Status and incentives.

Figure 1 shows earnings by age and education in the automobile and electrical industries in Japan and the United States; Figure 2 presents earnings profiles by age and education at the national level. In Japan, differentiation in earnings appears after age 35, and the earnings gap between different types of workers widens with age. (29) In contrast, in the United States, earnings increase (sharply for educated workers) in junior years but less so afterward, and the earnings gap between educated and less-educated workers widens up to age 35-39. As the industries here are fairly standardized, this probably does not reflect any differences in technology, but rather different management practices.

We have treated job mobility differences between these two countries as given, and argued that they could explain differences in compensation policy in a way that is consistent with our theoretical analysis. There could be any number of other underlying differences between the two economies that we do not control for which might jointly explain both mobility differences and differences in work compensation practices. Furthermore, our theoretical predictions are, in some respects, similar to those resulting from other theories which try to explain internal labor markets. In the next section, we place our contribution in this existing literature.

5. Related work on internal labor markets

* Our analysis provides a novel theoretical underpinning for understanding why promotions might be preferable to direct monetary incentives, and also predicts how individual earnings profiles over time are affected by the expected duration of the work relationship. Although these two issues are closely intertwined, they have been to a large extent considered separately in the existing literature.

[FIGURE 2 OMITTED]

The relationship between tenure and pay in internal labor markets has attracted a great deal of attention. The use of large prizes attributed only at specific times over a career is often interpreted as an attempt by firms to improve employee attachment (see, for instance, Becket, 1962; Salop and Salop, 1976; Lazear, 1979). Lazear (1979) argues that firms that want to invest in firm-specific human capital offer back-loaded compensation structures in order to retain their workers. In light of this theory, the different compensation policies in the United States and Japan might reflect greater investment in firm-specific human capital in Japanese than in American firms. The interpretation proposed here provides additional insights in two ways. First, the firm-specific human capital explanation establishes a causal link between commitment and delayed monetary rewards. As delayed pay raises are a means of fostering commitment by employees, any exogenous increase in this commitment will reduce the firm's incentive to delay rewards. The firm-specific human capital hypothesis is thus inconsistent with the data if, as is often argued, there are cultural reasons for the differences in job mobility for young workers in Japan and the United States. (30) Our analysis, on the contrary, assumes that commitment is exogenous but would also hold were commitment to be induced by the prospect of garnering future rewards. Second, empirical tests of the firm-specific human capital motive are at best inconclusive. Farber (1999) tries to explain the high returns to tenure with this theory, but only finds little support for it. (31) By way of contrast, our theory states that monetary rewards are delayed so as to match with the change in status resulting from promotion, as this is the most cost-effective way of providing incentives to young employees. We jointly explain the timing of monetary rewards and the use of promotions as an incentive tool. As we argue above, the coincidence of pay increases with promotions is a well-documented characteristic of Japanese internal labor markets.

The extensive use of promotions for incentive purposes has also been widely discussed in the literature. Direct monetary transfers allow for fine tuning of the incentive scheme, contrary to promotions which are discrete and irregular. One reason for the use of discrete incentive schemes is that it is not always possible to assess absolute performance, whereas relative performances are somewhat easier to evaluate. Promotions can then be viewed as prizes in a tournament between employees, as in Lazear and Rosen (1981). This leaves open the question of why, in practice, promotions involve both changes in status and pay raises, and why they are used so extensively (and not only when absolute performance is unobservable). The theory presented here provides a link between wage profiles, hierarchical structures, and tenure in firms.

Another way of linking the wage profile to the worker's position in the firm's hierarchy is to think of promotions as a way of screening employees. Gibbons and Waldman (1999) propose a model where there is no room for work incentives and workers' productivity is heterogeneous. Promotions are then used as a screening device to match more productive workers with tasks where performance is more sensitive to productivity. They argue that their setting explains many observed characteristics of compensation schemes, and notably the fact that pay raises are larger when they coincide with promotion. These pay raises reflect higher productivity and would not have occurred were the move up the hierarchy not have corresponded to a change in the individual's job. Yet, as Milgrom and Roberts (1992) note, some companies such as 3M or IBM have sought to avoid any conflict between the incentive and screening objectives by creating separate career ladders for scientists and engineers, so that they can be promoted without having to go into management. Similarly, faculty members in universities, or doctors in hospitals, are generally promoted without changing jobs. Furthermore, as Lazear (1991) points out, when promotions do involve an actual change in the employee's tasks, the associated wage increases are oftentimes out of proportion with any reasonable estimate of the rise in marginal productivity associated with a job higher up in the firm's hierarchy. In our setting, promotions involve no job changes, and the concentration of rewards toward the end of the career implies that those who are promoted are paid above their marginal productivity. (32)

One difficulty with using data on promotions to test this theory is that promotion systems and hierarchies must meet various functional goals such as production efficiency and screening, encouraging investment in firm-specific human capital as well as providing work incentives to employees by creating stimulating career paths within the firm. These potentially conflicting objectives lead to identification problems. It would therefore be extremely useful to appeal to different types of data to examine how status differentiation is used jointly with monetary rewards to provide work incentives. For instance, our results are consistent with the common practice of offering executives a variety of perks. Rajan and Wulf (2004) use a panel of 300 publicly traded U.S. firms, over the period 1986-1999, to see whether perks (i.e., executive jets, chauffeur-driven cars, and country club memberships) are managerial excesses, as generally argued in the corporate finance literature, a strategy to minimize income tax liability, or rather are designed to enhance managers' status or productivity. They find little empirical support for the tax explanation and, at best, mixed evidence for the private benefit explanation. However, they do find that pay and perks are positively correlated (even when controlling for firm size, industry, and year), and that larger, older, and more hierarchical organizations offer more perks. They also find that the more productive employees at the top of a firm's hierarchy tend to receive more perks. They conclude that perks might likely serve to enhance managers' status and firms' productivity. Oyer (2005), focusing on broader types of benefits, argues that benefits might be motivated by productive efficiency. For instance, company-provided meals or child-care services are found empirically to enhance employees' effort. He explains this result via a process of substitution between domestic tasks and work. Unfortunately, he does not consider status. Additional insights could likely be gained from the exploitation of large panels of firms and individuals such as that of the LEHD program at the U.S. Census Bureau described by Abowd, Haltiwanger, and Lane (2004). In particular, it would be interesting to consider how personnel management practices differ across firms characterized by different turnover rates, and thus different degrees of commitment. (33)

6. Conclusion