Intergenerational Inequality Samuel Bowles and ... - Semantic Scholar

Intergenerational Inequality Samuel Bowles and Herbert Gintis∗ March 31, 2001

1

Introduction

People differ markedly in their views concerning the appropriate role of government in reducing economic inequality through progressive taxation, transfers to the less well off and similar programs. While differing values and self-interest explain part of these differences in support for or opposition to redistribution, by far the most important fault line concerns contested factual beliefs about why the rich are rich and the poor are poor. Those who think that “getting ahead and succeeding in life” depends on “hard work” or “willingness to take risks” tend to oppose redistributive programs while those who think that the key to success is “money inherited from family,” “parents and the family environment,” “connections and knowing the right people,” or being white, support redistribution (Fong forthcoming, Bowles, Fong and Gintis forthcoming). Unfairness, not inequality per se, excites opposition, and high on most peoples list of unfair economic advantages are the things one inherits from ones’ parents: wealth, family connections, the quality of schooling, and race. Handing down success strikes many people as unfair even if the stakes are small, while achieving success is not objectionable even with high stakes, as long as the playing field is level. Is it? No one doubts that the children of well-off parents may tend to receive more and better schooling and benefit from material, cultural, and genetic inheritances. But until recently, the consensus in economics has been that, in the U. S., success ∗ To appear in J. Econ. Perspectives. We would like to thank Jere Behrman, Anders Bjork-

lund, Williams Dickens, Marcus Feldman, James Heckman, Tom Hertz, Arjun Jayadev, Christopher Jencks, Casey Mulligan, Robert Plomin, and Cecelia Rouse for their contributions to this paper, Bridget Longridge for research assistance and the MacArthur Foundation for financial support. Institutional affiliations: Bowles, Department of Economics, University of Massachusetts and Santa Fe Institute, Gintis: Department of Economics, University of Massachusetts. The authors can be reached at [email protected], http://www-unix.oit.umass.edu/˜bowles, [email protected], http://www-unix.oit.umass.edu/˜gintis.

1

Intergenerational Inequality

2

is something that is won or lost in every generation. The rather weak statistical relationship between parents’and children’s’adult economic status uncovered in the early research of Blau and Duncan (1967) seemed to confirm that the United States was indeed the “land of opportunity.” The simple correlations between parents’ and sons’ income or earnings (or their logarithms) in the United States reported by Becker and Tomes (1986) averaged 0.15, leading the authors to conclude that “Aside from families victimized by discrimination…[a]lmost all earnings advantages and disadvantages of ancestors are wiped out in three generations.” (S32) Becker (1988) expressed a widely held consensus when, in his Presidential address to the American Economics Association, he concluded: “…low earnings as well as high earnings are not strongly transmitted from fathers to sons.” But the appearance of such high levels of intergenerational mobility was an artifact of two types of measurement error: mistakes in reporting income, particularly when individuals were asked to recall the income of their parents, and transitory components in current income uncorrelated with underlying permanent income (Bowles 1972, Bowles and Nelson 1974, Atkinson, Maynard and Trinder 1983, Solon 1992, Zimmerman 1992). The high noise-to-signal-ratio in both generations’ incomes (as well as the restricted variance in some of the nonrepresentative samples used) depressed the intergenerational correlation, and when corrected using a variety of methods and data sources, the intergenerational correlations for economic status now appear to be quite substantial, on the order of twice or three times the average of the United States studies surveyed by Becker and Tomes (1986). The upwards adjustment of the consensus estimates of the extent of intergenerational inequality has stimulated a revival of empirical research on the mechanisms accounting for parent-offspring similarity in economic status (Behrman, Pollak and Taubman (1995), Mulligan (1997), Bjorklund and Jantti (1999) and Solon (1999) are important contributions). The relevant facts on which most researchers agree include the following: brothers’ incomes are much more similar than those of randomly chosen males of the same race and similar age differences; the income of identical twins are much more similar than fraternal twins or non-twin brothers; the children of well off parents obtain more and higher quality schooling, and wealth inheritance makes an important contribution to the wealth owned by the offspring of the very rich. On the basis of these and other empirical regularities, the intergenerational transmission of economic status is generally thought to reflect the combined effects of the genetic and cultural transmission of traits, such as cognitive functioning, that contribute to economic success, as well as the inheritance of income-enhancing group memberships and property. The superior education enjoyed by the children of higher status families contributes to this process of economic inheritance. This paper will present data on the extent to which economic status is inherited March 31, 2001


3

and will evaluate what is known about the channels of influence accounting for the observed degree of inheritance. We will also identify some as yet overlooked individual traits that enhance economic success in the members of both generations and are transmitted across generations. Our survey of the relevant evidence and models supports three conclusions.1 First, the extent of intergenerational economic status transmission is considerable. In the United States, knowing the income or wealth of someone’s parents is about as informative about the person’s own economic status as is the person’s own years of schooling attained, or score on a standardized cognitive test. Second, the genetic inheritance of traits contributing to higher earnings appears to account for a substantial part of the intergenerational income transmission process, but the cognitive skills measured on IQ and related tests explains very little, even if the estimated heritability of IQ used in these calculations is quite high. Behavioral genetics provides some clues (but no compelling estimates) concerning the remainder of the genetically transmitted component. The estimates of the genetic component are quite sensitive to small changes in assumptions concerning unobservable parameters and are at best imprecise approximations. Third, the combined inheritance processes operating through superior wealth, cognitive levels, and educational attainments of those with well-off parents, while important, do not fully explain the intergenerational transmission of economic status. The transmission of economic success across generations remains something of a black box. We first introduce measures of the extent of intergenerational status transmission, and then present a method for decomposing the intergenerational transmission correlation into components reflecting the influence of genetic and cultural transmission as well as wealth transfers in the process. In Section 4 we explore the contribution of the genetic inheritance of cognitive skills to the intergenerational transmission of status, and in Section 5 we take up the role of inherited wealth and schooling. We then provide some estimates of the relative importance of genetic, cultural and wealth inheritance in the process. In the penultimate section we consider the importance of parent offspring similarities in personality and group membership. 1 See Bowles and Gintis (2001) for the relevant formal models and other technical aspects of

this research, also available at http://www.santafe.edu/sfi/publications/working-papers.html. Bowles, Gintis and Osborne (2002b) presents a collection of recent empirical and theoretical research.

March 31, 2001


4

2 The Intergenerational Transmission of Economic Status Economic status may be measured in discrete categories—by membership in hierarchically ordered classes, for example—or continuously, by earnings, income, an occupational prestige index, or wealth. Continuous measures allow a simple metric of persistence, the intergenerational correlation coefficient ρ, the square of which measures the fraction of the variance in this generation’s measure of economic success that is statistically associated with the same measure in the previous generation. A second and closely related index of persistence based on continuous measures of status is that first used by Francis Galton (1889):97 in his study of the inheritance of stature. He plotted the height of sons against the height of fathers and used the slope of the best fitting line (which was 2/3) as his measure of intergenerational persistence. A commonly used extension of this approach (e.g., Goldberger 1989) represents the income of a member of the current generation as the sum of the effect of the parents’ income, mean income in the second generation and an error term. y = (1 − βy )y¯ + βy y p + y .

(1)

We use subscript‘p’ to refer to parental measures, so y is an individual’s economic status, adjusted so that its mean, y, ¯ is that of the parental generation, βy is a constant, y p is the individual’s parental y, and y is a disturbance uncorrelated with y p. Rearranging terms we see that y − y¯ = βy (y p − y) ¯ + y ;

(2)

i.e., the deviation of the offspring’s income from the mean income is βy times the deviation of the parent from mean income, plus an error term. We will term βy the “Galton measure” of intergenerational persistence. The influence of mean income on the income of the offspring, 1 − βy measures what is called regression to the mean for, as (2) makes clear, one may expect to be closer to the mean than one’s parents by the fraction 1 − βy .2 The relationship between the Galton measure and the intergenerational correlation is given by σy p ρ y = βy , σy where σy is the standard deviation of y. If y is the natural logarithm of wealth, income or earnings, the standard deviation of y is a common measure of inequality. 2 In this setup the intergenerational transmission process is represented as a first order Markov

process. We gain little explanatory power by using higher order Markov processes to take account of the effects of earlier generations (Warren and Hauser 1997, Behrman and Taubman 1989).

March 31, 2001


5

Thus, if inequality is unchanging across generations, so σy p = σy , then ρy = βy . When inequality is changing across generations the two measures diverge, the Galton measure exceeding the correlation when income inequality is declining and conversely. The Galton measure is preferred in these cases as it measures the normalized gross effect of parental income on offspring income independently of changes in the distribution of income, while the intergenerational correlation coefficient confounds the two. In the empirical work reviewed below earnings, income, wealth and other measures of economic success are measured by their natural logarithm unless otherwise noted. Thus, βy is the percentage change in offspring’s economic success associated with a one percent change in parents’ economic success. Economic Characteristic log family consumption log family wealth log family income log earnings or wages years of schooling

Number of Estimates 2 9 10 16 8

Range 0.59–0.77 0.27–0.76 0.14–0.65 0.11–0.59 0.14–0.45

Average 0.68 0.50 0.43a 0.34a 0.29a

Table 1: Intergenerational Persistence of Some Economic Characteristics, βi . Source: Mulligan (1999) a If recent studies of the U.S. only are included these averages are 0.35, 0.33, and 0.38 respectively.

Table 1 presents estimates of the Galton measure. The extent of persistence— especially for income, wealth and consumption—is substantial. The range of estimates is also considerable. The underlying studies indicate that persistence generally rises with age, is greater for sons than daughters, and is greater when multiple years of income or earnings are averaged. Behrman and Taubman (1989), using the Michigan Panel Survey of Income Dynamics, found that the estimated intergenerational correlation of parental income and offspring earnings is 0.58 when ten years of earnings are used compared to 0.37 for a single year. Mazumder (2000) averages income in both generations over many years and finds intergenerational earnings correlations on the order of 0.6. There is some evidence that the degree of income persistence, as measured by the Galton coefficient, rose between the 1970s and 1990s, having fallen prior to the 1970s (Levine 2001). How different are the probabilities of economic success for the children of the poor and the well off? Can the measures of persistence in Table 1 be translated into probabilities of obtaining high or low incomes conditional on the income level of one’s parents? We can illustrate the degree of conditioning implied by given values of ρy assuming that the underlying relationship is linear in y and y p and that both are normally distributed. Figure 1 represents the transition probabilities implied March 31, 2001


6

by these assumptions and two different intergenerational correlation coefficients, ρ = 0.4 and ρ = 0.5. It can be seen that when ρ = 0.4, an individual whose parents are in the top income decile is roughly sixteen times more likely to attain the top decile than an individual whose parents are in the bottom decile. For ρ = 0.5, the relative advantage of the child of the rich is forty-four times that of the child of the poor. Studies allowing for nonlinear effects in (1) suggest that our assumption of normality may lead these figures to understate the actual degree of persistence in the tails of the income distribution and to overstate it over the middle range of the distribution (Corak and Heisz 1999, Cooper, Durlauf and Johnson 1994, Hertz 2001).

March 31, 2001


7

0.350 0.300 0.250

Transition Probability

0.200 0.150 0.100 0.050 0.000 1 2 4

Parental Decile

6

8

10

10

8

6

4

2

1

Offspring Decile

Figure 1: Transition Probabilities for Two Intergenerational Status Correlations. The height of the bar in cell (i, j ) is the probability that a child of parents in the i-th decile of income will have an income in the j -th decile assuming that log income is binormally distributed. The black bars are for the case where ρ = 0.4 and the white bars for ρ = 0.5. Thus the black bar on the far left indicates that the child of parents in the poorest percentile will attain the highest percentile with probability 0.016 while the black closest to the reader indicates the child of parents in the richest percentile will attain the highest percentile with probability 0.267 or sixteen times more likely than the poor child. The corresponding figures for ρ = 0.5 are 0.007 and 0.322, which means that the prospects the children of the rich themselves being rich are 44 times as great as the prospects of the children of the poor.

March 31, 2001


3

8

Sources of Persistence: Cultural, Genetic and Bequest

We turn now to the role of various forms of inheritance in the transmission process. Equation (1) is merely a summary of the results of a number of distinct processes having little in common, except that they result in parent-offspring similarities for traits statistically associated with the degree of economic success in both generations. While candidates for the list of income-generating traits with strong parent-offspring similarity are many, there are few for which both economic relevance and parent-offspring similarity have been empirically demonstrated. Among these are cognitive performance, the level of schooling, and ownership of wealth, each illustrating distinct transmission processes. We treat income—that is, the sum of labor earnings and returns to assets—as a phenotypic trait influenced by the individual’s genotype g, environment e, and ownership of income-earning assets. Genotypic and environmental influences jointly determine individual skills and other traits relevant to job performance, sometimes termed ‘human capital.’ Among the environmental influences are cultural transmission from parents, schools and other learning environments. We transform income, human capital, and assets to their natural logarithms, and then normalize all variables to have zero mean and unit variance, represented by y, l, and k, respectively. Here and below, y˜ is total income, while y is earnings from labor, represented as a return to human capital. So we have the causal model illustrated in Figure 2, illustrating the paths from the income of the parents through the genetic, cultural and asset based determinants of income, to the income of the next generation. The constants ω, h, βle , and π give the effect in standard deviation units of a standard deviation change in the relevant variable. Thus h2y is the heritability of earnings, namely that portion of the variance of individual earnings that would be explained by the variance of g, were e and g uncorrelated. Taking advantage of the properties of correlation coefficients and normalized regression coefficients (Goldberger 1991), we can decompose the intergenerational correlation ρy˜ , yielding additive terms expressing the contribution of each of the above variables. Each term in the decomposition is the product of the normalized effect of one of the endogenous variables on the offspring’s income multiplied by the simple correlation between the variable and parental income. Thus we have: Intergenerational = genotypic + environmental + bequest Correlation effect effect effect To illustrate the underlying calculation, the environmental effect is the correlation between the child’s environment and parents’ income ry pe , multiplied by the normalized effect of environment on child’s income operating through the child’s level of human capital, ωβle . This decomposition expresses three fundamental mechanisms of intergenerational transmission of economic status: genetic, cultural,

March 31, 2001


ry pg y˜ p Parental Income

9

Genotype ✯ g

Human Capital ✲ l

hy βle

ry pe

✸

✲ e

π

Environment r y pk

ω s ✸ y˜

Income

z k

Wealth Figure 2: Cultural, Genetic, and Bequest Source of Persistence. Note: The three paths from parental income designated by an r refer to simple correlation coefficients, the remaining paths are normalized regression coefficients. Error terms are not shown.

and asset-based. What can we say about the relative importance of each? To answer this question we consider particular components of the genetic and environmental influences on economic success.

4 The Role of Genetic Inheritance of Cognitive Skill Both the similarity of parents’ and offspring’s scores on cognitive tests and the statistical association of test scores and earnings are well documented. This suggests an appealing explanation of the process of intergenerational status transmission via inheritance of cognitive skills. Correlations of IQ between parents and offspring are substantial, ranging from 0.42 to 0.72, the higher figure referring to average parental vs. average offspring IQ (Bouchard and McGue 1981). The contribution of cognitive functioning to earnings has been established using survey data to estimate the natural logarithm of earnings y as a function of a measure of parental economic and/or social status y p, years (and perhaps other measures) of schooling s, and performance on a cognitive score c— often, in U.S. data sets, the Armed Forces Qualification Test (AFQT), a cognitive test developed to predict vocational success—as well as an error term and other variables, such as work experience, race, and sex, which will not concern us. We have located sixty-five estimates of the normalized regression coefficient of a test score in a an earnings equation for the U.S. over a period of three decades. These appear in Figure 3, where the vertical axis is the estimated coefficient and the horizontal axis gives the year to which the data apply. The mean of these estimates 0.15, indicating that a standard deviation change in the cognitive score, holding constant the remaining variables, changes the natural logarithm of earnings

March 31, 2001


10

by about one seventh of a standard deviation. By way of contrast the mean value of the normalized regression coefficient of years of schooling in these studies is 0.22, suggesting a somewhat larger independent effect of schooling.3 βyc

0.35 r

0.3 0.25 r

0.2 0.15

r

r

rr rr

r r

0 −0.05 −0.1 1960

r

1965

1970

1975

r r

r r r r r r r r r r r r rr r r

r r r

r

r

r

r

r

r

0.1 0.05

r

r

r

r r r r r r r rr r r r r r r r r r

r

1980

1985

1990

1995 year

Figure 3: Normalized Regression Coefficient of Cognitive Score on the Logarithm of Income or Earnings by Year: Sixty Five Estimates from Twenty Four Studies (Bowles, Gintis and Osborne, 2002)

Does the fact that measures of cognitive functioning are robust predictors of individual earnings, along with the similarity of the cognitive scores of parents and offspring imply a major role for genetic inheritance of cognitive ability in the process of intergenerational economic status transmission? A way to formulate this question precisely is to ask how much lower would the intergenerational correlation be if there were no genetic inheritance of IQ, that is, if the correlation of parental and child genotypic IQ were zero rather than 0.5 (or somewhat higher of parents genotypic IQ’s are positively correlated through a process of assortative mating). Inspecting the causal model in Figure 4 one can see that this involves severing the genetic link (r g ) and then calculating the implied hypothetical correlation between parental earnings and offspring earnings. The difference between this hypothetical calculation and the observed correlation is the genetic contribution via IQ to the intergenerational transmission of economics status. The hypothetical ceteris paribus 3 We checked to see if these results were dependent on the weight of overrepresented authors, the

type of cognitive test used, at what age the test was taken and other differences among the studies and found no significant effects (Bowles, Gintis, and Osborne 2002a).

March 31, 2001


11

nature of this thought experiment is obvious: if IQ were not inherited it is likely that other aspects of the intergenerational transmission and earnings generating process would change. It is informative nonetheless to do the calculation.

❃

Parental IQ cp

βyc βsc

hc gp Parental Genotypic IQ

rg ❥

g

Genotypic IQ

hc

❥ sp

βys

Parental Schooling

z ❄✾ ✲ c IQ

✲ ✯

Parental Income yp

Income

βyc βsc

s

βys

✲ y ✸

s

Schooling

Figure 4: A Causal Model of Intergenerational Earnings Transmission. Note: the causal paths generate the intergenerational status correlation ry py . Solid lines indicate the causal paths we use to calculate the genetic contribution (via IQ) to the similarity of incomes across generations.

To answer this question we need the answers to two further questions. First, what role does genetic inheritance of IQ play in the covariation of parental and offspring cognitive performance? Second, how important is cognitive performance as a direct and indirect (through educational attainments) determinant of earnings? The answer to the first question depends on two factors: the heritability of IQ, which is probably about 0.5, but cannot be greater than unity, and the genetic correlation (also 0.5). The answer to the second question depends on three factors: the influence of IQ on educational attainment, the influence of educational attainment on earnings, and the direct influence of IQ on earnings, independently of schooling. The causal paths on which this calculation is based appear in Figure 4 as continuous arrows, the others as dashed arrows. We use representative estimates from the literature (most of them summarized in Bowles, Gintis and Osborne 2002a). The details of our calculations are outlined in the accompanying box. The estimate of the normalized effect on earnings of the child’s IQ (both directly and indirectly via schooling) is substantial: 0.266. We take this to be the relevant value for the parents’ generation as well. The genetic contribution to the correlation of parental and offspring IQ’s is r g times h2c , or half the heritability of IQ, with the result that the contribution of genetic inheritance of IQ to the intergenerational transmission of economic status is 0.5h2c (0.266)2 = 0.035h2c . Even if the heritability of IQ were 100% (h2c = 1), the genetic inheritance March 31, 2001


12

of IQ would account for a rather modest portion of the observed levels of intergenerational economic status transmission. The considerably lower values for h2c suggested by recent research of Devlin, Daniels and Roeder (1997), Feldman, Otto and Christiansen (2000), and Plomin (1999) are consistent with the conclusion that less than one twentieth of the observed intergenerational status transmission is due to genetic inheritance of IQ. The Genetic Inheritance of IQ as an Intergenerational Transmission Mechanism Let γc be the component of the correlation of parental and child IQ that is due to the parent-offspring similarity in genotypic IQ and let b be the normalized (direct and indirect) effect of IQ on earnings in both generations. Supressing the paths not involved in our calculation, and rearranging terms to highlight the logic of the above paragraphs, the causal structure in Figure 4 reduces to b ✲ y c γc

✒

po

p

❘ c

b

γy

✲ y✠

From this it can be seen that the component of the intergnerationalincome correlation acccounted for by the genetic inheritance of IQ is γy = b2 γc . Taking account of the indirect (through schooling) and direct effects of c on y, we obtain b = βys βsc + βyc . We have already introduced representative values for βys (0.22) and βyc (0.15). We take as an estimate of βsc the preferred estimate of Winship and Korenman (1999), which is 0.53. From Figure 4 we see that γc = r g h2c , giving us the calculation in the text. Some of the data used in this exercise are not very precisely estimated (h2c in particular) and taking account of assortative mating would raise the estimate marginally, but the main point is robust with respect to reasonable alternative values. If the genetic inheritance of IQ were the only mechanism accounting for the intergenerational income correlations, Figure 1 would represent a set of poorly laid bricks on a barely tilted surface rather than the urban skyline it in fact resembles. The likelihood that a child of the richest decile would attain the top income decile would exceed that of the poor by twelve percent, rather than by the sixteen times observed in Figure 1 (or 44 times for the case of ρ = 0.5).4

March 31, 2001


13

Could it be that cognitive skills not measured on existing test instruments are both highly heritable and have a major impact on earnings, thereby possibly explaining a more substantial fraction of the transmission process? The search for general cognitive measures that are substantially uncorrelated with IQ and predictive of success in adult roles began with Edward Thorndike’s paper (Thorndike 1919) on “social intelligence.” Some alternative test instruments, such as Robert Sternberg and collaborators’ “practical intelligence” (Sternberg, Wagner, Williams and Horvath 1995, Williams and Sternberg 1995), predict economic success in particular occupations. But the fact that the quest that Thorndike launched has yielded no robust alternative to IQ cautions against treating the possible existence of economically important but as yet unmeasured general cognitive skills as anything more than speculation. Indeed we are inclined to think that available estimates overstate the importance of general cognitive skill as a determinant of earnings for the simple reason that taking a test is more than a little like doing a job—the results measure performance, which is the joint effect of skill along with the disposition to follow instructions, persistence, work ethic, and other traits likely to contribute independently to one’s earnings. H. J. Eysenck (1994):9 writes: Low problem solving in an IQ test is a measure of performance; personality may influence performance rather than abstract intellect, with measurable effects on the IQ. An IQ test lasts for up to 1 hour or more, and considerations of fatigue, vigilance, arousal, etc. may very well play a part. Thus some of the explanatory power of the cognitive measure in predicting earnings does not reflect cognitive skill but rather other individual attributes contributing to the successful performance of tasks. The importance of the performance rather than capability aspect of the test score as a determinant of earnings is suggested Bishop (1991), who found a measure of computational speed on time tested exams to be a more robust predictor of earnings than two alternative measures—a normalized sum of academic tests as well as a measure of technical competence.

5 The Inheritance of Wealth and Educational Attainment If the genetic inheritance of cognitive performance does not explain parent offspring similarity in economic success, what does? The intergenerational inheritance of 4 We are assuming the consensus estimate of the heritability of IQ is 50%. A heritability of 100% would change the relative likelihood just cited from twelve percent to twenty three percent.

March 31, 2001


14

income-earning assets, both material and human, provides an alternative explanation. Table 1 shows that the intergenerational correlation for wealth status is quite substantial—within the range of parent offspring similarity in cognitive scores— and that it exceeds the persistence of family income (to which it contributes) which in turn exceeds the persistence of earnings (which do not include income from property). While comparisons of imprecisely estimated parameters across differing data sets are notoriously unreliable, Mulligan (1997) uses a single data set and common methods for his estimates of persistence with respect to earnings, income, wealth and consumption. His results confirm that consumption, wealth and income regress to the mean more slowly than do earnings or wages. These data are consistent with the inherited wealth account of economic status persistence. But for most individuals and families, income from property constitutes a negligible fraction of their total income. Only among the very well-to-do is property a major source of income. Correspondingly, very few individuals receive inheritances of significant magnitude. Mulligan (1997) estimates that estates passing on sufficient wealth to be subject to inheritance tax in the United States constituted between two and four percent of deaths over the years 1960-1995. Intervivos transfers, however, are considerable. Because wealth inheritance contributes to intergenerational income persistence, one would expect to find that persistence varies with the level of wealth of the individual’s parents. For instance, in a rare study with sufficient sample size to derive good estimates of persistence for different positions in the distribution of income, Corak and Heisz (1999) found that the degree of income persistence for the very rich was markedly greater than for the rest of the distribution, with ρy˜ = 0.8 for the top percentile. However, the apparent simplicity of the inherited wealth account is misleading. The intergenerational persistence of wealth is not explained simply by bequests but reflects as well parent-offspring similarities in traits influencing wealth accumulation, such as orientation towards the future, sense of personal efficacy, work ethic, schooling attainment, and risk-taking. Some of these traits covary with the level of wealth: less well off people are more likely to be risk averse, to discount the future and have a low sense of efficacy, for example (Bardhan, Bowles and Gintis 2000, Fong 2000). Because of this positive effect of wealth on the traits conducive to wealth accumulation, along with the fact that the wealthy often enjoy higher rates of return on their assets than the poor, the transmission processes need not be characterized by regression toward the mean. Thus, if in an equation analogous to (1) for wealth, there is no reason to rule out situations for which βk > 1, implying high levels of wealth persistence and growing wealth inequality over time (progression away from the mean). Periods of apparently increasing wealth inequality such as the United States between the late 18th and late 19th centuries (Williamson and March 31, 2001


15

Lindert 1980) may have been characterized by extraordinary levels of persistence, though there is no direct evidence to confirm this. Thus wealth accumulation over generations has been modeled as a non-ergodic dynamic process in which two or more classes differentiated by wealth level persist indefinitely (Galor and Zeira 1993). Like wealth levels, schooling attainments persist across generations although, as Table 1 indicates, less so. Nonetheless, father-offspring correlations for years of schooling reported by Behrman and Taubman (1989) are substantial (0.34, and would be closer to 0.4 were they corrected appropriately for errors in measurement). As in the case of wealth, the persistence of schooling attainments results from actions taken by parents and offspring, and these are influenced by beliefs and preferences that are themselves subject to intergenerational transmission. Some of the individual dispositions favoring high levels of school attainment—such as a sense of personal efficacy and a low discount rate—are correlated with parental schooling levels and incomes, thereby fostering higher levels of educational attainment by the children of the well educated. This tendency is exacerbated where the children of the less well off experience schooling as a hostile or unpleasant environment, and where their families limited incomes and inability to borrow make the students’ potential labor services valuable to the family. Moreover in a number of countries, including the United States, the rate of return to education is considerably higher for those with more than twelve years of schooling than for others (Ashenfelter and Rouse 2000, Hauser, Warren, Huang and Carter 2000), thereby dampening the incentive for schooling attainment for the children of the less well educated. Unlike investment in capital, however, investment in one’s own earning capacities must eventually confront diminishing returns given the limited nature of one’s working life, thus limiting the degree of persistence and strongly favoring regression toward the mean. Because schooling attainment is persistent and has clear links to skills and perhaps other traits that are rewarded in labor markets, the human capital based account of intergenerational status transmission has strong prima facie plausibility, especially when deployed with the inherited cognitive skills account. Indeed it was once commonly assumed that once adequate measures of schooling quality were developed, the only effects of parental economic status on offspring earnings would operate through effects on cognitive functioning and schooling, with the direct effect of parental on offspring earnings vanishing. But as the measurement of school quality has improved over the years, estimates of the direct effect of parental incomes on offspring incomes have proven resilient. Casey Mulligan (1997), controlling for a large number of measures of school quality as well as the AFQT and standard educational and demographic variables, finds that an estimate of parental income is an important and statistically significant predictor March 31, 2001


16

of the natural logarithm of the hourly wage rate in 1990 and 1991 in the National Longitudinal Study of Youth. A bit more than two fifths of the gross statistical association between parental and offspring economic success (ρy ) apparently operates independently of the influences of these conditioning variables. Specifically, the effect of a change in the logarithm of parental income on offspring’s logarithm of income in a regression conditioned on measures of schooling quantity and quality, employment, and cognitive performance is between two fifths and one half of the gross statistical variation of parental and offspring income unconditioned on these variables. Mulligan’s work repeats the finding in Atkinson et al. (1983) for a sample with direct measures of incomes of fathers and sons in the United Kingdom in which two thirds or more of the substantial intergenerational transmission of income status is independent, in the sense just defined, of the covariation of parental income with a range of measures of son’s schooling including the number of Ordinary and Advanced level exams taken, and the selectivity of the school. Finally, Bowles and Nelson (1974) found that between thirty three and sixty percent of the covariation of parental economic status and respondent’s income (not its logarithm) was not accounted for by the statistical association of parental status with childhood IQ or years of schooling. It is possible, of course, that direct effect of parental incomes measures the effects of wealth inheritance (none of these three studies includes measures of parental wealth or bequests) directly on earnings or indirectly via effects on unmeasured determinants of earnings, such as quality of schooling. But this seems unlikely given the very small number of individuals receiving any significant bequest in the samples in question. A more cautious interpretation is that it indicates an important gap in our understanding of how economic status is passed on from generation to generation: roughly half of the intergenerational transmission coefficient is unaccounted for.5

6 The Relative Importance of Genetics, Culture and Bequests Our approach thus far has been to assess the contribution to intergenerational transmission of measured individual traits such as years of schooling and IQ, building up the intergenerational correlation from well-established empirical estimates. Our finding that schooling and IQ fall far short of an explanation of the intergenerational correlation, however, tells us little about the importance of environmental and genetic contributions to the transmission process, for it is likely that genetic 5 It is also true that we can typically statistically account for less than half of the variance of the

earnings or income using the conventional variables described above. But this fact does not explain our limited success in accounting for the intergenerational correlation, as this measures only that part of the variation of earnings that we can explain statistically by parental economic status.

March 31, 2001


17

influences extend far beyond measured cognitive skills and that environmental influences other than years of schooling (and other educational measures) are also important. Is there evidence that the genetic transmission of traits other than cognitive performance contributes to income persistence across generations? We will measured the genetic contribution by the hypothetical amount by which the intergenerational correlation coefficient (or the Galton measure) would fall if parents and children were genetically unrelated with respect to the traits in question, holding constant all other aspects of the trait transmission and income determination processes. We will proceed cautiously because the concept of the genetic contribution to the intergenerational persistence of a phenotypic trait is often misunderstood. It is sometimes supposed that if the contribution is large then social policy must be ineffective in attenuating the disadvantage suffered by the children of the less well off. This is not so. Let us consider a concrete case. In South Africa roughly a third of the intergenerational correlation of incomes is attributable to the fact that parents and children are of the same race, and race is a powerful determinant of earnings (Hertz 2001). Were parents and children genetically unrelated, or were the phenotypic traits designated by “race” not heritable, parent-offspring similarity in this respect would vanish, and as a result the intergenerational transmission of economic status would decline sharply, demonstrating a substantial role of genetic inheritance on the transmission process. Yet it is clear that the economic importance of the genetic inheritance of physical traits derives from environmental influences. The very definition of races, assortative mating by race, and the discrimination suffered by nonwhites are what make the genetic inheritance of skin color and other racial markers important. Thus the determination of the genetic component in a transmission process says little about the extent to public policy can level a playing field. All genetic effects work through environments, many of which may be changed through appropriate public policies.

March 31, 2001


18

Heritability and Environmental Influence Estimated from Twin Data Suppose an attribute xi of individual i depends linearly on genetic endowment gi , environmental influence ei , and an error term i . We write this as xi = hgi + βei + i .

(1a)

If we normalize xi , gi , and ei , as they vary over all individuals in the population, to have zero mean and unit variance, we call h2 the heritability of the trait x. Now take two individuals, say i = 1, 2, and let rx , re , and rg be the correlation between their x-traits, their environments, and their genetic endowments, respectively. If we assume that within a given family g and e are uncorrelated, and the error terms are uncorrelated with each other and with the other dependent variables, we find rx = E[x1 x2 ] = h2 E[g1 g2 ] + β 2 E[e1 e2 ] = h2 rg + β 2 re ,

(2a)

where E is the expectation operator. Suppose the degree of similarity of the twins’ environments, re , is independent of whether they are identical or fraternal, so refr = re = reid . Then, if the twins are identical, rg = 1, whereas if they are fraternal, rg = 0.5. With these assumptions, and writing rxfr and rxid for the correlation of the x-trait for fraternal and identical twins, respectively, then (2a) gives us rxid = h2 + β 2 reid (3a) and rxfr = h2 /2 + β 2 refr ,

(4a)

h2 = 2(rxid − rxfr ).

(5a)

and, subtracting, we have If, as seems likely, reid > refr , the β and h must be estimated simultaneously from (3a) and (4a). Small differences between reid and refr substantially reduce the estimate of h2 . Accounting for assortative mating makes little difference in the estimates. The analysis here assumes that the effect of genes and environment on income are independent of the levels of each (equation 3a). While common in behavioral genetics, the assumption is quite strong, and a nonlinear model may be more appropriate, see Goldberger (1979), Jencks (1980), and Dickens and Flynn (1999). The fact that the earnings and schooling attainments of identical (more technically, monozygotic) twins appear to be substantially more similar than fraternal March 31, 2001


19

(dizygotic) twins suggests that genetically inherited traits other than cognitive skills may account for some of the intergenerational correlation of earnings.6 Figure 5 presents data from the Twinsburg, U.S. sample, the Australian Twin Register, and the Panel Study of Income Dynamics (U.S.). With a few (critical) ancillary assumptions, these data allow an estimate of the heritability of income considered as a phenotypic expression of individual genotypes (see box). Direct inferences about the degree of heritability of income from these data are likely to be misleading for the following reasons. First, the identical twins appear to have shared more similar environments than fraternal twins and other siblings (Loehlin and Nichols 1976), so environmental differences contribute to the greater similarity of the identical twins. Second, the twin data are not corrected for errors in measurement. An adequate treatment of this problem requires information on the degree of correlation of errors, which seems likely to be higher for the identical than the other twins. If this supposition is correct, error correction would increase the identical correlations by less than the fraternal correlations, lowering the estimate of heritability. Third, the identical twins are all same-sex pairs, while the fraternal twins include mixed sex pairs, and sex contributes to the within pair difference in economic success.7 The brothers data set addresses this latter issue, but includes “social siblings”—unrelated members of the same household, so the genotypic correlation among them is less than 0.5. All three problems imply that the naive calculation of heritability from these data will constitute an overestimate. Study Twinsburga Twinsburgb Australian Twinsc U.S. Brothersd

Identical Correlation 0.56 0.63 0.68

Fraternal Correlation 0.36 0.37 0.32

Brothers’ Correlation

0.45

Figure 5: Earnings and Genetic Similarity (ry )of Identical Twins, Fraternal Twins, and non-Twin Brothers. Notes: (a) Ashenfelter and Krueger (1994), with earnings measure log wage; (b) Rouse (1999) and personal communication, with earnings measure log wage; (c) Miller et al. (1995), with earnings measure log occupational income; (d) Bjorklund and Jantti (1999), with earnings measure log annual earnings. 6 Taubman (1976) uses a sample of identical and fraternal twins to measure the relative importance of genes and environment in income determination. 7 However the difference in the earnings correlation between identical and same sex pairs of fraternal twins in the NRC (Taubman 1976) sample and the Swedish Twin Registry Isacsson (1997) is 0.24 (in both data sets).

March 31, 2001


20

It may be of interest, nonetheless to see what this estimate yields. Twin correlations from Ashenfelter and Krueger (1994) for hourly wages and from Behrman and Taubman (1989) for years of schooling attained yield estimates of h2i = 0.4 for both traits. Assuming that the correlation of parental genotypic and phenotypic income is hy itself, the estimated contribution genetic inheritance to the intergenerational transmission of these measures of status is just the genetic correlation (0.5) times the heritability of the trait (r g h2i ) or one-fifth. While this estimate may overstate its influence, we do not doubt that genetic inheritance plays a role, perhaps a substantial role in the intergenerational transmission of economic status, even if (as we have seen) this process operates substantially independently of the cognitive skills measured on available tests. Similar methods presented in detail in the technical paper mentioned above, yield estimates for the environmental effects of exactly the same magnitude as the genetic effects. Our estimate of the effect on earnings, in standard deviation units, of a standard deviation change in the environmental influences on human capital √ and hence on earnings is 0.2 = 0.45. This may be compared with the normalized regression coefficient for a measure of schooling in an earnings equation, the mean estimate of which in our metaanalysis is just short of half of this magnitude, suggesting that while educational attainment captures important aspects of the relevant environments it is far from inclusive. To complete the analysis we turn to the intergenerational correlation for income, including returns to property, and thus take account of the intergenerational transmission of assets. The contribution of wealth inheritance to the transmission process is simply the correlation of parental income and offspring wealth (ry p,k ) multiplied by the contribution of offspring wealth to offspring income, expressed as a normalized regression coefficient, π , the value of which will vary with the share of returns to wealth, as opposed to earnings or other sources of income, in the individuals’s income. Using plausible estimates of the rate of return on assets, our calculations give the property inheritance contribution to the intergenerational income correlation for parents of average income as 0.18. Because higher income households typically receive a larger share of their income in the form of returns to wealth, the property inheritance contribution will be larger for higher income parents, and because this contribution is simply additive to the other contributions, we would expect persistence to be greater for wealthier families. To calculate the contribution of genetic and environmental contributions to the intergenerational correlation of income we simply multiply the contributions estimated above by the (normalized) path from earnings (or human capital) to income.8 The implied contributions to the intergenerational persistence of income based on 8 We need to assume, plausibly, we believe, that h ∼ = re y˜ p . y = hy˜ and rey p ∼

March 31, 2001


21

the above assumptions are genetic is 0.14, the environmental is 0.14, and the asset based is 0.18, giving an intergenerational correlation of 0.46 which, as one would expect, somewhat exceeds the analogous correlations for earnings alone. These results are illustrated in Figure 6. Earnings environmental genetic

re y˜ p βle = 0.2 h2y˜ r g

= 0.2

ωre y˜ p βle = 0.14 ωh2y˜ r g = 0.14 π rk y˜ p = 0.18

wealth intergenerational correlation

Income

= 0.4

= 0.46

Figure 6: Contribution of Environmental, Genetic, and Wealth Inheritance to Intergenerational Transmission. The estimated contribution of wealth varies with parental income level. The estimates here are for mean parental incomes. Most families are below the mean because of the right skewness of the income distribution, so these estimates of the role of wealth pertain to the relatively well off.

The substantial contribution of wealth inheritance to intergenerational transmission of economic status is unsurprising, at least among the wealthy, and the mechanisms through which it works relatively transparent. But the same cannot be said for the substantial genetic and environmental contributions, for as we have seen, neither IQ nor schooling, nor a combination of the two, provides an adequate account of these influences. Thus both genetic and environmental influences are important but we do not know why.

7

Understanding Intergenerational Status Transmission

There are two common ways to misunderstand the process of intergenerational status transmission. One is to think of the intergenerational transmission process as a literal handing-down of such things as good genes and material wealth—what we call the “family piano” theory of inheritance. The other is to limit the benefits being handed down to skills. What offspring get from their parents may be less important than what parents and children alike are or do. Rather than focusing on the handing-down of skills, a more inclusive approach might be to ask what a transmission process requires. A list of variables contributing in an accounting sense to the intergenerational correlation is of course not an explanation of how transmission works, but it is a very useful start. It is clear that transmission is accomplished by variables meeting just two criteria: (i) they March 31, 2001


22

account for variations in the economic success of the current generation (ii) they are correlated with the economic success of the previous generation. Of course any variable—like good schooling, good health, or good looks—that contributes to economic success in both generations, and for which parent offspring similarity is substantial meets these two criteria. Students of the persistence of economic status frequently identify “skills” as the critical individual traits subject to transmitted on the grounds that skills earn rewards in competitive labor markets. But other traits are persistent across generations and are arguably as important as determinants of income—for example, race, national citizenship, number of children, and as we have seen, wealth ownership. Another example of high levels of parent-offspring similarity for a non-skill trait contributing to economic success is obesity in women, which is a strong predictor of low earnings. Height predicts high earnings for men, and good looks predicts high earnings for both men and women, the latter independently of whether they hold jobs interacting with the public. Bowles, Gintis, and Osborne (2002a) provide a survey of empirical evidence concerning these and many other nonskill determinants of economic success. While many of these traits are inherited from parents (in at least one of the usual senses of vertical cultural transmission, bequest, and genetic transmission) in some cases parent-offspring similarity arises by other means—common nationality, for example. Two such variables will illustrate our argument: (i) group membership and (ii) personality and other individual traits that contribute to earnings but are not skills in the usual sense. Suppose economic success is influenced not only by a person’s traits, but also by characteristics of the individuals with whom the person typically interacts. Suppose also that these interactions are not random, but are more likely to take place with individuals who share membership in a group. Groups may differ in the average level of schooling, economic success, cognitive functioning, and wealth level. Groups may be residential neighborhoods, ethnic or racial groups, linguistic groups, citizens of a nation, or any other set of individuals who typically interact more with one another than with those who do not share common membership. Group effects on economic success are well documented and may arise for a number of reasons (Durlauf 1997, Borjas 1995), including discrimination, conformist effects on behavior, differential access to information, complementarities in production (a person’s productivity is higher when working with better educated people, for example). The importance of group effects on intergenerational persistence of economic status is suggested in the work of Cooper et al. (1994) on the importance of one’s neighborhood, and by the fact that the correlation among brothers’ earnings for the United States estimated by Bjorklund, Eriksson, Jantti, Raaum and Osterbacka (1999), namely 0.43, falls by 0.1 when the sample is restricted to March 31, 2001


23

whites; i.e., brothers typically share racial status, and when that contribution to the similarity of economic outcomes between siblings is removed from the sample, the brothers’ correlation falls by roughly a quarter. A less transparent but common source of group effects is the fact that individuals typically coordinate their activities by the use of conventions, i.e., behaviors that are individually beneficial as long as most other members of the group behave the same way. Driving on the right side of the road is an example, as are many rules of division of resources such as fifty-fifty, finders keepers, or first-come first-served, and customs such as mode of dress, language use, and manners. In residential communities, behaviors contributing to maintaining personal safety, environmental quality and other amenities may also be conventions. Conventions may persist over long periods of time, and they differ from group to group. Some contribute to the economic success of the members of the group (e.g., truth telling or respect for property rights) while others are economically counter-productive (e.g., toleration of corruption), thereby giving rise to group effects on individual income. Individuals are born into many groups (residential, racial, national, linguistic and others) and group membership often persists over many generations. Because group membership contributes to economic success in current and parental generations and persists across generations, it is thus a contributor to the intergenerational transmission process. Our second example concerns individual dispositions such as a sense of personal efficacy, work ethic, or a low time discount rate (strong orientation toward the future). These traits are not skills in any sense. Rather they contribute to individual economic success because they increase the gains from exchange in situations where contracts are incomplete. The importance of these aspects of personality stems from the fact that in an important class of exchanges, including the hiring of labor, borrowing and lending, or the exchange of goods of uncertain quality, it is impossible to specify all relevant aspects of the exchange in a contract enforceable by the courts. Where this is the case, the actual terms of the exchange are influenced by the degree of trust, honesty, hard work, and other dispositions of the parties to the exchange. The empirical importance of these traits are suggested in a number of studies (Duncan and Dunifon, 1998, Heckman, Hsee and Rubinstein, 2001, Bowles, Gintis, and Osborne, 2002a). As earnings constitute by far the largest share of income for almost all individuals, the workings of the labor market are especially important in the process of intergenerational transmission. A widespread form of employment is the following: the employer offers a wage and the promise to retain the employee in subsequent periods as long as the employees work performance is found adequate. To make the threat of termination effective, of course, the wage and other aspects of the job must be worth keeping, that is preferable to termination, subsequent job search and March 31, 2001


24

the expected alternative employment. The employee’s personality will influence how effective this contract is, and hence the value of the employee in the eyes of the employer (Bowles, Gintis and Osborne 2001). Consider, for example, the extent to which the employee discounts the future (the rate of time preference). A very present-oriented employee will not value the promise of continued employment in the future, hence will require a higher wage to motivate hard work in the present, and hence is less likely to be employed. Another example is the sense of personal efficacy: fatalistic workers who believes that the probability of termination is unaffected by their own actions will be costly to motivate under this type of contract. Osborne (2000) has studied the economic importance and intergenerational persistence of fatalism, as measured by the Rotter Scale, a common measure of the degree to which individuals believe that important events are caused by external events rather than by their own actions. Her study of a sample of United States women and their parents has produced two major findings. First, the Rotter Scale has a statistically significant and large influence on earnings. Second, the Rotter Scale is persistent from parents to offspring. Osborne has also studied a sample of women in the England and found that measures of social maladjustment taken at age eleven (the Bristol Social Adjustment Scale), such as aggression and withdrawal, are strong predictors of earnings at age thirty three. The influence of the Rotter Scale in Osborne’s United States study is somewhat smaller than the average influence of IQ in our metaanalysis. The influence of aggression and withdrawal is considerably larger. There are no measures of intergenerational persistence in the Osborne’s English data set, but other studies suggest that parent child similarity in measures of social maladjustment may be quite high. Osborne’s work thus suggests that the intergenerational transmission of personality traits (whether genetic or cultural) may be an important channel explaining the intergenerational persistence of income. We know relatively little about the intergenerational transmission process for personality traits relevant to economic success, other than cognitive functioning. However Melvin Kohn’s 1969’s study of child rearing values of parents suggests that at least for some traits, experiences in the workplace are generalized and passed by the process of vertical cultural transmission. Kohn categorizes his adult sample by the degree of self-determination that each experiences on the job, ranging from those who are relatively unsupervised to those who are closely directed by superiors. Kohn found that parents with high levels of occupational self-direction emphasize curiosity, self-control, happiness and independence as values for their children, while those who are closely monitored by supervisors at work emphasize conformity to external authority. Two thirds of the statistical association between class position and parental values was explained by the covariation between class and occupational self-direction. He concluded: March 31, 2001


25

Whether consciously or not, parents tend to impart to their children the lessons derived from their own social class and thus help prepare their children for a similar class position. Kohn did not measure the effects of parental values on their behaviors as parents, or effects on child development. However, subsequent work by Osborne does suggest that the degree of self-direction has significant effects on earnings later in life. Recent work by Yeung, Duncan and Hill (1997) shows that parental behavior, including church attendance, membership in social organizations, and such precautionary behavior as seat belt usage, have significant impacts on their children’s earnings.

8

Conclusion

How level, then, is the playing field, and what is it that accounts for its tilt? Recent evidence points to a higher level of intergenerational transmission of economics position than was previously thought to be the case. Moreover, although the level of intergenerational IQ inheritance is also considerable, the latter accounts for little of the former. Some combination of environmentally and genetically transmitted physical characteristics, expectations, preferences, behavioral patterns and personality traits probably accounts for a significant portion of the correlation between the economic positions of parents and children. These noncognitive individual traits, like group membership, affect earnings and display parent-offspring similarity. They thus join the genetic and cultural inheritance of cognitive skills, property bequests (and the other influences of wealth) and parent offspring similarity in schooling attainments as aspects of the process of intergenerational transmission of economic status. A policy maker with a mandate to reduce the influence of parental economic status on the likely economic success of their children might find Figure 2 and the data we have surveyed illuminating. The mandate would be to impede the influence of parental income as it proceeds along the paths in the figure eventually influencing the income of the offspring, to the extent that this could be accomplished in ways consistent with other valued objectives, such as allocative efficiency, privacy, and freedom of choice. The policy maker’s task then would be to determine which links in this network of paths account for most of the intergenerational transmission of status, and which ones might be broken or reduced in importance. Obvious candidates whose importance is confirmed in our study would be the correlations between parental income on the one hand and both the environmental influences on human capital acquisition and the wealth level of the offspring on the other. The elimination of racial discrimination would reduce the genetic heritability of income, March 31, 2001


26

as would a school system which was more homogeneous by income class and which therefore did not promote assortative mating by potential income. But the policy maker seeking to level the intergenerational playing field would also know that these policies would touch at most a half of the observed transmission processes at work. The black box that remains is a challenge to ongoing research in the social sciences, behavioral genetics, and other fields.

References Ashenfelter, Orley and Alan Krueger, “Estimates of the Economic Return to Schooling from a New Sample of Twins,” American Economic Review 84,5 (December 1994):1157–1172. and Cecilia Rouse, “Schooling, Intelligence, and Income inAmerica,” in Kenneth Arrow, Samuel Bowles, and Steven Durlauf (eds.) Meritocracy and Economic Inequality (Princeton University Press, 2000) pp. 89–117. Atkinson, A. B., A. K. Maynard, and C. G. Trinder, Parents and Children: Incomes in Two Generations (London: Heinemann, 1983). Bardhan, Pranab, Samuel Bowles, and Herbert Gintis, “Wealth Inequality, Credit Constraints, and Economic Performance,” in Anthony Atkinson and François Bourguignon (eds.) Handbook of Income Distribution (Dortrecht: NorthHolland, 2000). Becker, Gary S., “Family Economics and Macro Behavior,” American Economic Review 78 (1988):1–13. and Nigel Tomes, “Human Capital and the Rise and Fall of Families,” Journal of Labor Economics 4,3 (July 1986):S1–39. Behrman, Jere R. and Paul Taubman, “Is Schooling ’Mostly in the Genes’?,” Journal of Political Economy 97,6 (December 1989):1425–46. , Robert A. Pollak, and Paul Taubman, From Parent to Child: Intrahousehold Allocations and Intergenerational Relations in the U.S. (Chicago: University of Chicago Press, 1995). Bishop, John, “Achievement, Test Scores and Relative Wages,” in Marvin Kosters (ed.) Workers and their Wages (Washington, DC: AEI Press, 1991) pp. 146–186. Bjorklund, Anders and Markus Jantti, “Intergenerational mobility of Socioeconomic Status in Comparative Perspective,” 1999. Stockholm University and Abo Akademi University. , Tor Eriksson, Markus Jantti, Oddbjorn Raaum, and Eva Osterbacka, “Brother Correlations in Earnings in Denmark, Finland, Norway and Sweden Compared to the United States,” 1999. unpublished.

March 31, 2001


27

Blau, Peter and Otis Dudley Duncan, The American Occupational Structure (New York: Wiley, 1967). Borjas, George, “Ethnicity, Neighborhoods, and Human Capital Exernalities,” American Economic Review 85,3 (June 1995):365–390. Bouchard, T. J., Jr. and M. McGue, “Familial Studies of Intelligence,” Science 212 (1981):1055–1059. Bowles, Samuel, “Schooling and Inequality from Generation to Generation,” Journal of Political Economy 80,3 (May/June 1972):S219–251. and Herbert Gintis, “The Inheritance of Economic Status: Education, Class and Genetics,” in Marcus Feldman and Paul Baltes (eds.) International Encyclopedia of the Social and Behavioral Sciences: Genetics, Behavior and Society (New York: Oxford University Press and Elsevier, 2001). and Valerie Nelson, “The ‘Inheritance of IQ’and the Intergenerational Reproduction of Economic Inequality,” Review of Economics and Statistics 56,1 (February 1974). , Christina Fong, and Herbert Gintis, “Reciprocity and the Welfare State,” in Samuel Bowles, Richard Boyd, Ernst Fehr, and Herbert Gintis (eds.) Foundations of Cooperation in Society (, forthcoming). , Herbert Gintis, and Melissa Osborne, “Incentive-Enhancing Preferences: Personality, Behavior and Earnings,” American Economic Review 91,2 (May 2001):??–??. , , and , “The Determinants of Individual Earnings: Skills, Preferences, and Schooling,” Journal of Economic Literature (2002). in press. , , and , Intergenerational Inequality (New York and Princeton: Russell Sage and Princeton University Press, 2002). Cooper, Suzanne, Steven Durlauf, and Paul Johnson, “On the Evolution of Economic Status across Generations,” American Economic Review 84,2 (1994):50– 58. Corak, Miles and Andrew Heisz, “The Intergenerational Earnings and Income Mobility of Canadian Men: Evidence from the Longitudinal Income Tax Data,” Journal of Human Resources 34,3 (1999):505–533. Devlin, Bernie, Michael Daniels, and Kathryn Roeder, “The Heritability of IQ,” Nature 388 (31 July 1997):468–471. Dickens, William T. and James R. Flynn, “Heritabilty Estimates vs. Large Environmental Effects: The IQ Paradox Resolved,” 1999. The Brookings Institution and The University of Otago, version 3.2. Duncan, Greg J. and Rachel Dunifon, “’Soft-Skills’and Long-Run Market Success,” Research in Labor Economics 17 (1998):123–150. March 31, 2001


28

Durlauf, Stephen, “The Memberships Theory of Inequality: Ideas and Implications,” 1997. University of Wisconsin. Eysenck, H. J., “Intelligence and Introversion-extraversion,” in Robert J. Sternberg and Patricia Ruzgis (eds.) Personality and Intellegence (Cambridge University Press, 1994) pp. 3–31. Feldman, Marcus W., Sarah P. Otto, and Freddy B. Christiansen, “Genes, Culture, and Inequality,” in Kenneth Arrow, Samuel Bowles, and Steven Durlauf (eds.) Meritocracy and Economic Inequality (Princeton University Press, 2000) pp. 61– 85. Fong, Christina, “Social Insurance or Conditional Generosity: The Role of Beliefs about Self- and Exogenous-Determination of Incomes in Redistributive Politics,” 2000. Washington University Department of Political Science. , “Social Preferences, Self-Interest, and the Demand for Redistribution,” Journal of Public Economics (forthcoming). Galor, Oded and Joseph Zeira, “Income Distribution and Macroeconomics,” Review of Economic Studies 60,1 (1993):35–52. Galton, Francis, Natural Inheritance (London: Macmillan, 1889). Goldberger, Arthur S., “Heritability,” Economica 46 (1979):327–347. , “Economic and Mechanical Models of Intergenerational Transmission,” American Economic Review 79,3 (June 1989):504–513. , A Course in Econometrics (Cambridge, MA: Harvard University Press, 1991). Hauser, Robert M., John Robert Warren, Min-Hsiung Huang, and Wendy Y. Carter, “Occupational Status, Education, and Social Mobility in the Meritocracy,” in Kenneth Arrow, Samuel Bowles, and Steven Durlauf (eds.) Meritocracy and Economic Inequality (Princeton University Press, 2000) pp. 179–229. Heckman, James, Jingjing Hsee, and Yona Rubinstein, “The GED as a Mixed Signal,” American Economic Review 91,2 (May 2001):??–??. Hertz, Thomas, “Essays on Education and Poverty in South Africa,” 2001. Unpublished Ph.D. Disseration, University of Massachusetts. Isacsson, Gunnar, “Estimates of the Return to Schooling in Sweden From a Large Sample of Twins,” 1997. Jencks, Christopher, “Heredity, Environment, and Public Policy Reconsidered,” American Sociological Review 45,5 (1980):723–736. Levine, David I., “Choosing the Right Parents: Changes in the Intergenerational Transmission of Inequality Between the 1970s and the early 1990s,” 2001. Haas School of Business, University of California, Berkeley. Loehlin, John and Robert Nichols, Heredity, Environment, and Personaltiy (Austin, TX: University of Texas Press, 1976). March 31, 2001


29

Mazumder, Bhashkar, “Earnings Mobility in the U.S.: A New Look at Intergenerational Inequality,” November 2000. Department of Economics, University of California at Berkeley. Miller, Paul, Charles Mulvey, and Nick Martin, “What do Twin Studies Reveal About the Econmic Returns to Education: a Comparison of Australian and U.S. Findings,” American Economic Review 85,3 (June 1995):586–599. Mulligan, Casey, Parental Priorities and Economic Inequality (Chicago: University of Chicago Press, 1997). , “Galton vs. the Human Capital Approach to Inheritance,” Journal of Political Economy (1999). Osborne, Melissa, “The Power of Personality: Labor Market Rewards and the Transmission of Earnings,” 2000. University of Massachusetts. Plomin, Robert, “Genetic and General Cognitive Ability,” Nature 402,2 (December 1999):c25–c29. Rouse, Cecilia Elena, “Further Estimates of the Ecoomic Return to Schooling From a New Sample of Twins,” Economics of Education Review 18 (1999):149–157. Solon, Gary R., “Intergenerational Income Mobility in the United States,” American Economic Review 82,3 (June 1992):393–408. , “Intergenerational Mobility in the Labor Market,” in Orley Ashenfelter and David Card (eds.) Handbook of Labor Economics (Amsterdam: North-Holland, 1999). Sternberg, Robert J., Richard K. Wagner, Wendy M. Williams, and Joseph Horvath, “Testing Common Sense,” American Psychologist 50,11 (November 1995):912– 927. Taubman, Paul, “The Determinants of Earnings: Genetic, Family, and Other Environments; A Study of White Male Twins,” American Economic Review 66,5 (December 1976):858–870. Thorndike, Edward L., “Intelligence and Its Uses,” Harper’s Monthly Magazine 140 (December/January 1919):227–235. Warren, John and Robert Hauser, “Social Stratification Across Three Generations: New Evidence from the Wisconsin Longitudinal Study,” American Sociological Review 62 (August 1997):561–572. Williams, Wendy M. and Robert J. Sternberg, Success Acts for Managers (Florida: Harcourt Brace, 1995). Williamson, Jerry and Peter Lindert, American Inequality: A Macroeconomic History (New York: Academic Press, 1980). Winship, Christopher and Sanders Korenman, “Economic Success and the Evolution of Schooling with Mental Ability,” in Susan Mayer and Paul Peterson March 31, 2001


30

(eds.) Earning and Learning: How Schools Matter (Washington, DC: Brookings Insitution, 1999) pp. 49–78. Yeung, W. J., Greg J. Duncan, and M. S. Hill, “Putting Fathers Back in the Picture: Parental Activities and Children’s Adult Attainments,” 1997. Paper presented at the NICHD Family and Child Well-Being Network’s Conference on Father Involvement, Washington, D.C. Oct. 10-11, 1996. Zimmerman, David J., “Regression Toward Mediocrity in Economic Stature,” American Economic Review 82,3 (June 1992):409–429. e\Papers\JEP-paper\Intergenerational Inequality March 31, 2001

March 31, 2001