Risk Attitudes and Microfinance Preferences

Description

This study hypothesizes that risk attitudes dynamically update during repeated decision tasks and significantly determine household microfinance preferences and borrowing choices. Microcredit decisions are governed by both unobserved preference and scale heterogeneity, whereby dynamic risk preference changes moderate microfinance choices conditional on individual and household socioeconomic attributes. Specifically, temporal risk variations, captured by multi-stage behavioral experiments, alter respondents’ valuations of loan attributes. Empirical data were collected through a field survey integrating a multi-stage Behavioral Risk Elicitation Task (BRET) and a Discrete Choice Experiment (DCE). The BRET produces sequential risk indicators (K0–K4) and temporal risk-change terms (DiffK) to track dynamic risk-taking patterns across decision rounds. The DCE generates repeated individual-level discrete choice data via randomized microcredit alternatives that vary in interest rate, loan term, loan amount, mortgage type, repayment mode, and financial service intensity. To address individual preference heterogeneity and response scale heterogeneity in discrete choices, this study employs the Generalized Multinomial Logit (G-MNL) model. Unlike standard logit specifications, the G-MNL relaxes the Independence of Irrelevant Alternatives (IIA) restriction and accommodates flexible, decision-maker-specific preference and scale parameters. We estimate a hierarchical series of G-MNL specifications: (1) a baseline model without heterogeneous covariates; (2) a controlled benchmark model with full socioeconomic characteristics; (3) models incorporating static stage-wise risk indicators (K0–K4); (4) models with temporal risk-difference terms (DiffK); and (5) dynamic interactive models that capture path-dependent effects between risk preference shifts and prior decision outcomes. The results yield four key findings. First, rural households’ microfinance choices exhibit significant preference and scale heterogeneity, justifying the G-MNL framework. Second, static risk attitudes systematically explain microcredit preferences, suggesting that inherent risk profiles shape borrowing behaviors. Third, temporal risk changes exert robust heterogeneous effects on loan attribute preferences, highlighting the incremental explanatory power of dynamic risk traits over static risk measures. Fourth, prior decision outcomes, dynamic risk adjustments, and their interactions significantly influence choice scale with stage-dependent varying effects. Residual scale dispersion further confirms heterogeneous consistency in households’ microcredit decision-making.

Institutions

• Sichuan Agricultural University

  Sichuan, Ya'an

Categories

Licence