Notion of labor informality

Labor informality is an intrinsic problem of labor markets in developing countries (^{Fields, 2011}). Specifically, in Colombia, there is a lack of labor demand combined with an excess of labor force, the consequence of rural-urban migration, displacement, and demographic growth. The informality has resulted in small businesses, employment conditions which violate legislation, scavenging activities, and self-employment with very low levels of productivity becoming short-term solutions for generating sufficient income to subsist (^{Sanchez, 2020}).

There are several approaches for measuring labor informality, using various units of measurement and often arbitrary criteria. This article adopts the approach proposed by the International Conference of Labor Statisticians (^{International Labor Organization [OIT in Spanish], 2003}), which distinguishes between “Informal Employment” and “Employment in the Informal Sector.” The former relates to the type of employment relationship and compliance with labor regulations (institutionalist perspective), while the latter concerns economic activities and their accumulation capacity (structuralist perspective). Estimates were made according to each perspective. However, for reasons of space and given the number of estimates to be presented for each zone of Bogotá and its MA, all findings and calculations presented relate to Informal Employment, consisting of the following:

Levels of labor informality

In Bogotá and its surrounding area, 41.1% of employment is informal. Figure 1 shows informality levels by zone; it should be noted that this figure illustrates the percentage of informal employment relative to the total number of workers living in each zone, given that many workers commonly do not work in the same locality in which they live.

Source: Compiled by the authors based on EM17 (^{DANE, 2020}).

Figure 1 Informality rates by zone of Bogotá and its MA 

Gaps due to informality: methodology and findings

Labor informality is not only a problem linked to the vulnerability in employment continuity and security and the lack of protection resulting from precariousness in labor relations; it is also a situation that constricts living conditions due to the lower levels of income derived from these activities.

Table 3 shows the informality rate and the observed and estimated gaps in the average hourly labor income of formal and informal workers. In general terms, what is observed in income levels and their inequality is verified: the areas with the highest income and heterogeneity of strata present the highest gaps in remuneration observed due to informality.

This article estimates the individual gaps as proposed by ^{Jenkins (1994}), as well as those estimated for some Latin American countries in the studies by ^{Beccaria and Groisman (2008}) and ^{Maurizio (2015}). ^{Mincer (1974}) equations corrected for selection bias are estimated using the two-stage methodology of ^{Heckman (1979}). In this case, these are estimated separately for formal and informal employment, and therefore the selection correction must factor in that each estimation omits either formal or informal employment, as well as the non-employed who could potentially be so. Consequently, in the first stage of the Heckman approach, a multinomial logit model is estimated as in [1], where the dependent variable is the probability that the individual is formally employed, informally employed, or not employed (^{Pradhan and Van Soest, 1995}). The vector of covariates included (s _i ) in this estimation are age, age squared, categorical variables of educational levels, sex, categorical variables of position in the household, existence of children under 5 years old, and existence of children and young people between 5 and 15 years old; in other words, variables are included that affect the probability of belonging to any of the three activity conditions.

In the second stage, the inverse of the Mills ratio is included as a new dependent variable in the income equation, estimated, following ^{Lee (1983}), from the cumulative density and distribution function, calculated from the results of [1]. Thus, the second stage is the estimation of the income equation [2]:

In this case, the dependent variable is the logarithm of hourly labor income and the independent variables are the same as those included in [1], plus seniority in the occupation and λ^i, the Mills quotient. Once the Mincerian equations for formal and informal workers have been estimated, the estimated income of each group is calculated, as is the simulated income of informal workers as if they were paid as formal workers; thus, the individual gap that explains the magnitude of the potential penalization of the worker for being informal is estimated. Equation [3] illustrates the simulated labor income for an informal worker, considering both the vector of remunerations of the formal workers and their selection process.

The individual pay gap [4] is the difference between the labor income of informal workers and the income they would have if they were paid as formal workers, as a proportion of the latter.

Figure 2 and Table 3 illustrate the levels of pay gaps associated with informality by zone in and around Bogotá. The findings indicate that the highest observed differences in remuneration between formal and informal workers are found in the localities with the lowest gaps, Usaquén and Chapinero. This finding can effectively be explained by the phenomenon of labor segmentation due to informality. This, in turn, is explained by the fact that in these areas informal workers have pro-productive characteristics negatively associated with high salaries, limited education and experience, and household characteristics that affect their productivity (greater number of children in the household, higher share than other members).

Source: Compiled by the authors based on EM17 (^{DANE, 2020}).

Figure 2 Pay gaps by informality: zones of Bogotá and its Metropolitan Area 

Monetary poverty

Monetary poverty refers to the low income of the population, which prevents them from accessing a basic set of food and non-food goods and services (^{ECLAC, 2019}). Although this perspective is an important point of reference, it should be complemented by the multidimensional perspective analyzed in the following section, which includes other essential aspects of well-being.

The source of information for this research (the EM17) allows us to calculate, with a broad set of questions, the levels of labor and non-labor income of households. The Per Capita Household Income (PCHI) is calculated for all households based on this figure. The PCHI is then compared with the October 2017 Bogotá poverty line (270,782 pesos), given that the EM17 was conducted between September and November 2017.

To measure monetary poverty, the ^{Foster, Greer, and Thorbecke (FGT) (1984}) indicators were used, which are traditionally used in unidimensional approaches to poverty. Equation [5] shows this family of indicators. The indicator is the sum of the gaps of the PCHI (x _i ) relative to the poverty line (z _i ), for the case of poor individuals or households (x _i <z _i ) relative to the total population N. The FGT indicator depends on the coefficient ∝, which measures the levels of poverty aversion, when it is 0 all gaps equal 1. The indicator, meanwhile, is the proportion of poor, when it is 1, it is the average gap relative to the poverty threshold (line). If this is 2, it is an indicator known as poverty severity that weights each gap by itself.

Table 4 shows the estimated FGT indicators for each zone of Bogotá and its MA. It can be seen that US-SUM is the area of Bogotá with the highest poverty levels considering all indicators, as more than a third of its inhabitants live with less income than necessary to access a basic set of goods and services; SUR OCC has a similar level, although the latter presents lower levels of poverty gap and severity, which confirms US-SUM as the locality with the highest chronicity of poverty, with figures that exceed even those of the national average (^{DANE, 2019}).

In several areas of Bogotá, more than one-fifth of the population is poor (CBOL, BOSA, and CRIS-SFE-CAN), as well as the S. OCCIDENTE MA. These areas are also where the highest proportion of informal workers live and where the informality gap is greatest. In light of this, as presented in the following section, reducing informality would have a notable impact on improving the income levels of households living in poverty.

Multidimensional poverty

The most widely used methodology for measuring multidimensional poverty is the approach devised by ^{Alkire and Foster (2011}); however, there are other methodological proposals, such as that of ^{Bourguignon and Chakravarty (2003}), which use the union criterion to define poverty, i.e., the individual or household can be classified as being poor if it is deprived in at least one dimension. Likewise, this last proposal incorporates various degrees of substitution between dimensions, a parameter that modifies the sensitivity of the indicator to higher deficits in certain dimensions; the lower the substitution, the greater the impact of dimensions with greater levels of deprivation.

^{Bourguignon and Chakravarty’s [BYCH] (2003}) approach utilizes the FGT indicators [5], adapting them to the multidimensional approach [6], thus allowing the calculation of adjusted incidence indicators (α = 0), adjusted gap (α = 1), and adjusted depth (α = 2), adding different degrees of substitution between dimensions (𝜃=1), perfect substitution; 𝜃=2, moderate substitution; and 𝜃=5, low level of substitution among dimensions. Thus, the multidimensional poverty indicator is given by:

This article factors in several variables, dimensions, and thresholds, adjusted to the specific circumstances. Table 5 details how these criteria are operationalized, following on from ^{Sánchez et al. (2020}) which proposes an alternative measurement of multidimensional poverty for Colombia. Five dimensions of well-being (housing, education, work, health, and perceptions) were considered, consisting of 29 variables. Using this estimate, each qualitative variable of housing, work, health, and perceptions is classified with achievement (1) or deficit. The dimensional threshold, meanwhile, indicates the minimum number of variables that a household must attain achievement to not be considered deprived in that dimension. Understood in this light, the gaps are between dimensions rather than variables, meaning they quantify the number of variables for which a household has a deficit (or the distance from the threshold, similar way to the monetary measurement).

The housing dimension consists of eight variables that seek to evaluate housing conditions and access to household public services. Education is measured using a continuous variable, which aims to capture the educational level of the responsible members of the household, based on the education level of the head of household and their spouse.

The work dimension consists of seven variables, two of which are directly related to labor informality, namely "informal employment" and "labor income," and others that seek to evaluate different labor problems such as unemployment, economic dependence, and well-being at work. The health dimension was constructed by dichotomizing seven variables, including affiliation to the health system, regular attendance, physical activity, and diet.

Finally, perceptions consist of six variables designed to evaluate the internal experience of how individuals understand their material conditions; despite being a subjectively-evaluated dimension, it is important to understand how the heads of household or their spouses perceive their material living conditions regarding security, poverty, income sufficiency, and evaluations in the face of changing living standards.

As Table 5 illustrates, the thresholds refer to the minimum level of achievement for not being poor in each dimension (as in monetary poverty); in turn, given that a household is poor if it is deprived in at least one dimension (union criterion), those who are identified as deprived in one dimension have high degrees of deprivation, which is why the thresholds are relatively low.

Figure 4 shows the incidence of unidimensional poverty, that is, the percentage of households that are deprived according to each dimension. Results indicate that the housing and education dimensions present the lowest incidence percentages; for housing, most zones do not exceed 1% deprivation, with the exception of SUR OCC MA, while for education, the average household deprivation is 5%.

Source: Compiled by the authors based on EM17 (^{DANE, 2020}).

Figure 4 Deprivation by dimension and area of Bogotá and MA 

From the multidimensional perspective, one aspect of Figure 5 which stands out is that the SUR OCC zone has the highest poverty incidence rates relative to the other zones; meanwhile, the Bogotá CBOL, US-SUM, BOSA, CRIS-SFE-CAN, and TUN-RU zones present rates of over 30%.

Source: Compiled by the authors based on EM17 (^{DANE, 2020}).

Figure 5 Incidence of multidimensional poverty by area of Bogotá and MA 

Regarding the gap and severity of multidimensional poverty, it should be noted that, in general, Bogotá and the MA do not exhibit particularly high percentages of gap and severity (compared to the other regions of Colombia). It can therefore be hypothesized that people living in multidimensional poverty may be on the path towards breaking out of this situation. Table 6 indicates that the SOUTH OCC zone of the MA and the Bogotá zones of CBOL, BOSA, and US-SUM show high gap and severity rates when considering low levels of substitution; thus, poor households in these domains are farther away from leaving this condition in comparison to the rest of the domains.

Labor informality and monetary poverty

This section analyzes the potential effect that closing the informal labor informality pay gap would have on changes in income, and the consequent reduction in monetary poverty. The first step is to consider the income estimates of equation [2] and the simulated income of informal workers, assuming that they were paid like formal workers (equation [3]). Based on these estimates, the PCHI is calculated and the FGT indicators are estimated (equation [5]). The estimated ( PCHIEsth) will be the sum of all the labor income of the estimated formal and informal workers, and the non-labor income of all household members; finally this total family income is then divided by the number of household members. On the other hand, to observe the effect that the elimination of informality gaps, the simulated ( PCHISimh) is calculated, assuming that all informal workers in all households are remunerated based on the vector of coefficients of formal workers, maintaining the characteristics of informal workers.

Consequently, the effect of formalization on the FGT indicators of poverty is presented in equation [7]. Based on this, it can be hypothesized that the improvement in income derived from the remuneration of informal workers as if they were formally employed, implies a decrease in the FGT indicators. Therefore, the reduction in poverty due to the change from the estimated PCHI to the simulated PCHI would be the potential effect of the closing of segmentation gaps (^{Maurizio, 2015}; ^{Sanchez, 2020}).

Table 7 presents the results of this hypothetical scenario of closing informality gaps for each zone of Bogotá and all monetary poverty FGT indicators. As is to be expected, the zones that would see the greatest reduction in monetary poverty are those where the informality gap and poverty itself are highest. Particularly noteworthy are the reductions in poverty in CRIS-SFE-CAN, SUR OCC, and S. OCCIDENTE. By contrast, the localities with higher income levels and less poverty, and which are also those with lower differences in remuneration due to informality, would witness an effect that was quantitatively very low, indicating that the existing poverty (which is between 5% and 8%, lower than other localities) is due to informality, but rather other factors such as low educational levels, unemployment, and the high rate of economic dependency in households.

There is a pattern common among most areas of formalization having a greater impact on reducing the incidence of poverty rather than gaps or severity of poverty. This is because this hypothetical situation would help households to surpass the poverty line, but would have a lesser effect on improving the income of those households that continued to be poor.

The scenario hypothesized above assumes a complete closing of informality gaps; however, any potential formalization policy would have a gradual effect that would affect workers with different characteristics in various ways. In light of this, a scenario with progressive formalization was simulated, considering the three monetary poverty indicators and all zones of Bogotá and the MA. The estimation approached the formalization process using the estimations from the multinomial model (equation [1]), and specifically the probability that an individual is informally employed (assuming that they are currently informally employed). Thus, the simulation of the formalization and gap closure process began with workers with the lowest probability of being informally employed, and progressively increased the number of formalized workers until reaching total formalization (see Table 7), with the last worker to be formalized being the one with the highest probability of being informal.

Figure 7 shows how poverty indicators would change with progressive formalization for the total of Bogotá and the MA. The initial scenario assumes current levels of poverty and informality and, as formality decreases, poverty indicators also decrease. However, it is important to highlight that even without gaps in remuneration due to informality, the incidence of poverty would still be above 10%, and despite being the indicator that decreases the most, it does not show a significant decrease (barely 20% lower). The severity of poverty is less affected by the formalization process, implying the existence of other factors that affect the persistence and depth of poverty.

Source: Compiled by the authors based on EM17 (^{DANE, 2020}).

Figure 7 Changes in poverty indicators due to progressive formalization in Bogotá and MA 

Labor informality and multidimensional poverty

To evaluate changes in the dynamics and relationship between labor informality and poverty from a multidimensional perspective, this section calculates all multidimensional indicators for the 16 geographic domains and Bogotá and the MA, for three simulated labor scenarios: incidence elimination, gap elimination, and joint effect. To examine the potential effect of these scenarios on the living conditions of the population, the multidimensional poverty indicators are calculated using the BYCH methodology (equation 6).

Initially, the real informality and income gaps and the multidimensional poverty index are estimated based on the BYCH methodology, after which variable 1 (Informal Employment) of the work dimension is modified, thus constructing the "Formalization Incidence Effect" [8], where the entire population is formalized so that no individual is deficient in this variable.

The next step is the elimination of wage gaps between formal and informal workers. To achieve this, variable 2 (Labor Income) of the work dimension is modified by assigning remunerations to informal workers as if they were formal, thus creating the "Formalization Gap Effect." This scenario is presented in equation [9], which simulates the effect of closing the informality gap and the effect on the measurement of multidimensional poverty.

The third simulation is based on the first two; the "Aggregate Formalization Effect" makes simultaneous modifications to variables 1 and 2 of the work dimension. Thus, the effects of equations [8] and [9] are added together.

Table 8 shows how the multidimensional poverty indicators would change with the formalization of the entire population and the consequent increase in labor income (aggregate effect). These results demonstrate a reduction in the three indicators of multidimensional poverty (α = 0, 1, and 2) for all areas of Bogotá and the MA, with more significant decreases in the peripheral localities, as well as in the gap and severity indicators when there is perfect substitution between dimensions. Thus, potential formalization (elimination of incidence and gaps in remuneration) would undoubtedly improve the welfare of the population, even when there are households that continue to be poor due to their deprivations in other dimensions, in which even the deficits in the variables are more acute than in work, which is why the effect is smaller when the degree of substitution decreases.

Figure 8 illustrates the impact of each of the effects of formalization (incidence and gap), as well as the joint effect on the potential reduction of the multidimensional poverty incidence indicator in Bogotá and its MA. The Incidence Effect clearly reduces poverty more significantly, however; despite the lesser impact of the Gap Effect, the levels of income inequality in Bogotá stand out, since, despite the elimination of the gaps in remuneration, the peripheral localities and geographic domains continue to lag behind localities such as Chapinero, Barrios Unidos-Teusaquillo, or Usaquén. On the other hand, there is little distinction between the multidimensional poverty indicators, which would indicate that the improvement of the working conditions of the population from the elimination of the gap would not have a great impact on the reduction of poverty in the domains with greater vulnerability.

Source: Compiled by the authors based on EM17 (^{DANE, 2020}).

Figure 8 Changes in the incidence of multidimensional poverty due to formalization and closing of gaps in informality in areas of Bogotá and MA