Abstract
Chapters 1-3 of this thesis are comprised of three essays on debt aversion, discount rates and present bias and, finally, the joint effects of entrepreneurial orientation (EO) and credit utilization on poverty. Each chapter has its own chapter-specific set of appendices. An additional Appendix is included following Chapter 3, providing additional methodological detail regarding how and why the primary data analyzed in this thesis from Soc Trang Province, Vietnam, were collected. This extended Appendix also included stylized facts, background and context regarding how financial poverty (FP) and nonfinancial poverty, referred to throughout this thesis as multidimensional poverty (MP), have been measured in Vietnam, based primarily on publications by Vietnam's General Statistics Office (GSO) and the World Bank.
The thesis address three primary questions:
(1) Why do so many farmers who are eligible to borrow--low-income farmers whom policymakers often assume would benefit from greater credit utilization--refuse to borrow?
(2) Which observable characteristics of farmers are correlated with their time preferences?
(3) Given the mixed results in the previous literature regarding the effects of credit utilization on poverty (i.e., multidimensional poverty and income poverty), does entrepreneurial orientation reveal new insights regarding how well farmers are able to use borrowed funds?
The contents of the three chapters are summarized as follows:
Chapter 1: Debt averse nonborrowers are defined as those who have never borrowed in the past and prefer avoiding debt in the future, even when offered generous borrowing terms such as zero interest rates, zero collateral and easy debt-forgiveness. The other nonborrower preference type we consider has either borrowed in the past or is open to doing so in the future. To better understand nonborrowing behavior, three credit preference types (CPTs) were measured among 575 low-income farmers in Mekong Delta, Vietnam. Among 208 current nonborrowers, 156 had never borrowed and only seven were credit rationed. Among never-borrowers, 102 were debt averse. Thus, more than half of eligible nonborrowers (102 out of 201) were debt averse. This high prevalence challenges the assumption of unmet credit demand among the poor. Disaggregating CPTs reveals heterogeneity among nonborrowers that would remain observationally equivalent otherwise. We report reasons for not borrowing and investigate how observable characteristics influence the likelihood of debt aversion.
Chapter 2: Time preferences is in debates concerning the determinants of poverty. We cite a few papers that do indeed blame the poor for being too impatient or impulsive. However, there is little agreement about the basic descriptive facts regarding impatience (i.e. how high are their discount rates?) and impulsivity (typically measured as how much less than 1 is their estimated value of beta in Laibson's famous quasi-hyperbolic discounting model). There is a famous AER paper by Tanaka et al. (2010) that estimates extraordinarily high discount rates (r) and a large degree of impulsivity, which is substantially less than 1). Using data from poor Vietnamese rural dwellers, Tanaka, Camerer and Nguyen (2010) found: (i) an extraordinary degree of impatience overall (mean annualized exponential discount rate of r=7.95); (ii) inversely correlated impatience and income (i.e., the discount rate was decreasing in income); and (iii) quasihyperbolic present-bias (beta=0.644 or 0.820) across all income levels (i.e. no significant effect of income on beta). The three-parameter discounting function they estimated was a restricted version of Benhabib, Bisin and Schotter's (2010) more general 4-parameter model, which includes a fixed cost parameter (b) measuring willingness to pay to avoid time delay (independent of duration). Therefore, we set out to measure r and beta using all of Tanaka et al. (2010)'s time delays but extending it longer to include 6- and 12-month delays.
We also substituted a technique of elicitation from Benhabib et al.'s paper using NYU students. Benhabib et al. (2010) also gave us a nested discounting function that includes two competing theories of impulsivity: Laibson's beta-delta model which Tanaka et al. (2010) used versus Benhabib et al. (2010)'s fixed psychic cost of having to wait any duration whatsoever called b. Using indifference points elicited in Benhabib et al. (2010)'s matching task, we compare seven discounting functions to investigate whether Laibson's (1997) quasihyperbolic (beta-delta) model or the fixed cost mechanism better explains time preferences. Our sample of smallholder farmers in Mekong Delta provides useful sample variation in rates of financial poverty (9% FP-poor and 13% FP-near-poor) and nonfinancial or multidimensional poverty (15% MP-poor and 23% MP-near-poor). The overall proportion of households with at least one of these four types of hardship is 49.6%. Our sample includes 30 villages with heterogeneous farm types, soil quality, river-water salinity and distances to the nearest roads, industrial zones and formal credit institutions, in an agricultural region where farm entrepreneurship has helped reduce poverty rates from over 65% to under 15% in less than three decades, providing a unique opportunity to investigate empirical relationships between time preferences and farmers' observable characteristics. In a head-to-head test of quasihyperbolic (beta<1) versus fixed-cost (b>0) explanations for present bias, our data provide strong support for Benhabib et al. (2010)'s fixed cost mechanism in terms of fit, likelihood ratio tests and model selection criteria (e.g., the Bayesian Information Criterion). We show that quasihyperbolic present bias is largely an artefact of omitting the fixed cost parameter, with estimates of beta <0.920 and 0.987 in models without b and beta =1.020 and 1.090 with. Relative income correlates somewhat with patience (lower r) and smaller fixed-cost discounts (lower b). We also show that demographic factors commonly theorized to influence discounting, such as age, education and income are surprising weak. Big-5 personality shows up as significantly associated with r, beta and b in the models we estimate.
Chapter 3: We employ a survey-based instrument widely used in the management science literature called entrepreneurial orientation (EO), which has three primary components consisting of innovativeness, proactiveness, and risk-taking of individuals in income-generating activities as a factor that, when applied to how borrowed funds are used, is investigated as a novel explanation for poverty. We applied the so-called configurational model (with three-way interactions) in which EO interacts with credit to influence how well it is used.
Regarding policy implications, the novel empirical findings in this thesis provide a new look at the mechanisms that governments use, such as subsidized credit programs, from the standpoint of whether one-size-fits-all policies could be segmented to offer different incentives, messages or programs that achieve a better match with the decision-making processes that farmers of different types or segments use.
Positionality statement: As a native of the region who was born into a family of rice farmers, the author wishes to find solutions for multidimensional poverty alleviation and contribute to improving the living standards of farmers in the Mekong Delta, Vietnam. Finally, this research is also the promise of the PhD student to his father, who passed away in 2017, that he would contribute his modest knowledge to help poor farmers improve their lives.