Gender, Work Intensity, and Well-Being of Thai Home-Based Workers

Abstract

The contribution explores the time-use dimensions of the individual well-being of home-based workers in Thailand's urban squatter communities to demonstrate how time-use patterns provide information regarding individual experiences in performing economic activities that affect quality of life. The study focuses on two groups of home-based workers: the self-employed, and those who work for a contractor. Using an individual-level well-being index that takes into account income, the capabilities related to education, and work intensity, the authors examine by OLS and GME techniques the varied factors that affect the well-being of home-based workers. The findings show that women workers experience a higher incidence of work intensity and hence lower quality of life compared with men. A better understanding of the factors that promote or lower well-being can help policy-makers design more effective programs and economic and social policies.

Keywords:

• Well-being
• time use
• work intensity
• home-based workers
• informal sector

ACKNOWLEDGMENTS

We would like to acknowledge the support of HomeNet Thailand, United Nations Development Fund for Women (UNIFEM) Bangkok, and Aphitchaya Nguanbanchong during the fieldwork for this article. We thank Assen Asenov and Hazel Malapit for their assistance and the referees of this manuscript for their helpful comments.

Notes

¹ Amartya Sen (1993), Martha Nussbaum (2000), Lourdes Benería (2003), Ingrid Robeyns (2003), and Martha MacDonald, Shelley Phipps, and Lynn Lethbridge (2005), among others, have argued that discourses on well-being should go beyond the human development index (HDI). A. Robert Pollak and Michael L. Wachter (1975), F. Thomas Juster and Frank P. Stafford (1985), Nancy Folbre (1997), and MacDonald, Phipps, and Lethbridge (2005), among others, have raised the importance of exploring time-use patterns to determine a person's well-being.

² The LIMEW is constructed based on the notion that economic well-being refers to household members' command over, or access to, the goods and services produced in a market economy during a given period of time; it includes not only the state's provisioning of goods and services, but also those produced and consumed by household members. One component of the LIMEW measure is the estimated hours spent on total work (paid work plus housework) by the average household.

 The Process Well-Being (PWB) index was developed by F. Thomas Juster and Greg K. Dow (1985); it captures each individual's satisfaction from the economic and noneconomic activities, which serves as a measure of well-being. The measure is based on responses of time-use survey participants to rank, on a scale of 0 through 10, a set of twenty-two prespecified activities – for example, market work, cleaning the house, watching TV, playing with children. The PWB is based on the time spent in various activities as stated above, weighted by the process benefits for each activity. Note that the weight is fundamentally subjective based on each respondent's perspective.

³ Grace Baruch, L. Beiner, and Rosalind Barnett's study (1987) of the relationship between work and stress found the association between role conflict (typically defined as feeling pulled apart by conflicting demands) as well as levels of role overload (typically defined as having too much to do) and high levels of stress, especially among married women, to be statistically significant. Thomas Juster and Frank Stafford (1985) and Martha Roldan's (1985) findings also suggest that women have predominantly shorter resting hours, greater intensity and fragmentation of work, and more frequent recourse to multiple simultaneous tasks.

⁴ Examples include minding a child while selling goods on the street, or cleaning the house while managing a small food eatery at home.

⁵ The research project is a comparative study of urban labor market informalization and home-based work in Bolivia, Ecuador, Thailand, and the Philippines, with the support of local research institutions and community organizations. It involved data collection using multi-visit interviews and a multi-topic survey questionnaire during the period May to December 2002. The survey instrument included questions on household and community characteristics as well as on informal employment, savings, assets, credit, and participation in household decision-making.

⁶ The small sample size resulted in part from the time-consuming nature of the time-use questionnaire; many potential respondents said they simply “[didn't] have time” to participate.

⁷ However, if we base productivity on the unit of time (linearity in time and ignoring the individual effort), then the productivity increases.

⁸ Wolfe and Haveman examine the determinants of women's health, specifically the impact of time allocation. Recognizing that changes in health status are the result of a wide variety of exogenous characteristics and endogenous behaviors and status, they develop an empirically testable model to examine the impact of time allocation on women's health. Their regression results indicate that time allocations do have a significant effect on health status. Market work does not, in itself, cause health problems, and may in fact contribute to improvement in health. But the combination of both childcare and housework demands on women and the dual role of working and having young children appear to be associated with health deterioration (Wolfe and Haveman 1983, 138–9).

⁹ For example, an individual tends to be under greater stress than normally when working on two or more tasks simultaneously if one (or both) of the tasks require uninterrupted attention because of a due date or some other condition.

¹⁰ Thailand was described in a 1993 World Bank report as a model for economic development (Medhi Krongkaew 2001).

¹¹ The term “home-based workers” refer to those workers who carry out remunerative work within their homes. As noted by Carr, Chen, and Tate (2000), the term “home-workers” refers only to dependent subcontract workers.

¹² For a more detailed discussion of home-based work, see ILO (2002); Carr, Chen, and Tate (2000); and Martha Chen, Jennifer Sebstad, and Lesley O'Connell (1999).

¹³ In some countries, dependent contracted workers have been found to have fairly long relationships with a single contractor. This is not the case for many of the contracted workers here in Bangkok.

¹⁴ The communities were selected to reflect the diversity of urban poor communities in terms of size, proximity to commercial areas, length of existence, and strength of social networks or community organizations. An additional consideration was the presence of local contacts such as community leaders and researchers in the areas. The 170 households in the sample were then selected in a purposive manner: one of every six households in a neighborhood roster or mapping, such that it has at least one member engaged in informal sector work, whether as piece-rate home-worker or as self-employed.

¹⁵ Home-based worker respondents tend to estimate their working times in relation to routine activities rather than on a watch or clock. For example, when asked, “What time do you start working?,” a typical answer by a home-worker would be: “after I drop my children at school,” or “after I have finished my breakfast.” In addition, a proportion of our sample is illiterate, so diary or recall methods are not appropriate or are prone to serious errors. In addition, home-based workers, unlike most salaried workers, do not have regular or well-defined work patterns. They work whenever they have an interval of time in which to work, or when they get work from the contractor. Respondents may work every day for a week, and then do nothing for 2 weeks before getting another job contract. Finally, home-based workers tend to have busy lives. They typically combine paid market work with unpaid household work. For instance, a home-based grocery store owner might cook dinner while waiting to serve customers. Hence, the time-use data collection method was designed to capture both the main activity and the secondary activity.

¹⁶ A shortened and simplified time-use diary was given to the home-based worker respondents at the first interview session. The interviewer then explained the concept of time and related definitions such as which kinds of activities are considered work or leisure to the survey respondents. The interviewer provided detailed instructions and orientation to the respondents, as it is essential that survey respondents clearly understand all definitions and concepts to ensure consistency across the sample. During the next visit, the interviewer conducts a short recall interview with the respondent to ensure that the time-use diary has been filled out correctly. During this session, the interviewer could add any missing information. This multi-visit approach also affords the interviewer an opportunity to know the survey respondents better and gain their confidence over time. These steps were taken to reduce errors or biases in the data collected.

¹⁷ This magnitude is consistent with the average household size collected by the National Statistical Office of Thailand in the Household Socio-Economic Survey of 2001, with an average of 3.6 persons per household.

¹⁸ Casual workers are workers who have an explicit or implicit contract of employment that is not expected to continue beyond a short period, with the duration to be determined by national circumstances. Workers in short-term employment are workers who hold explicit or implicit contracts of employment that are expected to last longer than the period used to define “casual workers” (ILO 2002).

¹⁹ We conducted a sensitivity analysis in order to examine the effect of the change in relative weights on the three components on the GME tests. We calculated two alternative, weighted well-being indices, namely, WBI₂ and WBI_3. We gave more weight to income (40 percent) relative to education (30 percent) and inverse work intensity (30 percent) in the estimation of WBI_2. In the case of WBI_3, we gave more weight to both income (40 percent) and education (40 percent), while inverse work intensity has only 20 percent. Interestingly, the use of these alternative weights didn't change the direction of the signs, and there are only minor changes in the parameter estimates. If more permutations of different weights were conducted, it may be possible that we would obtain different results, but we suspect this would happen only with very extreme weights.

²⁰ In the OLS approach, researchers typically rely on the traditional sample-theoretic estimation and inference framework. That is, observed outcomes are first treated as a sample from a larger population with some known (true) model that is suggested by theory. Assumptions are made regarding the distribution of the outcomes of interest. In this study, we are dealing with nonexperimental, small samples where, among other things, inferences are sensitive to the assumptions we make especially with regard to the identifiability of distinct processes. In the GME approach, we treat the unknown signal (the expected outcome from our model) as an expectation over several possible points of support, whereby no assumptions are made about the probabilities other than that they are proper. We make use of prior information or knowledge about various probabilities to define the signal support space accordingly. As in all modeling exercises, we allow for some imperfections by stipulating that the expected (modeled) outcome will only approximate the observed outcomes. We bring into the picture theoretically derived predictors and utilize them to constrain the values that the probabilities take. In the Maximum Entropy (ME) formalism, the Shannon's entropy is used as the criterion to maximize, subject to all available constraints, in order to derive conservative inferences from a small sample (Amos Golan 2002, 2006).

²¹ The Cook–Weisberg test for heteroskedasticity using fitted values of the WBI indicates the problem of heteroskedasticity in the model.

²² A model specification error can occur when one or more relevant variables are omitted from the model, or if one or more irrelevant variables are included in the model. If relevant variables are omitted from the model, the common variance they share with included variables may be wrongly attributed to those variables, and the error term is inflated. On the other hand, if irrelevant variables are included in the model, the common variance they share with included variables may be wrongly attributed to them.

²³ In a study that investigates the validity of the individual WBI using a comparative analysis with respondents' own perception of their well-being, Anant Pichetpongsa (2004) made use of the respondents' response to the general “life satisfaction” question (translated from Thai): “We would like to ask you about your satisfaction with your life in general. Please answer by using the following scale: 0 means totally unhappy and 10 means totally happy. How happy are you at present with your life as a whole?” Applying OLS tests, Spearman Rank Correlation method, and Borda Rank approach on the same Thailand sample data, their findings tentatively show that the inverse work intensity index statistically predicts subjective well-being of the respondents, more so than monetary income alone.