Labor Informality in Latin America and the Caribbean: Patterns and Trends from Household Survey Microdata*

Informalidad laboral en América Latina y el Caribe: patrones y tendencias a partir de microdatos de encuestas de hogares

Leonardo Gasparini
Leopoldo Tornarolli **

* This paper is a follow-up of our contribution to the World Bank LAC Flagship Report on Informality in Latin America and the Caribbean (Perry, Maloney, Arias, Fajnzylber, Mason and Saavedra-Chaduvi, 2007). We are very grateful to the encouragement and comments of Omar Arias, Jaime Saavedra, Bill Maloney, Sebastián Galiani, and Jamele Rigolini, and seminar participants at UNLP, World Bank and LACEA. All the statistics were computed at CEDLAS-UNLP by the authors and Georgina Pizzolitto, Francisco Haimovich, Victoria Fazio, Julieta Pron, Pablo Gluzmann, Ana Pacheco, Hernán Winkler, Matías Horenstein, Evelyn Vezza, Javier Ibarlucia, Elena Cadelli, Rocío Carbajal, Sergio Olivieri, Gimena Ferreyra and Rafael Brigo. We are especially grateful to the excellent research assistance of Carolina García Domench. The usual disclaimer applies.

** CEDLAS, Universidad Nacional de La Plata. E-mails: leonardo@depeco.econo.unlp.edu. ar and ltornarolli@depeco.econo.unlp.edu.ar. CEDLAS is the Center for Distributional, Labor and Social Studies at Universidad Nacional de La Plata (Argentina). Web page: www.cedlas.org

This article was received july 18, 2008, modifed january 17, 2009 and finally accepted march 18, 2009.

Abstract

This paper documents the main patterns and trends of alternative definitions of labor informality in Latin America and the Caribbean, by exploiting a large database of more than 100 household surveys covering the period 1989-2005. The evidence suggests that there are no signs of a consistent pattern of reduction in labor informality in the region. Regardless of the definition used, labor informality remains a pervasive characteristic of labor markets in LAC. In several countries the increase in labor informality seems to have been associated more to a sizeable increase in the propensity to set informal arrangements within groups, than to changes in the national employment structure toward more informal sectors.

Key words: informality, employment, Latin America, Caribbean, labor market.

JEL Classification: J01, J21, J31, J42, J8.

Resumen

Este artículo documenta los principales patrones y tendencias de definiciones alternativas de informalidad laboral en América Latina y el Caribe (ALC), explotando una base de datos de más de 100 encuestas de hogares en el período 1989-2005. El trabajo no encuentra evidencia empírica a favor de un patrón consistente de reducción de la informalidad laboral en la región. Independientemente de la definición usada, la informalidad continúa siendo una característica dominante de los mercados laborales en ALC. En varios países el incremento en la informalidad laboral parece estar asociado más a un importante aumento en la propensión a fjar arreglos informales en todos los sectores productivos, que a cambios en la estructura nacional de empleo hacia actividades más informales.

Palabras clave: informalidad, empleo, América Latina, Caribe, mercado laboral.

Clasificación JEL: J01, J21, J31, J42, J8.

Introduction

Academics, policy-makers and commentators have extensively argued about the size of the informal labor market, its welfare implications and the adequate policy prescriptions. The debate, however, is often obscured by the fact that the term informality is ambiguous from a theoretical point of view, and difficult to implement empirically¹. Labor informality usually means different things to different people. While some researchers associate informal labor to low-productivity marginal jobs, others prefer to limit the concept to employment not complying with the legal norms in terms of labor taxes, regulations, and social protection.

This paper makes a contribution to the analysis of labor informality in Latin America and the Caribbean (LAC) by presenting evidence on the main patterns and trends of alternative definitions of informal labor. In particular, we implement a "productive" definition for which informal workers are those in low-productivity jobs in marginal small-scale and often family-based activities, and a "legalistic/social protection" definition for which informal workers are those with no access to social protection or right to certain labor benefits.

The evidence presented in this paper is based on microdata from a large set of more than 100 household surveys covering the period 1989-2005, taken from the Socioeconomic Database for Latin America and the Caribbean (SEDLAC), a project jointly developed by CEDLAS at the Universidad Nacional de La Plata and the World Bank’s LAC poverty group. This database allows us to provide a broad picture of the main trends and patterns of labor informality in LAC, and hence, hopefully, contributes to a better informed debate on this issue in the region.

The study of labor informality has a long tradition in the economics literature in Latin America. Most of the studies, however, are limited to a single definition of informality and restricted to one country or a small set of economies. The aggregation of this large set of studies into a survey that provides a unifed body of evidence is difficult, as individual researchers construct variables from household surveys in different ways, and take different methodological decisions, many of them not reported in the papers.

This article contributes to the study of labor informality in LAC by providing, for the whole region, an extensive set of statistics for alternative definitions of informality, constructed by applying similar definitions and methodologies across countries. The paper is mostly descriptive: it offers a broad view of labor informality in the region, without attempting to get deep into its determinants. However, by showing the results of correlations, multivariate regressions and microsimulations, the paper provides useful preliminary evidence that helps to think about labor informality, and hopefully, motivates future research.

The rest of the paper is organized as follows. Section I discusses the concept of labor informality, the alternatives to empirically estimate it, and proposes specific implementations with the information available in the national household surveys of the region. In particular, we implement a productive definition of labor informality associated to the type of job, and a legalistic definition associated to the access to social protection linked to the employment.

Section II is the core of the paper, as it presents the main patterns and trends of labor informality at the country level using alternative def-nitions. Most unskilled workers in LAC are informal for any of the definitions. They are self-employed or salaried workers in small firms without a signed contract in compliance with labor regulations, and without access to social protection and labor related benefits. In fact, that is also the labor condition for a sizeable share of skilled workers in the region. This situation does not seem to be the consequence only of economic stagnation. Despite a positive performance during some periods, most countries in the region have not experienced significant increases in the share of workers in the formal sector. Labor informality remains a pervasive characteristic of labor markets in LAC. The incidence of this phenomenon substantially differs across countries (e.g. from 70% in Bolivia and Paraguay to 40% in Chile, according to the productive definition).

In section III we look at wages and hours of work of informal workers. In particular, we provide estimates of the conditional wage gap of being informal. On average, informal (in the productive sense) male workers without a secondary education earn 30% less than their formal counterparts. In nearly all countries salaried workers without social protection also earn substantially less than formal salaried workers. In contrast, hours of work do not differ much across groups. Entrepreneurs and large-firms employees work in general more hours than in the public sector, while hours of work are approximately the same for the rest of the groups.

Section IV takes a look at changes in informality over the business cycle to assess whether informal employment moves pro or anti-cyclically with the economy and relative wages across sectors. We find that in the recessions informality increased along with a fall in relative wages. However, the symmetric story for the economic expansions did not take place: in several LAC economies informality also increased during periods of strong GDP growth. The evidence of increasing informality both in expansions and downturns in several countries is challenging as it calls for explanations that go beyond the economic cycle.

Section V is aimed at characterizing changes in labor informality at the country level over time. A given increase in the level of labor informality in an economy could be the consequence of either a change in the structure of employment in favor of groups with high propensity to informal arrangements (e.g. unskilled services), or a generalized increase in the propensity to informality for all groups. We examine this issue by applying a microeconometric decomposition methodology. We find that in some South American countries the growth in informality is mainly associated to a sizeable increase in the propensity to informality in most groups, and not to a change in the employment structure. El Salvador is the only country in our sample where a fall in informality is driven entirely by a change in the employment structure.

In section VI we carry out some counterfactual simulations to characterize the differences in informality across countries. In particular, we compare the actual informality rate in a given country to the coun-terfactual rate that would arise if that country "imported" only the observable characteristics of some other economy. The results of the decompositions can be used to assess scenarios under which a country may reduce informality.

Section VII closes with some brief concluding comments.

I. Measuring labor informality

There are at least two different concepts that are referred by the term labor informality². The "productive" definition pictures informal workers as those in low-productivity, unskilled, marginal jobs, while the "legalistic" or "social protection" definition stresses the lack of labor protection and social security benefits³. It is important to make clear from the outset that the definitions do not correspond to competing views about informality, with different welfare implications and policy prescriptions. Instead, they refer to different phenomena in the labor market. The productive definition is concerned with the type of job (e.g. salaried vs. self-employed, large vs. small firms), while the legalistic definition is concern with the compliance of the labor relationship with some rules, mainly labor protection. We follow the tradition of using the same term informality to refer to these two different aspects of the labor market.

The "productive" view classifes as informal those workers in low-productivity jobs in marginal small-scale and often family-based activities. ILO (1991) defines the informal sector as economic units "with scarce or even no capital, using primitive technologies and unskilled labor, and then with low productivity". Maloney (2004) includes in the informal sector the "small-scale, semi-legal, often low-productivity, frequently family-based, perhaps pre-capitalistic enterprises".

Naturally, it is very difficult to empirically implement this notion, since things like "productivity" are unobservable, others like "capital endowment" are not usually reported in surveys, while others like "marginal", "pre-capitalistic activities" or "primitive technologies" are difficult to define. In practice researchers have tried to adjust this notion of informality to the information usually contained in surveys. Hence, the empirical implementation of informality has been linked to (i) the type of job (salaried, self-employment), (ii) the type of economic unit (small, large, public sector), (iii) and the worker’s skills. Following this practice we divide the working population into seven groups: (1) Entrepreneurs (patrones), (2) Salaried workers in large private firms, (3) Salaried workers in the public sector, (4) Salaried workers in small private firms, (5) Skilled self-employed, (6) Unskilled self-employed and (7) Zero-income workers.

To implement this classification we include as unskilled all individuals without a tertiary or superior education degree, and we define as small all firms with 5 or fewer employees⁴. Given that an individual could have more than one job, we apply the classification only to his/ her main occupation. We implement the following definition of labor informality:

Definition 1 (productive definition): An individual is considered an informal worker if (s)he belongs to any of the following categories: (i) unskilled self-employed, (ii) salaried worker in a small private firm, (iii) zero-income worker.

Labor informality is closely related to self-employment. However, we exclude the self-employed with a tertiary degree from the group of informal workers. The group of skilled self-employed is mainly comprised by professionals and technicians usually with high productivity and fully incorporated into the modern economy. In fact, the professional self-employed is the group with the highest earnings in most countries in the region (see section III). Following a standard practice, we include salaried workers in small firms into the definition of informality. The assumption, which of course is debatable, is that most salaried workers in those firms operate using primitive technologies and with low productivity. In fact, many of these small firms are run by individuals who declare themselves being self-employed. Finally, we also add the group of zero-income workers into the informal sector. Household surveys in the region have this category to include mostly family workers, i.e. individuals who perform some activity in a family-based enterprise but who are not formally paid for that job.

The inclusion of patrones (entrepreneurs/employers) into the formal sector is debatable, since in practice some of them are just self-employed in a low-productivity activity using scarce capital and some few unskilled workers. There are two practical problems regarding this group: (i) it is difficult (probably impossible) in theory to set a line separating out the entrepreneurs from just the self-employed employing some workers, and (ii) even when we attempt to do it, there are some data limitations. For instance, most surveys do not report the number of employees working for a patrón. We have decided to include the patrones into the formal sector following a usual practice, and because earnings in that group are much higher than for the self-employed in all LAC countries⁵.

This discussion confirms that the productive definition of labor informality is theoretically weak and empirically difficult to implement. However, it has lasted for decades and it is extensively used in the academic and policy debate, because it refers, although in an ambiguous way, to a relevant characteristic of the labor markets in Latin America.

Although having statistics (and hence a definition of) labor informality is sometimes useful, in many of the following sections we work with the seven categories defined above separately. For many uses the binary formal/informal definition implies too much aggregation. Also, in some cases we find useful to stress the distinction self-employed-salaried workers, instead of the formal-informal grouping discussed above. A second strand of the literature has stressed the "legalistic" or "social protection" notion of informality. Informal firms are those not complying with the norms in terms of labor contracts, labor taxes, and labor regulations, and then their workers have no rights to labor protection or social benefits linked to employment. ILO (2002) defines an informal worker as one "whose labor relationship is not subject to labor legislation and tax rules, and has no access to social protection or right to certain labor benefits"⁶.

This second notion is also difficult to implement. There are at least two severe problems. The frst one arises from the fact that the number of dimensions to be included under labor protection and social security is large and varies across countries. Labor protection includes contracts, severance payments, advance notice, right to be unionized, workplace safety, vacations, working hours and many more. Social security includes pensions, health insurance, unemployment insurance and other insurances and benefits. Countries differ in the extent of their labor protection and social security systems. Moreover, even in a given country regulations and social security rights differ by sector, by tenure, or other work characteristics, and change over time. Therefore, it is difficult in theory to come up with a legalistic definition of a formal worker that is suitable for all countries and situations.

The second problem is practical. Even if we agree to a simple defini-tion of an informal worker (e.g. signed contract and right to pensions when retired), household surveys widely differ in terms of coverage of labor protection and social security issues. Some surveys ask about contracts and some do not. The type of questions aimed at capturing the right to health insurance is very different across countries, and in some cases it is impossible to know whether health insurance is linked to employment. The coverage on severance payments and unemployment insurance is very low, while the questions on insurance for accidents in the workplace are almost inexistent. In fact many LAC countries do not have comprehensive systems of insurances on many risks (including unemployment), so the National Statistical Offices do not include questions on these issues.

The right to receive a pension when retired is the social security beneft most asked in LAC household surveys. However, not all countries have questions on this item, and in those that have, questions are different. Moreover, in most countries the questions apply only to salaried workers, leaving all the self-employed as missing. In this paper we implement the following legalistic/social-protection definition of informality:

Definition 2 (legalistic or social protection definition): A salaried worker is informal if s(he) does not have the right to a pension linked to employment when retired.

Table 1 shows the specific social-protection definition of labor informality adopted in each country with relevant information in its household survey.

The productive and social protection definitions of informality are surely highly correlated. However, as mentioned above, we do not keep one and discard the other in this study, since we are interested in the two definitions for different conceptual reasons. The next section shows statistics on both definitions and discusses the possible overlapping.

II. Labor informality: patterns and trends

In this section we document the structure and patterns of informality under the two definitions discussed above. But frst we introduce the source of information for our study.

A. The data

All the statistics in this paper are obtained by processing microdata from household surveys, and are part of the Socioeconomic Database for Latin America and the Caribbean (SEDLAC), jointly developed by CEDLAS at the Universidad Nacional de La Plata and the World Bank’s LAC poverty group (LCSPP). The SEDLAC contains information on more than 100 household surveys in 21 LAC countries. Table 2 lists the surveys used in the study. The sample covers all countries in mainland Latin America, and four of the largest countries in the Caribbean (Dominican Republic, Haiti, Jamaica and Suriname). In each period the sample represents around 93% of LAC total population. Most household surveys included in the sample are nationally representative. The three exceptions are Argentina and Uruguay, where surveys cover only urban population which nonetheless represents more than 85% of the total population in both countries, and Suriname, where the survey is restricted to the city of Paramaribo (around 50% of the population of the country).

Household surveys are not uniform across LAC countries. The issue of comparability is of a great concern. We have made all possible efforts to make statistics comparable across countries and over time by using similar definitions of variables in each country/year, and by applying consistent methods of processing the data. However, perfect comparability is far from being assured. A trade-off between accuracy and coverage arises. The particular solution adopted contains an unavoidable degree of arbitrariness. We tried to be ambitious enough to include all countries in the analysis, and accurate enough so not to push the comparisons too much. In any case, we provide the reader with relevant information to assess the trade-offs⁷.

B. Informality I ("productive" definition)

Table 3 reports information on the share of workers in each of the seven categories defined above according to the type of work. Although the employment structures are roughly similar across countries, there are some relevant differences. Several countries have around 30% of their workers in large firms. That share is lower in less developed and more rural countries. Public sector employees are more than 10% of the labor force in the most developed countries of the region: Argentina, Brazil, Chile, Costa Rica, Dominican Republic, Mexico, Panama, Uruguay and Venezuela⁸. Self-employed professionals are a minority in LAC. Only in Argentina, they represent more than 3% of total employment. In contrast, the unskilled self-employed are a sizeable group in all countries. In fact, it is the largest group in Bolivia, Colombia, Dominican Republic, Ecuador, Guatemala, Haiti, Honduras, Nicaragua, Paraguay, Peru and Venezuela. More rural countries have a large size of their population as zero-income workers. That is the case of Bolivia, Ecuador, Guatemala, Honduras, Jamaica, Nicaragua, Paraguay and Peru⁹.

Figure 1 shows a substantial dispersion in informality rates across countries. While the share of informal workers according to the productive definition is above 70% in Bolivia and Paraguay, the corresponding share is below 40% in Chile. Labor informality seems negatively related to per capita GDP (at PPP) and positively related to the share of rural population in the survey (figure 2). However, when including both variables in a simple OLS regression, the latter becomes non-significant.

Labor informality has not changed much in the region (see figure 3 and table 4). Only Brazil and Chile have experienced drops in the share of informal workers. In the rest of the countries, informality either increased or did not significantly change. Colombia, Honduras, Panama, Paraguay, Peru, Uruguay and Venezuela seem to have experienced a sizeable increase in the share of informal workers, according to the productive definition. That has occurred mainly in correspondence with a fall in the share of workers in large firms. The share of informal workers has not changed much in Argentina, Bolivia, Costa Rica, Dominican Republic, Ecuador and El Salvador.

The conclusions are similar when restricting the analysis to urban areas. In fact, in most countries the performance of the rural areas in terms of labor informality changes was not worse than that of urban areas, while in some countries rural areas did better (e.g. Brazil, Costa Rica, Nicaragua and Paraguay).

The probability of being informal is decreasing in the worker’s education (table 4). Instead, the profle for age has a U shape.

Figure 4 shows relative employment and wages of the self-employed compared to the wage earners for the sample of those workers without a tertiary degree living in urban areas. The figures may be consistent with the idea of voluntary self-employment (Maloney, 2004). Unskilled young people enter the labor market as wage earners, accumulate knowledge, capital and contacts, and then set up their own informal businesses.

Informality differs by sector of activity. In table 5 we divide the working population of each country into 10 sectors and record the share of informal workers¹⁰. Workers in primary activities are mostly informal (either unskilled self-employed, salaried workers in small farms, or family workers). On average, about half of the workers in the food and cloth industries in LAC are informal, most of them being self-employed. Informality is lower in the rest of the manufacturing industry: on average (LAC unweighted) around 40% of workers are either self-employed or wage earners in small firms (only a small fraction declare themselves being family workers). Construction workers are mainly informal: around 60% are either self-employed or salaried workers in small establishments (in roughly the same proportion). Informality is even higher in the commerce sector (on average 65%). Differences across countries are considerable: while 56% of Bolivian workers in the commerce sector are unskilled self-employed, 50% of Panama’s workers in that sector are employed by large firms. Informality is substantially lower in the skilled-services sectors (banking, business services, professionals). On average, informality is around 25%. Most workers in that sector are employees of large firms. In theory all public administration employees should be registered as public sector salaried workers, and hence be classifed as formal. This occurs for the vast majority of workers, but there are exceptions that could be due to measurement errors, or situations where people work for the public administration through small private firms (e.g. consulting jobs). In any case the registered informality rate in the public administration is around 1%. On average, around 30% of workers in the education and health sector are informal, being most of them unskilled-self employed. The relative low level of informality in the sector is mainly driven by the large share of the public sector in the provision of education and health. Finally, almost all domestic servants are informal. In most countries they are classifed as salaried workers in small "firms" (houses).

Informal workers are poorer than formal workers. This means that household income adjusted for demographic is lower for informal workers, not that they earn less than formal workers controlling for observable characteristics (next section has data on this). Table 6 provides details about the position of formal and informal workers (and of each of the labor categories) in the household income distribution of each country. On average, while 5% of formal workers belong to the poorest quintile of the household per capita income distribution, that share climbs to 22% for informal workers. In the other extreme, whereas more than 40% of formal workers are in the top quintile of the household income distribution, 15% of informal workers manage to get there.

The last panel for each country in table 6 divides the formal and informal working population into poor and non poor according to the international standard of USD 2 a day per person (at PPP). A worker is poor if her household per capita income is lower than USD 2 a day. In Argentina 2005, while 3.1% of formal workers are poor according to that measure, the proportion of the informal workers that are poor climbs to 16.6%. In all countries the difference in the poverty head-count ratio between informal and formal workers is sizeable (around 4 times on average).

Although most entrepreneurs are not poor, in some countries a non-negligible proportion of patrones is located in the low-income quintiles. Several measurement errors may cause this allocation. Surveys record current, not permanent income. Specifically, they report incomes in the month previous to the survey. Entrepreneurs’ incomes are usually volatile, and hence some of them may report low earnings in a given month¹¹. The second measurement error was already mentioned. Some patrones may be just self-employed workers with low-productivity and hence low earnings.

In Brazil while 30.7% of workers in large firms belong to the top quintile of the household income distribution, that proportion rises to 45.9% for the public sector employees and to 88.3% for the skilled self-employed. That pattern is valid for nearly all LAC countries, although with different intensities. A relatively robust ranking also holds for the three informal categories: the poverty headcount ratio for the zero-income workers is higher than for the self-employed, which in turn is higher than for salaried workers in small firms.

C. Informality II ("legalistic/social protection" definition)

As commented above, the Latin American household surveys have a weak coverage of labor and social protection issues. We could implement the social protection definition of labor informality in only 14 countries of the sample. Moreover, several of them have questions only in some years, and the type of question differs across countries (see table 1).

Table 7 displays the share of salaried workers without the right to receive pensions when retired. That informality rate is presented for several socioeconomic groups. Informality is relatively low in Chile and Uruguay (around 25%) and somewhat higher in Argentina, Brazil and Venezuela (around 40%). The share of unprotected salaried workers is around 60% and higher in Bolivia, Colombia, Ecuador, Guatemala, Mexico, Nicaragua, Paraguay and Peru (see figure 5). As with the productive definition, labor informality in the social protection sense seems negatively correlated to per capita GDP and positively correlated to the share of rural population in the survey (figure 6). Again, when including both variables in a simple OLS regression, the latter becomes non-significant.

The likelihood of having the right to pensions when retired is decreasing in education and has a U-shaped pattern with respect to age. The youth and the elderly are less covered by the social security system linked to employment than the adult population. While in some countries women are more likely to be informal than men, that situation is not generalized in the region. In contrast, labor informality is always higher in rural areas than in the cities.

We cannot provide a complete picture of what has happened with the social protection dimension of labor informality over the last decade in LAC with household survey data, since there are few countries with enough observations. Labor informality has increased in Argentina, Nicaragua and Venezuela, has remained roughly unchanged in Chile and Paraguay, and has slightly decreased in Brazil and Peru (figure 7). Probably the main conclusion from the evidence is that there are no signs of a pattern toward less labor informality in the region. Most results hold when restricting the analysis to urban areas.

Social protection is low among salaried domestic servants, construction workers and rural workers (table 8). Informality is in general lower in the manufacturing industry, the skilled services, and in particular in the education, health and public administration sectors. However, notice that while in principle we expect all public sector workers to be covered by basic social protection, on average 20% of them report not having access to pensions.

Household income for the formal employees is substantially higher than for informal salaried workers (table 9). The poverty headcount ratio for the USD-2-a-day line is on average 6 times higher for the latter group.

The presence of a formal contract is a key feature of a labor relationship. Signing a contract makes the relationship more visible, and then increases the likelihood for the compliance with the labor legislation. Unfortunately, only few surveys include questions on labor contracts. Table 10 reports the share of salaried workers having signed a contract. That share is above 75% in Chile, above 50% on Mexico and Panama, and below 45% in the rest of the countries in the sample. As with pensions, signed contracts are more common among prime-age adults, the skilled and urban workers. From the scarce information of the table there are no signs of a fall in informality. In fact, the share of salaried workers with contracts has fallen in Chile and Mexico, the only two countries for which data goes back to the early 1990s.

D. Comparing the two definitions

To what extent the two definitions of labor informality overlap? In table 11 we compute the share of workers without the right to pensions when retired (i.e. our definition of social-protection informality) by labor category (i.e. the basis for our definition of productive informality). An initial observation is that a sizeable share of workers classifed as formal by the productive definition are informal in the social-protection sense. Even in the public sector, pensions seem not to be a universal right. In 10 out of the 14 countries in the sample the share of uncovered public sector workers is above 10%. That share climbs for the other two formal labor categories. In particular, the share of uncovered self-employed professionals is high (around 90% in many countries). As it will be shown in the next section this group enjoys the highest earnings of all groups. The typical Latin American self-employed professional has high relative earnings, but (s)he is out of the social security system. The share of large-firms employees without right to pensions is also high on average, although with large variations across countries: while around 20% of those workers are uncovered in the Southern Cone, the share goes up to more than 60% in Ecuador, Bolivia, Paraguay and Peru.

The great majority of informal workers in the productive sense are also informal in the legalistic sense. The mapping is not perfect, particularly for the salaried workers in small firms. In some countries a significant fraction of these workers has rights to pensions (around 20% in Argentina and Venezuela, 30% in Brazil and Uruguay and 50% in Chile).

Table 12 classifes workers in each country according to the two def-nitions of informality. The last column records the share of workers which are consistently classifed as formal or informal by the two definitions. On average, more than 75% are in that group. That share is higher when considering all workers (instead of just salaried workers).¹² There are few workers who are informal in the productive sense but have access to social security (column (iii)). The relatively large social security systems in the Southern Cone account for most of these cases. Instead, there are more formal workers in the productive sense which are informal in the legalistic sense: the low social-security coverage of the self-employed professionals, and to a lesser extent the employees of large firms are behind the figures in column (ii).

III. Wages and hours of work

In this section we document relative wages and hours of work of different labor categories. We start by showing unconditional statistics and then turn to a multivariate regression analysis.

Table 13 shows relative hourly wages by type of work. In the frst panel the base group is wage earners, while in the second panel wages of public sector employees are set at 100. In our companion paper we also show statistics for hours of work. On average for the region, entrepreneurs work 10% more hours than salaried workers and earn per hour 2.5 times more. Compared to the wage earners, the self-employed work 10% fewer hours and earn 10% less per hour. However, this average hides a variety of situations across countries. In Chile and some Central American countries, for instance, hourly wages are higher for the group of self-employed.

The second panel breaks down the working population into more labor categories. In general, the ranking of hourly wages is leaded by the self-employed professionals followed by the entrepreneurs, the salaried workers in the public sector, the salaried workers in large firms, the unskilled self-employed, and the salaried workers in small firms. On average, the skilled self-employed earn around 60% more than public sector employees. Large firm’s employees earn 30% less than in the public sector. That percentage climbs to 50% for the case of the unskilled self-employed and to 60% for the wage earners in small firms.

Hours of work do not differ much across groups. Entrepreneurs and large-firms employees work in general more hours than in the public sector, while hours of work are approximately the same for the rest of the groups. The exception is the group of zero income workers for whom hours of work are 20% lower than in the public sector.

To further analyze wage differentials across groups we run regressions of the log of hourly wages against several controls and dummies for informal workers. The conditional measures of the earnings gap of being informal arising from these regressions should be interpreted with much care¹³. In particular, welfare comparisons drawn from these results may be misleading. An informal job differs from a formal one in many dimensions, not only in the hourly wage paid. If we find that hourly wages are the same in both sectors, the informal job may still be inferior since it precludes the access to social protection¹⁴, but it could be also superior, at least for some workers, since informality usually implies more flexibility: "being your own boss" is certainly a work amenity for many people.

There is a second reason why regressions should be interpreted with care. The informality coefficients may be biased if unobserved worker characteristics that affect productivity infuence the sector an individual chooses to work. It could be that only people with entrepreneurial ability choose to be self-employed, and then become successful. Or on the other hand, it could be that people with low work attachment and without ability to tolerate authority, responsibilities and punctuality choose to be self-employed, and then probably get low earnings, in part precisely because the lack of these characteristics.

Table 14 shows the results of estimating log hourly wage regressions using Heckman maximum likelihood for a sample of urban workers aged 15 to 70. We exclude skilled workers (i.e. with a tertiary degree) and the group of patrones from the analysis, and run the regressions for men and women separately. In addition to the usual set of controls (education, age, regional dummies) we include interactions between education and informality. In particular, we construct interaction variables by multiplying the informal binary variable with two educational dummies: one for those without any secondary education, and one for those with some high-school education. We also include interactions with dummies variables for the youth (15-24) and the elderly (56-70). Table 14 is divided into three panels according to the definition of informality. Panel A considers the productive definition. Since as said above we exclude skilled workers and employers, the regressions report the wage gaps between the (i) unskilled-self employed + small-firms salaried workers, and (ii) salaried workers in large firms and the public sector. In panel B we compare unskilled self-employed with unskilled salaried workers. Finally, in panel C we restrict the analysis to unskilled salaried workers and divide them according to the social protection definition of informality. In each panel the table shows the coefficients of the interaction variables.

In most countries being informal in the productive sense implies lower wages, even when controlling for observable factors. On average, informal male workers without a secondary education earn 30% less than their formal counterparts. The wage gap for those with secondary education is also significant, although somewhat smaller in most countries. Wage gaps of roughly the same magnitude are also present in the case of female workers. The coefficients of the interaction variables with age groups are mostly non-significant. In some few countries being informal is associated to higher wages for the youth and lower wages for the elderly.

Panel B indicates that while in half of the countries in the sample being an unskilled self-employed implies lower wages than being an unskilled salaried worker, in the other half there are no significant differences in wages. In panel C the results are more conclusive: in nearly all countries salaried workers with social protection also earn substantially more than informal salaried workers. That seems to be true for males and females and for both educational groups.

IV. Informality over the cycle

In this section we take a look at the behavior of informality over the business cycle. Do informal employment and relative wages across sectors move pro or anti-cyclically with the economy? It has been argued that the co-movements of these variables over the cycle can provide some preliminary evidence over the relevance of the dualistic view of informality.¹⁵ According to this hypothesis when the economy enters a recession, sticky wages in formal firms force them to fre workers, who find in the informal sector a way to survive waiting for better times. The informal sector serves as disguised unemployment by absorbing displaced workers during downturns. The fow of entrants into the "flex-wage" informal sector drives wages down relative to the formal sector which remains downwardly rigid. Hence, relative (informal/ formal) sector size and wages should move oppositely.

In contrast, under other assumptions and shocks, the two variables may go in the same direction. For instance, if informality is perceived as a close substitute for a formal job, an autonomous increase of the informal sector relative wage (e.g. after an autonomous increase in the relative price of non-tradables) should attract workers and hence increase the size of that sector.

We do not have enough data to carry out a rigorous test of the co-movements between the size of the informal sector, relative wages and the cycle.¹⁶ Instead, we present a preliminary analysis of these variables for the countries in the sample. Table 15 shows the ratio informal/formal for the number of workers and median hourly wages.¹⁷ As in the previous section, these ratios are shown for men and women separately, and for three alternative definitions of informality: (i) self-employed+salaried workers in small firms, (ii) self-employed, and (iii) salaried workers without right to pensions. In each country we also show an index of real per capita GDP based on purchasing-power-parity (PPP).

Some cases are consistent with the dualistic view of informality, while some others fit better into the voluntary view of informality. In Argentina, and according to the prediction of the labor-market-segmentation hypothesis, the share of informal workers greatly raised during the crisis that started around 1998. There is also some fall in the relative wage of informal workers, although that result does not hold when considering only the self-employed as informal. In Chile, the relative number of informal workers went down during the expansion 1990-1998, while relative wages for that sector increased. From 1998 to 2003 changes have been erratic.

The case of Brazil seems more consistent with the voluntary view of informality. During the economic expansion in the frst half of the 1990s both the relative size and wages of the informal sector grew. When the economy came to a halt in the late 1990s the share of workers in informal jobs remained roughly constant, along with relative wages.

Most LAC countries have experienced an economic expansion in the early and mid 1990s, followed by stagnation and even recessions in the late 1990s and early 2000s¹⁸. Table 16 summarizes the direction of the changes in relative size and wages between urban unskilled self-employed and their formal salaried counterparts. The patterns are similar across countries during recessions: the relative size of the informal sector increases, while relative wages fall. There are few exceptions to this behavior. Instead, during expansions the patterns have been different. Some few countries experienced similar changes as those commented above for Chile. That is the case of Mexico. The rest, instead, has shared the experience of Brazil with higher informality, although in half of the countries the increase in the informal sector size was not accompanied by a raise in relative wages.

Summarizing, during the recent recessions informality has increased along with a fall in relative wages, in accordance with the dualistic view of the labor market. However, the symmetric story for the economic expansions did not take place in most LAC countries. In several economies informality increased during periods of strong GDP growth. That fact may respond to a voluntary view of the labor market: in good times people take advantage of the larger set of opportunities and decide to be self-employed. Of course, the evidence of increasing informality both in expansions and downturns is also consistent with some structural changes that induced an increase in self-employment and that operated regardless of the economic cycle.

V. Changes in employment and informality

A given surge in the level of informality in an economy could be the consequence of either an increase in the propensity to informality within groups, or to a change in the structure of employment toward groups with high propensity to informal arrangements. In this section we examine this issue for the case of salaried workers and the social protection definition of informality.

Informality varies across groups. As discussed above, the access to social protection linked to the job is not uniform across age, gender and education groups. The heterogeneity is significant also across economic sectors, type of firms and jobs. Due to the need for more labor flexibility, high monitoring costs for the government, and other reasons some sectors have high propensity to informality. Construction workers and domestic servants are more likely to be informal than public sector employees. Also, part-timer workers, small-firm employees and newly-recruited staff tend to have, ceteris paribus, lower access to social protection in their jobs. If for some reason the structure of employment changes toward one of these groups, the average rate of informality in the economy will probably increase. On the other hand, the propensity to informality may increase within each group, making the overall rate to grow.

We carry out a decomposition in order to assess the extent to which observed changes in the overall rate of informality in a country are the consequence of changes in the structure of employment or in the propensity to informality within groups. To that aim we follow the microeconometric decomposition methodology of Gasparini (2002). The main inputs are the estimated coefficients of models for the informality status of a worker. The actual change in the informality rate between time t₁ and t₂ in a country is the consequence of changes in the characteristics of the population (the matrix of the independent variables in the regression) and changes in the estimated coefficients of the informality regression. We label these effects as characteristics and parameters effects.

Table 17 shows changes in the structure of employment of urban salaried workers, while table 18 documents changes in the share of informal workers (social protection definition) by group. The main results of the informality regressions are presented in table 19. We estimate probit models for the informality variable (social protection definition) for the sample of urban salaried workers. As regressors we include gender, age, age squared, educational dummies, equivalent household income, categorical variables for the type of firm, seniority, a dummy for part-time worker, and dummies for regions and economic sectors. All regressions are similar across countries, except for the definitions of the regional dummies.

The results of the decomposition exercises are shown in table 20. Given data availability we carried out the microsimulations only for seven countries. The results can be read as follows. Labor informality increased 6 points among urban salaried workers in Argentina between 1995 and 2003. If only the parameters linking observable characteristics to informality (i.e. the estimated coefficients in the frst two columns of table 19) had changed in that period, and all observable characteristics had remained fixed, informality would have increased by 7 points. On the other hand if only the observable characteristics of workers (including those of their jobs) had changed, informality would have fallen 1 percentual point. In fact, although the employment structure changed in some informality-increasing directions as the fall in the share of large firms, and the sizeable growth in part-time jobs, other changes were informality-decreasing, as the raise in the share of education, health and skilled services in total employment, and the reduction in the share of workers with low seniority (see table 17). On average, these changes between 1992 and 2003 were slightly informality-decreasing¹⁹. The large growth in informality seems to have been associated to a sizeable increase in the propensity to informality in most groups (the parameters effect) (see table 18). A similar story applies to the rest of the Southern Cone countries: Chile, Paraguay and Uruguay. In Brazil the characteristics effects was similar to that of their neighbors, but the parameters effect was smaller, averaging out a negligible change in overall informality. In contrast, Venezuela has large values of both effects, leading to a large increase in informality. El Salvador is the only country in the sample with a significant fall in informality driven entirely by a change in the employment structure in favor of prime-age adults, the skilled, and those employed in large firms.

IV. Characterizing differences in informality across countries

Recorded informality rates considerably vary across countries. Differences are in part due to noise in the information, since household surveys are not uniform in the region. But there are genuine differences rooted in the variety of productive and employment structures across the region. One of the main relevant differences is the rural-urban mix of the population. In more rural countries informality is expected to be higher. Table 21 shows rates for national, urban and rural areas. The standard deviation for the urban observations is 2 points lower than for the national observations.

But even ignoring rural areas differences in informality across countries remain large (see figure 8). In this section we characterize these differences using microsimulation techniques similar to those applied in section V (also based in Gasparini, 2002). In particular, we compare the actual informality rate in a country A to the counterfactual rate that would arise if that country "imported" only the observable characteristics of some other country B. That exercise implies keeping the parameters that govern the relationship between observable characteristics and informality fixed at the country A’s values.

Country A may have a higher informality rate, measured as lack of social protection, than country B due to a different employment structure, even when within each group informality is the same as in the other economy. For instance, country A may have a larger construction sector or a larger fraction of its labor force as part-time workers. But it could also be the case that for each particular group for some reason informality is higher in A. For instance, it could be that construction is carried out mainly by big urban development firms in country B which tend to be more formal, and that the government in B has more effective instruments to audit labor regulation for part-time workers. The decompositions allow us to have an idea of the relative magnitude of these two channels. Of course this is not a general equilibrium exercise. When we import the characteristics of country B into country A the parameters would probably change. A larger part-time labor force may induce the government to increase the efforts to auditing the compliance with labor regulations (or to give up, given the size of the task…). In this sense, the microsimulations are partial-equilibrium exercises that illustrate the size of the direct channels through which each change operates.

The results of the decompositions can be used to assess scenarios under which a country may reduce informality. A larger characteristics effect implies that by transforming the employment structure country A may reduce informality to the country B’s level. That may require progress in education, demographic transitions or sectoral changes in production, all phenomena related to economic development. Instead, a large parameters effect suggests that for some reason informality is larger in A for each group (or most groups), and that may be more related to specific policy issues, as high tax pressure, low auditing efforts, or insufficient legislation.

The decompositions are carried out for both definitions of informality. In the social protection case we restrict the analysis to urban salaried adult workers, while in the productive definition the sample includes urban adult workers. The regressions that estimate the parameters of the informality models are similar to the ones explained in section V and shown in table 19.

The results of the decompositions are shown in table 22 for the productive definition of informality, and in table 23 for the social protection definition. The frst panel in table 22 shows that if Argentina imported the parameters of Chile, informality would fall from 44% to 35%, that is, a parameter effect of -9 points (see third panel). If Argentina kept its parameters but took the observable employment characteristics of Nicaragua, informality would increase from 44% to 53%, i.e. a characteristic effect of +9 points (see second panel).

Take the case of Paraguay to illustrate the results in table 23 That South American economy has the highest levels of informality under both definitions in the sample. If Paraguay manages to change its employment structure to mimic a more developed economy like Argentina, Chile or Uruguay, informality in the labor protection sense would fall by around six points. The effect would be much larger if Paraguay manages to "copy" the parameters of other countries. For instance, informality would fall 33 points by taking the parameters of Chile or Uruguay while keeping the same structure of observable characteristics. In general, the parameter effects are substantially higher than the characteristic effects under the social protection definition of informality. The difference in general is not large under the productive definition.

VII. Concluding remarks

We have presented a general picture of labor informality in Latin America and the Caribbean by showing a wide set of statistics for a sample of 21 countries. The evidence suggests that there are no signs of a consistent pattern of reduction in labor informality in the region in the last two decades. Regardless of the definition used, labor informality remains a pervasive characteristic of labor markets in LAC. The evidence of increasing informality both in expansions and downturns in several countries is challenging as it calls for explanations that go beyond the economic cycle.

The cross-section evidence seems to be consistent with the idea of voluntary self-employment. Unskilled young people enter the labor market as wage earners, accumulate knowledge, capital and contacts, and then set up their own informal businesses. However, on average, being informal implies lower wages, even when controlling for observable factors. Informal male workers without a secondary education on average earn 30% less than their formal counterparts. Accordingly, in all countries the difference in the poverty headcount ratio between informal and formal workers is sizeable. In most countries informal workers have lost ground against their formal counterparts in terms of hours of work, but not in terms of hourly wages.

In several countries the increase in labor informality, as defined by the lack of social protection, seems to have been associated to a sizeable increase in the propensity to informality in most groups. The same conclusion arises when comparing labor informality across countries. Understanding differences in informality over time and across countries seems to be much more complicated than accounting for different labor structures.

The legalistic or social protection definition of informality is probably the most interesting to study, and the most relevant for many policy issues. One way to learn about labor informality in this sense is by comparing country experiences on social protection. Although certainly subject to many caveats, the country comparisons are often in practice the most compelling pieces of evidence over economic policy arguments. Unfortunately, the information on social protection contained in the LAC household surveys is still scarce, heterogeneous and volatile. A generalized effort toward a better and more homogeneous coverage of social protection issues in household surveys would surely be socially very productive.

FOOT NOTES

¹ In this paper the term informality always refer to labor informality. There is a large literature that studies the broader issue of the informal or shadow economy. See, for instance, Schneider and Enste (2000) who measure the size of the shadow economy in 76 developing, transition, and OECD countries using various methods.

² See Fields (1990), Guha-Khasnobis, Kanbur and Ostrom (2006), Maloney (1999), Perry et al. (2007) Portes and Schauffer (1993), Pradhan and van Soest (1995), Saavedra and Chong (1999), for surveys and discussions.

³ In recent volume, Guha-Khasnobis et al. (2006) also link informality to the degree of structuring of the organization.

⁴ Given differences in surveys, the cut-off point is not 5 employees in all countries. See our companion paper (Gasparini and Tornarolli, 2006) for details.

⁵ Gasparini and Tornarolli (2006) show that most results are robust to the change in the clas-sification of patrones.

⁶ See also Merrick (1976), Portes, Blitzner and Curtis (1986) and Saavedra and Chong (1999).

⁷ Information is provided throughout this paper and in the SEDLAC webpage.

⁸ That happens also in the city of Paramaribo (the only city included in the household survey of Suriname).

⁹ The employment structure does not dramatically change when restricting the analysis to only urban areas. The main differences are the higher share of workers in large firms and the public sector in urban areas, and the higher share of unskilled self-employed and, in particular, zero-income workers in rural areas.

¹⁰ To save space we show results for only seven countries. See Gasparini and Tornarolli (2006) for the complete analysis for all 21 countries.

¹¹ The problem is not symmetric, since we expect most entrepreneurs to be non-poor when "permanent" (e.g. yearly) income is measured.

¹² Presumably, the share would be even higher if we increased the cut-off point for firm size to define formality in the productive sense.

¹³ See Maloney (2004).

¹⁴ Under the legalistic view, that is true by definition. Under the productive view social protection is not precluded for informal workers but it is rarer.

¹⁵ See Fiess, Fugazza and Maloney (2002) and Maloney (2004).

¹⁶ Using multivariate co-integration techniques Fiess et al. (2002) find periods of co movements of relative earnings and sector size in Mexico and Brazil.

¹⁷ The analysis is carried out for the sample of urban workers aged 15 to 70 without tertiary education who are not in the patrones group.

¹⁸ The recovery that started around 2003 is not well captured in our sample.

¹⁹ Notice that when using the EPH Continua 2004 some results change. In particular, the characteristic effect becomes positive. Unfortunately it is difficult to trace the causes of that change, since the survey was modifed in various dimensions.

References

1. FIELDS, G. (1990). Labor market modelling and the urban informal sector: Theory and evidence. In OECD, The informal sector revisited. Paris.        [ Links ]

2. FIESS, N.; FUGAZZA, M. and MALONEY, W. (2002). "Exchange rate appreciations, labor market rigidities, and informality", Policy Research Working Paper 2771. Washington, The World Bank.        [ Links ]

3. GASPARINI, L. (2002). "Microeconometric decompositions of aggregate variables. An application to labor informality in Argentina", Applied Economics, 34:2257–2266.        [ Links ]

4. GASPARINI, L. and TORNAROLLI, L. (2006). "Labor informality in Latin America and the Caribbean: Patterns and trends from household survey microdata", Working paper, CEDLAS.        [ Links ]

5. GUHA-KHASNOBIS, B.; KANBUR, R. and OSTROM, E. (2006). "Beyond formality and informality", Introduction to Linking the formal and informal economy: Concepts and policies, forthcoming, EGDI-WIDER.        [ Links ]

6. HECKMAN, J.; LOCHNER, L. and TODD, P. (2003). "Fifty years of Mincer earnings regressions", NBER Working Paper Series 9732.        [ Links ]

7. MALONEY, W. (1999). "Does informality imply segmentation in urban labor markets? Evidence from sectoral transitions in Mexico", World Bank Economic Review 13.        [ Links ]

8. MALONEY, W. (2004). Informality revisited. World Development, 32(7):1159-1178.        [ Links ]

9. MERRICK, T. (1976). "Employment and earnings in the informal sector in Brazil: the case of Belo Horizonte", Journal of Developing Areas, 10(3):337-354.        [ Links ]

10. PERRY, G.; MALONEY, W.; ARIAS, O.; FAJNZYLBER, P.; MASON, A. and SAAVEDRA-CHADUVI, J. (2007). "Informality", Exit and exclusion. Washington DC, World Bank, Latin American and Caribbean Studies.        [ Links ]

11. PRADHAN, M. and VAN SOEST, A. (1995). "Formal and informal sector employment in urban areas of Bolivia", Labor Economics, 2:275-297.        [ Links ]

12. PORTES, R.; BLITZNER, S. and CURTIS, J. (1986). "The urban informal sector in Uruguay: its internal structure, characteristics and effects", World Development, 14(6):727-741.        [ Links ]

13. PORTES, R. and SCHAUFFLER, R. (1993). "Competing perspective on the Latin American informal sector", Population and Development Review, 19(1):33-60.        [ Links ]

14. SAAVEDRA, J. and CHONG, A. (1999). "Structural reform, institutions and earnings: Evidence from the formal and informal sectors in urban Peru", Journal of Development Studies, 35(4):95-116.        [ Links ]

15. SCHNEIDER, F. and ENSTE, E. (2000). "Shadow economies around the world: Size, causes, and consequences", IMF Working Papers 00/26, International Monetary Fund.        [ Links ]