Econometric analyses of microfinance credit group formation, contractual risks and welfare impacts in Northern Ethiopia

Key words Microfinance, joint liability, contractual risk, group formation, risk-matching, impact evaluation, Panel data econometrics, dynamic panel probit, trend models, fixed-effects, composite counterfactuals, propensity score matching, farm households, Ethiopia. Lack of access to credit is a key obstacle for economic development in poor countries. The underlying problem is related to information asymmetry combined with the poor’s lack of collateral to pledge. New mechanisms in microfinance offer ways to deal with this problem without resorting to collateral requirements. The objective of this thesis is to examine the mechanisms of providing credit through microfinance and assess the long-run borrowing effects on household welfare in Ethiopia. The Ethiopian environment provides a suitable setting to examine these issues. To meet this objective, two unique data sets - a five-wave panel data on 400 and a cross-sectional data on 201 households - from northern Ethiopia are used. Borrowing decision is first conceptualized using a dynamic stochastic theoretical framework. Two types of risks involved in joint liability lending are incorporated, i.e., risk of partner failure and risk of losing future access to credit. Empirical analysis using recent dynamic panel data probit techniques show that these contractual risks indeed impede participation in borrowing. The impediment is higher for the poorer, and for new than repeat participants. Second, group formation is analyzed within the framework of alternative microeconomic theories of joint liability where the commonly held hypothesis that groups formed are homogeneous in risk profiles is tested. Empirical results reject this hypothesis indicating that the formation of heterogeneous risk profiles is an inherent feature in group formation and repayment. In fact, there is evidence that borrowers take advantage of established informal credit and saving, and other social networks, which also suggests that group formation outcomes vary depending on underlying socioeconomic contexts. Third, the impact of long-term borrowing on household welfare is assessed from the dimension of intensity and timing of participation in borrowing. Panel data covering relatively long period enabled to account for duration and timing concerns in program evaluation. Recent parametric and semi-parametric panel data techniques are innovatively employed to mitigate participation selection biases. Results from both approaches indicate that borrowing has increased household welfare significantly: the earlier and more frequent the participation the higher the impact partly due to lasting effects of credit. This also suggests that impact studies that are based on a single-shot observation of outcomes and that do not account for the timing and duration of participation may underestimate microfinance credit impacts.