The Rise and Fall of in the American South

Karen Clay* Ethan Schmick† Werner Troesken‡

This draft: August 2016

Abstract

No other -related in American history caused as many as pellagra. The by-product of insufficient consumption, pellagra reached epidemic proportions in the American South, killing roughly 7,000 Southerners annually at its peak in 1928. We document the rise and fall of pellagra in the American South and present three main findings. First, pellagra resulted, in part, from Southern ’s heavy emphasis on cotton, which displaced local production and effectively raised the price of niacin consumption. Evidence for this proposition derives in part from the arrival of the boll weevil. Although the boll weevil reduced Southern incomes and cotton production, it was also associated with increases in local food production and sharp reductions in pellagra. Second, pellagra was largely eliminated through voluntary fortification of cereal-grain products starting in 1937 and a series of state fortification laws passed in the 1940s. These laws, for the first time in Southern history, broke the strong positive correlation between cotton production and pellagra. Third, exposure to early- interventions that reduced cotton production and/or increased niacin consumption were associated with improved stature and wages. These findings have important implications for our understanding of the economic history of the American South and economic development in general.

* Contact: Karen Clay, Associate Professor of Economics and Public Policy, Carnegie Mellon University, Heinz College, 5000 Forbes Avenue, Pittsburgh, PA 15213. Email: [email protected] † Contact: Ethan Schmick, University of Pittsburgh, Department of Economics, 4901 Wesley W. Posvar Hall, 230 South Bouquet, Pittsburgh, PA 15260. Email: [email protected] ‡ Contact: Werner Troesken, Professor of Economics, University of Pittsburgh, Department of Economics, 4901 Wesley W. Posvar Hall, 230 South Bouquet, Pittsburgh, PA 15260. Email: [email protected]

1 1. Introduction

While virtually unheard of in the today, no other nutrition-related disease in American history has caused as many deaths as pellagra. The by-product of insufficient niacin consumption, pellagra reached epidemic proportions in the American

South during the first half of the twentieth century, killing roughly 7,000 Southerners annually at its peak in 1928 (Bollet 1992).1 Put another way, at its peak in 1933, killed around 4,500 Southerners annually, or only two-thirds the number of people killed by pellagra at its peak.2 If one looks at the incidence of pellagra, the data are even more striking. According to one estimate, the case fatality rate for pellagra was only around 3% (Goldberger 1928). Thus, it is possible that in 1928 there were approximately

230,000 cases of pellagra in the United States, which is equivalent to the population of

Atlanta at the time. In places where pellagra was endemic, it was estimated that around

20 percent of all households had at least one person sick with the disease (Goldberger

1928). While pellagra was not as pervasive as hookworm, which infected around one- third of the Southern population, it is important to remember that pellagra is an extreme indicator: one need not have developed a full-blown case of pellagra to have been poorly nourished, and the high incidence of pellagra is suggestive of widespread in the South (Etheridge 1972, Youmans 1964).

Although there is a small literature on pellagra in the American South, that literature is mostly qualitative and has not been subjected to formal hypothesis testing.

As a result, our understanding of the origins and effects of pellagra is incomplete and predicated on a limited evidentiary base. In this paper we document the rise and fall of pellagra in the by conducting a series of falsification exercises to test the central predictions of the existing historical literature. Additionally, we extend the literature by offering and testing new hypotheses on the long-term effects of early-

1 The Mortality Statistics of the United States report 6,824 deaths from pellagra in the year 1928. 2 The Mortality Statistics of the United States report 4,678 deaths from malaria in the year 1933. This number is likely overstated since in several years it is reported that the term malaria “is used somewhat loosely in some sections of the country” and that this “undoubtedly has much to do with some of the higher rates recorded” (Mortality Statistics of the United States 1924; pg. 25).

2 life exposure to high pellagra environments. Our results do not call for a wholesale revision of the prevailing understanding of pellagra in the American South, but they do bring to the fore economic and nutritional mechanisms that have important implications for our understanding of the American South in particular, and cash crop economies in the developing world more generally.

We focus on three sets of questions. First, what were the origins of pellagra, and why was it so pronounced in the American South? Existing historical accounts suggest the disease stemmed from the Southern production of cotton. According to the standard logic, cotton displaced local food production and drove poor Southern farmers and mill workers to consume milled Midwestern corn, which was relatively cheap but was also devoid of niacin (Park et al. 2000, Goldberger et al. 1920, Rajakumar 2000). Second, pellagra rates plummeted during the mid-twentieth century. What drove this sharp decline? Existing historical accounts attribute the eradication of pellagra to a series of state laws passed during the 1940s mandating breads and grains be fortified with niacin and other (Humphreys 2009, Park et al. 2000). Third, we ask a question that the existing qualitative literature has left unaddressed: what were the effects of pellagra, and nutritionally deficiencies more generally, on the long-term of Southerners?

We begin our analysis by searching for systematic evidence that cotton production displaced local food production, which in turn limited the availability of nutritionally-rich locally-sourced and discouraged adequate consumption of nutrients, such as niacin. Consistent with the prevailing historical understanding, we find that in times and places of high cotton production, pellagra rates were also higher.

One might wonder if these estimates conflate changes in income with changes in food availability: perhaps years of high cotton production were also years of low income, and it was the reduction in income, not the reduction in local food production, that drove the increase in pellagra. Two pieces of evidence, however, suggest this is not the case. First, the positive correlation between cotton production and pellagra survives the inclusion of controls for changes in farm revenue generated per acre of cotton. Second, the arrival of the boll weevil during the early 1900s reduced both income and cotton production,

3 and yet was associated with increased food production and reduced pellagra rates.

Moreover, when cotton production rebounded after the boll weevil, pellagra rates returned to their pre-boll-weevil levels.

After exploring the relationship between cotton production and pellagra, we search for evidence on the forces behind the decline in pellagra rates during mid- twentieth century. The raw data reveal two sharp drops in the pellagra rate. The first break occurs during the early 1930s, at the peak of the Great . The second sharp drop in the pellagra death rate is permanent and takes place during the late 1930s and 1940s. Because this drop happens at around the same time that state and federal governments introduced laws encouraging and mandating that breads and grains be enriched with niacin, historical observers have long attributed a causal connection between the laws and the reduction in pellagra. Using a panel of state-level data and a standard difference-in-differences set up, we exploit the interstate variation in the timing of niacin-fortification laws to identify their effects on pellagra and other nutrition-related . Consistent with the prevailing historical literature, the results suggest fortification laws significantly reduced the mortality associated with pellagra and low niacin consumption more generally.

In the final part of our analysis, we look at the long-term effects of early life exposure to high pellagra (low-niacin consumption) environments. Building on work in an earlier section of the paper and using a difference-in-differences estimation strategy, we first study how the arrival of the boll weevil, and the subsequent increase in local food production, affected the heights (a now common barometer of early-life health and nutrition) of World- II army recruits. The results suggest that the arrival of the boll weevil gave rise to taller, healthier soldiers, despite the reductions in farm income likely associated with the arrival of the boll weevil. The results are, moreover, most pronounced in counties producing high levels of cotton and in states with high pre-boll weevil pellagra rates. Following Neimish (2015), we also explore how early-life exposure to state-level niacin-fortification laws affected wages in the long-term. We find that fortification increased wages by approximately 2% amongst Southerners born after a

4 state passed a fortification law. The results are, again, larger for states that had higher levels of pre-fortification pellagra.

Taken together, our results contribute to the following literatures. The first, and probably the most important contribution, is to our understanding of the economic history of the American South. The American South has long lagged behind the North in economic performance, and only after World War II did incomes begin to converge.

Until 1940, income per capita in the South was 45 to 60 percent of the U.S. average; by 1980, that deficit had been reduced to 80 to 95 percent (Wright 1987). Standard explanations for these patterns fall into one of three categories. Institutional explanations consider national labor standards (Wright 1987); Civil Rights legislation (Wright 2013,

Collins 2003); and the decline of paternalism and other institutions hostile to black economic progress (Alston and Ferrie 1993, 1999). Technological explanations focus on air-conditioning (Biddle 2008, 2012), electrification (Downs 2014), and the mechanization of agriculture (Alston and Ferrie 1993, 1999). Disease-based explanations consider the eradication of hookworm and malaria (Bleakley 2007, 2010; Kitchens, 2013). The results here suggest a fourth possible mechanism: improved nutrition.

Second, there is a large and developing literature exploring how exposure to poor nutrition and parasitic diseases early in life adversely affect later life outcomes in terms of income, health, and educational attainment (see, for example, Bleakley 2007,

2010; Rivera et al. (1995); Hoddinott (2008)). To achieve identification, the current literature often focuses on negative shocks that adversely affect both nutrition and income (broadly construed). In a recent survey, for example, Lumey et al. (2011) reviews papers focusing on the Siege of Leningrad of 1941–1944 (Barber and Dzeniskevich 2005); the Dutch Winter of 1944–1945 (Trienekens 2000); crop failures in nineteenth- century Sweden and Finland (Bygren et al. 2000; Kannisto et al. 1997); seasonal in the Gambia between 1949 and 1994 (Moore et al. 1997); the Chinese Great Leap

Forward of 1959–1961 (Smil 1999) and recent seasonal famines in

(Moore et al. 2004). The results here complement and extend this literature. Of particular interest, we believe, are the effects of the boll weevil. What makes the boll weevil

5 interesting is that it breaks the usual correlation between nutrition and income: most shocks that reduce nutrition also reduce income, and it is difficult to disentangle the two effects as they are mutually reinforcing; but the boll weevil forced farmers to abandon a profitable crop (cotton) for less profitable crops (local food production) and yet historical observers believed the arrival of the boll weevil promoted better nutrition because of this shift. The data bear out this hypothesis.

Third, the observation that malnutrition can emerge in poor agricultural societies that rely heavily on cash crops is not unique to the American South. Describing conditions in the developing world in the late 1960s and early 1970s, the National

Academy of Sciences (1978, p. 44) explained that “changes in agricultural practices from mixed cropping to monocroppoing for cash and export . . . [might] result in health problems ranging from nutrition disorders to increased incidence of parasitic and infectious diseases.” A recent World Bank study of Malawi linked cash cropping, and subsequent displacement of local food production, to higher and stunting in later life (Wood et al. 2013). Similarly, a longstanding critique of British policy in colonial

India is that it fostered an undue reliance on cash crops, which disrupted local food production and left the indigenous populations poorly fed and vulnerable to frequent famines (Bhatia 1963).

2. Historical background and preliminary observations

Physicians and experts have long characterized pellagra by the four

D’s: , , , and death. Other symptoms include: sensitivity to sunlight; ; emotional disturbances; ; inflammation of the tongue; skin lesions; ; ; mental ; (loss of coordination) and paralysis in the extremities; and enlarged heart. Much like smallpox, which gave rise to a distinctive rash that made diagnosing severe cases of the disease easy, the dermatitis and skin discoloration associated with severe cases of pellagra facilitated a proper diagnosis. That said, for less severe cases, where the pathognomonic features of the disease were less pronounced, particularly those in relation to the skin, pellagra would

6 have been harder to distinguish from other ailments and may well have been underreported. These observations have clear implications for our empirical analysis. In particular, death rates for pellagra are probably well estimated because of the distinctiveness of the disease in severe cases, but incidence rates likely understate the prevalence of the disease.

Today the causes of pellagra are well understood. As noted above, the primary cause of the disease is inadequate niacin consumption, and secondary causes include a deficiency in (which is found in meat, fish, and eggs, and the body converts to niacin) and excess (which inhibits niacin ). Historically, however, the causes of pellagra were poorly understood. During the early 1900s, Southerners steadfastly rejected any claim that the disease might be associated with and poor diets. They saw such claims an indictment of Southern culture and habits, and instead argued that pellagra was spread by flies or spoiled corn (Mooney et al. 2014).

The first person to begin effectively arguing that pellagra was a nutritional disease was

Joseph Goldberger, who worked for the United States Public Health Service in the late

1910s and early 1920s. Through a series of detailed studies of diets in orphanages, sanitariums, and mill towns, Goldberger and his colleagues documented a tight correlation between and pellagra (Goldberger et al. 1915, 1920, 1929). But it was not until 1937 that showed definitively that pellagra was caused by inadequate niacin consumption (Elvehjem et al. 1937).

The first reported case of pellagra in the United States occurred in Georgia in

1902 and reached broader proliferation around 1906 (Bollet 1992). Historical observers have attributed the emergence of pellagra to changes in the milling of Midwestern corn.

Previous milling technology had removed less of the germ, retaining some niacin.

Newer technology removed the germ more completely, leading to a finer corn meal with a longer shelf life but much less niacin and other . Expansion of large- scale milling and movements of goods by railroad meant that this corn reached the

South in increasing quantities. Bollet (1992, p. 219) notes that “in the textile mill towns, surrounded by cotton fields, food was shipped in by railroad, and the cornmeal that

7 could be purchased in the company stores was processed in the Midwest, where it had been degerminated.” In addition, a report from the Thompson-McFadden Pellagra

Commission found that in six mill towns in South Carolina 2,515 residents (49%) consumed shipped cornmeal daily, 1,563 residents (30%) consumed shipped cornmeal habitually, and 1,052 residents (21%) consumed ship cornmeal rarely. This in contrast to the consumption of locally grown corn. They found that 628 residents (12%) consumed local corn daily, 291 residents (6%) consumed local corn habitually, 120 residents (2%) consumed local corn rarely, and 4,050 residents (80%) never consumed local corn (Siler et al. 1914).

Figure 1 maps the regional variation in pellagra rates and shows that this change in corn milling and the subsequent emergence of pellagra hit the South particularly hard.

While there was some pellagra in the border states and the Southwest, the disease reached its peak in the deep South, particularly, Alabama, the Carolinas, Florida,

Georgia, , , and Tennessee.

Why did the change in corn milling hit the South so hard? Aside from pork and molasses (neither of which was particularly high in niacin), the Southern diet was based mostly on corn. Niacin rich foods, such as fish, lamb, and poultry, were expensive and were not regularly consumed by poor Southern households (Hilliard 1969). Table 1 documents the Southern proclivity to eat corn. This table is based on The Study of

Consumer Purchases in the United States, 1935-1936, which collected data on food purchases from a large sample of households including Southern residents, Southern rural residents, and non-Southern residents. The table shows that while rural dwellers in the South ate less than non-Southerners, they consumed far more corn meal and grits (which are made from corn). In particular, Southern rural residents consumed 4.25 pounds of corn meal and hominy and 1.13 pounds of white bread per week. Non-Southern residents consumed 0.09 pounds of corn meal and hominy and 4.33 pounds of white bread per week. These differences are all highly significant.

The introduction of degerminated Midwestern corn probably would not have mattered as much had Southerners produced more corn for local consumption. Corn

8 that was grown and marketed for local consumption would have been consumed relatively quickly and the incentives to preserve it through milling and degermination would have been less pronounced. Locally-sourced corn could also be used to make hominy grits; milled corn from the Midwest could not. If prepared correctly, hominy grits would have contained more niacin than ordinary corn, corn meal, or corn bread.

Along these lines, while corn-based diets are generally correlated with pellagra, many

Indian societies with corn-based diets have managed to avoid pellagra because of their heavy reliance on hominy grits which involved soaking the corn in a lime/alkali solution that catalyzed the grain’s potential niacin and tryptophan (Carpenter 1983). In addition,

Southerners tended to plant and harvest , which contained relatively high levels of niacin, while the corn varieties grown in the Midwest contained 30 to 50 percent less niacin than sweet corn (Burkholder et al. 1944; Ayer 1895, p. 12-13). The upshot of all this is that as long as Southerners were consuming fresh, locally-grown corn that was not milled they would have been consuming more niacin than they did with imported Midwestern corn.

One of the central themes of the extent literature on pellagra is that the disease stemmed directly from the South’s cotton monoculture. In his study of South Carolina,

Edgar (1992) argues that a high debt burden forced many farmers to plant cotton, a cash crop with higher expected returns than other crops such as foods for the local market. In

1910, one of every five cultivated acres was devoted to cotton; by 1919 one of every two cultivated acres was devoted to cotton. Furthermore, despite the population of South

Carolina tripling from 1850 to 1935, the amount of food production remained about the same. As a result, South Carolina “had to import $70-$100 million worth of food annually. For poverty-stricken tenant farmers with little ready cash, this meant that there was less to eat. The consequent increased dependence on a diet of pork, cornbread, and molasses made poor Carolinians more susceptible to disease” (Edgar 1992, pg. 47).

A more general way to state this argument is the following: the production of cotton displaced local food production, and prompted poor Southern cotton farmers and laborers to import and consume corn from the Midwest, which though cheap, was also

9 degerminated and devoid of niacin. (See Lange et al. (2009) for evidence that cotton displaced corn production.) If so, one would expect a strong positive correlation with cotton production and pellagra. In years when cotton production was high (and crowded out the harvesting of food for local markets) pellagra rates should have risen, while in years of low cotton production (and by implication, relatively high rates of local food production) pellagra rates should have fallen.

Looking at trends in pellagra and cotton production, the data are broadly consistent with this view. Panel A of Figure 2 plots two time-series: the pellagra death rate in the American South from 1910 to 1950 and cotton-acres harvested from 1880 to

1950. These data suggest that pellagra erupted in the American South after two decades of steady growth in the amount of Southern farmland dedicated to cotton production.

The explosive growth in pellagra stops during the late 1910s at around the same time that the cotton economy stagnates with the penetration of the boll weevil. Pellagra rebounds during the late 1920s, shortly after the effects of the boll weevil begin to recede and the cotton economy recovers. Pellagra plummets again during the late 1920s and early 1930s, with the onset of the Great Depression and a sharp decline in cotton-acres harvested. It is only after the discovery of niacin in 1937, marked by the horizontal line in the graph, and the passage of laws mandating the fortification of grains and breads with niacin, that the correlation between cotton production and pellagra seems to break down. Panel B, C, and D of Figure 2 plot similar graphs for North Carolina, South

Carolina, and Louisiana. North Carolina and South Carolina will be particularly important for our analysis since they are the only states that report pellagra deaths at the county level during this time period.

Plotting pellagra directly against annual cotton acres harvested, Figure 3 highlights the correlation between pellagra and cotton production more clearly. What remains unclear, however, is the extent to which one might attribute causality to this relationship. One obvious concern is the large number of potentially confounding events during the 1900-50 period. In particular, there is a well-developed literature documenting the long-term economic significance of the events like the Great Influenza

10 (Almond 2006; Clay, Lewis and Severnini 2015; and Noymer 2010); enrichment (Neimish 2015); the Great Migration (Black et al. 2015); various New Deal programs, particularly those aimed at agriculture (Depew et al. 2013); the eradication of malaria and hookworm (Bleakely 2010, 2007); milk fortification; and salt iodization

(Adhvaryu et al. 2015, Feyrer et al. 2013). Any of these events might confound the time series above and/or interact with pellagra in important ways.

In the empirical analysis that follows we, therefore, take steps to control for these and other potential confounders. We believe that many of the confounders will be differenced out with year and state/county fixed effects. We will also control for the malaria death rate and a measure of cash crop revenue in some specifications. Finally, for interventions like milk fortification, migration, and salt iodization to drive our results they would have to work through very specific channels. For instance, for migration to be driving our results it must be the case that only unhealthy individuals migrate out of a location. Similar stories would need to be told for milk fortification and salt iodization, whereby adding iodine to salt somehow leads people to consume more niacin.

Building on an older literature, one might also argue that the South was over- producing cotton, and that as a result, years of high cotton production were also years of low income.3 In this case, pellagra would not have been the result of cotton displacing local food production (and raising the price of such food); instead, the disease would have stemmed from the fact that cotton production was driving down income and the reduction in income, not the displacement of local food, caused families in the cotton

South to consume less food.

3. A Simple Economic Framework

In this section, we sketch out a simple economic framework that can explain how the South’s reliance on cotton might have promoted higher pellagra rates in particular, and poor nutrition more generally construed. One of the central messages of

3 See McGuire and Higgs (1978) and McGuire (1980) for evidence on the profitability of cotton.

11 this framework is that cash cropping does not necessarily result in poor nutrition; it only does so under certain conditions.

Consider, then, a largely agricultural economy, where farmers are deciding how to allocate their land, labor, and capital: they can either plant, harvest, and sell a cash crop in an international market, in this case, cotton; or they can deploy their resources to produce food for the local market. In deciding how to allocate resources, farmers look toward expected prices. To the extent expected cotton prices are relatively high, farmers would allocate their land and resources to planting and selling cotton and produce little, if any, food for the local market. Driving up the relative price of locally-produced foods, consumers in this economy would substitute away from such foods and begin importing degerminated corn from the Midwest, which as noted above, was niacin deficient. As an empirical matter, we cannot observe the expected price of cotton or the price of locally produced foods, and instead rely on acres dedicated to cotton and food production, which will usually be highly correlated with expected prices. However, we will also discuss and exploit situations were expected prices do not drive planting decisions.

For the process above to have yielded high rates of pellagra, two conditions need to be satisfied. First, it must be the case that there are, in fact, nutritional differences between imported corn and locally-sourced foods. While we cannot directly test this proposition, there is anecdotal evidence to suggest this was the case. In the case of corn, locally-produced corn did not require degermination to extend shelf life and so it is thought to have been healthier and richer in niacin. There is also evidence (discussed below) that aside from corn, Southern farmers would switch to growing when cotton prices were low or when cotton production was not feasible. Peanuts were rich in niacin and other micronutrients. Second, it must the case that the price effect dominates any income effect. More precisely, if demand for nutrition grows with income and higher cotton prices, it is possible that consumers in the region might begin importing relatively expensive niacin-rich foods, rather than degerminated corn. Put another way, if cash cropping generates sufficiently high wages and income and nutrition is a normal good, cash cropping need not imply poor nutrition.

12

4. Data and empirical strategy

4.a. Data

County and individual level data from several sources are used to document the rise and fall of pellagra in the American South. Consistent data on pellagra deaths at the county level during our study period are, to our knowledge, only available for North

Carolina and South Carolina. Pellagra deaths for counties in North Carolina from 1915-

1950 come from The Annual Report of the Bureau of Vital Statistics of the North Carolina State

Board of Health.4 We have also collected county-level death counts for malaria, , typhoid, and from 1915-1930. We use these diseases as placebo tests in our analysis. The North Carolina State Board of Health did not issue a Vital Statistics report for the years 1918 and 1919 so we do not have data on deaths for these years. Pellagra deaths for county in South Carolina from 1915-1925 come from Annual Report of the State

Board of Health of South Carolina.5 To calculate the pellagra death rate at the county-level every year we perform a linear interpolation of the county population in-between censuses.

Data on crop acreage at the county-level comes from Haines et al. (2015) United

States Agriculture Data, 1840-2010. These data are a digitized version of data originally provided in publications from the United States Bureau of the Census. These data provide county-level crop production for years in which the Census of Agriculture was performed. Specifically, we use data on cotton, corn, , and sweet potato acreage.

To calculate acreage per capita in every year we linearly interpolate both the county population and crop acreage.

To study the long-run effects of nutrition we first look at the height of World

War II recruits. Heights of World War II recruits come from the World War II Army

Enlistment Records 1938-1946 provided by the National Archives and Records

4 In 1920 and 1921 these reports are found in The Health Bulletin published by the North Carolina State Board of Health. 5 We are in the process of digitizing the remainder of the South Carolina death data, which should eventually give us a time-series to 1950.

13 Administration. These micro-level data provide height for approximately 9.2 million

World War II soldiers. We make several sample restrictions to obtain a sample that is likely to be representative for each cohort. We restrict to white men who were drafted, born in the South, born between 1915 and 1921, enlisted between 1940 and 1946, report a valid height (between 5 feet and 6.5 feet), and report a valid weight (between 100 pounds and 400 pounds).6 Since we do not observe county of birth, we further restrict to soldiers that were living in the same state they were born in at the time of enlistment.

For these soldiers we assume that at the time of enlistment they were living in the same county they were born in.

Finally, we look at the long-run effects of fortification laws on income using the

1% Integrated Public Use Micro Sample (IPUMS) for the United States Census in 1970, 1980, and 1990 (Ruggles et al. 2015). We restrict to individuals who are born in the United

States, since individuals born outside the United States might not have been exposed to fortification during childhood. We further restrict to individuals born 4-6 years prior to the passage of a mandatory fortification law and individuals born 1-3 years after the passage of a law. As discussed in section 7.b., most of the effect of nutrition on later life outcomes occurs during the first three years of life. By restricting to individuals born 4-6 years prior to the passage of a law we are removing individuals who might have been partially exposed to improved nutrition during early life. We, therefore, compare individuals who are not exposed to improved nutrition during childhood to those born just after the passage of a law and, therefore, completely exposed to improved nutrition during early childhood.

4.b. Empirical Strategy

We estimate the effect of the boll weevil, voluntary fortification, mandatory fortification, and long-run outcomes using a standard difference-in-differences intensity of treatment . Our baseline model is given by:

6 We define Southern states to include: Alabama, Arkansas, Florida, Georgia, Louisiana, Mississippi, Missouri, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, and Virginia.

14

[표푢푡푐표푚푒]푐푡 = 훼 + 휃1 ∗ [푡푟푒푎푡푚푒푛푡]푐푡 + 휃2 ∗ [푡푟푒푎푡푚푒푛푡]푐푡 × [푖푛푡푒푛푠푖푡푦]푐 + 휃푐 (1)

+ 휃푡 + 휀푐푡

In equation (1), [표푢푡푐표푚푒]푐푡 is our outcome of interest (usually the log of the pellagra death rate) in county c in year t. When using height and the log of income as the outcome the variable is recorded at the individual level not at the county level. When analyzing the boll weevil, [푡푟푒푎푡푚푒푛푡]푐푡 is an indicator variable that takes a value of one after the boll weevil has arrived in a county. When analyzing voluntary fortification,

[푡푟푒푎푡푚푒푛푡]푐푡 takes a value of one starting in 1937. When analyzing mandatory fortification, [푡푟푒푎푡푚푒푛푡]푐푡 takes a value of one starting in the year a state passed a mandatory fortification law. We interact [푡푟푒푎푡푚푒푛푡]푐푡 with a county level measure of the intensity of the treatment: [푖푛푡푒푛푠푖푡푦]푐. The intensity of treatment variables will be explained as they are introduced. All intensity of treatment variables have been normalized so they have a mean of zero and a standard deviation of one. Finally, we include county fixed effects to control for unobserved time invariant county characteristics and year fixed effects to control for any unobserved shocks in a particular year. The standalone term, [푖푛푡푒푛푠푖푡푦]푐 , is completely absorbed by the county fixed effects. Standard errors are clustered at the county-level.

As mentioned in Section 2 there are a number of factors that might confound our analysis. We believe that most of these factors will be differenced out as a result of our difference-in-differences analysis. For example, if the Great Influenza Pandemic of 1918 interacts with pellagra in some way, than this should only be true in the year 1918 and will be differenced out in our year fixed effects. According to Bleakley (2010) the malaria intervention in the American South occurs almost concurrently with the arrival of the boll weevil in North and South Carolina. Again, the malaria intervention might interact the pellagra and confound our results. Accordingly, we control for the malaria death rate in certain specifications. Finally, a valid potential confounder of our boll weevil analysis would involve migration out of counties that were hit by the boll weevil. We

15 discuss the type of selection that would be required for migration to drive our results in section 5.b.

5. Cotton and the rise of pellagra

5.a. Relationship between cotton and pellagra

Although the existing literature on pellagra posits a causal relationship between cotton production and pellagra, there is a dearth of systematic data establishing even a correlation between the two. Accordingly, in this section, we analyze state and county- level data for evidence of a correlation between cotton production and pellagra. The state-level analysis provides the advantage of having annual data for both pellagra death rates and crop acres harvested. The county-level analysis provides the advantage of a much finer geographic unit despite only being able to measure crop acreage when the Census of Agriculture is taken. We also search for evidence that cotton production displaced the production of local foods, particularly corn, peanuts, and sweet potatoes.

We begin our empirical analysis by exploring the relationship between land devoted to cotton production and pellagra, as previously demonstrated in Figures 2 and

3. To examine this relationship we first collect data on the number of pellagra deaths in a state from the Mortality Statistics of the United States starting in 1900. We combine these data with state-level data on the number of acres of cotton, corn, peanuts, and sweet potatoes harvested yearly and the number of 480-pound cotton bales that were produced.7 We restrict our attention to only the nineteen states that produced cotton over this period and we note that we do not have a balanced panel. States enter our sample as they begin reporting to the Mortality Statistics of the United States. The first cotton producing states to report to the Mortality Statistics are California and North

Carolina in 1910. Furthermore, we believe that the relationship between cotton, corn, peanut, and sweet potato acreage and pellagra should be most pronounced prior to the

7 These data come from the United States Department of Agriculture, National Agricultural Statistics Service Database.

16 discovery of niacin and the commencement of fortification in 1937. For these two reasons, we restrict our attention to the period 1910-1936. We estimate the following equation:

푙푛[푝푒푙푙푎푔푟푎]푠푡 = 훼 + 휃1 ∗ [푐푟표푝 푎푐푟푒푠 푝푐]푠푡 + 휃2 ∗ [푐표푡푡표푛 푟푒푣푒푛푢푒]푠푡 + 휃푠 + 휃푡 (2)

+ 휀푠푡

Where ln[pellagra]st is the pellagra death rate in state s in year t. The variable [crop acreage]st is the number of acres per capita of a given crop. The variable [cotton revenue]st is the revenue per acre of cotton harvested.8 State and year fixed effects are included in the model. All crop acreage per capita variables are normalized to have a mean of zero and a standard deviation of one. Standard errors are clustered at the state-level.

The results from estimating equation (2) are given in Table 2. In column (1), a one standard deviation increase in cotton acres per capita is associated with an increase in the pellagra death rate by approximately 6%. Columns (2)-(4) show that when states use their land to produce local, nutritional food, the pellagra death rate is lower. A one standard deviation increase in acres per capita devoted to corn, peanuts, or sweet potatoes significantly decreases the pellagra death rate by 1.4-4%.

Using the same data we show evidence that, because land is fixed, food production will likely decrease when acres devoted to cotton production increases. To show this relationship we estimate the following equation:

[푓표표푑 푎푐푟푒푠 푝푐]푠푡 = 훼 + 휃 ∗ [푐표푡푡표푛 푎푐푟푒푠 푝푐]푠푡 + 휃푠 + 휃푡 + 휀푠푡 (3)

where [food acres pc]st is the number of acres per capita devoted to the production of corn, peanuts, or sweet potatoes and [cotton acres pc]st is the number of acres per capita devoted to the production of cotton. Again, all crop acreage per capita variables are

8 National cotton prices come from the Historical Statistics of the United States Database; series Da755-765.

17 normalized to have a mean of zero and a standard deviation of one. We now begin our analysis in 1900 since, unlike pellagra, crop acreage is reported back to 1900.

Table 3 provides the results from estimating equation (3). Column (1) shows that increasing cotton acres per capita by one standard deviation decreases corn acres per capita by 0.58 standard deviations. In column (2), a one standard deviation increase in cotton acres per capita is associated with a decrease in peanut acres per capita by over one standard deviation. Finally, in column (3), a one standard deviation increase in cotton acres per capita is associated with a decrease in sweet potato acres per capita by approximately 0.13 standard deviations.

We further test relationships between crop acreage and pellagra at the county- level. The county-level relationship between pellagra and crop acreage is shown in columns (5)-(8) of Table 2. Column (5) shows that a one standard deviation increase in cotton acres per capita is associated with an increase in the pellagra death rate by approximately 19%. In column (6), a one standard deviation increase in corn acres per capita is associated with a decrease in the pellagra death rate by approximately 16%.

Finally, the coefficients in columns (7) and (8), which report the relationship between pellagra, peanuts, and sweet potatoes, are both small and insignificant.

Finally, we show evidence that cotton production displaces food production at the county-level in Table 3. In column (4), a one standard deviation increase cotton acres per capita is associated with a decrease in corn acres per capita by approximately 0.32 standard deviations. The coefficients in columns (5) and (6), which show the relationship between cotton, peanut, and sweet potatoes, are once again small and insignificant.

Tables 2 and 3 provide evidence about one of the primary causes of the rise of pellagra in the United States. Table 2 demonstrates that, in the pre-fortification period, the pellagra death rate increased when land devoted to cotton production increased, and decreased when the land devoted to cotton production decreases. Table 3 demonstrates that a likely reason for this relationship is the crowding-out of local food production by cotton. Since land is fixed, if farmers decide to dedicate more land to cotton production they, necessarily, must dedicate less land to the production of locally grown foods. The

18 decline in local food production means that individuals must consume more imported food that was, usually, lower in nutritional value. These results hold at both the state and the county level.

Despite the consistent results in Tables 2 and 3, there is still the potential concern that high cotton production is associated with unobserved factors that increase the pellagra death rate. For example, perhaps cotton production decreases income, which leads to worse diets and increases in pellagra. To deal with these concerns, we next turn our attention to an exogenous shock to cotton production that was brought about by the arrival of the boll weevil.

5.b. The Boll Weevil

The boll weevil, a beetle that feeds on cotton leaves, squares (flower buds), and bolls, appeared in Texas in 1892. Figure 4 shows its progression through the cotton belt.

By 1922, the boll weevil had infected the entire cotton region. The arrival of the weevil has significant impacts on agriculture. Lange et al. (2009) show that the arrival of the boll weevil reduced county cotton production, yields, and land values. They also find evidence that farmers shifted crops after the arrival of the boll weevil (p. 710): “The decline in cotton acreage raises the question of what southern farmers did with the released land. … Overall, the corn results indicate a greater movement to alternative crops than suggested in the literature, which has downplayed the boll weevil’s effects on diversification.” They also write in a footnote that (p. 710): “Based on the census data, we also find production of hay, Irish potatoes, peanuts, , and sweet potatoes; cane, among other crops, showed statistically significant increases after the arrival of the weevil.”

The invasion of the boll weevil allows us to exploit an exogenous change in crop mix that arguably affected health. Data on the year the boll weevil first arrived in a county are taken from Lange et al. (2009), which originally came from USDA boll weevil maps. To examine the impact of the boll weevil on pellagra we estimate equation (1). We use two different variables to measure intensity of treatment. Our first variable is the average pellagra death rate in 1915 and 1916, just before the boll weevil arrived in North

19 Carolina and South Carolina. The idea is that places with higher pre-boll weevil pellagra death rates likely had worse baseline nutrition and, therefore, had more to gain from the arrival of the boll weevil. Our second intensity of treatment variable is cotton acres per capita in the county in 1909, which is the year closest to the arrival of the boll weevil that the Census of Agriculture was taken in.

Table 4 shows the impact the boll weevil had on pellagra death rates in the North

Carolina and South Carolina. Column (1) indicates that the pellagra death rate decreased by 25% after the arrival of the boll weevil in a county. Column (2) indicates that for the county with the average amount of pellagra prior to the arrival of the boll weevil, the pellagra death rate decreased by 25%. However, a one standard deviation increase in the pre-boll weevil pellagra death rate is associated with an additional decrease in the pellagra death rate of approximately 29%. In column (3), we show that for the county with the average amount of cotton acres per capita prior to the arrival of the boll weevil, the pellagra death rate decreased by 25% after the arrival of the boll weevil. A one standard deviation increase in cotton acres per capita is associated with an additional 4% decrease in the pellagra death rate. Finally, in column (4) we restrict to only counties that produced over 100 acres of cotton in 1889.9 We, again, find that the arrival of the boll weevil is associated with a 19% reduction in the pellagra death rate.

Columns (1)-(4) are estimated over the time period 1915-1950 and, as discussed in the introduction, there were several other events that lowered the pellagra death rate over this time period; specifically the Great Depression and fortification laws. In column

(5) we restrict the period of our study to 1915-1929; before the start of the Great

Depression and the passage of fortification laws. We also control for the county-level malaria death rate in each year since the effort to reduce malaria in the American South occurred almost concurrently with the arrival of the boll weevil in North Carolina and

South Carolina. We, again, find that the arrival of the boll weevil in a county is associated with a decrease in the pellagra death rate by approximately 20%. Finally, for

9 As shown in Figure 4, cotton cannot be grown in several counties in the western part of North Carolina.

20 our story to be consistent the arrival of the boll weevil should not be associated with significant decreases in non-nutrition related diseases. In column (6) we examine typhoid, in column (7) we examine tuberculosis, and in column (8) we examine measles.

The coefficient for the arrival of the boll weevil for all three diseases and is positive and not significant. Thus, the arrival of the boll weevil was not associated with general improvements in health, but instead it was only associated with improvements in nutrition. After all, the only way to cure pellagra is to consume more niacin.

As already mentioned, a potentially valid confounder of our boll weevil analysis involves migration out of counties that were hit by the boll weevil. Indeed, Lange et al.

(2009, p. 715) find “the weevil appears to have unleashed a wave of internal migration, leading to local population gains before contact and substantial losses after the onset of significant crop damage.” The placebo tests in Table 4, columns (6)-(8), show that the least healthy members of the population are not selectively leaving a county after the arrival of the boll weevil. Therefore, for selective migration to be driving our results it must be the case that only individuals who are so poorly nourished that they develop pellagra migrate out of the county, while other unhealthy individuals do not.

6. Niacin and the fall of pellagra

6.a. Voluntary Fortification

Shortly after 1937, when niacin was identified as the cause of pellagra, voluntary fortification of cereal-grain products began (Park et al. 2000, 2001). In 1939, the Council on Foods and Nutrition of the American Medical Association encouraged “with some qualification, fortification of certain staple foods with and minerals, specifically the restorative additions of , niacin, , iron, and calcium to white flour and white bread” (Wilder 1956, pg. 1540). In 1941 the FDA established standards for the fortification of bread, which are displayed in Table 5. Up to January of 1943, fortification was increasingly common, but remained voluntary. Further, it was primarily restricted to bread and flour. Fortification was briefly required at the national level, due to War

21 Food Order No. 1. When federal war powers ended, regulation devolved to the states, at which point, several states passed mandatory fortification laws.

Table 6 provides estimates of equation (1) and shows the effect of voluntary fortification on the pellagra death rate. As mentioned in section 4.b., The variable

[treatment]ct takes a value of one beginning in 1937. For the state-level analysis the intensity of treatment variables are the pre-niacin pellagra death rate and pre-niacin cotton acres per capita; both variables are averages over the years 1932-1936. In the county-level analysis we measure pre-niacin cotton acres per capita as the average from the years 1929 and 1934. Finally, since our treatment variable turns on in 1937 we do not include year effects, but instead, include a quadratic time trend.

Columns (1)-(3) examine the state level impact of voluntary fortification. In column (1), voluntary fortification is associated with a decrease in the pellagra death rate by 2%. In column (2) we show that virtually the entire decrease in the pellagra death rate is located in states with high pre-niacin pellagra death rates; a one standard deviation increase in the pre-niacin pellagra death rate is associated with a decrease in the pellagra death rate of 6%. Finally, a one standard deviation increase in pre-niacin cotton acres per capita is associated with a 4.5% decrease in the pellagra death rate.

Columns (4)-(6) examine the county level impact of voluntary fortification in NC.

In North Carolina, voluntary fortification is associated with an 18% decrease in the pellagra death rate (column (4)). Column (5) shows that a one standard deviation increase in the pre-niacin pellagra death rate is associated with a decrease in the pellagra death rate of 25%. Finally, column (6) shows that a one standard deviation increase in pre-niacin cotton acres per capita is associated with a 6% decrease in the pellagra death rate [p-value = 0.109]. Table 6 provides strong evidence that the voluntary fortification of cereal-grain products, beginning in 1937, was associated with significant decreases in the pellagra death rate.

6.b. Mandatory Fortification Laws

22 Pellagra was virtually eliminated in the United States by the passage of fortification laws that mandated that cereal grain products be enriched with niacin and other , as well as iron. Twenty-eight states passed some form of a mandatory fortification law over the decade 1940-1949. Figure 1 shows the states and years in which fortification laws were passed. The year of fortification law passage was taken from Park et al. (2001). Most of these laws required bread and flour to be enriched with thiamine, niacin, riboflavin, iron, and calcium, however, many laws in Southern states also pertained to cornmeal and hominy grits, as these were staples of the Southern diet.10 As a result, major producers of corn meal and grits in the Midwest began fortifying their products. A 1957 survey found that nearly all hominy grits sold were enriched, and that cornmeal was generally enriched, except in Florida and Virginia, where enrichment was less typical (Park 2001, NRC 1958).

We again, estimate equation (1) using the intensity of treatment measures as they were defined in section 6.a. The variable [treatment]ct takes a value of one beginning in the year that a mandatory fortification law was passed by a state. We once again include state fixed effects in the state-level analysis since there is variation in the years that states passed enrichment laws. In the county-level analysis the quadratic time trend is used since we only have data for North Carolina.

Table 7 shows the effect of mandatory fortification laws on the pellagra death rate. Columns (1)-(3) examine the state-level impact. The passage of a fortification law is associated with a decrease in the pellagra death rate by approximately 6% (column (1)).

In column (2), we show that a one standard deviation increase in the pre-niacin pellagra death rate is associated with a decrease in the pellagra death rate by 5%. In column (3), a one standard deviation increase in the pre-niacin number of cotton acres per capita is associated with a 2% decrease in the pellagra death rate.

Columns (4)-(6) examine the county-level impact of the North Carolina mandatory fortification law, which was passed in 1945. The passage of the mandatory

10 States that passed laws pertaining to corn meal and hominy grits were: Alabama, Georgia, Mississippi, North Carolina, South Carolina, and Texas.

23 fortification law is actually associated with a 13% increase in the pellagra death rate in

NC (column (4)). Note however, that if the quadratic time trend is not included in the model, the passage of the mandatory fortification law is associated with a 5% reduction in the pellagra death rate. This indicates that our regression analysis has difficulties separating the effects of the law passage from the pre-existing trend, which is potentially the result of voluntary fortification.

In column (5), a one standard deviation increase in the pre-niacin pellagra death rate is associated with a decrease in the pellagra death rate by 19%. Finally, a one standard deviation increase in the pre-niacin amount of cotton acres per capita is associated with a 5% decrease in the pellagra death rate. Table 7 demonstrates that the passage of mandatory fortification laws were associated with significant decreases in the pellagra death rate. The passages of mandatory fortification laws were also associated with the virtual elimination of pellagra in the American South by 1950 as shown in

Figure 2.

7. Long-run effects of nutrition

Finally, we turn to the examination of the long-run effects of nutrition. We believe that the long-run effects of nutrition during early childhood might be considerable given previous studies. The most important of these studies took place between 1969 and 1977 in four rural villages in Guatemala. Two villages were randomly selected to receive a high-energy, high- nutrition supplement (atole) and two villages were selected to receive a low-energy, no-protein nutrition supplement (fresco).

Supplements were provided freely to pregnant and lactating mothers and children under the age of seven. Schroeder et al. (1995) find that children receiving the nutritious supplement were taller and weighed more during the first three years of life. Rivera et al.

(1995) find that children receiving the nutritious supplement were taller and weighed more during adolescence. Finally, Hoddinott et al. (2008) find that exposure to the nutritious supplement during the first three years of life significantly increased wages

24 by approximately 46%.11 Virtually all studies of the long-run impacts of nutrition find that nutrition matters most during the first three years of life. As a result of these findings, we focus on the effects of early childhood nutrition on height and income.

We examine the long-run impact of the improvement in nutrition brought about by the arrival of the boll weevil on the height of World War II recruits. Using these data we estimate a model very similar to equation (1), but with individual, rather than state or county-level data. We include county of birth, year of birth, and year of enlistment fixed effects in our model.

The results from estimating the model are shown in Table 8. In column (1), the arrival of the boll weevil is associated with an increase in the height of World War II recruits by over a tenth of an inch (0.13; p-value 0.107). Column (2) shows that people born in states with average pre-boll weevil pellagra death rates experienced an increase in height of about 0.1 inches after the arrival of the boll weevil. A one standard deviation increase in pre-boll weevil pellagra levels is associated with an additional 0.21-inch increase in height. Finally, column (3) shows that for recruits born in counties with average pre-boll cotton acres per capita the arrival of the boll weevil is associated with an increase in height of about 0.02 inches. A one standard deviation increase in pre-boll weevil cotton acres per capita is associated with an additional 0.18-inch increase in height. Table 8 provides strong evidence that the nutritional improvement associated with the arrival of the boll weevil increased heights for World War II recruits.

Finally, we turn to the long-run impact of mandatory fortification laws on income. We, again, estimate a model very similar to equation (1), and all intensity of treatment variables are defined analogously to section 6.a.

Table 9 shows the long-run impact of mandatory fortification laws on income.

Column (1) indicates that individuals born after the passage of a mandatory fortification law earn approximately 3% higher incomes (p-value = 0.169). In column (2) we see that a one standard deviation increase from the average pre-niacin pellagra death rate is

11 Yet another study, Haas et al. (1995) finds that consuming the nutritious supplement during the first three years of life increased physical work capacity (VO2max).

25 associated with a significant 2.5% increase in income. Finally, in column (3) a one standard deviation increase from the pre-niacin cotton acres per capita is associated with a 1% increase in income (p-value = 0.105). Table 9 provides convincing evidence that incomes increase as a result of state-level fortification laws.

8. Conclusion

In this paper we have documented the rise and fall of pellagra in the United States. The rise of pellagra was associated with increases in cotton production in the

American South and the substitution of locally grown corn for corn that was milled in the Midwest. The Midwestern milled corn was degerminated, which stripped the corn of much of its nutritional value. We show strong empirical evidence that pellagra is positively associated with land devoted to cotton production and that cotton did, indeed, crowd-out the production of local corn and other crops such as peanuts and sweet potatoes.

To establish the causal relationship between cotton acreage and pellagra more definitively we exploit an exogenous shock in cotton production that occurred from the arrival of the boll weevil. Farmers diversified crops after the arrival of the boll weevil, which led to improved nutrition and reductions in the pellagra death rate. Furthermore, we provide evidence that the arrival of the boll weevil was not associated with overall improvements in health. The death rates for typhoid, tuberculosis, and measles experience no significant changes after the arrival of the boll weevil. This implies that the boll weevil’s only short-run impact on health was through improved nutrition.

The fall of pellagra in the American South was associated with the discovery of niacin as anti-pellagrant in 1937. Shortly after this discovery, bakeries and mills began to voluntarily fortify their products with thiamine, niacin, riboflavin, iron, and calcium.

Pellagra was virtually eliminated in the United States with the passage of state-level mandatory cereal-grain fortification laws from 1940-1949. We show that both voluntary and mandatory fortification significantly decreased the pellagra death rate and led to pellagra’s elimination around 1950.

26 Finally, we demonstrate that the improvements in nutrition brought about by the arrival of the boll weevil resulted in significant increases in the height of World War II recruits. We also show that the improvements in nutrition brought about by mandatory fortification resulted in significant increase in income for individuals later in life. These results are consistent with the fact that nutrition in early life leads to permanent, long- run effects on health and economic performance.

27 References

Adhvaryu, Achyuta, Steven Bednar, Teresa Molina, Quynh Nguyen, and Anant Nyshadham (2015). “Iodization and the Enfranchisement of the American Worker.” Working paper.

Almond, Douglas (2006). “Is the 1918 Influenza Pandemic Over? Long-Term Effects on In Utero Influenza Exposure in the Post-1940 U.S. Population.” Journal of Political Economy 114(4): 672 – 712.

Alston, Lee J., and Joseph Ferrie (1993). "Paternalism in Agricultural Labor Contracts in the U.S. South: Implications for the Growth of the Welfare State," The American Economic Review 83(4): 852‐ 876.

Alston, Lee J., and Joseph P. Ferrie (1999). Paternalism and the American Welfare State: Economics, Politics, and Institutions in the U.S. South, 1865‐1965. Cambridge University Press.

Ayer, Hartwell M. (1895). Resources and Manufacturing Industries of the State of South Carolina. Charleston: Lucas & Richardson Company.

Barber, John and Andrei Dzeniskevich (2005). Life and Death in Besieged Leningrad, 1941- 1944. Basingstoke: Palgrave Macmillan.

Bhatia, B.M. (1963). Famines in India: A Study in Some Aspects of the Economic History of India. New Asia Publishing House.

Biddle, Jeff (2008). “Explaining the Spread of Residential Air Conditioning, 1955-1980.” Explorations in Economic History 45(Sept): 402-423.

Biddle, Jeff (2011). "Making Consumers Comfortable: The Early Decades of Air Conditioning in the US", Journal of Economic History 71(4): 1078-1094.

Black, Dan A., Seth G. Sanders, Evan J. Taylor, and Lowell J. Taylor (2015). “The Impact of the Great Migration on Mortality of African : Evidence from the Deep South.” The American Economic Review 105(2): 477-503.

Bleakley, Hoyt (2007). “Disease and Development: Evidence from Hookworm Eradication in the American South.” The Quarterly Journal of Economics 122(1): 73-117.

Bleakley, Hoyt (2010). “Malaria Eradication in the Americas: A Retrospective Analysis of Childhood Exposure.” American Economic Journal: Applied Economics 2(April 2010): 1 – 45.

28

Bollet, A. J. (1992). “Politics and pellagra: the epidemic of pellagra in the US in the early twentieth century.” The Yale journal of biology and , 65(3), 211.

Bygren, Lars Olov, Soren Edvinsson, and Goran Brostrom (2000). “Change in Food Availability During : Is It Related to Adult Sudden Death From Cerebro- and Cardiovascular Disease in Offspring?” American Journal of Biology 12(June): 447- 453.

Depew, Briggs, Price Fishback and Paul Rhode (2013). “New Deal or No Deal in the Cotton South: The Effect of the AAA on the Agriculture Labor Structure.” Explorations in Economic History 50 (October): 466-486.

Burkholder, Paul R., Ilda McVeigh, and Dorothy Moyer (1944). “Niacin in .” Yale Journal of Biology and Medicine 16(6): 659-63.

Carpenter, K.J. (1983). “The Relationship of Pellagra to Corn and the Low Availability of Niacin in Cereals.” Experientia Supplement 44: 197-222.

Clay, Karen, Joshua Lewis and Edson Severnini (2015). “Pollution, Infectious Disease, and : Evidence from the 1918-1919 Spanish Influenza Pandemic.” Working paper.

Edgar, Walter (1992). South Carolina in the Modern Age. Columbia: University of South Carolina Press.

Elvehjem, Conrad A., R. J. Madden, F. M. Strong, and D. W. Woolley (1937). "Relation of nicotinic acid and nicotinic acid amide to canine black tongue." Journal of the American Chemical Society 59(9): 1767-1768.

Etheridge, Elizabeth W. (1972). The butterfly caste. A social history of pellagra in the South. Westport, CT: Greenwood Publishing Company.

Feyrer, James, Dimitra Politi, and David N. Weil (2013). “The Cognitive Effects of Deficiency: Evidence from Salt Iodization in the United States.” NBER Working Paper No. 19233.

Goldberger, Joseph, C. H. Waring, and David G. Willets (1915). “The Prevention of Pellagra: A Test of Diet among Institutional Inmate.” Public Health Report 30(43): 3117- 3131.

29 Goldberger, Joseph, G. A. Wheeler, and Edgar Sydenstricker (1920). “A Study of the Relation of Diet to Pellagra incidence in Seven Textile-Mill Communities of South Carolina in 1916.” Public Health Report 35(12): 648-713.

Goldberger, Joseph, G. A. Wheeler, Edgar Sydenstricker, and Wilford I. King (1928). “A Study of Endemic Pellagra in Some Cotton-Mill Villages of South Carolina: An Abstract.” Public Health Reports 43(41): 2645 – 2647.

Goldberger, Joseph, G. A. Wheeler, Edgar Sydenstricker, and Wilford I. King (1929). “A Study of Endemic Pellagra in Some Cotton-Mill Villages of South Carolina.” Hygience Laboratory Bulletin No. 153.

Haas, Jere D., Elkin J. Martinez, Scott Murdoch, Elizabeth Conlisk, Juan A. Rivera, and Reynaldo Martorell (1995). “Nutritional Supplementation during the Preschool Years and Physical Work Capacity in Adolescent and Young Adult Gautemalans.” The Journal of Nutrition 125: 1078S-1089S.

Haines, Michael, Price Fishback, and Paul Rhode. United States Agriculture Data, 1840 - 2010. ICPSR35206-v2. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor], 2015-08-05. http://doi.org/10.3886/ICPSR35206.v2

Hilliard, Sam (1969). “Hog Meat and Cornpone: Food Habits in the Ante-Bellum South.” Proceedings of the American Philosophical Society 113(1): 1-13.

Hoddinott, John, John A. Maluccio, Jere R. Behrman, Rafael Flores, and Reynaldo Martorell (2008). “Effect of a nutrition intervention during early childhood on economic productivity in Guatemalan adults.” The Lancet 371(February 2, 2008): 411-416.

Humphreys, Margaret (2009). "How four once common diseases were eliminated from the American South." Health Affairs 28(6): 1734-1744.

Hunter, W.D. and B.R. Coad (1922). The Boll weevil Problem. Washington, D.C.: United States Department of Agriculture.

Kannisto, Vaino, Kaare Christensen, and James W. Vaupel (1997). “No Increased Mortality in Later Life for Cohorts Born during Famine.” American Journal of Epidemiology 145(11): 987-994.

Kitchens, Carl (2013).“The Effects of the Works Progress Administration's Anti Malaria Programs in Georgia 1932-1947” Explorations in Economic History 50(4): 567-581.

Lange, Fabian, Alan L. Olmstead and Paul W. Rhode (2009). “The Impact of the Boll weevil, 1892–1932.” The Journal of Economic History 69(3): 685 – 718.

30 Lumey, L.H., Aryeh D. Stein, and Ezra Susser (2011). “Prenatal Famine and Adult Health.” Annual Review of Public Health.

McGuire, Robert A. (1980). “A Portfolio Analysis of Crop Diversification and Risk in the Cotton South.” Explorations in Economic History 17(4): 342-71.

McGuire, Robert A., and Robert Higgs (1977). “Cotton, Corn, and Risk in the Nineteenth Century: Another View.” Explorations in Economic History 14(2): 167-82. Mooney, Stephen J., Justin Knox, and Alfredo Morabia (2014). “The Thompson- McFadden Commission and Joseph Golderger: Constrasting 2 Historical Investigations of Pellagra in Cotton Mill Villages in South Carolina,” American Journal of Epidemiology 180(3): 235-44.

Moore, Sophie E., Timothy J. Cole, Elizabeth M.E. Poskitt, Bakary J. Sonko, Roger G. Whitehead, Ian A. McGregor, and Andrew M. Prentice (1997). “Season of birth predicts mortality in rural Gambia.” Nature 388(6641): 434.

Moore, Sophie E., Anthony J.C. Fulford, P. Kim Streatfield, Lars Ake Persson, and Andrew M. Prentice (2004). “Comparative analysis of patterns of survival by season of birth in rural Bangladeshi and Gambian populations.” Internation Journal of Epidemiology 33(1): 137-143.

National Academy of Sciences, Institute of Medicine (1978). Strengthening U.S. Programs to Improve Health in Developing Countries. Washington, D.C.

National Research Council Food and Nutrition Board (1958). Cereal enrichment in perspective, 1958. Prepared by the Committee on Cereals. Washington, DC: National Academy of Sciences; 1958.

Niemesh, Gregory (2015). “Ironing Out Deficiencies: Evidence from the United States on the Economic Effects of .” Working paper.

North Carolina State Board of Health (1914-1918; 1922-1950). Annual Report of the Bureau of Vital Statistics of the North Carolina State Board of Health. Raleigh, NC: Edwards & Broughton Printing Company.

North Carolina State Board of Health (1920). “Vital Statistics for 1920.” The Health Bulletin 36.7.

North Carolina State Board of Health (1921). “The Vital Statistics Law.” The Health Bulletin 37.6.

31 Noymer, Andrew (2010). “The 1918 influenza pandemic affected sex differentials in mortality: Comment on Sawchuk.” American Journal of Physical Anthropology 143(4): 499 – 500.

Park, Youngmee K., Christopher T. Sempos, Curtis N. Barton, John E. Vanderveen, and Elizabeth A. Yetley (2000). “Effectiveness of in the United States: The Case of Pellagra.” American Journal Of Public Health 90: 727-738.

Park, Youngmee K., Margaret A. McDowell, Eric A. Hanson and Elizabeth A. Yetley (2001). “History of Cereal-Grain Product Fortification in the United States.” Nutrition Today 36(3): 124 – 137.

Rajakumar, Kumaravel (2000). “Pellagra in the United States: A Historical Perspective.” Southern Medical Journal 93(3): 272-277.

Rivera, Juan A., Reynaldo Martorell, Marie T. Ruel, Jean-Pierre Habicht, and Jere D. Haas (1995). “Nutritional Supplementation during the Preschool Years Influences Body Size and Composition of Guatemalan Adolescents.” The Journal of Nutrition 125: 1068S- 1077S.

Schroeder, Dirk G., Reynaldo Martorell, Juan A. Rivera, Marie T. Ruel, and Jean-Pierre Habicht (1995). “Age Difference in the Impact of Nutritional Supplementation on Growth.” The Journal of Nutrition 125: 1051S-1059S.

Siler, J.F., P.E. Garrison, and W.J. Macneal (1914). Pellagra II: Second Progress Report of the Thompson-McFadden Pellagra Commission. Chicago: American Medical Association.

Smil, Vaclav (1999). “’s : 40 years later.” BMJ 319: 1619-1621.

Smith, David F. and Jim Phillips, eds. (2000). Food, Science, Policy and Regulation in the Twentieth Century. Abingdon: Routledge.

South Carolina State Board of Health (1915-1925). The Annual Report of the State Board of Health of South Carolina. Columbia: Gonzales and Bryan, State Printers.

Steven Ruggles, Katie Genadek, Ronald Goeken, Josiah Grover, and Matthew Sobek (2015). Integrated Public Use Microdata Series: Version 6.0 [Machine-readable database]. Minneapolis: University of Minnesota.

Trienekens, Gerard (2000). “The food supply in The Netherlands during the Second World War.” In Smith and Phillips, eds.

32 United States Department of Commerce: Bureau of the Census (1900-1936). Mortality Statistics. Washington, D.C.: United States Government Printing Office.

United States Department of Commerce: Bureau of the Census (1937-1950). Vital Statistics of the United States. Washington, D.C.: United States Government Printing Office.

United States Department of Labor. Bureau of Labor Statistics. Cost of Living Division, United States Department of Agriculture. Bureau of Home Economics. Economics Division , United States National Resources Committee. Consumption Research Staff. Industrial Section, United States Central Statistical Board, and United States Works Progress Administration. Study of Consumer Purchases in the United States, 1935-1936. ICPSR08908-v3. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor], 2009-06-29. http://doi.org/10.3886/ICPSR08908.v3

Wilder, Russell M. (1956). “A Brief History of the Enrichment of Flour and Bread.” Council on Foods and Nutrition 162(17): 1539-1541.

Wood, Benjamin, Carl Nelson, Talip Kilic, and Siobhan Murray (2013). “Up in Smoke? Agricultural Commercialization, Rising Food Prices, and Stunting in Malawi.” World Bank. Policy Research Working Paper Series: 6650.

Wright, Gavin (1987). “The Economic Revolution in the American South.” The Journal of Economic Perspectives 1(1): 161-178.

Wright, Gavin (2013). Sharing the Prize: The Economics of the Civil Rights Revolution in the American South. Cambridge, MA: Harvard University Press.

Youmans, J. B. (1964). “The changing face of nutritional disease in America.” Journal of the American Medical Association, 189(9), 672-676.

33 Figures

Figure 1: Distribution of pellagra (1930) and year of fortification law passage

34 Figure 2: Pellagra and cotton acreage in the South and three southern states

Figure 3: Relationship between cotton acres and pellagra in the American South

35 G en erated for S ch m ick, E than Jam es (U n iversity of Pittsb u rg h ) on 2 0 1 5 -0 4 -2 1 2 0 :4 1 G M T / h ttp ://hd l.h an d le.n et/2 0 2 7/u iu g .3 0 1 1 2 0 1 9 2 8 1 6 14 Pu b lic D om ain , G oog le-d ig itized / h ttp ://w w w .h ath itru st.org /access_u se# p d -g oog le

Figure 4: The progression of the boll weevil

Source: Hunter and Coad (1922) The Boll weevil Problem

36 Tables

Table 1: Southern Diets 1935 - 1936

Difference Southern Non- All Southern between p-value of rural Southern residents columns (2) difference residents residents and (3) (1) (2) (3) (4) (5)

Pounds of corn meal used in past 2.81 3.67 0.07 3.6 0.00*** week

Pounds of hominy grits used in 0.62 0.59 0.02 0.57 0.00*** past week

Pounds of corn meal and hominy 3.43 4.25 0.09 4.16 0.00*** grits used in past week

Pounds of white bread used in past 2.31 1.13 4.33 -3.2 0.00*** week

Observations 1473 909 2740

Notes: Data comes from the Study of Consumer Purchases in the United States, 1935-1936 accessed on ICPSR.

37 Table 2: Relation between crop acreage and pellagra

log pellagra death rate Geographic level: State (1910-1936) Counties in NC (1915-1936) and SC (1915-1925) (1) (2) (3) (4) (5) (6) (7) (8)

Cotton acres per capita 0.0603*** 0.189*** (0.0197) (0.0669)

Corn acres per capita -0.0242** -0.160** (0.00897) (0.0651)

Peanut acres per capita -0.0136* 0.00349 (0.00703) (0.0212)

Sweet potato acres per capita -0.0390*** -0.00467 (0.0104) (0.00475)

State FE Yes Yes Yes Yes No No No No County FE No No No No Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Revenue per acre of cotton Yes Yes Yes Yes Yes Yes Yes Yes

Observations 293 293 293 293 1906 1906 1906 1906 States or counties 19 19 19 19 117 117 117 117

Notes: Standard errors, reported in parentheses, are clustered at the state-level in columns (1)-(4) and at the county level in columns (5)-(8). Columns (5)-(8) restrict to counties that harvested over 100 acres of cotton in the year 1889. * p<0.1, ** p<0.05, *** p<0.01"

38 Table 3: Relation between cotton and food crops

Sweet potato Sweet potato Corn acres per Peanut acres acres per Corn acres per Peanut acres acres per capita per capita capita capita per capita capita Geographic level: State (1900-1936) Counties in NC (1915-1936) and SC (1915-1925) (1) (2) (3) (4) (5) (6)

Cotton acres per capita -0.582*** -1.100* -0.127 -0.320*** -0.0134 0.00396 (0.178) (0.554) (0.155) (0.0685) (0.111) (0.130)

State FE Yes Yes Yes No No No County FE No No No Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes

Observations 575 575 575 1923 1923 1923 States or counties 19 19 19 118 118 118

Notes: Standard errors, reported in parentheses, are clustered at the state-level in columns (1)-(3) and at the county level in columns (4)-(6). * p<0.1, ** p<0.05, *** p<0.01"

39 Table 4: The boll weevil and pellagra

log log typhoid log measles log pellagra death rate tuberculosis death rate death rate death rate Counties in NC (1915-1950) and SC (1915- Geographic level: 1925). Column (4): only counties that Counties in NC (1915-1929) produced over 100 acres of cotton in 1889 (1) (2) (3) (4) (5) (6) (7) (8)

Post boll weevil -0.252*** -0.249*** -0.250*** -0.190*** -0.197*** 0.0321 0.0379 0.0205 (0.0532) (0.0500) (0.0535) (0.0553) (0.0601) (0.0600) (0.0397) (0.0470)

Post boll weevil * pre-boll weevil pellagra -0.287*** death rate (1915-1916 average) (0.0295)

Post boll weevil * pre-boll weevil cotton acres -0.0371 per capita (1909) (0.0401)

County FE Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Malaria Death Rate No No No No Yes Yes Yes Yes

Observations 3591 3591 3591 2931 1294 1294 1294 1294 Counties 130 130 130 118 100 100 100 100 Notes: Standard errors, reported in parentheses, are clustered at the county level. * p<0.1, ** p<0.05, *** p<0.01"

40 Table 5: Enrichment Standards proposed by FDA (1941)

Minimum (mg) Maximum (mg)

Thiamine 1.1 1.8 Riboflavin 0.7 1.6 Niacin 10 15 Iron 8 12.5 Optional Ingredients: D 150 750 Calcium 300 800

Source: Food and Bread Enrichment 1949 - 1950; National Research Council Committee on Cereals

41 Table 6: Voluntary fortification and pellagra

log pellagra death rate Geographic level: State (1910-1950) Counties in NC (1915-1950) (1) (2) (3) (4) (5) (6)

Post voluntary fortification -0.0211*** 0.00816 -0.0282** -0.184*** -0.184*** -0.203*** (0.00684) (0.00962) (0.0111) (0.0318) (0.0198) (0.0313)

Post voluntary fortification * pre-niacin pellagra -0.0576*** -0.254*** death rate (1932-1936) (0.00443) (0.0240)

Post voluntary fortification * pre-niacin cotton acres -0.0446*** -0.0570 per capita (0.0126) (0.0352)

State FE Yes Yes Yes No No No County FE No No No Yes Yes Yes Quadratic time trend Yes Yes Yes Yes Yes Yes

Observations 534 534 534 3394 3394 3394 States or counties 19 19 19 100 100 100 Notes: Standard errors, reported in parentheses, are clustered at the state-level in columns (1)-(3) and at the county level in columns (4)-(6). * p<0.1, ** p<0.05, *** p<0.01"

42 Table 7: Fortification laws and pellagra

log pellagra death rate Geographic level: State (1910-1950) Counties in NC (1915-1950) (1) (2) (3) (4) (5) (6)

Post fortification law -0.0634*** 0.00386 -0.0558** 0.137*** 0.137*** 0.120*** (0.0207) (0.0114) (0.0218) (0.0300) (0.0422) (0.0324)

-0.0552*** -0.188*** Post fortification law * pre-niacin pellagra death rate (0.00345) (0.0215)

Post fortification law * pre-niacin cotton acres per -0.0169 -0.0509* capita (0.0157) (0.0269)

State FE Yes Yes Yes No No No County FE No No No Yes Yes Yes Year FE Yes Yes Yes No No No Quadratic time tred No No No Yes Yes Yes

Observations 534 534 534 3394 3394 3394 States or counties 19 19 19 100 100 100 Notes: Standard errors, reported in parentheses, are clustered at the state-level in columns (1)-(3) and at the county level in columns (4)-(6). * p<0.1, ** p<0.05, *** p<0.01"

43 Table 8: The boll weevil and height

Height (1) (2) (3)

Post boll weevil 0.129 0.0940 0.0233 (0.0801) (0.0793) (0.0832)

0.208*** Post boll weevil * state pellagra death rate (1928) (0.0717)

0.179*** Post boll weevil * 1910 cotton acres per capita (0.0612)

County of birth FE Yes Yes Yes Year of birth FE Yes Yes Yes Year of enlistment FE Yes Yes Yes

Observations 217600 217600 217600 Counties 991 991 991 Notes: Standard errors, reported in parentheses, are clustered at the county level. * p<0.1, ** p<0.05, *** p<0.01"

44 Table 9: Fortification and income

log income (1) (2) (3)

Post enrichment law 0.0303 0.0165 0.0297 (0.0205) (0.0232) (0.0211)

Post enrichment law * pre-niacin discovery pellagra 0.0253*** death rate (0.00653)

Post enrichment law * pre-niacin discovery cotton 0.00935 acres per capita (0.00524)

Census year FE Yes Yes Yes Year of birth FE Yes Yes Yes Age FE Yes Yes Yes State of birth FE Yes Yes Yes State of residence FE Yes Yes Yes

Observations 90364 90364 90364 States of birth 11 11 11 Notes: Standard errors, reported in parentheses, are clustered at the state of birth level. * p<0.1, ** p<0.05, *** p<0.01"

45