Explaining Fluctuations in Cuba’s Participation Rate
Cuba’s labor force participation rate (the ratio of the labor force to the population of working age) has fluctuated widely over the past three decades The participation rate is important for policy makers because it measures the degree of labor utilization, and therefore the pressure of aggregate demand on productive capacity. It is particularly helpful in a country like Cuba where the official (published) unemployment rate is basically useless as a measure of economic slack.
From 1989 to 1994 the participation rate fell abruptly as the economy went into its post-Soviet depression (See Figures1 and 2). It bottomed out in 1995-1996 as the economy began to recover following a stabilization plan. And it increased steadily during the expansion period from 1997 -2011 owing partly to large-scale investments and loans from Venezuela. After 2011, however, the participation rate fell sharply against the background of substantial cuts in the state labor force that were offset only in part by a rise in private employment. The ageing of the Cuban population also appears to have played a role in the recent fall in the participation rate. The participation rate recovered somewhat in 1989 and 1990.
LABOR FORCE PARTICIPATION AND REAL EARNINGS
In both the post-Soviet depression and the recovery that followed, the participation rate had a strong cyclical component. And both episodes can be explained by a labor supply model relating the participation rate to real earnings: Figures 1 and 2 show a close correlation between the two variables for the period 1989 to 2012. The figures use two different price variables to construct real wages: (1) the GDP deflator and (2) the consumer price index, or CPI. The problems with both variables are explained in the Annex.). But no matter what price variable was used to deflate nominal wages, the two figures show a correlation between the participation rate and real wages for the period 1989 to 2012. This result is strongly confirmed by the regressions presented in the Annex, and they confirm that workers move in and out of the labor force in response to changes in remuneration.
 The wage variable used in Figures 1 and 2 is the nominal wage in the state sector. There is currently no information on earnings in the private sector; data used to be published by ONEI but were eventually discontinued. As defined here, the state sector includes government agencies, state enterprises, and government controlled agricultural cooperatives.
 It is more familiar to observe the movement of workers from employment to unemployment (and vice versa). As noted above, however, open unemployment in Cuba is small and does not seem to reflect the tightness of the labor markets (see Hernández-Catá , 2019). Thus, in the Cuban context, workers are likely to move from employment to outside the labor force (and vice versa) in response to changes in real earnings or other factors affecting the attractiveness or discomfort of labor.
The labor supply model works well until 2012. But it collapses thereafter as the participation rate plunges despite rising wages. The participation rate now appears to be affected by a completely different set of factors. This is an important but, in my view temporary phenomenon associated with special factors; and the basic labor supply model will probably continue to play a role in the furfure. Nevertheless, the recent drop in participation is unusual and startlingly large. The rest of this paper seeks to provide an explanation.
THE IMPACT OF THE PLAN TO CUT DISGUISED UNEMPLOYMENT.
As shown in Fig. 3 and in Table 1, the fall in the participation rate reflected a decline in the labor force more than a rise in the population of working age. So, the question is why did the labor force fall so abruptly after 2011? From a policy perspective, this fall represented an abrupt brake with the past. Before 2011, the Cuban authorities influenced the level of state employment primarily by setting wages in the official sector and by providing subsidies to government agencies and state enterprises to avoid open unemployment—at the cost of raising disguised unemployment. (The concept and measurement of disguised unemployment is explained in Hernández-Catá (2019).
But this system was radically modified in 2011. In that year, the Cuban government, understandably dissatisfied with the magnitude of redundant employment in the official sector, decided to bring about a sharp decline in the number of state employees. At the same time, the government introduced several measures to encourage the state employees that were laid off to join the private sector: (i) private farmers were allowed to work formerly state-owned agricultural land in usufruct; (ii) the number of categories in which the private agents were allowed to operate was increased somewhat, although key sectors like health, education and energy, remained the province of the official sector; and (iii) a few regulations that had stifled the private sector were removed.
The expectation was that the employees released from the state sector would find a job in the private sphere, thereby cutting hidden unemployment and increasing labor productivity without adverse effects on total employment. The participation rate would remain unchanged and only the composition of employment would change. It was a good idea, and the plan worked smoothly in 2001, its first year of operation. As shown in Fig. 3 and Table 1, state employment dropped by an unprecedented 314 thousand workers in 2011, virtually all of which was absorbed by the private sector. The labor force remained roughly unchanged. However the plan ran into trouble afterwards. Large cuts in state employment continued through 2017, albeit at a slower pace. But private employment failed to grow at a pace sufficient to absorb these cuts. The result was a net contraction of the workforce that reached more than half a million workers.
The reason why job creation fell short of what was required is that, given existing restrictions on the size of the private sector, the administration’s plan simply ran out of room. After the success in 2011, there was no room for further expansion of the private sector, and any further cuts in state employment would trigger outflows from the labor force. The only solution was an official decision to allow the private sector to expand.
The net decline in employment (the sum of the declines in state and private sector jobs from the right column of the table divided by the population of working age in 2011), shows the impact of the employment cuts on the participation rate (the orange cell on the right-hand side of the table). The last line of the table indicates what the participation rate would have been in the absence of the labor market plan. It registered a decline of 4.8 percentage points during the period, compared with an actual decline of 12.6 percentage points.
An important question is how do workers cope when they leave the ranks of the employed and give up the benefit of a salary? The literature does not provide a clear answer, but there are several possible guesses. Those who leave the labor force might join the informal secgor, where income might be higher (and work less annoying) than in the state sector. Indications are that migration to the underground economyis is substantial, which might indicate that the the conventional participation rate is understaterd.
Fired workers might survive, at least for a while, by living out of remittances from relatives abroad; or stay at home temporarily while they look for a satisfying job in the private sector. Finally, those who are married could live out of their spouses’ salary while still performing productive duties by waiting in long lines to buy rationed goods, thus relieving their spouses from a time-consuming task andallowing them to continue working
There is a vast literature on the structural determinants of the participation rate, originating mainly in the United States: Aaronson. et.al (2014); (Barnichon (2019); Hornstein and Kidlyak (2017); Hornstein, Kudlyak and Schwuenert (2018); Valletta and Barlow (2018); Hall and Petrosky-Nadeau, (2016); and, most importantly Krueger (2017. This literature has been motivated in part by the sharp fall in the U.S. participation rate—that started a few years before the decline in Cuba’s rate. It has relied on the rich data base covering the U.S. labor market, which, of course is unavailable in Cuba. However, the data published by ONEI is sufficient to analyze some of the basic questions raised in this paper.
The analysis of structural factors affecting the U.S. labor market focuses mainly on the age and gender composition of the population and particularly on participation rates. However, it does not always achieve this goal. Most of models in this define a nation’s aggregate participation rate (Rt) as the weighted sum of participation rates for each age group (Rit), where t indexes time and i indexes the share of the population in each age group (θi). The precise equation is:
Rt = ∑ Rit θit
The formula can then be used to calculate the impact of each age group on the aggregate participation rate. As Krueger (2017) correctly emphasizes, however, the formula is simply the result of an algebraic decomposition, and it has no empirical content. Any interpretation of the result for the aggregate participation rate Rt must informed by prior knowledge about the individual participation rates. As Krueger notes, the formula is “just an accounting identity”, and “no factors are considered apart from demographics”—a major shortcoming, particularly when earnings variables are absent..
Independently of the formula, however, Krueger does complement the calculation of age groups by empirical elements, and so do Aaronson and his associates. Krueger’s important contributions are to examine extensively the role of opioid consumption, and the degree of job satisfaction in explaining shifts away from the labor force. force. The Cuban database does not allow for such refinements, but the issue of opioids may be irrelevant for Cuba because the Island’s people are too poor to afford these drugs. This may be true. As for the degree of job satisfaction, it is certainly relevant to the worker’s decision to remain employed but it is difficult to measure. It may be captured, at least in part, by the real wage variables included in the equations of the Annex.
Hornstein et.al. (2018) also consider the level of education of the U.S. labor force, and Mary Daly et.al. (2018) find that extensive parental leave policies in Canada explains most of the gap between Canadian and U.S. participation rates’
 They find that higher average levels of education in the United States tend to labor force participation.
Sources: ONEI and author’s calculations.
Note. The solid black line is the actual level of the dependency ratio; the broken red line is the value of this ratio held constant at its 2011 level The dotted line measures participation rate as a function of the dependency ratio. The distance between the dotted line and the constant 1917 red line shows the evolving gap between the actual participation rate and what it would have been had it remained at its 2011 level. The vertical double arrow shows the 2017 value of this gap,
ONEI publishes data on the age composition of the population featuring three age groups: the young (aged from zero to 14 years); the adult (from 15 to 59 years); and the old (60 years and older). The evolution of these groups shows some interesting features. (i) The share of the young group falls gradually throughout the period since 1989. (ii) The adult group is unchanged until 2009, but it falls by about two percentage points thereafter. (iii) The share of the old age group increased steadily from about 12% of the population in the early 1990s to almost 21% in 2019. The share of the old age group is projected to increase to 26% in 2025.
ONEI also calculates a “dependency ratio” by dividing the combined populations shares of the young and the old by the adult population share. This ratio is displayed in Fig.4 (black solid line) and is used below to calculate the impact of the rising old age population on Cuba’s participations rate.
The model used in this paper can be summarized by considering two equations. The first,
Rt = δ Dt, defines the relationship between the participation rate Rt , and the dependency ratio Dt times them negative parameter δ—a relation for which we have econometric evidence (see Annex). This relation applies for all time periods and therefore the second equation can be written as Ro = δ Do, where the subindex zero refers to 2011 values (the doted red line in Fig.4). Subtracting the second equation from the first yields:
Rt -R0 = δ (Dt – D0)
Where the right-hand side is equal to δ multiplied by the difference between the actual level of the dependency ratio and its constant 2011 level (i.e., the vertical distance between the solid black line in Fig. 4 and the red horizontal 2011 line). The left–hand side is the difference between the actual level of the participation rate and its constant 2011 level). It is also the extent to which changes in the dependency ratio since 2011 affected the participation rate (represented geometrically by the gap between the dotted line and the 2012 horizontal red line).
The final step is to evaluate the variables at their 2017 levels, for which have known numerical values, and select the value of δ from its regression coefficient in equation 2b in the Annex table. This yields a right-hands side total of 2.14; this is the impact of the rise in the dependency ratio on the participation rate from 2011 to 2017. Note that the dependency ratio is projected to increase further after 2017, and therefore it will continue to exert downward pressure on the participation rate in subsequent years.
This paper attributes the recent sharp drop in Cuba’s participation rate to two factors: first, the net decline in employment resulting from the government’s attempt to eradicate hidden unemployment (7.9 percentage points); and second, the ageing of the Cuban population (2.1 percentage points). Together these factors accounted for almost 80 percent of the actual decline in the participation rate from 2011 to 2017. This does not mean that all these factors were taken into consideration—in fact it is possible that participation would have increased in the absence of these factors. Possible candidates for further research in Cuba, include the possible role of education and the gender composition of the labor force.
From a policy perspective, there is not much the authorities can do about the ageing of the population, at least in the short run. It is not impossible that, in the longer run, and if the government is willing, population growth might be boosted by the return of part of the Cuban diaspora. But even that is highly uncertain.
In contrast, this article’s conclusions have clear implications for policy in the short to medium term. After the initial success of the government plan in 2011, the fall in state employment exceeded the creation of private jobs, causing a large outflow of workers from the labor force. The way to avoid this outcome in the future is to allow the private sector to operate freely in areas that, so far, have been the exclusive province of the state. This would seem to face insurmountable ideological obstacles. But it should not. Here’s a way to move in the right direction.:
Authorize private clinics, hospitals, and private medical consultancies.
Allow private schools, institutes and universities.
Authorize private consulting firms dealing with engineering, economics, legal, and accounting issues.
Allow Cubans to own and operate stores, hotels, restaurants, and warehouses without restriction of size, even if it involves replacing state owned businesses.
Last but not least, authorize free trade unions and allow workers to assemble, to participate in the management of firms, and to strike
* * *
One last comment. The plan to cut back disguised unemployment led to the elimination of thousands of redundant jobs, allowing budgetary saving and creating some incentives to search for openings. Moreover, the outflow of workers released from the state labor force did not necessarily imply a decline in active employment, because they were previously redundant. Disguised unemployment was transformed into open unemployed, and there has no reduction in aggregate production. This is not a bad outcome, but it would be better if those workers could find a job in an expanded private sector.
This annex presents a set of regressions explaining Cuba’s labor force participation rate over the period 1989-2012. In this period, the participation can be explained using a simple labor supply model relating participation to the real wage. The regressions do not cover the period after 2012, when the participation rate suffered a large decline. In my view, the behavior of participation in that period is related to special and partly transitory factors.
The annex table includes two blocks of four regressions each. The first block uses the GDP deflator to construct the real wage rate (equations1a to 1d); the second relies on the consumer price index (2a through 2d). Both variables have problems: the GDP deflator because it is heavily affected by price controls; the CPI because it covers only a subset of prices. The consumption deflator would have been more appropriate, but it is not available for the required time period. An experimental approach using sequential use of consumption deflators based on different time periods yielded results like those obtained using the GDP deflator.
Equation 1a reports on the simplest form of the labor supply model. The real wage rate is correctly signed and significantly positive at the 0.01 confidence level, based on a one tailed t-test (the criterion applied to all coefficients in this annex). Equation 1b adds the dependency ratio; both variables are also correctly signed and statistically significant. Equations 1c an 1d test for the hypothesis money illusion by introducing the nominal wage and the price level separately. Both variables are independently significant, but do not differ significantly from one another in absolute value.
The CPI-based equations 2a through 2d do not differ markedly from the GDP deflator-based regressions, except that they show some support for the money illusion hypothesis. The coefficients of the nominal wage variables in equations 2c and 2d are significantly larger (in absolute value) than those of the price variables.
The equations based on the labor supply model collapsed after 2012. By way of example, in equation 3, estimated for the period 1989-2019, the coefficients of the explanatory variables are insignificantly different from zero. Compared to the equations estmated for the shorter period 1989-2012, equation 3 implies is a sharp drop in explanatory power.
 D2017 = 547; D0 = D2011 = 567; and δ = -0.1068.
Aaronson, Stefanie, Bruce Fallikck, Andrew Figura, Jonathan Pingle, and William Wascher (2006). “The recent Decline in the Labor Force Participation Rate and its Impications for Potential Labor Supply”. Brookings Papers on Economic Activity.
Barnichon, Regis (2019).Is the Hot Economy pulling New Workers into the Labor orce?” FRBSF Economic Newsletter. Federal Reserve Bank of San Francisco. (May)
Daly Mary C., Joseph H.Pedtke, Nicolas Petrosky-Nadeau, and Annemarie Schweiner (2018). “Why Aren’t U.S. Workers Working?”. FRBSF Economic Newsletter. Federal Reserve Bank of San Francisco. (November).
Hernández-Catá, Ernesto (2019). “Why is Cuba’s Unemployment Rate So Low? Or is it really that Low?“ Cuba in Transition, vol 29. Association for the Study of the Cuban Economy. Miami FL.
Hernández-Catá, Ernesto (2015). “Estimating Hidden Unemployment in Cuba “. Cuba in Transition. Association for the Study of the Cuban Economy. Miami FL.
Hornstein, Andreas, Mariana Kudyak, and Annemarie Schweiner (2018). “The Labor Force Participation Rate rend and its Projections”. FRBSF Economic Newsletter. Federal Reserve Bank of San Francisco. (November).
Krueger, Alan B. (2017). Where have All the Workers Gone? An Inquiry into the Deecline of the U.S Labor Force Participation Rate. Brookings Papers on Economic Activity. Washington D.C. (Fall).
Valetta and Borlow (2018). FRBSF Economic Newsletter. Federal Reserve Bank of San Francisco.