Microfinance and poverty reduction: Evidence from Djibouti

• Published date: 01 May 2022
• Author: Mohamed Abdallah Ali,Mazhar Mughal,Dina Chhorn
• Date: 01 May 2022
• DOI: http://doi.org/10.1111/rode.12856

Rev Dev Econ. 2022;26:1189–1224. wileyonlinelibrary.com/journal/rode

|

1189

© 2022 John Wiley & Sons Ltd.

Received: 25 August 2020

|

Revised: 25 October 2021

|

Accepted: 27 December 2021

DOI: 10.1111/rode.12856

REGULAR ARTICLE

Microfinance and poverty reduction: Evidence

from Djibouti

MohamedAbdallah Ali1,2

|

MazharMughal1,2

|

DinaChhorn3,4

1Universite de Pau et des Pays de l’Adour,

E2S UPPA, CNRS, TREE, Pau, Aquitaine,

France

2ESC Pau Business School, Pau,

Aquitaine, France

3University of Bordeaux, Bordeaux,

France

4Royal University of Law and Economics,

Phnom Penh, Cambodia

Correspondence

Mazhar Mughal, ESC Pau Business

School, 64000 Pau, France.

Email: mazhar.mughal@esc-pau.fr

Abstract

Does access to microfinance improve household wel-

fare? We seek the answer to this question using data

on 2,060 borrower and nonborrower households based

in six major urban centers of Djibouti. We construct a

composite index of multidimensional poverty that cap-

tures various aspects of household well- being, includ-

ing ownership of agricultural and livestock assets, land,

transportation, employment, quality of housing, and

sanitation facilities. We carry out estimations using an

instrumental variable– based empirical strategy and a

number of econometric techniques. Our results show

that neither access to microcredit nor its ostensibly

productive use is significantly associated with poverty

regardless of the duration of time since the loan was

acquired. This holds for both access to and the amount

of microcredit obtained. The results are robust across

specifications and econometric techniques employed.

The lack of significant beneficial effect of microfinance

found in the study adds to the growing literature ques-

tioning the effectiveness of microfinance as a tool for

poverty alleviation. The findings raise doubts on the

usefulness of Djibouti's microfinance program.

KEYWORDS

Djibouti, microfinance, poverty, productive loans

JEL CLASSIFICATION

G21; G2; I32; I3; O16; O1

1190

|

ABDALLAH ALI et al.

1

|

INTRODUCTION

Extreme poverty is endemic to the countries located in the Horn of Africa, such as Djibouti.

With limited natural resources, a poorly developed industry, and climatic conditions unfavor-

able to agriculture, Djibouti has experienced economic crises since the 1990s. The deterioration

of the Djiboutian economy was due not only to domestic issues (proliferation of internal con-

flicts) but also to external difficulties (decline in development aid, increased regional competition

from the port of Assab,1 and influx of refugees from Somalia and Ethiopia). Between 1992 and

1996, the economic situation deteriorated sharply, and the gross domestic product (GDP) de-

clined (Foch,2010).2 Since 2010 however, the economic growth rate has increased substantially,

from 3.5% in 2010 to 7% in 2018. This strong growth fueled by investments in trade infrastruc-

ture has enabled the country to rank among the fastest- growing economies in the region (World

Bank,2018).

Thanks to a strong government focus on poverty alleviation, Djibouti has been able to reduce

extreme poverty from 42.2% to 21.3% between 2002 and 2017 (INSTAD,2018). A major social safety

net has been set up to combat poverty. The concept of microfinance has also been adopted as a

vital tool to improve the socioeconomic conditions of those segments of the population excluded

from the traditional banking system. After some initial steps taken in the 1990s, the microfinance

sector was recognized in the country during the 2000s with the establishment of the Social Fund

for Development (FSD), the Djiboutian Social Development Agency (ADDS), and an authority

managing microloans through credit unions (Caisse Populaire d’Epargne et de Crédit [CPEC]).

By 2016, one- third of Djibouti's households had gained access to microcredit mainly provided by

one of the three principal microfinance institutions (MFIs). These institutions together provided

US$12.3 million in loans. Despite this impressive growth of microfinance, there is little evidence

so far to show its effectiveness or its contribution to poverty reduction in Djibouti.

Microfinance is considered by some as a “magic wand” against poverty, a means to solve all

problems. However, evidence on the effectiveness of microfinance as a means of poverty reduc-

tion in developing countries is mixed. At one end of the spectrum, there are studies affirming that

microfinance is a positive and effective mechanism (e.g., Akotey & Adjasi,2016; Imai etal.,2010;

Imai & Azam, 2012; Rahman etal.,2015). Imai and Azam (2012) adopted the propensity score

method to analyze the effects of microcredit on household poverty reduction in Bangladesh and

Pakistan. Their results confirm the positive effects of microcredit on household poverty. Rahman

etal. (2015) quantified the effects of microcredit on household welfare by applying the differ-

ence in difference (DID) method to data collected in Shaanxi Province in China. Their findings

also suggest the desirable effects of microcredit. However, the DID approach cannot consider

unobservable factors such as borrower wealth, entrepreneurial skills, and motivation that can

potentially affect both microcredit utilization and outcome variables, while the DID approach

may still be subject to certain biases such as reverse causality and omitted variable bias. Other

empirical studies have attempted to address endog eneity issues associated with microcredit use

by using the endogenous treatment effects approach (e.g., Imai etal.,2010) to assess the poverty

reduction effects of productive lending by MFIs in India and the Heckman's sample selection

approach (e.g., Akotey & Adjasi,2016) to analyze the welfare- enhancing effects of microcredit

for MFIs in Ghana. These studies similarly found that the effects of microcredit were beneficial.

In contrast, some studies argue that microfinance actually plunges the population further into

poverty, affecting women in particular (e.g., Adams & Von Pischke,1992; Coleman,1999, 2006;

Duflo etal.,2015; Seng,2017). Studies by Coleman (1999, 2006) on households in northeastern

Thailand show that microcredit leads women to a vicious circle of debt with high interest rates

|

1191

ABDALLAH ALI et al.

coupled with the use of microcredit for nonproductive purposes. The failure of microcredit to

break the cycle of women's poverty is due to the fact that women typically invest in low- profit

businesses and can repay loans by borrowing from other financial sources, resulting in a heavy

debt burden (Seng,2017). These results support Adams and Von Pischke’s (1992) argument that

microfinance loans do little to lift poor households out of the poverty trap, nor can they improve

vulnerable households, but most likely exacerbate household poverty. Duflo etal. (2015) con-

ducted a study on 5,551 rural Moroccan households to estimate the effects of microcredit on

poverty. Their results confirm the negative effects of microcredit on household poverty.

Others have warned against considering microfinance as a miracle solution, claiming that it

helps poor households only to a limited extent, and have advocated its use with “cautious opti-

mism” (Banerjee et al.,2009; Duvendack & Palmer- Jones,2012; Karlan & Zinman, 2010). The

divergent conclusions drawn from existing literature are possibly a reflection of the diverse con-

texts of the studies as they focus on different geographical areas and use different methodologies.

In this study, we present evidence in this regard from Djibouti, a country hitherto not examined.

The interest in the country is particularly due to the significant penetration of MFIs and also due

to the fact that a large proportion of the country is vulnerable to climatic and geopolitical shocks

that increase the usefulness of microcredit to poor households. The study is based on a 2015 sur-

vey of 2,060 households based in the country's six major urban areas. The study contributes to

the literature in a number of ways.

First, the impact of microfinance on poverty is analyzed in a number of dimensions: the inci-

dence of a household acquiring microcredit, its ostensible productive use, the amount borrowed,

and the duration since the loan was acquired.

Second, the study goes beyond the use of headcount poverty indicator and generates a com-

posite indicator to estimate multidimensional poverty. The index captures various aspects of

household well- being, including employment; ownership of agricultural, livestock, transporta-

tion, and household assets; quality of housing; and availability of sanitation facilities.

Third, a number of empirical strategies are employed to tackle the possibility of reverse cau-

sality, omitted variable bias, and self- selection. The indicator for microfinance is instrumented

using information on the household's membership of a savings and credit cooperative. Though

a requirement for obtaining microcredit, the membership is free, widely acquired, and open to a

large segment of the society, for example, households (both poor and rich), small businesses, and

associations. Only a third (35%) of credit cooperative members end up acquiring a microcredit.

In addition, the study applies two alternative instruments as robustness check: the availability of

formal banks at the village or community level (which proxies for the proportion of households

with access to formal banks) and the availability of formal cooperatives at the village or commu-

nity level (which proxies for the proportion of households with access to formal cooperatives).

The estimations are also carried out using inverse probability weighting (IPW) and augmented

inverse probability weighting (AIPW). The results of the study are robust across specifications

and econometric techniques employed. The results show that neither access to microcredit nor

its ostensibly productive use is significantly associated with poverty regardless of the duration of

time since the loan was acquired. This holds for both access to and the amount of microcredit ob-

tained. The lack of significant beneficial effect of microfinance that the study reports adds to the

growing literature questioning the effectiveness of microfinance as a tool for poverty alleviation.

Section 2 summarizes the microfinance sector and the profile of poverty and vulnerabil-

ity in Djibouti. Section 3 presents the data and methodology used. Section4 discusses the re-

sults. Section5 describes the robustness measures. Section 6 concludes and provides policy

recommendations.