The number of workers who quit working before they reach the formal age of retirement is surprisingly high. In 2005 the average labor force participation rate in OECD countries of men in ages 55-64 was 65.5 percent, while the labor force participation rate for men in ages 25-54 was 92.1 percent. Furthermore, it is quite a recent phenomenon. Labor participation rates for US men in ages 55-64 dropped from 86.7 percent in 1948 to 62.6 percent in 1996, recovering slightly to 68.7 percent in 2004. Early retirement is usually attributed to bad health, wealth, and to incentives by generous retirement plans like social security. This paper offers an additional explanation to early retirement: erosion of human capital by technical progress.
Technical progress changes continuously the way we produce goods and services. It introduces new goods, new machines and equipment, and new production methods. Simultaneously it creates new professions and destroys old ones. New technologies frequently make some existing human capital obsolete, while creating demand for new types of human capital. This paper proposes that human capital erosion by technical progress also reduces labor of older workers. It affects older workers more than younger ones, since their career horizon is much shorter. Hence it is less beneficial for them, or for their employers, to invest in learning the new technologies. The paper models this idea theoretically, examines its implications and finds significant empirical support for it, using micro US data.
The theoretical model describes a growing economy with many sectors. Each sector uses a specific technology, which requires specific human capital. Individuals learn and acquire technology-specific professions when young and then work using these technologies. Then new innovations arrive, replacing existing technologies while requiring workers to learn how to use these new technologies, via retraining. While younger workers learn the new technologies, many older workers choose not to learn, and retire early. Thus, technical progress raises the probability of early retirement. We call this the erosion effect of technical progress. Note that while in the model the worker decides on early retirement, in real life the decision is often made by the employer, who prefers not to retrain an older worker. The worker becomes unemployed and usually retires after some futile search for a new job. The basic result is the same.
But the model also enables us to consider the result of the positive correlation between rates of technical progress across sectors, which is implied by the data. This creates an opposite effect of technical progress, which we call the wage effect. In a period of high technical progress in all or most sectors, wages rise in the economy. As a result older workers tend to work more and delay retirement. The model shows that as a result, the average economy-wide rate of technical progress has an ambiguous effect on early retirement, because the erosion effect and wage effect work in opposite directions, while the sector component of technical progress has an unambiguous positive effect on early retirement, since it reflects only the erosion effect.
The empirical cross-sector prediction of the model is examined by US data on the labor status of a sample of men over the age of 50 from the Health and Retirement Study (HRS), which also includes information on job histories. This information is merged with sector productivity growth data, measured by Jorgensen (2000), which represents technical progress. We test how working status is affected by the sector specific component of technical progress, which we measure by the sector’s rate of TFP growth minus the aggregate TFP growth. We find that the coefficient of the sector specific component of technical change on the probability of not working by older men is positive, while it is insignificant for younger workers. We also examine the possibility that this result is biased due to reverse causality, which might occur if sectors differ by lay-offs of older workers, and if the less productive workers are laid-off. We test an implication of this possibility and reject it. We also test the theory using an alternative measure of technical change, the ratio of equipment capital to labor, and the results are similar.
This paper is related to two different literatures, one on early retirement and the other on the effects of technical change on labor markets. The literature on early retirement has mostly focused on wealth, health and on greater financial incentives to retire. The first paper that suggested that technical progress and early retirement are strongly related through erosion of human capital is Bartel and Sicherman (1993), from here on BS. Our paper follows this insight, but differs significantly in identifying the wage effect. We show that by not differentiating between the erosion effect and the wage effect, the test in BS is not well specified and leads to some wrong interpretations. Our main contribution to this literature is therefore the general equilibrium theoretical model, which leads us to identify the wage effect. This theoretical model leads us to use variables that can better differentiate between the wage and the erosion effects.
The paper is also related to recent research on the effect of technical progress on the labor market. Some papers in the new growth literature, like Aghion and Howitt (1994), Helpman and Trajtenberg (1998), Hornstein and Krusell (1996), and Galor and Moav (2000), have claimed that technical progress might reduce employment due to costs of learning new technologies. This paper shows that this effect is stronger for older workers, whose career horizon is short.
