Do bottom-up and independent agricultural cooperatives really perform better? Insights from a technical efficiency analysis in Ethiopia

ABSTRACT

The cooperative landscape in Ethiopia is very heterogeneous with a mixture of remains of the pre-1991 government-controlled system and new post-1991 bottom-up collective action initiatives. This heterogeneity, coupled with a large growth in the number of cooperatives in the country, offers an interesting perspective to study the determinants of the (in)efficiency of cooperatives. In this paper, we analyse the performance of Ethiopian agricultural cooperatives, focusing on the degree of technical (in)efficiency and its determinants. We use the stochastic frontier approach in which we account for heteroskedasticity and the monotonicity of production functions, presenting a methodological improvement with respect to previous technical efficiency studies. The results show that NGO- and government-initiated cooperatives are less efficient than community-initiated ones, implying that governments and NGOs should not interfere too strongly in cooperative formation. Cooperatives with a high degree of heterogeneity in members’ participation are found to be about 98% less efficient, while cooperatives that have paid employees are 33% more efficient. Besides, results show that cooperatives in Ethiopia function more efficiently if they incentivize committee members through monetary compensation.

KEYWORDS:

• Technical efficiency
• agricultural cooperatives
• stochastic frontier approach
• cooperative performance
• cooperative heterogeneity
• Ethiopia

JEL CLASSIFICATION:

• J54
• O13
• Q13

Acknowledgement

We thank seminar participants in Leuven for their useful comments on earlier versions of the paper.

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability statement

Data supporting the results in this paper are available from the corresponding author upon reasonable request.

Notes

1 The selected districts include Atsbi-Wenberta, Ganta-Afeshum, and Gulomkeda from the Eastern zone; Abergele, DeguaTembien, and Mereb-Leke from the Central zone; Alaje, Enderta, and Samre from the South & Southeastern (S&SE) zone; Asigede-Tsimbela, Tsegede, and Wolqayt from the West & Northwestern (N&NW) zone.

2 $\mathrm{E}\mathrm{T}\mathrm{B}\left(\mathrm{E}\mathrm{t}\mathrm{h}\mathrm{i}\mathrm{o}\mathrm{p}\mathrm{i}\mathrm{a}\mathrm{n}\mathrm{B}\mathrm{i}\mathrm{r}\mathrm{r}\right)\approx 0.04\mathrm{U}\mathrm{S}\mathrm{D}$ at the time of the study.

3 However, the SFA confounds the effects of misspecification of functional form with inefficiency. On the other hand, since DEA is non-parametric, it is less prone to specification error, but lumps noise and inefficiency together, calling the combination inefficiency.

4 With low-variance values, the probability that u = 0 is high, implying high probability that firms will be fully efficient. If the market is competitive, inefficient firms will be forced out of the market in the long run (i.e., it is very likely that the surviving firms will cluster around the fully efficient level). By contrast, if firms are from a regulated industry, one would expect convergence in efficiency to have occurred (efficiency levels would be similar though not necessarily close to 100%). If regulatory incentives are strong, including those for the more efficient firms, convergence should tend toward the frontier, again suggesting that the half-normal model would be appropriate (Kumbhakar et al. 2015).

5 That is, they collect members’ products for sale, and therefore, their customers (buyers) are not their own members to whom they may not charge the highest possible price.

6 Unlike a classical linear model in which heteroscedasticity affects only the efficiency of the estimators and not their consistency, ignoring heteroscedasticity in the SFA framework leads to inconsistent estimates (Wang and Schmidt 2002).

7 Since our output variable is proxied by sales, its level has implications for the marketing performance of cooperatives as well.

8 Self-initiated refers to member- and community-initiated cooperatives.

Additional information

Funding

The authors acknowledge funding from the VLIR-UOS TEAM Program (VLIR-UOS-ZEIN2015PR406 (13V95615T), Belgium.