LONG-TERM TRENDS IN AMERICAN INTERGENERATIONAL OCCUPATIONAL MOBILITY

Boris A. Simkovich
Harvard University

1 Introduction

An interest in social and economic mobility is a common thread in the social sciences. In fields such as sociology, study of the topic is central to the discipline. In other fields, such as history, investigation of mobility occupies a prominent position, but is a comparatively new addition to the standard research agenda. Finally, in fields such as economics, mobility is not always an explicit subject of concern, but one that is nonetheless implicit in many important analyses. In economics, for example, almost every discussion of inequality or poverty implicitly involves issues related to the economic mobility of individuals or groups of individuals.

While a concern for mobility is shared by social scientists of almost all persuasions, research on the subject has not been entirely catholic. This is particularly evident if one partitions mobility research into two sub-fields: (a) the study of intragenerational mobility, or changes in the status of individuals over the course of their adult lives (where "status" is measured in terms of social class, occupation, or income level), and (b) the study of intergenerational mobility, or changes in the status of adults relative to the status of their parents. Whereas researchers from a variety of the social sciences — including economics — have carried out investigations of intragenerational mobility, the analysis of intergenerational mobility has been primarily the domain of historians and sociologists.

Recently, however, a number of economists have produced important contributions to the literature on intergenerational mobility. Works such as Solon (1992) and Zimmerman (1992) add markedly to our understanding of economic mobility in the United States. The Solon and Zimmerman papers, for example, provide some of the first estimates of intergenerational income correlations based on data from nationwide samples of individuals, and their studies also employ new techniques to eliminate potential biases in the measurement of parent-child income correlations.

Nevertheless, by calling attention to the extent of intergenerational economic mobility in contemporary America, these and other related studies raise two obvious questions. First, how does the pattern of intergenerational mobility today compare to that of the past? Second, to the extent that changes in mobility have occurred, how can one account for those changes? The answers to these questions are the focus of this paper.

Documenting trends in American intergenerational mobility is the task of the first half of this paper. This analysis concentrates on data from four nationally representative samples of white men aged 25–34 at different points of time during the past 100 years. While the data in these samples refer to a somewhat narrow segment of the overall population (young white males), the results they supply are nonetheless useful in gauging the mobility experience of a sizable fraction of all Americans. Similarly, although data constraints dictate that occupation, not income, be used to measure economic status when tracking long-term trends in intergenerational mobility, the use of this metric has advantages as well as disadvantages. On the one hand, the categorical nature of occupational information greatly complicates attempts to determine trends in the actual correlation of parents' and children's economic status. A discussion of such trends thus is not included in this paper. On the other hand, observed patterns of intergenerational occupational mobility are direct evidence of the extent and direction of intergenerational flows between specific segments of the economic hierarchy. Such patterns therefore disclose a great deal about the importance — or unimportance — of parental economic status in the determination of an individual's own economic status.

It is for this reason that the results summarized in the first half of this paper are so important. Those results reveal that, during the past century, there has been both significant change as well as surprising continuity in the patterns of American intergenerational occupational mobility. Much of the change reflects the dramatic shifts which have occurred in the composition of the American labor force. Thus, for men of all family origins, there has been a sharp decline in the proportion entering farm occupations, and the growth in the share of the labor force in white collar occupations has been mirrored in improved opportunities for the advancement of blue collar and farm sons into such occupations. Not all changes in mobility patterns can be attributed, however, to long-term shifts in the composition of the labor force, nor have all changes taken place over the entire course of the last century. In recent decades, for example, there has been for men of all family origins a perceptible decline in the proportion achieving the economic status associated with a high white collar occupation.

In addition, while change has characterized a number of important trends in the overall pattern of American mobility, there has also been marked continuity in many aspects of that pattern. Intergenerational flows between a number of occupational categories have remained startlingly similar over the past 100 years. In 1900, for example, the share of sons from middle status origins ending up in low blue collar jobs was 22.5 percent. In 1987, it was 23.3 percent. Likewise, relative flows between a number of occupational categories have remained essentially constant. Thus, in both 1900 and 1987, the share of farm sons entering low blue collar occupations was about twice the share entering professional, technical, and managerial positions.

The observed continuity in a number of mobility patterns should not overshadow, however, the many important changes which have occurred in such patterns. One of those changes — the long-term rise in the accessibility of white collar occupations — is arguably the single most important aspect of the evolution of American intergenerational mobility. The second half of this paper therefore considers the forces which may have been responsible for that development.

The discussion in the second half of the paper begins with a description of two competing hypotheses concerning the evolution of intergenerational mobility in the United States. The first hypothesis argues that the increased accessibility of white collar occupations has been the result of the tremendous expansion during the past century in the provision of public education. In contrast, the second hypothesis contends that increased spending on education has been merely a response to a more fundamental economic stimulus: technological change.

To determine which of these competing hypotheses is best supported by available evidence, the paper introduces a general equilibrium model of intergenerational occupational mobility. The parameterization of that model is not yet sufficiently advanced, however, to admit any definitive conclusions concerning the merits of the two hypotheses.

The remainder of this abstract is organized as follows. In the next section, I briefly summarize the existing literature on trends in American intergenerational occupational mobility. In section 3, I detail new evidence concerning such trends. Section 4 presents alternative explanations for the observed trends and describes a model for the evaluation of those explanations. Section 5 concludes.

2 The existing literature

Sociologists and historians have been responsible for most of the existing evidence on long-term trends in American intergenerational occupational mobility. That evidence is not as substantial, however, as one might hope. On the one hand, the research conducted by sociologists has carefully documented mobility in the United States during the 1960s and 1970s, and it also has motivated subsequent — albeit less detailed — research on mobility in the 1980s. In addition, sociologists have drawn attention to studies in other fields which provide insight into the mobility regime of the 1950s.

For perspectives on mobility in the United States prior to World War II, however, one must turn to the results of historical research. This research often documents mobility patterns for periods as early as the first half of the 19th century. Nevertheless, much of the historical research suffers from two disadvantages. First, it focuses only on specific cities or towns in the United States. Second, it generally tracks mobility in those areas over relatively limited periods of time.

As a result, while there is a large body of evidence concerning intergenerational occupational mobility in America during particular decades of the post-war era, knowledge concerning mobility patterns in other time periods is considerably more limited. This has seriously hampered attempts to draw accurate conclusions concerning the long-term course of intergenerational mobility in the United States.

3 Intergenerational occupational mobility during the past 100 years

The data

While investigators traditionally have faced serious obstacles to documenting long-term trends in American intergenerational mobility, such obstacles are beginning to crumble in the face of fresh research. Two developments in this regard are particularly noteworthy. First, a team of sociologists has recently constructed a dataset with extensive evidence on intergenerational mobility in the entire United States during the period 1880–1900. In addition, data collected by the Census Bureau in 1987 provide important new insights into mobility in contemporary America. As a result, sufficient evidence now exists to sketch the major trends in the evolution of American intergenerational occupational mobility over the past century.

In the analysis that follows, I make use of the two datasets just described as well two other datasets created during the past 35 years. Together, these datasets document mobility patterns for nationally representative samples of individuals at four different points in time: 1900, 1962, 1973, and 1987. The first dataset is the National Panel Study (NPS), which was compiled at the University of Washington during the late 1980s. It is based on a sample of matched records from the 1880 and 1900 population census manuscripts. The second dataset is a subset of the well- known Occupational Changes in a Generation (OCG) study conducted by Blau and Duncan in 1962. The third dataset is a subset of the 1973 follow-up to the 1962 OCG. Finally, the fourth dataset is a subset of the results in the Census Bureau's 1987 Survey of Income and Program Participation (SIPP), Wave 2.

All of the data analyzed in this paper pertain to the mobility experience of white, native- born males aged 25–34 in the relevant years. This group was the focus of the compilers of the National Panel Study in their efforts to match a sample of 1880 census records with records from 1900. As a result, subsamples of individuals with the same characteristics were selected from the more comprehensive datasets available for the postwar era. The numbers of observations in each of the samples used in this study are as follows: NPS, 1,814; OCG 1962, 3,816; OCG 1973, 7,324; SIPP 87/2, 1,401.

All of the samples described above provide information on the occupation of the individuals in the sample. In addition, information on each individual's parents' occupations is also available. The parental information relates to occupations held when the sample members were children. In this paper, discussion centers on comparisons between sample members' occupations and the occupations of their fathers. If an individual's father was not present when that person was a child, the occupation of the head of the childhood household — whether the head was male or female — is used as the sample member's "father's" occupation.

Documenting the trends: methods

To facilitate the description of mobility trends, the analysis in this paper employs a four- category occupational classification system. That system categorizes jobs as either high white collar, middle status, low blue collar, or farm. The first category consists of professional, technical, managerial, and official occupations. The second includes sales, clerical, and crafts occupations. The third covers operative, laborer, and service occupations, and the fourth consists of farm owners, farm tenants, and farm laborers. The detailed occupational information available in the four samples has been used to classify each sample member's own and parental occupations into one of these four categories.

The classification of all occupations into four categories enables the creation of a compact mobility table for each sample. A mobility table is simply a cross-classification of individuals' occupations and the occupations of their parents. Table 1 presents the mobility table for the 1900 NPS sample.

TABLE ONE

Mobility tables can be used for a number of different purposes. In this paper, they are used primarily to facilitate the calculation of a convenient measure of mobility, the outflow percentage. One outflow percentage exists for each cell in a mobility table. The percentage is simply a given cell's share of all observations in its row. It thus describes the fraction of all men from a particular family origin who are employed in a specific category of occupations. Table 1 indicates, for example, that in 1900 14.5 percent of all men whose fathers were farmers became workers with middle status jobs. This, then, is the outflow percentage for farm-to-middle status intergenerational occupational flows.

Documenting the trends: results

Figures 1 and 2 plot time series of outflow percentages for sons from farm and low blue collar families. In each figure, the labels "farm," "low," "middle," and "high" mark the lines plotting flows into farm, low blue collar, middle status, and high white collar occupations, respectively.

The most dramatic change is evident in Figure 1, which focuses on the experience of men from farm origins. There, the decline in the share of the American labor force in farming is reflected in a concurrent decline in the farm-to-farm outflow percentage. In 1900, 60 percent of the young sons of white American farmers were farmers themselves; by 1973, the share had dropped to only 13.5 percent. Since 1973, however, the share of farm sons entering farm occupations has rebounded, reaching half what it had been in 1900.

Perhaps the most remarkable aspect of the flows depicted in Figure 1 is not the variation in the share of young men entering farm occupations, but rather the way in which the flows to the other three occupational categories have changed. In particular, those three flows have maintained essentially constant proportions since 1900. For example, at the beginning of the century, the share of farm sons entering low blue collar occupations was 2.1 times the share of farm sons entering high white collar occupations. In 1987, the equivalent ratio was also 2.1. The lock-step fashion in which farm-to-non-farm flows have changed over the past century raises interesting questions about the forces affecting such mobility.

While the trends in Figure 2 are not nearly as dramatic as those in Figure 1, they are intriguing nonetheless. The flows in Figure 2 document the occupational attainment of young men from low blue collar family origins. Once again, the historical decline in farm employment is mirrored in declining intergenerational flows into the farm sector. The declining flows into farm employment have been offset, however, by rising opportunities for advancement into middle status and high white collar occupations. This, in turn, has meant surprising continuity in a flow which has been of particular interest to a number of previous researchers: the share of men from low blue collar origins who have remained low blue collar workers themselves. Throughout the past century, that flow has remained essentially constant at the relatively high level of 40 percent.

Although the results in Figures 1 and 2 depict only the experience of men from farm and low blue collar family origins, it is possible to carry out similar analyses for men from middle status and high white collar families. Unfortunately, space limitations preclude a presentation of such results in this abstract. It would be inappropriate, however, to end the foregoing discussion without a description of an important development involving occupations in these higher status categories. This development is the long-term rise in the share of men from all family origins who have been able to gain employment in white collar jobs. Table 2 documents this development.

INSERT TABLE 2

The results in Table 2 indicate that, between 1900 and 1962, fathers in all occupational categories saw significantly greater shares of their sons entering white collar occupations. Since 1962, trends in the share of men entering such occupations have differed somewhat from family origin to family origin. Nevertheless, it is still the case that, regardless of family background, a young, white male today has a significantly greater chance of entering a white collar occupation than a young, white male at the turn of the century.

4 Explaining the trends

Some hypotheses

The increased accessibility of white collar occupations has been one of the most important developments in American intergenerational occupational mobility. It has meant significantly improved economic opportunity for individuals from all family origins. Yet this important development raises an equally important question: why? What forces have been responsible for this critical change in the pattern of American intergenerational mobility?

While there are any number of possible answers to such a question, two explanations appear to offer the greatest promise. Both explanations stress the link between education and mobility.

The first explanation stresses supply-side forces. It begins by noting the tremendous expansion in public education which has occurred during the past century. It then argues that it has been precisely this expansion which has been the source of the increased accessibility of white collar occupations.

The second explanation focuses on demand-side forces. It contends that increased spending on education has been merely a response to a much deeper force in the economy: technological change. Such change, it is argued, has motivated parents to demand more education for their children, and the resulting rise in educational expenditures has thereby enabled increased mobility into white collar occupations.

There are actually two variants of the second explanation. The first argues that the most important aspect of technological change has been its effect on the level of income. The second contends that the relevance of technological change lies in the bias it has exhibited. To summarize what would otherwise be a rather lengthy discussion, the first variant relies on assumptions concerning the income elasticity of demand for children's education, while the second focuses on altered incentives for such education.

Weighing the hypotheses

None of the explanations just outlined are necessarily mutually exclusive. It is conceivable that elements of each have played a role in the observed change in the American intergenerational mobility regime. To determine, however, which of the explanations best accounts for that change, one needs a framework for weighing the relative significance of the different forces which have influenced mobility in the United States.

Until recently, such a framework has been unavailable. Most theoretical discussions of the forces affecting long-term trends in intergenerational mobility have been either relatively informal or have focused on intergenerational income mobility. Fortunately, research concurrent to the work summarized in this paper has led to a framework well-suited to an analysis of the hypotheses discussed above. That framework consists of a general equilibrium, overlapping generations model of intergenerational occupational mobility. The essential components of the model are as follows:

[1] Individuals in the economy live for two periods. During their first period of life, individuals are "children" who are incapable of any productive activity. In their second period of life, individuals become "adults." At the beginning of this second period, every individual gives birth to a child. Throughout the second period, adults inelastically supply their labor services in a competitive labor market.

[2] There are two types of labor services, manual and non-manual. All individuals are innately capable of supplying manual labor services when they are adults. Individuals are only capable of non- manual work, however, if they receive an "education" as children.

[3] A child receives an education if her parent contracts for the "teaching" services of non-manual workers. A specific amount of non-manual labor is required to educate every child. This amount varies, however, from child to child depending on the child's "learning ability," which is determined stochastically at birth but is correlated with the learning ability of the parent. Every parent observes her child's learning ability as soon as the child is born.

[4] A parent desires an education for her child because the lifetime utility of the child enters the parent's utility function. At the same time, the parent's own consumption is also an argument in her utility function. Preferences for consumption and the utility of children do not vary between parents.

[5] There are no capital markets for the funding of education. There is, however, a government subsidy financed by an income tax. Whether or not a parent invests in an education for her child depends, therefore, on the parent's after-tax income, the cost of the education net of the subsidy, and the expected benefit to the child of education. All of these depend, in turn, on the level of wages in the economy.

[6] Wages are determined as the marginal product of each type of labor in the aggregate production function for the economy's single consumption good. All manual labor is employed in the goods- producing sector, but non-manual labor is supplied to both the goods-producing and educational sectors. Non-manual workers employed in the education sector are assumed to receive the same wage as non-manual workers in the goods-producing sector.

[7] Manual and non-manual labor are the only inputs in the aggregate production function. There is no capital accumulation; all income is spent on either consumption or educational services. Technological change is exogenous to the model, and may be biased or factor neutral.

To characterize equilibrium in a model such as the one just outlined, one needs to specify and parameterize a number of different functions. That parameterization must be done for two time periods: the late 19th century and the late 20th century. At the present time, complete parameterization of the model is still underway, and it is therefore not possible to draw any firm conclusions concerning the hypotheses discussed earlier.

5 Concluding comments

While the modeling necessary to account for the evolution of American intergenerational mobility is still underway, the results in this paper nevertheless add markedly to our understanding of that evolution. The results indicate that both continuity and change have characterized the various patterns of intergenerational occupational mobility in the United States during the past century.

That there has been continuity in the midst of such dramatic shifts in the composition of the American labor force is perhaps the most surprising of the results summarized above. Yet it is the long- term rise in the accessibility of white collar occupations that has been especially relevant for so many young American men. An explanation of that trend— as well as further details concerning the evolution of American intergenerational mobility — are the promise of the full version of this paper.