Labour (im)mobility and monopsonistic exploitation of workers in the urban informal sector: Lessons from a field study

Abstract

It is well-recognised in labour economics that employers in monopsonistic competition have market power to set wages below the productivity of workers if workers are immobile, which causes monopsonistic exploitation of workers. Monopsonistic exploitation is propelled by employers’ expectations about the degree of (im)mobility of workers and therefore by ex post mobility of workers. It hence begs a question of what determines the relative (im)mobility of informal workers, which has significant implications for monopsonistic exploitation and working poverty in the urban informal sector. From our field study of urban and peri-urban labour markets in Bangladesh, by modifying the method of stochastic frontier analysis, we explain the behavioural foundation to the relative (im)mobility of informal workers. We thereby unravel the effects of individual and household characteristics, sociological, institutional factors and household ties (including worker’s kinship) on the relative immobility of a worker that can, in turn, explain monopsonistic exploitation and working poverty widely prevalent in the urban informal sector of South Asia.

Keywords

frontier approach kinship labour (im)mobility monopsonistic competition monopsonistic exploitation stochastic workers’ characteristics

Introduction

We envisage that an ‘oligopsony’, or ‘monopsonistic competition’, fits the most accurate descriptions of the informal labour market that we examine in this paper. Oligopsony describes a situation where the market power of an employer in setting wages persists despite competition with other employers – as a result, the number of employers does not need to be small (see Bhaskar and To, 1999; Bhaskar et al., 2002). If wages are set purely according to labour productivity, then high-wage workers and low-wage workers will have little reason to quit jobs. In models of oligopsony, or monopsonistic competition, however, the labour supply curve facing an individual employer is imperfectly elastic. The imperfect elasticity of labour supply suggests that a monopsonist can practise price discrimination in the labour market by setting a wage less than the productivity of a worker and, if workers are immobile, lower wages will perpetuate. We call this phenomenon of wages below one’s productivity monopsonistic exploitation (see Robinson, 1933) and seek to explain monopsonistic exploitation in the informal labour market. It is long held that immobility of workers contributes to monopsonistic exploitation and two main factors that impede labour mobility are employment and family conditions (see Dickie and Gerking, 1998; Topel, 1986). Our paper seeks to establish how family ties and employment conditions can seriously compromise labour mobility and, thereby, lead to monopsonistic exploitation of workers. The informality in the urban labour market has led to a deep cleavage in urban economics: the traditional economic models argue that informal workers are the involuntary and disadvantaged residual of segmented labour markets created and perpetuated by labour market interventions.¹ The rebuttal to this view poses the informal sector to be peopled by ‘micro-entrepreneurs’ who voluntarily choose an optimal work–life(style) balance, or package. In this paper, by appealing to the basic tenets of search models and behavioural economics, we posit that informal workers often fail to achieve an optimal package because of their relative immobility caused by informational problems, sociological and behavioural issues, so that workers make systematic errors and biases in their choices. As a direct consequence of search costs and these errors and biases, a spatial segmentation of the informal labour market, without any form of wage rigidity, can endogenously take place that will, in turn, result in monopsonistic exploitation of workers to perpetuate urban poverty. Our main contribution in our paper is to explain the major determinants of workers’ relative immobility.

In this work we try to empirically ascertain the factors responsible for monopsonistic exploitation of workers if informal workers have informational problems, sociological ties and behavioural issues. According to the second view, the informalisation is a voluntary process, in the sense that workers ‘vote with their feet’ as they choose to hold the current informal jobs vis-à-vis other jobs available elsewhere in the economy. Given their labour market characteristics, workers’ choices hence display their revealed preferences for a set of job characteristics that best suit their preferences given their economic, educational and social constraints.² This second view has a potent economic intuition that human rationality and perfect information often blend together to motivate workers to choose a desired lifestyle package and the informality is an integral component of the package voluntarily seized by many workers. The problem with this view is that information is often sketchy and costly to acquire, which calls forth efficient mechanisms for the transmission of information (Greif, 1993, 1998). As a result, there is nothing sacrosanct that workers can come to choose the most desired lifestyle package as Balán et al. (1973) argue in their early contribution that some workers have an incentive to ‘shop around’ for learning the desirability of various possible life choices. It is a moot point whether the learning process in the form of ‘shopping around’ in the informal sector can eventually lead workers to their most desired lifestyle packages.

At any point in time, because of an ongoing information dynamics from a learning process, the informal labour market will be segmented into submarkets – not because of wage rigidity as the traditional literature highlighted – but because of spatial wage distribution owing to costly search activities. For the relevance and consequences of search activities in economics see Gale (1987), Pissarides (2000) and Shimmer (2005). Though workers seek to choose the optimal lifestyle package, they fail to achieve the optimal package instantly because of informational constraint, which one may call monopsonistic exploitation resulting from relative immobility of informal workers. It is also instructive to note that workers may not choose the optimal package even if they are fully informed of the wage distribution for sociological and behavioural reasons. This has important implications for our study. At any point in time workers are either unaware of the optimal choice or fail to achieve the optimal choice, which creates relative immobility of workers since they fail to respond to available incentives. The relative immobility, in turn, creates and perpetuates monopsonistic exploitation of workers. The main goal of the paper is to explain monopsonistic exploitation in informal sector due to immobility from a field survey undertaken in Bangladesh. The study will offer valuable insights, to the best of our understanding for the very first time, into the choice of informal workers about their lifestyle packages and the study will thereby unravel the impacts of sociological and institutional variables on the choice of workers.

Related literature

The informal labour market thrives in the urban as well as peri-urban sectors mainly as a result of increased urbanisation of city fringes in developing nations driven by cheap oil, prosperous economies and increased private transport (Houston, 2005). We use the term urban informal sector for both traditional urban areas and peri-urban areas in our study. In many such markets, the decentralised nature of the interaction prevents the efficient (Walrasian) exchange. In the urban informal labour market, an unemployed worker who looks for a job might not be the only job-seeker, in which case one of the other job-seekers might be hired and he remains unemployed. Notwithstanding this, when the intensity of search is sufficiently high, it is expected that the labour market allocation will be approximately Walrasian. This is indeed the much-celebrated finding in sequential search models in which workers apply to one firm at a time, as long as the period length between applications becomes sufficiently small (e.g. Gale, 1987; Lauermann, 2006; Mortensen and Wright, 2002; Satterthwaite and Shneyerov, 2007).

‘Shopping around’ for the optimal lifestyle package: Search activities and wage formation

In most search environments, especially in the developing economies, the period length is not small. The hiring process creates a substantial lag between the time when a worker looks for a job and the point at which the final hiring decision is reached. More importantly, the urban informal labour market in developing nations is usually propelled by kinship-based transactions, which can induce significant lags and frictions for moving from one job to another. The implication is that urban informal labour markets in developing nations can therefore experience inefficient and non-Walrasian outcomes.³ However, the threat of not obtaining a job, or ‘good’ job, and having to wait another period is exactly what provides incentives for workers in the urban informal sector to simultaneously look for jobs from several employers. This idea to model job search as a simultaneous application process dates back to the foundations of labour search in Stigler (1962). In the labour markets of advanced nations, the ability of this channel to generate efficient equilibrium outcomes has adequately been explored by Acemoglu and Shimer (1999), Gautier and Moraga-González (2005), Albrecht et al. (2006) and Galenianos and Kircher (2008). Though none of these papers finds convergence to the Walrasian limit when the search costs vanish, they do find labour market discrimination to assume less importance with simultaneous search. They highlight that there is some form of inefficiency in the labour market, but no systematic exploitation of workers because of the prevalence of simultaneous and intensive search.⁴ There is a significant difference between the new work and the classical work on search in labour markets: the initial work on equilibrium search by Butters (1977) and Burdett and Judd (1983) also considered simultaneous search but their focus was on the distribution of wages. Except for the costs of information acquisition, these classic papers lack a margin of efficiency, since each firm can service the entire market. Efficiency questions are also missing from the original contribution by Stigler (1962), since he considered a model of the workers in a partial equilibrium setting in which the benefit from matching for the firms is not considered. Our main contribution is to document how household and social ties can influence search intensities and, thereby, monopsonistic exploitation of a worker.

Behavioural economics and search activities in the informal sector

Individual workers can incur systematic errors and biases in their decisions. Therefore, workers are said to be persistently irrational in their decision-making so that they fail to achieve the optimal package. They are irrational because their choices deviate from neoclassical norms of rationality. Because they are irrational, inducing rational cum neoclassical behaviour becomes an issue of critical importance from the perspective of the paradigm developed by Maloney and his colleagues. Behavioural economics has always been concerned about psychological as well as sociological and institutional variables as determinants of choice, which together lend themselves to a better understanding choice behaviour in (see Altman, 2005; Becker, 1996; North, 1990, 1994). While previous migration studies have indicated that family ties deter labour mobility (Krumm, 1983; Mincer, 1978; Shields and Shields, 1993), there is no convincing evidence that the reduced mobility is reflected in interregional wage variation. Dickie and Gerking (1998) provide the limited conditional evidence by examining wage disparities across Canadian provinces. Ours is a more detailed study of family ties being a major determinant of mobility of workers, especially in the context of the urban informal sector in a developing country.

Mechanism to detect monopsonistic exploitation of workers

In this paper we introduce the question of monopsonistic exploitation in the context of wage setting by oligopsonists in the urban informal labour market in Bangladesh. To be more specific, we estimate a new form of wage equations for male and female workers in urban and peri-urban areas, in the capital city of Dhaka of Bangladesh. To label what a worker could earn, we introduce the maximum potential wage of each worker at different locations that depends on a vector of observed individual, household characteristics and family ties, or kinships, of a worker. In the frictionless world, a worker has perfect information and zero immobility and will therefore achieve the maximum potential wage instantly, which implies a costless and perfect matching of workers with their employers. In an imperfect world, with incomplete information and lack of perfect mobility of workers, a worker will accept a wage offer from an oligopsonist below one’s productivity. The difference between actual wages and potential wages will give us an index of relative immobility for each worker, which is a direct implication of the model articulated by Acemoglu and Shimer (1999). Estimates of imperfect mobility are calculated by computing the percentage of potential wage received by each worker and the potential wage is computed from the stochastic frontier analysis. The approach, though modified, is adapted from the model of Aigner et al. (1977) and Meeusen and van den Broeck (1977) and the stochastic frontier and immobility/inefficiency effects are investigated as a function of individual specific factors in a one-step procedure (see Polachek and Robst, 1998). We then model the relative (im)mobility of workers as a function of the independent variables including labour-market specific characteristics of workers (such as human capital and skills), past employment conditions, search experiences and also from a host of other factors. From the simultaneously calculated stochastic frontier estimates and the (independent) immobility variables, we then explain the determinants of monopsonistic exploitation.

Labour mobility and stochastic frontier estimates: An index of monopsonistic exploitation of workers in the urban informal sector

Labour mobility is an important phenomenon in both modern and traditional societies. Labour mobility is usually captured by changes in the location of workers both across physical space (geographic mobility) and across occupations (occupational mobility). Our main concern is geographic mobility. Labour mobility is an important vehicle to reduce monopsonistic exploitation.⁵ At the micro level, mobility leads to improvements in the economic circumstances of those whose skills or aspirations are a poor match with their current employers. The standard result from the theory of job search suggests that mobility of workers depends on their reservation wages and search efforts of workers (Lippman and McCall, 1979; Mortensen, 1977). Consequently, there are various economic, institutional and social factors that either impede or promote geographic mobility by impacting on the net benefits, or productivity, from search activities (see Ben-Horim and Zuckerman, 1987).

To assess the validity of mobility of workers interesting studies are undertaken to understand the impacts of a worker’s information about prospective wages on one’s wage (see Groot and Oosterbeek, 1994; Hofler and Murphy, 1992, 1994; Hofler and Polachek, 1982, 1985; Polachek and Yoon, 1987, 1996).⁶ While the results obtained using the stochastic frontier estimates are consistent with search theory, however, there is a concern whether this frontier measure accurately depicts information (see Polachek and Robst, 1998). The main problem with the stochastic frontier approach is that incomplete worker information is measured as a residual. There is uncertainty whether the estimates of the stochastic frontier estimation really capture a measure of incomplete information. To rectify this deficiency, Polachek and Robst (1998) developed a direct measure of a worker’s knowledge of labour market opportunities to assess the estimates based on residuals from the stochastic frontier method. In the section ‘Research methodology: Stochastic frontier estimates and labour mobility’ we will apply and extend their model to assess the impacts of mobility factors on the wage of a worker, or monopsonistic exploitation of a worker. The details are provided in the subsection ‘Mobility and monopsonistic exploitation in the informal sector’.

Mobility and monopsonistic exploitation in the informal sector

Our main goal is to utilise the wage data, individual and household characteristics and kinship variables from the urban informal sector, without a comparison with the formal sector, to empirically estimate monopsonistic exploitation for each individual worker and determine the factors responsible for such exploitation. Family and social ties matter in labour markets as it pays to be socially connected: for example, employees find new jobs through their network of friends and colleagues (Granovetter, 1973, 1974), employers fill vacancies using their employees’ contacts (Holzer, 1987). Social networks play an important role in the organisation of many significant economic relationships. Informal social networks, such as kinships, are often the means for communicating information, sharing favours and allocating goods and services which are costly for acquiring them from markets. Economists have long argued, borrowing mainly from sociologists (see Freeman, 2004), that social networks help economic organisations by transmitting critical information, influencing decisions and motivating agents about job and business opportunities, reputation of employers and employees and thereby influence productivity of individuals in groups (see Jackson and Wolinsky, 1996).

Economic development and the urban informal sector

In simplistic terms, there are two contrasting views about the informal sector in development economics: the traditional view is that the informal sector is a disadvantaged residual of segmented labour markets, see Maloney (2001, 2003) for a critique of the traditional view. The informal sector thus represents, according to the traditional view, ‘the small-scale, semi-legal, often low-productivity, frequently family-based, perhaps pre-capitalistic enterprise’ that absorbs a significant proportion of the urban work force in developing nations. Extending the Harris and Todaro (1970) model, Chandra and Khan (1993), Chandra (1992), Loayza (1994) highlighted informal workers as the building blocks of the ‘less-advantaged sector’ of a segmented labour market. In other words, the above-market-clearing wages in the formal sector due to institutional factors, such as the minimum wage legislation and effective unionism, sends unemployed workers of the formal sector to an informal sector for their survival in a dualistic labour market. The dualistic labour market, according to the traditional view, is characterised by an absence of benefits, lower wages, high turnover precarious employment conditions and deep poverty. The informal sector is thus viewed as a residual comprising of disadvantaged workers who are ‘rationed’, or forced, out of good jobs because of inflexibility in the formal labour market. This view of informality is in consonance with the International Labour Organization (ILO) definition of ‘decent work’ as those jobs in the formal sector that are under the cover and protection of formal labour institutions. According to the World Bank, formal sector jobs are ‘good jobs’ that abide by labour standards and operate within the accepted institutional frameworks of labour markets (see Canagarajah and Sethuraman, 2001).⁷

The alternative view is that the informal sector is primarily an unregulated micro-entrepreneurial sector, which is similar to the sector comprising of ‘voluntary entrepreneurial’ small firms predominantly observed in the urban sector of many advanced nations (see Maloney, 2004; Maloney et al., 2003). This is especially the case if there is no wage rigidity in the formal labour market (see Maloney, 2001; Maloney and Nunez, 2003; Maloney and Ribeiro, 1999). Maloney (2003) argues that the informal sector arises as a micro-entrepreneurial sector since the minimum wage laws are not enforced as labour unions are overtly concerned with preserving jobs rather than securing healthy working conditions and wages. As a result, there is no segmentation of the labour market into formal and informal sectors in the absence of wage rigidity. There is no direct, or indirect, coercion of workers to leave the formal sector as they are free to ‘optimally choose’ the informal sector as their most preferred location.

Research methodology: Stochastic frontier estimates and labour mobility

We estimate a new form of wage equations for male and female workers in urban and peri-urban areas, in the capital city of Dhaka and Nilphamari District of Bangladesh. To label what a worker could earn, we introduce W_i to be the maximum potential wage of worker i:

W_{i} = β * R_{i} + φ_{1} {(Ω_{2})}_{i} + φ_{2} {(Ω_{3})}_{i} + φ_{3} {SEMPDUM}_{i} + ψ_{0} + e_{i} i = 1, 2 \dots 660,

(1a)

where ψ₀ is the constant term, R_i is a vector of household and individual and personal attributes such as urban residency, marital status and age etc.; and (Ω₂)_i and (Ω₃)_i are two variables of kinship measures, SEMPDUM_i is a dummy variable to capture the self-employment status of the ith worker.

A complete listing of all exogenous variables is provided in Table 2A and 2B. β and φ₁ and φ₂ are vectors of coefficients and e_i is a stochastic error term that is normally distributed N(0, σ²). In the frictionless world, a worker has perfect information and zero immobility and will therefore achieve W_i instantly, which implies a costless and perfect matching of workers with their employers. In an imperfect world, with incomplete information and lack of perfect mobility of workers, a worker will accept a wage offer W_i^O such that:

W_{i} - W_{i}^{O} = U_{i}

(1b)

Note that U_i≥0 and is an index of relative immobility and informational frictions of worker i. From equations (1a) and (1b) we get:

W_{i}^{O} = β * R_{i} + φ_{1} {(Ω_{2})}_{i} + φ_{2} {(Ω_{3})}_{i} + φ_{3} {SEMPDUM}_{i} + v_{i}

(1c)

v_{i} = e_{i} - U_{i} - ψ_{0}

(1d)

From the mean and mode of (e_i/v_i) for each individual worker i, following Polachek and Robst (1998), one can calculate an index of relative immobility for each worker i. Estimates of imperfect mobility are calculated by computing the percentage of potential wage received by each worker. The potential wage is computed from the stochastic frontier analysis. The approach is similar to the Aigner et al. (1977) model except the stochastic frontier and inefficiency effects are modelled as a function of individual specific factors in a one-step procedure. We model the relative (im)mobility U_i of worker i as a function of the independent variables that are demographic attributes and (past and experienced) employment conditions, human capital measures such as education and literacy, various search factors, household responsibilities etc. which determines the mobility of a worker and given as:

U_{i} = \partial * Z_{i} + θ_{1} K_{i} + θ_{2} L_{i} + ω_{1} J_{i} + ω_{2} M_{i} + ω_{3} {(Ω_{1})}_{i} + ω_{4} {STATUS}_{i} + α

(2a)

The individual mobility is calculated as:

{Mobility}_{i} = 1 - (\partial * Z_{i} + θ_{1} K_{i} + θ_{2} L_{i} + ω_{1} J_{i} + ω_{2} M_{i} + ω_{3} {(Ω_{1})}_{i} + ω_{4} {STATUS}_{i} + α)

(2b)

Where K_i is a vector of informational variables, L_i is a vector of variables that capture the household inertia to migrate, Z_i is a vector of past employment conditions and past search experiences, what we call experienced goods. ∂,θ₁, θ₂ are the coefficients, J_i is a variable to capture past contract type and M_i is a variable that captures the reason for past mobility such as severe push factors. Note that (Ω₂)_i is a variable that captures social kinship. Note that ω₁, ω₂, ω₃ are relevant parameters to be estimated. ∂ = θ₁= θ₂= ω₁= ω₂= ω₃= ω₄=0, then the model reduces to the model of Aigner et al. (1977) while non-zero values will lend meaning to our model of postulated (relative) immobility. Note that α is a random variable defined by the truncation of the normal distribution N(0, σ²).

The above stochastic frontier estimator can be used to compare mobility across regions and sectors.

Data

The relevant data set is derived from the primary survey as described in the Appendix. In order to obtain the relevant data set, we divide the field of study into 13 neighbourhoods, or urban and peri-urban locations. Each location is controlled by a small group of oligarchs, both in urban and peri-urban areas, who enjoy significant political and economic decision-making power within a location. Within each location j employer j has the monopsonistic power to set the wage rate and employment conditions. In principle, urban and peri-urban workers can vote with their feet in response to monopsonistic exploitation. However, because of the role of kinship in determining employment relations and household ties, we expect some immobility among workers, which in turn allows their employers to practise significant monopsonistic exploitation. In what follows we provide some of the basic statistics of the 13 regional locations.

Descriptive statistics of the regional locations

We selected the Nilphamari district for the peri-urban area, which is located 400 km from Dhaka and situated close to the Indian border town of Siliguri. The district was selected since it has Saidpur that is the third largest city of Bangladesh after Dhaka and Chittagong. Since the time of the British the district has been the home of several manufacturing sectors such as railway workshops, telephone exchanges and their ancillary industries, despite the fact that agriculture still plays an important economic role in the district. The district has six subdistricts and Tista is a trans-boundary river between India and Bangladesh. The district came into prominence during early British rule for Indigo cultivation and peasant revolts to forcible farming and packaging of Indigo. Towards the end of the British rule, the region also witnessed the Tebhaga movement by the local peasants led by the Communist Party of Undivided India for a just sharing of agricultural output between tillers and landlords. During the second week of December 2008 data collection took place for the peri-urban areas of the Nilphamari district that has a population of 135,000 over a geographic area of 1821 km².The population density is about 950 per km². We randomly chose 248 households out of 25,000 households in the region in four regions of the district: roughly 50 households were interviewed from each of the four regions – Nilphamari Sadar (Region 1), Saidpur (Region 2), Kishoreganj Sadar (Region 3) and Palashbari (Region 4). For the urban sector the metropolitan city of Dhaka has been selected for the survey. From the urban area, we selected 412 respondents from the non-farm sector. We collected data from the urban area till the end of March 2009. The Dhaka metropolitan area has a population of 15 million and an area of 360 km² with 49 Thanas (blocks). We chose eight regions spread across the city of Dhaka and interviewed 412 households from Hazaribagh (Region 5), Mirpur (Region 6) both in the North, Savar (Region 8) in the North West, Rupganj (Region 9) in the Northeast, Demra (Region 10) in the East, Shyampur (Region 11) in the South, Tejgaon (Region 12) in the Southeast and Ruhitpur (Region 13) in the West.

Table 1 summarises the descriptive statistics of the 13 urban and peri-urban locations surveyed.

Table 1.

Descriptive statistics.

Γ_j	N_j	SAW_j	SDW_j	AA_j	AE_j	AM_j	AY_j	H/P
1	51	86	25.21	37	4.7	2	14.2	P
2	50	82	36.28	39	3.38	2	15.8	P
3	50	72	29	40.78	3.48	2	16.5	P
4	48	74	27.14	42	4.22	2.04	19.3	P
5	50	74	20.96	37	3.66	2	14.8	H
6	49	68	7.26	26	2.54	2	4.5	H
7	51	76	7.6	35	2.96	2	12.7	H
8	49	84	18.46	48.24	2	1	4.6	H
9	50	94	20.17	25	2	1	5.4	H
10	50	115	20.88	38	4.7	1.96	15.9	H
11	50	75	19.83	39	4	2	15.2	H
12	50	93	43.5	36	4.46	2	13.4	H
13	51	63	27.4	40	4.72	3	14.8	H

Notes: Γ_j: Location j; N_j: Number of Workers Surveyed in Location j; SAWj: Average Wage of Surveyed Workers in Location j; SDW_j: Standard Deviation of Wages of Surveyed Workers in Location j; AA_j: Average Age of Surveyed Workers in Location j; AE_j: Average Education of Surveyed Workers in Location j; AM_j: Average Marital Happiness of Surveyed Workers in Location j; AY_j: Average Years of Service for Surveyed Workers in Location j; H: Urban location; P: Peri-urban location

Source: Created by the authors from the survey data.

In Figure 1 we plot the efficiency scores for the 13 regions constructed from seven alternative models, labelled as seven series, as explained in Table 3. The following plot, based on seven possible models termed series, clearly shows that the efficiency measures, from all models, display significant variations across regions, which suggests some variations in relative immobility of informal workers across regions. In what follows in subsection ‘Descriptive statistics of the regional locations’ we explain the causes of relative immobility of informal workers. Two urban regions, Demra in the East (Region 10) and Tejgaon in the Southeast (Region 12), register markedly lower efficiency levels than other regions, while others reveal some variations in efficiency scores. We thus note both intra and interregional variations in efficiency levels, or immobility, of informal workers.

Figure 1.

Regional variations of efficiency.

Regressors and relevant variables

To measure kinship variables from the survey data, the survey instrument asked three important questions about relevant kinship: first, all 660 households were asked whether they receive, and the amount of, any financial benefit from social networks outside one’s household (Variable Ω₂). Second, each respondent was asked if they receive, and the amount of, financial benefit the household receive from family members (Variable Ω₃). Third, each respondent was asked the nature of recruitment process associated with the current job (Variable Ω₁). We choose Ω₁ = 0 if the recruitment was through a formal channel and there was a written contract, Ω₁ = 1 if the recruitment was semi-formal and there was a verbal contract, Ω₁ = 2 if the recruitment process was informal and the job came through family and social networks without any form of contract – written or verbal. Thus, we posit that the higher the value of Ω₁ the larger is the influence of kinship relationship and household ties in the choice of workplace.

The survey also includes a number of measures of individual and household characteristics including sex, age, education, working conditions and family responsibilities, etc. The survey includes a measure of income, wage rates, working hours and decomposition of family income into various types of consumption. The survey has a precise geographical identifier for each family. Note that the details of variables, except the kinship variables, are described in Table 2B.

Table 2.

A: Kinship variables.

Ω₁: It captures the nature of recruitment process in the current job and takes three possible discrete values: Ω₁ = 0 for a written and formal contract, Ω₁ = 1 for a semi-formal and verbal contract, Ω₁ = 2 for informal recruitment through social networks and without any contract

Ω₂: The total financial/monetary benefits that the respondent received from one’s social network excluding one’s household circle

Ω₃: The total financial/monetary benefits that the respondent received from one’s household members

Source: Constructed by the authors from the survey.

Table 2.

B: Other variables of interest

Variable name	Survey question/description
R₁: GENDER	Male = 1, Female = 0
R₂:UR	1 = Urban Sector, 0 = Peri-urban Sector
R₃: AGE	Age in years
R₄: Highest Level of Education in Household (LE)	LE Ranges from 0 to 8 (increasing scale)^a
R₅: Marriage Dummy (MAR)	MAR = 1 if currently married, otherwise = 0
STATUS: Social Status	Social Status of Current Occupation (increasing scale)^a
Current Wage (w)	Log of hourly wage rate in Bangladeshi currency
SEMPDUM	Dummy Variable = 0 if self-employed or employed by family, otherwise = 1
Past search experiences
Search Time to Find the Current Job (Z₄)	In days
Search Unemployment, if any, during search for the Current Job (Z₅)	In weeks
(Im)mobility indicators
1. Informational variables
Respondent’s Literacy Index (K₁)	0 = Illiterate, 1 = Literate
Respondent’s Schooling Index (K₂)	Never Attended School = 0, Attended School = 1
2. Contract type
Nature of Past Contract (J₁)	0 = Daily Contract, 1 = Non-daily Contract
3. Mobility restrictions resulting from family responsibility (L_i)
Index of Free Domestic Services (L₁)	An increasing scale in terms of hours of free domestic work (0–15), L₁ = 0 for zero free domestic work
Self-evaluation if free domestic services lower wages (L₂)	0 = No, Yes = 1
Self-evaluation if job mobility is due to household restrictions (L₃)	1 = Yes, 2 = Maybe, 3 = No
4. Reasons for past mobility
Reason for past mobility (M₁)	1–9 = Severe Push Factors, 10–18 = Pull and Push Factors, 19–32 = Only Pull Factors^a
5.Workplace conditions (experienced goods)
Safety measure dummy (Z₆)	Z₆ = 0 No safety measure at work experienced, otherwise Z₆ = 1
Safety at work dummy (Z₇)	Z₇ = 0 if work is experienced as unsafe, Z₇ = 1, otherwise
Shelter of law dummy (Z₈)	Z₈ = 0 if no shelter of law experienced, Z₉ = 1, otherwise
Injury responsibility of employer dummy (Z₉)	Z₉ = 0 if employers did not take responsibility for injury, Z₉ = 1, otherwise

Note: ^aSee Appendix (A4) for details of the variable.

Source: Constructed from the survey.

Findings

To avoid multicollinearity and potential endogeneity problems we have dropped variables R₇ (health expenditure) and R₈ (education expenditure) and Z₁ (occupation of parents), Z₂ (worker’s occupation) and Z₃ (delay in wage payment). Our findings are summarised in Table 3A and Table 3B.

Table 3.

A: OLS estimates.

	Model 1Coef
LNAGE	−0.09
LNΩ₂	0.003
LNΩ₃	0.005
GENDER	−0.12
UR	0.06
LE	−0.001
MAR	−0.07
SEMPDUM	−0.040
Constant	1.89
ADJ R ²	0.24
N	660

Table 3.

B: Stochastic frontier estimates.

	Model 2Coef	z-score	Model 3Coef	z-score	Model 4Coef	z-score	Model 5Coef	z-score	Model 6Coef	z-score	Model 7Mean	z-score
LNAGE	0.012	1.03	0.011	0.97	0.013	1.21	0.013	1.17	0.010	0.91	0.0037	0.33
LNΩ₂	0.0055	3.80^a	0.0014	3.86^a	0.005	4.03^a	0.0058	4.09^a	0.00557	4.10^a	0.006	4.20^a
LNΩ₃	0.001	1.08	0.0010	01.12	0.0010	1.11	0.001	1.13	0.0012	1.38	0.001	1.22
GENDER	−0.061	−7.51^a	−0.061	−7.68^a	−0.053	−6.93^a	−0.054	−7.05^a	−0.048	−6.61^a	−0.05	−7.61^a
UR	0.035	2.99^a	0.038	3.28^a	0.410	3.53^a	0.043	3.77^a	0.048	4.23^a	0.04	4.06^a
LE	−0.0356	−5.11^a	−0.0027	−3.54^a	−0.0035	−5.12^a	−0.002	−3.84^a	−0.0035^a	−5.18^a	−0.002	−3.69^a
MAR	−.026	−3.22^a	−0.026	−3.24^a	−0.028	−3.58^a	−0.028	−3.61	−0.029^a	−3.72^a	−0.026	−3.43^a
SEMPDUM	−.025	−2.41^a	−0.024	2.38^a	−0.021	−2.05^a	−0.020	−2.02^a	−0.014	−1.42	−0.015	−1.64
Constant	1.86	43.07^a	1.86	43.05^a	1.85	43.55^a	1.84	43.5^a	1.85^a	44.34^a	1.84	44.15^a
ADJ R ²
Constant	−6.35	30.61^a	−6.30	−31.08^a	−6.4	−31^a	−6.42	−31.33^a	−6.5	−31.94^a	−5.95	−37.95^a
STATUS	.046	3.05^a	0.037	2.44^a	0.039	2.59^a	0.032	2.1^a	0.042	2.81^a	0.042	1.92
K₁	−.085	−0.76			−0.05	−0.47			−0.11	−1.03
K₂			0.43	2.64^a			0.33	2.16^a			0.47	2.06^a
L₁	.040	2.43^a	0.046	2.72^a
L₂					0.54	3.23^a	0.53	3.12^a
L₃									0.64	4.98^a	0.76	3.98^a
J₁	−.45	−2.11^a	−0.37	−1.72	−0.55	−2.63^a	−0.5	−2.40^a	−0.44	−2.18^a	−0.23	−0.76
M₁	.0057	0.73	0.0034	0.43	0.0056	0.73	0.004	0.51	0.004	0.56	0.005	0.51
Z₄	.057	0.50	0.076	0.67	0.056	0.49	0.071	0.63	−0.056	−0.48	0.015	0.09
Z₅	−0.009	−0.31	0.003	0.1	−0.004	−0.16	0.006	0.21	0.0005	0.02	0.004	0.11
Z₆	1.02	4.58^a	1.13	4.89^a	0.98	4.47^a	1.05	4.73^a	0.86	3.89^a	1.30	4.10^a
Z₇	1.40	4.36^a	1.49	4.53^a	1.35	4.32^a	1.41	4.42^a	1.34	4.33^a	1.49	3.30^a
Ω₁	−.97	−9.73^a	−1.02	−9.85^a	−0.89	−9.01^a	−0.93	−9.16^a	−0.90	−9.30^a	−1.14	−7.60^a
Z₈	−.411	−2.16^a	−0.52	−2.69^a	−0.45	−2.42^a	−0.54	−2.82^a	−0.607	−3.18^a	−0.86	−3.05^a
Z₉	.266	1.33	0.346	1.69	0.29	1.51	0.36	1.80	0.19	0.96	0.32	1.09
Constant	−6.03	−11.68^a	−6.47	−11.4^a	−6.75	−11.81^a	−7.01	−11.96^a	−6.02	−11	−8.82	−10.82^a
$σ^{2}$	0.0037	−30.61^a	0.0041	−31.08^a	0.0042	−31.1^a	0.0025	31.33^a	0.00244^a	−30^a	0.0025	−10^a
$γ$	0.45	−11^a	0.576	−11.4^a	0.41	−11.81^a	0.47	−11.96^a	0.37	−11^a	0.4	−10.1^a
N	660		660		660			660		660		660
LogL	585.5		588.92		587.67		590	597.83		594.67		602
Wald Chi²	199.37		186.12		195.89		183.97	172.55		189.00		171.55

Notes: Dependent variable is hourly wages. ^a Indicates significance at least at the 5% level. The parameter $γ$ lies between zero and one. For details about $γ$ see Battese and Corra (1977).

OLS and maximum likelihood results are presented in Tables 3A and 3B.

We estimate several different frontier and inefficiency specifications to compare residual based (im)mobility measures to various indicators of mobility. It should be noted that the frontier will vary depending on the choice of the mobility indicators as the frontier and the model of mobility are estimated simultaneously. Column 1 of Table 3A reports the OLS results. Some of the important observations are the following: the role of kinship variables in the wage formation is very interesting: first, the proxy of social networks outside household (Ω₂) shows economically meaningful, statistically significant and positive impact on the potential wage of a worker. Second, the impact of household network (Ω₃) on wage formation is not meaningful. Third, the informality of recruitment process (Ω₁ has a meaningful impact on the mobility of workers and thereby on their ability to achieve of a greater percentage of the potential wages on offer. Finally, the SEMPDUM variable shows that self-employment and employment within household firms increase the potential wage of a worker in the informal sector.

Other findings are as follows: wages in the informal sector are positively related to age (LNAGE), and, hence, experience matters – though the relationship is not statistically significant. We find that there is a wage premium in the urban sector vis-à-vis the peri-urban sector as given by the UR variable. The marriage dummy (MAR), by reducing workers’ mobility, has negative impacts on potential wages and is statistically significant. There is also no evidence of gender bias against female workers (GENDER) in the informal sector. As opposed to the standard finding in the formal sector, wages are negatively associated with the highest level of education in the household (LE), which we take as a proxy for the social status of the household. Thus there is no evidence of additional returns from experience and social status of a household. There is also no evidence of gender bias in the informal sector. These are significant anomalies, which deserve further research in the future. One plausible rationale for the absence of education premium and gender bias is as follows: both female workers and workers from household with lower social status have a higher probability of choosing self-employment, or family employment, and use family ties for securing employment, which are noted to increase their maximal wages and also to achieve a larger percentage of their (larger) maximal wages.

The results in columns 4–7 establish that household factors responsible for mobility restrictions (L₁, L₂, L₃) do inhibit workers from achieving a larger fraction of their potential wages. Workers with a higher value of L₁, or L₂, or L₃ have significantly less mobility and, hence, achieve a smaller fraction of their maximal wages. We also find an inverse relationship between the type of past contract (J₁) and the frontier measure of incomplete mobility – in column 3. This is expected since the search cost of a worker with a daily contract is larger than the one with a longer contract. We note that the past search activities and experiences (Z₄, Z₅) have little consequences for achieving the potential wages in current jobs. The role of M₁ (reasons for past mobility) is also noted to be very limited for influencing current mobility.

Our frontier specifications compare the accumulated formal knowledge of a worker (K₁ and K₂) with the stochastic frontier estimates of incomplete mobility of a worker. These results are mixed: we note that literacy (K₁) increases mobility as workers are able to achieve a larger percentage of their potential wages, yet the effects are not statistically significant. We then drop K₁ for avoiding endogeneity and choose K₂ and find that schooling has a counter-intuitive (negative) effect on the ability of workers to achieve their potential wages. The role of education in the informal sector is thus a serious issue that will need a further investigation in our future research. Finally, when we consider workplace conditions and safety as ‘experienced goods’ we find Z₆, Z₇, Z₈ to be economically meaningful and statistically significant to reduce mobility as workers sacrifice potential wages for having experienced workplace safety. In a similar vein, the perception of a worker about the rule of law (Z₈) enhances one’s mobility and thereby motivates one to achieve a larger fraction of one’s potential wage. Quite paradoxically, Z₉, revealed injury responsibility of employers, does not have a statistically significant impact, which can partly explain the deaths of innumerable workers from fire incidents in factories of Bangladesh. This effect of Z₉ is an important behavioural anomaly that also calls forth further investigation.

Conclusion

We developed simplistic tests for our models to show how, if we are to understand informal sector employment issues, we need to study the social and economic circumstances and also informal institutions of informal sectors. It is certainly not the case that the formal sector is necessarily good (or, bad) and the informal sector necessarily bad (or, good). Yet, it is widely held that improvement in the informal sector will be a primary driver of urban poverty reduction in developing nations. We evinced that policies towards the informal sector would generally be better if they attempted to reduce labour market frictions resulting in relative immobility of workers. These reductions can result in significant amelioration in the welfare of informal workers without sacrificing competitive advantage of the informal business enterprises. To ensure that appropriate policies are put in place, the informal workforce needs to be recognised and understood by policy-makers. To ensure that the policy-makers are well-informed, it needs to be evidence-based. Our work underscores the need that greater priority to be given to the collection of data on informal employment at the micro level, which is a relatively new topic in labour statistics.

From the primary data, in conclusion, we noted the following: the monopsonistic exploitation of workers in the urban informal sector is caused by the inertia of workers to vote with their feet for better wages and working conditions, while exploitation of workers is known to shrink and working conditions improve if workers are highly mobile. So the key issue for us is the determinants of mobility of workers that will, in turn, prompt workers to optimally choose the ideal balance between work and lifestyle. If there are informational frictions, sociological and behavioural issues, workers can systematically fail to obtain an optimal choice of their working conditions and the inertia can perpetuate incorrect choices paving the ground for their monopsonistic exploitation. The labour market in the urban informal sector being a complex mechanism of kinship-based exchanges, individual and household attributes of workers, their employment conditions and family and social ties will impede labour mobility and, thereby, allow monopsonistic competitors to exploit workers by setting wages below workers’ productivity. For assessing the determinants of workers’ immobility and consequent monopsonistic exploitation, we develop a simple model to test monopsonistic exploitation, by modifying the stochastic frontier model, in the informal sector. We identified a set of premarket attributes of workers, their household characteristics and ties, kinship factors to explain their relative immobility and monopsonistic exploitation.

Footnotes

Appendix: Description of the survey

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Notes

References

Acemoglu

Shimer

(1999) Holdups and efficiency with search frictions. International Economic Review 40: 827–849.

Aigner

Lovell

CAK

Schmidt

(1977) Formulation and estimation of stochastic frontier production function models. Journal of Econometrics 5: 21–38.

Albrecht

Gautier

Vroman

(2006) Equilibrium directed search with multiple applications. Review of Economic Studies 73: 869–891.

Altman

(2005) Behavioral economics, rational inefficiencies, fuzzy sets, and public policy. Journal of Economic Issues 34: 683–706.

Balán

Browning

Jelin

(1973) Men in a Developing Society. Austin, TX: Institute of Latin American Studies, U.T.

Battese

Corra

(1977) Estimation of a production frontier model: With application to the pastoral zone of eastern Australia. Australian Journal of Agricultural Economics 21: 169–179.

Bhaskar

(1999) Minimum wages for Ronald McDonald monopsonies: A theory of monopsonistic competition. The Economic Journal 109(455): 190–203.

Bhaskar

Manning

(2002) Oligopsony and monopsonistic competition in labor markets. Journal of Economic Perspectives 16(2): 155–174.

Becker

(1996) Accounting for Tastes. Cambridge, MA and London: Harvard University Press.

10.

Ben-Horim

Zuckerman

(1987) The effect of unemployment insurance on unemployment duration. Journal of Labour Economics 5(3): 386–390.

11.

Burdett

Judd

(1983) Equilibrium price distributions. Econometrica 51: 955–970.

12.

Butters

(1977) Equilibrium distributions of sales and advertising prices. Review of Economic Studies 44: 465–491.

13.

Canagarajah

Sethuraman

(2001) Social Protection and the Informal Sector in Developing Countries: Challenges and Opportunities. Social Protection Discussion Paper Series, No 0130.

14.

Chandra

(1992) Informal sector in developing countries: A theoretical analysis. Unpublished Doctoral Dissertation. Johns Hopkins University.

15.

Chandra

Khan

(1993) Foreign investment in the presence of an informal sector. Economica 60(237): 79–103.

16.

Dickie

Gerking

(1998) Interregional wage disparities, relocation costs, and labor mobility in Canada. Journal of Regional Science 38: 61–87.

17.

Freeman

(2004) The Development of Social Network Analysis: A Study in the Sociology of Science. Empirical Press: Vancouver.

18.

Gale

(1987) Limit theorems for markets with sequential bargaining. Journal of Economic Theory 43: 20–54.

19.

Galenianos

Kircher

(2008) Directed search with multiple job applications. Journal of Economic Theory 114: 445–471.

20.

Gautier

Moraga-González José

(2005) Strategic wage setting and coordination frictions with multiple applications. Manuscript, Tinbergen Institute.

21.

Granovetter

(1973) The strength of weak ties. The American Journal of Sociology 78(6): 1360–1380.

22.

Granovetter

(1974) Getting a Job. Chicago, IL: University of Chicago Press.

23.

Greif

(1993) Contract enforceability and economic institutions in early trade: The Maghribi Traders’ Coalition. The American Economic Review 83: 525–548.

24.

Greif

(1998) The historical and comparative institutional analysis. The American Economic Review Papers and Proceedings 83: 80–84.

25.

Groot

Oosterbeek

(1994) Stochastic reservation and offer wages. Labour Economics 1: 383–390.

26.

Harris

Todaro

(1970) Migration, unemployment, and development: A two sector analysis. American Economic Review 60(1): 126–142.

27.

Hofler

Murphy

(1992) Underpaid and overworked: Measuring the effects of imperfect information on wages. Economic Inquiry 30: 511–529.

28.

Hofler

Murphy

(1994) Estimating reservation wages of workers using a stochastic frontier. Southern Rconomic Journal 60: 961–976.

29.

Hofler

Polachek

(1982) Ignorance in the Labor Market: A New Approach for Measuring Information Content. Proceedings of the 1982 Business and Statistics Section American Statistical Association, Washington, DC, pp. 422–425.

30.

Hofler

Polachek

(1985) A new approach for measuring wage ignorance in the labor market. Journal of Economics and Business 37: 267–276.

31.

Holzer

(1987) Hiring Procedures in the Firm: Their Economic Determinants and Outcomes. NBER Working Paper No. 2185.

32.

Houston

(2005) Re-valuing the fringe: Some findings on the value of agricultural production in Australia’s peri-urban regions. Geographical Research 43(2): 209–223.

33.

Jackson

Wolinsky

(1996) A strategic model of social and economic networks. Journal of Economic Theory 71: 44–74.

34.

Krumm

(1983) Regional labor markets and the household migration decision. Journal of Regional Science 23: 361–376.

35.

Lauermann

(2006) Dynamic matching and bargaining games: A general approach. Manuscript, Department of Economics, University of Michigan.

36.

Lippman

McCall

(1979) Studies in the Economics of Search. Amsterdam: North Holland.

37.

Loayza

(1994) Labor Regulations and the Informal Economy. Working Paper No. 1335, Washington, DC: World Bank Policy Research, August.

38.

Maloney

(2001) Self-Employment and Labor Turnover: Cross-Country Evidence. Proceedings. Washington, DC: World Bank Economists Forum, World Bank.

39.

Maloney

(2003) Informal self employment: Poverty trap or decent alternative. In: Fields

Pfeffermann

(eds) Pathways Out of Poverty, Private Firms and Economic Mobility in Developing Countries. Boston, MA: Kluwer Academic Publishers.

40.

Maloney

(2004) Informality revisited. World Development 32(7): 1159–1178.

41.

Maloney

Nunez

(2003) Measuring the impact of minimum wages, evidence from Latin America. In: Heckman

Pages

(eds) Law and Labor Markets. National Bureau of Economic Research, NBER Working Paper 9800. Chicago, IL: University of Chicago Press, pp. 108–130.

42.

Maloney

Ribeiro

(1999) Efficiency Wage and Union Effects in Labor Demand and Wage Structure in Mexico. Working paper 2131, Washington, DC: World Bank.

43.

Maloney

Cunningham

Bosch

(2003) Who Suffers Income Shocks during Crises: An Application of Quantile Analysis to Mexico 1992–1995. Working Paper, Washington, DC: World Bank.

44.

Meeusen

van den Broeck

(1977) Efficiency estimation from Cobb-Douglas Production Functions with composed error. International Economic Review 18: 435–444.

45.

Mincer

(1978) Family migration decisions. Journal of Political Economy 86(5): 749–773.

46.

Mortensen

(1977) Job search, the duration of unemployment and the Phillips curve. Industrial and Labor Relations Review 30: 505–517.

47.

Mortensen

Wright

(2002) Competitive pricing and efficiency in search equilibrium. International Economic Review 43: 1–20.

48.

North

(1990) Institutions, Institutional Change and Economic Performance. New York: Cambridge University Press.

49.

North

(1994) Economic performance through time. American Economic Review 84: 359–368.

50.

Pissarides

(2000) Equilibrium Unemployment Theory. 2nd edition. Cambridge, MA: MIT Press.

51.

Polachek

Robst

(1998) Employee labor market information: Comparing direct world of work measures worker’s knowledge to stochastic frontier estimates. Labour Economics 5: 231–242.

52.

Polachek

Yoon

(1987) A two-tiered earnings frontier estimation of employer and employee information in the labor market. Review of Economics and Statistics 69: 296–302.

53.

Polachek

Yoon

(1996) Panel estimates of a two-tiered earnings frontier. Journal of Applied Econometrics 11: 169–178.

54.

Robinson

(1933) The Economics of Imperfect Competition. London: Macmillan.

55.

Satterthwaite

Shneyerov

(2007) Dynamic matching, two-sided incomplete information, and participation costs: Existence and convergence to perfect competition. Econometrica 75(1): 155–200.

56.

Shields

(1993) A theoretical and empirical analysis of family migration and household production: U.S. 1980–1985. Southern Economic Journal 59: 768–791.

57.

Shimmer

(2005) The assignment of workers to jobs in an economy with coordination frictions. Journal of Political Economy 113(5): 996–1025.

58.

Stigler

(1962) Information in the labor market. Journal of Political Economy 70: 94–105.

59.

Topel

(1986) Local labor markets. Journal of Political Economy 94: S111–S143.

60.

Van Ours

Ridder

(1992) Vacancies and the recruitment of new employees. Journal of Labor Economics 10(2): 138–155.