, but the book provides ample materials beyond this theme. Biodemography is “an emerging field of research that seeks to integrate and translate findings from a variety of disciplines (e.g., demography, evolutionary and molecular biology, genetics, epidemiology, ecology) into their effects on population health status and individual/social behavior” (p. 340, glossary). With biodemography comes a range of measures that are based on biological materials (such as blood or urine), physiological measurements (such as blood pressure or height), or environmental exposure measurements (such as lead or specific air pollutants). The chapters in Cells and Surveys describe recent technical progress in biodemography and its implications for survey-based research, examining the kind of information that can be included in surveys and suggesting the potential benefits and costs of including such information. In the brief closing chapter Kenneth Wachter recalls the introduction of an- thropometric data collection (height and weight) in surveys in England in the late nineteenth century. The addition of these biological indicators to social surveys led to new insights on the relationship between socioeconomic environment and human height, to increased collaboration between the biomedical and social sciences, and to new statistical approaches (linear regression by Francis Galton). Wachter argues that in many ways the new interest in biological indicators in the context of social surveys is likely to have similar consequences: more studies on gene–environment interaction, more interdisciplinary research, and development of new statistical approaches. How can these new developments help survey research in the social sciences? And how can social surveys help genetic and biomedical research by including bioindicators? Maxine Weinstein and Robert Willis suggest that traditional survey research will benefit because self-reports can be calibrated with biological and clinical measures of health and diseases. In addition, research on causal linkages between the social environment and health may become much more elaborate and many more causal pathways could be explored. One of the challenges will be how to deal statistically and conceptually with the large increments in numbers of variables in biodemographic research. Demographic research on fertility and mortality has made extensive use of conceptual frameworks—especially the proximate determinants framework—that link the social and biological systems. Eileen Crimmins and Teresa Seeman build upon this approach, presenting a conceptual

Click to return to Table of Contents 606 PDR 27(3) B OOK REVIEWS framework that links demographic variables (such as age, sex, and socioeconomic status) as well as social, psychological, and behavioral influences to health outcomes operating through a set of biological mechanisms. There is little doubt, however, that including bioindicators in surveys will present a major methodological challenge, if only because bioindicators abound and even more are expected to become available in the near future. Several contributors to the book express the need for a more integrative model of biological risk that focuses on the combined effects of multiple biological parameters as they influence risks for various health outcomes. An example is the concept of allostatic load, which is considered to represent the cumulative toll of dysregulation across multiple physiological systems over time, reflecting both a multisystem and life- span orientation. Allostatic load is hypothesized to affect subsequent mortality, disease pathology, and aging. In studies of aging, the operational measure of allostatic load includes information from about ten biological parameters that reflect functioning of different physiological systems (e.g., cardiovascular and metabolic systems) and is strongly related to health outcomes. With the rapid advancement of genetic knowledge, surveys are increasingly important to obtain data from nonclinical samples. Wachter warns that social scientists need to temper overly enthusiastic interpretations of genetic information that may lead to crude biological determinism. Indeed, it is likely that news will outpace evidence during the next decade and lead to an outburst of supposedly biologically driven reports (a gene for lying, a gene for math, etc.). Several authors note the key role of surveys in analyzing gene–environment interaction and emphasize the public health dimension of genetic study findings. The rapid advances in the measurement of environmental exposure provide an additional dimension. Surveys offer the opportunity to make environmental observations and collect environmental specimens that are potentially important in defining health outcomes. In his chapter on genetics and demography, Douglas Ewbank coins the term demogenes. The defining characteristic of such genes is that they have a notice- able effect at the population level. At present, there is only one gene in that class. The apolipoprotein E gene (APOE) is clearly related to a number of major health outcomes in the older population (mortality, heart disease, cognitive functioning). The impact of genetic research on demography during the next 10–15 years will depend on how many additional demogenes are discovered. Ewbank identifies several criteria for demogenes. They should be associated with the most common diseases, causes of death, or other variables of interest to demographers and must be major genes associated with substantial variation in risk (e.g., genes with frequen- cies of at least 5 percent in the population of interest). These two criteria determine the attributable fraction, that is, the proportion of cases of a disease or death that are associated with a given factor. Other criteria for demogenes are substantial variation across populations and interaction with social and behavioral variables of interest to demographers. Finally, inherited mutations are likely to be more useful for demographic research based on large surveys than are somatic mutations, as the latter tend be localized in individual organs or certain cells, which may affect screening feasibility in surveys.

Click to return to Table of Contents B OOK REVIEWS PDR 27(3) 607

The book also includes chapters on the value of other methods for biodemographic research, including the use of sibling and other ”relational” data, autopsy, animal models, and functional assessment. Additional chapters address the potential disadvantages of including biological measurements in social surveys and discuss the complicated ethical issues that face researchers who attempt to incorporate bioindicators in surveys. This book is one of the first to review and discuss the implications of advances in the biological sciences for demographic research in a broad perspective. In recent years, epidemiologists have been engaged in a lively debate about the consequences of the progress in genetics and molecular biology for their discipline (e.g., Susser and Susser 1996; The Lancet 1997). Several epidemiologists are wary of the possible explanatory power of new biological measurements, fearing they may affect the orientation of the discipline and pull it too much in the biomedical direction. There are concerns such micro-level research may expand at the cost of social epidemiology and public health research. Demography, as an observational discipline that contributes significantly to public health by describing processes and exposing inequalities in health, has an important role to play in these new developments. While most epidemiologists seek to understand the causes of specific diseases, demographers have primarily been occupied with the social and economic context of health. Biodemographic research is needed to put biological advances in a social and population health context, and population-based surveys are a crucial instrument to study interaction between social and biological systems and to expose inequalities. Therefore, the inclusion of biological measurements should be considered in the design of social surveys, for there is little evidence that doing so affects the traditional nonbiological sections of such surveys. The greatest potential lies in carefully designed longitudinal studies that include multiple measurements of a range of bioindicators and nonbiological variables over time. But as technology advances, single-round surveys may also become increasingly valuable in the measurement of past exposures, disease incidence, and disease history. Given the complexity and rapid changes in the field of biological measurement, wide interdisciplinary consultation—among geneticists, environmental scientists, epidemiologists, molecular biologists—is a requirement, as will be very evident to the reader of this book.

Carolina Population Center J. TIES BOERMA University of North Carolina at Chapel Hill

References

Susser, M. and E. Susser. 1996. “Choosing a future for epidemiology: II. From black box to Chinese boxes and eco-epidemiology,” American Journal of Public Health 86(5): 674– 677. The Lancet. 1997. “Putting public health back into epidemiology” (editorial), 26 July 250(9073): 229. Wachter, Kenneth W. and Caleb E. Finch (eds.). 1997. Between Zeus and the Salmon: The Biodemography of Longevity. Washington, DC: National Academy Press.

Click to return to Table of Contents

608 PDR 27(3) B OOK REVIEWS

SHORT REVIEWS

by John Bongaarts, Susan Greenhalgh, Geoffrey McNicoll, Michael P. Todaro, Etienne van de Walle, Zachary Zimmer

GAY BECKER The Elusive Embryo: How Women and Men Approach New Reproductive Technologies Berkeley, Los Angeles, and London: University of California Press, 2000. x + 320 p. $45.00; $17.95 (pbk.). The rapid spread of new reproductive technologies over the last quarter-century has given rise to a mushrooming literature exploring the social, political, cultural, and ethical issues surrounding the technologizing of childbirth. In this incisive study, Gay Becker, medical anthropologist and professor at the University of California, San Francisco, draws on her interviews with some 300 American men and women who used these technologies to conceive a child, relating their stories about the effect of that decision on their bodies, identities, and life plans. Becker demonstrates that what is at stake in the high-risk, high-cost, high-tech quest for a biological child is not just the possibility of having a biologically constituted family. What drives people with the financial wherewithal to undergo these costly, physically trying, and emotionally agonizing procedures is also the desire to live out deeply embedded cultural scripts depicting the “normal” American life course and “proper” womanhood and manhood, scripts that entail biological reproduction. Those who “fail” must reconfigure their gender identities, life courses, and much more. Written in readable, nontechnical prose, and filled with the absorbing stories of her informants, Becker’s account covers every stage of the process that couples undergo, from the stigma attached to infertility, to struggles with the body, decisions about donors and adoption, and, finally—for those who are unsuccessful in their quest—“letting go” of biological continuity and designing new blue- prints for life. Becker finds, as others have, that the new technologies and associated practices tend to reinforce old ideologies—in particular, the imperative of parenthood and the patriarchal association of women with biology and childbearing. Having herself lived through, and then moved beyond, the experience of infertility that was never “cured,” Becker is frustrated by the stubborn perseverance of cultural narratives that sustain ideologies equating parenthood with personhood. Such ideologies, she believes, limit our view of life’s possibilities. Becker might have expanded her sense of “life’s possibilities” had she included among her informants some people who never accepted as “normal” or “good” the dominant cultural narrative that makes biological parenthood the defining core of American personhood. In The Elusive Embryo, those cultural dissenters are marginalized and rendered invisible by statements such as “Parenthood equals normalcy for people

Click to print article Click to return to Table of Contents B OOK REVIEWS PDR 27(3) 609 in their middle years” (p. 1). Another puzzling shortcoming is Becker’s failure to situate her informants’ narratives about reproduction in the context of their lives as a whole. We learn much about their anxieties regarding bad sperm and bad eggs, loss of control over their lives to doctors, and so on, but nothing about the way these apprehensions fit into the larger picture of their lives: their jobs, their hobbies, their histories, and so on. These, however, are relatively small quibbles. Becker has written an accessible and honest account that will enlighten specialists in reproductive politics and culture while helping infertile individuals undergoing the traumatic quest for a child make sense of what is happening to them. Highly recommended to students of reproductive technology, medicine, and gender. Ref- erences, index.—S.G.

THEODORE CAPLOW, LOUIS HICKS, AND BEN J. WATTENBERG The First Measured Century: An Illustrated Guide to Trends in America, 1900– 2000 Washington, DC: The AEI Press, 2001. xv + 308 p. $20.00 (pbk.).

STEPHEN MOORE AND JULIAN L. SIMON It’s Getting Better All the Time: 100 Greatest Trends of the Last 100 Years Washington, DC: Cato Institute, 2000. xviii + 294 p. $29.95; $14.95 (pbk.). US social, economic, and environmental trends over the last century are presented in simple and colorful charts in these two volumes, issued by prominent conservative think-tanks. In both of them the text amounts to only a few paragraphs on each chart, but the data sources are fully documented. There is a large measure of overlap of content between the two. Moore and Simon, as their title indicates, are resolutely focused on what the authors see as the good news. The book is described in the foreword as “one of the last” of Julian Simon’s manuscripts left unfinished at his death in 1998. It is the sixth of them so far to appear in print. Simon’s ebullience is evident throughout: the charts end with “the greatest trend of all: 270 million Americans and growing.” The authors write: “All of the evidence in this book documents that in every material way, life in the United States is much better today with 270 million people than it was in 1900 with 70 million people.… the American people are net resource creators, not resource depleters—protectors of the environment, not destroyers. Each generation leaves the ecological fate of the planet and our continent in better condition for future generations.” Caplow et al.’s The First Measured Century, which originated as empirical backup for a television program of the same name, has a broader range and is more fully informative, with greater reliance on primary sources. Its authors are interested in changes for the worse as well as for the better—for example, in rates of drug use, divorce, and incarceration. Some of its data come from a 1999 resurvey of Muncie, Indi- ana, the site of the classic Middletown study of the 1920s, allowing comparisons of small-town behaviors and attitudes over a 75-year period.—G.McN.

Click to return to Table of Contents 610 PDR 27(3) B OOK REVIEWS

ELISABETH CROLL Endangered Daughters: Discrimination and Development in Asia London and New York: Routledge, 2000. xiv + 207 p. $75.00; $24.95 (pbk.). The Asian giants China and India have made a mighty contribution to the decline in global birth rates, but at a heavy cost: the loss of perhaps 100 million girls. In this short primer on the subject, anthropologist and development consultant Elisabeth Croll, who has written extensively on gender, family, and development in China, provides a timely overview of work by others, primarily demographers and anthropologists, on the “missing girls” in East, South, and Southeast Asia, adding her own observations about the “voices” of people in the region and the programs that have been initiated to put an end to discrimination against daughters. Drawing on the research of many demographers, the book presents the available longitudinal data on sex ratios at birth and sex ratios of child mortality for China, South Korea, Taiwan, Vietnam, India, Bangladesh, and Pakistan. Although the most recent data included are nearly a decade old, Croll has performed a useful service in bringing together information culled from a variety of sources. The story the numbers tell is that excess female mortality is not only widespread in these countries, but is increasing as fertility falls and incomes rise. Focusing on India and China, Croll explains the widespread daughter discrimination as a product of a “culture of gender,” rooted in the patriarchal family, that treats sons and daughters as complementary but unequal family members and leads parents to a “gender reasoning” that encourages them to pare back the number of daughters they raise, whether by infanticide, selective neglect, or, more recently, sex-selective abortion. Croll gives this analysis ethnographic flavor by sharing the “voices” of some people (who? when? where?—we are not told) whose experiences support the narrative. The treatment here is less anthropological than journalistic, for a serious anthropological analysis would draw on sustained fieldwork in one or more locales and would include identified voices that challenge the dominant narrative as well as those that support it. To be sure, Croll’s account of family-level dynamics captures a critical part of a complex reality. Yet this master narrative of the “Asian patriarchal family,” long an anthropological staple, tends to treat Chi- nese and Indian families as homogeneous and unchanging, neglecting the great variation from place to place and the paradoxical changes that have, in some situations at least, disrupted the patriarchal plot by favoring daughters. Also problem- atically, Croll’s analysis places the blame on families, even though families make decisions within larger political and economic contexts that constrain their choices. The author’s decontextualized treatment of the microlevel sources of daughter dis- appearance is not only empirically thin, it is politically consequential, for it de- flects attention from the role governments may play in fostering the gender disparities that she, like many others, is concerned about. Indeed, the author evinces a touching faith that governments, once confronted with gender-disaggregated data on child mortality, morbidity, and human development, will promptly act to alle- viate the disadvantage that girls suffer. Once they see the terrible truth of gender inequality, she reasons, they will make the needed changes. But the policymakers of China and India already possess the relevant figures, yet they have been unable or unwilling to devote the political capital necessary to successfully promote girls’

Click to return to Table of Contents B OOK REVIEWS PDR 27(3) 611 survival and development. Without sustained analysis of the political reasons that the myriad programs and policies already in place have had so little effect, this call to arms is unlikely to produce an effective policy response. What is needed is a much more robust and politically informed micro-cum-macro analysis of the causes of discrimination against daughters and of the failure of current policies to stop it. One hopes that Croll’s readable if insufficient account will inspire others with greater breadth of analytic vision to undertake that larger project. Meanwhile, Croll’s book should serve as a handy introduction for those unfamiliar with the topic. Notes, index.—S.G.

BARBARA ENTWISLE AND GAIL E. HENDERSON (EDS.) Re-Drawing Boundaries: Work, Households, and Gender in China Berkeley, Los Angeles, and London: University of California Press, 2000. xii + 344 p. $50.00; $19.95 (pbk.). This ambitious volume was originally envisioned as a vehicle for the interdisciplinary creation of a synthetic model of the links between work, households, and gender in reform-era China. This was a tall order—indeed, so tall that the editors, professors of sociology (Entwisle) and social medicine (Henderson) at the Univer- sity of North Carolina, Chapel Hill, eventually abandoned it, shifting the focus to elucidating some of those links while simultaneously documenting the slipperi- ness, fickleness, and boundary confusions of the core concepts. Each of the book’s four sections addresses a particular part of this revised agenda: work, gender, internal migration, and the household. Contributors to each section are concerned to take the constructs apart, revealing their historical variability and contexualized meanings, while exposing their relationships with each other. Each contribution provides fascinating glimpses into a society undergoing rapid transformation. In- dividual chapters show that the political and economic upheavals of the last two decades have revived male advantage in some domains, while bringing women empowerment and enrichment in others. Despite the broad and interesting topics covered in the book’s 18 chapters, however, the volume is less satisfying than one might have hoped. Empirically, it offers a kind of anti-conclusion: that in contemporary China, gender, work, and household are elusive constructs about which no generalizations can be made. Theoretically, the volume presents much hypothesis-testing, but little if any explicit theorizing. The volume includes the work of a prominent cast of sociologists, demographers, historians, and anthropologists spe- cializing in China, many well known to the population studies community. (In addition to the editors, the list includes Sidney and Alice Goldstein, William L. Parish, Wang Feng, and Martin King Whyte.) Although all of the authors address related topics, the wide variety of data types and methodologies employed results in loosely connected chapters that sit awkwardly together in a single volume. De- spite these problems of integration—which can be traced to the editors’ admirable if not easily realizable ambitions—the arguments challenge our ideas about contemporary Chinese society in fresh and useful ways. Re-Drawing Boundaries will have much to offer both China hands and comparativists looking to China for sociological insight. Consolidated reference list, index.—S.G.

Click to return to Table of Contents 612 PDR 27(3) B OOK REVIEWS

DAVID T. GRAHAM AND NANA K. POKU (EDS.) Migration, Globalisation and Human Security London: Routledge, 2000. x + 222 p. $85.00. The framework loosely linking this collection of essays, set out in an early chapter, contrasts orthodox ideas of national security and their backing in realist international relations theory with concepts of “human security” that emerged in the 1990s. In a system of sovereign states migration is one component of interstate relationships— unsymmetrical, sometimes violent or messy (particularly in the case of refugee movements), but regulated as a matter of right by government. From a human security standpoint, concerned with people’s freedom from fear and want, such sovereign rights are attenuated at best. Migration becomes an issue for moral consideration rather than for calculation of economic or political interest. The remaining chapter authors fall variously between the two positions, some emphasizing the national security threats posed by migration, others the obstacles (more or less effective) that states put in the way of migrants’ search for safety and opportunity. (A couple of chapters are not about migration at all.) Topics include global diasporas, the “securitisation” of migration in Sweden and Malawi, the politics of North African and Kurdish migrants in Europe and the United States, the “metasocieties” of Polynesians in the Pacific rim, and the post-Soviet brain drain from Russia. Index.—G.McN.

PAUL HARRISON AND FRED PEARCE AAAS Atlas of Population and Environment Berkeley: University of California Press, 2000. xi + 204 p. $65.00; $29.95 (pbk.). A summary of data and research findings on time trends in primary production, resource supplies, environmental conditions, and population is assembled in a readable volume packed with attractive charts and maps. There are three parts: overview, including a chapter on “the theory of population–environment links”; Atlas, consisting of short chapters on particular products, resources, and types of ecosys- tem, where apparent effects of population increase can be pointed to; and case studies—six potted stories of population-caused devastation, contributed by the World Wildlife Fund and The Nature Conservancy. The Atlas conveys a strong environmentalist message, reinforced in a foreword by Peter H. Raven: things are getting worse, quickly. The contribution of population growth to environmental degradation is backed by reference to I=PAT. The complexity of the actual relationship is acknowledged, but little attention is paid to data controversies or to apparently favorable trends (on which, for example, see Bjørn Lomborg’s The Skep- tical Environmentalist). This is popular science rather than an authoritative data source. A conscientious user of the Atlas would have to track the information in its highly abbreviated tables and simple charts back to the original source for con- firmation and complicating footnotes. Thus it does not seek to compete with the mass of data to be found in the country tables of the World Resources report series issued by the World Resources Institute. (The 2000–01 edition of the WRI report, subtitled People and Ecosystems: The Fraying Web of Life, is specifically concerned with how ecosystems have changed under population growth. It is available online at www.wri.org/wr2000.) The Atlas might also be criticized for allowing graphical

Click to return to Table of Contents B OOK REVIEWS PDR 27(3) 613 design sometimes to take precedence over content, as with use of Mercator-type projections (a world dominated by Russia, Canada, and Greenland) for the many global maps—giving a visually misleading picture especially for data such as land cover—or the squashing of the numerous pie charts into ovals. The volume is produced under the auspices of the American Association for the Advancement of Science. Index.—G.McN.

RUSSELL KING, PAOLO DE MAS, AND JAN MANSVELT BECK (EDS.) Geography, Environment and Development in the Mediterranean Brighton, UK and Portland, OR: Sussex Academic Press, 2001. xii + 292 p. $69.50; $29.95 (pbk.). This volume is based on papers originally presented at the 28th International Geo- graphical Congress, held in The Hague in August 1996. Although the Mediterra- nean encompasses a diverse group of countries, the contributors confine themselves almost exclusively to a European perspective on the region. Other parts of the Mediterranean Basin such as North Africa and the eastern Mediterranean states receive only passing attention. Given that the basic perspective is geographic, it is natural to find several chapters devoted to questions regarding the definition and character of the Mediterranean as a macro-region. This is done most effectively in the opening chapter by Russell King, Berardo Cori, and Adalberto Vallega, who highlight the tension between the conceptualization of the Mediterranean as a unified region and as one of fragmentation and diversity. They come down on the side of unity but caution that this conceptualization must also encompass flexibility to account for the many subdivisions and particularities of the region. Ewan Anderson’s chapter on Mediterranean geopolitics describes the many ways in which the geographical setting influences political events and political decisions. Michael Dunford and Russell King present a wide range of data to demonstrate the multifaceted nature of uneven development among the countries bordering the Medi- terranean Sea. They focus on variables such as wealth, human development, employment trends, agricultural change, industry, oil, and indebtedness. Other variables that have a strong spatial development component, such as foreign trade, international migration, and tourism, are dealt with in other chapters. Three other chapters, in particular, appear noteworthy. W. Jan van den Bremen questions the validity of the perceived regional unity of the Mediterranean Basin, analyzing the wide gap between “perceived reality” and the “factual reality” of regional unity. Focusing on commodity trade as a proxy for the unity hypothesis, he concludes that while patterns of intraregional trade have intensified over the past three decades, indicating growing unity, diversity still exists in the form of separate clusters of countries with their own regional unity. Lila Leontidou examines patterns of urbanization in southern Europe, arguing that urban landscapes and urban life in this region are culturally rather than economically conditioned in the sense that they are not centered on industrialization or economically motivated urban growth. Finally, John Thornes assesses three aspects of regional environmental concerns in the Mediterranean: shortage of water, land degradation, and pollution. He identifies water as the most critical resource for the region’s future development, using data from Spain as his primary example. Effectively

Click to return to Table of Contents 614 PDR 27(3) B OOK REVIEWS dispelling the concept of the Mediterranean as a “Garden of Eden,” he stresses the tough environment of the region and the constant battle between humans and natural forces. On balance, this is a well-conceived volume that should interest readers from a variety of disciplinary backgrounds. Its major drawback, as noted above, is its Euro-centered orientation and consequent neglect of many of the compelling issues facing the North African and eastern Mediterranean states. Bibliography, index—M.P.T.

MICHAEL T. KLARE Resource Wars: The New Landscape of Global Conflict New York: Metropolitan Books, 2001. 289 p. $26.00. Competition for natural resources, it is sometimes argued, is diminishing in importance as a source of international conflict. Economic performance increasingly relies on technology and human capital rather than resources. Many recent conflicts have had more to do with national and ethnic rivalries (if not as yet with “clashes of civilizations”). Michael Klare is unpersuaded: resources remain central, both for the United States and for the world as a whole. For the US, the Cold War in retrospect was a temporary interruption, for a time defining strategic interests in political and ideological rather than economic terms. With its ending, “resource issues reassumed their central role in U.S. military planning.” For the world in general, he writes, “a new geography of conflict” is emerging, “a global landscape in which competition over vital resources is becoming the governing principle behind the disposition and use of military power.” Not all resources matter: most of those once thought to be limits to growth have proved ample in supply or the need for them has been skirted by shifts in technology. Geopolitical concern is mainly with oil (and natural gas) and, in significant regions, water. The three oil and gas regions Klare discusses are the Persian Gulf, the Caspian basin, and (still mainly potential) the South China Sea. The water conflicts treated are those of the Jordan, the Nile, the Tigris and Euphrates, and the Indus. Population growth is appreciably implicated in water conflicts, but not much in oil. The new geography thesis is extended to encompass timber, diamonds, and certain mineral ores, which have been recent sources of conflict within states, particularly in Af- rica. For the most part these are lootable commodities rather than “vital resources” affecting strategic balances. However, Klare sees a progression through militarization of claims and increasing involvement of outside powers. Appendix listing oil- and gas-related territorial disputes, index.—G.McN.

KOREA INSTITUTE FOR HEALTH AND SOCIAL AFFAIRS AND UNITED NATIONS POPULATION FUND Low Fertility and Policy Responses to Issues of Ageing and Welfare Seoul: Korea Institute for Health and Social Affairs, 2000. xii + 400 p. (pbk). The papers included in this volume address issues and policy responses in countries experiencing below-replacement fertility, with a focus on East and Southeast Asia. Yap Mui Teng, writing about Singapore, rehashes the familiar story that not

Click to return to Table of Contents B OOK REVIEWS PDR 27(3) 615 long ago that city-state, concerned about population growth, was promoting antinatalist policies and advertising the benefits of small families. Fertility reduction has occurred with such rapidity, though, that Singapore’s policies have turned pronatalist. Current incentives include tax rebates for early births. But a reading of this volume suggests that the reversal in government fertility policy in Sing- apore is an anomaly. Other countries, as is noted several times, are recognizing and reacting to the consequences of low fertility more slowly. The publication contains selected presentations from the International Sym- posium on Population and Development Policies in Low-Fertility Countries, held in Seoul in 1998. The title is slightly misleading since the topic of aging is prominent in only a few of the essays. In his Introduction, Warwick Neville provides a review of social and economic issues facing countries with low fertility. These include declining family size, changing family dynamics, higher dependency ratios, increasing costs of care associated with rapid population aging, and changing employment and retirement structures. Thirteen papers follow, organized into policy and case studies and addressing the challenges laid out by Neville. Although most of the writings concern East and Southeast Asia, the UK and Australia make an appearance, and Wolfgang Lutz presents an overview on low fertility and population policy in Europe. Conspicuously absent from Lutz’s assessment is any discussion of eastern Europe, despite the fact that the region has been experiencing the effects of low fertility for some time. Besides Singapore, chapters on Thailand and China are informative. In an essay of particular interest, John Caldwell, Pat Cald- well, and Peter McDonald place policy responses to lower fertility in a global and historical perspective. They remind us that low fertility is not unique to the current period; what is unusual is that there has been very little response from either governments or the public—thus again the anomalous position of Singapore.—Z.Z.

NATIONAL ASSESSMENT SYNTHESIS TEAM Climate Change Impacts on the United States: The Potential Consequences of Climate Variability and Change. Report for the US Global Change Research Program Cambridge: Cambridge University Press. Overview: 154 p., 2000, $16.95 (pbk.); Foundation: 612 p., 2001, $39.95 (pbk.). These companion volumes—a concise “overview” and a “foundation” report with more comprehensive discussion—summarize the results from a decade-long effort to understand and document the implications of future climate change for the United States. This review was undertaken by the National Assessment Synthesis Team, which collaborated with regional experts from universities, government agencies, industry, and nongovernmental organizations, under the sponsorship of the United States Global Change Research Program. Hundreds of researchers were involved and all chapters were written by multiple authors. The report is broadly similar in nature to the large international effort by the International Panel on Climate Change (IPCC) to examine global changes in climate, their impact, and potential responses. Despite considerable uncertainty about what lies ahead, several key findings emerge from this review of impacts on different geographic regions and sectors. In

Click to return to Table of Contents 616 PDR 27(3) B OOK REVIEWS the absence of major interventions, the average rise in temperature by 2100 in the United States is expected to be anywhere from 5° to 9°F. This rise will likely cause more precipitation overall—and more often in the form of extreme downpours— and greater frequency of both very dry and very wet conditions in different parts of the United States. This in turn will affect natural ecosystems and water resources, some of which will undergo major changes. A rise in the sea level will cause a loss of coastal wetlands and put coastal communities at greater risk of storm surges, especially in the Southeast. Reduction in snowpack will likely alter the timing and amount of water supplies, potentially exacerbating shortages of and conflicts over water, mostly in the Western United States. The ski industry will suffer and the need for air conditioning will rise. Not all expected effects are negative. Agriculture and forestry in much of the country will probably see productivity increases, owing to the fertilizing effect of higher carbon dioxide concentrations and an extended growing season. Other likely benefits, particularly in the Northern United States, include a longer construction season, reduced heating requirements, and reduced cold-weather mortality. For the nation as a whole the direct economic impacts of climate change are expected to be modest. These findings are generally consistent with previous national and global assessments. The major contribution of this report is its regional focus, an often ne- glected approach. One reason for this neglect is the massive uncertainty inherent in local projections: regional forecasts are more uncertain than national forecasts, which in turn are more uncertain than global ones. This problem is illustrated in these reports by the sometimes contradictory regional forecasts produced by different computer models. The “foundation” report also contains very brief discussions of adaptation responses, for example the building of sea walls. Virtually nothing is said about efforts that could slow climate change by reducing the growth of emission of manmade greenhouse gases. The highly respected assessments by the IPCC have led to an international consensus on the desirability of strong efforts to halt the ongoing rise in emissions. The Kyoto Protocol negotiated recently by 178 governments represents an important step toward preventing future climate change. Op- position from the US government has put the implementation of the protocol in doubt. These reports, aimed at a wide audience of students, scientists, and policymakers, should strengthen the conviction in the United States that action is desirable. —J.B.

JACQUES VALLIN AND THÉRÈSE LOCOH (EDS.) Population et développement en Tunisie: La métamorphose Tunis: Cérès éditions, 2001. 801 p. This handsome and bulky volume is the result of a collaboration. It was sponsored by the government of Tunisia and benefited from financial support from UNFPA, but the two editors are well-known French demographers working at INED, the French National Institute for Demographic Studies. The 38 authors are from gov-

Click to return to Table of Contents B OOK REVIEWS PDR 27(3) 617 ernment and international agencies, research institutes, and universities, and, according to the editors’ count, represent at least ten disciplines. It is difficult to maintain a unity of purpose and tone in such a large enterprise. Much of the volume presents official statistics and official government positions, with a tone of self-congratulation spiked with criticism of failed policies of the past. The editors’ intention was to highlight the process of demographic transition in Tunisia, representing the adoption of a family planning program in 1966 as a courageous choice on the part of the government that led to a spectacular decline of total fertility from 7 to 2.2 children per woman by 1998. Implicit is the idea that other countries in the region have made different choices, with other ideological premises and other social priorities, and that the results vindicate the Tunisian model. The editors claim that the decline of fertility paid off in terms of demographic, but also of economic and social changes. This is a plausible hypothesis, but the evidence presented here is not adequate to test it. No comparison is attempted with Tunisia’s neighbors or with other developing areas of the world. The various authors pay lip service to the benefits brought by fertility changes (the effect, for example, of reduced cohort size on the education and health systems, or on employment), but there is no attempt to measure these effects. De- spite valuable theoretical introductory and concluding chapters by the editors, the volume’s main interest is as a source book on Tunisia. A small but fascinating country, Tunisia has initiated progressive policies in areas other than family planning, for example to promote gender equality. In addition to various chapters on the country’s demographic and socioeconomic characteristics, the book contains several “case studies”: on noteworthy development efforts in small areas of the country, on urban growth, on the informal sector’s contribution to employment, and so on. A “case study” is even devoted to “La population tunisienne à l’épreuve du tourisme,” a nicely ambiguous title that suggests that Tunisians may be tourist- proof, but that the seasonal invasion of their beaches by hordes from Europe represents for them “une épreuve,” a challenge or even an ordeal.—E.W.

Click to return to Table of Contents Click here to print article Click to return to Table of Contents DOCUMENTS

On the Socioeconomic Impact of the HIV/AIDS Epidemic

The UN General Assembly Special Session on HIV/AIDS met 25–27 June 2001 and adopted a Declaration of Commitment on HIV/AIDS. The Declaration, in 103 paragraphs, sets out a comprehensive response strategy for governments and UN agencies, supports establishment of a global HIV/AIDS and health fund, and calls for an annual progress report to be reviewed by the Assembly. As part of the Special Session, four “round tables” were conducted on substantive topics: prevention and care, human rights, socioeconomic impact, and international funding. Round Table 3, Socioeconomic impact of the epidemic and the strengthening of national capacities to combat HIV/AIDS, was led by the United Nations Development Programme. The background document prepared for it is reproduced in full below . It argues that the brunt of the epidemic’s impact on human development has been borne by households, communities, and civil society organizations. The emphasis of national and international action has been on prevention and care rather than on counteracting that impact. “Extraordinary efforts” are now required to intensify poverty-reduction measures, to assist caregivers and orphaned children, to prevent the collapse of public services, and to promote workplace tolerance and flexibility. “While HIV/AIDS must be seen as an emergency of the highest order, steady progress in reducing poverty is still the long-term and sustainable solution to the health crisis in the developing world. In the long run, prevention and care will only succeed if people and nations can lift themselves out of poverty.” (The Declaration was not much influenced by such arguments. It devotes two paragraphs to socioeconomic impact, both setting diffuse goals: “By 2003, evaluate the economic and social impact of the HIV/AIDS epidemic and develop multisectoral strategies [on poverty alleviation, etc.]” and “By 2003, develop a national legal and policy framework that protects in the workplace the rights and dignity of persons living with and affected by HIV/AIDS.…”) The Millennium Summit referred to in the document was the meeting on the role of the UN in the twenty-first century held in September 2000 as part of the 55th session of the General Assembly. The Declaration of Commitment on HIV/AIDS and the Round Table 3 document can both be found at http://www.unaids.org/ungass/index.html.

I. Background: The multifaceted prove to be the biggest single obstacle for reach- impact of HIV/AIDS ing the Millennium Summit development goals. Given that AIDS kills mostly people in 1. HIV/AIDS is having a disastrous impact on the 15–49 year age group, it is depriving fami- the social and economic development of coun- lies, communities and entire nations of the tries highly affected by the epidemic and will young and most productive people. It is there-

POPULATION AND DEVELOPMENT REVIEW 27(3) (SEPTEMBER 2001) 619

Click here to print article Click to return to Table of Contents 620 D OCUMENTS fore uniquely devastating in terms of increas- as many households lose their breadwinner ing poverty, reversing human development to AIDS, livelihoods are greatly compromised achievements, eroding the ability of Govern- and savings are eroded by the cost of health ments to provide and maintain essential serv- care and funerals. One study has shown that ices, reducing labour supply and productivity households that have lost one breadwinner and putting a brake on economic growth. to HIV/AIDS see their incomes drop by 80 per cent. In one country, the proportion of Survival people living under that poverty line has already increased by at least 5 per cent as a re- 2. AIDS has already taken a devastating toll in sult of HIV/AIDS. Without addressing this terms of increased mortality and morbidity. In impact, the Millennium Summit develop- the 35 highly affected countries of Africa, life ment goal of halving the proportion of people expectancy at birth is estimated at 48.3 years living in extreme poverty by 2015 cannot be in 1995–2000, 6.5 years less than it would have reached. been in the absence of AIDS. By 2005–2010, average life expectancy at birth in the 11 worst affected countries is projected to decrease to 44 Labour force years, instead of rising to 61 years as projected 6. In sub-Saharan Africa, the size of the in the absence of the disease. In addition, un- labour force will be 10 to 30 per cent smaller der-five child mortality rates in some of the by 2020 than it would have been without worst affected countries are now on the rise as HIV/AIDS. Erosion of human capital, loss of a result of HIV/AIDS, eroding progress towards skilled and experienced workers and reduc- the goal of reducing child mortality rates by two tion in productivity will result in a mismatch thirds by 2015, as agreed at the Millennium between human resources and labour re- Summit. quirements, with grave consequences for the private sector and public sector employers. Education The problem of child labour is exacerbated by HIV/AIDS as children who have lost their 3. As teachers die and orphans drop out of parents have to rely on themselves for basic school, gains in literacy and enrolment ratios survival. are being quickly eroded. In some of the worst affected countries, nearly one half of children who lose their parents to HIV/AIDS drop out Food security of school. Given that the number of AIDS or- 7. The epidemic is intensifying existing labour phans is projected to reach 40 million by 2010, bottlenecks in agriculture, increasing malnu- this is evidence that progress towards the Mil- trition and adding to the burden on rural lennium Summit goal of ensuring universal pri- women, especially those who head farm house- mary education by 2015 is now under threat. holds. Reduced food production is already being reported in some areas, and the Millennium Economic growth Summit development goal of halving the pro- 4. In the worst-affected countries, the epidemic portion of people suffering from hunger by is putting a brake on economic growth by at 2015 is under threat as a result of HIV/AIDS in least 1 to 2 percentage points a year, greatly some countries. jeopardizing efforts to reduce poverty through equitable growth. Many countries will see their Governance gross national product (GNP) shrink by a fifth to a quarter by 2020, some even more than 8. HIV/AIDS has a disastrous impact on the that, and private sector growth and enterprise capacity of Governments to deliver basic so- development are severely affected. cial services. Human resources are lost, public revenues reduced and budgets diverted towards coping with the impact. Similarly, the Income poverty organizational survival of civil society insti- 5. In both rural and urban areas, HIV/AIDS tutions is under threat, with a correspond- pushes people into deeper income poverty, ing impact on democracy.

Click to return to Table of Contents D OCUMENTS 621

Women pact on human development and poverty reduction efforts. 9. HIV/AIDS has a particularly severe impact 13. However, despite intensifying efforts on women in their productive as well as re- focused on prevention and care, the epidemic productive roles. Women tend to be more continues to spread unabatedly, and as vulnerable to HIV infection for both biologi- people infected by HIV become ill and die the cal and social reasons, and infection rates brunt of the impact is now being felt in the among young women are up to four times worst affected countries. Assuming that life- higher than among young men in many prolonging treatment will not be universally countries. Women are also the principal care available in poor countries overnight, death providers both for those sick with AIDS and rates from AIDS will continue to soar. Re- for the children orphaned by AIDS. cent estimates from the United Nations Popu- lation Division show that the population of Social cohesion the 45 most affected countries will be 97 million smaller in 2015 than it would have been 10. HIV/AIDS poses a threat to the very fab- in the absence of HIV/AIDS. Most of this loss ric of society, and is increasingly recognized is due to sharp increases in mortality among as a risk factor for social and political insta- young adults. bility. AIDS is decimating entire generations 14. In the absence of national and glo- of productive young adults, while leaving bal action to mitigate the developmental im- behind a huge cohort of children without pact of HIV/AIDS, households, communities parents and adequate community support, and civil society organizations continue bear- vulnerable to exploitation and lacking edu- ing the brunt of this tragic disaster. They are cation and livelihood opportunities. at the front lines of efforts to mitigate the im- 11. To address these impacts on socio- pact of HIV/AIDS, responding directly to the economic development, the draft declaration needs of people and often working with little of commitment of the special session includes outside or government support. Communi- time-bound targets for developing and imple- ties are mobilizing themselves and showing menting strategies to ensure the maintenance great resilience and solidarity despite their of essential services and the intensification vulnerability to external shocks, such as the of poverty reduction efforts, including pro- premature death of their most productive grammes specifically targeted at households members. and communities hardest hit by the epidemic 15. The global response to HIV/AIDS has (see relevant section in the draft declaration tended to ignore the bigger picture of the im- of commitment). plications for development and poverty reduction. Much research has been undertaken II. Current response to understand the impact of the epidemic, but less has been done to operationalize these 12. The global response to HIV/AIDS has fo- findings and implement measures to coun- cused, rightly so, on the challenge of contain- teract the impact. Discussions on the impli- ing the epidemic and preventing new infec- cations of HIV/AIDS among development tions through advocacy, information and experts and policy makers have been ex- education campaigns, behaviour change tremely limited, and global development tar- communication, condom distribution, pro- gets and goals have been agreed on without grammes targeting groups that are particu- taking into account the added challenges re- larly vulnerable to infection and other key sulting from sharp increases in AIDS-related interventions. The other part of the response adult mortality rates in most of Africa and in has focused on care and support for people some parts of other regions as well. living with HIV and AIDS, efforts that are 16. To say that nothing has been done expected to intensify as new treatments be- to mitigate the impact of HIV/AIDS, as de- come more accessible and affordable. Both fined by the present paper, would be incor- prevention and treatment are top priorities, rect. Overall poverty reduction efforts have not only in saving lives and reducing human intensified in the last decade, partly thanks suffering but also in limiting the future im- to global commitments made at the World

Click to return to Table of Contents 622 D OCUMENTS

Summit for Social Development in 1995. ther fuelling the spread of HIV. Research and Over the past year, HIV/AIDS has gradually experience over the last 20 years have estab- been incorporated into poverty reduction lished a clear link between these conditions strategies, especially in Africa. A recent re- and increased susceptibility to infection and view of 20 national strategies showed that lack of access to treatment. While HIV/AIDS AIDS was mentioned as a factor in deepen- must be seen as an emergency of the highest ing poverty, and a few of them outlined ma- order, steady progress in reducing poverty is jor actions to fight AIDS as part of poverty still the long-term and sustainable solution reduction interventions. But much more to the health crisis in the developing world. needs to be done. In the long run, prevention and care will only 17. In addition, some countries have put succeed if people and nations can lift them- in place more targeted programmes to sup- selves out of poverty. port people, households and communities 20. Countries ravaged by the HIV/AIDS ravaged by the epidemic. Partly thanks to in- epidemic are facing a double jeopardy. On the tensifying global support, some progress has one hand, their capacity for planning and been made in promoting efforts to support implementing development strategies is greatly children orphaned by the epidemic. But compromised by the loss of human capital and given that the world can expect to see over diversion of scarce resources. On the other 40 million children orphaned by HIV/AIDS hand, strong national capacity is becoming by 2010, these efforts are far from adequate. even more crucial as countries face the formi- dable challenge of responding to the epidemic. Such capacity is essential not only in the health III. Implementing the sector, in coping with the added disease bur- declaration of commitment: den and delivering new treatments, but in all The way forward sectors of government, the private sector and civil society, which must be mobilized around 18. Given the reality of current and future broad-based prevention and social mobilization impacts on human development, extraordi- efforts to reverse the epidemic. nary efforts are now required to intensify 21. The present paper sets out below four poverty reduction efforts. This includes en- concrete priorities for the implementation of suring that basic social services are main- those sections of the draft declaration of com- tained despite loss of human resources, as mitment related to the socio-economic im- well as policies to generate equitable eco- pact of the epidemic, and for consideration nomic growth, notwithstanding the loss of during the discussion of round table 3. productivity and deficit-creating pressures on public health budgets. Such efforts must be 1. Intensification of poverty undertaken in the context of overall national reduction efforts development plans and poverty reduction strategies. 22. In order to reach the Millennium Sum- 19. There are two main reasons why mit development goals and nationally deter- tackling the impact of HIV/AIDS on devel- mined human development targets in coun- opment is such an essential part of the glo- tries affected by the epidemic, current bal response to the epidemic. First of all, poverty reduction strategies need to be re- without policies, strategies and adequate re- evaluated and adjusted in order to address sources to compensate for the poverty-cre- the unique challenge posed by the impact of ating effect of high mortality among produc- HIV/AIDS. The impact on rural communities tive age groups, the Millennium Summit needs special attention, as they are often un- development goals cannot be reached in der-served with regard to social services and much of Africa, or in other parts of the world infrastructure, and are absorbing urban where the epidemic is likely to continue to dwellers returning to their villages when they spread, as explained above. Second, low hu- fall ill. Efforts to promote equitable growth, man development, widespread poverty and generate employment, raise incomes, im- inadequate access to education and health, prove agricultural production and promote greatly exacerbated by the epidemic, are fur- informal sector livelihoods must be scaled up

Click to return to Table of Contents D OCUMENTS 623 in order to compensate for the poverty-cre- sick and dying. In addition, government rev- ating effect of high mortality rates among the enues are expected to fall by as much as one most productive age groups. The optimal al- fifth in the worst affected countries due to location of scarce domestic resources becomes the impact on overall economic activity and an even more crucial challenge, leaving even shrinking GNP. All these effects must be ad- less room for budget items not directly con- dressed in national budgets, medium-term tributing to poverty reduction and the im- expenditure plans and sector development provement of access to basic social services. plans. Pro-poor budgeting becomes even Poverty reduction strategy papers (PRSPs) more crucial in AIDS-affected countries, and need to be formulated in such a way that they special measures must be introduced to safe- take into account the current and expected guard against the collapse of public sector impact of the epidemic. For example, Burkina functions. Such actions include fast-track Faso and Kenya have made progress in ad- training and recruitment of new teachers, justing their poverty reduction strategies to nurses and other key civil servants, realloca- the impact of HIV/AIDS and have started to tion of budgets towards the most essential allocate debt relief savings towards HIV pre- services and efforts to prolong the working vention and care. life of people living with HIV through care, support and team work, much along the lines 2. Special programmes targeted at of changes in human resources management children, women, the elderly and needed in the private sector. Malawi is one other groups country taking proactive steps to assess the impact of HIV/AIDS in the public sector and 23. In addition to the scaling up of national implement workplace policies to ensure the poverty reduction strategies, special social continued functioning of essential public serv- protection programmes are required to sup- ices, notwithstanding the impact on human port the people, households and communi- resources. ties that are hardest hit by the epidemic. Given the heavy burden that the epidemic is 4. Addressing the impact on putting on women as caretakers and bread- labour markets winners, social security arrangements are needed to respond to their needs. In addi- 25. Labour market and workplace policies tion, an extraordinary effort is needed to pro- need to be adjusted to address the impact on vide for the needs of children orphaned by the availability of skilled workers, productiv- the epidemic, including special efforts to en- ity and human resource development. Efforts sure access to primary education, food, health are needed to support and protect the rights care and other social support. A central part of workers living with HIV and AIDS, maxi- of this effort must be to support existing com- mizing their productivity through access to munity-based solidarity mechanisms for or- care and support and changes in work rou- phan care. tines. For example, Volkswagen Brazil has implemented a successful prevention and treatment programme, preventing many new 3. Preventing the collapse of infections among its workers and reducing essential public services and HIV/AIDS-related absenteeism by 90 per institutions of democratic governance cent. Social dialogue is essential between government, workers and employers to develop 24. Special efforts are needed to ensure the a legal and policy framework to address the maintenance of essential public services, such impact of the epidemic on the workforce. A as education, health, security, justice and in- code of practice on HIV/AIDS and the work- stitutions of democratic governance. The place should be adopted at national and en- public sector in the worst affected countries terprise levels. is crumbling under the weight of the epi- 26. Countries affected by the epidemic demic, as irreplaceable human resources are cannot successfully address these challenges decimated and public budgets diverted to- without adequate international solidarity, wards the immediate needs of caring for the cooperation and financial support. Although

Click to return to Table of Contents 624 D OCUMENTS this issue is covered by round table 4 on fi- rate with the magnitude of the challenge. Si- nancing the response to HIV/AIDS, it is es- multaneously, the possibility of full debt can- sential to stress the need for much higher lev- cellation for the hardest hit countries may els of official development assistance (ODA) need to be explored, justified by the devas- to the worst affected countries, in support of tation caused by the epidemic and provided overall poverty reduction strategies and im- that a substantial portion of debt cancellation provement of social services. Since 1990, savings are allocated to HIV/AIDS prevention ODA flows to the 28 countries with highest and care. Without such support, the inter- adult HIV prevalence rates (more than 4 per national community will fail in its obligation cent) have fallen by nearly one third, from to help countries reach the Millennium Sum- 12.5 to 8.6 billion United States dollars.1 This mit development goals in much of Africa and, trend must be reversed and resource flows depending on the course of the epidemic, in substantially increased to levels commensu- many other parts of the world as well.

1Excluding South Africa.

Click here to print article Click to return to Table of Contents

Click to return to Table of Contents ABSTRACTS

Life Span Extension in Humans Is proposes an alternative explanatory model Self-Reinforcing: A General Theory and tests it empirically with reference to of Longevity Kenya. Access to political power and, through power, access to a state’s resources—including resources devoted to clin- JAMES R. CAREY ics, schools, labor opportunities, and other DEBRA S. JUDGE determinants of demographic behavior—are This article proposes that longevity is not advanced as the key factors underlying eth- merely the result of an absence of mortality nic differences. District-level estimates of but a self-reinforcing and positively selected “political capital” are introduced and merged life-history trait in social species. It argues with two waves of Demographic and Health that a small increase in longevity is ampli- Survey data. The effects on models of con- fied as (1) reductions in mortality at young traceptive use are explored. Results confirm ages increase natural selection for mecha- that measures of political capital explain re- nisms of maintenance and repair at all older sidual ethnic differences in use, providing ages as well as increasing the potential for strong support for a political approach to the intergenerational transfers; (2) intergenera- analysis of demographic behavior. tional transfers of resources from old to young increase fitness (e.g., through improved health, skill, and competitive ability) Is There a Stabilizing Selection of the young and thus favor the presence of older individuals in a population; and (3) the Around Average Fertility in Modern division of labor increases both efficiency and Human Populations? innovation at all levels, resulting in increased resources that can be reinvested. This theory ULRICH MUELLER is framed around the longevity-oriented Possibly the greatest challenge for an evolu- question posed two decades ago by the ger- tionary explanation of demographic transi- ontologist George Sacher, “Why do we live tion is the fact that fertility levels universally as long as we do?,” rather than the more start to fall first among the well-to-do, well- prevalent question today, “Why do we grow educated, healthy classes, which can be ex- old?” The article describes the foundational plained only by some voluntary or at least principles and the main phases of a model adaptive action. The problem of how re- for the evolution of longevity mediated straints on fertility could have evolved by through social organization, and applies the natural selection has been tackled with group concept specifically to human populations. selection models as well as with stabilizing selection models. The latter model, which is critically discussed in this article, posits that First Politics, Then Culture: some intermediate (rather than maximal) Accounting for Ethnic Differences in level of fertility is optimal for long-term re- Demographic Behavior in Kenya productive success. Tests of stabilizing selection in human populations are rare, their results inconclusive. Here four sets of data are ALEXANDER A. WEINREB analyzed: they are samples drawn from the Ethnic differences in demographic behavior class of 1950 of the US Military Academy at tend to be disguised behind analytically West Point (cohorts 1923–29), retired US opaque labels like “district” or “region,” or noncommissioned officers (cohorts 1913– else subjected to simplistic cultural explana- 37), and western German and eastern Ger- tions. Drawing on new political economy, so- man physicians (cohorts 1930–35), all con- ciological theory and the political science lit- taining fertility data over two generations, erature on sub-Saharan Africa, this article and from European royalty (cohorts 1790–

POPULATION AND DEVELOPMENT REVIEW 27(3) (SEPTEMBER 2001) 625

Click here to print article Click to return to Table of Contents 626 A BSTRACTS

1939) containing fertility data over four gen- sive immigration policies reflecting “client” erations. Deterministic as well as stochastic politics. Australia has since pursued a more fitness measures are used. It is found that restrictive and selective course while the maximal, not average, fertility in the first United States has resisted pressures toward generation leads to maximal long-term re- such a stance. The authors account for these productive success. Also against prediction, differences by assessing the theoretical per- no decreasing marginal fitness gains by in- spectives of interests, rights, and states. Con- creasing fertility can be observed. The find- flicts among groups with direct interests in ings leave little space for considering stabi- policy outcomes are the principal source of lizing selection as a plausible mechanism immigration politics, but a comparison of the explaining the course of demographic tran- roles of rights and state institutions helps ex- sition but indicate instead that biological evo- plain peculiarities of the two cases. The dis- lution today is as fast and vigorous as ever tinctive Australian policy trajectory is shaped in human history. Even in large populations, by greater volatility of public opinion about all people living today may be the descen- immigration and multiculturalism, and by dants of just some few percents—a much political institutions that are more respon- smaller proportion than generally believed— sive to popular sentiment. of the people living some generations ago.

The Age of Migration in China The Retrenchment of Marriage: Results from Marital Status Life ZAI LIANG Tables for the United States, 1995 Using data from the 1987 and 1995 China One Percent Population Sample Surveys, this ROBERT SCHOEN article examines migration patterns during NICOLA STANDISH 1982–95, a period of sweeping social and Marital status life tables were calculated us- economic changes in China. Several major ing 1995 US rates of marriage, divorce, and patterns are evident: the increase in overall mortality. Compared to figures for 1988, the migration and especially in temporary mi- proportion of persons surviving to age 15 gration, the increasing importance of inter- who ever marry remained fairly steady at provincial migration, and the concentration about five-sixths of all men and seven- of migrants in the coastal region. Over time, eighths of all women. The average age at first migrants of rural origin were more likely to marriage rose substantially: to 28.6 years for choose cities as destinations than towns. The men and 26.6 years for women. The prob- consequences and implications of the ability of a marriage ending in divorce changes in migration patterns are explored. changed little and was .437 for men and .425 for women. It is likely that no US period or cohort will ever have half of all marriages Divergent Paths of Immigration end in legal divorce, though the highest co- Politics in the United States and hort may reach 47 percent. Patterns of mar- Australia riage and divorce observed since 1970 show the effect that cohabitation continues to have on the American family, where it is delay- GARY P. FREEMAN BOB BIRRELL ing, but not replacing, marriage. The United States and Australia converged by the mid-1980s on receptive and expan-

Click to return to Table of Contents A BSTRACTS / FRENCH 627

L’extension de la longévité humaine que avec des étiquettes analytiques opaques est auto-renforçatrice : une théorie comme “district”, “région”, ou encore à uti- générale de la longévité liser des explications culturelles simplistes. Puisant dans la documentation sur l’écono- mie politique, la théorie macro-sociale et la JAMES S. CAREY science politique de l’Afrique subsaharienne, DEBRA S. JUDGE le présent article propose un modèle expli- Le présent article suggère que la longévité catif alternatif et le met à l’essai de façon n’est pas seulement l’absence de mortalité empirique par rapport au Kenya. L’accès à mais qu’elle constitue un trait auto- la puissance politique et, par le truchement renforçateur et positif des cycles évolutifs du pouvoir, aux ressources de l’État—y com- dans l’espèce sociale. Il allègue qu’une lé- pris les ressources consacrées à la santé, à gère augmentation de la longévité est am- l’éducation, aux chances en matière d’em- plifiée car, (1) les baisses du taux de morta- ploi et aux autres facteurs déterminants du lité en bas âge avantagent la sélection comportement démographique—est pré- naturelle des mécanismes d’entretien et de senté comme facteur clé sous-jacent des dif- réparations à tous les âges plus avancés, en férences ethniques. Les prévisions plus d’accroître les possibilités des transferts d’“avantage politique” à l’échelon du district intergénérationnels; (2) les transferts sont présentées et amalgamées à deux pha- intergénérationnels des ressources des vieux ses des données d’Enquête démographique aux jeunes augmentent le fitness (p. ex. et sanitaire. Les effets sur les modèles d’uti- l’amélioration de la santé, des habiletés et lisation de contraceptifs sont étudiés. Les ré- de la capacité compétitive) chez les jeunes, sultats confirment que les mesures d’avan- favorisant ainsi la présence d’individus plus tage politique expliquent les différences âgés dans une population; et (3) la division ethniques résiduelles en usage et supportent du travail augmente l’efficacité et l’innova- fortement une approche politique à l’analyse tion à tous les niveaux, si bien que les res- du comportement démographique. sources accrues peuvent être réinvesties. Cette théorie part de la question sur la lon- gévité posée il y a deux décennies par le gé- rontologue George Sacher : « Pourquoi vi- vons-nous aussi longtemps ? », plutôt que de Y a-t-il une sélection stabilisatrice la question courante « Pourquoi vieillissons- dans le taux de fécondité moyen nous ? » Le présent article décrit les princi- chez les populations humaines pes généraux et les principales phases d’un modernes ? modèle de l’évolution de la longévité modi- fié par le biais de l’organisation sociale et ap- ULRICH MUELLER plique le concept expressément aux populations humaines. Il se peut que le plus grand défi de toute ex- plication évolutionniste de la transition dé- mographique soit qu’à l’échelle mondiale, la baisse des taux de fécondité se fasse d’abord sentir chez les gens aisés, érudits et en santé, D’abord la politique, puis la ce qui ne s’explique que par une action vo- culture : les différences ethniques lontaire ou, du moins, adaptative. La ques- prises en ligne de compte dans le tion « Dans quelle mesure le contrôle de la comportement démographique au fécondité est-il fonction de la sélection na- Kenya turelle ? » a été étudiée au moyen de modè- les de sélection de groupe ainsi que de mo- dèles de sélection stabilisatrice, modèle ALEXANDER A. WEINREB discuté de façon éclairée dans le présent ar- On est enclin à masquer les différences eth- ticle. Ce modèle suggère qu’un certain ni- niques dans le comportement démographi- veau de fécondité intermédiaire (plutôt que

Click to return to Table of Contents 628 A BSTRACTS / FRENCH maximal) est optimal pour le succès de re- gration temporaire, l’importance croissante production à long terme. Les essais sur la sé- de la migration interprovinciale et la concen- lection stabilisatrice dans les populations tration des migrants dans les régions côtiè- humaines sont rares et leurs résultats ne sont res. Au fil du temps, les migrants d’origine pas concluants. Le présent article analyse rurale étaient plus susceptibles de choisir quatre séries de données : les données com- éventuellement les grandes villes que les vil- pilées sur les individus ayant gradué en 1950 les de taille moyenne ou petite. Les consé- du US Military Academy de West Point (co- quences des changements de ces courants hortes 1923–1929), des sous-officiers amé- migratoires sont examinées. ricains à la retraite (cohortes 1913–1937), des médecins d’Allemagne de l’Ouest et d’Al- lemagne de l’Est (cohortes 1930–1935)—ces trois séries contenant des données sur la fé- condité sur une période de plus de deux gé- Courants de pensée politiques nérations—et enfin, les membres de la divergents sur l’immigration aux royauté européenne (cohortes 1790–1939)— États-Unis et en Australie contenant des données sur la fécondité sur une période de plus de quatre générations. GARY P. FREEMAN Les techniques déterministes et stochastiques BOB BIRRELL ont été utilisées et il en ressort que la fécon- dité maximale—et non intermédiaire—dans Au milieu des années 1980, des politiques la première génération mène au succès de d’immigration réceptives et expansives, reproduction maximal à long terme. En axées sur le « client », étaient en vigueur aux outre, contre toute prévision, on n’y observe États-Unis et en Australie. Depuis ce temps, aucun gain marginal décroissant de fitness en l’Australie a suivi une ligne de conduite plus accroissant le taux de fécondité. Les résul- restreignante et sélective, alors que les États- tats n’accordent pas beaucoup de place à Unis ont résisté aux pressions dans cette di- l’étude de la sélection stabilisatrice comme rection. Les auteurs expliquent ces différen- mécanisme plausible pouvant expliquer le ces en évaluant les perspectives théoriques cours de la transition démographique; ils in- des intérêts, des droits et des états. Les con- diquent plutôt que l’évolution biologique est flits parmi les groupes ayant des intérêts di- aujourd’hui aussi rapide et vigoureuse que rects dans les conséquences des politiques depuis le début de l’histoire de l’homme. Qui sont la source principale des politiques sur plus est, même les membres des grandes po- l’immigration. Cependant, en comparant le pulations actuelles pourraient bien n’être les rôle des institutions des droits et des états, descendants que de quelques pourcentages les particularités des deux pays sont mises d’individus ayant vécu il y a de ça quelques en évidence. En Australie, le courant de pen- générations—une proportion beaucoup plus sée politique est façonné par une opinion faible qu’on ne l’avait cru jusqu’ici. publique plus volatile sur l’immigration et le multiculturalisme ainsi que par les institutions politiques plus sensibles à l’opinion po- pulaire.

L’ère de la migration en Chine

ZAI LIANG Le repli du mariage : résultats des À l’aide de données tirées des enquêtes de tables de survie selon l’état populations échantillon « One Percent » de matrimonial pour les États-Unis, 1987 et de 1995 en Chine, le présent article 1995 étudie les courants migratoires de 1982 à 1995, période ayant connu des changements ROBERT SCHOEN économiques et sociaux radicaux en Chine. NICOLA STANDISH De nombreux modèles majeurs en ressor- tent : l’augmentation de la migration dans Les tables de survie selon l’état matrimonial son ensemble et particulièrement de la mi- ont été calculées à partir des taux de mariage,

Click to return to Table of Contents A BSTRACTS / SPANISH 629 divorce et mortalité aux États-Unis en 1995. blance, aux États-Unis, aucune cohorte ou Comparée aux chiffres de 1988, la propor- période ne verra la moitié des mariages dé- tion des individus survivant jusqu’à l’âge de boucher sur un divorce légal, quoique le taux 15 ans, qui se sont mariés, est demeurée as- de divorce de la cohorte enregistrant le taux sez stable—environ 5/6 de tous les hommes de plus élevé puisse bien atteindre 47 pour et 7/8 de toutes les femmes. L’âge moyen lors cent. Les tendances de mariage et de divorce du premier mariage s’est considérablement observées depuis 1970 démontrent l’effet élevé, à 28,6 ans chez les hommes et 26,6 que les unions de fait ont sur la famille amé- chez les femmes. La probabilité d’un divorce ricaine : le mariage, s’il est retardé, n’est pas a peu changé : 0,437 chez les hommes et remplacé par l’union de fait. 0,425 chez les femmes. Selon toute vraisem-

Extensión de vida en seres humanos de la longevidad mediada a través de la ores autoreforzada: Una teoría general ganización social, y aplica el concepto de longevidad específicamente a las poblaciones humanas.

JAMES R. CAREY DEBRA S. JUDGE En este artículo se propone que en las espe- Primero políticas, luego cultura: cies sociales la longevidad no es simplemen- Explicando diferencias étnicas en el te el resultado de la ausencia de mortalidad comportamiento demográfico de sino que una característica autoreforzante y Kenya positivamente seleccionada de la historia de vida. El artículo sostiene que un pequeño ALEXANDER A. WEINREB aumento de longevidad es amplificado cuan- do (1) reducciones de mortalidad a edades Las diferencias étnicas en el comportamien- tempranas incrementan la selección natural to demográfico tienden a ocultarse detrás de para los mecanismos de mantenimiento y re- rótulos analíticos opacos como ser “distrito” paración a todas las edades más avanzadas, o “región”, o sino están sujetas a explicacio- como también aumentando el potencial de nes culturales simplísticas. Basándose en la transferencias intergeneracionales; (2) las economía política, la teoría macrosocial, y en transferencias intergeneracionales de recur- la literatura de ciencias políticas sobre Afri- sos—de ancianos a jóvenes—aumentan el ca al Sur del Sahara, este artículo propone fitness de los jóvenes (por.ej. a través de me- un modelo explicativo alternativo y lo pone joramientos en la salud, destreza y habilidad a prueba empíricamente con referencia a competitiva) y por consiguiente favorecen la Kenya. El acceso al poder político y, a través presencia en la población de individuos más del poder, el acceso a los recursos de un es- ancianos; y (3) la división del trabajo aumen- tado—incluyendo los recursos dedicados a ta tanto la eficacia como la innovación en clínicas, escuelas, oportunidades laborales, y todos los niveles, lo cual resulta en aumen- otros determinantes del comportamiento tos de recursos que pueden reinvertirse. Esta demográfico—es propuesto aquí como el fac- teoría se ha estructurado alrededor de la pre- tor clave subyacente de las diferencias gunta relacionada con la longevidad que el étnicas. Estimaciones distritales de “capital gerontólogo George Sacher formuló hace dos político” son introducidas y fusionadas con décadas: “¿Por qué vivimos tan largamente datos de dos Encuestas Demográficas y de como lo hacemos?” más bien que en la pre- Salud. Los efectos sobre modelos de uso de gunta actual más prevalente “¿Por qué en- anticoncepción son examinados. Los resul- vejecemos?”. El artículo describe los princi- tados confirman que las medidas de capital pios en que se fundamenta y las fases político explican las diferencias étnicas principales de un modelo para la evolución residuales en uso, proveyendo un firme apo-

Click to return to Table of Contents 630 A BSTRACTS / SPANISH yo a un enfoque político del análisis del com- lógica actual es tan rápida y vigorosa como portamiento demográfico. lo ha sido siempre en la historia humana. Aun en poblaciones grandes, todas las personas que viven hoy en día pueden ser los descendientes de tan sólo unos pocos por- ¿Existe una selección estabilizadora centajes—una proporción mucho menor de alrededor de una fecundidad media lo que se cree generalmente—de personas en las poblaciones humanas que vivieron hace varias generaciones. modernas?

ULRICH MUELLER La era de migración en China El reto mayor para cualquier explicación evolucionaria de la transición demográfica ZAI LIANG posiblemente sea el hecho que los niveles de fecundidad comienzan universalmente a des- Usando datos de las Encuestas de Muestreo cender primero entre la clase más pudiente, de China del Uno Por Ciento de la Población de mejor educación y de buena salud, lo que de 1987 y 1995, este artículo examina los puede explicarse solamente por alguna ac- patrones de migración durante 1982–95, un ción voluntaria o por lo menos adaptiva. El período de vastos cambios sociales y econó- problema de cómo podrían haber evolucio- micos en China. Se evidencian varias impor- nado por selección natural las limitaciones tantes corrientes: un aumento en el total de de fecundidad ha sido abordado con mode- la migración y especialmente en la migración los de selección por grupo como también con temporal, la creciente importancia de la mi- modelos estabilizadores. Este último mode- gración entre provincias, y la concentración lo, examinado críticamente en este artículo, de migrantes en la región costera. A través postula que un nivel de fecundidad interme- del tiempo, los migrantes de origen rural dio (más bien que maximal) es óptimo para tienden a escoger como lugar de destino ciu- un éxito reproductivo a largo plazo. Pruebas dades grandes más bien que pequeñas. Se de una selección estabilizadora en poblacio- examinan las consecuencias y efectos de los nes humanas son raras, sus resultados incon- cambios en los patrones de migración. clusos. Aquí se analizan cuatro grupos de datos. Estos son muestras obtenidas de la clase de 1950 de la Academia Militar de los Es- tados Unidos en West Point (cohortes de Divergentes trayectorias de 1923–29), de oficiales jubilados no comisio- inmigración en los Estados Unidos y nados (cohortes 1913–37), de médicos de la en Australia Alemania Occidental y Oriental (cohortes 1930–35), todos conteniendo datos de fecun- GARY P. FREEMAN didad de más de dos generaciones, y de rea- BOB BIRRELL leza europea (cohortes 1790–1939) conteniendo datos de fecundidad de más de cuatro A mediados de la década de 1980 los Esta- generaciones. Se usan medidas del estado de dos Unidos y Australia convergieron en sus salud determinísticas como también políticas receptivas y expansivas de inmigra- estocásticas. Se encontró que una fecundi- ción basándose en políticas de “clientela”. dad maximal, no media, en la primera ge- Desde entonces Australia ha proseguido un neración lleva a un éxito reproductivo curso más restrictivo y selectivo mientras que maximal a largo plazo. También, contrario a los Estados Unidos ha resistido las presiones las predicciones, no se observaron decrecien- hacia tal postura. Los autores toman en tes beneficios en el fitness marginal al aumen- cuenta estas diferencias al examinar las pers- tar la fecundidad. Los resultados dejan poco pectivas teóricas de los intereses, derechos, espacio para considerar una selección y estados. Conflictos entre grupos que tie- estabilizadora como posible mecanismo ex- nen un interés directo en los resultados de plicador del curso de la transición demográ- las políticas son la fuente principal de las po- fica pero indican en vez que la evolución bio- líticas de inmigración, pero una comparación

Click to return to Table of Contents ABSTRACTS / SPANISH 631 del papel que juegan los derechos y las ins- dad de 1995. Comparado con cifras de 1988, tituciones estatales ayuda a explicar las pe- la proporción de personas que sobreviven a culiaridades de los dos casos. La trayectoria la edad de 15 y que se casan alguna vez per- distinctiva de la política australiana ha sido maneció bastante firme alrededor de cinco- determinada por una mayor volatilidad en sextos de todos los hombres y de siete-octa- la opinión pública con respecto a la inmigra- vos de todas las mujeres. La edad media del ción y multiculturalismo, y por instituciones primer matrimonio aumentó considerable- políticas que son más sensitivas a los senti- mente a 28.6 años para los hombres y 26.6 mientos populares. años para la mujer. La probabilidad que un matrimonio termine en divorcio cambió poco y fue .437 para el hombre y .425 para la mu- Retrasando el matrimonio: jer. Es posible que en ningún período o nin- guna cohorte de los Estados Unidos la mitad Resultados de tablas de mortalidad de todos los matrimonios terminen en un del estado matrimonial para los divorcio legal, aunque la cohorte más alta Estados Unidos, 1995 podría alcanzar un 47 porciento. Los patrones de matrimonio y divorcio observados ROBERT SCHOEN desde 1970 muestran el efecto que la coha- NICOLA STANDISH bitación continúa teniendo sobre la familia estadounidense, en la cual está retrasando, Tablas de mortalidad del estado matrimonial pero no reemplazando, el matrimonio. fueron calculadas utilizando tasas estadouni- denses de matrimonios, divorcios y mortali-

Click here to print article Click to return to Table of Contents AUTHORS FOR THIS ISSUE

BOB BIRRELL is Reader in Sociology and Director, Centre for Population and Urban Research, Monash University.

JAMES R. CAREY is Professor, Department of Entomology, University of California, Davis, and Senior Scholar, Center for the Economics and Demography of Aging, University of California at Berkeley.

JOEL E. COHEN is Professor of Populations, Rockefeller University and Columbia University.

GARY P. FREEMAN is Professor of Government and Director, Public Policy Clinic, University of Texas at Austin.

JACK A. GOLDSTONE is Professor, Department of Sociology, University of California at Davis.

DEBRA S. JUDGE is Research Ecologist, Department of Entomology, and lectures in the Department of Anthropology, University of California at Davis.

ZAI LIANG is Associate Professor of Sociology, City University of New York–Queens College.

ULRICH MUELLER is Professor, Institute of Medical Sociology and Social Medicine, Medical School, University of Marburg, Germany.

ROBERT SCHOEN is Hoffman Professor of Family Sociology and Demography, Department of Sociology, Pennsylvania State University, University Park.

NICOLA STANDISH is with the Department of Finance, State of California, Sacramento.

ALEXANDER A. WEINREB is a National Institute of Child Health and Human Development (NICHD) postdoctoral fellow at the Population Research Center, NORC and the University of Chicago.

632 POPULATION AND DEVELOPMENT REVIEW 27(3) (SEPTEMBER 2001)

Click to return to Table of Contents