برآورد کشش عرضه نیروی کار فریش در ژاپن

Using Japanese data from the 1990s aggregated by prefecture, age group, and sex, we estimate Frisch labor supply elasticity, which has been seldom estimated in Japan. The change in labor supply can be decomposed into two labor-supply behaviors: extensive margin, indicating workers' entry and exit from the labor market; and intensive margin, indicating changes in hours of work in response to a wage change. Our estimates of the Frisch elasticity on the extensive and intensive margins combined are in the range of 0.2 to 0.7 for males, 1.3 to 1.5 for females, and 0.7 to 1.0 for both sexes. Our estimates of the Frisch elasticity on only the intensive margin are in the range of 0.1 to 0.2 for all three categories. These results suggest that extensive margin explains the bulk of labor-supply changes in Japan. As for the changes in the estimates of the Frisch elasticity in Japan from the 1990s, it has been either unchanged or in a declining trend on the extensive and intensive margins combined, either unchanged or in a slight rising trend on only the intensive margin, and in a declining trend on only the extensive margin. J. Japanese Int. Economies22 (4) (2008) 566–585.

Using Japanese data from the 1990s aggregated by prefecture, age group, and sex, we estimate Frisch labor supply elasticity (hereafter, Frisch elasticity) in Japan. As pointed out by Prescott (1986), Frisch elasticity is one of the most important structural parameters in economics. In the dynamic general equilibrium models used in macroeconomics, for example, the responses of endogenous variables to shocks largely depend on Frisch elasticity. To our knowledge, however, previous studies have not done much to estimate the elasticity in Japan. Given this situation, the main objective of this paper is to show parameter values of Frisch elasticity in Japan. Frisch elasticity is a type of intertemporal labor supply elasticity, which is derived from standard dynamic models that solve the intertemporal utility maximization problem of a representative agent. We estimate Frisch elasticity based on a life-cycle model, a line of analysis that stems from the permanent income hypothesis from Friedman, 1957 and Friedman, 1976. According to the life-cycle model, a representative agent changes his/her labor supply over the business cycle in response to the temporal wage changes induced by shocks (that is, to the deviation of actual wages from the permanent or expected wage). Frisch elasticity indicates the extent to which people change their labor supply in response to these temporary wage changes. There is a wealth of literature estimating Frisch elasticity using either aggregate or longitudinal data in the United States and European countries. Examples of such research include Lucas and Rapping (1969), Altonji (1982), Mankiw et al. (1985), and Algoskoufis (1987), which use aggregate data, and MaCurdy (1981), Heckman and MaCurdy (1982), Browning et al. (1985), and Altonji (1986), which use longitudinal data. As for Japan, however, there is a paucity of empirical research on Frisch Elasticity. Consequently, the parameter value derived from US or European research is typically used without modification when conducting simulations on dynamic general equilibrium models in Japan. Should there be substantial differences in labor market characteristics across countries, however, this may be reflected in different labor supply behavior in Japan. In this regard, it is worth estimating the elasticity for Japan using its own data. In addition, because of this lack of estimates for Frisch elasticity, we do not yet know whether there is an inconsistency between theory and empirical evidence1 in Japan as has been the case in the US and European literature. That is, when the parameter values of Frisch elasticity obtained in empirical analyses have been plugged into simulations of theoretical models, those simulations have usually done a poor job in describing the real economy. As stated above, there is no prior research specifically aimed at estimating Frisch elasticity in Japan. The only related research would be Osano and Inoue (1991), which uses Japanese time series data to estimate Euler equations for consumption and leisure. Although their main purpose is to examine the validity of life-cycle models and implicit contract theory, we can infer the Frisch elasticity from their estimation results—about 0.06 to 0.13. Our approach differs from that of Osano and Inoue (1991) in several respects. First, our primary objective is to provide the estimates for Frisch elasticity. To achieve this objective, we also estimate m-supply, Marshallian, and Hicksian labor supply elasticities. We do this because these other elasticities are robust to liquidity constraints, and at the same time they provide additional information regarding the size of Frisch elasticity. Since there is a theoretical relationship in size between Frisch and other elasticities, we can specify the upper and lower bound of the Frisch elasticity. Second, we estimate Frisch elasticity using not only time series variation but also cross-sectional variation with data on hours of work, wages, and other variables aggregated by prefecture, age group, and sex. Third, using aggregate data enables us to estimate labor-supply behavior on the extensive margin and intensive margins combined, while Osano and Inoue (1991) estimate Frisch elasticity only on intensive margin. 2 By extensive margin, we mean the marginal change in the number of people who enter into/exit from the labor market in response to temporary marginal wage changes. By intensive margin, we mean the marginal changes in hours of work for those already employed in response to temporary marginal wage changes. In the review of empirical studies for the United States, Heckman (1993) concludes that most of the change in the labor supply reflects changes in the number of people entering and exiting the labor market. That is, extensive margin explains most of the fluctuation in the labor supply. If this also applies to Japan, the parameter value for Frisch elasticity in macro models shall include not only intensive but also extensive margin to describe a representative agent's labor supply behavior. As is done in Lucas and Rapping (1969), using aggregate data enables us to estimate Frisch elasticity in both extensive and intensive combined. 3 Lastly, to examine the possibility that Japan's labor market underwent structural change during the so-called “lost decade” of the 1990s, we also examine whether or not Frisch elasticity changed during the 1990s. 4 The results of this paper can be summarized as follows. First, Frisch elasticity on the extensive and intensive margins combined is in the range of 0.7 to 1.0 when we pool data on males and females together (in our paper, we call this “both sexes”). We conclude that our estimates of Frisch elasticity are not very different from those suggested in the US empirical studies, such as Lucas and Rapping (1969), Algoskoufis (1987) and Mulligan (1999).5 Second, we find large differences in Frisch elasticity by sex, with males having elasticity in the range of 0.2 to 0.7, and females in the range of 1.3 to 1.5. Third, Frisch elasticity on the intensive margin only is very low for both sexes, males, and females, in the range of 0.1 to 0.2. The last finding is consistent with Heckman (1993), that extensive margin dominates intensive margin in most of the changes in labor supply behavior. Regarding possible changes in Frisch elasticity in the 1990s, it has been either unchanged or slightly declining on the extensive and intensive margins combined. When we decompose it into the two margins, however, intensive margin was either unchanged or slightly rising whereas the extensive margin was possibly declining. This paper is organized as follows. In Section 2, we derive our estimation models and discuss how we deal with the caveats of the models. In Section 3, we explain the data we use, and then in Section 4 we present estimation results. We end with conclusions in Section 5.

In this paper, using Japanese data from the 1990s aggregated by prefecture, age group, and sex, we estimate the plausible range of Frisch elasticity, one measure of intertemporal labor supply elasticity. Our estimation results are as follows. First, Frisch elasticity on the extensive and intensive margins combined is in the range of 0.7 to 1.0 when we pool data on males and females together (in our paper, we call this “both sexes”). We conclude that our estimates of Frisch elasticity are not very different from the estimates suggested in the US empirical studies, such as Lucas and Rapping (1969), Algoskoufis (1987) and Mulligan (1999). Second, we find large differences in Frisch elasticity by sex, with males having elasticity in the range of 0.2 to 0.7, and females in the range of 1.3 to 1.5. Third, Frisch elasticity on the intensive margin only is very low for both sexes, males, and females, in the range of 0.1 to 0.2. The last finding is consistent with Heckman (1993), that extensive margin dominates intensive margin in most of the changes in labor supply behavior. Regarding the possible changes in Frisch elasticity in the 1990s, it has been either unchanged or slightly declining on the extensive and intensive margins combined, although it was either unchanged or slightly rising on the intensive margin and possibly declining on the extensive margin. These changes may be generated by the increase in part-time workers. In Japan, there is a huge discrepancy between full-time and part-time hourly wages even after controlling for individual attributes. The substantial increase in the ratio of part-time workers since 1990s, therefore, may influence our results to some extent. An analysis using longitudinal data is needed to check this possibility.21 There is also a possibility that the life-cycle model does not hold up, which suggests another issue for further study: an estimation of labor supply elasticities based on other theoretical models.