Academic journal article Review of Social Economy

Structural Change in U.S. Labor Markets

Academic journal article Review of Social Economy

Structural Change in U.S. Labor Markets

Article excerpt

Two debates over labor market structure have occurred in the U.S. since World War II: the structuralist/aggregate demand school debate in the early 1960s and the deindustrialization debate of the 1980s. In both controversies one side offers a structural explanation in which difficulties in the labor market derive from an overall transformation of the economy. The opposing view is that labor market trends cannot be explained by a structural transformation, since no such transformation has taken place. Despite heated policy debates and impressive data collection the essential point of disagreement in both debates remains unresolved: the existence of structural change in the labor market.

This paper argues that two fundamental problems have hindered a satisfactory outcome to these recurring debates over structural change in the labor market. First, each side in the debate has a different concept of structure. If a consensus cannot be reached on what structural change in U.S. labor markets means, any debate over the existence and implications of this change is bound to be fruitless. With an imprecise or inconsistent definition of structure the question cannot be settled empirically. The second and more serious difficulty is that both sides use concepts of structure that have more to do with product markets than labor markets. The focus on product markets results from the fundamental assumption that labor demand is a derived demand. This assumption is axiomatic in neoclassical theory and exerts a powerful influence even on structuralist labor economists who reject much of the orthodox approach to labor.

The first part of this paper looks at the consequences of this approach to structure. Despite the fact that the debates were sparked by high unemployment and other disturbing trends in the labor market, the debates address labor market structure only indirectly. The theoretical arguments underlying the assorted views reflect a paradigmatic dispute over the nature of economic growth. The second part of the paper discusses the concept of labor market structure and suggests an alternative approach. The paper then returns briefly to the debates over structure to demonstrate how this alternative approach would clarify this controversy.

The Two Labor Market Debates

The initial impetus to both debates was a secular rise in average unemployment rates. The structuralist/aggregate demand school controversy resulted from an increase in the average unemployment rate to 5.8% between 1958 and 1964 (Frumkin, 1987).(1) Killingsworth (1965) argued that the trend toward higher unemployment rates was due to a structural transformation of the labor market. The disproportionate incidence of high rates of unemployment in specific segments of the labor market led to the Killingsworth twist, a forerunner of the declining middle of the 1980s. Killingsworth's conclusions were based on the observation of lower unemployment rates among more highly skilled and educated workers. However, despite the use of statistics on labor supply to support his argument, Killingsworth was arguing that it was the structure of labor demand that was undergoing transformation.

Killingsworth's structuralist arguments were countered by the Council of Economic Advisors (Bowen, 1965). The Council's contention was that although specific sub-groups of the labor force were affected differently over the business cycle, there had been no secular increase in the level of structural unemployment nor had there been significant increases in the unemployment rates of groups considered to be dominated by unskilled workers. The higher unemployment rates after 1957 were attributed to a deficiency of aggregate demand.

During the 1980s unemployment reached its highest level since the Depression. Many manufacturing industries were particularly hard hit by the recessions of the early 1980s leading to widespread layoffs and plant closings. Economists such as Robert Reich (1983), Barry Bluestone and Bennett Harrison (1982) argued that high unemployment rates and the loss of manufacturing jobs reflected long-term structural changes in U.S. labor markets. These economists pointed to the continued persistence of high unemployment.(2) In the first half of the 1980s the average level of unemployment rose to 7%, and in late 1982 unemployment surpassed 10% for the first time since World War II. Reich and others also stressed the decline of manufacturing that seemed to be producing a polarization of the labor market dubbed the declining middle by Robert Kuttner (1983). Trends in earnings were also less than encouraging to the average worker. Real wages grew at an average annual rate of 2.3% between 1951 and 1961 and averaged a 2.6% per year rise between 1961 and 1971. In contrast, weekly earnings dropped 13% between 1972 and 1980. Real wages had fallen to 1967 levels by 1982 (Bluestone and Harrison, 1988).

The anti-structural argument formulated principally by Robert Lawrence (1983a and 1983b) and Robert Klein (1983) echoed the views of the earlier aggregate demand school: "These labor force developments are significantly affecting present unemployment rates and their expected future movements, but they are more in the nature of cyclical swings than structural changes" (Klein 1983, p. 3).

The apparently alarming trends in the U.S. labor market during the 1970s and 1980s were attributed to one-time exogenous influences such as demographic changes, the oil price hikes, increased competition from newly industrializing countries, and inappropriate economic policies. The significant job losses of the manufacturing sector in the early 1980s were explained by the higher sensitivity of manufacturing to downturns in GNP. As a result of these destabilizing, but transitory factors there was a misallocation of resources in the short-run. While it is true that most of these factors can no longer be considered "short-run" or transitory, the implication was that once these factors either disappeared or were adjusted to, the U.S. manufacturing sector would return to its former strength. Temporary price effects required a reallocation of resources between industries. Given sufficient mobility of resources, the presumption was that incorrect prices would adjust and there would be a return to equilibrium.

The wide assortment of views in both the structural and anti-structural camps are rather easily classified once it is realized that the disagreement is not about labor market structure but about the nature of economic growth. Deindustrialization, the shift to services, increasing international competition, and movements in TABULAR DATA OMITTED aggregate demand over the business cycle all refer to some aspect of the market for output. The two "great labor market debates" have, in fact, been debates about product markets. The opinion of the Aggregate Demand School in the 1960s' debate was that there had been no secular shifts in the structure of the labor market. This view reflects a presumption of balanced economic growth. In the 1980s' debate, the dispute is over the nature of unbalanced growth. The point in contention was not whether economic growth was balanced or unbalanced but whether the uneven growth would lead to eventual convergence or divergence.

Balanced Growth.

In the structuralist-aggregate demand controversy of the early 1960s the Council of Economic Advisors attributed higher unemployment rates after 1957 to a deficiency of aggregate demand. Lawrence and others echoed this view early in the deindustrialization debate, when emphasis was placed on cyclical influences. The poor performance of the manufacturing sector of the U.S. economy was attributed to the greater cyclical sensitivity of manufacturing. The theoretical framework underlying a focus on the conventional business cycle is a textbook macroeconomic model. Perceived economic problems constitute deviations from equilibrium caused by exogenous forces.

The fact that the aggregate demand school and neoclassical economists in the deindustrialization debate did not find evidence of structural change using a standard macro model is not surprising. First, a short-run analytical framework is not suitable for a long-run phenomenon such as structural change. Second, the usual macro model is based on an aggregate production function for the economy. It is a "one-sector" model. If structural change is defined to mean shifts in the sectorial composition of the economy, the use of a one-sector model precludes an identification of such shifts. The choice of such a model implies that the economist has made an a priori assumption of balanced secular growth. Secular balance suggests that shifts due to cyclical expansions and contractions do not extend beyond the current business cycle.

Neoclassical economists in the 1980s recognized that trends such as the relative decline of manufacturing employment, lagging productivity, and higher structural unemployment had endured through several business cycles. The persistence of these trends was blamed on a series of exogenous shocks to the U.S. economy: oil price hikes; increasing exports from newly industrializing countries; the baby boom; and the labor force participation of married women. There are several implications of a preoccupation with "exogenous shocks." First, that these trends did actually occur can hardly be disputed; however, the analysis of their impact outside of any analytical framework precludes a finding of structural change. Second, the question is not whether these factors are exogenous or endogenous, but whether they are temporary or permanent. The presumption by Lawrence and Klein seemed to be that exogenous influences are temporary and do not reflect structural change. Thus, the identification of labor market trends as due to forces that are exogenous to the analytical model merely begs the question of structural change in the labor market.

The Wharton econometric model uses a definition of balanced growth as one in which real economic growth and real interest rates are equal with a budget deficit approaching zero. Evidence of imbalance in the economy includes internal and external deficits and inflationary pressure (Klein, 1983). That the U.S. economy has shown one or more of these imbalances since the early 1970s is apparent to everyone. The relative decline of manufacturing sector employment has been observed since World War II. Moreover, many of the exogenous forces cited by Lawrence reflected permanent structural changes to the economy. For example, it is now apparent that there has been both an increase in the level and a shift in the composition of structural unemployment (Frumkin, 1987). Klein cites deregulation of the financial sector and changes in tax laws as other major structural changes in the economy. The existence of structural change in the U.S. economy is no longer in dispute.

Convergent Unbalanced Growth.

In contrast to the aggregate demand school approach, most neoclassical economists in the deindustrialization debate admitted the possibility of uneven growth by the use of "two-sector" models. Ordinarily two-sector, growth models are either demand driven or supply driven. On the demand side an "exogenous" change in real income affects product demand unevenly due to Engel effects. Uneven growth could also occur with a decline in demand for the products of one sector due to changing tastes. On the supply side are models dealing with differences in capital intensities between industries. In neoclassical theory the capital/labor ratio will readily change in response to changes in relative input costs or factor supply. A shift in international comparative advantage may also cause industries to show a decline.

Analyses using the types of two-sector model described above are not difficult to find. For example, the possibility of a shift in international comparative advantage was investigated by Ira Magaziner and Robert Reich (1982) among others. Competitiveness problems are frequently blamed on poor management by American business and inappropriate policies. The impact of relative price changes or a shift in the terms of trade due to an overvalued dollar is a recurring theme (Artus, 1981; Klein, 1983; Lawrence, 1983a, 1983b). Changes in input costs due to the oil price increases of 1973 and 1979 have also been cited as sources of the uneven growth of various industries. Finally, the impact of shifts in demand was emphasized by Lawrence (1983b) who claimed that rising income levels have led to a decline in the share of manufacturing products in total consumer spending.

Relative shifts between sectors are an inevitable result of unbalanced growth, but do not indicate fundamental structural change. An emphasis on the distinction between relative and absolute shifts led Lawrence to conclude that the U.S. was ". . . virtually no more a service economy in 1980 than it was in 1960" (Lawrence, 1983a, p. 34). Relative shifts were presumed to lead to eventual convergence, and a presumption of convergent economic growth rules out fundamental structural change. During the 1970s neoclassical models were devised that did address the possibility of an absolute decline in the manufacturing sector. The booming-export or "Dutch disease" models depicted a situation in which an exogenous shock of sufficient magnitude destabilizes the economy (Cordon and Neary, 1983; and Long, 1983). The expanding sector draws resources away from the traditional export sector resulting in a fall in output. Since this situation implies labor and resource shortages in the manufacturing sector, they were inappropriate to U.S. economic conditions in the early 1980s.

Divergent Unbalanced Growth.

In neoclassical "two-sector" models the primary cause of unbalanced growth is the lack of mobility between two sectors that are fundamentally alike. The two sectors may represent a manufacturing industry and a service industry or two different manufacturing industries. The only meaningful difference between the two sectors in the model is the capital/labor ratio. In contrast, a "dual-sector" model suggests that there are inherent differences between the two sectors. These differences stem from institutional, behavioral, and technological factors not captured by measures of capital intensities in production. The shift from one sector to another involves more than a reallocation of labor. It involves broader sociological, political, and economic changes. The structuralists' argument is that the shift from manufacturing to services involves these broader, more fundamental, changes in the structure of the economy. In mainstream economic theory relative shifts are considered inevitable in a growing economy due to price fluctuations but are assumed to culminate in convergence. In contrast, Killingsworth and the later structuralists saw the problems in the labor market not as temporary, and possibly reversible, imbalances, but as fundamental changes in the way U.S. labor markets work.

The structuralists' argument is that there is uneven development in a framework of industrial dualism. Colin Clark (1940) pioneered the division of an economy into the agricultural, manufacturing, and service sectors. The Clark hypothesis is that the dominant contribution to GNP evolves from the agricultural sector to the industrial sector and then to the service sector during the process of economic growth. Although this shift in sectorial balance is founded more in empirical observation than formal analytics, it is nonetheless implicit in many models of economic development. It is also a common way of describing long-term trends in the U.S. economy. The proponents of the deindustrialization hypothesis departed from the economic development view in that current trends derived more from changes in the institutional structure of the economy than from long term economic growth. Reich, Bluestone and Harrison emphasized the changes in the legal and tax environment that encouraged "paper entrepreneurialism" by corporations. Despite differences in opinions over the causes and implications of the uneven growth Killingsworth (1965), Averitt (1987), and Bluestone and Harrison (1982 and 1988) all use this dual-sector approach.

Both the 1960s' Killingsworth twist and the 1980s' declining middle are based on the notion that the shift from manufacturing to services along with the expansion of process control technology is permanently changing the pattern of jobs available in the U.S. economy. A high proportion of manufacturing jobs are in the so-called primary labor market that includes jobs with above average wages and low turnover. In contrast the secondary labor market is characterized by low-paid, unskilled or general skill based jobs with high turnover. While the division into the primary and secondary labor market does not parallel the division between the manufacturing and the service sector, there was an underlying presumption among proponents of the deindustrialization thesis that the shift to services would lead to a decline in the size of the primary market and an expansion in the size of the secondary labor market.(3)

Derived Demand vs. Derived Structure

The emphasis on the composition and level of output as the appropriate way to approach problems in the labor market is based on a familiar axiom in microeconomic theory: demand for inputs is derived from the demand for outputs. If labor is assumed to be homogeneous, it is a simple matter to determine labor demand by finding the profit maximizing level of output. The crucial problem is that both neoclassical and structuralist approaches implicitly assume that labor market structure is also derivative. Apparently it is assumed that it is only a small leap from derived demand to derived structure.

Conventional microeconomic theory applies to levels of demand. Due to the standard assumption of homogeneous labor, the composition of labor demand or labor market structure has no meaning within the analytical framework. Some analysts depart from the assumption of homogeneous labor by specifying two types of labor input. The two categories are usually labeled, "skilled" and "unskilled," where skilled means expensive and unskilled means cheap (Nelson and Norman, 1977). But the only difference between the two groups is cost or the equilibrium wage levels that are determined in two separate labor markets. The central focus remains the demand for levels of categories of labor input derived from cost minimization of the given production function. Changes in labor demand are the direct result of changes in product demand that, in turn, depend on aggregate economic growth.

The Dual Labor Market (DLM) hypothesis is based on a more complex view of labor market structure than the neoclassical approach of labor categories based on cost. However, dualistic theories follow the neoclassical approach and treat the labor market as largely derivative, although other factors are also taken into account. Labor market characteristics are explicitly linked to the product market in the core/periphery analysis of Michael Piore (1975) and, to a certain extent, David Gordon, Richard Edwards, and Michael Reich (1982). The emphasis is stability rather than the absolute level of product demand. Primary sector jobs are said to occur in those firms controlling the stable portion of product demand, the so-called "core" firms. Secondary sector jobs reflect the instability of that portion of product demand allocated to peripheral firms.

In the DLM hypothesis, structure refers to the primary and secondary job markets. The characteristics of jobs in the primary and secondary markets have been extensively described and are generally agreed upon. To assert that the lower tier primary market is shrinking conveys an ambiguous message regarding qualitative changes in the labor market. Structure in regard to economic sector or industrial structure commonly refers to the primary, intermediate, and final sectors or to the agricultural, manufacturing, and service sectors. What occurred in the deindustrialization debate was that the DLM labor market categories were grafted onto growth theory resulting in sort of a hybrid theory. An assertion of a shrinking manufacturing sector was intended to imply the same qualitative changes in the labor market as an assertion of a shrinking lower tier primary market.

The Structure of Labor Demand

The two labor market debates were stimulated by a need to explain the changing structure of labor demand. But, what is the structure of labor demand? In the absence of an unambiguous and generally accepted definition each side filled the void with a proxy. Both sides then proceeded to explain their proxies. Neo-classical theory has no concept for labor demand structure. In its place, mainstream economists substituted the structure of product demand.(4) The DLM hypothesis does provide a fully developed concept of labor demand structure. There is a close link between structure of labor demand and industrial structure, but in the 1980s' debate these two structures became virtually indistinguishable. Thus, even in the structuralist camp the structure of output markets became the central focus. The phenomenon to be explained was the share of manufacturing in GNP.

The important question is, of course, How good are the proxies? Do shifts in economic sectors provide a useful explanation of current trends? In many cases the answer is, yes. Jobs in manufacturing have historically paid higher wages than jobs in services. Many workers displaced from manufacturing jobs in the 1980s have found lower paying jobs in services. Unemployment rates in the 1980s declined along with the recovery of manufacturing. Shifts away from traditional manufacturing industries in favor of high tech have affected the structure of labor demand. Workers displaced from clothing mills in the South are not likely to find new jobs in the computer software industry. However, the links between these changes in the labor market and shifts in economic sector are tenuous and subject to continuing debate. The notion that an analysis of product market structure will tell us all we need to know about labor market structure requires three analytical steps: 1) a definition of product market structure; 2) a definition of labor market structure; and 3) the link between them. These preliminary steps are missing on both sides of the two debates. One remedy would be to try to provide the missing steps: steps two and three for the anti-structuralists or step three for the structuralist camp. An alternative solution is to make the structure of labor demand the central focus and abandon the proxy, product market approach. The following section is intended as a step toward this alternative solution.

Structured Labor Markets

This paper has argued that two problems hindered a satisfactory outcome to these recurring debates over structural change. First, the definition of structure when used in reference to the labor market is ambiguous. Second, both sides use concepts of structure that have more to do with product markets than labor markets. As a result the debates address the labor market only indirectly. The intent of this section is to explore these two related problems. A productive dialogue on the changing U.S. labor market requires a greater recognition of the separate role of structure.

The importance of structure in the functioning of labor markets is perhaps the major contribution of the labor economists of the 1940s and 1950s. In this literature there are two essential characteristics of U.S. labor markets: 1) structure; and 2) segmentation. These two characteristics while related are not equivalent. To say that labor markets are structured means that they do not operate solely on price signals as in the labor market of orthodox theory. Segmented means essentially divided. For example, labor markets in the steel and automobile industries may be described by their wage patterns, occupational categories, customary work rules and so on. These markets have a particular structure. The idea of segmentation means that there are barriers to mobility between one industry and another.

Barriers to mobility may exist independently of structured markets. For example, in orthodox theory movement of workers is slowed principally by information difficulties. Workers must find out about jobs, and employers must find out about workers. Markets may be seen as segmented but not structured in the sense of operating differently. The structuralist view is that markets are also structured. Jobs are associated into clusters, workers move up along identifiable internal job ladders, and all workers in a firm find themselves in structured markets whether they are production workers, managers, or staff employees (Dunlop, 1957; Kerr, 1957). An unemployed auto worker in Detroit may never become a nurse in Dallas but not because of lack of information about a shortage for nurses in Dallas.

The DLM hypothesis and the later literature primarily investigated segmentation between a structured and structureless market. Markets possessing a distinct structure were grouped into a "primary" market, and a substantial part of the economy was in a structureless or "secondary" market. Lloyd Fisher (1951) had described the "structureless" market as one composed of temporary, manual labor paid on a piece rate basis and using little or no capital equipment. With Doeringer and Piore (1971) the secondary market included any unstable, unskilled work, not just piece rate work and not necessarily lacking in capital equipment. In fact, in the 1960s the large secondary market was considered pathological and received by far the most attention. Because only the segmentation between the structured and structureless markets was noteworthy, in the later literature the existence or absence of structure became effectively synonymous with segmentation.

In looking primarily at the distinction between the structured market and the structureless market, the emphasis was on the characteristics of the structured market. The first research task appeared to be an explanation of the origin of this market. The structured primary market was linked to the "core" economy, to the manufacturing sector, or to a social structure of accumulation, and the central focus shifted away from labor markets to product markets.(5) On the other hand, if we begin with the idea that all labor markets possess a distinct structure including the so-called secondary market, the research agenda is very different. If all markets are structured, there is not one structured market but many. The question then is not whether or not a labor market is structured, but what type of structure it has. The focus remains the labor market, since the first requirement is to identify and understand different structures.

The Elements of Structure

While placing structure at the center of the analysis is an important step, a second objective is to view structure as a multifaceted phenomenon. Labor economists have described structure by focusing on one of three principal elements: 1) the characteristics of the product market; 2) the technology of the production process; and 3) the organizational rules and procedures. Orthodox, structural and industrial relations economists alike recognize the importance of these three elements. What is recognized less often is that all three aspects must be considered. Job structure cannot be predicted from technology alone, from market characteristics alone or solely from the institutions governing the allocation of work within the organization.

The debates discussed in the first section of this paper demonstrate the prevalence of a product market approach to labor market structure. Product markets by themselves do not adequately predict structure. The same product market characteristics can lead to several different structures depending on the technology and institutional environment. Compare, for example, job structures in the textile industry with those in the apparel industry. Distinctly different labor market structures have also been observed in the same industry or even in the same firm (Maurice, Sellier, Silvestre, 1984; Kanter, 1984; Massey and Meegan, 1982).

Similar difficulties arise with an overemphasis on technology. The same technology can produce a variety of structures. For example, narrowly defined, rigid job classifications and authority structures are held to be incompatible with computer process control (Cyert and Mowery, 1987; Helfgott, 1988; Levin, 1987). On the other hand, computerized process control has been introduced into hierarchical organizations without substantial reorganization (Baran, 1985). This is the essence of Harley Shaiken's (1983) argument that computerized automation is used primarily to increase control and monitoring of workers. Clearly, technology alone is also insufficient to explain structure.

Structures are also commonly described by a list of rules and procedures. For example, Paul Osterman emphasizes the variability of structures by expanding the concept of the internal labor market:

Although the term "internal labor market" is sometimes used to denote a particular pattern -- the seniority-based closed job ladder system that characterizes much of America's heavy industry -- it is more useful to recognize that all firms have rules and procedures, but they may vary in important ways. The high turnover and dead-end clerical job is organized by a set of rules, just as are more stable occupations, but the rules and procedures differ (1988, p. 62).

Osterman describes two structures, the "salaried model" and the "industrial model."(6) But a listing of rules and procedures does not provide enough flexibility for real world structures. For example, where would we classify a temporary office worker at a large corporation? According to the criteria regarding stability of employment, method of hiring, and benefits, it is apparent that this worker could not be placed in either Osterman's "salaried" or "industrial" category. Temporary office work must operate according to some third model. But, placing all three models in the same corporation means they are 1) subject to the same product market characteristics; 2) influenced by the same product and process technology; and 3) subject to the same internal institutional environment. All three models are related or part of some larger structure.

Attempting to understand labor market structure by looking at one element and neglecting the remaining elements contains numerous pitfalls. Disputes in labor economics are often the result of a narrow focus on only one aspect of job structure without explicit assumptions regarding the remaining aspects. For example, the deskilling debate is ostensibly an argument over the impact of technology on job structure, but the real dispute is over the institutional aspects of job structure. A discrete structure emerges only if the interdependence of all three elements is recognized.

The Labor Market Debates Revisited

The importance of conceiving of structure as an integrated whole can be demonstrated by returning to the labor market debates. In the 1960s' debate Killingsworth argued that the pattern of labor demand was undergoing a "twist" in which demand for highly skilled workers was rising, and the demand for unskilled workers was falling. The notion of the Killingsworth twist is that both groups of workers were affected as part of a single structural change. Failure to conceive of the Killingsworth twist as a single structure led to disputes over pieces of the puzzle such as unemployment among unskilled workers, automation and the need for job training. Indeed much of the literature following the 1960s' debate dealt with these related issues. While certainly significant these issues do not address the central question of structure raised by the Killingsworth twist. As a result the 1960s' debate remained unresolved.

The declining middle of the 1980s' debate is a distinct structure somewhat similar to the Killingsworth twist. The meaning of the declining middle hypothesis is in the total configuration of jobs and the relative size of each category. The declining middle refers to a particular array of jobs or occupations: managerial, professional, and technical are at the top of the hierarchy; semiskilled or blue collar workers are in the middle; and unskilled workers at the bottom. What is critical is the overall pattern.

In contrast to a coherent structure such as the declining middle, one of the principal characteristics of the DLM hypothesis is the separateness of each market segment(7). Each market segment of the DLM is self-contained and may occur not just in different turns or industries but in different economic sectors. In fact, the existence of both a lower tier primary structure and a secondary type market within the same firm has been considered an anomaly (Edwards, 1979; Bluestone and Stevenson, 1981).

The failure to grasp structure as an integrated whole occurred in the 1980s' debate, when the declining middle was equated to a shrinking lower tier primary market.(8) This is a logical implication of the DLM hypothesis. Manufacturing sector jobs are located in the lower tier primary market. Thus, a declining manufacturing sector implies a shrinking lower tier primary market. However, equating the declining middle to a declining lower tier primary led to the awkward task of trying to explain the decline of something when no satisfactory explanation exists for its rise. If the lower tier primary disappears, what remains? Does this leave the upper tier primary and the secondary markets intact? Are industrial workers in a "structureless" market, while corresponding managerial and administrative staffs operate in a structured market? In some industries, production jobs have been structured into several tiers: employees with seniority who retain relatively high pay and benefits, and new employees who perform the same work at a lower rate of pay (Walsh, 1988). In other industries, a core of permanent, full-time workers is supplemented with a ring of part-time, temporary, or sub-contracted employees who receive low pay and no benefits (Appelbaum, 1984). A disappearing lower tier primary will not necessarily leave a "hole" or a "missing middle," but may result in new categories altogether.

The Declining Middle Structure

An entirely different understanding of the declining middle emerges when it is viewed as an integrated structure. A structure is composed of all three elements: product market characteristics; the technology of the production process; and the organizational rules and procedures. It is the combination of elements rather than any single element that defines structure. Interpreted as a complete structure the declining middle becomes a substantive hypothesis. A debate in the 1980s over the prevalence of this specific structure would have had a greater change of resolution.

Organizationally, a declining middle means fewer intermediate jobs. Concern over declining manufacturing industries is not only due to the numbers of lost jobs, but also due to the loss of opportunities. There are literally millions of workers who began at low level entry positions and rose to higher paying jobs though promotion along internal job ladders. In fact, the prevalence of job structures with well-defined job ladders in manufacturing industries generated what Eileen Appelbaum (1986) calls "working class jobs paying middle class wages." A declining middle job structure implies that the middle rungs of the internal job ladder have been removed. Barbara Baran describes this situation in her extensive case study of the insurance industry:

|A~lthough skill levels may be rising, opportunities for occupational mobility seem to be diminishing. As more and more professional functions are turned over to clerks, lower level professional categories which provided clericals some possibility of advancement to the career track are being eliminated. Because the remaining professional jobs . . . are highly skilled, there is in the words

of one manager increasingly a "quantum leap" between the computer-linked clerical positions and the next step up the occupational ladder (1985, pp. 137-8).

The lack of internal advancement opportunities is a key characteristic of a declining middle structure. The ease with which employees can obtain the requisite skills to move to a new rung of the job ladder is a function of the type of training required. For example, the traditional industrial model of an internal labor market allows movement up from the entry port position through a combination of informal on-the-job training and acquired experience (Osterman, 1988). Little or no formal training may be necessary to move up the job ladder. In fact, appropriate training may not be available except through tenure on the job. In this case, one cannot move up the job ladder without experience. A declining middle implies the opposite situation. Appropriate training may not be available except through formal schooling. In this case, one cannot move up the job ladder without education or formal training.

A shift in training from on-the-job and internal training to external training and formal education implies a shift from specific to general skills. Process control technologies have led to an increase in general training requirements in some industries, and these requirements are often significant. A college degree is a necessary qualification for professional and technical workers. Managerial positions also may be upgraded through a reduction of low level managerial and supervisory positions.(9) There is, however, considerable variation in the impact of process control technology on job structures. A gap between educated professionals and specifically skilled workers is often less a consequence of the technical difficulty of new technologies than organizational choices. For example, Helfgott (1988) found a wide variation in the location of programming of computer controlled machines. In some cases, programming is assigned exclusively to engineers and programmers. In other cases, the function is divided between production workers and professional programmers.

Firms have adopted process control in response to highly competitive and unstable market conditions (Markusen, Hall, and Glasmeier, 1986; Massey and Meegan, 1982). Instability may be due either to rapid growth or fears of decline. Whether process control is adopted in response to rapid growth or to forestall decline affects the direction of organizational change. On the one hand, a large, established corporation primarily concerned with retaining market share in an environment of increasing competition would be more likely to retain bureaucratic, hierarchical organizational structures with rigidly defined jobs and responsibilities. Job enlargement in this case could very well truncate career paths and remove the middle rungs of the ladder. At the other extreme, a corporation less wedded to traditional forms of work organization, less fearful of delegating responsibility, and more willing to adopt looser organizational boundaries could avoid a declining middle structure.


The two labor market debates raised serious questions about the future of the U.S. labor market. Unfortunately, these questions could not be answered by a dispute over the nature of economic growth. The link between economic growth, product demand, and the derived demand for labor is only a part of the story. Once labor market analysis is broadened to include a wage structure, internal labor markets, segmented external labor markets, and jobs differentiated according to stability, tenure, and the level and acquisition of skills, it is no longer appropriate to have the labor market implied. While it is safe to assert that economic growth will have an impact on labor market structure, it is not immediately obvious what the nature or significance of this impact will be.

The notion that the labor market can be derived completely from the product market precludes any separate role for technology or the institutional environment. To use product market structure as a proxy for labor market structure has the same effect of ignoring technology and institutions. Since the Killingsworth twist and the declining middle deal principally with the technological and institutional dimensions of structure, the product market approach effectively misses the point.

A clarification of the concept of structure is essential to more sensible debates about the U.S. labor market. If we are to come up with appropriate and useful policy options we must grapple with the dimensions of real world structures. The purpose of this paper is to demonstrate that an approach based on a multidimensional view of structure yields very different conclusions. If the cause of a declining middle is a weak manufacturing sector or simply a shift to services, then there are three policy options: 1) do nothing since a shift to services is "inevitable"; 2) use standard macro policy to stimulate growth; or 3) institute an industrial policy for declining manufacturing industries. Questions of political feasibility aside, it is not clear that any of these options would be an appropriate response to declining middle structures. Investigations into the declining middle have shown that it is not necessarily linked to downturns in the economy. On the contrary, the declining middle structure may be associated with a moderate growth industry.

1 The unemployment rate exceeded 5% in only 9 years between 1947 and 1969. The "full employment" level of unemployment during the 1950s and 1960s was between 4% and 5%.

2 Unemployment was in "full employment" range in only one year during the 1970s and averaged 6.5% between 1975 and 1979.

3 A presumption that does not, in fact, appear to be correct. See Noyelle (1987) for investigations of internal labor markets in the service sector.

4 Neoclassical theory does have a concept for structure of labor supply. In many cases, labor supply also serves as a proxy for labor demand. For example, economists use data on labor supply to analyze the impact of new technologies on jobs.

5 Gordon, Edwards, and Reich (1980) link the various segments of the labor market, the upper tier primary, lower tier primary and secondary to types of control originating in social structures of accumulation.

6 These two models are essentially the upper and lower tier of the DLM primary market. In the 1985 edition of Internal Labor Markets and Manpower Analysis, Doeringer and Piore state that they also view internal labor markets as more variable and widespread than in the 1971 edition of their book.

7 The primary distinction in Piore and Doeringer (1985) is between the primary and secondary markets. Gordon, Edwards, and Reich (1982) see the upper tier primary occurring independently of the lower tier primary.

8 Another view of the declining middle hypothesis is that it means a polarization in the distribution of income (Rosenthal, 1985). However, this interpretation fails to understand the declining middle as an hypothesis about labor markets.

9 There are numerous case studies of process control. See Flynn (1988), Baran (1985), Menzies (1981), Hirschhorn (1984).


Appelbaum, Eileen. "High Tech and the Structural Employment Problems of the 1980s," in Eileen Collins and L.D. Tanner (eds.), American Jobs and the Changing Industrial Base, Cambridge, MA: Ballinger, 1984.

-----. "Restructuring Work: Temporary, Part-Time, and At-Home Employment," in Heidi I. Hartmann, Robert E. Kraut, and Louise A. Tilly, (eds.), Computer Chips and Paper Clips, vol. I. Washington, D. C.: National Academy Press, 1986.

Artus, Jacques. "Persistent Surpluses and Deficits on Current Account Among Major Industrial Countries," in J. R. Sargent (ed.), Europe and the Dollar in the World-Wide Disequilibrium, Rockville, MD: Sijthoff and Nordhoff, 1981.

Averitt, Robert T. "The Dual Economy Twenty Years Later," Journal of Economic Issues, 21 (June 1987), pp. 795-802.

Baran, Barbara. "The Technological Transformation of White Collar Work: A Case Study of the Insurance Industry," Working Paper #9, Berkeley: Berkeley Round Table on the International Economy, University of California, January 1985.

Bluestone, Barry and Bennett Harrison. The Deindustrialization of America. New York: Basic Books, 1982.

-----. The Great U-Turn: Corporate Restructuring and the Polarizing of America. New York: Basic Books, 1988.

Bluestone, Barry and Mary Hugg Stevenson. "Industrial Transformation and the Evolution of Dual Labour Markets: The Case of Retail Trade in the United States," in Frank Wilkinson, (ed.), The Dynamics of Labour Market Segmentation. New York: Academic Press, Inc., 1981.

Bowen, William G., ed., Labor and the National Economy. New York: W.W. Norton, 1965.

Clark, Colin. The Conditions of Economic Progress. London: Macmillan, 1940.

Cordon, W. and S. Neary. "Booming Sector and De-Industrialization in a Small Open Economy," Economic Journal, December 1982.

Cyert, Richard M. and David C. Mowery (eds.), Technology and Employment: Innovation and Growth in the U.S. Economy. Washington, D.C.: National Academy Press, 1987.

Doeringer, Peter B. and Michael J. Piore. Internal Labor Markets and Manpower Analysis. Armonk, NY: M.E. Sharpe, Inc., 1985.

Dunlop, John T. "The Task of Contemporary Wage Theory," in George W. Taylor and Frank C. Pierson (eds.), New Concepts in Wage Determination. New York: McGraw-Hill, 1957.

Edwards, Richard. Contested Terrain. New York: Basic Books, Inc., 1979.

Fisher, Lloyd H. "The Harvest Labor Market in California," Quarterly Journal of Economics, November 1951.

Frumkin, Norman. Tracking America's Economy. Armonk, NY: M.E. Sharpe, 1987.

Markusen, A., Peter Hall, and A. Glasmeier. High-Tech America. Boston, MA: Allen and Unwin, 1986.

Gordon, David, Richard Edwards, and Michael Reich. Segmented Work, Divided Workers. Cambridge: Cambridge University Press, 1982.

Helfgott, Roy B. Computerized Manufacturing and Human Resources. Lexington, Massachusetts: D.C. Heath, 1988.

Kanter, Rosabeth Moss. "Variations in Managerial Career Structures in High-Technology Firms: The Impact of Organizational Characteristics on Internal Labor Market Patterns, in Paul Osterman (ed.), Internal Labor Markets. Cambridge, MA: The MIT Press, 1984.

Kerr, Clark. "The Balkanization of Labor Markets," in E. Wight Bakke (ed.), Labor Mobility and Economic Opportunity. Cambridge, MA: MIT Press, 1954.

Killingsworth, Charles C. "Automation, Jobs, and Manpower: The Case for Structural Unemployment," in Garth L. Mangum (ed.), The Manpower Revolution. Garden City, New York: Doubleday, 1965.

-----. "The Fall and Rise of the Idea of Structural Unemployment," Proceedings of the Thirty-First Annual Meeting, Industrial Relations Research Association, Chicago, IL, 1978, pp. 1-23.

Klein, Lawrence R. "Identifying the Effects of Structural Change," in Industrial Change and Public Policy. Federal Reserve Bank of Kansas City, 1983.

Kuttner, Bob. "The Declining Middle," Atlantic Monthly, July 1983.

Lawrence, Robert Z. "Changes in U.S. Industrial Structure: The Role of Global Forces, Secular Trends and Transitory Cycles," in Industrial Change and Public Policy. Federal Reserve Bank of Kansas City, 1983b.

-----. "The Myth of U.S. DeIndustrialization," Challenge, July/August 1983a.

-----. Can America Compete? Washington, D.C.: The Brookings Institution, 1984.

Levin, Henry M. "improving Productivity through Education and Technology," in Burke, Gerald and Russell W. Rumberger. The Future Impact of Technology on Work and Education. Philadelphia, PA: The Falmer Press, 1987.

Long, N.V. "The Effects of a Booming Export Industry on the Rest of the Economy," The Economic Record, March 1983.

Magaziner, Ira C. and Richard B. Reich. Minding America's Business. New York: Vintage Books, 1982.

Massey, Doreen B. and Richard Meegan. The Anatomy of Job Loss. New York: Meuthen, Inc., 1982.

Maurice, Marc, Francois Sellier, and Jean-Jacques Silvestre. "The Search for a Societal Effect in the Production of Company Hierarchy: A Comparison of France and Germany," in Paul Osterman, (ed.), Internal Labor Markets, Cambridge, MA: The MIT Press, 1984.

Nelson, Richard R. and Victor D. Norman. "Technological Change and Factor Mix over the Product Cycle," Journal of Development Economics, 4 (March 1977), pp. 3-24.

Noyelle, Thierry. Beyond Industrial Dualism: Market and Job Segmentation in the New Economy. Boulder, CO: Westview Press, 1987.

Osterman, Paul, ed. Internal Labor Markets. Cambridge, MA: The MIT Press, 1984.

Osterman, Paul. Employment Futures: Reorganization, Dislocation and Public Policy. New York: Oxford University Press, 1988.

Piore, Michael J. "Notes for a Theory of Labor Market Stratification," in Richard C. Edwards (ed.), Labor Market Segmentation. Lexington, MA: D.C. Heath, 1975.

Reich, Robert B. The Next American Frontier. New York: New York Times Books, 1983.

Shaiken, Harley. Work Transformed. New York: Holt, Rinehart and Winston, 1985.

Walsh, David J., "Accounting for the Proliferation of Two-tier Wage Settlements in the U.S. Airline Industry, 1983-1986," Industrial and Labor Relations Review, October 1988, 42(1), pp. 50-62.

Figure 2


Managerial, professional, and technical

* high pay

* formal training

* permanent, full time

* job security

Skilled, paraprofessional

* relatively high pay

* on-the-job training

* permanent, full time

* relative job security

Unskilled or semi-skilled

* low pay

* temporary, part time

* low job security for permanent or full-time workers

Figure 3


Upper Tier Primary Market

Managerial and Professional jobs

* high pay

* career advancement

* high turnover

* informal work rules

Lower Tier Primary

Blue Collar Manufacturing jobs

* relatively high wages

* internal job ladders

* low turnover

* formal work rules


Unskilled and "Other" jobs

* low pay

* no career advancement

* high turnover

* informal work rules

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.