The Home as Factory Floor: Employment and Remuneration of Home-based Workers1 Wendy Cunningham and Carlos Ramos Gomez The World Bank Home-based work, defined as non-professionals who perform market work from their homes, is an increasingly recognized form of employment in Latin America. The majority of the research on this segment of the labor force relies on small sample, qualitative data, which find that home-based workers are women, children, and adults with disabilities with low skills who work long hours for low wages. Using a large random sample, control groups of non-home-based workers, and including men in the analysis, this paper examines the home-based work sector in Brazil, Mexico, and Ecuador in 1999. The results show that in all three countries, women are over-represented among home-based workers, particularly older women, those with low levels of education, and those with children or spouses, unlike men for whom these factors do not matter. Female home- based workers earn 25-60 percent less per hour than do non-home-based working wome and they work one-third to one-half as many hours each week. Home-based working men, on the other hand, earn 0-17 percent less than do men who do not work from their homes, and they only work 10 percent fewer hours per week. The wage and work hour gaps for women are largely related to marital status, not the presence of children, suggesting that simply being the primary caregiver in the households, regardless of the actual time constraints (children) is the key factor to differences between home-based working women and those who work outside of their homes. 1Prepared for the Latin American and Caribbean Annual Study: "From Natural Resources to the Knowledge Economy." We would like to thank the Bank-Netherlands Partnership Program's Economic Policy and Gender initiative, which provided funding for this study. Also, thanks to William Maloney for valuable feedback. All errors, of course, are the sole responsibility of the authors. I. Introduction Home-based work has recently gained increasing attention, particularly in the context of globalization. One line ofliterature argues that as firms compete to survive in the global economy, they seek out modes of production that maximize flexibility in terms of production schedules and factor inputs while generating quality outputs at internationally competitive prices. This requires a reorganization of production processes such that the final assembly, packing, data entry, or other low skilled tasks occurs less in factories and more in private homes through contracted-out production (Carr 2000, WIEGO 2000).2 The absence of a permanent labor force or fixed capital allows firms to easily change production methods, the goods produced, or the location of production, thereby lowering costs (ILO 1995, Prugl 1997, Benería and Roldón 1987). Additionally, a portion of the labor force that has constraints to working outside of the home ­ either due to limited mobility (people with disabilities) or due to frequent duties that must be performed throughout the day in the home (primary caregivers in the household) ­ provides a large pool of potential home-based workers, thereby depressing wages and further reducing costs to employers. Other literature counters that home-based work is a preferred organization of labor force participation for workers who have difficulty working outside of their homes, and is only facilitated by, rather than caused by, globalization (ILO 1995). This argument is not easily resolved since there is little evidence of the motivation for or the nature of home-based work. While economists are increasingly recognizing the 2The reorganization of production to minimize costs at all levels of production has lead to the development of "global value chains", which are multi-level production processes that are headed by multinational corporation and move around the globe searching for inexpensive production costs. Carr, Chen and 2 heterogeneity of the informal sector and the need to analyze each of its segments independently, their studies do not discuss the home-based work sector (Cunningham and Maloney 2001, Maloney forthcoming). This is partly due to the absence of large sample data sets that would allow for economic analysis (Chen 1999). Instead, the majority of the evidence is derived from qualitative case studies and small-sample interview data by sociologists and anthropologists (Chen 1999, Prugl 1997, Carr 2000). The stylized picture that is forwarded by these researchers is that home-based work is primarily undertaken by women and their children (Chen 1999) who perform simple, repetitive, labor-intensive tasks (Arriagada 1998) for long hours and low wages. Since the work is performed in the privacy of one's home, the workers are subject to poor work conditions, few legal protections,3 and isolation (WIEGO 2000). While some of these conclusions are indisputable, others would benefits from further analysis with a large, random sample. Jointly analyzing the data on home-based workers and data from a control group of workers with similar characteristics who are not home-based workers will reveal whether home-based workers are paid less and work more because of their particular labor market constraints, or if they are doing just as well as non-home-based workers with similar constraints. This paper uses household surveys in Brazil, Ecuador, and Mexico to empirically test the findings from the small sample ethnographic research. The paper addresses three primary questions: who are home-based workers?, are they underpaid?, and do they work extraordinarily long hours? The analysis selects comparison groups of non-home based workers and attempts to control for characteristics that may be responsible for the O'Connell (1999) give a full exposition on this concept. 3Many countries have legislation governing market work that is performed in the home (ILO 1995), but 3 observed differences in the qualitative analysis. Particular attention is given to sex, marital status, and children variables, in order to further examine their role in constraining labor market choice, as suggested by the qualitative research. The next section defines home-based work and lays out the challenges to analyzing the sector. Section III presents the findings from the small sample, ethnographic data. Section IV discusses the data, research methodology, and the mean characteristics of the samples. Section V uses the large samples from household surveys to empirically test the hypotheses. Section VI summarizes the findings and discusses public policies to increase the efficiency of the sector while protecting workers' rights. II. Defining home-based work Home-based work is a difficult empirical concept due to an absence of a single definition of the sector and a lack of data to aid in the development of one (Prugl 1997). The most general definition of a home-based worker is an individual who works in his or her home (WIEGO 2000, Arriagada 1998), encompassing the self-employed;4 piece workers; salaried employees who work for a "middle-man" or a firm; or unpaid workers in a family enterprise. Refinements have led to a wide-range of concepts to identify a home-based worker, for example: § DegreeofSubordination: aworkerwhose"client"hascontroloverthe worker and the production process (Prugl 1997, ILO 1995); reporting of violation of the laws and enforcement is judged to be low. 4The self-employed may be defined as those who own their own firm but do not have any paid employees. The only employees would be family members who do not earn a wage. 4 § LevelofRisk: aworkerwhoreceives a fixed salary from the "client" and is not subject to the risk associated with price or demand fluctuations (Prugl 1997); § NatureofOutput: workers in industrial home-work,craftproduction,making and selling of foods, and telecommunications-based services (Prugl 1997); § OwnershipoftheProductionProcess: assembly where the inputs and production tools may be provided by the employer or by the employee (Pollack 1998, Arriagada 1998); § Skilllevel: those who work out of their homes but are not professionals (the latter are classified as entrepreneurs); and § TypeofClient: those whose client is an intermediary (i.e. not the public) as opposed to those who sell directly to the public (ILO 1995). The wide variation in definition leads to two problems in analysis. First, many are difficult to quantify. For example, "degree of subordination" or "level of risk" can only be proxied by a number of other characteristics about autonomy in decision-making. Also, the workers themselves may not have the necessary information for classification; she/he may not know if the output is sold to an individual for own use or if that buyer will turn around and re-sell it to another client. Only firm surveys that follow a product through the global value chain can provide these type of information. Second, studies of home-based workers often use different definitions, thus hindering the construction of a single, coherent picture of the sector. Seemingly contradictory findings from qualitative data may in fact be a result of different methods of measurement rather than true differences in the realities studied. The absence of single 5 set of characteristics to define the sector makes it difficult to design surveys in order to systematically collect data on the sector (Chen, Sebstad, and O'Connell 1999). III. Lessons from the Qualitative Data: A Literature Review Keeping in mind these measurement difficulties, Table 1 shows that the proportion of the labor force that engages in home-based work ranges from 1.6 percent to 23 percent of the working population across the world. An array of definitions and questions are used to construct these data, thus making comparisons meaningless. Nonetheless, it should be noted that home-based work exists in both developed and developing countries.5 Table 1 also suggests that this organization of work does not necessarily arise from the globalization process since some of the countries in the table had sizeable home-based work sectors prior to lowering their trade barriers. 3.1 Characteristics of home-based workers The ethnographic research suggests that home-based workers are primarily individuals who are unable or unwilling to work outside of their homes. Generally, workers who are confined to the home fall into three general categories: women (or men, to a lesser extent) with children or other household responsibilities that require them to be near home throughout the day, those who have physical or mental conditions that limit their ability to travel to a worksite, and individuals who have difficulty entering the labor force due to legal constraints, discrimination, or information asymmetries. 5It should be noted that the conditions of and remuneration for home-based work differs between countries. Thus, a comparison of the incidence of home-based work does not allow us to draw inferences about the quality of the employment nor to pass welfare judgments on the quality of labor in the countries in Table 1. 6 Regarding the first category ­ those with household responsibilities - more than half of the home-based workforce across the world is female; estimates range from 64 percent to over 90 percent, compared to 40 percent of the non home-based workforce (ILO 1995, Chen, Sebstad, and O'Connell 1999). In Latin America, female home-based workers report that flexible work hours and the home as the work site allow them to more efficiently fulfill their home and market responsibilities than if they worked in a remote production location (Prugl 1999, Tomei 2000, Arriagada 1998). Benería and Roldón (1987) report that 60 percent of Mexican home-based workers in their sample would decline the offer of a better paying factory job in favor of the time and location flexibility of working from the home while Jelin (2001) reports that her sample of Argentine female home-based workers with children hope to find employment outside the home once their children leave home. The constraints faced by these women due to their roles as caregivers in the absence of or the prohibitive cost of market or public alternatives to substitute for the women's home care ­ such as child care, running water, labor saving household appliances ­ make home-based work an optimal form of market participation; without these constraints, they may choose to work outside of their homes. An alternative hypothesis is that even if women do not have childcare responsibilities, those from more traditional households may have difficulty obtaining permission from their husbands (or themselves) to work outside the home (Benería and Roldón 1987). When traditional gender roles ­ a husband who works in the market to materially provide for the family and a wife who cares for the household - guide the division of household labor, a husbands may feel threatened or embarrassed if his wife also works in the market, since it may signal his inability to perform his role as sole 7 breadwinner of the family (Akerlof and Kranton 2002). Home-based work can be performed clandestinely, though. Also, market work in the home may be less valued than work outside of the home - husbands and the women themselves who work from the home often classify these earnings as women's "pocket money", even when it is the primary source of household income (Geldstein 2001) - thereby maintaining the appearance that the husband is adequately performing his breadwinning role. A second group of people for whom home-based work may be optimal, given their constraints, are those whose physical or mental capacities make it difficult to travel to external work sites or to perform in workplaces that are not adapted for their particular needs. They may include individuals of any age with disabilities (ILO 1995) and older people who have lost mobility with age (Carr 2000, Tomei 2000). Again, while home- based work may not be the preferred type of work, it may offer the best opportunity for employment under the existing constraints. Finally, home-based workers may be those who have difficulty entering the regular labor market due to legal constraints, discrimination, or a lack of information. Women, migrants, children, low-skilled individuals, and older workers tend to have difficulty finding jobs, as demonstrated by high unemployment rates among these groups.6 This may be due to demand side factors, since employers often assume that these groups are less skilled due to little job experience, low education levels, innate ability (i.e. pure discrimination), or skill obsolescence and are more costly due to legal limitations (child labor, illegal migrants). Since home-based work is low-skilled, hidden from the employer's eyes (thus circumventing the legal issue), and often paid by output (so skill is less important), employer bias may be less prevalent in this work. On the 8 other hand, these individuals may have less information about job opportunities or how to search for jobs, particularly among women, migrants, and children, so they accept piece work that is handed out in their marginal neighborhoods, thus never fully searching the labor market (ILO 1995). Both of these scenarios possibly contribute to the existence of a home-based labor force and a potentially ologoposonistic-type labor market; i.e. a few potential employers who may collude to keep wages low and benefits at a minimum.7 3.2 Nature of the work The production processes and outputs vary by home-based workers, but most of the qualitative studies agree on two characteristics: home-based workers earn low wages and they work more than the legal number of hours each week.8 For example, Beneria and Roldon (1987) find that 90 percent of the women in their sample earn less than the minimum wage; other studies support this assertion (WIEGO, Carr 2000, ILO 1995). WIEGO (2000) and ILO (1995) find that home-based workers have work shifts that are much beyond the legal work week, as do Benería and Roldón (1987), for home-based workers who provide most of the household income. However, Jelin (2001) finds the opposite result for Argentina home-based workers, whose weekly work shifts are 2/3 of those of shift workers. 6Authors' calculations; available upon request. 7Empirical evidence for oligosonies in the home-based work is scarce, but the potential for such a market does exist since home-based workers in a particular community tend to work for a the same employer, largely due to an absence of information about alternative employers that would permit the workers to compete in a competitive labor market. 8Multiple papers address other characteristics of home-based work that are beyond the empirical scope of this paper. For a discussion of these characteristics, please see: regulation (ILO 1995, Tomei 2001), employment benefits (ILO 1995, WIEGO, Beneria and Roldon 1987), cyclicality and riskiness of work (ILO 1995), occupational health hazards (WEIGO 2000), capital (Arriagada 1994, Carr 2000), type of work (Arriagada 1998), types of output (Chen 2000, Prugl 1997, Arriagada 1998, Beneria and Roldon 1987), skills (Prugl 1997), market access (Prugl 1997), gender roles (Beneria and Roldon 1987), global value chains (Chen 1999, Carr 2000), and the social context (Beneria and Roldon, 1987), to name a few. 9 IV. Data, Methodology and Sample Characteristics 4.1 Data The remainder of the paper uses household survey data from Brazil, Ecuador and Mexico to test the qualitative findings. These countries were selected solely on the basis of data availability ­ they were the only labor market surveys for Latin America that could be obtained and that permitted identification of market work in the home. The data sets used are: the Brazilian National Survey of Households (Pesquisa Nacional de Amostra do Domicilio - PNAD) for 1999; The Mexican National Urban Employment Survey (Encuesta Nacional de Empleo Urbano - ENEU) for 1999 (second quarter); and the Ecuadorian Living Conditions Survey (Encuesta de Calidad de Vida - ECV) for 1999. For all countries, the sample is limited to urban men and women who are between the ages of 15 and 70 years old.9 Men and women are analyzed separately since the qualitative analysis suggests that they have very different patterns in home-based work and distinct motivations for engaging in this organization of work. The regression analysis only includes those who have no more than a secondary education, thus excluding professionals. The resulting sample size for Brazil is near 60,000 and for Mexico is close to 100,000, while the Ecuador sample has fewer than 3,000 individuals. Due to data limitations, the most basic definition of home-based worker is used: an individual whose place of employment is his/her home. The surveys permitted identification of four general types of home-based workers: 9The motivation for the spatial limitation is two-fold: First, the ENEU only covers the urban population so to allow for some standardization across countries, the data was limited to urban population in the other two sample countries. Second, the nature of home-based work in rural areas is more complex than in urban areas due to the great overlap in home and market production activities, particularly in agricultural households, and the greater in-kind payment of rural work. 10 § Informalself-employed. Individualswhodeclarethemselvesself-employed, work at home, and have no more than secondary education completed. § Professional. Individualswhodeclarethemselvesself-employed or employees, work at home and have more than secondary education completed. § Contract. Individuals who declare themselves as employees, work at home for a wage, have no more than completed secondary education, and work in the manufacturing sector.10 § Unpaid. Individuals who work from their homes and are not remunerated. Table 2 shows the proportion of home-based worker in each sample country by sex and type of home-based worker for 1999. The "other" category consists of firm owners, service workers (including domestic servants), and agricultural laborers. Home-based workers are a substantial fraction of the work force. In 1999, approximately five percent of the Brazilian and Mexican labor force (and 10 percent of the informal sector) and more than 17 percent of the Ecuadorian labor force worked out of their homes. Women are more than three times as likely as men to be home-based workers in all three countries. Most home-based workers are informally self-employed. In Brazil and Mexico, approximately three percent of the labor force are self-employed individuals with less than a secondary education who work out of their homes. In Ecuador, the proportion is nearly double with over seven percent in this category. Women are more likely than men to hold this position. The other categories of home-based workers constitute less than 10A very large portion of employees with no more than a secondary education who claim that they work at home are, in fact, domestic servants, who, perhaps, are not home-based workers in the sense of piece or assembly type jobs. To limit the sample to those in the more traditional form of home-based work, the category was limited to the manufacturing sector. 11 one percent of the labor force with the exception of Ecuador, with approximately four percent of the labor force in unpaid home-based work.11 Due to the small number of professionals and unpaid workers, and their unique situations ­ professionals have high human capital and unpaid workers may be considered as having misreported their earnings12 ­ the rest of the analysis will only consider informal entrepreneurs and contract workers who have no more than a completed secondary level of education. 4.2 Demographic Characteristics of Home-based Workers in the Sample Women are over-represented in the home-based work sector in all three countries analyzed (Tables 3a-3c). In Mexico, while women comprise 35.5 percent of all workers, they are nearly 64.8 percent of informal entrepreneurs and 41.9 percent of contract workers. The proportions are higher in Brazil, with 74.7 percent and 81.7 percent, respectively, as compared to 39.6 percent in the rest of the economy, and Ecuador with 74.1 percent and 61.5 percent, respectively, compared to 34.9 percent of all workers. In general, home-based workers have less human capital than do workers in the rest of the economy. Informal entrepreneurs tend to be older and thus have more potential work experience, than workers in the rest of the economy while contract 11Data from 1993 in Brazil, 1994 in Mexico and 1995 in Ecuador show that the proportion of home-based workers declined over the second half of the 1990s in Brazil, but increased in Ecuador and Mexico. In Brazil, home-based workers were over seven percent of the labor force in 1993, but this declined to 5.5 percent by 1999, primarily due to a decrease in informal self-employment and the "other" category. Most of the decrease was seen among women, whose share fell from 13.2 percent of the female labor force to 9.9 percent seven years later. In contrast, home-based work has increased slightly in Mexico and Ecuador. This is due to an increased propensity of both men and women to be home -based workers in the later period. In Ecuador, the growth occurred in the unpaid sector while in Mexico, both men and women increased their informal self-employment. 12Unpaid workers may be considered as misreporting their earnings since most work in family firms and thus receive in-kind remuneration. For example, the household head may own the household firm and report firm profits as his/her income, but she/he uses this income to buy food, clothing, lodging, and other material goods for other household members who were partly responsible for generating this income 12 workers are mixed. Only 8.5 percent of Mexican informal entrepreneurs, 11.6 percent of Brazilian, and 7.1 percent of Ecuadorian entrepreneurs are in the age category 15-25 while approximately 30 percent of the rest of the economy is. However, in Mexico, nearly half of all contract workers are in the young age category while Brazilian contract workers are more likely to be in the age category 26-45, i.e. prime-aged workers and in Ecuador they tend to be in the oldest category: 46 years or more. Education levels also tend to be lower than those of the rest of the economy, especially among contract workers. Married women and people who are not head of the household are most likely to be home-based workers. In Mexico, 35.5 percent of the informal entrepreneurs and 20.4 percent of the contract workers are wives while in Brazil, the proportions are 49.7 percent and 60 percent, respectively, and in Ecuador they are 31.2 percent and 23.1 percent, respectively. In comparison, only 13.8 percent of Mexican workers, 19.6 percent of Brazilian workers, and 13.3 percent of Ecuadorian workers whose workplace is outside of the home identify themselves as married women. The only exception that emerges from the data is a greater proportion of Mexican contract workers who identify themselves as unmarried males than as wives: 34.4 percent as opposed to 20.4 percent. Less than half of the home-based workers are identified as household heads. Only in Mexico are informal entrepreneurs as likely to be household heads as are workers in the rest of the economy. In the other countries, they are 6-10 percent less likely to be head of the household. Contract workers are even less likely to head the household, with approximately 20 percent in each country being the household head. This reflects the high proportion of women and wives in this category. through their unpaid labor. 13 Contract workers are more likely to have children age 0-12 in their homes in Mexico and Brazil than do workers in the economy as a whole or informal entrepreneurs. Approximately 29 percent of Mexican workers have children age 0-5 in their households while more than 40 percent have children age 6-12 or 13-18 living in their home. The proportions are similar among Mexican home-based informal entrepreneurs, but the proportions of children age 0-5 and 13-18 are 10 percentage points higher in Mexican contract worker homes and 20 percentage points higher for children age 6-12. Similar differences emerge in Brazil. In Ecuador, home-based workers are a few percentage points more likely to have children in the household than are workers outside of the home, but the differences are not as large as in Mexico and Brazil.13 4.3 Employment Characteristics of Home-based Workers in the Sample Home-based workers are less likely to be in the formal sector than are non-home- based workers, but the proportions differ among the three countries in the sample (Tables 4a-4c). In Brazil and Ecuador, approximately seven percent of contract workers receive benefits from their employers, compared to 58.4 percent of Brazilians and 31.7 percent of Ecuadorians who do not work out of their homes. On the other hand, nearly 20 percent of Mexican home-based workers report having a formal employer, as opposed to 54.9 percent of the other workers.14 13The sample size of contract workers in Ecuador is small, so the proportions are not significantly different between contract workers and the other categories of workers. 14This difference between countries is likely driven by the definition of "contract worker" created from the data sets. The ENEU explicitly asks if an individual is a contract workers while this category was defined for Brazil and Ecuador as the home-based worker who identified his/her occupational category as "manufacturing." 14 Contract workers who work from their homes are more likely to be employed by small firms than are those in a formal work site.15 Forty-five to 80 percent of home- based contract workers are employed in firms with five or fewer employees while the proportions are more than twenty percentage points less for non-home-based workers. In contrast, the share in firms with more than 10 employees (15 employees, for the case of Mexico) is lower for home-based than non-home-based workers in Mexico and Brazil, while the proportions are nearly equal in Ecuador. This high proportion in small firms may be due to the contract worker's identification of the "middle men" as their employers, who commonly receive orders from firms then farm out the work among producers (Arriagada 1998). Both contract workers and informal entrepreneurs who work out of their homes have lower average hourly earnings than do workers who do not work from their homes. Using the legal minimum wage in the first period for each country as a unit of measure (thereby not allowing comparisons across countries), a higher proportion of home-based workers earn less than one minimum wage than do non-home-based workers while a lower proportion earn more than three minimum wages. Finally, home-based workers report working fewer hours on the job than do individuals who work outside of their homes. In Brazil and Mexico, home-based entrepreneurs and contract workers spend approximately 36 hours weekly in their jobs, compared to nearly 46 hours among Mexican and 37 hours among Brazilian non-home- based workers. In Ecuador, the disparity is even larger with home-based workers reporting work weeks of 32.3 hours compared to 43 hours among those who work outside 15Informal entrepreneurs, by definition, own small firms, so even if they sub-contract, the survey does not ask them the size of the firm with which they contract. 15 of the home.16 This is similar to the finding by Jelin (2001) in Argentina but contradicts other qualitative literature (WIEGO 2000, ILO 1995). The contradiction may be a result of reporting errors due to the sporadic nature of home-based work, difficulty in identifying activities as market or non-market based, the multiple jobs that home-based workers often hold (Benería and Roldan 1987), the intermixing of market and home activities that makes it difficult to estimate time spent in market work in the reference week, and the participation of other household members in the production. 4.4 Methodology Simple regression methodologies are used to better identify the characteristics that lead to home-based work and whether or not wages and hours are lower than non-home- based workers, conditioning on their propensity to be less educated and to have family duties. The sample is the entire workforce, age 15-70, who identify their place of work, declare a wage and declare their hours worked; the sample is described in tables 3a-4c. The characteristics of home-based workers are identified by using a logit estimate of the probability of being a home-based worker, where home is a dummy variable that takes a value of 1 if the individual works in his or her own home and 0 if not.17 The dependent variable is regressed on the demographic variables given in Tables 3a-3c. Female takes a value of 1 if the individual is a female. The variables age15, age2645, and age46 are dummies that take a value of 1 if the observation's age is within the range; 16The hours worked per week decrease between 1995 and 1999 for home-based workers but not those who work outside of the home. 17In Mexico, the ENEU asked to classify the establishment where work-related activities took place, having "in your own home" as one option. In Brazil, they PNAD asked "The work establishment is located in...." the chosen option is "the address where I lived." In Ecuador, the question is "During the last week, you worked as .... in" and "your house" option was chosen for the purpose of home workers identification. 16 the last dummy is omitted. For all countries, nosch is a dummy that takes a value of 1 if the individual never attended school; pri is a dummy variable that takes a value of 1 if the individual completed no more than six (eight) years of school in Mexico and Ecuador (Brazil). The omitted education dummy is 7-12 years of school for Mexico and Ecuador and 9-11 years of school for Brazil. Head is a dummy that takes a value of 1 if the individual is reported as the household head. To control for household constraints and/or additional potentially productive household labor, the dummy variables presch, elemsch, teen, and adult are included. Presch takes a value of 1 if there are children aged 5 or younger in the household, elemsch takes a value of 1 if there are children aged between 6 and 12 in the household, teen takes a value of 1 if there are children aged between 13 and 18 in the household, and adult takes a value of 1 if there are (non-spouse) adults age 19 or older in the household. The married variable takes a value of 1 if the person declared being in a marital or consensual union and a 0 otherwise. For the Brazilian regressions, the dummy variable non-white is included to control for race, where the variable takes a value of 1 if the individual self-identifies him/herself as black (preto) or mixed race (pardo); this variable was not available for Mexico and Ecuador. The regression is estimated three times for each country; first with the whole sample (where the female variable is of most importance) and then by sex.18 To test whether or not home-based workers earn less than other workers simply by being based out of their home or if the observed wage differentials are due to human capital characteristics, the log hourly wage (income, for the self-employed) earned by home-based and non-home-based workers is regressed on variables that determine labor 18Additional specification that interact sex with marital and child variables were also estimated. The results are presented in footnotes, where relevant. The full set of regressions can be obtained from the 17 market productivity or that employers may use to price discriminate between workers: age, race, sex, and education level. The variable of interest in the regression is home.19 The regression is estimated separately for the pooled sample, men, and women. The exercise is repeated by estimating log hourly wages conditional on being in the labor force. A Heckman selection two-step procedure is used, where the first step is to estimate the likelihood of being in the labor force, a correction factor is calculated for each observation, and it is included as an independent variable in the wage equation. Finally, to test whether or not home-based workers have longer or shorter work shifts than do those with similar time constraints and who work outside of the home, the log of weekly hours worked is regressed on sex, age, headship, and other household characteristics that may constrain labor market time (marital status, the presence of children of different ages, the presence of other adults), and the home dummy. Again, the home dummy is the key variable of interest. V. Regression Results 5.1 Characteristics home-based workers The regression results for all three countries strongly support the findings in the qualitative literature that women dominate the home-based work sector (column 1 in Tables 5a-5c). Even when controlling for households constraints that more often limit women's rather than men's time use ­ the presence of children or a spouse- women are authors. 19It may be argued that individuals who become home -based workers have certain characteristics that are associated with low earnings, and thus the regression may be spuriously attributing low earnings to home- based work while it is actually due to other causes. To correct for this, an instrument for home-based work would need to be identified, but such a variable was not found in the data. 18 4.4 percent, 8.8 percent and 22.9 percent more likely to be home-based workers in Mexico, Brazil, and Ecuador, respectively.20 The results also support the qualitative findings that those with greater domestic responsibilities are more likely to engage in home-based work, namely women with household responsibilities and older individuals. Unambiguously, women with household constraints are more likely to be home-based workers than are women who do not have spouses or children (Columns 3 in Tables 5a-5c).21 In Brazil and Mexico married women are approximately three percent more likely to work from their home than are unmarried women. In Ecuador, the coefficient is positive, but not significant, possibly due to the small sample size.22 With respect to children, Mexican and Brazilian women with children age 0-5 (presch) were about two percent more likely to be home- based workers than were women without young children in their households and those with children age 6-12 (elemsch) were 1.0 percent more likely to be home-based workers. In Ecuador, the coefficient estimate on children age 0-5 is positive and significant at the 10 percent level, thereby weakly suggesting that these women are six percent more likely to be home-based workers. Men, on the other hand, show weak tendencies for home- based work as a result of household characteristics: their participation in home-based work is uncorrelated or negatively correlated with marriage and the presence of young children. 20In Mexico and Ecuador, these likelihoods increased between 1994 (1995 for Ecuador) and 1999 while in Brazil, the probability fell. 21Coefficient estimates on interaction terms of sex and married, sex and presch, and sex and elemsch in the pooled sample are highly significant and greater than zero for the Mexico and Brazil data, but not for Ecuador, with its much smaller sample size. 22Using 1995 data from Ecuador, despite the small sample size, Ecuadorian married women were 6.9 percent more likely to be home -based workers; the coefficient was statistically significant at the 5% level. 19 Even when controlling for the presence of children, less educated women have a higher propensity for home-based work in Brazil and Mexico (12 and six percent for nosch and primsch only) while the education coefficient for men are very small. The strong effects for women and absent effects for men suggest that education alone does not lead to home-based work. Instead, perhaps the combination of being female and uneducated leads to a higher value for her domestic work since uneducated families tend to be poor (Wodon 2000) and poor families have fewer labor saving devices and more children (Wodon 2000). Older individuals are more likely to be home-based workers, thus supporting the qualitative findings that those with obsolete market skills and limited mobility tend to work at home. The probability of being a home-based worker is monotonically increasing with age for men and women in all three countries, with young women being particularly unlikely to accept this type of work in Brazil and Ecuador. This reflects the propensity for the informal self-employed, who are a sizeable portion of the home-based worker in the sample, to be older workers (Cunningham and Maloney 2001). There is less evidence that those with limited access to the labor market, whether due to demand or supply constraints, leads people to home-based work. The regression using the Brazilian data shows that non-white individuals are neither more nor less likely to be home-based workers compared to the (favored) white workers. 5.2 Hourly remuneration Both men and women who are home-based workers earn lower hourly wages than do non-home-based workers in all three countries. When controlling for human capital differences, Brazilian home-based workers earn 22 (21), 26 (20), and 19 (21) percent less 20 per hour than do non-home-based workers in Brazil, Ecuador, and Mexico, respectively, with selection corrected estimates presented in the parentheses.23 The wage premium to worksite wages is larger among women than men, where women who work from their homes earn approximately 26 (11), 41 (21), and 27 (29) percent less than non-home- based Brazilian, Ecuadorian and Mexican women. Men's discrepancies, on the other hand, are in the range of a 0-17 percent premium to non-home-based work in all three countries.24 While children and a spouse decrease the earnings for working women in general, only the presence of a spouse decreases the wages for home-based working women even more. The coefficient estimates for the interaction of the home and the married variable is negative in all three countries and significant in Brazil and Mexico. However, the interaction term between home and the child variables is not significant for any country. The wages of men who worked from and outside of their homes did not differ by marital or child status. Perhaps the gap for women reflects a compensating wage differential. Qualitative interview data in Latin America show us that women prefer home-based work due to the flexibility and opportunity to fulfill both home and market responsibilities, so maybe women who work outside the home are compensated for the foregone goods and services (for auto-consumption) that they would have produced had they not had to spend time traveling to work or could take breaks from market work to perform domestic work 23When breaking apart the sample into type of home-based worker, Mexican and Ecuadorian home-based wage workers earned 38 and 61 percent less than wage workers outside of the home, with the differential between home-based and other self-employed was only half as large. In Brazil, the differential for home - based and outside workers was the same for wage and self-employed workers (regression results available from authors upon request). 24The exception is the wages of Ecuadorian men, with a 50 percent difference. 21 throughout the day. Women working in the home "pay" for the benefits of flexibility in work schedules and locations via a lower wage.25 5.3 Work shift duration Contrary to the qualitative findings, but supporting the unconditional estimates in tables 4a-4c, the data show that home-based workers have shorter work weeks than do non-home-based workers. Even when controlling for proxies for time constraints (presence of children and marital status), in Mexico and Ecuador, their hourly work weeks are approximately one-third less than the time that non-home-based workers spend on the job while in Brazil, they spend about one-fourth less time at work. This is mainly due to women's shifts, with women in all three countries working 32 ­ 46 percent fewer hours than women who work outside the home. Male home-based workers, on the other hand, work 9-18 percent shorter work weeks than their non-home-based counterparts.26 In Mexico and Brazil, family structure affects the time that both men and women spend in home-based work, but their time spent on the job is still lower than that of those who work outside the home. Men who are home-based workers and are married or have children work more hours per week than do home-based working men without a spouse or children. Conversely, married women work about 3 percent less than do home-based working women who are not married. Surprisingly, the presence of children does not 25The wages of home -based workers has fallen over time in all three countries, but especially in Mexico. In Mexico, home-based workers earned eight percent less than non-home-based workers in 1994, but this nearly tripled by 1999. The change was the largest for men, whose differential increased eightfold while women's only doubled. However, women's new differential in 1999 was 27.8 percent compared to men's differential of 10.2 percent. In Brazil and Ecuador, on the other hand, the difference between home -based and non-home-based hourly wages increased by five percentage points. The change was primarily felt in the differential between home-based and non-home-based women, not men. 26The discrepancy between work shifts is decreasing over time, in Mexico and Brazil, but it has increased in Ecuador. 22 affect the time women spend in home-based work any more than it does for those working outside of the household (the exception is pre-school children in Brazil). VI. Conclusions This paper has shown that many of the findings on home-based work from qualitative studies can be supported by large, randomly sampled quantitative surveys. Women are more likely to be home-based workers. This is due to their roles as mothers, wives, and women. Wives select themselves into this type of work in order to better balance home and market work, but sacrifice higher wages in exchange; children seem to have little impact on hourly wages or hours worked for women. Women with low levels of education, who are likely to subscribe to more traditional gender roles and have fewer exit options, also tend to enter this type of work, perhaps in order to preserve their own and their husband's gender identities. Women in home-based work also are more likely to be part-time workers, allowing them to hold multiple jobs, a common phenomenon among poor women (Geldstein 2001, Chant 1993, Benería and Roldón 1987), and perform household duties. Men with low levels of education, spouses, or children, on the other hand, are not more likely to be home-based workers, have less of a wage penalty, and have work shifts that are more similar to men who do not work out of their homes. The paper has also shown that home-based workers are a sub-set of the labor force for whom labor market opportunities are scarce or expensive. Women with families, uneducated women, and the elderly do not have the time flexibility (women) or the skills and mobility (older workers) to work outside the home. Instead, home-based work offers perhaps the best form of market work in which they may engage given their constraints (Beneria and Roldon 1987). 23 Finally, it should be noted that the analysis in this paper is limited due to an absence of a quantifiable definition and good country data regarding the motivations for being in this sector and the patterns. Only once a definition is agreed upon and country statistical agencies use this definition to standardize questionnaires with questions to examine these issues, will the incidence of, reason for employment in, and quality of the jobs in home-based work be well understood. Given the lessons learned in this paper, to lower the incidence of home-based work, it would be necessary to decrease the barriers to work outside the household. Among poor women with high domestic responsibilities, time saving devices ­ such as running water, electricity, and childcare services with flexible hours (for some countries) ­ may lower the cost of these women working outside the home. Additionally, more frequent, adequate, and safe transportation that would lower the time to travel to a job site, provide means for people with disabilities to leave their homes, and lower the danger of traveling alone would open the labor market options for women and the elderly. Finally, if, indeed, a working wife is a threat to men's identity, socialization in schools and community services to break traditional gender tasks would decrease the stigma of an employed woman. Information may be a key to improve the wages of home-based workers. Although this study could not explicitly explain why home-based workers (especially women) had lower hourly earnings than their counter-parts in more formal work sites, we may hypothesize that it is due to participation in an oligopsonistic labor market. In the short run, collectivization of home-based workers would lead to an exchange of information on salaries (and work conditions) throughout the market, thereby increasing 24 the competitiveness of the market. In the longer run, as constraints to working outside the household are decreased, home-based workers may fully enter the competitive market, only working at home if the wage and non-wage benefits in home-based work is preferable than working outside of the home. 25 Tables Table 1: Proportion of the labor force engaged in home-based work Estimated Year of Year of trade proportion estimate liberalization Algeria1 3.3 1989 1991 Australia1 2.9 1989 1964 Brazil2 5.5 1999 1991 Ecuador2 17.3 1999 1991 India1 2.5 1981 1994 Japan1 1.6 1988 1964 México2 4.4 1999 1986 Philippines1 23.0 1980s 1988 Peru1 10.5 1987 1991 United Kingdom1 2.3 1981 mid-century United States1 7.53 1985 mid-century 1ILO (1995); 2author's calculations Table 2: Prevalence of Home-based Work, by Category of Home-based Work and Gender (1999) Home-based workers as a proportion of the respective labor Brazil force Male Female Total Informal self-employed 1.6 6.3 3.6 Professional 0.2 0.5 0.4 Contract 0.01 0.08 0.04 Unpaid 0.2 0.7 0.4 Other* 0.3 2.4 1.2 All 2.3 9.9 5.5 Ecuador Informal self-employed 3.4 12.8 7.4 Professional 2.05 3.7 2.8 Contract 0.1 0.3 0.2 Unpaid 2.5 5.9 4.0 Other* 1.4 4.8 2.9 All 9.5 27.6 17.3 México Informal self-employed 1.8 5.5 3.2 Professional 0.5 0.5 0.5 Contract 0.06 0.08 0.07 Unpaid 0.2 1.0 0.5 Other* 0.08 0.3 0.2 All 2.6 7.4 4.4 * firm owners, service workers, and agricultural workers 26 Table 3a: Characteristics of Home-based Workers in Brazil (1999), in percent Home-based Workers Rest of the Economy* Informal Entrepreneurs Contract Sample size 4851 60 57811 Female 74.7 81.7 39.6 Age 15-25 11.6 20.0 31.06 Age 26-45 52.4 63.3 49.4 Age 46+ 36.08 16.7 19.5 No school 14.6 11.4 11.5 Primary education 57.3 82.9 58.2 Secondary education 28.09 5.7 30.3 Household head 41.3 16.7 49.4 Husband 17.8 3.3 37.5 Wife 49.7 60.0 19.6 Single Mother 16.6 10.0 10.7 Unmarried Female, no children 5.09 5.0 2.8 Unmarried Male 3.5 3.3 5.3 Child age 0-5 in household 22.2 28.3 25.8 Child age 6-12 in household 34.6 38.3 33.3 Chile age 13-18 in household 39.1 41.7 39.2 Non-white** 50.2 43.3 50.01 % Household income from this job? 52.0 42.0 63.0 *Rest of the economy are those who are working, but not in their homes ** Self-identified as black (preto) or mixed race (preto). 27 Table 3b: Characteristics of Home-based Workers in Ecuador (1999), in percent Home-based Workers Rest of the Economy* Informal Entrepreneurs Contract Sample size 464 13 2228 Female 74.1 61.5 34.9 Age 15-25 7.1 38.5 29.5 Age 26-45 50.2 15.4 47.6 Age 46+ 42.7 46.2 22.9 No school 3.9 0.0 3.3 Primary school 44.3 53.9 38.9 Secondary school 51.8 46.2 57.8 Household head 42.5 23.1 47.0 Husband 15.3 15.4 25.9 Wife 31.7 23.1 13.2 Single Mother 22.6 15.4 12.7 Unmarried Female, no children 0.4 0.0 0.2 Unmarried Male 4.7 23.1 21.7 Child age 0-5 in household 25.4 30.8 30.1 Child age 6-12 in household 35.8 15.4 39.01 Child age 13-18 in household 40.1 76.9 38.9 Indigenous** N/A N/A N/A % Household income from this job? 51.0 35.0 56.0 *Rest of the economy are those who are working, but not in their homes ** Ethnicity was not reported in the survey 28 Table 3c: Characteristics of Home-based Workers in Mexico (1999), in percent Home-based Workers Rest of the Economy* Informal Entrepreneurs Contract Sample size 4322 93 89683 Female 64.8 41.9 35.5 Age 15-25 8.5 48.4 31.7 Age 26-45 51.6 40.9 48.6 Age 46+ 39.9 10.8 19.6 No school 13.05 11.0 5.05 Primary education 71.7 75.3 73.5 Secondary education 15.2 13.7 21.4 Household head 46.3 20.4 47.5 Husband 23.7 14.0 37.3 Wife 35.5 20.4 13.8 Single Mother 3.6 2.2 3.01 Unmarried Female, no children 5.04 12.9 11.0 Unmarried Male 5.8 34.4 19.0 Child age 0-5 in household 27.8 37.6 29.4 Child age 6-12 in household 41.4 62.4 40.03 Child age 13-18 in households 42.2 59.1 43.02 Indigenous** N/A N/A N/A % Household income from this job? 47.0 28.0 54.0 *Rest of the economy are those who are working, but not in their homes ** Ethnicity was not reported in the survey 29 Table 4a: Firm Characteristics in Brazil (1999), in percent Home-based Workers Rest of the Economy* Informal Entrepreneurs Contract Sample size 4851 60 57811 Formal sector employera 0.0 6.7 58.4 Firm size 0-5 100.0 45.0 29.4 Firm size 6-10 0.0 11.7 14.2 Firm size 11+ 0.0 43.3 56.4 0-1 minimum wages 40.3 51.7 29.4 1-3 minimum wages 43.8 44.8 50.6 3-5 minimum wages 9.1 1.7 11.3 5-10 minimum wages 5.0 1.7 6.5 10+ minimum wages 1.8 0.0 2.3 Hours worked per week 35.4 36.3 37.0 Manufacturing 16.3 100.0 13.9 Commerce 14.6 0.0 18.3 Services 63.6 0.0 28.2 Agriculture 0.0 0.0 8.6 Other 5.5 0.0 30.9 * Rest of the economy are those who are working but not at home aFormal sector employer is defined by the worker having a work card that was signed by the employer upon hiring the employee (carteira assinada). 30 Table 4b: Firm Characteristics in Ecuador (1999) Home-based Workers Rest of the Economy* Informal Entrepreneurs Contract Sample size 464 13 2228 Formal sector employera 0.0 7.7 31.7 Firm size 0-5 100.0 69.2 60.4 Firm size 6-10 0.0 0.0 8.0 Firm size 11+ 0.0 30.8 31.6 0-1 minimum wages 67.1 84.6 49.9 1-3 minimum wages 25.6 15.4 38.1 3-5 minimum wages 3.8 0.0 6.7 5-10 minimum wages 2.9 0.0 3.5 10+ minimum wages 0.7 0.0 1.8 Hours worked per week 32.3 31.4 43.3 Manufacturing 37.7 100.0 14.4 Commerce 36.2 0.0 28.7 Services 25.2 0.0 38.5 Agriculture 0.9 0.0 7.6 Other 0.0 0.0 10.8 *Rest of the economy are those who are working but not at home aFormal sector employer is defined by a positive response to the question of having a written contract (contrato escrito/nombramiento) 31 Table 4c: Firm Characteristics in Mexico (1999), in percent Home-based Workers Rest of the Economy* Informal Entrepreneurs Contract Sample size 4322 93 89683 Formal sector employera 0.0 19.7 54.9 Firm size 0-5 100.0 80.7 44.4 Firm size 6-15 0.0 12.9 9.7 Firm size 16+ 0.0 6.5 46.0 0-1 minimum wages 19.4 17.5 9.3 1-3 minimum wages 51.2 62.5 49.0 3-5 minimum wages 16.9 15.0 20.4 5-10 minimum wages 9.0 4.2 14.9 10+ minimum wages 3.5 0.8 6.5 Hours worked per week 34.8 36.2 45.9 Manufacturing 43.0 100.0 22.9 Commerce 21.7 0.0 21.7 Services 23.4 0.0 32.0 Agriculture 0.0 0.0 1.7 Other 11.9 0.0 21.8 *Rest of the economy are those who are working but not at home aFormal sector employer is identified by the receipt of benefits and a firm size with greater than five workers. 32 Table 5a: Brazil - Probability of Home-based work (relative to non home-based)* All Men Women Female 0.81 --- --- (34.23) Age1525 -0.74 -0.43 -0.83 (-22.13) (-4.7) (-19.7) Age2645 -0.36 -0.25 -0.39 (-13.31) (-5.65) (-11.13) Head -0.12 0.12 0.018 (-4.43) (2.51) (0.43) Married 0.041 -0.13 0.22 (1.61) (-3.09) (5.45) presch 0.014 -0.12 0.1 (0.57) (-2.79) (3.25) elemsch 0.011 -0.061 0.058 (0.52) (-1.72) (2.06) teen 0.0078 -0.017 0.028 (0.38) (-0.49) (1.06) adult -0.08 -0.018 -0.051 (-3.49) (-0.43) (-1.76) non-white 0.029 0.0046 0.048 (1.54) (0.15) (1.96) nosch 0.36 0.039 0.56 (10.33) (0.74) (11.72) pri 0.25 -0.036 0.37 (11.62) (-1.04) (14.16) constant -1.75 -1.63 -1.22 (-33.55) (-20.78) (-21.05) Pseudo R2 0.12 0.023 0.066 Sample size 52,722 33,035 19,687 * z-values in parentheses 33 Table 5b: Ecuador - Probability of Home-based work (relative to non home-based)* All Men Women Female 0.96 --- --- (12.75) Age1525 -1.1 -0.086 -1.28 (-9.32) (-4.48) (-8.13) Age2645 -0.44 -0.061 -0.43 (-5.49) (-3.8) (-3.83) Head 0.0011 0.012 0.1 (0.01) (0.72) (0.09) Married 0.11 0.017 0.074 (1.68) (1.16) (0.75) presch 0.08 -0.0047 0.17 (1.05) (-0.3) (1.71) elemsch -0.18 -0.035 -0.12 (-2.59) (-2.51) (-1.32) Teen 0.18 0.011 0.24 (2.74) (0.8) (2.7) Adult 0.31 0.035 0.36 (2.24) (1.7) (1.84) nosch -0.36 -0.056 -0.28 (-1.98) (-1.56) (-1.35) pri 0.043 0.007 -0.26 (0.64) (0.5) (-0.29) constant -1.34 -1.36 -0.47 (-7.96) (-5.53) (-2.21) Pseudo R2 0.17 0.0813 0.079 Sample size 2705 1638 1067 * z-values in parentheses 34 Table 5c: Mexico - Probability of Home-based work (relative to non home-based)* All Men Women Female 0.044 (27.72) Age1525 -0.042 -0.024 -0.071 (-25.77) (-12.49) (-19.46) Age2645 -0.023 -0.012 -0.040 (-15.92) (-7.4) (-12.15) Head -0.007 0.007 0.007 (-4.9) (3.57) (1.82) Married 0.005 -0.008 0.036 (3.97) (-5.49) (10.02) presch 0.004 -0.002 0.017 (3.00) (-1.58) (4.41) elemsch 0.002 -0.001 0.006 (1.21) (-1.00) (2.24) Teen -0.000 0.001 -0.005 (-0.08) (1.29) (-1.78) Adult -0.003 0.003 -0.002 (-2.22) (2.00) (-0.69) nosch 0.04 0.004 0.095 (12.28) (1.35) (13.28) pri 0.014 0.003 0.033 (10.99) (2.09) (10.41) Pseudo R2 0.0895 0.0308 0.0966 Sample 94098 60483 33615 size * z-values in parentheses 35 Table 6a: Brazil - Log hourly wages for the labor force as a whole and with Heckman correction* Hourly wage estimation Heckman corrected All Men Women All Men Women Wage earners Female -0.26 -0.11 (-40.73) --- --- (-11.77) --- --- Age1525 -0.59 -0.63 -0.51 -0.57 -0.61 -0.52 (-61.75) (-52.34) (-32.6) (-58.55) (-49.77) (-31.85) Age2645 -0.14 -0.12 -0.16 -0.22 -0.24 -0.17 (-15.74) (-10.9) (-11.13) (-22.89) (-20.03) (-10.42) Nonwhite -0.24 -0.24 -0.21 -0.24 -0.24 -0.21 (-39.64) (-32.68) (-22.07) (-39.77) (-32.56) (-22.11) Nosch -0.97 -1.0 -0.88 -0.85 -0.94 -0.84 (-83.01) (-72.19) (-40.01) (-64.95) (-64.84) (-25.45) Pri -0.44 -0.46 -0.41 -0.38 -0.43 -0.39 (-68.67) (-54.23) (-41.63) (-53.21) (-49.04) (-25.37) Home -0.22 -0.11 -0.26 -0.21 -0.11 -0.26 (-17.23) (-4.61) (-17.31) (-16.39) (-4.49) (-17.18) constant -2.51 -2.49 -2.8 -2.39 -2.3 -2.75 (-256.3) (-207.9) (-187.1) (-211.54) (-164.1) (-90.05) R2 57057 0.25 0.19 --- --- --- Sample 0.24 36002 21055 111548 53223 58325 size * t- and z-values in parentheses 36 Table 6b: Ecuador - Log hourly wages for the labor force as a whole and with Heckman correction* Hourly wage estimation Heckman corrected All Men Women All Men Women Wage earners Female -0.13 --- --- 0.33 --- --- (-3.55) (6.47) Age1525 -0.43 -0.41 -0.44 -0.19 -0.27 0.00071 (-8.27) (-7.0) (-4.46) (-3.18) (-4.04) (0.01) Age2645 -0.031 0.0044 -0.055 -0.19 -0.087 -0.38 (-0.7) (0.08) (-0.71) (-3.74) (-1.52) (-3.8) Nosch -0.57 -0.7 -0.51 -0.54 -0.57 -0.64 (-5.48) (-4.55) (-3.37) (-4.68) (-3.49) (-3.27) Pri -0.23 -0.2 -0.28 -0.28 -0.22 -0.37 (-6.17) (-4.61) (-4.16) (-6.66) (-4.93) (-4.28) Home -0.26 0.042 -0.41 -0.2 -0.051 -0.21 (-5.36) (0.56) (-6.05) (-4.26) (0.69) (-3.41) Constant 7.3 7.24 7.25 7.65 7.46 8.64 (157) (139.61) (87.41) (136.22) (114.2) (76.67) R2 0.063 0.063 0.059 --- --- --- Sample 2597 1573 1024 4836 2163 2673 size * t- and z-values in parentheses 37 Table 6c: Mexico - Log hourly wages for the labor force as a whole and with Heckman correction* Hourly wage estimation Heckman corrected All Men Women All Men Women Wage Earners Female -0.13 --- --- -0.16 --- --- (-32.0) (-42.99) Age1525 -0.35 -0.33 -0.37 -0.47 -0.48 -0.47 (-56.74) (-45.06) (-34.49) (-81.98) (-66.41) (-48.37) Age2645 -0.048 -0.031 -0.08 -0.098 -0.096 -0.1 (-8.53) (-4.52) (-8.0) (-18.67) (-14.85) (-11.39) Nosch -0.76 -0.67 -0.87 -1.07 -1.03 -1.13 (-77.03) (-52.57) (-54.91) (-111.3) (-82.57) (-74.27) Pri -0.35 -0.29 -0.42 -0.63 -0.61 -0.65 (-71.52) (-44.59) (-55.93) (-162.8) (-123.85) (-106.2) Home -0.19 -0.038 -0.27 -0.21 -0.074 -0.29 (-18.69) (-2.34) (-20.54) (-21.67) (-4.83) (-23.4) Constant 1.69 1.62 1.63 2.05 2.029 1.92 (252.46) (196.3) (154.9) (342.3) (266.37) (193.5) R2 0.12 0.09 0.16 --- --- --- Sample 104936 67653 37283 156642 97662 58980 size * t- and z-values in parentheses 38 Table 7a: Brazil - Log hours worked (1999)* All Men Women Female -0.16 (-42.91) --- --- Age1525 0.028 -0.010 0.076 (5.38) (-1.64) (7.28) Age2645 0.048 0.023 0.081 (10.67) (4.81) (8.72) Head 0.092 0.076 0.057 (22.34) (4.81) (5.5) Married 0.019 0.020 -0.011 (4.88) (4.38) (-1.11) Presch -0.001 0.012 -0.023 (-0.33) (2.88) (-2.66) Elemsch -0.002 0.011 -0.021 (-0.5) (2.94) (-2.78) Teen -0.011 -0.018 0.001 (-3.33) (-4.95) (0.16) Adult 0.026 0.014 0.032 (6.82) (3.42) (4.19) Nonwhite -0.021 -0.007 -0.045 (-7.09) (-2.33 (-7.13) Home -0.267 -0.092 -0.322 (-42.45) (-9.16) (-34.84) R2 0.10 0.022 0.063 Sample size 69372 45267 24105 * t-values in parentheses 39 Table 7b: Ecuador - Log hours worked* All Men Women -0.332 Female (-10.1) - - 0.213 0.140 0.334 Age1525 (4.67) (2.95) (3.52) 0.104 0.111 0.109 Age2645 (2.87) (2.95) (1.47) 0.213 0.128 0.205 Head (6.19) (3.12) (2.66) 0.061 0.066 0.087 Married (2.09) (2.05) (1.31) -0.064 0.078 -0.290 presch (-1.97) (2.32) (-4.3) -0.014 0.006 -0.037 elemsch (-0.48) (0.19) (-0.59) -0.036 0.021 -0.107 Teen (-1.24) (0.7) (-1.76) 0.038 -0.040 0.064 Adult (0.71) (-0.75) (0.5) -0.364 -0.135 -0.461 Home (-9.3) (-2.57) (-7.38) R2 0.1453 0.0306 0.091 Sample size 2950 1900 1050 * t-values in parentheses 40 Table 7c: Mexico - Log hours worked* All Men Women Female -0.18 (-61.03) --- --- Age1525 0.10 0.03 0.15 (24.55) (7.11) (17.33) Age2645 0.08 0.05 0.10 (24.38) (16.11) (13.71) Head 0.11 0.06 0.02 (32.9) (14.06) (2.26) Married -0.01 0.02 -0.1 (-4.61) (7.71) (-14.33) presch -0.00 0.02 -0.04 (-1.22) (6.37) (-6.08) elemsch -0.01 0.01 -0.03 (-2.40) (1.76) (-5.04) Teen -0.01 -0.01 -0.00 (-2.74) (-3.54) (-0.1) Adult 0.05 0.02 0.05 (19.43) (8.4) (8.8) Home -0.32 -0.18 -0.38 (-55.65) (-22.67) (-40.6) R2 0.12 0.026 0.086 Sample size 108415 70091 38324 * t-values in parentheses 41 References Akerloff, George and Rachel Kranton (1999) "Economics and Identity," Quarterly Journal of Economics (115:3), 715-753. Arriagada, Irma (1998) The Urban Female Labour Market in Latin America: The Myth and the Reality, Serie Mujer y Desarrollo #21 (ECLAC: Chile). Beneria, Lourdes and Martha Roldon (1987) Las Encrucijadas de Clase y Género (University of Chicago Press: Chicago, USA). de Ferranti, David, Guillermo E. Perry, Daniel Lederman, and William F. Maloney (2002) From Natural Resources to the Knowledge Economy: Trade and Job Quality (World Bank: Washington, DC). Carr, Marilyn, Martha Chen, and Jane Tate (2000) "Globalization and Home-Based Workers" Feminist Economics 6(3), 123-142. Chant, Sylvia (1991) Women and Survival in Mexican Cities (Manchester University Press: Manchester). Chen, Martha, Jenifer Sebstad, and Leslie O'Connell (1999) "Counting the Invisible Workforce: The Case of Homebased Workers" World Development 27(3), 603-610. Cunningham, Wendy and William Maloney (2001) "Heterogeneity in the Mexican Micro-Enterprise Sector: An Application of Factor and Cluster Analysis" Economic Development and Cultural Change (5), 131-156. Geldstein, Rosa (2000) "Non-Labor Market Coping Strategies in Argentina", LCSPG/World Bank, mimeo. Gonzalez de Rocha, Mercedes (1994) The Resources of Poverty: Women and Survival in a Mexican City (Blackwell Publishers: Massachusetts). Jelin, Elizabeth, Matilde Mercado and Gabriela Wyczykier (2000) "Home Work in Arentina" SEED Working Paper #6 Series on Homeworkers in the Global Economy (ILO: Geneva). Maloney, William (forthcoming) "Informality Revisited", World Development. Prugl, Elisabeth and Irene Tinker (1997) "Microentrepreneurs and Homeworkers: Convergent Categories" World Development 25(9), 1471-1482. Tomei, Manuela (2000) "Home Work in Selected Latin American Countries: A Comparative Overview", unpublished (ILO: Geneva). 42 WIEGO (Women in Informal Employment: Globalizing and Organizing) (2000) Women in the Informal Economy, www.wiego.org. Wodon, Quentin (2000) Poverty and Policy in Latin America and the Caribbean World Bank Technical Paper No. 467 (World Bank: Washington, DC). 43