2. Measurement of Intergenerational Mobility and ...faculty.arts.ubc.ca/nfortin/econ560/E560L135BwTF.pdf · Measurement of Intergenerational Mobility and Comparative ... H hourly

Fortin – Econ 560 Lecture 5B

V. Social Mobility 2. Measurement of Intergenerational Mobility and Comparative

Evidence

1. Early Studies

2. Use of Administrative Data

3. Strategies to Identify Causal Effects a. Adoptive Children b. Exogenous Variation

4. Cross-Country and Within-Country Studies


1. Early Studies The deceptively simple equation of the intergenerational transmission of income,

01ˆyy , (1)

where 0y denotes the long-run income of the father and 1y the long run income of the son, hides many estimation pitfalls.

Many early studies had not recognized the potential problems and some were still exploring the choice of functional forms, which quickly converge to the log-linear form.

With the log-linear form, represents the intergenerational income elasticity

(IGE). For fathers, a key challenge is to derive an accurate measure of long-run earnings.

Because of both response error and genuine transitory fluctuations in earnings t0 , single-year measures ty0 are error-ridden proxies for longer-run earnings: tt yy 000 (2)

Table 1 Regressions of Son's Income or Earnings on Father's Income or Earnings in Linear, Semilog, and Log-linear Form

Variables Location and Father's Son's Year Year Dependent Independent Other Coefficient t R Z N E Author

1957-60 E I P None 2069 Hauser, Sewell, and Lutterman (1975)

1957-60 Log E I P None N.A. Hauser (in press)? 1974 1957-60 Log E Log I P None 2493 Tsai (1983)t

United States, 1981-82 1981-82 Log E§ Log E§ None 722 Behrman and Taubman (1983)

United States: 1969 (young white) \!'hen son Log H Log I3 II 1607 Freeman (1981)

was 14 1966 (older white) When son Log H Log I3 I1 2131 Freeman (1981)

was 14 1969 (young black) When son Log H Log I3 I1 634 Freeman (1981)

was 14 1966 (older black) When son Log H Log I3 I1 947 Freeman (1981)

was 14 York, England:

1975-78 1950 Log H Log W None 198 Atkinson (1981) 1975-78 1950 Log W Log W None 307 Atkinson (1981)

Malmo, Sweden, 1963 1938 Log I I C D None 545 de \X'olff and van Slijpe (1973)

545 545

Geneva, Switzerland, 1980 1950 IHH I H H None 801 Girod (1984)

Sarpsborg, Norway, 1960 1960 Log I LORI None 115 Soltow (1965)

NOTE.-& = elasticity of son's income or earnings with respect to father's income or earnings; E = earnings; H = hourly earnings; I = income; I3 = income in three-digit occupation; I C D = income-class dummy; I H H = household income; I P = parents' income; W = weekly earnings.

'

* First 5 years in the labor force. t Also ~ o b e r t M. Hauser (personal communication, October 2, 1984). $Adjusted for response variability.

Adjusted for work experience. Sons with work experience of 4 years or less were excluded. The regression was weighted so that each father had equal weight. f\Vork experience, three dummies for re. .ion of residence at age 14, five dummies for type of place of residence at age 14, and a dummy for living in one parent/female home

at age 14. "The elasticities are values between pairs of income classes

Source: Becker and Tomes (1986)

Nicole

Rectangle

Nicole

Oval

Nicole

Oval


This sort of errors-in-variables problem in a regression equation’s explanatory

variable tends to dilute the estimated coefficient of that variable and produce a downward bias in the coefficient,

22

2

0

ˆ plimy

y

When the father’s income is not available, it has been typical to used predicted

earnings. This two-stage procedure that uses education, occupation or social class to predict father’s earnings is likely to lead to an upward bias. o In the second-stage regression, when father’s education, occupation or social

class is used only to predict father’s earnings, but not as a separate explanatory variable in its own right, the resulting omitted-variables bias may lead to overestimation of the intergenerational earnings elasticity.


A different problem has surfaced in measuring son’s earnings. There may be some measurement error in the sons’ earnings: tt yy 111

(3) Or, as argued by Haider and Solon (2003), the slope of the linear projection of ty1

on 1y may not equal to 1, but varies over the life-cycle: ttt yy 111 (3’) by contrast with the textbook case where 1t .

This latter type of measurement error generates the following type of bias

2

0,1

1

)(ˆ plimy

tyyCov

Numerous researchers have indeed reported that they estimate relatively small

intergenerational elasticities if they measure son’s earnings near the very beginning of his career, but that their estimates get larger as son’s earnings are measured further along in the lifecycle.


This generates a problem of mean-reversion (the tendency for a stochastic process to remain near, or to return over time to its long-run average value). As explained by Bound et al. (1994), mean-reverting measurement error in a regression’s dependent variable compresses its variation and consequently leads to a tendency to underestimate the magnitude of the regression’s slope coefficient.

A first solution to the estimation of long-run earnings was to average earnings over

multiple years. Given that there are a few panel data sets in the U.S., this strategy was used in a number of studies.

Several studies have also used IV as a second way to address the measurement error

problem, using education, sex, occupation, industry dummies (e.g. Mulligan, 1997) and the index Duncan index of socioeconomic status (Zimmerman, 1992) as instruments. This strategy has its own problems (Kim and Solon, 2005).

The summary of the U.S. literature is that studies that have used multi-year

measures of father’s earnings and have measured son’s earnings after his first few years in the labor market have estimated about 0.4 or higher.

Source: Solon (1999)

Nicole

Rectangle

Source: Solon (1999)

Nicole

Rectangle

Nicole

Rectangle


There are some problems with the use of these panel data. The intergenerational samples that can be constructed from the PSID and NLS are relatively small, and there is also considerable attrition in these data sets. Moreover, there are some cohort issues with the NLS.

2. Use of Administrative Data Another source of longitudinal data is provided by administrative data, such as

income tax data or social security data. These data have the advantage of large sample size and potentially long

observation window to estimate lifetime earnings that allow the relationship between current earnings and lifetime earnings to be evaluated (e.g. Haider and Solon, 2003).

Corak and Heisz (1999) is such an early study using Canadian income tax data on

about 400,000 father-son pairs, and they find intergenerational earnings elasticities to be about 0.2. substantially lower than for the U.S, when estimating the relationship:

Corak and Heisz 513

Table 3 Intergenerational Elasticities for Various Samples and Specifications: Father and Son Earnings and Market Income

Total Market Earnings Income

Standard Standard Elasticity Error Elasticity Error

A. Sample selection rules 1. Average income over five years 2 $1 2. Income in each of five years 2 $1 3. Income in each of five years 2 $100 4. Income in each of five years 3 $1,000 5. Income in each of five years 2 $3,000

B. Life cycle adjustments 1. Quadratic in age of fathers and sons 2. No controls for age 3. Dummy variables for age of sons 4. Quartic in age of fathers and sons 5 . Oldest sons born in 1963 6. Oldest sons born in 1966 7. All siblings 8. Controls for marital status

C. Choice of regressor 1. Eatnings 2. Total market income

D. Estimation method Average income over five years 2 $1

1. Least squares 2. Median regression

Income in each of five years 2 $1 3. Least squares 4. Median regression

horizon would appear to be long enough to reduce the bias due to transitory income fluctuations.

The robustness of these findings to the sample selection rules employed, the way in which adjustments for life cycle differences is made, and the choice of the father's income variable are assessed in Table 3. All of the results in this table are based upon five year averages of the father's income measure. The four rows labeled 1 in each of the panels of Table 3 repeat the results in the last column of Table 2. There are three major findings. First, the selection rules used to define the sample seem to have an important influence on the estimates of the elasticities. This influence seems to be restricted to whether individuals with zero (or negative) earningslincome are included in the sample before the average is calculated. If fathers must have at least $1 of earnings in each of the five years over which the average is calculated (as opposed to the average being at least $1) the earnings-earnings elasticity increases from 0.131 to 0.228 (see Rows 1 and 2 in Panel A). The elasticity does not change much beyond this as the cutoff is raised further, reaching 0.242 at a cutoff of $3,000.

Source: Corak and Heisz (1999)


20504

213120101 lnln ageageageageyy

They use nonparametric techniques to uncover significant nonlinearities and show

that I intergenerational earnings mobility is greater at the lower end of the income distribution than at the upper end, and displays an inverted V-shape elsewhere.

A common drawback of using administrative data is that human capital and labour

supply variables are typically unavailable, which can introduce considerable measurement error in the earnings measures.

Mazumder (2005) tries to overcome this problem by using the 1984 Survey of Income and Program Participation (SIPP) matched to the Social Security Administration’s Summary Earnings Records (SER) but ends with small sample sizes.

He argues that transitory fluctuations in parental incomes that can have effects

lasting more then 5 years so that using even 5-year averages of fathers’ earnings yields estimates that are biased down by approximately 30% and thus the IGE should actually be closer to 0.6.

this analysis. In the top panel of the table, fathers’ earningsmust be positive in each year. In the lower panel, some yearsof zero earnings are allowed. Within each panel, there arethree additional selection rules: noncovered fathers aredropped; noncovered fathers’ earnings are imputed; andgovernment and self-employed fathers and noncovered fa-thers are dropped. In the first set of results in the top panel(row 1 of table 4), it is not necessary to actually identifycovered status, because all fathers with years of zero earn-ings are dropped. Therefore, it is possible to constructaverages that include years prior to 1979. Under the secondrule (estimates in row 2), in contrast, averages can only beconstructed going back to 1979, because it is difficult toidentify covered status in prior years. Under the third rule(row 3), those identified as government or self-employedworkers at any time during the 1984 SIPP survey period aredropped.

The results from using the two-year average with SERdata are clearly lower than what was found using the SIPP.The highest coefficient is 0.289 when noncovered fathersare dropped from the analysis. The fact that many fathershave noncovered earnings (in addition to covered earnings)that are not captured in the SER data is the obvious expla-nation for the greater attenuation using the SER data. Infact, when noncovered fathers are dropped and earnings arerequired to be at least $3,000 in each year, thereby elimi-nating many of those whose covered earnings severelymisrepresent their true earnings, the estimated coefficient

rises to 0.334 (not shown), which is comparable to the SIPPresults from table 3. This suggests that the results based onthe SER may, in fact, be biased down by even more thanwould be the case with comparable survey data. It alsosuggests that the possibility of upward bias from correlatedmeasurement error between fathers and children when usingSER data is more than offset by the overall attenuation bias.Otherwise the estimates using the SER would have beenhigher than those found when using the SIPP. It is alsoapparent from table 4 that on the whole, the IGE is onlyslightly lower when the imputed noncovered fathers areadded to the sample.

The most striking finding is that the IGE rises dramati-cally as the fathers’ earnings are increasingly averaged overmore years. Indeed, the estimated father-son elasticity is0.613 when the fathers’ earnings are averaged over 16 years.The father-daughter elasticity is a bit lower at 0.570. Whenthe sample of fathers is restricted to private-sector, non-self-employed workers, however, the father-daughter elasticityis estimated at 0.754. Such a high degree of transmission israther surprising and may be due to the possible positivecorrelation between fathers’ earnings and daughters’ laborforce participation among this group, as discussed earlier.

C. Does Excluding Years of Nonemployment Matter?

Couch and Lillard (1998) argue that the results of Solon(1992) and Zimmerman (1992) are sensitive to the inclusion

TABLE 4.—INTERGENERATIONAL ELASTICITIES USING SER FOR FATHERS’ EARNINGS

Fathers

Elasticity (Standard Error) N

Sons Daughters Pooled

Log Avg. Earn. 84–85 82–85 79–85 76–85 70–85 84–85 82–85 79–85 76–85 70–85 84–85 82–85 79–85 76–85 70–85

Father Earnings Must Be Positive Each Year

Drop 0.253 0.349 0.445 0.553 0.613 0.363 0.425 0.489 0.557 0.570 0.308 0.388 0.470 0.559 0.600noncovered (0.043) (0.059) (0.079) (0.099) (0.096) (0.065) (0.087) (0.110) (0.140) (0.159) (0.039) (0.052) (0.067) (0.084) (0.093)fathers 1262 1218 1160 1111 1063 1178 1124 1070 1031 982 2440 2342 2230 2142 2045

Impute 0.289 0.313 0.376 — — 0.350 0.395 0.422 — — 0.322 0.358 0.404 — —noncovered (0.050) (0.052) (0.062) (0.062) (0.081) (0.096) (0.039) (0.048) (0.056)fathers 1485 1462 1433 1360 1339 1310 2845 2801 2743

Drop 0.273 0.419 0.474 0.533 0.652 0.526 0.563 0.635 0.750 0.754 0.393 0.487 0.553 0.643 0.707government & (0.060) (0.082) (0.096) (0.111) (0.135) (0.089) (0.137) (0.150) (0.173) (0.192) (0.057) (0.077) (0.086) (0.100) (0.118)self-employed 844 825 801 779 746 782 758 736 719 690 1626 1583 1537 1498 1436

Allow Some Years of Zero Father Earnings*

Drop 0.234 0.334 0.434 — — 0.312 0.423 0.506 — — 0.269 0.377 0.472 — —noncovered (0.043) (0.057) (0.069) (0.060) (0.065) (0.091) (0.034) (0.043) (0.056)fathers 1295 1268 1227 1201 1168 1127 2496 2436 2354

Impute 0.238 0.342 0.403 — — 0.295 0.384 0.474 — — 0.266 0.365 0.441 — —noncovered (0.042) (0.057) (0.059) (0.055) (0.061) (0.080) (0.033) (0.042) (0.049)fathers 1534 1550 1571 1394 1406 1424 2928 2956 2995

Drop 0.242 0.355 0.441 0.523 0.575 0.400 0.504 0.600 0.731 0.847 0.304 0.422 0.570 0.622 0.703government & (0.059) (0.080) (0.084) (0.101) (0.109) (0.084) (0.083) (0.113) (0.130) (0.145) (0.046) (0.061) (0.073) (0.081) (0.087)self-employed 874 869 862 895 917 803 794 785 825 831 1677 1663 1647 1720 1748

Dependent variable is children’s log average earnings, 1995–1998. All results use tobit specification.Note: For the dependent variable, probit models based on the 1996 SIPP matched to SER were used to determine if zero earnings reflected noncoverage or nonworker status and were imputed accordingly. For

fathers, earnings for those identified as noncovered are either dropped or imputed for the years 1979–1985 as indicated. For the years before 1979, no adjustment is attempted. Earnings for topcoded fathers are imputedusing March CPS data for 1970 to 1980 and using 1984 SIPP for 1981 to 1985. Standard errors are adjusted for within family correlation when more than one sibling is present.

*Required years of positive earnings are: 1 for 2-year averages; 2 for 4-year averages; 3 for 7-year averages; 7 for 10-year averages; and 11 for 16-year averages.

THE REVIEW OF ECONOMICS AND STATISTICS246

Nicole

Rectangle

Nicole

Rectangle

Nicole

Oval

Nicole

Oval


Using detailed information on wealth from the SIPP, he finds an higher IGE for families with low net worth, offering some empirical support for theoretical models that predict differences in IGE due to borrowing constraints, but again small sample sizes are involved.

How much difference does it really make if the IGE is 0.6 rather than 0.4? With an IGE of 0.6, it will take the descendants of a family, initially at the poverty threshold (roughly 75% below the national average) 5 to 6 generations (125 to 150 years) before their income would be within 5% of the national average. With an IGE of 0.4, the comparable time will be just 3 generations.

Yet, using these empirical strategies, these studies cannot claim to found that a causal interpretation can be given to the IGE.

What is needed is that strategy that allows the separation the intergenerational

transmission of income through the investment in schooling mechanism from the inheritability of the other traits.

intergenerational mobility.”43 One problem with this ap-proach is that it does not directly measure parents’ ability tofinance schooling for their children at the time that such aninvestment is made. Mulligan’s measure also does notcapture inter vivos transfers. Finally, the model focusessolely on an intergenerational budget constraint and doesnot analyze parents’ potential inability to borrow from theirown future income, which may be an important issue in itsown right.

Gaviria (2002) addresses some of these problems. Healso uses the PSID, but categorizes the nonconstrained asthose who have actually reported receiving large financialtransfers or whose parents have a high net worth. Gaviriaalso uses a split-sample estimation approach and findssome evidence that intergenerational mobility is in factlower among borrowing-constrained families. However,the differences are not large, and the samples are toosmall to find differences in the IGE at the 5% significancelevel.

The SIPP-SER data can bring several clear advantages tothis question. First, with a larger sample than the PSID it iseasier to detect differences among subgroups when using asample-splitting strategy. Second, the highly detailed wealthdata available in the SIPP make it possible to measureborrowing constraints more directly through net worth. Networth measures the ability of parents to borrow against theircurrent wealth or to draw down assets in order to financehuman capital acquisition for their children. Third, the dataon wealth are available for 1984, when the children arebetween the ages of 16 and 21, and at a time when criticaldecisions regarding college attendance or continuation arebeing made.

Table 10 shows the results of this exercise. First, on usingthe SIPP sample, pooling both sons and daughters together,and splitting the sample by the median level of net worth(approximately $65,000 in 1984 dollars) the results point toa sharp difference between those below the median andthose above. The IGE is 0.458 for those with lower thanmedian net worth, but only 0.274 for those above themedian level. Though the difference looks large, one couldnot reject the null hypothesis of equality at the 5% signifi-cance level. The second set of results compares those at orbelow the first quartile of net worth with those at the topquartile. In this case the difference is even more dramaticand is statistically significant at the 5% level. In fact, for thetop quartile the IGE appears to be close to 0. Indeed, thepermanent income model would predict this result if incomeis uncorrelated with ability.

Similar attempts were slightly less conclusive using SERdata for fathers’ earnings, as the bottom half of table 10shows. Whereas estimates for the low end of the net-worthdistribution were similar to that found using the SIPP, theestimates for those with high net worth were significantly

higher than that found in the SIPP. A possible explanationfor this result is that despite the imputation procedure, thetopcoding of fathers’ earnings resulted in a more com-pressed earnings distribution among those with high networth. In any case, the difference in point estimates whencomparing the top and bottom quartiles (0.310) is still quitelarge and significant at the 10% level.

Although the fact that net worth is measured duringcollege-going years has been presented as a strength, it mayactually be a limitation if the relevant period for measuringborrowing constraints should include earlier points in thechild’s educational career. Cameron and Heckman (2001)and Carneiro and Heckman (2003) have argued that muchof the inequity in human capital development arises early inlife. In this case, empirical evidence of differences in inter-generational mobility by wealth at such a late age mightactually be viewed as even stronger evidence in favor ofborrowing constraints.

One might argue that because net worth and earnings arehighly correlated, any nonlinearities in the IGE may also bereflected in differences in � by levels of net worth that mayor may not be due to borrowing constraints. To address thisissue, the same comparisons in table 10 were also doneusing quartiles of the earnings distribution rather than networth as sample cutoff points. The results (not shown)indicated that there are very modest differences between thehigh and low ends of the earnings distribution. In no casewere the differences remotely close to significant, and in43 See Mulligan (1997, p. 247).

TABLE 10.—INTERGENERATIONAL ELASTICITY BY LEVEL OF NET WORTH

Elasticity (Standard Error) N

OverallHigh Net

WorthLow Net

Worth Diff. t-Stat.

SIPP Results

Father earnings 0.369 0.274 0.458 0.184 0.855Log avg. 84–85 (0.069) (0.184) (0.112) (0.215)Low is �median 1514 757 757High is �medianFather earnings �0.044 0.465 0.508 2.795Log avg. 84–85 (0.135) (0.122) (0.182)Low is �25th percentile 380 379High is �75th percentile

SER Results

Father earnings 0.480 0.304 0.465 0.160 1.130Log Avg. 79–85 (0.068) (0.110) (0.090) (0.142)Low is �median 2,186 1,093 1,093High is �medianFather earnings 0.205 0.515 0.310 1.799Log Avg. 79–85 (0.113) (0.130) (0.172)Low is �25th percentile 547 547High is �75th percentile

Dependent variable is log average earnings, 1995–1998. All results use tobit specification. Sons anddaughters pooled.

Note: For the dependent variable, probit models based on the 1996 SIPP matched to SER wereused to determine if zero earnings reflected noncoverage or nonworker status and were imputedaccordingly. Fathers must have positive earnings in each year. When fathers’ earnings are from the1984 SIPP, they must be present for all interview months and have no cases of nonresponse toearnings or income questions. Only those fathers successfully matched to their wave 4 questionnaireare kept in all samples. Standard errors are adjusted for within-family correlation when more thanone sibling is present.


Source: Mazumder (2005)

Nicole

Rectangle


3.Strategies to Identify Causal Effects a. Adoptive Children A number of recent papers (e.g. Björklund, Lindahl, and Plug (2006), and Sacerdote

(2007)) have used adoption data to ascertain what portion of the transmission of income and education comes from nature (the birth parents) and from nurture (the adoptive parents).

If we assume that 1. Adopted children are randomly assigned to families as infants 2. Adopted children are treated exactly the same was as biological children

o Then adoption can be considered to be an experimental intervention that randomly assigns children to families.

Björklund, Lindahl, and Plug (2006) use the census of Swedish adoptees placed

during 1962–1966 that contains information on the children’s educational and economic outcomes and on the characteristics of both their adoptive and biological parents.


They measure children’s outcome in 1999 when children are 33–37 years of age and model the transmission of an outcome from the biological parent bp and the adoptive parent ap to the adopted child ac born in family j and adopted and reared in family i as

aci

api

bpj

qci yyy 210

where asi represents unobserved child-specific characteristics. Comparing 1 and 2

allow them to distinguish what comes from pre-birth and post-birth factors. They find that both pre- and postbirth factors contribute to intergenerational

earnings and education transmissions, but the prebirth factors are relatively more important for the transmission of education and postbirth factors relatively more important for the transmission of earnings.

They also find that intergenerational associations are much stronger in families that are better educated and generate more income.

A potential source of bias in this study is introduced by the combination of selective placement and measurement error accounts. The authors argue that it accounts for no more than 13 percent of the estimated impact of the parents’ characteristics.

their partner’s education is brought into the equation. Assorta-tive mating on education lies behind this pattern.

The estimated intergenerational elasticities with respect tofather’s earnings and income are virtually identical and equal to0.24. They are similar to those obtained by Bjorklund and Jantti[1997] and Bjorklund and Chadwick [2003] for Sweden.19

In the second panel we report the results for specification (2)where our four outcomes for adoptees are run on the same vari-ables for their adoptive and biological parents using the sameformat as before. The first three columns report the estimates for

19. The previous mobility models are also estimated for sons and daughtersseparately. The intergenerational transmissions for schooling are found to be verysimilar. The intergenerational transmissions for earnings and income are, how-ever, somewhat larger for sons. A similar pattern is observed when we estimatethese models on our samples of female and male adoptees.

TABLE IIESTIMATED TRANSMISSION COEFFICIENTS IN LINEAR MODELS

Years of schooling University Earnings Income

(1) (2) (3) (4) (5) (6) (7) (8)

Own-birth childrenBio father .240** .170** .339** .237** .235** .241**

(.002) (.002) (.004) (.004) (.005) (.004)Bio mother .243** .158** .337** .246**

(.002) (.002) (.004) (.004)Adopted childrenBio father .113** .094** .184** .148** .047 .059*

(.016) (.016) (.036) (.036) (.034) (.028)Bio mother .132** .101** .261** .229**

(.017) (.017) (.034) (.034)Adoptive father .114** .094** .165** .102** .098** .172**

(.013) (.014) (.024) (.026) (.038) (.031)Adoptive mother .074** .021 .145** .097**

(.014) (.015) (.024) (.026)

Sum of estimates .227** .188** .349** .249** .145** .231**for bio and (.019) (.029) (.040) (.059) (.049) (.040)adoptive fathers

Sum of estimates .207** .122** .406** .326**for bio and (.021) (.016) (.039) (.029)adoptive mothers

Standard errors are shown in parentheses; * indicates significance at 5 percent level, and ** at 1 percentlevel. All specifications include controls for the child’s gender, 4 birth cohort dummies for the child, 8 birthcohort dummies for biological/adoptive father/mother, and 25 region dummies of where the biological/adoptive family lived in 1965. The numbers of observations in the second panel for own-birth and adoptedchildren are 94,079/2,125 in columns (1)–(6), 87,079/1,780 in column (7) and 91,932/1,976 in column (8).

1013THE ORIGINS OF INTERGENERATIONAL ASSOCIATIONS

Source: Bjorklund, Lindahl and Plug (2006)

Nicole

Rectangle

Nicole

Rectangle


Sacerdote (2007) analyzes a new set of data on Korean American adoptees who were quasi-randomly assigned to adoptive families, placed by Holt International Children’s Services during 1964–1985 using a queuing (first-come first-served) policy.

He uses a variance decomposition analysis of the share of outcome (Y) that is

attributable to genetic inputs (G), shared (common) family environment (F) and unexplained factors or separate environment, (S), assumed to be uncorrelated:

2222SFGY ,

Dividing both sides by the variance in the outcome ( 2Y ) and defining 222 / YGh ,

222 / YFc , and 222 / YSe yields the standard BG (behavioral genetics) relationship:

2221 ech o The comparison of adoptive siblings, who share only the F component allow

the identification of 2c : Cov(Y1,Y2) = Cov(F1,F1) = Var(F1) = c2.

o Assuming that nonadoptive siblings share half of the same genetic endowment and the same common environment, the comparison of the non-adoptive siblings allow the identification of half G plus F components:


o Cov(Y1,Y2) = Cov(G1 + F1 + S1, G2 + F2 + S2)= Cov(G1 + F1, 1⁄2G1 + F1) =

1⁄2h2 + c2. He also explicitly estimates of treatment effects from assignment to a high

education, small family and finds positive and significantly effects on education, and drinking but not on income, which is however not very well measured.

FIGURE V Comparison of Coefficient of Transmission from Parent to Child Graph shows coefficient from a regression of child’s outcome on mother’s outcome for adoptees and nonadoptees in the sample.

those of Lichtenstein, Pedersen, and McClearn [1992] who findthat family environment explains 21 percent of the variance, andScarr, and Weinberg [1994] who find an adoptive sibling correla-tion of .13. Teasdale, and Owen [1984] find an adoptive siblingcorrelation of .43. All three of these studies use completely differ-ent samples. Teasdale and Owen are examining a small sample ofDanish siblings reared apart and Lichtenstein et al. are examin-ing a small sample of Swedish twins reared apart. Differentialselection of adoptees into families could explain the differences inresults, or perhaps there is something fundamentally differentabout outcomes for siblings reared apart.

Perhaps the more interesting fact is how much the percent ofvariance attributable to nurture varies across different outcomes.When I consider graduating from a US News ranked college, familyenvironment explains 25 percent of the variation, instead of the 14

TABLE VPROPORTION OF OUTCOME VARIANCE EXPLAINED BY HERITABILITY, SHARED FAMILY

ENVIRONMENT, AND NON-SHARED ENVIRONMENT USING A SIMPLE BEHAVIORAL

GENETICS MODEL

Outcome

Proportionexplainedby nurture

(shared familyenvironment)

Proportionexplainedby nature

(heritability)

Unexplainedportion

(non-sharedenvironment)

Has 4 years of college 0.135 0.406 0.459Highest grade completed 0.157 0.443 0.400Family income 0.110 0.334 0.556Log (family income) 0.139 0.324 0.537Drinks 0.336 0.055 0.609Smokes 0.152 0.273 0.575Height 0.014 0.858 0.128Weight 0.044 0.458 0.498BMI 0.115 0.308 0.577Overweight 0.087 0.172 0.741Attended US News

ranked school0.249 0.335 0.417

Acceptance rate ofschool

0.337 0.245 0.418

Married 0.076 0.056 0.979Number of children 0.105 0.196 0.699

I use the simple BG model described in the text to decompose the variance in each outcome into theportions attributable to genes (heritability), shared family environment, and non-shared family environment(i.e., the unexplained portion). See equations (2), (2A), and the paragraph that follows.

141A STUDY OF KOREAN AMERICAN ADOPTEES

Source: Sacerdote (2007)

Nicole

Rectangle

Nicole

Rectangle

Nicole

Rectangle

Nicole

Rectangle

TABLE VIITREATMENT EFFECTS FROM ASSIGNMENT TO HIGH EDUCATION, SMALL FAMILY

Treatment effect“middle group”of families vs.

large, lesseducated

Treatment effecthigh educationsmall family vs.

large, lesseducated

Nonadoptees: Higheducation smallfamily vs. large,

less educated

Effect from a 1standard deviationchange in family

environment index

Child’s years of education 0.314 (0.226) 0.749 (0.245)** 2.157 (0.264)** 0.845Child has 4� years college 0.060 (0.056) 0.161 (0.057)** 0.317 (0.031)** 0.179Log child’s household income 0.071 (0.081) 0.113 (0.089) 0.210 (0.089)* 0.263Child four-year college ranked by

US News0.082 (0.052) 0.231 (0.060)** 0.365 (0.052)** 0.224

Acceptance rate of child’s college 0.007 (0.035) 0.016 (0.036) 0.053 (0.032) 0.098Child drinks (yes/no) 0.099 (0.050)* 0.178 (0.049)** 0.229 (0.041)** 0.280Child smokes (yes/no) 0.013 (0.044) 0.006 (0.048) 0.075 (0.024)** 0.162Child’s BMI 0.509 (0.460) 0.941 (0.468)* 0.929 (0.498) 1.224Child overweight 0.030 (0.047) 0.077 (0.045) 0.088 (0.048) 0.121Child obese 0.020 (0.023) 0.044 (0.018)* 0.037 (0.018)* 0.047Child has asthma 0.005 (0.028) 0.013 (0.031) 0.005 (0.034) 0.085Number of children 0.070 (0.099) 0.199 (0.103)* 0.580 (0.132)** 0.267Child is married 0.014 (0.050) 0.000 (0.056) 0.092 (0.053) 0.123

I split the sample into three groups: High education small families are defined as those with three or fewer children in which both the mother and father have a college degree(Type 1). Twenty-seven percent of adoptees are assigned to such a family. Large lesser educated families are defined as those with four or more children and where neither parenthas a college degree (Type 3). Thirteen percent of adoptees are assigned to such a family. The remaining families (which are either small or have a parent with a college degree) areType 2. Column (1) shows the coefficient on the dummy for assignment to Type 2 relative to Group 3. Column (2) shows the coefficient on the dummy for assignment to Type 1 (smallhigh education) relative to Type 3 (large less educated).

Column (3) shows this Type 1 versus 3 “effect” for the non-adoptees. In a each row, the effects in Columns (1) and (2) are estimated together with a single regression while Column(3) uses a separate regression. Column (4) shows the effect for the adoptees from a one standard deviation move in an index of shared family environment. This is calculated by takingthe square root of the variance share explained by shared family environment in the previous table and multiplying by the standard deviation of the outcome variable: that is, R ��v � �vhat � predicted effect on the outcome from a one standard deviation change in an index of family environment. Standard errors are corrected for within family correlation(1 cluster by family).

146Q

UA

RT

ER

LY

JO

UR

NA

LO

FE

CO

NO

MIC

SSource: Sacerdote (2007)

Nicole

Rectangle

Nicole

Rectangle

Nicole

Oval


b. Exogenous Variation Another strategy is to appeal to the same exogenous instruments that have been

used to estimate the returns to schooling, namely compulsory schooling and parental income shocks.

For example, Black, Devereux and Salvanes (2005) examine the effects of an

increase in mandatory schooling from 7 years to 9 years, which was phased in across municipalities in Norway between 1959 and 1973.

Using the timing of the law changes to instrument for parental education, they find

weak (but not statistically significant) evidence of a causal relationship between exogenous increases in parental education and children’s outcomes.

They do find, however, that among mothers with low levels of education, the

mandated increase in education had a statistically significant effect on their children’s educational attainment.


Oreopoulos, Page and Stevens (2006) exploit the larger variation in compulsory school law changes in the United States and the larger U.S censuses samples to get more precise estimates.

They find that a 1-year increase in the education of either parent, arising from

changes compulsory schooling laws, reduces the probability that a child repeats a grade by between 2 and 4 percentage points.

Using a Canadian panel of administrative data that follows almost 60,000 father-

child pairs from 1978 to 1999 and includes detailed information about the firms at which the father worked, Oreopoulos, Page and Stevens (2005) compare outcomes among individuals whose fathers experienced an employment shock to outcomes among individuals whose fathers did not.

More specifically, their analysis is based on data from the Intergenerational Income

Database (IID) of Statistics Canada, which links tax information on children born between 1963 and 1970 to data on their parents, for all years between 1978 and 1999.

(1) (2) (3) (4) (5) (6) (7) (8)

Dep var:

Age Adj. Log Earnings, 1995-99



Age Adj. Log Earnings, 1995-99 UI Receipt UI Receipt

Social Assistance

Receipt

Social Assistance

Receipt

Father's Log Income 0.383 0.383 0.38 -0.047 -0.081Residual [0.023] [0.023] [0.024] [0.007] [0.004]

Father Displaced -0.086 -0.086 -0.088 0.016 0.018 0.025 0.026[0.040] [0.039] [0.039] [0.012] [0.012] [0.008] [0.008]

With Initial FirmCharacteristic Controls No No No Yes No Yes No Yes

Observations 57,683 57,683 57,683 57,683 59,681 59,681 59,681 59,681

R-squared 0.04 0.03 0.04 0.05 0.02 0.05 0 0.02

Table 3Estimated Effects of Father's Displacement

on Child's Earnings, Unemployment and Social Assistance Receipt

Note: The dependent variable in columns 1-4 is child's log real earnings averaged between 1995 and 1999 after de-meaning by age. All regressions include fixed effects for birth cohort interactedwith gender. The initial firm characteristic controls include fixed effects for regional location of firm (18 possible characters that identify provinces and smaller regions in large provinces)interacted with an indicator variable for whether the firm is in an urban or rural location. The controls also include 11 industry categories and the median log wage of the initial firm in 1978. Huber-White standard errors are shown from clustering by father ID.

Source: Oreopoulos, Page and Stevens (2005)

Nicole

Rectangle


This data base is matched with the LEAP data base, a company-level database that

includes all employers in Canada, both corporate and unincorporated, to identify firm closure.

The authors find an IGE of 0.38 (much closer to the U.S. estimates) and that

children whose fathers were displaced have annual earnings about 8-9% lower than similar children whose fathers did not experience an employment shock.


5. Cross-Country and Within-Country Studies Cross-country studies (Corak and Heisz (1999), Jantti and Osterbacka (1996),

Osterberg (2000) and Osterbacka (2001) strongly suggest that Canada, Sweden, and Finland) are more mobile societies than is the United States with around 0.25.

In contrast, Atkinson, Maynard and

Trinder (1983) and Dearden, Machin and Reed(1997) report higher elasticities (0.42 and 0.57) for the United Kingdom.

One must bear in mind that whether

the intergenerational correlation is 0.2 or 0.5 has enormous consequences for actual mobility rates.

Source: Corak (2007)

A son whose father’s status is in the


fifth percentile has a 37% chance to rise above the median in the intergenerational correlation is 0.2 and a 17% chance to rise above the median if it is 0.5.

In the United States almost one half of children born to low income parents become

low income adults. This fraction is also high in the United Kingdom at four in ten, and Canada where about one-third of low income children do not escape low income in adulthood.

In


the Nordic countries, where overall child poverty rates are noticeably lower, it is also the case that a disproportionate fraction of low income children become low income adults.

A pertinent question is whether cross-country differences in intergenerational mobility are connected with other cross-country differences in income inequality, such as lower cross-sectional inequality, remembering that in the equation (1):

01 / . This does work in the comparison of the Scandinavian countries with the U.S.,

where cross-sectional inequality has increased more substantially over time.

Studies that ask why estimates may differ across country/time focus on role of institutions/government o Solon (2004): IGE lower with progressivity of government education o Ichino, Karabarbounis, Moretti (2009): political economy consideration are

important for education policy and thus influence the IGE o Mayer and Lopoo (2008): IGE higher in low spending (education) states o Pekkarinen Uusitalo, Kerr (2008): a reduction in tracking in Finland leads to

increased intergenerational mobility, lower IGE.

Table 1Estimates of Intergenerational Earnings Elasticities in Countries Other than theUnited States

Study SampleEarnings Measure and Age

Range for SonsFather’s Earnings

Measure b

Atkinson, Maynardand Trinder(1983)

Fathers in working-classneighborhoods ofYork, England, in1950 and their sons

Log hourly earnings atsurvey date (1975–78)a

Log weekly earningsin 1950

0.42

Bjorklund andJantti (1997)

Swedish Level of LivingSurveys

Log annual earnings in1990; ages 29–38

Prediction of logannual earningsbased on educationand occupation

0.28

Corak and Heisz(1999)

Canadian income taxrecords

Log annual earnings in1995; ages 29–32

Log of � ve-yearaverage of annualearnings

0.23

Couch and Dunn(1997)

German Socio-Economic Panel

Log of multiyear (up tosix-year) average ofannual earningsb

Log of multiyear(up to six-year)average of annualearnings

0.11

Dearden, Machinand Reed(1997)

British National ChildDevelopment Survey

Log weekly earnings in1991; age 33

Prediction of logweekly earningsbased on educationand social class

0.57

Gustafsson (1994) Fathers in Stockholm,Sweden, in 1955 andtheir sons born in1939–46

Four-year average of logindividual income;ages 31–41

Log individualincome in 1955

0.14

Hertz (2001) Co-residing fathers andsons in two SouthAfrican surveys

Monthly earnings in1993 or 1998; ages16–39

Monthly earnings in1993 or 1998

0.44c

Jantti andOsterbacka(1996)

Finnish censuses Log annual earnings in1990; ages 30–40

Log of two-yearaverage of annualearnings

0.22

Lillard andKilburn (1995)

Malaysian Family LifeSurveys

Log annual earnings in1988d

Log annual earningsin 1976–77

0.26

Osterbacka (2001) Finnish censuses Log of three-yearaverage of annualearnings; ages 25–45

Log of two-yearaverage of annualearnings

0.13

Osterberg (2000) Swedish income taxrecords

Three-year average oflog annual earnings;ages 25–51

Three-year averageof log annualearnings

0.13

Wiegand (1997) German Socio-Economic Panel

Log monthly earningsin 1994; ages 27–33

Five-year average oflog monthlyearnings

0.34

a Atkinson, Maynard, and Trinder do not report an age range for their regression sample, but theirTable 4.4 for a broader sample shows a range from under 25 to over 65.b Couch and Dunn report a sample mean age of 22.8 in 1984, the second of the six years in which theyobserve earnings.c This elasticity estimate comes from multiplying Hertz’s 0.145 coef� cient estimate for the intergenera-tional regression of earnings levels by a 3.0 ratio of fathers’ sample mean earnings to sons’ sample meanearnings.d Lillard and Kilburn require their sons to be over 18, and they report a sample mean age of 25.

62 Journal of Economic PerspectivesSource: Solon (2002)


The higher the returns to education (and the presence of elite schools), the higher the IGE

Source: Corak (2007)


Differences in the IGE among immigrants and non-immigrants can provide another source of identification for parental vs. societal influences, which may include “ethnic” capital.

Aydemir, Chen, and Corak (2009) show that for Canada, on average the transmission of inequality is the same for immigrants and their children, but outcomes vary by country of origin.

There is lower earnings across the generations for sons in spite of higher than average education for the following countries: o Barbados, Columbia, Grenada, Guyana, Haiti, Jamaica, Japan, Oceania, St.

Lucia / St. Vincent, Trinidad, West Africa

The majority of these 11 countries are Caribbean or surrounding South American o Japan is an exception, the earnings of fathers and sons are essentially at the

Canadian averages

broadly defined samples in rows 1 and 2 are an understate-ment of about one-third. This said, the estimates are withinone standard deviation of each other. The only change

introduced in row 4 is to broaden the sample of sons toinclude those with one parent who was born in Canada,more akin to the definition used by Borjas (1993). The

TABLE 7.—LEAST SQUARES ESTIMATES OF REGRESSION TO THE MEAN MODELS OF EARNINGS FOR IMMIGRANT FATHERS AND THEIR SONS

Outcome

Sample Selection Rules Least Squares Regression Results

Fathers Sons ConstantSlope

Coefficient R2

1. ln weekly earningsfor sons

- male, immigrant - male 5.50[0.573]

0.207[0.084]

0.17- 16 to 65 years - 16 to 65 years- 1981 Census - both parents immigrants

- 2001 Census

2. ln weekly earningsfor sons at age 31


0.207[0.084]

0.17- 16 to 65 years - 16 to 65 years- 1981 Census - both parents immigrants

- 2001 Census



0.267[0.100]

0.25- married or in a common-law relationship

with another immigrant- 25 to 37 years

- children 5 to 17 years- both parents immigrants

- 1981 Census- 2001 Census


- male, immigrant or spouse is animmigrant

- male 5.11[0.607]

0.224[0.089]

0.22

- married or in a common-law relationship- 25 to 37 years

- children 5 to 17 years- at least one parent immigrant


5. ln annual earningsfor sons at age 31


0.176[0.087]

0.11- married or in a common-law relationship

with another immigrant- 25 to 37 years



Notes: Earnings are adjusted for age and region as described in the text. The number of observations in all cases is seventy, corresponding to the country of birth of the father. Estimations are based on weightedleast squares, with the sum of the number of sons and daughters from each group as the weight. Standard errors are presented in square brackets. All estimates are significant at least at the 5% level.

The sample selection rules in row 1 are similar to those in Card, DiNardo, and Estes (2000) and intended to facilitate a Canada-U.S. comparison. The use of annual earnings as the outcome in row 5 is intendedto facilitate comparisons to existing studies of generational mobility among the general Canadian population.

TABLE 8.—LEAST SQUARES ESTIMATES OF REGRESSION TO THE MEAN MODELS OF EARNINGS FOR IMMIGRANT FATHERS AND THEIR DAUGHTERS

Outcome

Sample Selection Rules Least Squares Regression Results

Fathers Daughters ConstantSlope

Coefficient R2

1. ln weekly earnings for - male, immigrant - female 6.86 �0.050 0.01daughters - 16 to 65 years - 16 to 65 years [0.623] [0.092]

- 1981 Census - both parents immigrants- 2001 Census

2. ln weekly earnings fordaughters at age 31

- male, immigrant - female 6.57 �0.050 0.01- 16 to 65 years - 16 to 65 years [0.623] [0.092]- 1981 Census - both parents immigrants

- 2001 Census


- male, immigrant - female 6.715 �0.048 0.01- married or in a common-law relationship

with another immigrant- 25 to 37 years [0.738] [0.108]




- male, immigrant or spouse is an immigrant - female 7.13 �0.114 0.03- married or in a common-law relationship - 25 to 37 years [0.649] [0.095]- children 5 to 17 years - at least one parent immigrant- 1981 Census - 2001 Census

5. ln annual earnings fordaughters at age 31

- male, immigrant - female 11.1 �0.093 0.02- married or in a common-law relationship

with another immigrant- 25 to 37 years [0.966] [0.091]



Notes: Earnings are adjusted for age and region as described in the text. The number of observations in all cases is seventy, corresponding to the country of birth of the father. Estimations are based on weightedleast squares, with the sum of the number of sons and daughters from each group as the weight. Standard errors are presented in square brackets. All estimates are significant at least at the 5% level.

The sample selection rules in row 1 are similar to those in Card, DiNardo, and Estes (2000) and intended to facilitate a Canada-U.S. comparison. The use of annual earnings as the outcome in row 5 is intendedto facilitate comparisons to existing studies of generational mobility among the general Canadian population.

INTERGENERATIONAL EARNINGS MOBILITY 387Source: Aydemir, Chen, and Corak (2009)

estimate of the slope falls from 0.267 to 0.224, a differenceof less than one standard error.

Finally, row 5 reports the estimation results when theoutcome of interest is defined to be the natural logarithm ofannual rather than weekly earnings. This is the outcomeused in much of the existing Canadian literature on gener-ational earnings mobility, as for example in Corak and Heisz(1999) and Grawe (2004a, 2004b), and as reported in table1. The resulting estimate at 0.176 is almost exactly in linewith these results.

In sum, the major conclusions from this table are (i) thatthe best estimate of the generational elasticity in father-sonweekly earnings is 0.27; (ii) that the generational elasticityamong the immigrant population in Canada is no different thanfor the population at large; and (iii) that this elasticity is lower,possibly about 50% lower, than in the United States.

Table 8 offers the least squares results for the father-daughter earnings relationship. All of the estimated elasticitiesare not statistically different from 0, though the point estimatessuggest a very weak negative correlation. This is in contrastwith both the existing Canadian literature for the population atlarge and the findings of Card, DiNardo, and Estes (2000). TwoCanadian-based studies examine the generational mobility ofdaughters, focusing on annual earnings. Fortin and Lefebvre(1998, table 4.3) use a similar estimator with Census data thatis based upon averages of occupational earnings to suggest thatin 1994 the father-daughter elasticity is in the neighborhood of0.22; though one of their estimates is as low as 0.14, it remainsstatistically significant. Corak (2001, table 1) uses administra-

tive data that directly link fathers with their children andreports a father-daughter earnings elasticity of 0.20. Card,DiNardo, and Estes (2000, table 6.7) report 0.21 for U.S.immigrants using their 1940–1970 sample, and 0.50 for their1970–1995 sample. The latter result is not significantly differ-ent from the 0.62 reported for fathers and sons.

Figures 1 and 2 are scatter plots of the seventy data pointsand the estimated regression lines from row 3 of tables 7 and8. The regression line is estimated with weighted least squares.In order to draw further insights we identify any particularlyinfluential data points by successively dropping a single obser-vation from the regression and reestimating equation (1) withthe remaining 69 observations. We do this for each observationand obtain 69 separate estimates of �, which are plotted infigure 3 for sons and figure 4 for daughters. The results arealways within one standard error of the preferred estimates inrow 3 of tables 7 and 8 based on all seventy observations. Thisexercise highlights that sons of fathers from China and theUnited Kingdom have a noticeable impact on the point esti-mate, suggesting that they are more mobile than the average.China also stands out in the results for daughters.10 Theopposite is the case for sons with fathers born in Greece.Though these observations stand out in figure 3, it should once

10 The relative shares of these countries in the entire population alsodetermines the extent of the change in the estimated elasticity. Forexample, as appendix table 1 illustrates, the United Kingdom has thesecond-highest number of children in the data and therefore carries arelatively large weight in the weighted regressions.

FIGURE 1.—SCATTER PLOT OF GROUPED DATA OF WEEKLY EARNINGS FOR FATHERS AND SONS

|CANADA

FRANCE

IRELAND

UK

FINLAND

NORWAY

SWEDEN

OTHER E/C AFRICA

EGYPT

S AFRICA

AUSTRALIA

NEW ZEALAND

OTHER C AMERICA

GRENADA

HAITI JAMAICA

\S. LUCIA

COLOMBIA

ECUADOR

GREECE

\W AFRICA

LEBANON

OTHER W ASIA

CHINA

TAIWAN

OTHER E ASIA

FIJI

--ROMANIA

TANZANIA

UGANDA

MOROCCO

--OTHER N AFRICA

TURKEY

IRAN/IRAQ

SYRIA

SRI LANKA

OTHER CARIBBEANS

6.2

6.4

6.6

6.8

7.0

Mea

n lo

g W

age

,2n

d G

ene

ratio

n S

ons

6.2 6.4 6.6 6.8 7.0 7.2

Mean log Wage, Immigrant Fathers

Notes: Sample selection rules are those described in row 3 of table 7. The weighted least squares regression line is depicted and has a slope of 0.267. Dotted lines represent the average log weekly earnings offathers and sons. Not all points are labeled. See appendix table A1 for details. An observation for Canadian-born children of Canadian-born fathers is included for reference but not used in the regression.


Source: Aydemic, Chen and Corak (2009)

Nicole

Highlight

Nicole

Highlight

Nicole

Highlight

Nicole

Highlight


In an ambitious study using population tax data covering all tax filers in the U.S. from 1996-2011, Chetty, Hendren, Kline and Saez (2013) measure intergenerational mobility at the local (census commuting zone) level based on the correlation between parents’ and children’s earnings (focusing on all children who were born in 1980 or 1981)

To study the effects of tax expenditures on the social mobility using spatial variation in tax expenditures across the United States, they estimates the following regression:

ܧܩܫ = ߙ + ݔܧ ߚ + ߛ (1)ߝ +for each CZi, where IGEi is the parent rank-child rank correlation using within-CZ income centile ranks, EXPENDi is the measure of tax expenditures, and Xi is a vector of CZ characteristic controls including CZ median income and percentage of the population that is a 4-year-college graduate, white, black, Hispanic and other.

They show that the level of local tax expenditures (as a percentage of AGI) is positively correlated with intergenerational mobility, a correlation that is robust to introducing controls for local area characteristics o An increase of overall tax expenditures by 1% of AGI in a CZ decreases the

parent-child income correlation by .5 percentage points, relative to the mean correlation of 0.33.

22

FIGURE 2

Progressivity of Tax Expenditures

(% Tax Expenditures/AGI, for Lowest - Highest AGI Class)

Notes: This figure maps CZ difference in mean tax expenditures as a percentage of average

Adjusted Gross Income (AGI) for individuals with less than $10,000 AGI and individuals with

over $200,000 AGI. Darker areas represent more progressive tax expenditures. Data are from the

IRS Statistics of Income ZIP Code Individual Income Statistics (2008).

Source: Chetty, Henren, Kline and Saez (2013)

23

FIGURE 3

Intergenerational Mobility (Parent-Rank and Child-Rank Income Correlation)

Notes: This figure maps CZ coefficients from OLS regressions of adult children’s income rank

on their parents’ income rank, with rank defined by income centiles within each CZ. Darker

areas represent lower intergenerational mobility. See Section 2.2 for details on the construction

of local IGE measures.


18

TABLE 5 Tax and other Correlations with Intergenerational Mobility

Dep. Var.: E[Child

Rank|Parent=p25] (1) (2) Local Expenditure 0.215 (0.076)

State Tax 0.199 (0.141)

State EITC Rate 0.231 (0.109)

Student Expenditure 0.251 (0.094)

High-school Dropout Rate -0.639 (0.064)

Score 0.557 (0.086)

College Return -0.276 (0.137)

College Tuition -0.003 (0.060)

Colleges per capita 0.102 (0.042)

Inc. at p75 - Inc. at p25 -0.475 (0.089)

Share of Income of Top 1% 0.178 (0.068)

Share Black -0.605 (0.065)

Black Isolation -0.513 (0.065)

Segregation of Poverty -0.405 (0.063)

Migration Inflow -0.184 (0.075)

Share Foreign Born -0.016 (0.060)

Migration Outflow -0.098 (0.069)

Mean Household Income 0.109 (0.075)

Income Growth Rate 0.561 (0.066)

Share Manufacturing -0.260 (0.081)

Trade Shock -0.274 (0.124)

Social Capital Index 0.617 (0.091)

Religiosity 0.510 (0.087)

Crime Rate -0.326 (0.101)

Share Single Moms -0.763 (0.078)

Share Single Moms (kids of married) -0.652 (0.094)

Divorce Rate -0.688 (0.108)

Teen birth Rate -0.550 (0.091)

Notes: The dependent variable is the fitted value (with parent rank at 25) of a within-CZ OLS regression of 100-bin parent income rank and 100-bin child income rank in 2011. Each row reports estimates from an OLS regression against the variable in the left-most column. Column 1 reports the slope coefficients and Column 2 reports standard errors that are clustered at the state level. We standardize our dependent variable (the fitted value) and all right-hand-side variables to have mean 0 and variance 1.


Nicole

Rectangle


Among various categories of local expenditures, they find significant positive

correlations between state EITC policy and intergenerational mobility.


Basic readings: Solon, G. “Intergenerational Mobility in the Labor Market,” in O. Ashenfelter and D. Card, eds., Handbook

of Labor Economics, vol. 3A, North-Holland, 1999, 1761-1800. Solon, G. “Cross-Country Differences in Intergenerational Earnings Mobility,” Journal of Economic

Perspectives, Vol. 16 (Summer 2002), 59-66. Aydemir, A., W.H. Chen, M. Corak “Intergenerational earnings mobility among the children of Canadian

immigrants”, Review of Economics and Statistics, Vol. 91(May 2009): 377-397

Documents

2. Measurement of Intergenerational Mobility and ...faculty.arts.ubc.ca/nfortin/econ560/E560L135BwTF.pdf · Measurement of Intergenerational Mobility and Comparative ... H hourly