Residential Mobility and Its Implications for Economic Mobility in the United States

No. 2 November 2015 e21 Report

Published by the Manhattan Institute Residential andEconomic WHEN MOVING America, 1880–2010 Mobility Trends in MATTERS Senior Fellow, ManhattanInstitute Scott Winship AT THE MANHATTANINSTITUTE

Executive Summary

In recent years, observers across America’s political spectrum have expressed concern over declining residential mobility and its implications for economic mobility in the United States. There is a widespread belief that Ameri- cans’ economic mobility has declined and that Americans are also less likely to “move to opportunity” than in the past. These two assertions have been linked to argue that falling residential mobility is an important factor behind diminished economic opportunity in America.

The reality is more complicated. The bulk of research on economic mobility—focused on earnings, income, occupation, and education—suggests very little change since at least the mid-twentieth century. While the share of Ameri- cans having moved in the previous year has fallen since the 1970s, this paper finds that other types of residential mobility are now as high as they have been in 100 years or more.

The author’s extensive analysis of different data sources—12 decennial censuses, extending back to the nineteenth century; the Census Bureau’s American Community Survey; and two panels from the Labor Department’s National Longitudinal Surveys—confirms a long-standing connection between residential mobility and economic outcomes; but this connection is much stronger when focusing on the kinds of residential mobility that have not declined—moves between birth and adulthood (and most likely between adolescence and adulthood) and moves across state boundaries.

While the trajectories of U.S. residential and economic mobility are thus less alarming than widely believed, there is still good reason to be concerned: though not lower than in the past, U.S. upward economic mobility remains low; and certain disadvantaged groups, including the less educated and African-Americans, are less willing, or able, to move to economic opportunity.

If the association between residential and economic mobility reflects a causal relationship—as recent research suggests—opportunity in America could be expanded through policies to promote greater residential mobility among groups with low upward economic mobility. Various policies to reform the country’s safety net, reduce housing-cost inflation, and deregulate housing and labor markets might effectively encourage migration to higher-opportunity areas.

When Moving Matters e21 Report 2 November 2015 ©Library ofCongress Cover photo:amigrantfamily fromIdabel,OklahomatravelstoCaliforniaduring theGreatDepression. CONTENTS

1 Introduction 2 I. Does Residential Mobility Promote Economic Mobility? Theory and Evidence 3 II. Earlier Research 4 III. Has Moving Become More Important for Intergenerational Economic Mobility? 5 IV. Transcending One’s Birthplace: How Economic Mobility Varies with Residential Mobility from Birth to Adulthood and over the Preceding Five Years 11 V. Transcending One’s Family Origins: How Economic Mobility Varies with Residential Mobility from Birth to Adolescence and from Adolescence to Adulthood 17 VI. Residential Mobility and Adult Outcomes, 1880–2010 27 VII. Changes in the Extent of Residential Mobility and Group Differences 39 Conclusion 43 Endnotes

When Moving Matters e21 Report 2 November 2015 urban studiesfromUniversityandaPh.D.insocialpolicy Northwestern Harvard University. hastestifiedbefore CongressWinship onpoverty, inequality, andjoblessness.HeholdsaB.A.insociology ReformsforaLimitedGovernmentandThrivingMiddleClass(2014). to theeBookRoomGrow:Conservative Quarterly,andBreakthroughJournal ; andhecontributedanessayonantipovertypolicy National Review,TheWilson Third Way. writes a column for Forbes.com; his research Winship has been published in he wasresearch manageroftheEconomicMobilityProject ofthePewCharitableTrusts andaseniorpolicyadvisorat Institution. Winship’s research interests includelivingstandards andeconomicmobility, inequality, andinsecurity. Earlier, SCOTT WINSHIP A The authorthankstheManhattanInstitute’s Preston Cooperforhisresearch assistance. A bout cknowledgment

the A is theWalter B.Wriston FellowattheManhattanInstitute. Previously, hewasafellowattheBrookings uthor City Journal, National Affairs, When Moving Matters Residential and Economic Mobility Trends in America, 1880–2010 Scott Winship INTRODUCTION

pward economic mobility has traditionally been among America’s core ideals. But in recent years, policymakers and analysts on both the left and on the right have grown concerned about the health of the American dream. Chil- Udren in the United States who start in the bottom of the income distribution are unlikely to achieve a solidly middle-class adulthood. Just 13 percent of today’s adults who were raised in the poorest fifth of families as children made it to the top two-fifths.1 Fully 43 percent remained in the bottom fifth as adults, and 70 percent were in the bottom two- fifths. While the poverty rate in America has declined significantly over the past half-century, public policy has failed to move the needle on upward economic mobility.2

One factor that may be related to economic mobility is residential mobility. Economic growth and job options vary with geography. Thus, people who are willing and able to relocate may be more likely to earn more money. Moving to better opportunities has been a prominent feature of American history. Westward expansion, for example, was one of the most important developments of the nineteenth century. During the first half of the twentieth century, mil- lions of African-Americans moved from the South to seek better lives in northern cities. During the 1930s, hundreds of thousands of farmers left the Dust Bowl.

When Moving Matters 1 2 e21 Report 2 November 2014 economic outcomes. The paperthenexplores how adulthood—is associated with arangeofbetter stateandbetweenticularly toanew childhoodand suggests thatbeingwillingandabletomove—par- mobility promotes economicmobility. The evidence The paperbeginsby assessingwhetherresidential search ofbetter economicopportunities? for publicpoliciestoencouragepeoplemove in economic mobility? What doestheevidencesuggest mobility andinitsprevalence—interacted toaffect have these trends—in ofresidential theimportance dential mobility declined? If so, among whom? How moving beenhistorically?Has resi toopportunity - ty mayaffecteconomicmobility. How has important and contemporary, related tohow residential mobili- This paperexplores anumberofquestions,historical lingly reinforced thatplacereally doesmatter. omist RajChettyandhiscolleagueshave compel- has beenmixed, butrecent papersby Harvard econ- neighborhood effectsonchildandadultoutcomes mixed-income projects. The evidenceforso-called hoods torental vouchers anddispersedlow-rise and from densehigh-riseprojects withinpoorneighbor- ing subsidieshave shiftedinrecent decadesaway been considered disadvantageous, and federal hous- larly severe among African-Americans, has long economic mobility. Concentratedpoverty,- particu move hasbeenthoughttobeofconsequencefor Even withincities,thewillingnessandabilityto higher crime,andslower jobgrowth. share oflower-skilled anddisadvantaged residents, found themselves facinganeroding taxbase,arising who were unableorunwillingtomove from thecity tion ofpeopleandjobswithinmetro areas. Those more living space, caused a dramaticdecentraliza- in automobile ownership andpent-up demand for policy,portation combinedwithexplosive growth from citytosuburb. Postwar housingandtrans- involves moves withinmetropolitan areas, especially forexpandingopportunity historically important Another formofgeographicmobilitythathasbeen detrimental normsandweakened localinstitutions. and culturalcapital,values, andaspirationsthrough tage mightaffectchildeducationaloutcomes,social has theorized waysinwhichconcentrateddisadvan- children are exposed.Arichsociologicalliterature vironmental, and institutional conditions to which Residential mobility canalsoaffectthesocial,en- closertopublictransportation. grab apartments in thesuburbs. Working-class familiesbreak leasesto the citysothattheirkidscanattendbetterschools ing. Codersflocktotechnologyhubs. Parents leave thatare boom- towns ofthecountry forotherparts ry der. Blue-collar workers leave behinddecliningfacto- up stakestomove upthecorporateoracademiclad- higher-growth areas. Professionals andmanagerspull attendcollege,orsetdown rootsarmed services, in opportunities. Young adultsleave hometojointhe When Americansmove, theyoftendosoforbetter THEORY ANDEVIDENCE PROMOTE ECONOMICMOBILITY? I. DOESRESIDENTIALMOBILITY es withadiscussionofpolicyimplications. among more disadvantaged groups. The paperclos- pecially low today;butresidential mobilityislower that appeartomatterforadultoutcomesare notes- to conventional wisdom,thetypesofmobility trary it varies across different demographicgroups. Con- residential mobilityhaschangedover timeandhow can hurt outcomes. Residential stability may promote outcomes.Residentialcan hurt stabilitymaypromote Of course,itisalsopossiblethattoomuch movement reasonably expected to improve their own outcomes. outcomes, andmigrationby grown children canbe ity. Parental moves mighthelpchild to opportunity ity shouldcorrespond withgreater economicmobil- It makes sense, then, that greater residential mobil- opment ofchildren inlow-income neighborhoods. physical andemotionalhealthcognitive devel- es andvermin, microbes, andviolence affectsthe exposure toleadpaintandothertoxins, cockroach- A nascentpublic-healthliterature isfocusedonhow 4 3

good psychological health while frequent moves could to make progress answering questions of cause and ef- be disruptive to children’s education and social devel- fect without first understanding the underlying facts. opment. More generally, if moves often occur between one disadvantaged locale and another, we would not II. EARLIER RESEARCH expect to see improvement in child outcomes. Some research has provided strong evidence from ran- Some researchers have suggested that rising economic domized controlled trials that where a child grows up inequality and differences in the cost of living have matters. The vast majority of research is focused on reduced the extent to which residential mobility can neighborhoods.7 The literature assessing lotteries for improve economic mobility. A higher cost of living school assignment (often involving charter schools) would be expected to lead to higher incomes for consistently shows sizable long-term effects on aca- poor and middle-class families that move. However, demic performance and adult educational attainment researchers have hypothesized that rising inequality from attending higher-performing schools.8 in many cities has driven a wedge between the cost of living—especially the cost of housing—and me- Because school attendance is generally determined dian incomes, partly because of restrictions on the by residential proximity rather than lotteries, it supply of housing and partly because of inefficien- follows that living in a neighborhood with lower- cies in labor markets.5 If, for instance, zoning regu- performing schools has a negative impact on edu- lations prevent the supply of housing from keeping cational outcomes, and therefore residential mobil- up with rising demand in areas with labor shortages, ity between neighborhoods (or larger geographies) then low- to moderate-income migrants to booming would improve those metrics. At least two studies areas like Silicon Valley may face unaffordable hous- outside the U.S. have established that refugees as- ing costs.6 To the extent that such dynamics operate, signed (without their input) to neighborhoods with residential mobility may be less obviously related to better schools or better-educated residents had bet- upward economic mobility than we might think. ter educational outcomes than those assigned to less advantageous neighborhoods.9 Assembling compelling evidence on this question is a formidable challenge. The basic problem is easily Other research is based on policy experiments— grasped. People who move might do better or worse usually not fully randomized—that move low- than those who do not move because moving tends income families to more advantaged neigh- to be helpful or harmful. But they might simply borhoods.10 For instance, evidence from the have strengths or weaknesses that would lead them “Gautreaux program”—a court-ordered effort to do better or worse than non-movers whether or in Chicago to place segregated public-housing not they move. A second problem is that the impact residents in predominantly nonblack neighbor- of residential mobility surely varies depending on a hoods—suggests that living in neighborhoods host of particulars, not the least of which is where with more nonpoor or nonblack residents im- someone is moving from and moving to. proved mothers’ employment and earnings and lessened their reliance on welfare. Living in a This paper will not surmount these methodological suburban neighborhood instead of one in the city problems and should be considered as a primarily reduced drug arrests among sons and appears to descriptive document. Finding or not finding that have improved educational and employment out- residential mobility goes hand in hand with better comes and earnings in young adulthood.11 Hav- economic outcomes is suggestive of causality or its ing more educated neighbors and more employed absence—but only suggestive. However, it is difficult as professionals reduced early male mortality.12

When Moving Matters 3 4 e21 Report 2 November 2014 as adults. the children inMTO familiestotheirtaxrecords thaniel Hendren, andLawrence Katz,wholinked on itsheadbystudyfrom anew RajChetty, Na- This disappointingconclusionhasbeenturned relatively unimportant. vantages remain; theneighborhooditselfmaybe them outofpoorneighborhoods,andthedisad- and familieslivinginthoseneighborhoods.Move disadvantageous characteristicsoftheindividuals fect” ofconcentratedpoverty simplyreflects the suggested insteadthatmuchoftheapparent “ef- States. ity across labormarkets throughout the United documented largedifferences ineconomicmobil- localities. Using tax-return data,histeaminitially valuable research on the benefits of moving between Chetty andhiscolleagueshave alsoprovided in- anything, didworsethanthecontrol group. dren whowere olderwhentheirfamiliesmoved, if also attended better colleges. At the same time, chil- rose while theirratesofsingleparenthood fell. They nearly one-third, and theircollegeattendance rates to receive avoucher. Their incomeswere higherby mid-twenties than children randomly assigned not before theyturned13didsignificantlybetterintheir voucher tomove to a higher-income neighborhood found thatchildren whosefamilieshadusedtheir ter neighborhoodswhenyoung—the Chettypaper studies found—even forchildren whomoved tobet- tional outcomesinadolescencethatprevious MTO economic mobilitythatwould makeaDane proud. ic andeducationaloutcomes. from moving—particularly inregard toeconom- that hadpreceded it,MTO foundlimitedbenefits much of the (mostly nonexperimental) evidence toaffected adultandchildoutcomes.Contrary lesspoorneighborhood inasomewhat apartment er requiring public-housing residents torent an (MTO)Opportunity project, considered wheth- icy experiments,thefullyrandomized Moving to andambitiousofthesepol- The mostimportant 15 Some areas, suchasSalt LakeCity, featured 14 Despite theabsenceofimproved educa- 13 MTO’s findings least beneficialforchildren. termined whichspecificcounties were themostand affect economicmobility. The Chettyteamalsode- theseresultsshort, suggestthatmoving persedoes counties didbetterorworsethanoneanother. In amounts of timein high- or low-economic-mobility age.Evena certain siblingsexperiencingdifferent counties, themore theiroutcomesimproved, upto rates. The more timethatchildren spentinthese - come, college attendance, marriage,andteenbirth to be trackedfrom their childhood homes to their Few datasetsexistthat allow thesameindividuals sesses outcomesforvarious demographicgroups. tween different typesofresidential mobilityandas- moving andeconomicoutcomes.It distinguishesbe- historicalevidenceabouttherelationshipnew between residential mobility, past and present, by providing This paperseekstoaddourknowledge regarding ECONOMIC MOBILITY? IMPORTANT FOR INTERGENERATIONAL III. HASMOVING BECOMEMORE do betterthemselves. lies move tocountieswithbetterchildoutcomes last concernby showing thatchildren whosefami- The mostrecent paperfrom Chettyaddresses this moving places)whileothersstayput. (andtocertain on their own attributes and preferences, with some intoneighborhoodsbased people mightsimplysort colleagues examined.Even more problematically, behind themeasurablefactorsthatChettyandhis ences inthepopulationsofdifferent labormarkets connections. There couldbeunderlyingdiffer- it appropriately urgedcautioninmakingcausal markets are correlated with economic mobility, but number ofeconomicandsocialfeatures oflabor nomic outcomes? The Chettyteamshowed thata Do thesedifferent placescausebetterorworseeco- tochildrenportunities raisedthere. Others, likeAtlanta, appeartoofferfarfewerop- 16 foradultin- This wastrue adult residences. Nevertheless, many sources of data households from previous census years and are not include information on adult outcomes at a point asked about parental circumstances.18 One clever in time and also ask where participants have lived in way around this problem was developed by Daniel the past or where their parents are from. Aaronson and Bhashkar Mazumder in a widely cit- ed paper looking at trends in economic mobility.19 To look at whether residential mobility affects economic outcomes, this paper assembles several types of Aaronson and Mazumder use various censuses to evidence. The first involves the use of decennial census impute family incomes to adult men, taking advan- data and the American Community Survey—point- tage of the fact that birth states and birth years are in-time data sets that will be collectively referred to as supplied in each data set. From earlier censuses, they the “Census Bureau data.”17 These analyses compare estimate the average income of families that had a the economic outcomes of adults with the income lev- son born in a given five-year window in a given state els typical of their birth state when they were children. and who continued to live in the state. They attach They consider whether the relationship to conditions that average family income to adults in subsequent in one’s birthplace is weaker among adults who change censuses born in the same state and during the same residences than it is among those who do not. five-year window. Finally, they estimate intergenerational economic mobility measures, relating “family The paper also turns to two surveys from the Bu- income” to the adult earnings of men.20 reau of Labor Statistics National Longitudinal Study, each of which allows individual children to Importantly, the Aaronson-Mazumder approach be tracked into adulthood. These two surveys allow differs from the conventional economic mobility true economic mobility estimates to be computed metric tied to real parental incomes. The vast major- for recent cohorts of men and women in their late ity of interfamily income inequality occurs within twenties and to see whether movers have more eco- birth states and cohorts. For that reason, this paper nomic mobility than non-movers. will characterize the indicator of child circumstances as “birthplace income” rather than “parental in- Finally, the paper presents long-run trends in a come.” The measure reflects the importance of the number of adult economic outcomes and how they economic conditions in one’s birth state when one have varied depending on adults’ moving histories. was a child (or of other factors correlated with these These outcomes do not constitute economic mobil- conditions), and only minimally does it reflect cir- ity measures per se because they are not conditioned cumstances specific to one’s family. on childhood circumstances, but they allow for a look back to the nineteenth century. Trends in “economic mobility” using the Aaronson- Mazumder measure need not reflect trends con- IV. TRANSCENDING ONE’S ventionally measured that relate adult incomes to BIRTHPLACE: HOW ECONOMIC actual parental incomes. Birthplace conditions can MOBILITY VARIES WITH RESIDENTIAL become more important without family conditions MOBILITY FROM BIRTH TO increasing in salience because family conditions vary ADULTHOOD AND OVER THE greatly within birth states and birth cohorts.21 PRECEDING FIVE YEARS The Rank-Rank Slope The limitations of the Census Bureau data sets Even with real information about parental in- preclude true intergenerational mobility analyses come, the metric used by Aaronson and Mazum- because adults cannot be linked to their childhood der suffers from several problems as a measure of

When Moving Matters 5 6 e21 Report 2 November 2014 within each year (birth cohort). within eachyear (birth ings andfamilyincomesare topercentiles converted 30years later.to 1980censusesbutobserved Earn- of years as thefive groups ofchildren from the1940 groups of adults, aged 30–44, born in the same set from 1936 to1950,andsoon. That creates five between 1926 and1940,thosein1980were born adults aged30–44,sothatthosein1970were born vey. Within eachyear, thesamplesare restricted to censuses andthe2010AmericanCommunitySur- The nextstepistoswitchthe1970through 2000 across allfamilieswithchildren uptoage14. comes ineachyear are topercentiles converted 1940 through 1980are firstused,andfamilyin- In theanalysesbelow, thedecennialcensusesfrom incomepercentile. birthplace outatany expected adultincomeofachildstarting place, andtogethertheycanbeusedtocomputethe pected adultincomeofachildinthepoorest- birth the slopeisestimatedsimultaneouslywithex- come percentiles change.Using astatisticalmodel, income percentiles increase in- orfallasbirthplace assessed. The rank-rankslopeindicateshow adult a given incomelevel, before theirrelationship is the percentage ofpeoplewhofallabove orbelow are intopercentile bothconverted ranks,indicating Essentially, incomes and adult incomes birthplace economic mobility. “rank-rank slope.” problems istoestimateanotherstatisticcalledthe each birth state. each birth the average incomepercentile iscomputedwithin Within eachyear (or, equivalently, cohort), birth dren bornintheyears 1936to1950,andsoforth. and 1940,1950percentiles are computedforchil- tiles are computedforchildren bornbetween 1926 are livingindependently.) That is,in1940,percen- few stateandwhileensuringthatvery their birth dence when they show up in the data differs from dren whilereducing thenumberwhosestateofresi- (Stopping atthatageensures asizablegroup ofchil- 26 23 22 A solution to most of these 24 27 25

the 25 birthplace incomewasatthe25 birthplace percentile iscomputedforeachyear foranadultwhose birthplace average familyincometomen’sbirthplace earnings. Figure 1 displaysthefirstsetof results, which relate Adult Earnings, 1970–2010 examining otheroutcomes. analyses includessupplementary r2_appendix.pdf) this paper(www.manhattan-institute.org/pdf/e21_ between movers andnon-movers. The appendixto betweenincome differs adult income and birthplace Mazumder paperby lookingathow therelationship These analysesbuildontheAaronson and ofmoving.assess theimportance is estimatedseparatelyformovers andnon-movers to hasthemedianincome.Economic mobility birthplace the 1 is atthe50 above the99 than 75percent ofthem. are The richestbirthplaces that ishigherthanjust25percent ofadultsandlower during the1970sbutwasthen basicallyflat. words, economic mobilitymeasured inthisway rose fell through 2000,and endedin2010at42.In other two. andtotheinteractionbetween the cohort) year (birth statearound thetimetheywerebirth children, tothe relates adults’ earningstotheaverage incomeintheir A simplelinearregression modelisestimatedthat er andbornin Virginia. across children in1970whowere aged14oryoung- in Virginia are given theaverage incomepercentile other words, allmenaged30–44in2000andborn 30yearsfrom earlier. cohort their15-year birth In percentile amongchildren state bornintheirbirth Adults are thengiven theaverage familyincome average family incomeatthe25 dicates thatanadultwhowasbornintoastatewith Among men generally, the value of 35 in 1970 in- the 35 earlier typicallyhadadultearningsthatputhimat 28 st From themodelestimates,expectedincome . inthemiddleofdistribution The birthplace th th percentile means having birthplace income income percentile meanshavingbirthplace percentile. That value rose to42in1980, th th percentile, andthepoorest are below percentile, and an adult from that percentile, andanadultfrom that th th percentile. Being at percentile. Being at percentile 30years 30 29

Figure 1. Expected Earnings Percentile for Man The primary questions of interest are with Low Birthplace Income, by Residential whether economic mobility rates are Mobility Since Birth, 1970–2010 higher for adults who move from their birth state and whether the importance 60 of residential mobility has grown over 54 time. Figure 1 shows that economic 48 49 48 50 mobility was significantly higher 41 42 42 40 among men who lived outside their 40 38 35 37 35 35 birth state than it was for those still 33 30 living there. In 1970, a man still living 27 in his birth state who had been at the th 20 25 percentile of birthplace income was typically at the 27th percentile 10 of male earnings. But if he lived in a Family Income Was at the 25th Percentile Family Income Was different state, he was typically at the

Expected Adult Family Income Percentile if Birthplace 0 41st percentile. 1960 1970 1980 1990 2000 2010 2020

All Men Moved from Birth State Did Not Move from Birth State The economic mobility trends for movers and non-movers were similar Source: Author’s calculations based on data from decennial censuses and 2010 American Community Survey, Integrated Public Use Microdata Series from 1970 to 2000, though the advantage of movers widened, especially after 2000. By 2010, the movers were Figure 2. Expected Earnings Percentile for Woman at the 54th percentile of adult earnings, with Low Birthplace Income, by Residential compared with the 35th percentile Mobility Since Birth, 1970–2010 for those who stayed put. Economic mobility among non-movers was lower 60 in 2010 than it had been in 1980. 52 52 51 50 48 49 50 46 45 43 In Figure 2, which analyzes women’s 42 41 41 39 40 37 earnings, the striking feature is the 35 steady decline in economic mobility 30 from 1980 to 2000. In 1980, a woman raised at the 25th percentile 20 of birthplace income was typically at the 48th percentile of female 10

Family Income Was at the 25th Percentile Family Income Was earnings. By 2000, such a woman was at the 41st percentile. Upward

Expected Adult Family Income Percentile if Birthplace 0 1960 1970 1980 1990 2000 2010 2020 economic mobility remained at that level in 2010, the first time that it All Women Moved from Birth State Did Not Move from Birth State was lower than for men.31 Figure 1

Source: Author’s calculations based on data from decennial censuses and 2010 American and Figure 2 show similar patterns Community Survey, Integrated Public Use Microdata Series in that movers do much better than non-movers, and the gap between them widens over time.

When Moving Matters 7 8 e21 Report 2 November 2014 5, formen,economicmobility falls incomes ratherthanearnings. In Figure incomeswithadult family birthplace Figures 5through 8switchtocomparing Adult Family Income, 1970–2010 and 2000. with arecent out-of-statemove in 1990 mobility rates thanmen, except for men women generallyhave highereconomic most sense economically. At any rate, earning husbandsdoingwhatmakesthe could beassimplecoupleswithhigher- husbands’ ambitionsover wives’, orit some familiescontinuetoaccommodate husbands ratherthanwives. Perhaps tobenefit age tenddisproportionately possibility isthatmoves inornearmiddle of upward economicmobility. One of womenhadpracticallyidenticalrates shown inFigure 4. The three categories to conferanyadvantage towomen,as Interestingly, recent moves donotappear oftheanalysestofollow.)true previous five years. This is also generally five years, rather thanmoves inthe driven by moves thatpreceded thepast mobility trends inFigure 1are primarily notshown, 1980. (While theeconomic saw declinesineconomicmobilityafter better thannon-movers, andbothgroups stayed in-stategenerallyfared onlyalittle moved intheprevious five years but upward economicmobility. Men who 30 years, withtheformerseeingmore other menwasfairlyconstantover the of state in thepreceding five years and The gapbetween menwhomoved out rather than birth-to-adulthood moves.rather thanbirth-to-adulthood of moves inthepreceding five years this time,itconsidersdifferent kinds state,but family incomeintheirbirth Figure 3compares men’s earningsto 32

Community Survey, IntegratedPublicUse Microdata Series Source: Author’s calculationsbasedondatafrom decennialcensusesand2010American Community Survey, IntegratedPublicUseMicrodata Series Source: Author’s calculationsbasedondatafrom decennialcensusesand2010 American

Expected Adult Family Income Percentile if Birthplace Expected Adult Family Income Percentile if Birthplace Figure 4.ExpectedEarningsPercentile forWoman Mobility over thePrevious Five Years, 1970–2010 Family Income Was at the 25th Percentile Family Income Was at the 25th Percentile Mobility over thePrevious Five Years, 1970–2010 Figure 3.ExpectedEarningsPercentile forMan 1 2 3 4 5 6 1 2 3 4 5 6 with LowBirthplaceIncome,byResidential with LowBirthplaceIncome,byResidential 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 9 9 6 6 0 0 Moved toaDifferent State All Women Moved toaDifferent State All Men 4 1 3 3 3 4 1 6 9 5 3 4 2 9 4 7 7 6 4 0 0 7 1 1 4 9 4 9 4 4 3 4 4 8 8 9 8 5 2 9 2 7 0 0 1 1 9 9 4 3 4 9 9 0 9 7 4 0 0 4 2 3 Did NotMovefrom Residence theSameState Moved Within Did NotMovefrom Residence theSameState Moved Within 2 2 0 4 0 3 3 4 0 4 0 6 9 7 0 0 3 4 8 1 2 2 0 4 0 4 1 1 1 2 0 0 2 2 0 0 2 2 0 0 Figure 5. Expected Earnings Percentile for Man between 1980 and 2000 but remains with Low Birthplace Income, by Residential higher in 2010 than in 1970. However, Mobility over the Previous Five Years, 1970–2010 the fall is confined to men who remain in their state of birth. Movers saw little 60 change after 1980 and had higher 51 economic mobility in 2010 than ever, 50 47 47 46 while non-movers’ expected incomes 41 41 38 fell steadily, even from 2000 to 2010. 40 37 38 35 37 33 The gap between movers and non- 32 30 movers doubled between 1980 and 30 28 2010. Strikingly, a man who was born th 20 to the 25 percentile of birthplace income typically ended up at just the th 10 30 percentile of family income if he Family Income Was at the 25th Percentile Family Income Was remained in his state, experiencing very

Expected Adult Family Income Percentile if Birthplace 0 little upward economic mobility. 1960 1970 1980 1990 2000 2010 2020

All Men Moved from Birth State Did Not Move from Birth State Nowhere is the widening economic mobility gap between movers and non- Source: Author’s calculations based on data from decennial censuses and 2010 American Community Survey, Integrated Public Use Microdata Series movers more apparent than in Figure 6, which compares birthplace incomes with women’s family incomes in adulthood. In Figure 6. Expected Earnings Percentile for Woman 1970, a woman from the 25th percentile with Low Birthplace Income, by Residential of birthplace income typically was at the Mobility over the Previous Five Years, 1970–2010 30th percentile of family income if she still lived in her birth state, compared 60 with the 39th percentile if she had 52 moved. By 2010, the expected incomes 50 47 47 45 were at the 28th and 52nd percentiles. 41 39 39 40 38 37 36 Non-movers experienced very little 36 34 32 economic mobility at all by 2010; where 30 30 28 an expected adult income percentile of 25 would indicate complete economic 20 immobility, the 2010 estimate is 28.

Family Income Was at the 25th Percentile Family Income Was The trends in short-run residential mobility in Figure 7 resemble those for Expected Adult Family Income Percentile if Birthplace 0 1960 1970 1980 1990 2000 2010 2020 male earnings in Figure 3, except that the gaps between men who moved out of All Women Moved from Birth State Did Not Move from Birth State state in the past five years and other men Source: Author’s calculations based on data from decennial censuses and 2010 American are smaller this time. Upward economic Community Survey, Integrated Public Use Microdata Series mobility for men who remained in the same state as five years earlier was practically the same in 2000 as in 1970.

When Moving Matters 9 10 e21 Report 2 November 2014 of jobs,educationalattainment, and amining the educational requirements presents additionalanalysesex pdf) institute.org/pdf/e21_r2_appendix. The appendix(www.manhattan- themselves—is striking. nothing ofthemagnitudegaps But theconsistencyofdata—tosay they already lined up.have opportunities move, andare successful,move because be lessbeneficial—orperhapsthosewho they wouldmove toplacesthatwould better even iftheydidmove. Perhaps those whodon’t move wouldfare no whether ornottheymoved—or that well compared with those who don’t, move would fromdo poor birthplaces outcomes. It couldbethatthosewho mobility leadstobettereconomic to saywithconfidencethat residential From theseanalyses,itisnotpossible one’s state confers little to no advantage. caveat thatmovingimportant within comes towomen’s earnings—withthe the previous five years, except when it forresidentialalso true mobilityover to-adulthood residential mobility. It is years. forbirth- This isgenerallytrue widened considerably over thepast35 those whodonotmove, andthegaphas have greater economicmobilitythan In sum,menandwomenwhomove in 2010than1970. economic mobilitysignificantlyhigher Only among out-of-state movers was else. after 1980butfellforeveryone mobility wasflatforout-of-statemovers state thaniftheymove in-state.Economic movesshort-term iftheymove outof see abiggerfamilyincomeboostfrom Figure 8suggeststhat,likemen,women Community Survey, IntegratedPublicUse Microdata Series Source: Author’s calculationsbasedondatafrom decennialcensusesand2010American Community Survey, IntegratedPublicUseMicrodata Series Source: Author’s calculationsbasedondatafrom decennialcensusesand2010 American

Expected Adult Family Income Percentile if Birthplace Expected Adult Family Income Percentile if Birthplace Figure 8.ExpectedEarningsPercentile forWoman Mobility over thePrevious Five Years, 1970–2010 Family Income Was at the 25th Percentile Family Income Was at the 25th Percentile Mobility over thePrevious Five Years, 1970–2010 Figure 7.ExpectedEarningsPercentile forMan 1 2 3 4 5 6 1 2 3 4 5 6 with LowBirthplaceIncome,byResidential with LowBirthplaceIncome,byResidential 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 9 9 6 6 0 0 Moved toaDifferent State All Men Moved toaDifferent State All Women 3 1 1 6 9 9 3 3 3 3 4 7 7 5 9 4 7 0 0 0 1 1 9 9 4 4 4 4 4 4 8 8 1 1 3 4 1 0 0 0 1 1 9 9 3 3 4 3 3 4 4 3 9 9 8 9 2 9 6 1 4 5 0 0 Did NotMovefrom Residence theSameState Moved Within Did NotMovefrom Residence the SameState Moved Within 3 2 2 6 0 0 4 4 3 3 3 3 0 0 4 3 7 5 6 7 0 0 2 2 0 0 3 3 1 1 8 6 0 0 2 2 0 0 2 2 0 0 employment. To summarize the evidence relating within-state moves, and it is not possible to estimate birthplace income to adult outcomes—whether movers true intergenerational economic mobility estimates did better or worse than non-movers in 1970 on some from them. To surmount these shortcomings, this indicator—trends were generally more favorable for section presents results from the National Longitu- those who lived outside their birth state than for those dinal Surveys of the Bureau of Labor Statistics (BLS). who remained in it. In 1970, movers already tended to have higher earnings and incomes than non-movers. BLS has fielded a number of surveys over the years There were only small differences—if any—between following nationally representative groups of adoles- movers and non-movers in terms of the educational cents into adulthood and tracking their economic requirements of jobs, high school graduation and and social outcomes. This section relies on two of college graduation rates, and employment rates. those surveys. The first, the National Longitudinal Movers actually had higher unemployment rates than Survey of Youth 1979 (NLSY79), began in 1979 non-movers. By 2010, however, existing advantages with a group of adolescents aged 14–21. An entirely among movers had grown, gaps had opened up where separate survey, the National Longitudinal Survey of none existed, and higher unemployment among Youth 1997 (NLSY97), was initiated in 1997 with movers had disappeared. a group aged 12–17. In both surveys, adolescents who were initially aged 14–17 may be followed Short-term residential mobility—moves over the for 14 years, until they are aged 28–31 (in 1993 or preceding five years—tended to correlate with better 2011). In each survey wave, participants report their outcomes in 1970, at least if adults moved out of earnings and family income from the previous year, state; in-state moves were generally not that beneficial. so the results compare parental income in 1978 or The gap between out-of-state movers and other adults 1996 (at ages 13–16) with adults’ own income in remained relatively stable over time. Men who moved 1992 or 2010 (at ages 27–30).33 out of state in the preceding five years tended to have employment levels that were as low as or lower than Relative Intergenerational Economic Mobility, those of other adults in 1970 and unemployment rates 1992 and 2010 that were higher than those of other adults. By 2010, A useful distinction in thinking about intergen- however, outcomes were more favorable among out-of- erational economic mobility is between relative and state movers than among other adults. An exception to absolute mobility. Relative mobility is about the ex- all these trends is that in terms of earnings and family tent to which children end up in the same economic income, adult women who had moved or not moved position as their parents. Ignoring whether children recently did about equally well in every year. are richer or poorer than their parents and by how much, relative mobility is concerned only with in- V. TRANSCENDING ONE’S FAMILY terpersonal rankings. If someone is raised by parents ORIGINS: HOW ECONOMIC MOBILITY who are poorer than 80 percent of their peers, what VARIES WITH RESIDENTIAL MOBILITY are the chances that the child will end up richer than FROM BIRTH TO ADOLESCENCE AND 80 percent of the child’s peers? How many children FROM ADOLESCENCE TO ADULTHOOD start in the bottom quarter of family income and end up there themselves as adults? How many starting in The Census Bureau data sets have two important the richest quarter end up in the bottom 75 percent? strengths: they are very large, allowing for precise estimates to be computed; and they provide information In contrast, rankings are irrelevant for assessing back to the nineteenth century, as discussed below. absolute mobility. The son of parents at the 40th However, they include only limited information for percentile might end up at the 40th percentile

When Moving Matters 11 12 e21 Report 2 November 2014 on average, at the 43 Bureau, itfellfrom 39to36(Figure 6). BLS datarose from 42to43;intheCensus expected adultincomepercentile in the 2010 (Figure 5).Amongwomen,the Bureau datashow 38 forboth1990and and 44in2010(notshown). The Census income percentile was43formenin1992 period. In theBLSdata,expectedadult with theCensusBureau dataforthesame upward economic mobility isconsistent This findingofessentiallynochangein percentile. sex. none oftheresults are disaggregated by increase thesamplesizes intheseanalyses, on relative economicmobility. In order to Figure 9provides thefirstsetofBLS results isgettingpoorerif everyone over time. tomakeitthetopislessmeaningful opportunity jobs have children fillingtheworstjobs years later; standards are lesssatisfyingifparents with theworst of economicmobilityare Risingliving important. poorer—between generations.Bothdimensions because the nation can grow richer—ortemporarily despite seeingnorelative mobility. This ispossible will have experiencedabsoluteupward mobility off ininflation-adjustedtermsthanhisparents. He himself, buthemightalsobesignificantlybetter the 25 adults whosechildhoodincomewasat or average, familyincomepercentile of in 1992were, onaverage, atthe42 family income, while their counterparts 25 adults aged27–30whohadbeenatthe unchanged duringthisperiod.In 2010, upward mobilityactuallymayhave been enough, given thesamplesizes, that over time. Yet thisincrease issmall who hadbeenpoorincreased slightly upward economic mobility among those th 34 percentile ofchildhoodincomewere, The measure here isthe“expected,” th percentile. 35 rd Among all adults, Amongalladults, percentile of adult percentile of adult 36 nd

and NationalLongitudinalSurveyofYouth 1997 Source: Author’s calculationsbasedondatafrom NationalLongitudinalSurveyof Youth 1979 and NationalLongitudinalSurveyofYouth 1997 Source: Author’s calculationsbasedondatafrom NationalLongitudinalSurveyofYouth 1979

Figure 10.ExpectedFamily IncomePercentile forAdults Expected Adult Family Income Expected Adult Family Income Figure 9.ExpectedFamily IncomePercentile forAdult Percentile if Childhood Family Income Percentile if Childhood Family Income with LowChildhoodIncome,byResidentialMobility Was at the 25th Percentile Was at the 25th Percentile Between BirthandAdolescence, 1992and2010 10 15 20 25 30 35 40 45 50 10 15 20 25 30 35 40 45 50 0 5 0 5 Mobility SinceAdolescence,1992and2010 with LowChildhoodIncome,byResidential is residential immobilitythatappearsbeneficial. went on to have higher upward mobility. In this case, it residence whentheywere asattheirbirth adolescents both 1992and 2010, adultswhoremained inthesame not affectupward mobilityinaclearway, except that,in from one’s andfarther farther residence birthplace does moves andadolescence.Living occurringbetween birth simple relationship between economicmobilityand the start ofthesurveys. the start adolescents lived whentheywere born,aswell asat provide The BLSsurveys informationonwhere 42 42 All All 43 43 in '93as'79 Same Tract Same Residence 37 at Birth 44 43 47 County 1992 42 1992 Same Same State 41 37 at Birth 39 Figure 9doesnotreveal a 2010 2010 43 43 Same State 45 Same Region at Birth 42 Region 44 45 Same 49 Different Region at Birth 43 Different Region 48 40 46 In Figure 10, the BLS data are broken out by different within or outside the region in which they lived as types of residential mobility between adolescence adolescents had the most upward mobility.38 and adulthood. This time, in both 1992 and 2010, upward economic mobility tends to rise with greater Another indicator of relative upward mobility is distance from one’s adolescent origins. Adults who the percentage of adults who were raised in the remained in the same census tract or county as in poorest quarter of adolescents remaining in the adolescence experienced the lowest upward mobility. bottom quarter of income in adulthood. Figure (A census tract typically comprises a few thousand 11 and Figure 12 provide the results. Once again, people.) Those who had moved to a different state there is little obvious relationship between birth- to-adolescence residential mobility Figure 11. Percentage of Adults Raised in the Bottom and upward mobility in Figure 11. Quarter of Family Income Who Remain in the Remaining in the same residence as at Bottom Quarter in Adulthood, by Residential Mobility birth appears to correspond to higher Between Birth and Adolescence, 1992 and 2010 upward mobility (fewer adults remaining in the poorest quarter). It appears that 60% 55% 52% upward mobility improved strongly 50% 47% 46% 44%44% over time among adults who had moved 39%40% 39% 40% 35% to a different state in the same region 30% between birth and adolescence, but the

20% sample of such adults is small enough

10% that the drop shown in Figure 11 is consistent with a true decline that was 0% All Same Residence Same State Same Region Different Region much smaller (or even nonexistent). at Birth at Birth at Birth at Birth 1992 2010 As before, adolescence-to-adulthood Source: Author’s calculations based on data from National Longitudinal Survey of Youth 1979 and National Longitudinal Survey of Youth 1997 residential mobility appears more strongly related to upward economic mobility (Figure 12). Among adults who lived in Figure 12. Percentage of Adults Raised in the Bottom the same census tract as adults as they did Quarter of Family Income Who Remain in the in adolescence, 59 percent remained in Bottom Quarter in Adulthood, by Residential the bottom quarter in 1992. Those who Mobility Since Adolescence, 1992 and 2010 lived in an entirely different region of the

60% 59% country had only a 39 percent likelihood 49% 48% of remaining in the bottom quarter. 50% 46% 44%44% 42% 43% None of the changes over time shown in 40% 36% 37% 37% 34% Figure 12 is statistically reliable, but the 30% improvement in mobility among those 20% remaining in their adolescent census 10% tract is most likely to reflect what actually 0% happened. All Same Tract Same Same Same Different in '93 as in '79 County State Region Region 1992 2010 In addition to upward economic mobility, Source: Author’s calculations based on data from National Longitudinal Survey of Youth 1979 the BLS data provide information on and National Longitudinal Survey of Youth 1997 downward mobility. For instance, one

When Moving Matters 13 14 e21 Report 2 November 2014 young adultsin2010(55percent, not of living).Somewhat surprisingly, more (after accountingfortherise inthecost their childhoodincomeasadolescents adults whosefamilyincomeexceeds section examinesthepercentage ofyoung measured inavariety ofways. This Absolute economicmobilitymaybe Mobility, 1992and2010 Absolute Intergenerational Economic the actualdeclinemayhave beensmaller. census tractisstatisticallylikely, though those remaining intheiradolescent large drop in downward mobilityamong downward economicmobility. The their adolescentcensustracthadlower Roughly, those whomoved from farther between adolescenceandadulthood. with more limitedresidential mobility economic mobilityoutcomesforthose Figure 14continuestoindicateworse is precisely estimated. occurred, butnoneoftheotherchanges is statistically likely to have really residencetheir birth asadolescents mobility for adults still living in 2010. The improvement indownward economic mobilityinboth1992and born hadthehighestratesofdownward from theoneintowhichtheywere asadolescentsthatwasdifferent country Adults who lived in a region of the residential immobilitybeingbeneficial. relationship canbediscerned,itpointstoward As withupward mobility, totheextentthatany way.economic mobilityinanystraightforward andadolescenceisrelatedbetween birth todownward 13 provides littleevidencethatresidential mobility asadults.Figuredropping tothebottomquarter second-poorest offamilyincomeasadolescents quarter can examinetheshare ofadultswhowere raisedinthe and NationalLongitudinalSurveyofYouth 1997 Source: Author’s calculationsbasedondatafrom NationalLongitudinalSurveyofYouth 1979 and NationalLongitudinalSurveyofYouth 1997 Source: Author’s calculationsbasedondatafrom NationalLongitudinalSurvey ofYouth 1979 Bottom QuarterinAdulthood,byResidentialMobility 10% 15% 20% 25% 30% 35% 40% Figure 14.Percentage ofAdultsRaisedintheSecond Figure 13.Percentage ofAdultsRaisedintheSecond 0% 5% 10% 15% 20% 25% 30% 35% 40% 0% 5% Between BirthandAdolescence, 1992and2010 Bottom QuarterinAdulthood,byResidential Mobility SinceAdolescence,1992and2010 28% Quarter ofFamily IncomeWhoFall tothe Quarter ofFamily IncomeWhoFall tothe 28% same (15.9percent) asin1992(15.8percent). unemployment duringtheyear wasroughly the theshareshort, ofthelaborforce experiencingany because mostunemployment spellsare relatively Long-term unemployment washigherin2010;but median family income in their business cycles. Both 1992and2010were closetolow pointsfor of young adultsin1992 was (46 true percent). shown) had exceeded theirchildhood incomes than All 24% All 24% in '93as'79 Same Tract 40% Same Residence 24% 27% at Birth 16% 30% County 1992 Same 1992 28% Same State 28% at Birth 2010 2010 26% 22% Same State 26% Same Region 17% at Birth 26% Region Same 20% 17% Different Region 34% Different 22% at Birth Region 18% 34% 40 39

The explanation for the greater absolute upward In any given year, it is harder to exceed income mobility of adults in the second BLS data set—a from an earlier year, the closer the starting result very likely, statistically, to reflect a true real- point is to a peak. Economic commentary often world increase—is the point at which childhood emphasizes the stagnation or decline of median incomes are assessed. In the first panel, 1978 was income over the past 15 years, but even according one year from a cyclical peak for median family to the official Census Bureau measures (which income. In contrast, in the NLSY97, parental understate income growth for various reasons), income is assessed in 1996, which was closer to the median income was over $2,000 higher in 2010 trough year of 1993 than to the peak year of 2000. than in 1996, after accounting for inflation. (It was about $6,500 higher in 2007 than Figure 15. Percentage of Adults Raised in the in 1996.) The point is that the trend Bottom Quarter with a Higher Family Income estimates in this section are likely to than Their Parents, by Residential Mobility be sensitive to the particular years Between Birth and Adolescence, 1992 and 2010 considered, which were chosen to maximize the comparability between 100% 93% 90% 85% 86% 83% 86% the BLS data sets and to avoid 80% 73% 75% 72% 70% 71% observing adults at younger ages. 70% 60% What is important is less the trends 50% and more the differences according to 40% 30% types of residential mobility. 20% 10% 0% It should be emphasized that the family Bottom Same Residence Same State Same Region Different Region incomes of adults were measured when Quartile at Birth at Birth at Birth at Birth 1992 2010 they were between the ages of 27 and 30. Their parents tended to be older Source: Author’s calculations based on data from National Longitudinal Survey of Youth 1979 and National Longitudinal Survey of Youth 1997 than this when childhood family income was assessed, so if it were possible to observe the adults in the NLSY97 at, Figure 16. Percentage of Adults Raised in the say, age 40, the share exceeding their Bottom Quarter with a Higher Family Income childhood income would be higher still. Percentile than Their Parents, by Residential In the NLSY79, where adults have been Mobility Since Adolescence, 1992 and 2010 followed longer and are older today

100% than in the NLSY97, only 46 percent 87% 88% 90% 85% 84% 83% 85% 78% of adults had higher family incomes 80% 74% 77% 73% 70% 70% 66% than their parents at ages 27–30; but 64 60% percent of these same adults exceeded 50% 40% their parents’ incomes in 2005, when 30% they were 40–43 (not shown). 20% 10% 0% Determining how residential mobility Bottom Same Same Same Same Different Quartile Tract County State Region Region is connected to absolute upward 1992 2010 mobility requires accounting for

Source: Author’s calculations based on data from National Longitudinal Survey of Youth 1979 parental family income because it and National Longitudinal Survey of Youth 1997 is easier for adults who were poor as

When Moving Matters 15 16 e21 Report 2 November 2014 lower downward economicmobility. Adults whomove awayhave farther quarter, andeven arough onein2010. mobility from thesecond-poorest residential mobilityanddownward between adolescence-to-adulthood 18 shows a clear relationship in 1992 relationship tobediscerned.Figure 2010. In 1992,there isagainlittle residence,their birthplace atleastin adolescentswerethe farther from measure ofmobilitylooksworse, ofchildhoodincome. quarter This mobility from thesecond-poorest Figure 17shiftstoabsolutedownward so in2010. bottom in1992,thoughnotobviously of absolute upward mobility from the correspond withagreater likelihood from one’s adolescenthomeroughly In but consistentpatternsare stillelusive. forabsoluteupwardimportant mobility, more and adulthoodmaybesomewhat Residential mobility between adolescence upward mobility. seems tocorrespond lower tosomewhat from one’sfarther residence birthplace is readily apparent; butin1992,living mobility andabsoluteupward mobility residentialbetween birth-to-adolescent low point.No starting relationship Strong majoritiesdosobecauseoftheir their childhoodincomeininflation-adjustedterms. ofchildhoodincomewhoexceedbottom quarter Figure 15considers the share ofadults raised inthe between richandpoorchildren. addition, residential mobilitypatternsmaydiffer children, there isnowhere togobutup. In in affluentfamiliesaschildren; forthepoorest children tomove upthanitisforadultswhowere Figure 16,moves thatare farther and NationalLongitudinalSurveyofYouth 1997 Source: Author’s calculationsbasedondatafrom NationalLongitudinalSurveyof Youth 1979 and NationalLongitudinalSurveyofYouth 1997 Source: Author’s calculationsbasedondatafrom NationalLongitudinalSurveyofYouth 1979 Poorest QuarterwithaLower Family IncomePercentile Figure 18.Percentage ofAdultsRaisedintheSecond– 100% Second–Poorest QuarterwithaLower Family Income 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 0% than TheirParents, byResidentialMobilitySince Between BirthandAdolescence, 1992and2010 Figure 17.Percentage ofAdultsRaisedinthe than TheirParents, byResidentialMobility Quartile Quartile 50% Second Second 50% mobility between adolescenceandadulthoodis not. The evidenceismore consistentthatresidential early lifestagesmaybeharmfulmore oftenthan outcomes, meaningthatresidential mobilityduring to where onewasborncorresponds tobetter economic mobilitythatlivinginadolescencenearer There issomeevidenceforbothrelative andabsolute Bureau becauseoffarsmallersamples. data,inpart In sum,theBLSdataare lessclearthantheCensus 34% 34% Adolescence, 1992and2010 Same Residence 62% Same Tract 49% at Birth 35% 26% 55% 1992 County 1992 Same 38% Same State 50% at Birth 2010 36% 2010 45% Same State 33% Same Region 48% at Birth 42% Region 33% Same 27% Different Region 51% 38% Different at Birth Region 47% 28% associated with better economic mobility outcomes Educational Requirements of Jobs, 1880–2010 and that outcomes are better for the adults who The Census Bureau data include, for each year and have moved the farthest away. for each worker with an occupation identified, the percentage of workers in that job who had com- Combining the BLS results with those of the pleted at least one year of college. Ranking jobs by Census Bureau, to the extent that residential this metric indicates which jobs are most popular mobility between birth and adulthood has among those who are highly educated. Some jobs benefits, they seem to arise from the adult’s own may not necessarily pay inordinately well but are residential mobility between adolescence and desirable for other reasons. adulthood rather than the parents’ residential mobility during their childhood. The BLS data The measure is somewhat less informative in the years offer no suggestion that residential mobility has before 1950 because the college enrollment estimates become more important over time. for each occupation are based on the 1950 educational attainment figures. Jobs that required a relatively high VI. RESIDENTIAL MOBILITY AND ADULT level of schooling in 1950 may not have required this OUTCOMES, 1880–2010 in, say, 1880. Nevertheless, this metric is still likely to be among the best indicators of adult opportunity The analyses above were able to connect childhood available before the mid-twentieth century. circumstances and adult incomes, either by linking adults to their parents or birthplaces in childhood. One residential mobility indicator that is available in Therefore, it was possible to make statements about the earliest years of the census involves comparing the “intergenerational economic mobility”—the ex- birth states of adults with their parents’ birthplaces.42 tent to which movers and non-movers did better or A daughter might be born in a different state from worse, taking their origins into account. her parents either because her parents moved to a new state as adults, before she was born, or because her Section VI extends the analyses to the nineteenth grandparents moved when her parents were young. It century, at the cost of not being able to observe is an intergenerational measure of residential mobility. childhood circumstances. Comparing adult outcomes between movers and non-movers is less infor- Early trends in the educational requirements of mative if childhood conditions cannot be taken into occupations and how they vary with residential account. The concern becomes greater that movers mobility are dominated by women (Figure 20). who do well were advantaged and would have done In 1880, there were already small differences well anyway. Nevertheless, the evidence may be sug- in these educational requirements, depending gestive, particularly to the extent that it is consistent on where women and their parents were born. with the results from previous sections. Thirtysomething women who were raised by one or two immigrant parents typically had jobs with Analyses in Section VI use the Census Bureau data higher educational requirements than their peers and are confined to native-born adults aged 30–39.41 with native-born parents. From 1900 to 1910, the Because the 1890 census records are unavailable, the gap between women with one immigrant parent (but trends in the charts below interpolate from 1880 to not two) and everyone else widened dramatically. 1900. The appendix (www.manhattan-institute.org/ pdf/e21_r2_appendix.pdf) includes estimates for men In 1910, the differences were still small between women and women combined (Figures A45–A65), while the born in the same state as both parents and women born following discussion presents them separately. in a different state from at least one native-born parent.

When Moving Matters 17 18 e21 Report 2 November 2014 in adifferent region. same geographicregion, andadultsliving adults livinginadifferent statewithinthe in thesamestatewhichtheywere born, section distinguish between adults living during childhood. The analysesinthis because theirparents moved sometime away from home;itmightdifferalso adults mightdifferbecausetheymoved year. The current stateof stateandbirth of residential mobilityisavailable inevery measure 1940, butthebirth-to-adulthood information onparents’ statesafter birth The CensusBureau datadonotinclude born inthesamestateasbothparents). held by men(theexception beingwomen in 1930and1940,compared withthose requirements ofthejobsheldby women figures isthemuchhighereducational even in1940(Figure 19). Of noteinthese residential mobilitycategorieswassmall much atall,sothedifference between of typical male jobs did not increase In contrast,theeducational requirements of 60years earlier. 7 percent, notmuchabove the3percent born in the same state as their parents was The corresponding figure for women had completedatleastoneyear ofcollege. percent–21 percent of1950incumbents from oneorbothparents, 18 or country who hadbeenborninadifferent state jobs typicallyheldin1940by women worse in1940thantheirpeers.In the same stateasbothparents fared far womenborninthe Thirtysomething state. By 1930,thatgaphadclosed. those withjustoneparent from adifferent parents hadmore prestigious jobsthaneven women borninadifferent statefrom both By 1920,alargegaphadopened,and Microdata Series Source: Author’s calculationsbasedondatafrom decennialcensuses,Integrated PublicUse Microdata Series Source: Author’s calculationsbasedondatafrom decennialcensuses,IntegratedPublicUse Percentage of Incumbents in One's Occupation in 1950 Percentage of Incumbents in One's Occupation in 1950 Figure 20.EducationalRequirements ofOccupation,

Who Complete at Least One Year of College Who Complete at Least One Year of College Figure 19.EducationalRequirements ofOccupation, 1 1 2 2 10 15 20 25 0 5 0 5 0 5 0 5 1 1870 Thirtysomething Adults,byIntergenerational Thirtysomething Adults,byIntergenerational 8 Residential Mobility, 1880–1940—Women 7 Born inSameStateasBothParents Born Other Born AbroadOther Born One ParentinSameState, Born Same StateasChild Parents inU.S.,OnlyOne Born inSameStateasBothParents Born AbroadOther Born One ParentinSameState, Born Same StateasChild Parents inU.S.,OnlyOne Born 0 Residential Mobility, 1880–1940—Men 1 1880 8 8 0 1 1890 8 9 0 1 1900 9 0 0 1 1910 9 1 0 1 1920 Different Statesfrom Child Both Parents inU.S.but Born Both Parents Abroad Born Different Statesfrom Child Both Parents inU.S.but Born Both Parents Abroad Born Other Born AbroadOther Born One ParentinDifferent Born State, AbroadOther Born One ParentinDifferent Born State, 9 2 0 1 1930 9 3 0 1 1940 9 4 0 1 1950 9 5 0 Figure 21. Educational Requirements of Occupation, Because the results in the earliest years Thirtysomething Adults, by Childhood–to–Adulthood are uninteresting, the figures begin with Residential Mobility, 1940–2010—Men 1940.43

80 According to Figure 21, from 1950 70 onward, non-moving men lagged

60 further and further behind other men in terms of the number of people in their 50 jobs who had attended a year of college. 40 It is unclear what accounts for the dip 30 in 2000; but apart from that year, men

20 who move from their birth state have similar outcomes regardless of whether 10 they stay in the region. Who Complete at Least One Year of College Who Complete at Least One Year 0 1930 1940 1950 1960 1970 1980 1990 2000 2010 Percentage of Incumbents in One's Occupation 1950 Before 1970, thirtysomething women Lives in Birth State Lives in Region of Birth, Lives in Region Different had jobs with higher educational Different State from Region of Birth requirements than their male Source: Author’s calculations based on data from decennial censuses and 2010 American counterparts, and that remained true Community Survey, Integrated Public Use Microdata Series after 1970 among non-movers (Figure 22). Men and women who moved had Figure 22. Educational Requirements of Occupation, essentially identical outcomes from 1970 Thirtysomething Adults, by Childhood–to–Adulthood forward, and women who continued to Residential Mobility, 1940–2010—Women live in their birth state did just as well as movers until 1990. Between 1980 and 80 2010, female movers fell behind other 70 women, though the gap was smaller

60 than among men.

50 Residential mobility over the preceding 40 five years may be assessed between 1940 30 and 2000. In the figures to follow, the

20 trend is extrapolated from 1940 to 1970, as information for 1950 and 1960 is 10 unavailable.44 Among both men and Who Complete at Least One Year of College Who Complete at Least One Year 0 1930 1940 1950 1960 1970 1980 1990 2000 2010 women, those who had moved to a Percentage of Incumbents in One's Occupation 1950 different region in the preceding five Lives in Birth State Lives in Region of Birth, Lives in Region Different Different State from Region of Birth years consistently did better than their peers in terms of the share of people in Source: Author’s calculations based on data from decennial censuses and 2010 American their occupation who attended college Community Survey, Integrated Public Use Microdata Series for a year. The gap among men rose between 1940 and 1970 and remained large (Figure 23); it was much smaller among women until 2000 (Figure 24).

When Moving Matters 19 20 e21 Report 2 November 2014 state made $2,800 less (2014 dollars) in worked andwhostilllived intheirbirth tial mobility, menwho thirtysomething Turning tobirth-to-adult for whichtheseestimatesare available. in 1940, this istheonly yearbirthplace census stoppedaskingaboutparental had thehighestearnings.Since the with oneorbothparents bornabroad the United States. Thirtysomethings while theotherwasbornelsewhere in with oneparent state sharingtheirbirth children were born,followed by adults in adifferent statefrom where their born parents whoboth were born women). Next were adultswithnative- A46 forresults combiningmenand and in 1940thantheirpeers(Figure 25 state asbothparents hadlower earnings any earnings,thoseborninthesame whohadand womenintheirthirties residential mobility, among both men year. Starting withintergenerational information abouttheprevious earnings and family income, recording The 1940censuswasthefirsttoaskabout Earnings, 1940–2010 closed in2000. to 1990,butthegendergapessentially than womenwhomoved from 1970 out ofregion hadbetteroccupations than non-moving men.Men whomove because non-moving women do better in-region-mover gapamongwomen region movers. There isnonon-mover/ between out-of-region movers andin- remains muchsmallerthanthegap gap thathasgrown over timebutthat better occupationsonlyamongmen,a within theregion isassociatedwith In contrast,moving state to a new Figure 26;seeAppendix Figure hood residen

Microdata Series Source: Author’s calculationsbasedondatafrom decennialcensuses,Integrated PublicUse Microdata Series Source: Author’s calculationsbased ondatafrom decennialcensuses,IntegratedPublicUse Thirtysomething Adults,byLast–Five–Years Residential Percentage of Incumbents in One's Occupation in 1950 Thirtysomething Adults,byLast–Five–Years Residential Percentage of Incumbents in One's Occupation in 1950 Figure 24.EducationalRequirements ofOccupation, Who Complete at Least One Year of College Who Complete at Least One Year of College Figure 23.EducationalRequirements ofOccupation, 10 20 30 40 50 60 70 80 10 20 30 40 50 60 70 80 0 0 1930 1930 as FiveYears Ago Lives inSameState as FiveYears Ago Lives inSameState 1940 1940 Mobility, 1940–2000—Women Mobility, 1940–2000—Men 1950 1950 1960 1960 Lives inSameRegion Lives inSameRegion 1970 1970 1980 1980 Lives inDifferent Region Lives inDifferent Region 1990 1990 2000 2000 2010 2010 Figure 25. Median Wage & Salary Income of Workers, Thirtysomething Adults, by Intergenerational Residential Mobility, 1940—Men

$20,000 $18,152 $18,152 $17,803 $18,000 $16,755 $16,378 $16,000 $13,963 $14,000

$12,000

$10,000

2014 Dollars $8,000

$6,000

$4,000

$2,000

$0 Born in Same Parents Born One Parent Both Parents One Parent Both Parents State as in U.S., Only One Born in Same Born in U.S. but in Born in Different Born Abroad Both Parents in Same State State, Other Different States State, Other as Child Born Abroad from Child Born Abroad

Source: Author’s calculations based on data from 1940 census, Integrated Public Use Microdata Series

Figure 26. Median Wage & Salary Income of Workers, Thirtysomething Adults, by Intergenerational Residential Mobility, 1940—Women

$20,000

$18,000

$16,000

$14,000 $12,413 $12,000 $11,589 $11,170 $10,053 $10,000 $8,894 $7,987

2014 Dollars $8,000

$6,000

$4,000

$2,000

Source: Author’s calculations based on data from 1940 census, Integrated Public Use Microdata Series

When Moving Matters 21 22 e21 Report 2 November 2014 again by 30percent. earnings doubledandthenincreased 2010. In the30years preceding 1980, only by about5percent during1980– non-moving mengrew thirtysomething medians inFigure 27,theearningsof how muchthey would addtothe under reasonable assumptionsabout share ofpayover time.However, even earnings measure, have becomearising Nonwage benefits,notincludedinthis over $3,000. median lower in2010than1980by downward trend inearnings,withthe among non-movers isthere aclear movers were thesameasin1980.Only as in1970,andthemedianearningsof stateweretheir birth atthesamelevel menremainingof thirtysomething in after the1960s.By 2010,theearnings clearly shows stagnationinearnings Across allthree groups, Figure 27 their region thaniftheylefttheregion. state didnobetteriftheyremained in unevenly. Men whohadmoved outof grew tonearly$5,500by 2010,though 1950s, and1960sthatwas intended of awagepremium duringthe1940s, trends is thatmenwere beneficiaries One explanation consistent with these decade formaleearningsgrowth. earnings during the 1950s, astrong Interestingly, working womensawflat percent to 80 percent) (Figure 28). much asbetween 1950and1980(71 percent from 1980to2010—nearlyas women rose bythirtysomething over 60 In contrast,themedian earningsof state(Figuretheir birth 27 ). 1940 thanmenwhonolongerlived in 46 45 That gap That Community Survey, IntegratedPublicUseMicrodata Series Source: Author’s calculationsbasedondatafrom decennialcensusesand2010 American Community Survey, IntegratedPublicUseMicrodata Series Source: Author’s calculationsbased ondatafrom decennialcensusesand2010American 2014 Dollars 2014 Dollars Thirtysomething Adults,byChildhood–to–Adulthood Thirtysomething Adults,byChildhood–to–Adulthood $10,000 $20,000 $30,000 $40,000 $50,000 $60,000 $10,000 $20,000 $30,000 $40,000 $50,000 $60,000 $0 $0 Residential Mobility, 1940–2010—Women Figure 28.MedianWage &SalaryIncome, Figure 27.MedianWage &SalaryIncome, Lives inBirthState Lives inBirthState 1930 1930 Residential Mobility, 1940–2010—Men 1940 1940 1950 1950 Different State Lives inRegionofBirth, Different State Lives inRegionofBirth, 1960 1960 1970 1970 1980 1980 1990 1990 from RegionofBirth Lives inRegionDifferent from RegionofBirth Lives inRegionDifferent 2000 2000 2010 2010 Figure 29. Median Wage & Salary Income, to support the patriarchal male- Thirtysomething Adults, by Last–Five–Years Residential breadwinner ideal, while women— Mobility, 1940–2000—Men particularly married women—faced labor-market discrimination. As more $60,000 and more wives worked significant hours, the rationale for the male- $50,000 breadwinner ideal broke down. Men, $40,000 finding their pay at inflated levels that their productivity could not $30,000 justify, saw decades of stagnation, while women enjoyed strong earnings

2014 Dollars $20,000 growth as they became more skilled $10,000 and experienced and as occupational barriers withered. $0 1930 1940 1950 1960 1970 1980 1990 2000 2010 After 1970, non-moving women began Lives in Same State Lives in Same Region Lives in Different Region to fall behind women who no longer as Five Years Ago lived in their birth state but still lived Source: Author’s calculations based on data from decennial censuses, Integrated Public Use Microdata Series in the same region; by 2010, a gap of roughly $3,000 had opened. Women who lived in an entirely different Figure 30. Median Wage & Salary Income, region from where they were born have Thirtysomething Adults, by Last–Five–Years Residential generally done better than women who Mobility, 1940–2000—Women moved to a new state within the region; since 1980, however, the difference has $60,000 been negligible. $50,000 As for short-term residential mobility $40,000 over the preceding five years: among

$30,000 women in their thirties, those who moved out of state but who stayed in

2014 Dollars $20,000 the region consistently did a bit better than women who remained in the same $10,000 state. The gap ranged from roughly

$0 $1,300 to $2,500. Women who 1930 1940 1950 1960 1970 1980 1990 2000 2010 moved to a new region did worse than other women, though the difference Lives in Same State Lives in Same Region Lives in Different Region as Five Years Ago between them and non-movers was

Source: Author’s calculations based on data from decennial censuses, Integrated Public Use only sizable in 1980 and 1990 (Figure Microdata Series 30). These patterns also held among men between 1970 and 1990. By 2000, the ordering had reversed: men who moved out of their region had the highest earnings (Figure 29).

When Moving Matters 23 24 e21 Report 2 November 2014 Source: Author’s calculationsbased ondatafrom 1940census,IntegratedPublicUseMicrodata Series Source: Author’s calculationsbasedondatafrom 1940census,IntegratedPublic UseMicrodata Series

2014 Dollars 2014 Dollars $10,000 $12,000 $14,000 $16,000 $18,000 $20,000 $10,000 $12,000 $14,000 $16,000 $18,000 $20,000 $2,000 $4,000 $6,000 $8,000 $2,000 $4,000 $6,000 $8,000 Figure 32.MedianFamily Wage &Salary Income,ThirtysomethingAdults, Figure 31.MedianFamily Wage &SalaryIncome,ThirtysomethingAdults, $0 $0 Born inSame Born Born inSame Born Both Parents Both Parents $11,868 $12,566 State as State as by Intergenerational ResidentialMobility, 1940—Women by Intergenerational ResidentialMobility, 1940—Men in U.S.,OnlyOne in U.S.,OnlyOne in SameState in SameState Parents Born Parents Born $13,963 $13,963 as Child as Child Born inSame Born Born inSame Born Born AbroadBorn Born AbroadBorn State, Other State, Other One Parent One Parent $18,152 $17,439 Born inU.S.but Born Born inU.S.but Born Different States Different States Both Parents Both Parents from Child from Child $13,963 $13,963 Born inDifferentBorn Born inDifferentBorn Born AbroadBorn State, Other Born AbroadBorn State, Other One Parent One Parent $16,853 $19,408 Both Parents Both Parents Born AbroadBorn Born AbroadBorn $18,850 $18,152 Figure 33. Median Family Income, Family Income, 1940–2010 Thirtysomething Adults, by Childhood–to–Adulthood The same broad pattern for inter Residential Mobility, 1940–2010—Men generational residential mobility (found when examining men’s and women’s $80,000 earnings) recurs with family income $70,000 (specifically, family-level earnings in 1940). Adults with one or two immigrant $60,000 parents have higher incomes than adults $50,000 with only native-born parents, and those born in the same state as both parents

2014 Dollars $40,000 have the lowest incomes (Figure 31 and $30,000 Figure 32). Women with a native-born

$20,000 parent born in a different state and an immigrant parent did especially well, $10,000 having the highest median family income $0 of any of the groups for men or women. 1930 1940 1950 1960 1970 1980 1990 2000 2010 Otherwise, the figures for men and

Lives in Birth State Lives in Region of Birth, Lives in Region Different women are nearly indistinguishable. Different State from Region of Birth

Source: Author’s calculations based on data from decennial censuses and 2010 American Men’s and women’s family income Community Survey, Integrated Public Use Microdata Series trends are remarkably similar when broken out by birth-to-adulthood residential mobility (Figure 33 and Figure 34. Median Family Income, Figure 34). The estimates show an Thirtysomething Adults, by Childhood–to–Adulthood Residential Mobility, 1940–2010—Women increase, between 1980 and 2010, of about 10 percent—significantly smaller $80,000 than in the preceding decades.

$70,000 Both figures show that adults still living $60,000 in their birth state have lower incomes than those living elsewhere. This gap $50,000 has grown over time, equaling roughly

2014 Dollars $40,000 $6,000 in 2010. A smaller gap has

$30,000 opened in recent decades between those who live in a different region from the $20,000 one in which they were born and those

$10,000 living in a different state but in the same region. Staying within the region, but $0 1930 1940 1950 1960 1970 1980 1990 2000 2010 outside one’s birth state, is associated with median incomes higher by about $1,000. Lives in Birth State Lives in Region of Birth, Lives in Region Different Different State from Region of Birth Family income trends according to Source: Author’s calculations based on data from decennial censuses and 2010 American Community Survey, Integrated Public Use Microdata Series residential mobility in the preceding five years are similar to those for male

When Moving Matters 25 26 e21 Report 2 November 2014 both parents alsohadworseoutcomesin rates. Men born inthesamestateas high schoolandcollege graduation family earningsbutintermsoftheir of theirjobsandindividual terms oftheeducationalrequirements worse thantheirpeers—notonlyin in thesamestateasbothparents fared results: in1940,adultwomenborn intergenerational residential mobility Toappendix.pdf). summarize the full manhattan-institute.org/pdf/e21_r2_ are includedintheappendix(www. tional attainmentandemployment Additional analyses examiningeduca Summary and incomeiscomplicated. residentialbetween short-term mobility At anyrate,itisclearthattherelationship environs.comparable jobsintheirnew collar workers whowere unabletofind times, necessitatingmoves by, say, blue- went through economically difficult regionsperhaps theanswer isthatcertain may have fewer skills than others. Or moving placewithintheregion) toanew filling outtheircensusform(insteadof out of the region in the years before people whohave gonesofarastomove some years. Perhaps itreflects selection— poor showing ofout-of-region movers in It isunclearwhatmightaccountforthe groups ofmovers were minimal. the differences between thethree incomes in1980and1990.By 2000, their region entirely hadlower family Men andwomenwhomoved outof else fromthan everyone 1970 to 1990. region hadhigherfamilyincomes statewithintheirexistingto anew Men andwomenwhorecently moved earnings (Figure 35andFigure 36). Microdata Series Source: Author’s calculationsbasedondatafrom decennialcensuses,Integrated PublicUse Microdata Series Source: Author’s calculationsbasedondatafrom decennialcensuses,IntegratedPublicUse

2014 Dollars 2014 Dollars $10,000 $20,000 $30,000 $40,000 $50,000 $60,000 $70,000 $80,000 $10,000 $20,000 $30,000 $40,000 $50,000 $60,000 $70,000 $80,000 $0 $0 Thirtysomething Adults,byLast–Five–Years Thirtysomething Adults,byLast–Five–Years Residential Mobility, 1940–2000—Women 1930 1930 as FiveYears Ago Lives inSameState as FiveYears Ago Lives inSameState Residential Mobility, 1940–2000—Men Figure 36.MedianFamily Income, Figure 35.MedianFamily Income, 1940 1940 1950 1950 Lives inSameRegion Lives inSameRegion 1960 1960 1970 1970 1980 1980 1990 1990 Lives inDifferent Region Lives inDifferent Region 2000 2000 2010 2010 1940, relative to other thirtysomething men—not growing more closely related to adult benefits over only in terms of individual and family earnings but time. It is much less clear that the willingness or in terms of high school and college graduation rates. ability of adults to move in the short run improves Adults with immigrant parents often had better their economic outcomes, or ever did. Residential outcomes than their peers during these years, except mobility by parents between the birth of a child and that they had lower educational attainment. the child’s adolescence may actually worsen adult outcomes more often than not. Unequal outcomes are also present when examining residential mobility between birth VII. CHANGES IN THE EXTENT OF and adulthood. Across nearly all the outcomes RESIDENTIAL MOBILITY AND GROUP examined, in most years, for both men and DIFFERENCES women, native-born thirtysomethings who remain in their birth state did worse than their The evidence from Section VI suggests that the abil- peers. In many cases, especially for the educational ity and willingness to move may be more impor- requirements of jobs and college graduation rates, tant today than in the past. If residential mobility the gap widened. The exceptions are employment has increased, that would be a force working in the and unemployment rates, where the patterns were direction of raising economic mobility. Conversely, not so closely related to residential mobility; and falling residential mobility would raise the concern high school dropout rates, where differences by that the decline has hurt economic mobility rates residential mobility narrowed. over time. Also potentially worrisome: Has residential mobility risen less or fallen more among disad- Compared with the patterns for intergenerational vantaged groups than among others? residential mobility and birth-to-adulthood moves, recent moves (over the preceding five Scholars have found a number of varying long- years) seem less strongly related to adult outcomes. term trends in residential mobility. Jason Long The familiar pattern—where adults who end up and Joseph Ferrie found that 50 percent of white geographically farthest away from where they began native-born men, aged 20–29, changed counties do significantly better than other adults—shows between 1870 and 1880, and 26 percent changed up clearly only in the educational requirements states. The figures for 1971–81 were just 42 of men’s jobs and in college graduation rates. percent and 22 percent, respectively.47 However, Indeed, adults who have recently moved to a new Joshua Rosenbloom and William Sundstrom region—not just a new state—regularly have worse found that the share of native-born 20- to outcomes than their peers, a result that may reflect 29-year-olds who had left their birth state was “selection” (certain types of people moving to or somewhat higher in 1980 than in 1880, hovering from certain types of regions). around 30 percent in both years.48

In light of the existing results—including those that Both studies found that residential mobility de- were conditioned on birthplace or family income— clined among older men. Long and Ferrie found the implication is that to the extent that residential that in 1870–80, 35 percent of white native-born mobility improves adult outcomes, it is the geo- men aged 45–59 changed counties, and 23 percent graphic movement that occurs between adolescence changed states. The figures for 1971–81 were 17 and adulthood that most clearly matters, and, by percent and 9 percent, respectively. Rosenbloom and large, it is out-of-state residential mobility that and Sundstrom found modest declines for native- is beneficial. These kinds of moves appear to be born adults aged 40–49 and 50–59.

When Moving Matters 27 28 e21 Report 2 November 2014 steadily between themid-1950sand2007. tion thatchangedresidences intheprevious year fell William Frey shows that the share of the popula- In contrast to the trends on interstate mobility, 1980s through the2000s. ous year among25-to44-year-olds fellfrom the Wozniak find that interstate mobility in the previ- continued rising until 1980. Molloy, Smith, and interstateresidentialthirtysomethings, mobility until 1960;andthatitwasflat1990. For then and1930or1940;thatitsubsequentlyrose thatitchangedlittle betweentwentieth century; between 1860andthelatenineteenthorearly bility between decades. They findthatitdeclined detailed lookattrends ininterstateresidential mo- Rosenbloom andSundstrom also provide a more cern aboutaseculardeclineinresidential mobility. to considermobility—hasbeenthebasisforcon- window short-term withinwhich constitutes avery within astate,county, orneighborhoodandwhich type of mobility—which includes residential moves Series Source: Author’s calculationsbasedondatafrom decennialcensusesand2010AmericanCommunitySurvey, IntegratedPublicUseMicrodata *Some columnsdonotsumto100percent duetorounding. 100% 10% 20% 30% 40% 50% 60% 70% 80% 90% 0% Figure 37.BirthplacesofThirtysomethings&TheirParents, 1850–2010* 1850 81 19 Born intheU.S. Born 1860 71 29 1870 73 27 49 1880 67 24 3 6 All Born Abroad Abroad 1890 Parents Parents Parent in U.S. Child Born Born Both One and 1900 22 57 16 Born AbroadBorn 5 1910 21 58 14 50 6 This 1920 22 59 13 7 1930 had beenbornabroad, afigure thathadfallento that in1860,29percent ofadultsintheirthirties is dominatedby immigration.Figure 37shows of U.S. residential mobility between generations Through themid-twentieth century, thestory Versus Parental Birthplaces Intergenerational Residential Mobility—Birthplaces adult outcomesintheprevious section. moves—the kindwiththeweakest connectionto perspective ortoinvolve orrecent short-distance by pastresearch appeartobesmallinhistorical decades. The declines inrecent decades found and hasbeenontherise,orstable,inrecent completed mobility(inthepreceding five years)— adulthood mobilityandin1970forrecently out several decadesago—in1940forbirth-to- and indicatethatresidential mobilitybottomed resemble thoseofRosenbloom andSundstrom this conflictingliterature. The results most closely residential mobilityare presented toreconcile In Section VII, trends fordifferent kindsof 17 62 14 7 Born intheU.S. Born 1940 10 70 14 6 1950 73 17 4 6 1960 79 13 1 7 1970 81 7 5 6 1980 91 9 Born intheU.S. Born 1990 88 12 2000 82 18 2010 76 24 22 percent by 1900. The immigrant share of the abroad or had a parent born abroad. By 1940, 19 population then rapidly declined to 1 percent by percent of thirtysomethings (born in the first decade 1960, in the aftermath of legislative restrictions of the twentieth century) had been born in a state on immigration in the 1920s and administrative that differed from one or both of their American- decisions by the Roosevelt administration during born parents, but 30 percent had been born abroad the Great Depression. Similarly, the share of or had a parent born abroad. thirtysomethings either born abroad or with at least one parent born abroad was 43 percent in 1900 Information on the birth state of parents is but 19 percent in 1970. After immigration reform unavailable in the census data after 1940, but it in 1965, the share of thirtysomething Americans was still the case in 1950 that 27 percent of adults born abroad rose steadily, from 1 percent in 1960 in their thirties were immigrants or children of to 24 percent in 2010 (higher than in any year immigrants. It is likely that as late as 1970, among since the nineteenth century). thirtysomething adults, about as many were immigrants or children of immigrants as had been Through the mid-twentieth century, having been born in a different state from one of their native- born in a different state from one’s American-born born parents. It is also likely that by 2010 (or even parents was less common among thirtysomething 2000), immigration again accounted for most adults than having been born to immigrant parents. intergenerational residential mobility in this sense. In 1880, 24 percent of adults in their thirties had American-born parents, at least one of whom was Native-born adults with native-born parents have born in a different state from where they were born tended to live in the same state as their parents for (Figure 38). In contrast, 33 percent were born well over a century. This tendency has grown stronger

Figure 38. Birthplaces of Thirtysomethings & Their Parents, 1880–1940*

100% 10 90% 17 24 Born Abroad 22 22 22 14 80% 14 2 6 14 13 70% 1 16 9 Born in U.S., 2 Diﬀerent State 2 3 4 60% 16 2 9 from Both Parents 9 10 11 5 13 50% 1 4 8 4 8 3 8 8 40% 8 Born in U.S., 30% Same State as At 52 Least One Parent 43 41 44 20% 36 39

10%

0% 1880 1890 1900 1910 1920 1930 1940

Born Abroad Born in U.S., Both Parents Born Abroad Born in U.S., One Parent Born in Same Born in U.S., One Parent Born in Different State, Other Born Abroad State, Other Born Abroad Born in U.S., Parents Born in U.S., Born in U.S., Both Parents Born in U.S. Only One in Same State as Child but in Different States from Child Born in U.S., in Same State as Both Parents

*Some columns do not sum to 100 percent due to rounding. Source: Author’s calculations based on data from decennial censuses, Integrated Public Use Microdata Series

When Moving Matters 29 30 e21 Report 2 November 2014 thirtysomethings lived inthesamestate asboth thirtysomethings born parents. Within thisgroup, 78percent ofblack whites, even amongthenative-born withnative- African-Americans moved lessthannon-Hispanic with just49percent ofwhites). parents. By 1940,82percent were (compared born intheU.S.andsamestateasboth adultswere77 percent ofblackthirtysomething immigrants orchildren ofimmigrants.In 1880, the 1960s,practicallynoAfrican-Americanswere intergenerational residential mobility. Through and A92–A95). a whole (Appendix Figures A68–A71, A80–A83, the samestateasbothparents thanAmericansas less likely to haveand somewhat been born in moresomewhat likelytohave beenbornabroad race andeducation.Non-Hispanic whiteswere emerge whenadultsare considered separatelyby A78, A79,A90,andA91).Interesting differences are generallysimilar(Appendix Figures A66,A67, but theresidential mobilitypatterns than womentohave beenbornabroad, adult menwere more somewhat likely In thedecadesof heavy immigration, mobility significantly. reduced intergenerationalresidential parents. The decline in immigration born there and in the same state as both and children ofimmigrants—hadbeen in theU.S.—includingimmigrants adults that amajorityofthirtysomething Figure 38shows thatitwasonlyin1940 idential mobility during this period. Immigration created substantialres percent in1880to12percent in1940. from bothparents fellby half—from 24 contrast, theshare borninadifferent state That share rose to74percent by 1940.In same stateasbothparents (Figure 39). them (aged 30–39) had beenborn in the over time.Asearlyas1880,64percent of 51 African-Americanshadmuchless Microdata Series Source: Author’s calculationsbasedondatafrom decennialcensuses,Integrated PublicUse 100% Figure 39.Parent andChildBirthplaces,Children Bornin 10% 20% 30% 40% 50% 60% 70% 80% 90% 0% 1880 the U.S.withU.S.–BornParents, 1880–1940 64 12 24 Born inU.S.,SameStateasBothParentsBorn inU.S.,ParentsOnlyOneSameStateasChild Born inU.S.,BothParentsU.S.butDifferentBorn Statesfrom Child abroad fellfrom 64percent to20percent. and 1960,theshare adultsborn ofthirtysomething however, declinedtothepointwhere, between 1920 majority of Hispanic adults. Hispanic immigration, parents. From 1910to1940,theforeign-born were a those born in the U.S. and in the same state as both born Hispanic outnumbered thirtysomethings born abroad. However, from foreign- 1910forward, same stateasbothparents stilloutnumbered those the numberofHispanics bornintheU.S.and in the percent, compared with 27 percent of whites). But likely thanblacksandwhitestobebornabroad (40 whites andblacks.In 1880,Hispanics were more The Hispanic patterndiffersfrom thoseofboth almost entirely ofinterstatemoves. mobility among African-Americans consisted because oftheabsenceimmigration,residential while the1940level forblackswas83percent. Still, percent—level forwhitethirtysomethings—72 parents, as of1880. That was higher than the1940 1890 1900 64 14 22 1910 67 14 19 1920 70 14 16 1930 72 14 14 1940 74 14 12 Among the native-born with native-born parents, Among the native-born children of native-born Hispanics resemble blacks during this period in adults in 1940, residential mobility rose with educa- terms of their low residential mobility: fully 89 per- tional attainment. For those without a high school cent of these thirtysomething adults were born in diploma (two-thirds of this group), 76 percent were the same state as both parents in 1880, and 77 per- born in the same state as both parents, while the fig- cent were in 1940. The trend for Hispanics was de- ures were 68 percent and 63 percent for high school clining, however, unlike among blacks and whites. graduates and college graduates, respectively (who constituted 25 percent and 6 percent of American- Among nonwhite, nonblack, and non-Hispanic born thirtysomethings with American-born par- adults—primarily Asian-Americans in recent de- ents, respectively) (Appendix Figures A74–A77, cades, but including Native Americans, too—the A86–A89, and A98–A101).53 residential mobility story is dominated by sharply declining immigration. Nearly all adults in this In sum, intergenerational residential mobility fell in “other” category were born abroad in 1860, the first decades of the twentieth century, though falling to 54 percent by 1910 and 1 percent by the pattern differed by race. It was generally low- 1960.52 Among the native-born adults of native- er among nonwhites and those less educated than born parents, these thirtysomethings were mostly among whites and those more educated. immobile, with 75 percent having been born in the same state as both parents in 1940 (though, as Birth-to-Adulthood Residential Mobility with Hispanics, down from nineteenth-century Consistent with the trends for intergenerational levels). (By 1940, it should be noted, just 1.6 per- residential mobility, the birth-to-adulthood mea- cent of thirtysomething Americans were Hispanic sure of mobility indicates that American thirty- and just 0.3 percent were something other than somethings were less mobile in 1940 than in the white, black, or Hispanic.) second half of the nineteenth century. Figure 40 shows trends in birth-to-adulthood mobility for Data on educational attainment are not available all thirtysomething adults, while Figure 41 dis- before 1940, but there is information on whether plays the same information but excludes thirty- respondents can read and whether they can write somethings born abroad. (in any language) (Appendix Figures A72–A73, A84–A85, and A96–A97). Interpreting these re- In Figure 40, the decline and subsequent increase sults is made complicated by immigration, but in immigration is again apparent. From 1860 to looking just at the native-born children of native- 1920, it was nearly as common for thirtysomething born parents, residential mobility fell between adults to have immigrated to the U.S. as it was 1880 and 1930 among both the literate and the il- for them to have moved from their state of birth. literate—though it was always lower for the illiter- But immigration then dropped over the next 30 ate. In 1920, 91 percent of the illiterate were born years. Only in 2010 did immigration again reach in the same state as both parents, compared with levels comparable with those of 1920. By that 72 percent of the literate. Immobility was higher year, almost a quarter of adults in their thirties among the illiterate in 1880 than among the lit- had been born abroad, and just over a quarter of erate in 1930. Interpreting these trends requires thirtysomethings lived outside their birth state. keeping in mind that the share of American-born Half of adults remained in the state in which they thirtysomethings with American-born parents who were born. That was fewer than at any time since could both read and write rose from 80 percent in the nineteenth century, suggesting historically 1880 to 97 percent in 1930. strong levels of residential mobility.

When Moving Matters 31 32 e21 Report 2 November 2014 Microdata Series Source: Author’s calculationsbasedondatafrom decennialcensusesand2010AmericanCommunitySurvey, IntegratedPublicUse *Some columnsdonotsumto100percent duetorounding. Microdata Series Source: Author’s calculationsbasedondatafrom decennialcensusesand2010 AmericanCommunitySurvey, IntegratedPublicUse *Some columnsdonotsumto100percent duetorounding.

Percent Percent 100 100 10 20 30 40 50 60 70 80 90 10 20 30 40 50 60 70 80 90 0 0 1850 1850 49 24 19 60 10 30 8 Lives inBirthState 1860 1860 10 31 22 29 41 58 7 Figure 40.Birth–to–AdulthoodResidentialMobility, Lives inRegionofBirthbutDifferent State Lives inBirthState Figure 41.Birth–to–AdulthoodResidentialMobility, 1870 1870 Native–Born Thirtysomething Adults, 1850–2010* 10 29 21 27 44 61 8 1880 1880 10 26 20 24 48 64 8 Thirtysomething Adults,1850–2010* 1890 1890 Lives inRegionofBirthbutDifferent State 1900 1900 21 16 22 54 70 9 7 1910 1910 10 20 16 21 55 70 8 1920 1920 16 22 10 20 55 70 8 1930 1930 17 17 11 21 57 68 9 1940 1940 Lives inDifferent Regionfrom BirthRegion Born AbroadBorn 18 10 10 19 64 70 9 1950 1950 22 10 22 65 68 9 4 Lives inDifferent Regionfrom BirthRegion 1960 1960 24 10 25 62 65 9 5 1970 1970 25 28 58 63 8 8 9 1980 1980 26 10 29 56 62 9 9 1990 1990 24 12 27 56 63 8 9 2000 2000 22 18 10 27 52 63 8 2010 2010 20 24 26 49 64 7 9 Focusing only on native-born adults, the changes among native-born Americans. However, the results over time are less pronounced (Figure 41). Mobility, in the next section on recent (preceding-five-years) in the sense of living outside one’s birth state, fell residential mobility complicate this view. during the late nineteenth century, rose back to pre- 1900 levels by 1970, and then remained more or Appendix Figures A102–A125 provide trends for less flat. In 2010, nearly two-thirds of native-born the same subgroups analyzed above. The discussion thirtysomethings lived in their birth state, about the here will focus on Figures A114–A125, which show same as in 1960 and in 1880. results for the native-born population. The patterns for men and women (Appendix Figures A114– Over time, changes in mobility arise primarily from A115) are essentially the same as shown in the chart more or fewer adults moving from their birth state for all thirtysomethings shown above (and thus, the to an entirely different region. To be sure, some of same as each other). The patterns for non-Hispanic these “interregional” moves simply involve crossing whites (Appendix Figure A116) are also nearly in- a state border into an adjacent region, where cross- distinguishable from the overall picture. In every ing a different border would have kept a migrant case, mobility bottoms out in 1940 and is largely within the region. But while the distinction in any flat from 1980 to 2010 at nineteenth-century levels. one year between an interregional and intraregional move is ambiguous, the fact that the former ac- In contrast, mobility among native-born blacks was counts for most of the change in mobility over time lowest in 1900 and then rose over the next six decades is nonetheless meaningful. (Appendix Figure A117). This trend reflects the his- toric Great Migration of Southern African-Americans The fall in birth-to-adulthood mobility predates to northern and western cities in search of greater the fall in intergenerational residential mobility freedom and better opportunities. For 50 years, from by about 40 years. One way of thinking about 1930 to 1980, African-American thirtysomethings these patterns is that residential mobility became had higher rates of mobility than did whites. less common generally during the calendar years 1860 to 1900. Thus, we observe adults in their But unlike the white trend, which leveled off, black thirties with less mobility since birth in 1900 mobility began falling after 1960 or 1970. By 2010, than 40 years earlier, and we observe adults in 70 percent of black thirtysomethings lived in the their thirties in 1940 (who would have been born same state in which they were born, compared with around 1905, on average) having less mobility 62 percent of whites. Nevertheless, because histori- between their birthplace and those of their par- cally black residential mobility was relatively low, ents than adults in 1900 (who would have been the 2010 level was comparable with levels in the born around 1865). nineteenth century, as was also the case for whites.

The same nineteenth-century years that featured The Hispanic trend differs markedly from the white falling birth-to-adulthood mobility among adults and black patterns (Appendix Figure A118). Resi- approaching middle age also saw declines in mobil- dential mobility among native-born Hispanics has ity among parents and would-be parents that would always been low, reflecting their relative concentra- show up decades later as falling intergenerational tion (until recently) in a few parts of the country. residential mobility for their children. This interpre- In 1850, 90 percent of these thirtysomethings lived tation would suggest that the 1900–40 period that in the same state in which they were born. That fell followed the nineteenth-century drop in mobility to 68 percent by 1950, the first and only time that may have featured uniquely low residential mobility Hispanics had more mobility than non-Hispanic

When Moving Matters 33 34 e21 Report 2 November 2014 just ahighschooldiploma hadlower mobilityrates were backtotheir 1940mobilitylevel, thosewith without a high school diploma thirtysomethings to fallmodestlyamongallthree groups. By 2010, Beginning inabout1975,residential mobility began cent to50percent between 1940and1970. the share state fell from living in their birth 60 per- However, amongcollegegraduatesintheirthirties, degree, itfellonlyfrom 66percent to65percent. cent; amonghighschoolgraduateswithoutacollege fellonlyfrom 73percenttheir thirties to67per- statewhilein school dropouts livingintheirbirth tainment, from 1940to1970,theshare ofhigh Examining results disaggregated by educationalat- cent between 1850and1930. over these years, risingfrom 89percent to96per- of theliterate.Literacy, itshouldbenoted,spread state,comparedin theirbirth withjust68percent lived1930, 80percent ofilliteratethirtysomethings er amongtheilliteratethanliterate;in both groups. From mobilitywaslow 1860forward, - fell. Mobility leveled offfrom 1900to1930among mobility through 1900,asinterregional mobility Figures A120–A121).Bothgroups sawadeclinein ate maybetrackedfrom 1850to1930(Appendix Residential mobility amongtheliterateandilliter- rate wasashighitamongwhites. creased among these adults. By 2010, their mobility accelerating between 1960and1980—mobilityin- (Appendix Figure A119).Starting in1930—and ofanyotherracialgroup statethanwastrue birth more inthisgroup lived thirtysomethings intheir much ofthepast150years. From 1910to1970, adults, residential mobilitywasespeciallylow for Finally, among nonwhite,nonblack,non-Hispanic Hispanics relative tointraregional movement. gional mobilitybecameincreasingly for important blacks andwhitesafter1950.Over time,interre- panics fell modestly, remaining lower than among whites. From 1950 to2010,mobilityamongHis- in the same residence over the period plummeted, 79 percent had(back in 1996).But theshare living andin2010, 14, 15,or16,backin1978) asatbirth, adults hadlived inthesamestateasateen(atage13, state.Inbirth 1992,81percent ofnative-born young mobility—that toanotherresidence withinone’s This findingconcealsasizableshiftin residential includes foreign-born adults.) state wasthesamein2010as1992.(Figure 42 adults who,inadolescence,lived intheirbirth Figure 43reveals thattheshare ofnative-born statewasfairlyconstantfrom 1970to2010,birth that theshare ofnative-born adultslivingintheir two distinctstages.LikeFigure 41,whichshowed measures ofmobilityandsplitthemintothese it ispossibletotakethebirth-to-adulthood adolescence andadulthood.Using theBLSdata, moves andadolescenceorbetween between birth generallyprovidesurveys littleinformationabout Typically available questionsincensusesand Birth-to-Adolescence Residential Mobility in mobilityerasedover thisperiod. exception, havingseen mostoftheirearlierincrease college graduates.African-Americansare againthe cational attainmentandhighermobilityamong at relatively highlevels, thankstoincreases inedu- dential mobilityhasheldsteadyforthemostpart earlier andstoppedsooner).In recent decades,resi- (though theAfrican-Americanratesbeganrising increased amongallthegroups considered above period inwhichresidential mobilityrose, andit from 1940to1980represent theonlysustained tial mobilityintheUnited States. The fourdecades In sum,1940wasprobably alow pointforresiden- shifting peopleintohigher-mobilityeducationlevels. the factthat Americans’ educational attainment rose, reflects mobility among collegegraduates and partly duetoincreasingbetween 1940and2010ispartly residential mobilityrose inthenationasawhole graduates were backto1950levels. The factthat than in1940,andmobilityratesamongcollege

Figure 42. Distribution of Adolescents from 51 percent to 21 percent. This increased by Residential Mobility Between Birth and mobility may be more harmful than beneficial; Adolescence, 1992 and 2010* earlier in this paper, relative and absolute economic mobility outcomes tended to be a bit better for 100 3 4 those not having moved by adolescence than for 90 11 11 those who had moved elsewhere in the state, though 80 7 8 the differences were usually small. 70 29 60 Appendix Figures A126–A151 display residential

50 57 mobility patterns for subgroups. Focusing on the Percent 40 native-born, women were less likely than men to have

30 moved out of state by adolescence in 1992 but more 49 likely to have done so by 2010, though the relatively 20 small sample sizes do not statistically preclude the 10 20 possibility that the share was constant for men and 0 1992 2010 women alike (Appendix Figures A139–A140).

Same Residence at Birth Different Region at Birth Same State at Birth Born Abroad African-Americans were far less likely than Same Region at Birth whites in 1992 to have moved between birth and *1992 column does not sum to 100 percent due to rounding. adolescence, 31 percent to 52 percent, respectively Source: Author’s calculations based on data from National Longitudinal (Appendix Figures A141–A142).54 In 2010, the two Survey of Youth 1979 and National Longitudinal Survey of Youth 1997 groups were equally likely to have moved between birth and adolescence, and both were much more Figure 43. Distribution of Native–Born likely to have moved than in 1992. Blacks were less Adolescents by Residential Mobility Between likely than whites, in both years, to have moved out Birth and Adolescence, 1992 and 2010 of state between birth and adolescence. Hispanics were the only group to become more likely to have 100 12 12 remained in the same state in adolescence as at birth 90 7 8 (Appendix Figure A143). Like African-Americans, 80 Hispanics in 2010 were less likely than whites to 70 30 have moved out of state. 60 50 59 Appendix Figures A144–A147 reveal that residential Percent 40 mobility between birth and adolescence increased

30 as a parent’s educational attainment rose, for adults 51 55 20 in both 1992 and 2010. (Note that parental education is being considered here, unlike in the 10 21 analyses using the Census Bureau data.) However, 0 1992 2010 some convergence across educational groups

Same Residence at Birth Same Region at Birth occurred over the period. By 2010, adults in all Same State at Birth Different Region at Birth four categories of parental education were equally

Source: Author’s calculations based on data from National Longitudinal likely to have remained in their birth residence in Survey of Youth 1979 and National Longitudinal Survey of Youth 1997 adolescence (and all were less likely to have done so than adults in 1992 had been).

When Moving Matters 35 36 e21 Report 2 November 2014 increases withparental educationlevels; thesegaps attainment. The extentofresidential mobility adults withdifferent levels ofparental educational Appendix Figures A157–A160displayresults for they were in1992(Appendix Figures A154–A156). moved outsidetheiradolescentcountyin2010than blacks and Hispanics also were more likelyto have away from theiradolescentresidence. Nevertheless, less likelythanwhitestomove andtomove farther American and Hispanic adults in both years were county (Appendix Figures A152–A153).African- ly in2010tohave moved outsidetheiradolescent census tractasyoung adults;womenwere more like- women inbothyears toremain intheiradolescent Figures A152–A164.Men were more likelythan The subgroup breakdowns are shown inAppendix moving toanothercountywithinthestate. shift from moving withinone’s adolescentcountyto the samein2010as1992,thoughthere wasa in adolescence.Overall, thepicture looksmuch remained inthecensustractwhichtheylived to a state outside the region. Just 15 percent adolescent, andnomore than15percent moved moved from thestateinwhichtheyresided asan Fewer than25percent ofadultsinbothyears were intheU.S.asadolescents. data, including adults who were born abroad but careers. andprofessional service, education, military to adulthood,includingdecisionsabouthigher residential mobility:itencompassesthetransition might beconsidered for especiallyimportant The periodbetween adolescenceandadulthood Adolescence-to-Adulthood Residential Mobility residencebirth (Appendix Figures A148–A151). not, however, lesslikelytohave remained intheir in 2010 than in 1992. that was less true They were andadolescenceinbothyears,between though birth income experienced more residential mobility Adults ofchildhood raised in the richest quartile Figure 44displaysresults from theBLS based gapswidenedover time. income. Further, many of these childhood-income- for residential mobility to increase with parental income, there isarough tendencyinboth years ofchildhood second-richest, andrichestquarters adulthood increased for those raised in the poorest, While residential mobilitybetween adolescence and (Appendix Figurestell asimilarstory A161–A164). Residential mobilitypatternsby childhoodincome 25 percent ofthechildren ofcollegegraduates. asadults, compared region with new ofthecountry the children ofhighschooldropouts hadmoved toa figures were 8percent and37percent; 9percent of the children ofcollegegraduatesin2010,those and 65percent remained inthesamecounty, among remained inthesamecensustract,asadolescence, 20 percent ofthechildren ofhighschooldropouts elsebecamealittlelessso.while everyone While of collegegraduatesbecamealittlemore mobile, do notconverge over time.If anything,children Survey ofYouth 1979andNationalLongitudinalSurveyofYouth 1997 Source: Author’s calculationsbasedondatafrom NationalLongitudinal *Columns donotsumto100percent duetorounding.

Percent Residential MobilitySinceAdolescence, 100 Figure 44.Distribution ofAdultsby 10 20 30 40 50 60 70 80 90 0 Same Tract Same State Same County 1992 and2010* 1992 15 40 23 10 13 Same Region Different Region 2010 15 36 26 15 9 The two sets of BLS estimates—one Figure 45. Residential Mobility, Past Five Years, for residential mobility from birth Thirtysomething Adults, 1940–2000* to adolescence and the other for

100 1 2 4 2 3 4 adolescence-to-adulthood residential 6 11 2 90 12 11 11 mobility—reinforce the results from 12 the Census Bureau analyses showing 80 that residential mobility between birth 70 and adulthood has not declined in 46 39 45 45 45 60 61 recent decades. The BLS results indicate 50 that more disadvantaged groups tend to Percent 40 have less extensive residential mobility than more advantaged groups, though 30 this inequality has not worsened across 20 41 43 41 41 40 31 the board. 10

0 Residential Mobility in the Past Five 1940 1950 1960 1970 1980 1990 2000 Years Moved, Details Not Available Moved Within State Moved from Abroad Same House as Five Years Ago Recall that birth-to-adulthood residential Moved to Different State mobility fell from 1860 to 1900, while *Some columns do not sum to 100 due to rounding. Americans were not asked about where intergenerational residential mobility they lived five years earlier in 1950. fell from 1900 to 1940. The speculative Source: Author’s calculations based on data from decennial censuses, Integrated Public Use Microdata Series interpretation offered was that calendar years 1860 to 1900 featured declining mobility that translated into the Figure 46. Residential Mobility, Past Five Years, 1900–1940 decline in the likelihood Thirtysomething Adults in the U.S. that thirtysomething adults had been Five Years Earlier, 1940–2000* born in a state different from that of

100 their parents. If true, the period of flat 6 4 11 12 11 11 birth-to-adulthood mobility from 1900 90 12 to 1940 would mark a low point for 80 residential mobility in America over the 70 past 165 years. The short-run mobility 47 46 47 47 62 40 60 trends presented in this section, however,

50 are inconsistent with that conclusion. Percent 40 Figure 45 shows short-run mobility 30 with recent immigrants included. 20 42 44 42 42 42 32 Figure 46 omits adults who were living 10 abroad five years earlier. Mobility, in the

0 sense of living in a different home from 1940 1950 1960 1970 1980 1990 2000 five years earlier, fell between 1940 and Moved, Details Not Available Moved Within State 1960, fell a bit more in 1970, and rose Moved to Different State Same House as Five Years Ago slightly from 1970 to 1990, leveling off *Americans were not asked about where they lived five years earlier in 1950. by 2000. The changes between 1960 Source: Author’s calculations based on data from decennial censuses, Integrated Public Use Microdata Series

When Moving Matters 37 38 e21 Report 2 November 2014 percent and40percent by 1970 but fell to 43 Figures A175–A176). That gapwidenedto48 compared with 30percent ofmen (Appendix five years earlier were livinginthesamehome, womenwhohadbeenintheU.S.thirtysomething gender gap is narrowing. In 1940,34 percent of than menonthisdimension,though demographic groups. Women have less mobility breakdown ofthesemobilityresults forvarious Appendix Figures A165–A184provide theusual as theyhadfive years earlier. earlier, butjust42percent lived inthesamehome and 89percent lived inthesamestateasfive years state, stilllived intheirbirth thirtysomethings years earlier. By 2000,63percent ofnative-born just 32percent lived inthesamehomeasfive state asfive years earlier(Figure 46). However, born (Figure 41),and94percent lived inthesame stilllived in the stateinwhichthey’dthirties been (Figure 39),70percent ofAmericansintheir been borninthesamestateasboththeirparents 74 percent had ofnative-born thirtysomethings throughout thesedecades.In 1940,forinstance, degree ofwithin-stateresidential mobility Above all,Figures 45and46indicateasurprising person’s current region. statewithina thananew region ofthecountry much more likelytoleadresidence inanew Not shown inthefigures, out-of-statemoves are with 11percent whomoved toadifferent one. homeintheirexistingstate,compareda new movedcent ofestablishedthirtysomethings to cross statelines.In 2000,forinstance,47per- within aperson’s current statethantheyare to Recent moves are muchmore likelytooccur 2000 distributionsare essentiallythesame. and 42percent in2000. The 1960,1990,and years earlier, compared with42percent in1960 Americans lived inthesamehomeastheydidfive in 1940,32percent ofestablishedthirtysomething and 2000are fairlysmall.According toFigure 46, graduates thanotheradults. among whitesthannonwhites andamongcollege recent decades. Interstate mobilityishigher five years isquitehighandhasnotfallenin In residential short, mobilityinthepreceding bility wasaboutat1960levels in2010. lege graduates.Except forcollege graduates,mo- graduates, 1980fordropouts, and1990forcol- dential mobilityhitlows in1970forhighschool out oftheirhome,except in1980.Recent resi- likely thanhighschooldropouts tohave moved their peers.High schoolgraduateswere nomore moved from adifferent statecompared with the greater likelihoodthatcollegegraduateshad outs did. The difference waslargelybecauseof graduates and44percent ofhighschooldrop- years earlier, while45percent ofhighschool lege graduateslived inthesameresidence asfive col- 2010, only34percent ofthirtysomething other adults (Appendix Figures A181–A184). In common among college graduates than among residentialShort-run mobility is much more again (Appendix Figure A180). low in1960,increases until1980,andthenfalls black, non-Hispanics isunique—mobilityhitsa among whites. The patternamongnonwhite,non- with blacks,interstatemobilityislower thanitis same levels aswhites(Appendix Figure A179).As mobility fallsuntil1970andthenstabilizes atthe whites (Appendix Figures A177–A178).Hispanic generally lesscommonamongblacksthan 2000 wasatits1960levels. Interstate migrationis whites andadultsgenerally, residential mobilityin tomed outin1980ratherthan1970,butasfor Among African-Americans,recent mobilitybot- women ordelayed childbearingandmarriage. growth ingraduateandprofessional degrees among This increase couldhave somethingtodowiththe 1970 and2000,itgrew modestlyamongwomen. mobility didnotchangemuchbetween 1960or percent and41percent in2000. While men’s CONCLUSION middle-class incomes. Noah is skeptical of the efficiency of labor and housing markets but also The surprising conclusion from this paper’s analyses emphasizes what is likely to be a crucial market is that America’s residential mobility problem ap- distortion, the absence of which would likely ren- pears not to be that residential mobility has declined der rising inequality irrelevant: exclusionary zon- but that it has grown more important and is lower ing and other housing regulation that limits the among disadvantaged groups. construction of affordable housing.

Past research concluding that residential mobility From the Right, an essay in National Affairs by has declined since the mid-twentieth century, the Eli Lehrer and Lori Sanders emphasizes the pro- 1980s, or the 1990s generally highlights moves over liferation of federal safety-net programs with state- the past year, including those within states or coun- varying eligibility rules, such as Medicaid and the ties. These kinds of residential mobility have the Supplemental Nutrition Assistance Program (dis- weakest link to the adult outcomes examined here. ability benefits could be added to their list), as The types of mobility that have the strongest asso- well as a disproportionate emphasis on place-based ciation with adult outcomes involve residential mo- antipoverty policies at the expense of ones that bility that occurs between generations or between promote residential mobility.58 (Brian Cadena and childhood and adulthood, as well as mobility across Brian Kovak instead cite unemployment insurance not just states but regions. Those kinds of mobility as inhibiting mobility, since it reduces the pressure are not especially low. to find work elsewhere.)59

Nevertheless, because the gaps in adult outcomes Lehrer and Sanders also point to federal policies between people who do and do not make such that benefit homeowners instead of renters. Home- moves have generally widened and because those ownership discourages residential mobility. While with lower economic mobility also have less resi- Noah points out that mobility has fallen for own- dential mobility, crafting policies to promote resi- ers and renters alike, the increase in homeowner- dential mobility should be an important national ship might still have been an important cause of priority. Before turning to policy options, it is use- declining mobility if homeowners are, in fact, less ful to briefly review what is known about the fac- mobile than renters. Finally, Lehrer and Sanders tors behind the mobility declines that have been highlight two areas also emphasized by Noah—in- found in past research. sufficient public transportation networks and local housing regulations. From the Left, Timothy Noah reviewed some of the factors in the Washington Monthly in 2013.56 Several scholars have found that a decline in varia- Noah considers the aging of the baby boomers, the tion across local and state job markets is behind increase in two-earner couples, the recent bursting the drop in residential mobility. Places are less of the housing bubble, the greater opportunities for unique than they once were, the argument goes, so telecommuting offered by the Internet, improved lo- moving away offers less than it used to for workers cal job opportunities, and changes in state taxes. He seeking a change.60 convincingly argues that these possible explanations are unlikely to be major factors behind the decline.57 A number of other societal and economic chang- Less convincingly, he highlights rising income in- es—or continuities—could also be behind falling equality as an important factor, noting the widen- residential mobility. It is possible that the increase ing gap between housing costs and working- and in immigration has filled local demand for labor

When Moving Matters 39 40 e21 Report 2 November 2014 in themid-twentieth century, andthereturns to broad middleclassismuchbetter off thanitwas risk,oreven anecessity.as aworthwhile Today, the improve one’s circumstances may have been viewed standards were muchlower, to anyopportunity not prioritize professional growth. When living residential mobility, atleastamongthosewhodo Finally, itmaybethataffluenceitselfdiscourages strong local networks offamilyandfriends. states, especiallytotheextentthattheywillgive up in predominantly whiteneighborhoods,cities,and may beconcernedabouthow theywillbeaccepted crimination. Racialminoritiesconsideringamove the perception—correct or not—of prevalent dis- cial discriminationinhousingandhiringby Residential mobility isalsolikelyinhibitedby ra- pecially dauntingtothem. across statelines—oreven neighborhoods—ises- works offamilyandfriendsforhelp, soamove pressed mobility. Single parents rely more onnet- crease insingleparenthood couldalsohave de- careers thanitisforonespousetodoso—thein- it ismore their difficultfortwopeopletodisrupt be expectedtoreduce residential mobility—since While theincrease oftwo-worker familiesmight licensed allover again. to move toanotherstatewhere theywillhave tobe licensing requirements inonestatewillbereluctant recent decades. Workers whohave fulfilledarduous lines istheexpansionofoccupationallicensingin tentially affectedresidential mobilityacross state thathaspo- market distortion Another important graphically concentrated. mobility totheextentthattheytendbegeo- alsohave dampenedresidentialstart-ups—might evident, forinstance,inthefallingnumberof decline in innovation and economic dynamism— wages toareas ofjobcreation. The documented among native-born Americansinsearch ofhigher have encouragedmigration that wouldotherwise encourage markets forhomeequityinsurance. Another waytocounterNIMBYismwouldbe NIMBYism lessrewarding. fewer peopleinhigh-housing-costareas andmakes interestmortgage deduction, so thatit benefits incremental reform wouldbetosimplycapthe people tomove tohigher-productivity areas. One affect their homevalues but allow lower-income able housingdevelopment thatwouldpotentially to amassenoughpoliticalpower toblockafford- would dictate,we makeiteasierforhomeowners By makinghomeownership cheaperthanmarkets sponsored enterprisesinvolved inhousing. subsidies toFannie Mae andtheothergovernment- totheimplicit ministration insuranceofmortgages interestmortgage paymentstoFederal Housing Ad - for homeownership, from thetaxdeductionfor forms wouldreduce oreliminatefederalsubsidies licensing, whichhasexploded inrecent decades. mobility by reducing theamountofoccupational Deregulation could also help promote residential localities toderegulation oflanduseandhousing. the federalgovernment couldtieaidtostatesand that lowered thevalue oftheirproperty. In addition, stand tomakelessbecauseoflocaldevelopments This insurancewouldcompensatehomesellerswho areas isespeciallyimportant. drive up the cost of housing in high-productivity Addressing housingandland-useregulations that mensions appearstobegrowing. alongsomedi- tunity andbecauseitsimportance of itbecauseappearsrelated toeconomicoppor- on thedecline,policiesshouldaimtopromote more Regardless of whetherU.S. residential mobility is wagons—to improve ouropportunities? by westward-migrating pioneers—intheircovered ofriskstaken of ustodaywouldconsiderthesorts out andthelossofsocialnetworks). How many entailed (includingtheriskofthingsnotworking moving maynotbesufficientlyhigh,given thecosts 61 One setofpolicyre- 63 62

The federal government could promote certification Even policies in seemingly unrelated areas could pro- as an alternative to licensing, tie federal aid to de- mote more residential mobility. Deregulation, cor- regulation, provide federal funding and incentives porate tax reform, and intellectual property reform for states to develop common regional licensing could promote economic dynamism, which might agreements, and encourage the “sharing economy” create local and regional centers of innovation that as a way of subverting licensing (Uber’s weakening would encourage migration to high-productivity of the taxicab cartel is a prime example). areas.65 Rebalancing immigration policy so that it is less focused on family reunification—which, by As Lehrer and Sanders emphasize, consolidating increasing the supply of lower-skilled labor might federal antipoverty programs would make state dif- reduce migration of native-born Americans to plac- ferences in eligibility requirements somewhat less of es that would otherwise seek them out—and more a barrier to interstate mobility among low-income focused on skilled workers (who would raise pro- families. Federal policies could also encourage use ductivity levels and reduce inequality) could affect of housing vouchers to move between local hous- residential mobility. ing authorities rather than within them. Unemploy- ment insurance could be reformed to allow people Promoting planned and marital childbearing—by to take a lump sum—dubbed a “mobility grant” by eliminating marriage penalties in the tax code and Lehrer and Sanders—to pay for relocation expenses providing tax incentives to married parents—could instead of receiving monthly payments that dis- reduce single parenthood, which might increase courage job search. These sorts of safety-net reforms mobility.66 Policies to increase college graduation would have the advantage of being relatively acces- rates (and reduce dropout rates) would also pro- sible levers for federal policymakers. mote interstate and interregional migration.

Changing federal safety-net policies is a more direct Given the provisional evidence in this paper and way of encouraging residential mobility than some elsewhere that Americans who are willing and able of the bank-shot ideas that require the federal gov- to move have better economic outcomes, if we want ernment to persuade state and local policymakers to to increase economic mobility, we should try a vari- implement reforms or that depend on lowering in- ety of approaches to increasing residential mobility. flated housing prices to counter NIMBYism. At the Doing so would almost certainly pay off in higher same time, it may be that the payoff from enacting economic growth rates, too. safety-net reforms, in the form of greater residential mobility, would not be as large as that from other The analyses in this paper suggest that, just as we policies if they could be successfully implemented. should worry less than we do that economic opportunity is on the decline in the U.S., we should More federal, state, and local attention to dis- be less concerned than we are about declining mi- crimination in mortgage lending and among real- gration among Americans. Nevertheless, we should estate agents would also promote more residential strive for more economic mobility—regardless of mobility among African-Americans and other ra- whether it is falling—and, therefore, we should do cial minorities.64 more to increase residential mobility.

When Moving Matters 41 42 e21 Report 2 November 2014 Endnotes

1. Pew Economic Mobility Project (2012). Pursuing the American Dream: Economic Mobility Across Generations (Wash- ington, D.C.: Pew Charitable Trusts). 2. On poverty trends, see Christopher Wimer et al. (2013). “Trends in Poverty with an Anchored Supplemental Poverty Measure.” Working paper; Bruce D. Meyer and James X. Sullivan (2012). “Winning the War: Poverty from the Great Society to the Great Recession.” Brookings Papers on Economic Activity 45(2): 133–200. 3. Robert Sampson (2012). The Great American City: Chicago and the Enduring Neighborhood Effect (Chicago: Univer- sity of Chicago Press); Patrick Sharkey (2013). Stuck in Place: Urban Neighborhoods and the End of Progress Toward Racial Equality (Chicago: University of Chicago Press). 4. Ichiro Kawachi and Lisa F. Berkman, eds. (2003). Neighborhoods and Health (Oxford: Oxford University Press). 5. Peter Ganong and Daniel Shoag (2014). “Why Has Regional Income Convergence in the U.S. Declined?.” Working paper. 6. Enrico Moretti (2012). The New Geography of Jobs (New York: Houghton Mifflin Harcourt). 7. For a review of the neighborhoods literature, see Patrick Sharkey and Jacob W. Faber (2014). “Where, When, Why, and for Whom Do Residential Contexts Matter? Moving Away from the Dichotomous Understanding of Neigh- borhood Effects.” Annual Review of Sociology 40: 559–79; Robert J. Sampson, Jeffrey D. Morenoff, and Thomas Gannon-Rowley (2002). “Assessing ‘Neighborhood Effects’: Social Processes and New Directions in Research.” Annual Review of Sociology 28: 443–78; Narayan Sastry (2012). “Neighborhood Effects on Children’s Achievement: A Review of Recent Research.” In Rosalind King and Valerie Maholmes, eds. The Oxford Handbook of Poverty and Child Development (Oxford: Oxford University Press); Steven N. Durlauf (2004). “Neighborhood Effects.” In J. V. Hen- derson and J. F. Thisse, eds. Handbook of Regional and Urban Economics, vol. 4 (Elsevier). Notable neighborhoods research that uses nonexperimental evidence includes Jonathan T. Rothwell and Douglas S. Massey (2015). “Geographic Effects on Intergenerational Income Mobility.” Economic Geography 91(1): 83–106; Geoffrey T. Wodtke, David J. Harding, and Felix Elwert (2011). “Neighborhood Effects in Temporal Perspective: The Impact of Long-Term Exposure to Concentrated Disadvantage on High School Graduation.” American Sociological Review 76: 713–36; Patrick Sharkey (2012). “An Alternative Approach to Addressing Selection into and out of Social Settings: Neighborhood Change and African-American Children’s Economic Outcomes.” Sociological Methods and Research 41(2): 251–93; Robert J. Sampson (2008). “Moving to Inequality: Neighborhood Effects and Experiments Meet Social Structure.” American Journal of Sociology 114(1): 189–231. 8. Rothwell and Massey (2015) review this literature. They also note that one study finds a positive effect on adult earnings from having been randomly assigned to a better kindergarten class. See Raj Chetty and John N. Friedman (2011). “Does Local Tax Financing of Public Schools Perpetuate Inequality?.” National Tax Association Proceedings 103: 112–18. 9. Olof Åslund et al. (2011). “Peers, Neighborhoods, and Immigrant Student Achievement: Evidence from a Placement Policy.” American Economic Journal: Applied Econometrics 3(2): 67–95. Eric D. Gould, Victor Lavy, and M. Daniele Paserman (2004). “Immigrating to Opportunity: Estimating the Effect of School Quality Using a Natural Experiment on Ethiopians in Israel.” Quarterly Journal of Economics 119: 489–526. Similar quasi-random refugee studies have found evidence of other “neighborhood effects” on welfare use, crime, and labor-market outcomes. Olof Åslund and Peter Fredriksson (2009). “Peer Effects in Welfare Dependence: Quasi-Experimental Evidence.” Journal of Human Resources 44(3): 798–825. Anna Piil Damm (2009). “Ethnic Enclaves and Immigrant Labor Market Outcomes: Quasi-Experimental Evidence.” Journal of Labor Economics 27: 281–314. Anna Piil Damm (2014). “Neighborhood Quality and Labor Market Outcomes: Evidence from a Quasi- Random Neighborhood Assignment of Immigrants.” Journal of Urban Economics 79: 139–66. Anna Piil Damm and

When Moving Matters 43 44 e21 Report 2 November 2014 18. 17. 16. 15. 14. 13. 12. 11. 10. States Since1850.”JournalofEconomicPerspectives 19(3):199–215. 87(2): 217–34;JosephP. Ferrie(2005).“HistoryLessons:TheEndofAmerican Exceptionalism?MobilityintheUnited Sacerdote (2005).“SlaveryandtheIntergenerationalTransmission ofCapital.”ReviewEconomicsandStatistics 1109–1137; ClaudiaOlivettiandM.DanielePaserman(2015). AmericanEconomicReview105(8):2695–2724;Bruce “Intergenerational OccupationalMobilityinBritainandtheU.S. Since1850.”AmericanEconomicReview103(4): Mobility inBritainandtheU.S.Three Eras.”EconomicJournal117(519):C61–C71;LongandFerrie(2013). point inhistory. SeeJasonLongandJosephFerrie(2007).“ThePathtoConvergence: IntergenerationalOccupational research hasusedthedatasetsbetween1850and1940toexaminechangesinoccupationalmobilityupthis censuses are availableis1940,whichwasthefirstyearinpeoplewere askedabouttheirincomes.Some across censusyearsusingtheirnames,birthyears,andotherinformation.Themostrecent yearforwhichentire After 70years,completecensusrecords are madepublic,makingitpossibletoattemptlinkthesamepeople Dakota, SouthVermont, andWyoming. data setthatIuseexcludesstateswithoutlargemetropolitan areas, includingDelaware, Idaho, Montana,North data; andAlaskaHawaii(admitted1959)are absentfrom 1850through 1900,in1940, andin1950.The1970 mitted 1890)isabsentfrom 1850 and1860;Oklahoma(admitted1907)isabsentfrom the1850,1870,and1880 In thecensuses,contemporarystatesare notalwaysincludedbefore theywere admittedtotheunion. Wyoming(ad- Exposure Effects andCountyLevelEstimates.”Working paper. Raj ChettyandNathanielHendren (2015b).“TheEffects ofNeighborhoodsonIntergenerationalMobility:Childhood United States.”QuarterlyJournalofEconomics129(4):1553–1623. Raj Chettyetal.(2014).“Where IstheLandofOpportunity?TheGeographyIntergenerationalMobilityin Children: NewEvidencefrom theMovingtoOpportunityExperiment.”AmericanEconomicReview(forthcoming). Raj Chetty, Nathaniel Hendren, andLawrence F. Katz(2015a).“TheEffects ofExposure toBetterNeighborhoodson AJS-Ludwig-2008.pdf. Experiment?.” AmericanJournalofSociology114(1):144–88.Available athttp://home.uchicago.edu/ludwigj/papers/ Jens Ludwigetal.(2008).“WhatCanWe AboutNeighborhoodEffects Learn from theMovingtoOpportunity Youth: Evidence from Gautreaux, Chicago.”SocialScienceandMedicine 68(5):814–23. Mark Votruba andJeffrey R.Kling(2009).“Effects ofNeighborhoodCharacteristicsontheMortalityBlack Male assigned toneighborhoodswithouttheirowninputorchoice. program didnotinvolverandomized assignmentorincludeacontrol group, butmoverswere, forthemostpart, cago Gautreaux HousingProgram andRecentMobilityPrograms.” IndianaLawReview653:659–62.TheGautreaux 7–25; JamesE.RosenbaumandStefanieDeLuca(2008).“WhatKindsofNeighborhoodsChangeLives?TheChi- Stefanie DeLucaetal.(2010).“Gautreaux MothersandTheirChildren: AnUpdate.”HousingPolicyDebate20(1): for LatinoandAfrican-AmericanChildren(Washington, D.C.:U.S.DepartmentofHousingandUrbanDevelopment). County, Maryland”(NewYork: Century Foundation);AnnaMariaSantiagoetal.(2014).OpportunityNeighborhoods “Housing PolicyIsSchoolPolicy:EconomicallyIntegrativeHousingPromotes AcademicSuccessinMontgomery es ofLivinginaPoorNeighborhood.”QuarterlyJournalEconomics118(4):1533–75.HeatherSchwartz(2010). comes.” JournalofResearchonAdolescence17(2):249–84;PhilipOreopoulos (2003).“TheLong-RunConsequenc- borhood? Long-Term ImpactsofMovingtoLow-PovertyNeighborhoodsonPoorChildren’s andAdolescents’ Out- Place 18(2):180–90;RebeccaC.Fauth,Tama Leventhal,andJeanneBrooks-Gunn (2007).“Welcome totheNeigh- (2012). “NeighborhoodDisorder andAnxietySymptoms:NewEvidencefrom aQuasi-ExperimentalStudy.” Health kers, NewYork; Toronto; MontgomeryCounty, Maryland;andDenver. SeeRebeccaCascianoandDouglasS.Massey Examples ofthisresearch includeexperimentsandcourt-ordered policychangesinMountLaurel, NewJersey;Yon- American EconomicReview104:1806–32. Christian Dustmann(2014).“DoesGrowing UpinaHighCrimeNeighborhoodAffect Youth CriminalBehavior?.” 19. Daniel Aaronson and Bhashkar Mazumder (2005). “Intergenerational Economic Mobility in the United States, 1940 to 2000.” Journal of Human Resources 43(1): 139–72. 20. Aaronson and Mazumder’s measure is analogous to the conventional summary measure in mobility studies, the “intergenerational elasticity,” or IGE. The IGE is the coefficient on logged parental income in predicting the logged incomes of adult children. It indicates the extent to which economic inequalities between children persist into adulthood. If the IGE is 0.5, that means that a 50 percent difference in family income between two children typically leads to a 22 percent difference in earnings when the children are adults. When the IGE is 0.0, childhood gaps typically dissolve entirely by adulthood. When the IGE is less than 0.0, poorer children often end up better off than richer children. When the IGE is greater than 1.0, initial gaps between rich and poor children widen, rather than shrink, in adulthood. While much depends on the methodological details used, the IGE typically ranges between 0.2 and 0.7 in the U.S. and other countries (in the U.S., between 0.4 and 0.7). 21. By giving adults the average family income of children born around the same time and in the same state, Aaronson and Mazumder must discard income variation between children born in the same state and around the same time. That variation is a substantial amount of overall variation in childhood family income. In my analyses, typically 85 percent or more of the variation occurred within birth cohorts and birth states. Worse, the share of overall income variation that occurs within birth state and birth cohort might change over time, which would make estimated trends in the Aaronson-Mazumder measure unreflective of trends in the IGE. My analyses indicated that, over time, the “within” variation in income grew as a share of total variation. Consider the difference between the IGE—computed as the covariance of individual childhood incomes and adult incomes, divided by the variance of childhood incomes—and the Aaronson-Mazumder measure. Their measure is computed as the covariance of individuals’ cohort-birth-state income mean and individual adult incomes, divided by the between-cohort-birth-state income variance. The denominator in that measure becomes smaller relative to the denominator in the IGE as the within-cohort-birth-state variance grows, causing the Aaronson-Mazumder denominator to be biased upward as a measure of the IGE denominator and pushing their trend toward finding rising immobility over time compared with the IGE trend. The numerator in the Aaronson-Mazumder measure is also affected by their substitution of individual parental incomes with cohort-birth-state means, so the extent and direction of bias in their measure as an IGE proxy is indeterminate. Nevertheless, in my analyses, the rank-rank estimates rose over time—indicating falling mobility—not because the covariance in the numerator rose (it fell) but because the variance in the denominator fell even more. In contrast, the variance of individual parental incomes rose over time. If one assumes that the IGE was flat over time, as suggested by past research, then a rising denominator would indicate that the covariance between individual parental incomes and adult incomes rose correspondingly. When I simply held constant the covariance in the Aaronson-Ma- zumder measure used in this report at its 1970 level and let the variance in the denominator grow at the same rate as the variance in individual parental incomes, the resulting IGE fell over time (from 0.59 to 0.46) instead of starting and ending at 0.59. 22. Conceptually, the IGE is a measure of mobility that is affected by changes in inequality, making it an impure measure of the former. When inequality rises more between generations, the IGE will be larger than when inequality rises less or falls, even if the ability to rise from lower ranks to higher ranks in the income distribution has not changed. Com- putationally, the IGE is sensitive to very low incomes and changes in the number of very low incomes reported. The often substantial underreporting of income among poor families can also artificially affect mobility levels and trends. Another problem is that the IGE tends to be sensitive (at least in survey data) to the number of years of income used to estimate parent and child income and to the age at which these incomes are measured.

When Moving Matters 45 46 e21 Report 2 November 2014 25. 24. 23. my samplewhileavoidingthe problem ofchildren whodonotliveathome. the estimatesmore variableovertime(withoutchangingthe1970–2010 trend). Iuse0–14 and30–44tomaximize age groups to0–9and30–39madevirtuallynodifference tomyresults, thoughrestricting to0–4and30–34made and mobilityestimatestherefore tendtobebiaseddownward whenyoungeradultsare notexcluded.Narrowing my prior research hasshownthatincomesbelowage30are poorindicatorsoftheincomesthatadultseventuallyattain, more problematic theolderchild(seen.26below).Ialsochosetorestrict myadultsampletoage30–44because that oftheirparents. Inaddition,thedistinctionbetweenone’s birthstateand one’s stateasofthecensusyearis some children nolongerreside athome19,andtheir“familyincome”willreflect theirownfamilyratherthan include children asold19inimputingparental incomestoadults.Thischoiceispotentiallyproblematic because produced muchlessprecise results, giventhatIdonotpoolasmuchdatatheydo.Aaronson andMazumder five-year—rather than15-year—birth-cohortwindows;Ifound thatcomputingpercentiles usingfive-yearwindows the adultsample).Irestrict thesampletoonechildaged0–14ineachhousehold.Aaronson andMazumderused Minneapolis: MinnesotaPopulationCenter[producer anddistributor],2010.) tion Center. (Steven Rugglesetal.IntegratedPublicUseMicrodata Series:Version 5.0[Machine-readable database]. population from thatyear. AlltheCensusBureau datacomefrom theIPUMS-USAwebsiteofMinnesotaPopula- The datafrom eachyearconstitute a1percent (1940–70,2010)or5percent (1980–2000) sampleofthenational methodological decisionsthattheymade(eachrelying, inpart,onout-of-sampleprojections). results from 1950to2000.However, myanalysesrevealed thattheirestimatesfor1950and1960are sensitivetothe nately, thisallows metoshowtrends foradultsonlyfrom 1970to2010,whereas Aaronson andMazumderpresent parental incomeestimates from thisyearwillbeusedforadultsin2010,thelastdecennialavailable.Unfortu- I beginin1940becauseitisthefirstyearwhichafamily-incomemeasure isavailable.Iendin1980becausethe percentile of40doesnotcorrespond tothesamelevelofincomeovertime. over time.Inmyanalyses,Icomputedifferent childhoodandadultpercentiles foreachcensus, soanexpectedadult tion. Inthatcase,thepercentiles alwayscorrespond toanactualincomelevelthatdoesnotchangeacross groups or ate onlybecauseallcomparisonsare made usingpercentiles basedonasingleincomedistributionineachgenera- percentile ofbirthplaceincome. Chettyetal.(2014)refer tothismeasure as“absolutemobility,” butthatisappropri- regression modelstolookattheexpectedadultincomes(andotheroutcomes),conditionalonstarting25 move, versusaricherchildwhodoesn’t move.Forthisreason, Icombinetherank-rankslopewithconstantinmy a poorer childwhomoveswillgainmore, versusaricherchild whomoves,thanwillapoorer childwhodoesn’t to dobetterthanlow-incomenon-movers.Itcouldbethatallnon-moversmovers,eventhough convergence thanthere isamongnon-movers.Itdoesnottellyouthatalladults,low-incomemoverstend rank slopethanadultswhodidnotchangeresidence, thatsimplysaysamongmovers,there ismore income convergence withinagroup. Ifonefindsthatadultswhoexperienced residential mobilityhavealowerIGEorrank- lematic forassessingwhichgroup hasmore mobility. Thesemeasures examinetheextentofintergenerational income (2014). AswithAaronson andMazumder’s mobilitymeasure, comparingtherank-rankslopeacross groups isprob- measured (thoughnotbelowage30forchildincome)andtothenumberofyearsincomeaveraged.Chettyetal. changes ininequality. Atleastinadministrativedata,itisinsensitivetotheageatwhichparental andchildincomeis (i.e.,halfaslarge).Therank-rankslopeisnotaffectedcentile gapinadultearnings bylowreported incomesorby A rank-rankslopeof0.5indicatesthata50-percentile gapinchildhoodfamilyincometendstoleada25-per Using AdministrativeData.”InstituteforResearch onPovertyDiscussionPaperNo.1342-08.Chettyetal.(2014). Dahl andDeLeire (2008).“TheAssociationBetweenChildren’s Estimates andFathers’LifetimeEarnings: Earnings The rank-rankslopewasusedbyMollyDahlandThomasDeLeire andpopularizedbyChettyhiscolleagues.See I drop children ingroupoutsidethe50statesandWashington quartersandthoseborn D.C.(Idothesamefor In 1940,theonlyincomemeasure availableistotalwageandsalaryincomewithinthehouseholdhead’s family. - th

26. Following Aaronson and Mazumder, I average within the state in which a child resides at the time of the census after restricting to children who remain in their birth state. Children still living in their birth state are unlikely to be representative of all children born in the state. However, since I am interested in an indicator of birthplace circumstances, including for purposes of computing averages, the incomes from other states for children who have moved from their birthplace is a departure from the ideal indicator. Importantly, I compute percentiles across states before computing the average percentile within each state. Computing percentiles within each state would produce an average income percentile of 50 for every state in every birth cohort, which would then be assigned to every thirtysomething adult in every birth cohort in every state. 27. I include men who report $0 in earnings in all these mobility analyses, including in the computation of percentiles. I do the same for $0 reports of family income. Because of the great changes in women’s labor-force participation over time and because many women still remain out of the labor force to care for children, I drop women who report no earnings, including in the computation of percentiles, in the analyses for women. 28. This is a much simpler model than the one used by Aaronson and Mazumder, who additionally fit a quadratic in age, interacted with calendar year and logged income, and include fixed-cohort effects, fixed-year effects, and a cohort- by-income interaction. Their model is more complex because they impute parental incomes from multiple cohort-year cells, whereas cohort and year are equivalent in my setup, and because they are attempting to distinguish age, period, and cohort effects. My setup is selected so that age, period, and cohort are essentially equivalent. The exception is that I have age-by-birth-year variation within 15-year cohorts; but unless this variation changes over time, the results should be unaffected. Effectively, one can view my estimates as showing trends for men aged 30 to 44, from 1970 to 2010, or across cohorts of men aged 30 to 44, born 1926–40 through 1966–80. It should be noted that few mobility-trend studies attempt to distinguish between periods and cohorts. 29. The “earnings” estimates do not include self-employment income. Both earnings and family income are reported before taking out taxes. In the decennial censuses, respondents provide earnings or income for the previous year, so figures for, say, 1940 and 1970 actually refer to earnings or income in 1939 and 1969. In the 2010 ACS, respondents provide amounts for the previous twelve months, which represents some weighted combination of 2009 and 2010 earnings and income, depending on the month in which the household was surveyed. 30. Not shown in the figure, the rank-rank slope itself—a measure of “relative mobility” that is more comparable with conventional estimates—indicates a large increase in mobility between 1970 and 1980, followed by a smaller but sizable decline over the next 30 years. This pattern of falling mobility is in contrast with most of the literature measuring economic mobility between actual parents and children. The majority of these studies find minimal changes in economic mobility over time. See Chetty et al. (2014); Robert T. Reville (1996). “Two Essays on Intergenerational Earnings and Wage Mobility.” Ph.D. diss., Brown University; Mary Corcoran (2001). “Mobility, Persistence, and the Consequences of Poverty for Children: Child and Adult Outcomes.” In Sheldon H. Danziger and Robert H. Haveman, eds. Understanding Poverty (New York and Cambridge, Mass.: Russell Sage Foundation and Harvard University Press); David I. Levine and Bhashkar Mazumder (2002). “Choosing the Right Parents: Changes in the Intergenerational Transmission of Inequality—Between 1980 and the Early 1990s.” Federal Reserve Bank of Chicago Working Paper 2002-08; Angela R. Fertig (2003). “Trends in Intergenerational Earnings Mobility in the United States.” Journal of Income Distribution 12: 108–30; Yunju Nam (2004). “Is America Becoming More Equal for Children? Changes in the Intergenerational Transmission of Low- and High-Income Status.” Social Science Research 33: 187–205; Susan E. Mayer and Leonard M. Lopoo (2005). “Has the Intergenerational Transmission of Economic Status Changed?.” Journal of Human Resources 40(1): 169–85; David J. Harding et al. (2005). “The Changing Effect of Family Background on the Incomes of American Adults.” In Samuel Bowles, Herbert Gintis, and Melissa Osbourne Groves, eds. Unequal Chances: Family Background and Economic Success (New York and Princeton, N.J.: Russell Sage Foundation and Princeton University Press); Tom Hertz (2007). “Trends in the Intergenerational Elasticity of

When Moving Matters 47 48 e21 Report 2 November 2014 33. 32. 31. “Intergenerational IncomeMobility inPermanentandSeparatedFamilies.”EconomicsLetters 80:239–46;AngelaR. fornonresidentmissing earnings fathers mightbiasthetrend. SeeAnders BjorklundandLauraChadwick(2003). nonresident fathersare that unavailable inbothsurveys,andtherisesinglemotherhoodraises theconcern offathersormothersbecause thelatterarewith theearnings unavailable intheNLSY79.Further, of earnings “family.” includewageandsalaryincome,plus self-employmentincome.Icannotcompare Earnings childearnings of actualfamilyincomeinbothsurveys;buttheNLSY97,unmarried partnersoftheheadare includedinthe confined tothefamilymembersofhouseholdheadin NLSY79.Adolescents’adultfamilyincomeconsists includes theincomeofallhouseholdmembersinNLSY97 (includingunmarriedpartnersofthehead),while were aged13–16).Familyincomeisreported byparents inbothsurveysandreflects pretax income.Itactually subpopulations. start atabout3,700–4,000peopleintheNLSY79and3,200–3,500 intheNLSY97butare reduced inlookingat family incomeismeasured, orwho are missinginformationonparental income.Thesamplesizes for theseanalyses the yearinwhichadultincomeswere measured, whodidnotlivewithaparent intheyearwhich “parental” those inthemilitaryandresiding ininstitutions).Iexcludefrom thesamplethoseyouthwhowere inschoolduring relevant birthcohorts,while theNLSY79isrepresentative oftheentire nationalpopulationinthecohorts(including The NLSY97isdesignedtoberepresentative (afterweighting)oftheciviliannoninstitutionalizedpopulationfor This residential mobilityvariable isunavailablein2010. to drawinferences aboutgender gaps. Percentiles are Figure computedtakingintoaccountonlywomenwithearnings. 2cannotbecompared with Figure 1 decline inmobility. Keep inmindthatmobilityhere ismeasured withintheadultfemale population,excludingmen. tion oftheworkforce islikely tosubstantiallyaffect thetrends forwomenandisprobably themainfactorbehind changes infemaleworkforce participation reduce overtime.Still,thechangingcomposi- thenumberofnon-earners In Figure are 2,adultswithno earnings excluded(unlikeinFigure 1).Thismakestrend analysesmore meaningful,as the reasons discussedinthese endnotes,Ifocusontheexpected-incomemobilitymeasure inthissection. Because itisproblematic tocompare therank-rankslopesofdifferent groups, suchasmoversandnon-movers,for attainment ofadultchildren. Changes inthedistributionofincomeandeducationcantherefore affect theresults. mobility,” where the“mobility”measure isnotbasedonranksbutparental incomeandtheeducational For acritiqueoftheClarkbook,seeChettyetal.(2014).TheHilgerstudyisessentiallyanexamination“absolute Longitudinal Study, and inforthcomingwork,Ifindthatmobilitylooksloweragainthemost recent NLSpanel. Mobility Since1940.”NBERWorking PaperNo.21217(Cambridge,Mass.:NationalBureau ofEconomicResearch). Mobility (Princeton,N.J.:PrincetonUniversityPress); NathanielG.Hilger(2015).“TheGreat Escape:Intergenerational Mobility.” SocialForces90(2):375–95.Gregoryof Clark(2014).TheSonAlsoRises:SurnamesandtheHistory 7–21; Deirdre BloomeandBruceWestern (2011).“Cohort Change andRacialDifferences inEducationalandIncome Mazumder (2007).“TheGrowing ImportanceofFamily:Evidencefrom Brothers’IndustrialRelations46(1): Earnings.” out-of-sample extrapolations.Forotherexceptionsfindingdeclinesinmobility, seeDavidI.LevineandBhashkar mobility estimates.Theirestimatesalsomaybeimprecisely estimatedinearlieryears,where theymustmake and Mazumderpaperisanotableexception,but,asnoted,theirmeasure isnotcomparablewithconventional Worsened InequalityofOpportunityintheUnitedStates?.”SocialPhilosophyandPolicy31(2):28–47.TheAaronson little changeormixedevidenceovertime.Forareview, (2015).“HasRisingIncomeInequality seeScottWinship Intergenerational IncomeMobility.” ReviewofEconomicsandStatistics91(4):766–72. Family IncomeintheUnitedStates.”IndustrialRelations46(1):22–50;Chul-InLeeandGarySolon(2009).“Trends in I examinethe1978incomeofNLSY79adolescentsand1996 incomeofNLSY97adolescents(whenthey The LevineandMazumderpaperBloomeWestern paperusethetwoearlierpanelsofNational The literature oneducational“relative mobility”andoccupational “exchangemobility”alsotendtofind Fertig (2007). “Is Intergenerational Earnings Mobility Affected by Divorce?.” Center for Research on Child Well-Being Working Paper 2002-04. Available at http://hogwarts.spia.uga.edu/~afertig/research/Fertig_igemdivorce.pdf. All incomes and earnings are adjusted for inflation using the Bureau of Economic Analysis implicit price deflator for personal consumption expenditures (PCE); dollar amounts are shown at 2014 levels. For the estimates that consider the likelihood of moving up from the poorest quarter of family income or down from the next-poorest quarter, I adjust the family income of parents and adult children for needs by dividing by the square root of family or household size. I top-code income and earnings at the 97th percentile in all years in all surveys. All results include parents who report $0 in income and adult children who report $0 in earnings, the latter of which became more prevalent over time (even after excluding students). 34. I had originally hoped to analyze the relationship between economic and residential mobility with the BLS data for subgroups defined by sex, race, and parental education. Unfortunately, the sample sizes proved insufficient to yield meaningful results. 35. This measure is computed from the constant and regression coefficient in a bivariate regression of adult family income percentile on childhood income percentile. Another way of assessing relative mobility is to focus on the rank- rank slope, which ignores the constant from the bivariate regression. This measure is unhelpful for comparing different groups, however, because it indicates how much family incomes converge between generations, where “convergence” is confined to the particular group. We are generally more interested, for instance, in whether blacks are less able to transcend their family origins than whites are, but the rank-rank slope tells us only the extent to which the children of the richest black families and the poorest black families are converging. They might, as a group, generally be stuck at lower incomes. 36. As discussed in n. 29 (above), the census estimates are actually for 1989 and 2009 rather than 1990 and 2010. Interestingly, the “rank-rank slope” in the BLS data is not consistent with the Census Bureau data. (The expected percentiles analyzed in the paper depend on the rank-rank slope and the “constant” estimated from the regression equations.) While the Census Bureau data show the rank-rank slope rising for men (from 0.46 to 0.56) and women (from 0.46 to 0.62) between 1990 and 2010, the BLS data show it falling for men (from 0.30 to 0.25) and rising only slightly among women (from 0.32 to 0.34). This reinforces that the Aaronson-Mazumder estimates may prove unreli- able as an indicator of trends in how closely parent and child income ranks are connected. 37. There is limited information on geographic areas available in the public-use NLSY79 and NLSY97 data sets. In order to examine residential mobility in these surveys, I turned to restricted-use information from the surveys that included the birth states of survey participants, as well as their state, county, and census tract of residence in each survey year. 38. In these analyses, moves to states that share a border are never categorized as constituting a move between regions (unlike in the Census Bureau analyses). Otherwise, I use the Census Bureau’s nine “divisions” as my “regions,” not its four “regions.” Thus, my regions include (moving southward and westward) New England, Middle Atlantic, South Atlantic, East North Central, East South Central, West North Central, West South Central, Mountain, and Pacific. 39. See http://www.census.gov/hhes/www/income/data/historical/families/2014/f05.xls. 40. See http://www.bls.gov/news.release/history/work_102093.txt; http://www.bls.gov/news.release/archives/ work_12082011.htm. 41. Figures for “all men” and “all women” include the foreign-born. I focus on adults aged 30–39 to partly take life- cycle factors into account that would complicate analyses that included younger and older adults. Adults in their thirties are, by and large, done with schooling and have accumulated significant work experience; they have also generally formed families, though this is less true of more recent cohorts of thirtysomethings. Information for 1890 is unavailable; the census records for that year were destroyed by fire in 1921. For parental birthplaces, the 1950 census distinguishes only between Alaska, Hawaii, and everywhere else in the U.S., and the censuses of 1960 through 2000 (as well as the 2010 ACS) distinguish only between the countries or territories of birth.

When Moving Matters 49 50 e21 Report 2 November 2014 54. 53. 52. 51. 50. 49. 48. 47. 46. 45. 44. 43. 42. nonblack, non-Hispanicsexcept multiracialadolescents(whoare excludedfrom theseanalyses). In the1979panel,“white”includes allnonblack,non-Hispanicadolescents;inthe1997panel, “white”includesall of schoolingtypicalagivendegree. IPUMS datarecode therecent yearsforcomparabilitywiththepre-1990 data,assigningpeoplethenumberofyears from 1990forward, thehighestdegree wasrecorded earned forthosewhohadatleastcompletedhighschool.The least 16yearsofschooling.Before 1990,theinformationindatarelates tothehighestyearofschoolcompleted; obtain aGEDwithouthaving12yearsofschooling.Similarly, Iuse“collegegraduate”synonymously withhavingat I use“highschooldropout” synonymouslywithhavinglessthan12yearsofschooling;inreality, someonemight figures show. inclusion ofNativeAmericansdoesnotappeartoaffect thetrend muchbetween1880and1900,astheappendix atically countNativeAmericansonreservations or“roaming …overunsettledtractsofcountry.” Nevertheless,the Few NativeAmericanswere enumerated inthecensusuntil1890,firstyearthatattemptedtosystem- asked inthecensusandACS. Hispanic origintoindividualsbefore 1980; since1980,aconsistentlyreliable questionaboutHispanicoriginhasbeen 2010, predicted from individualandgeographicvariables.Italsoincludesacreated Hispanicindicatorthatimputes The IPUMSdataincludeacreated racevariablethatassignsasingletomultiracialrespondents in2000and migration%20frey/1209_migration_frey.pdf. Brookings Institution.Available athttp://www.brookings.edu/~/media/research/files/opinions/2011/1/12%20 H.Frey (2009).“TheGreatWilliam AmericanMigrationSlowdown:RegionalandMetropolitan Dimensions.” of EconomicPerspectives25(3):173–96. Raven Molloy, Christopher L.Smith,andAbigailWozniak MigrationintheUnitedStates.”Journal (2011).“Internal States: Evidencefrom 22:289–325. theIPUMS, 1850–1990.”ResearchinEconomicHistory A.Sundstrom (2004).“TheDeclineandRiseofInterstateMigrationintheUnited Joshua L.RosenbloomandWilliam Since 1850.”AmericanEconomicReview103(4):1109–37. Jason LongandJosephFerrie(2013).“IntergenerationalOccupationalMobilityinGreat BritainandtheUnitedStates who generallyhavenoearnings. 2007 wouldproduce somewhat largerincreases. CensusBureau estimatesincludeinstitutionalized menandwomen, depths oftheGreat Recession, while1979and1969were peakyearsinthebusinesscycle.Comparing1979with these percentages totheestimates inFigure 27produces theresults above.Itshouldbenotedthat2009wasatthe and 12percent in1969.Overthenext30years,thatshare rose to18percent; by2009,itwas20percent. Applying According totheNationalIncomeandProduct Accounts,nonwage compensationwas5percent ofpayin1939 These estimatescompare incomesin1949,1979,and2009ratherthan1950,1980,2010.Seen.29(above). All incomefigures inthis report are expressed in2014dollarsusingtheBureau ofEconomicAnalysis’s PCEdeflator. metropolitan-area mobilityisdoubtfuleveninanyoneofthese years. areas. Hence,aconsistentseriesacross yearsisimpossible, and ameaningfulunderstandingofwithin-between- many Americanscannotbedetermined,andtheproblem isspecifictoparticularstatesandtypesofmetropolitan ity withinandbetweenmetropolitan areas inthissection,too, butfrom 1980forward, themetropolitan statusof but noyearfeatures bothsetsofquestions.Neitherquestionwasaskedin1960.Ioriginallyhopedtoexaminemobil- In the1950censusand2010ACS,similarquestionswere askedaboutwhere Americansresided oneyearago, show especiallyinteresting trends. See n.38above.Istartin1940becausefewofmyoutcomesare availablebefore then,andthosethatare donot women” includetheforeign-born. All analysesinthissectionare restricted adults,exceptthatfigures tonative-born for“alladults,”men,”and 55. The variable is based on questions asked of parents about the number of years of schooling they had. In both panels, I take whichever is bigger between the mother’s and father’s educational attainment. 56. Timothy Noah (2013). “Stay Put, Young Man.” Washington Monthly, November/December. 57. Raven Molloy and her colleagues find that the increase in two-earner families is not an important factor. Greg Kaplan and Sam Schulhofer-Wohl conclude the same of telecommuting. Both find no support for housing-market fluctua- tions or demographic changes as explanations. Young and Varner dismiss taxes as an important factor. Contrary to Noah’s claim, Kaplan and Schulhofer-Wohl speculate that the Internet may have contributed to declining mobility by allowing people to be more deliberative and to assess their options before deciding on a move. See Raven Molloy, Christopher L. Smith, and Abigail Wozniak (2011). “Internal Migration in the United States.” Journal of Economic Perspectives 25(3): 173–96; Greg Kaplan and Sam Schulhofer-Wohl (2012). “Understanding the Long-Run Decline in Interstate Migration.” NBER Working Paper No. 18507 (Cambridge, Mass.: National Bureau of Economic Research); Cristobal Young and Charles Varner (2011). “Millionaire Migration and State Taxation of Top Incomes: Evidence from a Natural Experiment.” National Tax Journal 64 (2, pt. 1): 255–84. 58. Eli Lehrer and Lori Sanders (2014). “Moving to Work.” National Affairs 18. 59. Brian Cadena and Brian Kovak (2013). “Immigrants Equilibrate Local Labor Markets: Evidence from the Great Reces- sion.” NBER Working Paper No. 19272 (Cambridge, Mass.: National Bureau of Economic Research). 60. Kaplan and Schulhofer-Wohl (2012); Raven Molloy, Christopher L. Smith, and Abigail Wosniak (2014). “Declining Migration Within the U.S.: The Role of the Labor Market.” IZA Discussion Paper No. 8149; Mark D. Partridge et al. (2012). “Dwindling U.S. Internal Migration: Evidence of Spatial Equilibrium or Structural Shifts in Local Labor Mar- kets.” Regional Science and Urban Economics 42: 375–88. 61. Ryan Avent (2011). The Gated City (Kindle Single eBook); Edward Glaeser (2011). The Triumph of the City: How Our Greatest Invention Makes Us Richer, Smarter, Greener, Healthier, and Happier (New York: Penguin). 62. William A. Fischel (2001). “An Economic History of Zoning and a Cure for Its Exclusionary Effects.” Working paper. 63. Morris M. Kleiner (2015). “Reforming Occupational Licensing Policies.” Hamilton Project Discussion Paper 2015-01 (Washington, D.C.: Brookings Institution). 64. Douglas S. Massey and Nancy A. Denton (1993). American Apartheid: Segregation and the Making of the Underclass (Cambridge, Mass.: Harvard University Press). 65. Brink Lindsey, ed. (2015). Reviving Economic Growth (Washington, D.C.: Cato Institute). 66. Scott Winship (2015). “Up: Expanding Opportunity in America.” (New York and Washington, D.C.: Manhattan Insti- tute; Center on Budget and Policy Priorities; and Peter G. Peterson Foundation).

When Moving Matters 51

Director Diana Furchtgott-Roth

Senior Fellow Scott Winship

Fellow Jared Meyer

Policy Analyst Preston Cooper

Economics21 at the Manhattan Institute (e21) is dedicated to conducting rigorous economic research and promoting innovative public policies to meet twenty-first century challenges.

Economics21 aims to advance free enterprise, fiscal discipline, economic growth, and the rule of law. Drawing on the expertise of policymakers and academics, e21 encourages accessible, spirited debate about the way forward for democratic capitalism.