Cuba’s labor participation rate, the ratio of the labor force to the population of working age, has fluctuated widely over the past 2 1/2 decades. From 1989 to 1994 the rate fell abruptly as the economy went into its post-Soviet depression (see figure below). It bottomed out in 1995-96 as the economy began to recover; and it increased rapidly from 1996 to 2011, a period of strong economic expansion. The participation rate fell from 2011 to 2013 against the background of a slowdown in growth and large-scale transfers of employees from the state to the private sector.
Studies on the participation rate in other countries[1] have emphasized the role of two factors: (i) sociological and demographic factors, such as the age and gender composition of the labor force; and (ii) cyclical factors, such as economic activity and wages. This article contends that cyclical factors, notably wage rates, are a key determinant of labor participation in Cuba. The idea is that workers in Cuba, like in any other country, respond to monetary incentives: increasing their participation in the labor force and searching for jobs when wages rise; and withdrawing from the labor force when wages fall.
As shown in the figure, both the participation rate (F/N) and the employment rate (the ratio of employment to population of working age, E/N) display a pronounced cyclical behavior. The two series are strongly correlated, which is not surprising because the employment rate is a major component of the participation rate. However, the cyclical component is more pronounced in the employment rate than in the participation rate, because the latter includes the unemployment rate which is anti-cyclical. The difference between the population of working age (N) and the labor force (F) consists of two groups of individuals: (i) those who do not wish to work for various reasons (housewives who prefer to stay at home and take care of children, early retirees, and invalids) and are permanently out of the labor force; and (ii) discouraged workers, i.e., those who would like to work but remain out of the labor force at least for some time because they find current employment prospects so dim that job search is not worth the effort. The latter group, just like the members of the labor force, is likely to respond to wage incentives. A fall in wages, such as occurred during the post-Soviet crisis, is likely to induce unemployed workers to move out of the labor force.[2]
To test this hypothesis we relate the participation rate to the logarithms of the real wage rate (equations 1 and 2 in the table below). Two alternative price variables are used: the GDP deflator and the consumer price index. In equations 3 and 4, the nominal wage and price variables are entered separately to test our prior belief that there is a substantial degree of nominal wage rigidity in Cuba.
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
The regression results confirm that real wages have a significant effect on the decision of workers to move in and out of the labor force as the coefficient of the real wage is significantly larger than zero in all equations.[3] The results are ambiguous regarding the hypothesis of nominal wage rigidity; the equation using the GDP deflator indicates that the absolute difference between the wage and price coefficients is not significantly different from zero, suggesting that only real wages matter. In contrast the equation using the CPI shows that the coefficient of the nominal wage rate is significantly higher (in absolute value) than the price coefficient, pointing to nominal wage rigidity. It should be noted that the two price variables behave very differently in the initial phase of the sample period—with a very large CPI/deflator differential opening up in 1989-1993 and vanishing in the next several years—for reasons that are not entirely clear. A price deflator for price household consumption was also used in the regressions and performed roughly like the GDP deflator.
The four equations are in the table are labor supply functions. They were estimated by ordinary least squares (OLS), a procedure which maximizes the R2 of the equations but yields biased estimates of the estimated coefficients. In this context, this is because the labor market also includes a demand for labor function, which we model as the equality between the real wage and the marginal product of labor. Accordingly, equation 1 was re-estimated by two-stage least-squares (2SLS), yielding an estimated real wage coefficient of 15.4—significantly higher that the OLS coefficient of 12.2 shown in the table for equation 1. [4]
The results presented above are encouraging, but they must be interpreted with caution because several seemingly relevant variables were not included or turned out to play an insignificant role. First, the wage variable used related only to the state sector; there is no comprehensive data on wages (or earnings) in the private sector. Second, the equations did not include demographic variables even though studies in other countries have found them to be important (see Aaronson, et. al. 2006). As in other countries, the participation rate in Cuba is higher for men than for women and displays a different cyclical behavior. [5] The appropriate variable to capture these differences (a combination of the share of women in the population of working age and the participation rate differential) could not be constructed for the entire sample period, partly because of a sharp and unexplained break in series in 1998. Therefore, separate participation rate regressions for men and women were estimated, but only for the period 1999-2013. Both showed a significant relation between participation and real wages. However, the equation for women revealed a much higher wage sensitivity, suggesting that females, are less attached to the labor force—i.e., that they tend to be discouraged (or encouraged) more easily than men by a given change in real wages.
***
In conclusion, both the employment and the participation rates in Cuba are highly sensitive to changes in real wages, and this explains much of the wide fluctuations in these variables over the past 2 1/2 decades. In particular, the participation rate is significantly affected by changes in real wages, suggesting that workers, women in particular, move in and out of the labor force in response to changes in remuneration. An important implication of these results is that open unemployment is a very poor indicator of the tightness of labor markets and the level of unused resources: policy-makers need to look at a broader concept of unemployment, one that includes both open unemployment and discouraged workers.
All this is in addition to the fact that employment in Cuba has been strongly affected by the subsidization of loss-making enterprises which has generated a large and highly variable level of disguised unemployment. The decline in both inactive employment and discouraged workers during the past two decade suggests that the degree of unused resources narrowed considerably more than what could be inferred from the decline in open unemployment. This, of course, has implications for the sustainability of low inflation and exchange rate stability. Against that backdrop the recent decline in participation signals that pressures on capacity may have eased somewhat.
[1] See, for example, Stephanie Aaronson, Bruce Fallick, Andrew Figura, Jonathan Pingle and William Wascher (2006). “The Recent Decline in the Labor Force Participation Rate and its Implications for Potential Labor Supply”. Brookings Papers in Economic Activity, 1.
[2] The open (published) unemployment rate actually fell from 1989 to 1994, which may seem unbelievable given the dramatic depth of the output contraction. The Main reason for this phenomenon is that state employment was heavily subsidized by the government, leading to a surge in disguised (rather than open) unemployment. But a discouraged worker effect may also have prevented open unemployment from rising.
[3] The difference between the working age population and the labor force (expressed as a percentage of N) is
(N-F)/N, or 100% – F/N. Therefore, using equation 1 as an example, (N-F)/N = 39.9% + 12.2 ln(w/p).
[4] The procedure involved regressing the natural logarithm of the real wage, ln(w/P), to the independent variables of the model, i.e. the natural log of real GDP and the population of working age. In a second stage, equation 1 was re-estimated using the fitted value of ln(w/p) as a regressor. Relative to the OLS estimate, the 2SLS estimate has lower bias but higher variance.
[5] Another variable that is not included in the Table is the age structure of the labor force, which other studies have found to be important in explaining participation. Unfortunately, data for a sufficiently long period are not available. Information on hires and on the duration of unemployment is inexistent in Cuba.