SciELO - Scientific Electronic Library Online

 
vol.51 número2INTERGENERATIONAL MOBILITY AND ENTREPRENEURSHIP IN URUGUAYIS ENTREPRENEURSHIP INHERITED? A STUDY OF INTERGENERATIONAL SOCIAL MOBILITY IN MEXICO índice de autoresíndice de materiabúsqueda de artículos
Home Pagelista alfabética de revistas  

Servicios Personalizados

Revista

Articulo

Indicadores

Links relacionados

  • En proceso de indezaciónCitado por Google
  • No hay articulos similaresSimilares en SciELO
  • En proceso de indezaciónSimilares en Google

Compartir


Latin american journal of economics

versión On-line ISSN 0719-0433

Lat. Am. J. Econ. vol.51 no.2 Santiago nov. 2014

http://dx.doi.org/10.7764/LAJE.51.2.227 

 

INTRA-GENERATIONAL SOCIAL MOBILITY AND ENTREPRENEURSHIP IN URUGUAY*

 

DANIEL BUKSTEIN**

NÉSTOR GANDELMAN***

* The authors would like to thank Virginia Robano, Eduardo Lora, Francesca Castellani, and participants at the Inter-American Development Bank workshop "Strengthening Mobility and Entrepreneurship: A Case for the Middle Classes" for helpful comments and suggestions. All errors and omissions are the authors'.
** Daniel Bukstein dbukstein@hotmail.com) is affiliated with Universidad ORT Uruguay and Agencia Nacional de Investigación e Innovación of Uruguay.
*** Nestor Gandelman (gandelman@ort.edu.uy) is the Director of the Department of Economics of Universidad ORT Uruguay.


In this paper we follow an income-based, time-dependence approach to measure the impact of entrepreneurship on social mobility in Uruguay. The working definition of entrepreneur is business owners with employees. Using household surveys from 1982 to 2010 we show that their income level, income volatility, and evolution over the business cycle are consistent with them being opportunity entrepreneurs. Self-employed workers are more similar to necessity entrepreneurs. We find significant evidence that entrepreneurship is associated with greater social mobility while self-employment is not.

JEL classification: O15, L26, D31.

Keywords: income mobility, social mobility, entrepreneurship, pseudo-panels.


 

1. INTRODUCTION

Among Latin American countries, Uruguay has the lowest income inequality. Traditionally it has been considered a country with a large and stable middle class. However, inequality and segregation have been growing in Uruguay, accompanied by greater polarization between the rich and the poor.

The relatively large size of the Uruguayan government has often been considered responsible for the country's better income inequality statistics than other Latin American countries. The Organization for Economic Cooperation and Development (OECD) (2011) presented evidence that the government and the middle sectors are more closely intertwined in Uruguay than in other countries. For example, 21% of employed middle sector household members work in public administration, the highest figure in Latin America. Within the academic literature, there is an ongoing debate about the level of entrepreneurship among the middle classes. On the one hand, Acemoglu and Zilibotti (1997) and Doepke and Zilibotti (2007) consider that the larger a country's middle class, the more likely it is to have a strong business class, based on the belief that the values and attitudes that are typical of middle classes promote innovation. On the other hand, Banerjee and Duflo (2008) take a dissenting position based on the relatively lower percentage of entrepreneurs among the middle sectors when compared to wealthier sectors of society.1

Successful entrepreneurship is associated with upward movement on the social scale, but there are no recipes for success. A straightforward prediction of most economic models is that entrepreneurs in equilibrium have higher but also more volatile income levels. The implication of this is that entrepreneurship is likely to push many households up the income scale but it will also push down others. In times of economic volatility, the latter may exceed the former. In this paper we explore the relationship between entrepreneurship and intra-generational social mobility.

In doing so, one problem that must be addressed is attrition. Suppose that panel data shows that at age 18, all individuals must decide whether to apply for a salaried job or start a new enterprise. As time passes, many who chose to become entrepreneurs will see their enterprises fail and end up joining the labor force. Without controlling for this survival bias, we would overestimate the impact of entrepreneurship on social mobility.

To address social mobility, ideally we would have access to repeated observations of a sample of the population. Unfortunately, panel data is not available in most developing countries. The best alternative is to use repeated cross-sectional (RCS) surveys to construct pseudo-panels, as we do here. There are two positive attributes of RCS data worth mentioning. First, in a cohort of entrepreneurs, there are some who are successful and some who fail. The data from the cohort represent an average of all these individuals and, therefore, the problem of non-random sample attrition is somewhat minimized. Second, pseudo-panels have fewer measurement problems than other sources because they average individuals in adequately constructed cohorts. With large enough cohorts the average measurement error tends to zero.

In this paper, as in the papers of this special issue, an entrepreneur is a business owner who employs at least one person. This definition of entrepreneurship is in line with the OECD-Eurostat Entrepreneurship Indicator Programme (EIP), which defines entrepreneurs as "those persons (business owners) who seek to generate value through the creation or expansion of economic activity, by identifying and exploiting new products, processes or markets." Therefore, empirically entrepreneurship and business ownership are very similar. Our working definition of entrepreneurship does not include those business owners who do not have employees, since self employment is often the fallback position of those unable to insert themselves into the labor market.

The goals of this paper are:

a) to evaluate social mobility in Uruguay,

b) to evaluate whether entrepreneurs have more or less social mobility than the rest of society.

2. DATA

We use household surveys (Encuesta Continua de Hogares, ECH) from the National Institute of Statistics (Instituto Nacional de Estadística, INE). These surveys are conducted annually to gather household composition data including age, gender, educational level, and labor market variables. The ECH surveys cover Montevideo, the capital city, and urban areas with over 5,000 inhabitants in the rest of the country. Only since 2006 has the INE gathered information from rural settings, so our study is restricted to urban areas.2 We include heads of household 21-65 years old; the number of such households is reported in Table 1 below.

 

Table 1. Number of households 1982-2010

 

3. METOHODOLOGY

3.1 Measuring social mobility with pseudo-panels

The starting point of the income-based time-dependence approach to social mobility is the following regression:

where yit represents the log of per-capita income of household i at time t and uit is a disturbance term. The coefficient β of lagged income is the measure of social mobility. A value of β equal to 1 is interpreted as a situation of no social mobility, whereas a value of β below 1 represents a situation of income convergence. A situation of total income mobility occurs in the extreme case that β is equal to 0 when current income has no relationship to its past value. The coefficient β obtained from Equation (1) is usually referred to as a measure of unconditional convergence, as it is estimated in a regression with no covariates other than past income.


(1)

Including additional controls in the regression leads to an estimate of β that constitutes the conditional convergence:


(2)

where X is a vector of covariates and γ measures the impact of these covariates on present income.

To conduct this kind of analysis, the researcher ideally should have information about the same individuals over time, i.e., panel data. However, panel data are not usually available in developing regions such as Latin America. Deaton (1985) presented a way to address the paucity of panel data by constructing pseudo-panels using a series of RCS. A pseudo-panel is formed by creating synthetic observations obtained by averaging observations from groups of individuals, usually called cohorts, with similar time-invariant characteristics in a sequence of RCS data sets. The most commonly used of these characteristics is birth year, although it may also be combined with gender, place of birth and/or educational level, or other pre-determined household characteristics. This way, the cohorts can be viewed as being "followed" over time, the same way individuals are followed over time with true panel data; hence the name pseudo-panel.

Considering the pseudo-panel nature of the data, equations (1) and (2) take the following forms:

where the individual index i has been replaced by the cohort index c(t). The notation c(t) indicates that the cohort is time-dependent, while the flat lines above the variables indicate that the values represent sample averages of the cohort c(t) in period t. Like equations (1) and (2), the coefficient β of lagged income is interpreted as a measure of unconditional or conditional convergence. One problem with estimation of these latter equations in growing economies is that the cohort's current income should be higher than the cohort's past income. Therefore the estimated coefficient is biased upward. To prevent this bias, time dummies should be included in the estimations. There is a great deal of literature that addresses the conditions under which the parameters of equations (3) and (4) can be consistently estimated, given the limitations that arise when working with pseudo-panel data as opposed to real panel data. Examples of this literature are Deaton (1985), Moffitt (1993), Verbeek and Vella (2002) and Antman and McKenzie (2005), among others.


(3)

 


(4)

3.2 Measuring social mobility for groups of interest

In this section we extend the income-based approach to measure social mobility for specific sectors. We illustrate this by considering differences in social mobility that can be attributed to entrepreneurship, but the same approach can be used to address differences by gender or in terms of head of household education.

Examining the simpler case, suppose we have panel data and can follow the same set of households over time. One way to measure entrepreneurs' social mobility is to estimate a regression of the form:

(5)

where eit is a dummy variable valued at 1 if the ith household head is an entrepreneur and 0 otherwise. In this case, the slope coefficient β1 represents income mobility for non-entrepreneurs, while the sum β1 + β2 represents social mobility for entrepreneurs.

The cohort version of (5) is:


(6)

where ec(t)t is the percentage of entrepreneur households in cohort c(t). The interpretation of Equation (6) is similar: β12 is the income mobility for entrepreneurs, while β1 is the income mobility for non-entrepreneurs.

A key difference when addressing differences in social mobility by entrepreneurship or self-employment with other characteristics like gender or educational level is that the former may change over time. Therefore the percentage of entrepreneurs in a cohort varies in response to the business cycle in a similar fashion as the average per-capita income if the cohort changes. In other words it may be that entrepreneurship and income are jointly determined. To address this endogeneity we run instrumental variable regressions in which current cohort entrepreneurship is instrumented by the first two lags of this variable. In the regressions with self-employment we use their two lags as instruments. Since gender and educational level of the household head are pre-determined variables that do not change over time, there is no need for instrumental variables.

Pseudo-panel construction

In constructing the cohorts, we made sure they were large enough, otherwise the average characteristics per cohort would not result in good estimates for the population cohort means. If the cohort is too large, the cohorts that comprise the number of observations in our estimations will be small. Balancing cohort size and number of cohorts is essential for consistent estimation of pseudo-panels. In this vein, Verbeek and Nijman (1992) and Antman and McKenzie (2005) showed that large cohort sizes are necessary to overcome the "artificial" nature of pseudo-panel data, and to treat them as genuine panels that allow for consistent estimates of the parameters.

The cohorts were constructed using household heads between the ages of 21 and 65, divided into five-year spans according to birth year. In our estimations, we have expanded this definition to also define pseudo-panels by birth year and gender, by birth year and education level (above and below the birth cohort median), by birth year and entrepreneurship status, and by birth year and self-employment status. The cohorts defined by birth year and gender or by birth year and education level are relatively standard in the sense that gender and educational level are pre-determined characteristics that do not change over the time of analysis.3 The cohorts defined by birth year and entrepreneurship or self-employment status are more problematic since employment status can change over time. The estimations produced with these cohorts should be taken with precaution. We present them here for the sake of completeness and to check the robustness of the results obtained by using them and the general results using other alternatives. In all cases, frequency weights were used to construct the cohorts to appropriately mimic the structure of the Uruguayan population.

Given that we are working with household heads between ages 21 and 65 and that our first survey year is 1982, the first cohort observed contains individuals born between 1920 and 1924, and the last cohort in 2010 contains individuals born between 1980 and 1984. Note that the aggregation of individuals born over five different years causes each of the survey year cohorts to be measured over a span of ages, e.g., the 1920-1924 birth cohort in 1982 represents ages from 58 to 62. As we were not able to follow all the individuals, or cohorts, over time in an equal number of periods because of restrictions imposed by the available survey years and the ages we worked with, we ended up with an unbalanced pseudo-panel of 13 cohorts and 237 observations. When the cohort is defined by birth year and other characteristics, such as gender, education, and entrepreneurship/self-employment, we end up with twice as many cohorts and observations. Table 2 shows the distribution of the 237 observations per the birth year cohort definition, the average number of household heads in each cohort, and the percentage of entrepreneurs and females. The cohort defined by birth year and education is the result of taking the median cohort education level and dividing it between those more and less educated. The median cohort education level was calculated for each cohort for the entire time that it was observed.

 

Table 2. Cohorts

 

4. MEASURING INOOME AND ENTREPRENEURSHIP

Income is measured in per-capita terms adjusted by purchasing power parity (PPP) to 2005 U.S. dollars.4. This is sometimes called "absolute mobility" in contrast to a different measure of mobility where income is normalized by the year median.5

It is important to clarify what we mean by "entrepreneur" in this paper. Acs (2006) differentiates between opportunity and necessity entrepreneurs. The former are those who find unexploited business opportunities and transform them with their income-generating activity. The latter are individuals with low probabilities of successfully inserting themselves into the formal labor market who end up self-employed in low-productivity activities. We are mostly interested in effects for opportunity entrepreneurs rather than necessity entrepreneurs. Using household surveys, this distinction is difficult to make empirically because there are no good proxy variables for this classification that are uncorrelated with income and income mobility.

Our estimations are at the household level. The household surveys allow for classifying individuals by their labor status, i.e., distinguishing between the status of those who own a business and have employees, those who are self-employed, have no employees and have a fixed workplace, and those who are self-employed without a fixed workplace. In our definitions we consider a household an "entrepreneur household" if its main income depends on someone who is in charge of their own business and has employees. Those who run their own businesses but do not have employees are in an intermediate category between entrepreneurs and employees; they may be either opportunity or necessity entrepreneurs. We do not consider them entrepreneurs in our estimations; instead, we refer to them as self-employed.6

4.1 Descriptive statistics

Figure 1 presents an overview of income evolution during the period of study, illustrating the general growth trend and the years marking the two significant crisis episodes during the last 30 years in Uruguay: 1982 and 2002. Figure 2 presents the evolution of income by groups of interest. All groups follow the same trend and are similarly affected by the business cycle. However, there are sizeable income differences. Entrepreneurs' households have on average about three times the per-capita income of the self-employed who do not have a fixed workplace; they have 80% more income than the self-employed with a fixed workplace and other employed. The self-employed who lack a fixed workplace are stuck in low-productivity occupations, which accounts for the low income expected of necessity entrepreneurs. Households with more educated heads have about 100% higher incomes than households with less-educated heads. There are no sizeable differences in per capita income between male and female household heads, but this is not contrary to typical gender income differences. Female household heads are not a random sample of females and thus have different characteristics than other females.7

 

Figure 1. Income and GDP

Source: World Bank and authors' calculations based on household surveys.

 

Figure 2. Household per capita income
(PPP adjusted averages)

Source: Authors' calculations based on household surveys.

 

Although the evidence indicates that entrepreneurs tend to be wealthier than non-entrepreneurs, this has no implication for social mobility. Entrepreneurs have on average larger incomes, but they also experience more volatility. The standard deviation of entrepreneurs' income is twice that of the other employed. The standard deviation of income for both types of the self-employed is lower than that of income for other employed. During the 2002 crisis, income in households without entrepreneurial activity fell by 10%; in households with entrepreneurial activity, the decline in income was 15%. Here we find a sharp difference between entrepreneurs and other individuals, including the self-employed, since entrepreneurial activity involves substantially more risks than other activities. This lower volatility of income for the self-employed is surprising and suggests that the self-employed are not true entrepreneurs.

Figure 3 reports the percentages of households with entrepreneurship or self-employment activity. Entrepreneurship and self-employment with fixed workplace are pro-cyclical, which is what we would expect of opportunity entrepreneurs. It is interesting to note, however, that they differ in terms of the magnitude of their response to the cycle. In the 1999-2002 recessions, entrepreneurs' households experienced a larger decline than households whose main income came from a self-employed person with a workplace. By contrast, households with self-employment in a workplace show a larger income increase than entrepreneurs in the most recent years following the general economic bonanza. It may be that some of these self-employed will end up hiring employees and becoming entrepreneurs according to our definition. Unlike those two groups, the percentage of self-employed household heads without a fixed workplace is countercyclical. This is consistent with the latter being necessity entrepreneurs who prefer to be employees in a salary-based relationship when the economic situation improves.

 

Figure 3. Percentage of households with entrepreneurship or
self-employment activity

Source: Authors' calculations based on household surveys.

 

On average, in about 5% of households there is entrepreneurial activity, i.e., there are business owners with employees. The self-employed who have a fixed workplace represent 7% of households, and the self-employed who do not have a fixed workplace represent 3% of households. Kantis et al. (2012) report information on the occupational composition for Argentina, Brazil, Peru, Ecuador, and El Salvador. Our results suggest that Uruguay has about the same level of entrepreneurial activity as Brazil, more activity than Argentina, and less activity than Ecuador and El Salvador. The number of self-employed in Uruguay is well below that of other countries; this is likely due to the lower degree of informality in the Uruguayan labor market.

5. RESULTS

Table 3 reports the first set of social mobility results. The first three columns present estimations for different cohort definitions based on pre-determined characteristics. The last three columns correspond to cohorts defined by birth year and employment status. Since employment status may change over time, these pseudo panels do not follow exactly the same type of individuals over time. The results, therefore, are technically more questionable. We present them here to stress that the result on social mobility is robust to various cohort definitions.

 

Table 3. Social mobility according to various alternative cohort definitions
(absolute convergence, PPP adjusted income in logs)

 

The estimates are large, but they are statistically different from 1 in most cases. These estimates are similar, or somewhat below, those presented in Table 5, model I of Cuesta et al. (2011) for Uruguay. They show a small level of income convergence. The extreme results appear when the cohorts are defined by gender (large convergence) and educational level (almost no convergence).

Table 4 reports the estimation of Equation (3) for subsamples of the population. It shows the degree of mobility within the cohorts of these groups. As in Table 3 the first (four) columns refer to cohorts defined by pre-determined characteristics while the latter (four) refer to cohorts defined by birth and employment status. The table shows much larger convergence among females than among males. It also shows greater social mobility among the more educated than among the less educated and greater social mobility among the self-employed and entrepreneurs than among others.

 

Table 4. Social mobility within groups

 

The main results of this paper are presented in Table 5, which shows the measures of conditional convergence corresponding to Equation (6). According to this table, entrepreneurs experience greater social mobility than non-entrepreneurs. Entrepreneurship reduces the coefficient of social mobility by about 0.0869 or 0.0768, depending on the inclusion of additional controls. Although the point estimates suggest also that both types of the self-employed have greater social mobility than other individuals, these effects are not statistically significant.

 

Table 5. Impact of entrepreneurship and other household characteristics
on social mobility

 

The results for gender and educational level are less robust. Without other controls, we find no significant differences in social mobility for male and female headed households but we do find larger social mobility for the more educated. The last column, including all controls, shows the reverse picture with statistically significant effects for females but not for the more educated.

6. CONCLUDING REMARKS

In this paper we measure intra-generational social mobility in Uruguay using an income time-dependence approach applied to a large pseudopanel, finding evidence of low unconditional convergence. Our results suggest that there is greater mobility within the cohorts of certain groups of the population, i.e., females and those more educated, than between groups.

We address the link between entrepreneurship and social mobility, despite the difficulty of measuring the concept of entrepreneurship. We show that business owners with employees have more income than other employees, but also that they experience much greater income volatility. The self-employed have about the same, or even less, income volatility than other employees. Therefore, there is an important difference in risk-taking between entrepreneurs and the self-employed. We also show that the percentage of households whose main income depends on a business owner with employees evolves pro-cyclically and this is also the case for the self-employed with a fixed workplace. The percentage of self-employed without a fixed workplace is countercyclical.

These findings make clear that business owners with employees behave like opportunity entrepreneurs in that they take more risks and follow the business cycle, i.e., in booms they find more business opportunities than in recessions. It is also clear that the self-employed without a fixed workplace are necessity entrepreneurs who would rather have a salaried job. The situation of the self-employed with a fixed workplace is less obvious. Overall, we do not find that they face significant risks due to income variability, but they flourish in booms by following the business cycle.

Although in this paper the operational definition of entrepreneurs includes only business owners with employees, we also present the results on social mobility for both types of self-employed persons.

After controlling for the endogeneity of employment status, we find that mobility is much greater for entrepreneurs than non-entrepreneurs and we fail to find statistically significant differences patterns for the self-employed.

The methodology used in this paper does not allow for measuring upward and downward mobility. The greater mobility of entrepreneurs is a confirmation of the larger risks that they face, which are not only a part of their work, but affect their families and their households disposable income. With all other factors constant, entrepreneurs with lower socioeconomic status are more likely to move up the social ladder than non-entrepreneurs, but they are also more likely to fall into extreme poverty.

Policies promoting micro-entrepreneurship, such as microfinance programs, should bear this in mind and include careful evaluation of the probability of success for potential entrepreneurs. Promoting entrepreneurship is not a guaranteed method for fighting poverty. Duflo et al. (2013) report evidence from a randomized evaluation of the impact of introducing micro credit lending in India, finding significant differences in the probability of having a microcredit between the slums where microcredit was introduced and those where it was not, but no differences in expenditure per capita.

The underperformance of Latin American countries in terms of productivity is related to the existence of many low-producing micro-firms (Pagés 2010). These firms benefit from a relatively broad range of public policies designed to provide support for small enterprises (subsidized lines of credit, technical assistance, etc.). Our results suggest that governments should not confuse social assistance programs, e.g., transfers, with programs designed to improve the efficiency of resource allocation in society. Rather than social assistance, policies to foster entrepreneurship should be designed to increase productivity and efficiency.

NOTAS

1 For a more extensive literature review, see Castellani and Lora (2014).

2 This is not an important limitation since the urban population accounts for 94% of the country's total population.

3 Although years of schooling is a personal decision, by age 21 most individuals have completed their formal education. Those who have not yet completed their education are for the most part university students who (even before obtaining their university degree) fall in the higher education group. Therefore dividing the cohort by the median education level allows for following the same individuals over time.

4 The purchasing power parity (PPP) conversion factor is the local currency unit per dollar. Source:
World Development Indicators.

5 In the working version of this paper we also present estimates of relative income mobility.

6 Gandelman and Robano (2014) use the same working definition of entrepreneurship.

7 Similarly, Gandelman (2009) shows that, on average, female household heads in Latin America are more likely to own their homes. After controlling for the endogeneity of homeownership and female household heads, the author reports a negative association between females and homeownership for most countries.

 

REFERENCES

Acemoglu, D. and F. Zilibotti (1997), "Was Prometheus unbound by chance? Risk, diversification, and growth," Journal of Political Economy 105(4): 709-51.         [ Links ]

Acs, Z. (2006), "How is entrepreneurship good for economic growth?" Innovations Journal, Winter.         [ Links ]

Antman, F. and D. J. McKenzie (2005), "Earnings mobility and measurement error: A pseudo-panel approach." Stanford University mimeo.         [ Links ]

Banerjee, A. and E. Duflo (2008), "What is middle class about the middle classes around the world? Journal of Economic Perspectives 22(2): 3-28.         [ Links ]

Castellani, F. and E. Lora (2014), "Is entrepreneurship a channel of social mobility in Latin America? Latin American Journal of Economics 51(2): 179-194.         [ Links ]

Cuesta, J., H. Ñopo and G. Pizzolitto (2014), "Using pseudo-panels to estimate income mobility in Latin America," Institute for the Study of Labor (IZA) Discussion Papers 5449.         [ Links ]

Deaton, A. (1985), "Panel data from time series of cross sections," Journal of Econometrics 30: 109-26.         [ Links ]

______ . (1997), The analysis of household surveys: A microeconometric approach to development policy. Baltimore: Johns Hopkins University Press.         [ Links ]

Doepke, M. and F. Zilibotti (2005), "Social class and the spirit of capitalism," Journal of The European Economic Association 3(2-3): 516-24.         [ Links ]

Duflo, E., A. Banerjee, R. Glennerster and C. Kinnan (2013), "The miracle of microfinance? Evidence from a randomized evaluation." NBER Working Papers 18950.         [ Links ]

Kantis, H., J. Federico and L. Trajtenberg (2012), "Entrepreneurship, economic mobility and entrepreneurial propensity. A regional view based on the analysis of selected Latin American countries," Inter-American Development Bank Working Paper Series No. 315.         [ Links ]

Gandelman, N. (2009), "Female-headed households and homeownership in Latin America," Housing Studies 24(4): 525-49.         [ Links ]

Gandelman, N. and V. Robano (2014), "Intergenerational mobility and entrepreneurship in Uruguay," Latin American Journal of Economics 51(2): 195-226.         [ Links ]

Moffitt, R. (1993), "Identification and estimation of dynamic models with a time series of repeated cross-sections," Journal of Econometrics 59: 99-123.         [ Links ]

OECD (2011), Latin American economic outlook. Paris: OECD Development Centre.         [ Links ]

Pagés, C. (2010), The age of productivity: Transforming economies from the bottom up. Washington, D.C.: Inter-American Development Bank, Palgrave MacMillan.         [ Links ]

Verbeek, M. and T. Nijman (1992), "Can cohort data be treated as genuine panel data?" Empirical Economics 17(1): 9-23.         [ Links ]

Verbeek, M. and F. Vella. (2002), "Estimating dynamic models from repeated cross-sections," K.U. Leuven Center for Economic Studies, mimeo.         [ Links ]

 

Creative Commons License Todo el contenido de esta revista, excepto dónde está identificado, está bajo una Licencia Creative Commons