Ex Ante Risks & Returns and Occupation Sector Choice in Low-Income Countries: Evidence from subjective expectations in Mozambique

In this paper, we study occupation sector choice among college students in Mozambique, with a focus on the role played by students’ subjective expectations for education and labor market outcomes.
Economic models of human capital investments and career decisions posit that individuals make such decisions by comparing expected benefits (returns) and costs (risks) of alternative choices (e.g., Becker (1964), Mincer (1974), Siow (1984)). However, empirical research has long been plagued by a lack of data on individuals’ expectations about the risks and returns of alternative human capital or career options, (e.g., dropout risk, expected wages, wage dispersion, etc.), and by the associated difficulty of separating the roles of expectations, preferences, and other determinants of these decisions (e.g., Manski (1993)).
A recent and growing literature has begun addressing the problem by directly measuring individuals’ and households’ expectations for human capital investments and their consequences in economic surveys (see reviews by Manski (2004), Hartog and Diaz-Serrano (2013), and Giustinelli and Manski (2018)). These data have been used to estimate empirical models of human capital decisions under uncertainty and to investigate how individuals form choice-relevant expectations over time or following reception of new information (e.g., Dominitz and Manski (1996), Zafar (2011, 2013), Arcidiacono et al. (2012), Stinebrickner and Stinebrickner (2014a,b), Wiswall and Zafar (2015a,b), Giustinelli (2016), Patnaik et al. (2019)). These studies have shed light on the importance of students’ and families’ expectations for both pecuniary and non-pecuniary outcomes in influencing educational and early career decisions.
While most of the early expectations studies were carried out in developed countries, several recent efforts have directed attention to developing contexts where measurement and analysis of expectations hold the promise of fostering understanding of market imperfections and lack of development related to limited information and imprecise beliefs, as well as policy makers’ ability to design policies addressing those failures (e.g., Jensen (2010), Attanasio and Kaufmann (2014), Delavande and Zafar (2019)).
Expectations studies in developing countries have spanned an array of high-stake domains, including health, education, labor, and migration decisions (see reviews by Attanasio (2009), Delavande et al. (2011a,b), and Delavande (2014)). In this paper, we study a particularly relevant choice in developing contexts for economic productivity and growth, that is, the choice of occupational sector among public, private, and self-employment (the latter including entrepreneurs). Specifically, the main goal of our analysis is quantify the importance of individuals’ expectations for monetary outcomes of alternative occupational choices in driving selection of the occupation sector, with a focus on the potential trade-off between the perceived returns and risks of working in different sectors.
This topic is highly relevant for developing countries like Mozambique, where policies of intensive privatization, liberalization, and deregulation have been implemented for several decades now, but where the private sector and especially the self-employment sector have not expanded at rate and to the extent that those policies had intended.
The development literature has long emphasized the importance of self-employment/entrepreneurship for growth and development; see Banerjee and Newman (1993), Lloyd-Ellis and Bernhardt (2000), Ghatak and Nien-Huei Jiang (2002), and Eckhout and Jovanovic (2012) for theoretical frameworks and predictions, and Ray (2007), Naude’ (2010), and World Bank (2013) for reviews. However, the existing empirical evidence points to low self-employment take-up and high preference for the public sector.
Yet little is known about how individuals make occupation sector choices in these contexts. One exception is the work by Bianchi and Bobba (2013) in Mexico, which points to the role of liquidity constraints and business risk. Other works on MENA countries document and analyze the phenomenon of public-sector queuing (e.g., Assaad (1997, 2007, 2014) and Stampini and Verdier-Chouchane (2011)).
The low take-up of self-employment/entrepreneurship and the over-supply of public sector labor seem hard to reconcile with existing realizations-based estimates of the returns to self-employment/capital, which suggest that such returns are high (e.g., de Mel, McKenzie, and Woodruff (2008), Blattman, Fiala, and Martinez (2014)). However, surprisingly little is known about individuals’ subjective perceptions of the returns and risks involved in self-employment.
To address this gap, we collect and employ novel survey data from Mozambique, encompassing: students’ probabilistic beliefs about population earnings and own earnings (both mean and distribution), along with their subjective employment probabilities in alternative occupations (public sector, private sector, and self-employment), under different scenarios about human capital accumulation (graduating vs. dropping-out from university).
Following econometric approaches developed in Manski (1999) and Blass et al. (2010), and recently applied by Arcidiacono et al. (2019), Briggs et al. (2019), Giustinelli and Shapiro (2019), and Wiswall and Zafar (2018) to different choice domains, we interpret within-person differences in wage expectations (mean and higher moments) across graduation scenarios and/or occupational choices as individual-level subjective treatment effects of graduation and/or occupational choice on wages. Additionally, we provide structural estimates of the effect of expected returns and risks on occupational choices (the latter being measured in the form of choice probabilities).
Preliminary results point to a role of both expected monetary returns and perceived left-tail earnings risk in occupational choice plans, with higher expected returns positively affecting the probability of choosing the corresponding occupation and higher perceived left-tail earnings risk lowering the probability of choosing the corresponding occupation. Hence, these preliminary findings suggest that students in our sample trade off expected monetary returns and risks when planning their occupation sector choice.
We are currently in the process of extending the analysis in several directions. To aid identification of the model’s preference parameters, we have been working on a new set of estimates that exploit a particular feature of the survey design which embedded an informational treatment randomized across students. Specifically, different groups of students were given information about the mean and/or the dispersion of the actual earnings distributions in relevant subpopulations defined by their graduation status (college graduate vs. not) and occupation sectors (private, public, self-employment). This design generated a panel of beliefs consisting of students’ wage expectations and sector choice probabilities before and after receiving the information. Such data allow us to use individual fixed effects methods to address the potential endogeneity of subjective expectations (e.g., high expectations may reflect idiosyncratic tastes for a specific outcome).
Our analysis sheds light on several aspects of occupational choice among college students in Mozambique, most importantly: (i) how well students are informed about actual labor market opportunities and whether they over- or under-estimate earnings (mean and dispersion) in specific educational and occupational scenarios; (ii) the relative importance of pecuniary vs. non-pecuniary factors affecting occupational choices. With reference to the latter, a key aspect we investigate is the importance of perceived wage riskiness in occupation sector choice; that is, whether and, if so, to what extent individuals tend to avoid remunerative occupations (high expected wages) because they perceive them as high-risk (high expected wage volatility).