Figure 4.1 Kernel Density Plots of Hourly Wages by Employment Sector
Sources: Angel-Urdinola and Tanabe, 2011; Alloush and others 2011.
SUMMARY: Building on the analysis of earning differentials, job mobility patterns, self-reported attitudes toward moving jobs, and self-rated satisfaction, this chapter provides a characterization of the quality of informal jobs. The existing evidence corroborates the view that segmentations exist across formal and informal jobs and that much of the observed informality is likely due to exclusion factors. Returns to human capital differ sharply, with estimated formality premiums ranging up to 50 percent for salaried workers. Informal salaried workers also report lower access to nonpension benefits, including annual, maternity, and official leave, and to training and skills upgrading. Moreover, informal jobs in MENA countries have longer durations than in comparator countries, thus dispelling the notion that informal jobs are mainly an entry point to the labor market to help employers and new entrants (youth) overcome asymmetric information constraints. In some countries, notably Egypt, a prominent margin of segmentation is along the public/private employment margin, as indicated, for example, by the extremely long duration of public jobs. The lack of information on earnings for the self-employed for most countries limits a full analysis of self-employed work. However, where these data are available, as in Lebanon, the evidence suggests that the self-employed have higher earnings than informal salaried workers, but overall lower earnings than formal salaried workers, while a small percentage of them earns more than both. The large majority of self-employed workers, who tend to be older and have lower education than both informal and formal salaried workers, do not have access to social security.
The previous chapters, in particular chapter 2, discussed the profile and characteristics of informal workers, pointing at common trends but also important heterogeneities. Overall, a large informal sector and the resulting coverage gap represent a suboptimal equilibrium that calls for policy action: Welfare can be improved by extending instruments to better manage risk. However, the ultimate drivers of informality are critical for identifying the appropriate policies. This chapter investigates the attributes of formal and informal jobs, including earnings, job mobility, access to leave and other benefits, and investment in training, as well as workers’ subjective perceptions. In addition to providing a full characterization of informal work, this evidence can help document empirically the extent to which informality is the result of workers’ voluntary choice or exclusion. This chapter first uses a variety of techniques, from simple comparisons of wage distributions to multivariate regression analysis, including the use of interesting and innovative controls for measured ability, to estimate whether a formality premium exists. Although persistent formality premiums are prima facie evidence that returns to skills are not equalized across formal and informal jobs for individuals of observed similar skills, wage differences cannot be taken alone as evidence that segmentations prevent arbitrage between the two sectors. Many factors can confound comparisons between earnings from different jobs but are unmeasured (for example, flexibility and independence). To gain a better understanding, this chapter complements the information on wage differentials with evidence on job mobility. High mobility is an indicator of integrated labor markets, where workers and enterprises, following market signals, search for matches between skills and vacancies. In countries with integrated labor markets, informal jobs help reduce asymmetric information for new labor market entrants, who, once they have proven their skills, would move rapidly to formal employment or self-employment. Thus workers that are observed working informally might indeed choose to do so, because in labor markets with easy transitions they could move out of informality if they so wished. In line with evidence from around the world, formal jobs have a longer duration than informal jobs. However, high persistence in informal, lower-paying jobs (for workers of skills comparable to those in the formal sector) is likely to indicate that many barriers, including labor regulations, prevent informal workers from moving to better jobs. In addition to analyzing data on actual job transitions and associated wage differentials, this chapter discusses newly collected data on self-reported attitudes toward job search, as well as the role of networks. Finally, the chapter discusses access to benefits (beyond old-age pensions) and training, and overall working conditions.
On average, wages in the formal sector are higher than in the informal sector. Wage data from surveys in Egypt (2006, and youth, 2009), Lebanon (2010), Morocco (available for youth only, 2010), Syria (2010), and the Republic of Yemen (2006) are used to quantify whether a premium to formal work exists beyond the returns to specific measured skills and experience. To address the high degree of heterogeneity among informal workers, the analysis is restricted to those individuals living in urban areas who are wage earners and working full time (between 30 and 60 hours a week). Data from Egypt, Morocco, and the Republic of Yemen include public and private sector workers, but data for Lebanon and Syria include only the private sector.1 Whenever possible, a distinction is made between private and public sector formal wages. Table 4.1 provides information about average wages by gender of formal and informal workers across different education levels and sectors. Informal sector wages are uniformly lower than formal sector wages, especially if compared with the formal private sector. Women systematically command lower wages than men. In Egypt, public sector wages are higher than private formal sector wages for low-to-medium-skill workers, thus providing incentives for a large share of the population to queue for public sector jobs. Kernel density plots provide a useful representation of differences in wages among formal and informal workers because they plot (smoothed) densities at all wage levels. The heavy solid line in figure 4.1 shows informal salaried workers’ wages. In Egypt and Lebanon (to a larger extent), the distribution of wages for workers in the informal sector is such that wages in the informal sector are always “‘within’” the distribution of public and formal sector wages; that is, they are stochastically dominated by them, confirming the existence of a formality wage premium throughout the wage distribution and lending prima facie support to the view that informality is an exclusion phenomenon. In other words, workers would prefer to move from the low-paid informal sector to the high-paid formal sector if possible. In the Republic of Yemen, however, this is not the case. Although in that country the average formal worker earns slightly more than the average informal worker, a small pool of workers at the high end of the wage distribution is able to earn higher wages in the private informal sector than would be earned in the public sector (seen at the right tail of the wage distribution). For this small pool of workers, working in the informal sector is probably a profitable choice.2
Figure 4.1 Kernel Density Plots of Hourly Wages by Employment Sector
Sources: Angel-Urdinola and Tanabe, 2011; Alloush and others 2011.
Table 4.1 Average Hourly Wages in Local Currency
Sources: Angel-Urdinola and Tanabe, 2011; Alloush and others, 2011.
A wage gap in the earnings of formal and informal workers persists, especially for women, even after controlling for a number of observable factors. Although kernel density plots are informative, they represent unconditional distributions; that is, they compare wages in the formal and informal sectors without controlling for observed characteristics that might affect the distribution (such as lower education and/or age). To overcome this limitation, standard Mincerian regressions were used. In their basic form, Mincerian regressions estimate wages as a function of a number of key observable characteristics, such as gender, age (nonlinear), and years of education. A binary variable for working in the formal sector is then added to this standard specification. The coefficient on this variable captures an estimate of the formality premium.3 This estimate varies significantly from country to country, highlighting the heterogeneity of the informality phenomenon in the region (figure 4.2). In Lebanon, the average estimated formality premium is 29 percent, in Egypt, 32 percent, and among Moroccan youth, as high as 53 percent.4 The formality premium is higher among female workers in all cases. In Egypt this phenomenon is particularly notable, where, on average, a female worker in the formal sector earns 1.5 times as much than an otherwise similar woman in the informal sector. Very low wages paid to female workers in the informal sector often explains why many women decide not to participate in the labor market at all, especially if the opportunities to find formal jobs are limited. This finding is quite relevant in a country such as Egypt, where female participation rates remain low by international standards.5 In both Lebanon and Syria, the formality premium is higher for workers with a low level of education, which suggests that unskilled informal workers might have lower bargaining power in wage setting (Alloush and others 2011).
Figure 4.2 Estimated Formality Premiums by Gender in Different Countries
Sources: Angel-Urdinola and Tanabe, 2011; Alloush and others 2011.
The differences in wages between the formal and the informal sector are explained by differences in the characteristics of formal and informal workers, as well as by differences in how the labor market rewards those characteristics. It is interesting to assess whether the observed differences in wages of formal and informal workers are mainly due to differences in the characteristics (endowments) of workers in the two groups, or to differences in the extent to which formal and informal jobs reward those endowments. Returns to observed skills can differ for many reasons between formal and informal jobs, depending on firm scale, complementarities with physical capital investment, and incentives to acquire experience and upgrade skills. Although many observed and unobserved variables affect workers’ distribution into formal and informal occupations, a simple Oaxaca decomposition provides an alternative insight into the likely drivers of observed wage differences.6 Figure 4.3 plots the result of this decomposition: The dark gray bars illustrate the share of the difference in wages explained by differences in observable characteristics between formal and informal workers, as attributed to characteristics (endowments), returns (how formal and informal jobs reward these characteristics), and an interaction effect. For example, part of the difference in wages can be explained by the fact that, on average, formal workers are older and more educated than informal workers (Angel-Urdinola and Tanabe 2011). These differences in “endowments” explain about one-third of the wage differential in Lebanon and about half in Egypt. One could argue that this component of the wage differential is “fair,” because the market should positively reward higher education attainment and experience.
Figure 4.3 Oaxaca Decomposition of the Differences in Wages
Source: Based on Angel-Urdinola and Tanabe 2011.
Note: For Egypt these are wages for urban workers in private and public sectors. For Lebanon, the wages are for private sector employees with no distinction between rural and urban areas.
The light gray bars in figure 4.3 illustrate the gap in returns to endowment due to being an informal rather than a formal worker, or the extent to which equal endowment/work is not rewarded by equal pay across the two sectors (for example, having a university education is rewarded less if the degree holder works as a taxi driver than if he works in the financial sector). These differences in rewards explain a large share of wage differences in Lebanon but a smaller share in Egypt. The higher the share of the wage differential that is explained by differences in “rewards,” the higher the chance that informality is an exclusion phenomenon (because workers would prefer to be rewarded more for a given observable characteristic, such as educational attainment, if they could). Last, the black bars represent the share of the difference in wages that arises because of the interaction between how the market rewards characteristics of formal versus informal workers and their differences in endowments.
Looking beyond individual characteristics, where workers are employed (small firms or large firms) affects the formality premium. However, the formality premium still persists after controlling for size. Where people work can potentially make an important difference in the earnings that they command. In particular, it is expected that informal wage workers in micro-enterprises (two to five employees) will earn significantly less than workers in larger firms, because smaller firms usually have lower productivity. In turn, firm productivity could account for much, if not all, of the formality premium across workers. Omitting this variable from an analysis of wage differentials could lead one to erroneously attribute the formality premium to informal employment when it might instead reflect firm characteristics. Once firm size in included in the regressions, the earnings gap between formal and informal workers still persists. The resulting formality premium is now estimated to be between 10 and 50 percent (figure 4.2). However, ample heterogeneity is observed in how firm size affects the estimated formality premium. In Syria, working in micro-firms does not change the formality premium significantly. This could be because of the nature of the survey sample, since survey participants all work in registered firms. Small micro- “formal” firms might share some of the key productivity attributes of larger firms so that controlling for firm size might not make a significant difference. In Egypt the formality premium continues to be significant but increases. In Lebanon, up to one-third of the informality premium is attributable to firm characteristics, with workers in micro-firms earning 23 percent less than workers in medium size firms (10 to 49 employees). Unlike in Egypt, informal workers in Lebanon are not primarily concentrated in micro-enterprises, because a large share of them work in larger firms. In this sense, returns to informal workers’ skills might differ in important respects because of the widely different productivity levels of the firms where they work (ranging from micro-informal to large formal). If uncaptured, this difference would be spuriously attributed to workers’ status (formal versus informal) rather than to firms’ characteristics and productivity.
Note that, when direct measures of individual ability are explicitly accounted for, the formality premium is virtually unchanged. Quality of education and skills often differ starkly across social strata and are correlated with the likelihood of working in the informal sector. Assignment to the informal sector is not random, and so it is important to control in some way for individual ability/skills to ensure that the attribution of the formality premium does not instead reflect other factors. In particular, unmeasured individual ability, which is likely negatively correlated with selection into the informal sector, could be spuriously driving these results, because years of education are unlikely to capture well actual quality of education and skills. To overcome this limitation in the interpretation of the estimation, the results of direct cognitive and noncognitive tests are included in the regression. These tests were administered in conjunction with the Lebanon labor force survey and the Syria matched employer-employee survey. Both surveys fielded a battery of ability tests to all participants as well as a questionnaire to self-assess personality and relational skills (box 4.1 describes the methodology and main features of these tests). The results of these tests are translated into scores and added as additional controls in the wage regressions. Although concerns have been expressed related to possible endogeneity of cognitive and noncognitive score measures, the inclusion of these scores is likely a good proxy for individual ability. Note that when cognitive and noncognitive scores are included in the estimation, they do not affect the estimate of the informality coefficient. Even after taking ability and personal characteristics into account, formality still has a 20 percent premium in Lebanon, and a 12 percent premium in Syria (tables 4.2 and 4.3).7
Box 4.1 Cognitive and Noncognitive Tests
Traditionally, studies have focused on measuring skills through educational attainment and experience, without measuring ability directly. Cognitive and noncognitive tests can be good proxies for individual ability. In fact, an individual’s skill set can be defined as a combination of his or her cognitive, noncognitive, and technical skills (see table below). In turn, given that human capital is multidimensional, both cognitive and noncognitive skills can serve as good predictors of economic outcomes such as income and wealth (Bowles and others 2001; Cawley and others 2001; Hansen and others 2004; Heckman and others 2006). Two innovative labor market surveys conducted in the MENA region (in Syria and Lebanon) have measured cognitive and noncognitive skills of labor force participants and have produced interesting results.
Relationship between technical, cognitive, and noncognitive skills |
||
Type of skill |
What it measures |
Measurable concepts |
Cognitive |
Ability |
Verbal, math, logic reasoning |
Noncognitive |
Ability, interaction |
Personality, communication |
Technical |
Knowledge, experience |
Breadth, depth of field of knowledge |
The cognitive test used “Raven’s Progressive Matrices,” a nonverbal test in which individuals have to identify the missing piece of a particular pattern among multiple choices. Respondents are given five minutes to answer as many as they can of the 12 matrices included in the test, which become progressively more difficult and require greater cognitive capacity. This test is independent of language, reading, or writing skills and focuses on measuring observation skills, analytical ability, and intellectual capacity. A score (out of 12) is calculated for each respondent based on the number of correct answers completed in the allocated time.
The noncognitive test consists of a list of statements describing personal behaviors and characteristics corresponding to the five core dimensions of personality used in psychology literature to classify human personality (referred to as the “Big Five”):
• Conscientiousness: tendency to be organized, responsible, and hardworking
• Emotional stability: tendency to be predictable and consistent in emotional -reactions
• Agreeableness: tendency to act in a cooperative and unselfish manner
• Extraversion: tendency to orient one’s energies toward the outer world of people and things
• Openness to experience: tendency to be open to new cultural or intellectual experiences.
These traits are considered to be relatively stable throughout one’s lifetime. The noncognitive test used in the Syria and Lebanon surveys, an adaption of the Goldberg test, asks respondents to rank the applicability of a total of 15 traits/behaviors to themselves (three of each corresponding to one of the “Big Five”) from a scale of 1 to 7. Accordingly, a score for each of the “Big Five” traits is calculated.
Main observed patterns: Cognitive skills are positively correlated with educational attainment, but more strongly so in Syria than in Lebanon. A positive difference in the average cognitive score between formal and informal workers is observed, which is significant in Lebanon but not in Syria.
Note: In Lebanon, 9.4 percent of the total population surveyed did not take the test. The highest nonresponse rate was for the illiterate (50 percent). Note, however, that only 10 workers in Lebanon reported themselves as illiterate. The results in Lebanon are in line with the “Flynn effect,” which describes an intergenerational increase in scores.
Source: Alloush and others 2011.
Unobserved firm characteristics can also affect earnings differentials between formal and informal occupations. In most household surveys, information on the types of establishments where individuals work is not available, with the exception of information on firm size. For example, characteristics such as whether an individual firm is registered with tax authorities or the average level of education of its workforce are not known. These characteristics might matter and drive the ability of firms to reward skills. If unmeasured, they can also confound the size and attribution of estimated wage gaps between formal and informal workers. The Syrian data, which allow for these controls, are collected as part of a matched employees-employer survey, so that workers can unequivocally be mapped to their firms and firm-specific effects controlled for. Even after controlling for individual cognitive and noncognitive ability and for firm fixed effects, the formality premium is unaffected; workers who are hired as informal command a lower wage.
Table 4.2 Formality Premium, Lebanon: Ordinary Least-Squares (OLS) Estimates for Private Sector Formal and Informal Workers
Source: Annex tables 4.1 and 4.2 from Alloush and others 2011.
***Significant at 1 percent.
The wage analysis developed in this section suggests that informal workers command uniformly lower pay than formal workers with similar characteristics. One could argue that, given the overall worse terms of informal jobs (that is, lower duration and limited or no access to leave and pension benefits), informal jobs would have to pay higher wages to attract workers. However, in economies such as those in the MENA region where job creation is rationed, lower pay and worse conditions in the informal sector might be seen by workers as a preferable alternative to unemployment, especially for workers with low skills and education.
Although wage comparisons can be informative, differences in earnings alone cannot unequivocally prove segmentations in labor markets. Earnings differences between the formal and informal sector might reflect a range of unobservable job characteristics, including flexible hours, value of training, and value (or lack) of social security benefits, all of which can affect the salary wedge between observationally similar jobs (Maloney 1999, 2004). Moreover, selection into formal or informal jobs might be driven by unobservable workers’ skills and ability, which might in turn affect earnings. The concern that this might lead to estimation bias is mitigated in wage regressions for Lebanon and Syria, due to inclusion of direct test score measures of cognitive and noncognitive ability. These controls did not affect the size of the formality premium, hence suggesting that “informality” status did not, to a large extent, spuriously reflect ability. However, other unobservables might still persist. To complete and complement this analysis, patterns of labor mobility in and out of informality and wage differentials for individuals moving from informal to formal jobs and vice versa were examined using actual transitions from the Egypt panels as well as data on self-reported jobs search.
Table 4.3 Formality Premium, Syrian Arab Republic: Ordinary Least-Squares (OLS) Estimates for Private Sector Formal and Informal Workers
Source: Annex tables 4.1 and 4.2 from Alloush and others 2011.
Workers’ mobility patterns in MENA provide evidence that segmentations exist in the region’s labor markets. Mobility patterns can provide important insights on the extent to which labor markets are segmented or integrated. A good level of mobility and integration in labor markets indicates that workers and employers have the chance to find productive matches between labor demand and supply of skills. In this section, longitudinal data from two separate sources for Egypt are used to study mobility patterns between the formal and informal sectors: the ELMPS, which surveyed individuals in 1998 and 2006, and the HIECS panel, which surveyed individuals in 2008 and 2009. Over the eight-year span of the two ELMPS surveys, persistence in formal or informal status was high (figure 4.4): 46 percent of informal salaried workers in 1998 continued to be informal salaried in 2006. By 2006, only 19 percent had transitioned to formal employment, either in the private or in the public sector. In the second panel survey, 65 percent of informal salaried workers in 2008 remained so in 2009. Upward mobility was minimal, with only 9 percent of informal salaried workers transitioning toward jobs with social security coverage (public or private).
Only a minority of all informal workers in 2008 became unemployed in 2009 despite the financial crisis. It is interesting to note that when the financial crisis struck Egypt in 2008, formal private sector employment shrank substantively (35 percent) with a large share of workers transitioning to informal salaried jobs in the private sector (about 28 percent). In a crisis year, many of these transitions are likely involuntary and might have been the result of termination of temporary contracts (initially with benefits) and their subsequent conversion into informal jobs, as well as of firm closures after which workers rapidly found informal subsistence employment. It is notable that transitions into unemployment from formal jobs are almost nonexistent. Lacking effective unemployment insurance, informal work (which commands average lower wages) represents a safety net type of employment.8 Table 4.4 depicts the 1998–2006 transition matrix, and table 4.5 depicts the 2008–2009 transition matrix. The yearly transition probabilities in 2008 translate into an estimated job duration of about three years for informal salaried workers,9 above that observed in comparator countries in other regions, such as Mexico (where movement in and out of salaried informal jobs is high and average job duration is estimated at about two years).10 Bosch and Maloney (2010) offer a sophisticated approach to reading transition matrices, which adjusts for the extent to which vacancies open up and workers leave in different sectors (for example, formal private, public, or private informal).11 Applying their methodology to the Egypt context confirms that the public sector -continued to be an engine of employment growth in the crisis, second only to informal salaried work, absorbing 28 percent of formal private sector workers. Overall, persistence in informal status (including, in addition to informal wage workers, the self-employed, many employers, and unpaid workers) is extremely high, at about 90 percent.
Table 4.4 Labor Market Transitions in Egypt (1998–2006)
Source: Silva and others 2011.
Note: Percentages are reported in bold. The shaded area highlights transition among employed workers. Sample: Male workers, 15–64 years old in 2006.
Table 4.5 Labor Market Transitions in Egypt (2008–2009)
Source: Silva and others 2011.
Note: Percentages are reported in bold. The shaded area highlights transition among employed workers. Sample: male workers, 15–64 years old.
Panel data also allow for observing earning changes along transitions. On average, when a worker moves from an informal to a formal job, his or her net earnings increase by about 24 percent and even more if benefits are accounted for (table 4.6). Note that by tracking earnings changes for the same individual across the informal-formal transition, the concern that higher earnings in formal jobs might reflect unobserved workers’ characteristics is eliminated.
Table 4.6 Average Level of Earnings Increase When Moving from Formality to Informality
Source: Silva and others 2010 using Egypt HIECS 2008/09.
Note: Line (1) presents the mean percentage change in real earnings between 2008 and 2009. For this calculation, nonadjusted earnings (take-home pay), which exclude social security benefits, were used. Sample: Male, 15–64 years old, employed in 2008 and 2009.
Patterns of social security contribution densities from administrative data confirm the limited mobility between formal and informal jobs. Social security contributions offer an alternative data source to understand job duration patterns. Contribution density is defined as the share of time an individual contributes to the pension system during his or her active life period. Individual contribution densities are the longitudinal companion of the coverage rate, which measures the share of employed who are contributing to social security at a certain moment in time. Figure 4.5 illustrates the age-specific contribution density patterns toward the mandatory social insurance scheme administered by the Social Security Corporation (SSC) in Jordan. The SSC dataset used for the analysis includes everyone who contributed to or received benefits from the SSC between 1980 and 2010. The pattern indicates that in Jordan, individuals tend to complete either long contributory spells or long noncontributory spells (that is, they stay in or out of the system for long periods of time). As illustrated by the figure, adult individuals between ages 25 and 55 are expected to stay in the system for about 30 years (which is generally the time needed for full retirement). Across all individuals and cohorts, contribution densities are low for youth, but they increase rapidly as individuals reach adulthood, leave school, and establish themselves in formal employment. As expected, contribution rates drop significantly when individuals are in their late fifties, because early and statutory retirement rules drive down contributions (a pattern that is observed in most countries).12
Figure 4.5 Contribution Density and the Distribution of Contributory and Noncontributory Spells in Jordan
Source: Calculations using administrative datasets from the SSC of Jordan.
Contribution density patterns show the coexistence of two different groups of workers: some with an established contributory status, and some who hardly ever contribute. The average length of a continuous contributory spell in Jordan is longer than four years, with little difference between women and men (49 and 53 months, respectively), and with 40 percent of all contributors contributing for at least three years (figure 4.6). Only 16 percent of all contributors (mainly new entrants to the labor market) have a contributory period shorter than six months. The distribution of the length of noncontributory spells signals the existence of a group consistently out of coverage. For example, once workers who are enrolled in a contributory job (that is, registered with SSC) at some point in their career start a noncontributory spell, 69 percent of them stay out of coverage for a minimum of three years and an average of approximately six years (94 months).
Figure 4.6 Distribution of the Length of Social Insurance Contributory and Noncontributory Spells in Jordan and Chile
Source: Administrative datasets from the SSC of Jordan and the Chilean Pension Supervisory Agency.
Changes in contributory status in Jordan are significantly less frequent than in a comparator country such as Chile. The third and fourth panels of figure 4.6 illustrate the transition patterns in and out of contributory status based on monthly administrative data on social insurance contributions from Chile. In the case of Chile, the most frequent length category for both being in or out of social insurance contributory status is less than six months: 42 percent of those contributing to the social insurance system in Chile are likely to discontinue contributions within six months, and 37 percent of those registered with social security will restart contributions within six months. In contrast, the Jordanian contributory pattern is extremely skewed: 40 percent of contributors are likely to complete a spell longer than three years, and 69 percent of those once registered as contributors, but currently not contributing, will be outside of the system longer than three years. These results confirm the existence of an insider/outsider structure in MENA labor markets that is notably different from other parts of the world.13
This section aims to provide as assessment about the quality of informal employment. To begin with, available evidence indicates higher levels of job dissatisfaction among informal workers.14 Table 4.7 reports responses to the question “Would you like to change your job?” for Lebanon, Egypt, and Syria. In Lebanon and Egypt, informal workers were significantly more likely to want to change jobs or were searching for a job. A higher level of job dissatisfaction persists in Egypt and Lebanon even after many individual characteristics are accounted for, including age, education, and pay (with the expectation that workers with higher pay are less likely to want to change jobs). In Syria, where differences between formal and informal workers are less pronounced in the surveyed sample, formal workers actually appear to be marginally more willing to change jobs than informal workers, but this difference is insignificant when individual characteristics are accounted for (Alloush and others 2011).
Table 4.7 Individuals Who Want to Change Jobs
Sources: Egypt youth survey 2010, Lebanon Labor Force Survey 2011, Syria Employee-Employer matched module 2011.
In general, informal jobs entail worse working conditions. Table 4.8 presents descriptive statistics related to self-perception of employment quality for formal and informal workers in Lebanon and Syria. Results indicate that, in many dimensions, formal workers experience better job quality, as measured by higher earnings, having a written contract, longer employment tenure, and higher chances of choosing when to work overtime (and be paid for such work). A noticeable feature is that informal jobs have a significantly shorter tenure than formal private sector jobs, which suggest that informal jobs provide firms with the needed flexibility in hiring and firing (especially relevant in countries where labor regulations are particularly restrictive).15 Little difference is seen in the amount of work time (hours per day and days per week) between formal and informal workers in Lebanon and Syria. However, access to annual or other leave is much lower for informal workers (as discussed below), so that they may end up working a significantly greater number of days per year than formal workers.
Table 4.8 Differences in Job Quality between Formal and Informal Jobs
Source: Alloush and others 2011. — = Not available.
As expected, informal workers enjoy fewer benefits than formal workers. In Lebanon, 56 percent of informal workers do not have any annual leave versus 13 percent of formal workers. Similarly, informal workers have significantly less access to benefits of family, sick, or maternity leave (figure 4.7). Although no significant difference is seen between formal and informal workers in terms of their access to benefits, such as provision of a firm nursery or a lunch allowance, large differences emerge for benefits that are fairly common in the formal sector. The evidence from Egypt suggests that informal jobs require, on average, shorter commutes, indicating that they are more likely to be “local.” Still, transport expenses (which are not factored in the reported wage) can be substantial. For example, in Lebanon, daily transportation costs are anecdotally reported to be somewhat higher (at about 0.8–1.2 times the average hourly wage).
Figure 4.7 Lack of Access to Job Benefits between Informal and Formal Sectors (% of Workers without Access)
Furthermore, informal workers are more likely to be working in unsafe conditions. Data from the Morocco youth survey provide additional information on job safety and dignity at the job place for formal and informal workers (table 4.9). Results indicate that informal workers are systematically worse off: more likely to be physically or verbally abused, more likely work longer period of time for less pay, more likely to be harassed at the workplace, and more likely to feel they are exposed to exhausting workloads. Finally, informal workers are much more likely to feel that their job is not challenging and/or not interesting, which supports the view that informal jobs might entail waste of talents.
Informal workers appear to be less likely to receive training than formal workers, even after controlling for individual characteristics. Evidence in chapter 3 showed that firms with a higher share of informal workers are less likely to provide training. In the Syria sample, and after controlling for individual characteristics such as education and cognitive and noncognitive scores, informal workers are found to be significantly less likely to receive training. However, this association loses significance once firm fixed effects are included, indicating that the correlation identified cross-sectionally might all be due to differences in firms’ types. Among Moroccan workers who did not report benefiting from technical training, informal workers were significantly more likely to report that no training is accessible to them, and that they have little information on what training to take. Moreover, they were almost twice as likely as formal workers to report that they could not afford the fees.
For some self-employed, informality may be a matter of choice. As mentioned in chapter 2, self-employed workers are somewhat different from other informal workers (mainly wage earners). Results from Egypt and Lebanon indicate that self-employed workers are mainly male (72 percent in Egypt and 85 percent in Lebanon), belong predominantly to the age group 35–54 (about 50 percent in both countries), and have attained at most a secondary education (93 percent in Egypt and 87 percent in Lebanon). On average, self-employed workers are older and less educated than informal wage earners (see chapter 2). Nevertheless, as illustrated in figure 4.8, self-employed workers in Lebanon generally earn, on average, 32 percent higher wages than informal wage earners (although they still earn lower wages than formal wage earners in net terms). This holds true all along the wage distribution. Indeed, a small pool of self-employed is seen at the top end of the wage distribution (see figure 4.8) earning even higher wages than formal wage earners. These results would be consistent with self-employed workers preferring their employment status versus that of informal wage earners and even versus formal wage earners (that is, those at the very high end of the wage distribution).
Figure 4.8 Net Wages by Employment Status (Lebanon, 2010)
Self-employed workers enjoy being independent, although they would like to be covered by a social security scheme. Table 4.10 provides information on self-stated preferences and perceptions of being self-employed versus being a wage earner. Results indicate that 70 to 75 percent of all workers would like to be wage earners mainly for job security reasons (that is, having social security and greater employment security). Note that only 7 to 15 percent of all workers interviewed claimed that better opportunities and/or greater job satisfaction would be a reason for wanting to be a wage earner. At the same time, the majority of workers interviewed claimed that greater independence and higher earnings are the main two reasons for wanting to be self-employed. Although enjoying the benefits of independence and higher earnings, the majority of self-employed workers (53 percent) would like to be registered in a social security scheme.
Table 4.10 Preferences for Wage Earning versus Self-Employment (Lebanon, 2010)
By and large, informal workers would like to have access to social security benefits and are willing to pay for them. In Lebanon and Syria, 70 percent of informal workers indicate they would like to have social security benefits and are willing to give up on average 5.8 percent of their salary. This compares to a contribution rate of 2 percent for NSSF, which provides health insurance coverage with an additional 9 percent paid by the employer. Of the 30 percent of informal salaried workers who are not interested in being registered in NSSF, 37 percent say that having private health insurance from a different source is the main reason for their lack of interest (table 4.11). Although only limited inferences can be made because of the small sample size (these questions were asked only of informal workers who did not want to contribute to social security), it should be noted that both the poor perceived quality of the public service and lack of information are reported as leading reasons for opting out of the system. Similarly in Morocco, lack of information and how the system works are the top reasons for not wanting social security (table 4.11). Although no strong correlations emerge, generally males and the better educated are more likely to know how social security works. It is interesting that a high share of noncontributing workers in Syria report that they are not concerned about retirement. It is not clear whether this is attributable to myopia or to the existence of strong informal social safety nets that would substitute for retirement savings. Also in Syria, half of the noncontributing respondents report that “not knowing how it works” is an important reason for not contributing. This suggests that improving financial literacy and communication about pension systems’ roles and rules might be an important component of effective risk protection.
Table 4.11 Reasons Why Informal Workers Do Not Want to Contribute to Social Security
Source: Salaried workers (Lebanon Labor Force Survey 2011, Syria 2010, Morocco Household and Youth Survey 2010).
Networks constitute an important mechanism for workers to find formal and informal employment. The large majority of workers in the MENA region report having found jobs through personal connections. This is particularly the case for informal workers. For example, 85 percent and 74 percent of informal workers in Lebanon and Syria, respectively, report finding jobs through personal contacts (table 4.12). Overall, this evidence points at a very limited role for private and public institutional mechanisms to support broad matching of skills. Employers seem to fulfill their labor demand through mechanisms that bypass market signals of skills or quality of education, either because these signals might not be informative or because other factors, such as trust, are dominant and unlikely to be picked up by market matching mechanisms. To the extent that personal contacts and networks develop within, rather than across, specific social and economic strata, these job search methods are likely to perpetuate segmentations within labor markets, particularly across formal and informal jobs, thus preventing the best matches and highest returns to skills from being realized.
Table 4.12 How People Find Jobs
percent |
||
Lebanon |
Formal |
Informal |
Personal contacts |
77.3 |
84.5 |
Other |
10.4 |
6.3 |
Advertisement |
7.3 |
6.6 |
Online job search |
4.2 |
1.7 |
Recruitment agency |
0.66 |
0.99 |
National employment agency |
0.22 |
0.0 |
Syrian Arab Republic |
||
Personal connections |
62.3 |
74.1 |
Direct contact with company |
21.5 |
16.3 |
Advertisement (newspaper) |
6.1 |
6.6 |
Other |
4.6 |
2.2 |
National employment agency |
3.8 |
0.2 |
Private employment agency |
1 |
0.5 |
Online advertisement |
0.6 |
0.16 |
Source: Lebanon Labor Force Survey 2010 and Syria 2010.
Evidence from Lebanon and Syria stresses the importance of family relationships for finding informal employment. Table 4.13 presents the self-reported relationship to the firm owner for employed individuals in Lebanon and Syria. Results indicate that informal workers tend to be related to the firm owner (that is, be family or friend) more often than are formal workers. Among informal workers, this phenomenon is particularly present among family-owned companies and almost absent in large enterprises. The opposite is observed among formal workers, for whom the importance of networks does not change substantially across firm size. Consistent with these findings, in Lebanon and Egypt, informal jobs are less likely to require significant commutes because networks are generally built within the community of residence (for example, in Egypt, youth are less likely to have a lengthy commute if working informally).
Networks seem to be an important determinant for finding “formal” employment. Regression analysis indicates that in Morocco, having a father with a formal job significantly increases one’s chances of having a formal job, even after controlling for own and parental education, as well as for whether one’s father speaks French (used here as a proxy for social status; table 4.14). Moreover, the probability of being informal decreases significantly if another member of the household holds a formal job. In Morocco, the formality definition captures both pension and health insurance coverage, because the benefits are bundled. The literature on informality often makes the assumption that informality is higher because, with bundled and family coverage, incentives to participate in the system for family members of a formal worker are very limited. However, the evidence for Morocco points at the opposite: Formal jobs appear as a privilege of the few, and family connections are successfully leveraged to access them.
Table 4.14 Determinants of Informality: Networks in Morocco
Source: Gatti and others 2011.
***Significant at 1 percent.
Results indicate that the average worker in the formal sector earns higher wages compared to the average worker in the informal sector. Controlling for other factors such as education, experience, and gender, workers in the formal sector earn higher wages than otherwise similar workers in the informal sector. This “formality premium” is generally high, ranging from 10 percent in Syria to up to 53 percent among youth in Morocco. The formality premium is higher among female workers in all cases. On average, a female worker in the formal sector earns 1.5 times as much as an otherwise similar worker in the informal sector. Very low wages paid to female workers in the informal sector often explain why many women decide not to participate in the labor market at all, especially if the opportunities to find formal jobs are limited.
The observed differences in wages between the informal and formal can be attributed to differences in the characteristics of formal and informal workers (endowments), as well as to differences in how the labor market rewards those characteristics (returns). About one-third of the wage differential in countries such as Egypt and Lebanon can be explained by the fact that, on average, formal workers are older and more educated than informal workers. In Lebanon, the wage differential is explained mostly by the labor market treating similar skills of formal and informal workers differently.
Where workers are employed (small versus large firms) and their skills and ability can both potentially affect the observed formality premium, but the formality premium is found to persist even when these additional factors are accounted for. A priori there might be important differences in workers’ wages across small and large firms because of differences in firms’ productivity. Omitting this factor could be problematic in estimating the formality premium, especially when informal workers for the most part find employment in micro-firms. When firm size is controlled for, workers in micro-firms in Lebanon earn 23 percent less than workers in medium size firms (10 to 49 employees) and about one-third of the formality premium is attributable to firm characteristics. However, a formality premium of about 20 percent still persists. In Egypt and Syria, the formality premium is virtually unaffected by inclusion of controls for where workers are employed. Although firm size might be an imperfect measure of firm productivity, even using firm fixed-effects estimation (Syria) confirms that informal workers earn less than formal workers, independent of where they work.
Individual ability does not seem to be an additional important determinant of the formality premium. Quality of education and skills often differ starkly across social strata and are correlated with the likelihood of working in the informal sector. Assignment to the informal sector is not random, and so it may be important to control for individual ability and skills to ensure that the attribution of the formality premium does not instead reflect other factors. Data from Lebanon and Syria allow proxying for individual ability through results of direct cognitive and noncognitive tests. When cognitive and noncognitive scores are included in the estimation (and controlling for other factors), the observed informality premium remains largely unchanged. These results suggest that informal workers are generally worse-off, because individual ability is better rewarded by formal jobs than by informal jobs.
Limited mobility between formal and informal employment points at the existence of segmentations in the region’s labor markets. The fact that measured rewards (wages) to skills are consistently different across formal and informal workers cannot be taken alone as an indicator of labor market segmentation, because many other features of jobs are usually unmeasured. Mobility patterns provide important insights into the extent to which labor markets are segmented or integrated. In this chapter longitudinal data from two separate sources for Egypt are used to study mobility patterns between the formal and informal sectors. Results indicate that between 1998 and 2006, persistence in formal or informal employment status was very high and upward mobility was minimal, with only 9 percent of informal salaried workers transitioning into formal jobs. Transition from informal to formal jobs was associated with an average increase in earnings of 24 percent, confirming the hypothesis that formal workers are generally better off.
Beyond earnings, workers in the informal sector are generally disadvantaged in many other dimensions. Informal workers are more likely to be unsatisfied with their job relative to formal workers. This is explained by the fact that informal jobs entail worse working conditions. Results in this chapter indicate that, in many dimensions, formal workers display better job quality indicators, such as higher earnings, having a written contract, longer employment tenure, and higher chances to choose when to work overtime (and be paid for such work). Furthermore, informal workers enjoy fewer benefits than formal workers, such as paid leave and maternity leave. On the positive side, evidence from Egypt suggests that informal jobs require, on average, shorter commutes, which indicates that they are more likely to be “local.”
Finally, results in this chapter indicate that individual networks are important to explain the likelihood that individuals find formal and informal employment. The large majority of workers in the MENA region report having found jobs through personal connections. This is particularly the case for informal workers. For example, 85 percent and 74 percent of informal workers in Lebanon and Syria, respectively, report finding jobs through personal contacts. Informal workers tend to be somewhat more related to the firm owner (that is, a family or friend) than formal workers. Networks are also important for individuals to find formal employment. Evidence from Morocco indicates that, controlling for other factors, having a father with a formal job significantly increases one’s chances of having a formal job. In Morocco, the evidence also suggests that, everything else equal, workers are more likely to find a formal jobs if someone else in the family already has a formal job. In countries where pension and health insurance are bundled and where spouses and dependents can rely on the worker’s coverage, this evidence suggests that formality is likely driven by the use of network to secure the “good jobs.”
Annex table 4A.1 Estimating the Formality Premium in Lebanon
Source: Alloush and others 2011.
Note: Estimation with OLS, robust standard errors.
Annex table 4A.2 Estimating the Formality Premium in Syria
Source: Alloush and others 2011.
Note: Estimation with OLS, robust standard errors.
Source: Angel-Urdinola and Tanabe 2011, using the 2006 Egypt ELMPs.
Note: Omitted variables: Education: Primary; Region: Cairo; Sector of employment: Primary sector; Primary sector (agriculture); Secondary sector (manufacturing and construction); Tertiary sector (wholesale, transport, services); Public administration and social services (including education and health).
1. To appropriately control for hours worked, the analysis is based on hourly wages in current local currency (that is, wage rates).
2. Formal workers are required to pay taxes, and so some high-earning workers have an incentive to evade taxes and earn higher net wages, especially if they have access to health insurance through a family member and/or to a pension.
3. Because of the likely presence of omitted variables, the estimate of this coefficient is likely biased. In particular, if (unobservable) ability is positively correlated with formality, this coefficient is likely biased upward. However, when direct measures of individuals’ cognitive and noncognitive ability are introduced for Lebanon and Syria, the size of the coefficient remains unchanged.
4. Herrera and Badr (2011) estimate a formality premium for Egypt of 13.2 percent using a different definition of informality (workers not having a contract and not contributing to social security). Using the same definition, very similar results are found in this analysis and are available upon request.
5. Controls are year of education, age, age squared, and a gender dummy.
6. See, for example, Arias and Khamis (2007) and Pianto and others (2009) for approaches that correct for selectivity bias.
7. Two main sources of concern are found when it comes to endogeneity. First, ability influences education outcomes, which in turn affect measured ability. This would tend to overestimate the coefficients on skills and underestimate that of schooling. Second, reverse causality may exist between labor outcomes and socioemotional skills. Personal characteristics can influence labor market outcomes, but labor market outcomes can also influence the way one feels about oneself. For example, having a low paying job can lower one’s self-esteem and emotional stability (Heckman and others 2006).
8. Note that the two rounds of surveys are likely to capture the early crisis impact because they were fielded in April 2008 and March 2009. Unfortunately, because of the lack of information on type of contracts (temporary vs. permanent), it is not possible to further disentangle mobility across formal/informal states.
9. Following Maloney (1999), job duration can be calculated as 1/(1−Pii) where Pii is the individual probability of not leaving sector/state i. See Silva and others (2011) for a detailed discussion.
10. See Maloney (1999).
11. Bosch and Maloney (2010) compute an adjusted propensity to transit from sector i to sector j (C-stats). This statistic weighs the raw transition probabilities by the rates of leaving the origin sector and the relative opening in the receiving sector compared to the economy as a whole.
12. Note that contribution density here, in contrast to the coverage rate, is not defined in relation to the labor force, but rather to the population registered with the social insurance administration. This population includes (at least temporarily) inactive people who exited the labor force after a period of past contributions; therefore the contribution density defined in this manner is naturally smaller than the coverage rate for the identical population.
13. The Chile and Jordan datasets differ somewhat, which may confound the interpretation of these results. For example, the underlying Jordanian administrative data cover a period of 363 months, whereas the Chilean data cover only 288 months; that is, it might be more likely to observe long (in excess of 36 months) contributory and noncontributory spells in Jordan than in Chile because of data differences. However, it is unlikely that the radical difference in the shape of the distributions can be explained solely by this sample difference.
14. In this chapter “job quality” is referred to in a way that is akin to the “decent work” definition. Decent work involves opportunities for work that is productive and delivers a fair income, security in the workplace and social protection for families, better prospects for personal development and social integration, freedom for people to express their concerns, organize and participate in the decisions that affect their lives, and equality of opportunity and treatment for all women and men. The private sector and innovation literature usually refers to job quality as “high value-added jobs.” The discussion of which policy interventions can promote the development of high value-added production (for example, moving from textiles to electronics) is complementary to the treatment of informality herein. However, it is beyond the scope of this report.
15. An interesting exception is in Lebanon, where informal workers report having higher confidence in the stability of their jobs than do formal workers.
Alloush, M., C. Chartouni, R. Gatti, and J. Silva. 2011. “Informality and Exclusion: Evidence from Lebanon and Syria.” Mimeo. Washington, DC: World Bank.
Angel-Urdinola, D., and K. Tanabe. 2011. “Micro-determinants of Informal Employment in the Middle East and North Africa Region.” Mimeo. Washington, DC: World Bank.
Arias, O., and M. Khamis. 2007. “Comparative Advantage and Informal Employment.” Mimeo. Washington, DC: World Bank.
Bosch, M., and W. Maloney. 2010. “Comparative Analysis of Labor Market Dynamics Using Markov Processes: An Application to Informality.” Labour Economics 17: 621–631.
Bowles, S., H. Gintis, and M. Osborne. 2001. “The Determinants of Earnings: A Behavioral Approach.” Journal of Economic Literature 39 (4): 1137–1176.
Cawley, J., J.J. Heckman, and E. J. Vytlacil. 2001. “Three Observations on Wages and Measured Cognitive Ability. Labour Economics 8 (4): 419–442.
Gatti, R., M. Ivanic, and J. Silva. 2011. “It’s Who You Know: Networks and the Chance of Getting a Formal Job.” Mimeo. Washington, DC: World Bank.
Hansen, K. T., J. J. Heckman, and K. J. Mullen. 2004. “The Effect of Schooling and Ability on Achievement Test Scores.” Journal of Econometrics 121 (1–2): 39–98.
Heckman, J. J., J. Stixrud, and S. Urzua. 2006. “The Effects of Cognitive and Noncognitive Abilities on Labor Market Outcomes and Social Behavior.” Working Paper 12006. Cambridge, MA: National Bureau of Economic Research.
Herrera, S., and K. Badr. 2011. “Why Does the Productivity of Education Vary across Individuals in Egypt? Firm Size, Gender and Access to Technology as Sources of Heterogeneity in Returns to Education.” Mimeo. Washington, DC: World Bank.
Maloney, W. F. 1999. “Does Informality Imply Segmentation in Urban Labor Markets? Evidence from Sectoral Transitions in Mexico.” World Bank Economic Review 13 (2): 275–302.
________.2004. “Informality Revisited.” World Development 32 (7): 1159–1178.
Pianto, D. M., M. Tannuri-Pianto, and O. Arias. 2009. “Informal Employment in Bolivia: A Lost Proposition?” Mimeo. Washington, DC: World Bank.
Silva, J., M. Benyagoub, and G. Wallner. 2010. “Informality, Firm Performance and Labor Market Outcomes: Evidence from Matched Employer-Employee Data for Morocco.” Mimeo. Washington, DC: World Bank.
Silva, J., Y. Dukhan, and D. Marotta. 2011. “Determinants and Earnings Consequences of Mobility In and Out of Informality in Egypt.” Mimeo. Washington, DC: World Bank.
