Effectiveness of Microfinance Policy in Nigeria, 2010-2017

This study evaluates the effectiveness of microfinance policy in Nigeria from 2010 to 2017. The Study used AMJU Microfinance Bank as its sample and the objectives of the study are to examine appropriate client targeting mechanisms that enhance good microfinance practice aimed at poverty reduction and how client impact evaluation improves operational efficiency and effectiveness. Propensity Score Matching (PSM) technique of micro econometric framework was used to establish the counterfactual for participants. The fundamental evaluation problem of selection bias was treated in the study and primary data obtained through interviews were analysed. The findings show that microfinance client exist rate was on the increase for established clients being significant; and client loan size dissatisfaction for older clients was on the rise and customers that have benefited from microcredits were better-off than non-beneficiaries. The study concludes that Microfinance is effective in poverty reduction and recommends amongst others that the delivery methodology should be tailored after their operational strategy and target clients; and also, appropriate feedback mechanism be built into their product and services delivery to encourage impact evaluation of participants’ responses, thereby, providing relevant inputs for the formulation of effective National Microfinance Policy in Nigeria.


Introduction 1.1 Background to the Study
There has been a growing interest towards recognizing the need to extend financial services freedom to the poor to enable them pursue legitimate economic diversification geared towards survival as well as improving their economic goals. Several schemes were advanced by government to provide the much needed financial succor to this target group, but were characterized by problems such as shallow rooted policy, poor implementation, high default rates, interest rate barriers, corruption and bias allocation (Johnson & Royaly, 1997;and Morduch, 1999).
The attention paid to this subsector by funding institutions and donor agencies as well as research work by various scholars are yet to yield the expected result on the economy as Nigeria is still ranked among the nations with high poverty index. This study is aimed at investigating the effectiveness of the National Microfinance Policy adopted by the Nigeria government and assessing the extent to which the policy has positively affected poverty level in the country. Unlike the past microfinance impact studies on Nigeria (Anyanwu, 2004;Udy, 1993) that focused on service availability and poverty reduction depth, this study used econometric analysis in assessing the effectiveness of Microfinance policies on Nigeria economy using AMJU UNIQUE (MFB) LTD as sample study.

Objectives of the Study
The primary objective of this study is to evaluate the effectiveness of microfinance policy in Nigeria, studying one of the vibrant performing microfinance institution, AMJU Unique (MFB) Ltd.
The specific objectives are to: (i) To determine the appropriate client targeting to promote good microfinance practice; (ii) To ascertain whether impact evaluation of microfinance services helps in improving operational efficiency of microfinance institutions.

Hypotheses of the Study
For the purpose of this study the following hypotheses were tested. Ho1: Client targeting strategy cannot promote good microfinance practice in Nigeria. Ho2: Impact evaluation is not an effective tool for formulation and implementation of national microfinance policy.

REVIEW OF RELATED LITERATURE 2.1 Conceptual Issues
Microfinance is a programme that extends small loans (credit) to the poor for self-employment projects that have the capacity to generate income, allowing for sustain ace of the entire family (micro-credit Summit, 1997). from stable below poverty line to stable above poverty line (Osthoff, 2005) and the vulnerability effect in which families are able to reduce the effects of income fluctuations thereby enhancing consumption smoothening and other coping strategies.
The participation dimension of poverty comprises many forms of deprivation such as humiliation and isolation, powerlessness and social inferiority (Hulme & Mosley, 1996). In this dimension of poverty, microfinance takes a broader perspective of poverty reduction to include non-material possessions. A major is the extent of empowerment of women. Empowerment strategy breaks the vicious circle of poverty.

Microfinance Service Provision
Microfinance institutions are expected to take into account varying needs of the poor in the design of their products. That is, designing and implementing microfinance services, there is need to consider that credit has varying implications for different segment of the poor and as such could create additional risk for them if not properly directed (Hulme & Mosely, 1996). Microfinance encompasses both financial and social intermediation including group formation, and training in financial literacy and management practices (Kalpana, 2004). Expedient upon microfinance institutions to diversify their hitherto relatively homogenous products and services to include environmental considerations.

Prospective Clients
The issue of who should constitute a majority of microfinance clients, men or women have been a friction. Some literature have opined female dominated clientele, but recent studies have proven otherwise (Brau & Woller, 2004;Amm et al, 2003). The argument in favour of women as clients is the assumption of better usage of credits and focus on family. A view of the sustainable Development Goal (SDG) is that women are critical to achieving these goals, hence the motivation to target them. The access to financial services empower women both financially and socially, hence their large number as clients to microfinance (Tassel, 2004) and World Bank (2007) confirm that most microfinance programmes are targeted towards. Women a contrary view was expressed by (Murduch, 1996) suggesting microfinance did not perform better with women as target. However, gender is not a determinant of poverty or poverty being gender sensitive.

The Business of Microfinance Banking in Nigeria
Nigeria is endowed with a huge population of abo9ut 200 million people as at 2018, of which 60% is predominantly rural (NBC, 2016) with a physical spread of 923,768km 2 . A West African country and strong player in regional economics and politics, bounded in the North by the Nigeria Republic, west by Benin Republic, east by the Chad and Cameroun and Gulf of Guinea in the South. It comprises of 36 Administrative states and a federal capital reserve, Abuja. These broken down into 774 local government areas (LGAs). She got her political independence in October of 1960, an hitherto colony of Britain. For over two decades, it was administered by the military. In May of 1999 a new democratic setting emerged and still running till date. Despite its seemly progress, it still parade indices of poverty characterized by unchecked population growth, slow development, unstable macroeconomic environment buffeted by internal and external crisis and a micro economy, heavily reliant on crude petroleum. The discovery, which have relegated other sectors. The cry for economic diversification had been on, but only serious when oil and gas price shocks are experienced and relapses as soon as the price is stable. The current dispensation is giving an unserious try to this effort but laden with huge debt. As at the time of this study, it owed $6.7 trillion external debt and N22.4 trillion internal debt with 15% attracted bytes 36 states of the country.
Specifically, in terms of macroeconomic indicators, Nigeria witnessed a consistent decline in growth rate, Gross Domestic Product (GDP) and rising inflation consecutively between 2014-2017, thereby eroding the saving capability of most households. Nigeria economy is also experiencing an increase in debt to GDP ratio resulting in low foreign investment and inability of the government to meet developmental programmes in the areas of poverty alleviation, infrastructural development and jobs creation. Also, credit extension to the private sector is low, limiting their capabilities of investment activities vis-à-vis poverty reduction.

Theoretical Issues in Microfinance
There is increasing need among development agencies, donors, government and private concerns to undertake a robust evaluation of any intervention programme. The fundamental question to address about a programme is if objectives have been met within cost. In other words, impact evaluation can provide information on whether a programme measurably benefits participants in comparison with those who did not participate. Although, many stakeholders are eager to undertake formal evaluation of their programmes due to cost implications and the limitation of the outcomes of evaluation (Ogiogio, 2006). A main constraint encountered in most evaluation programmes is getting key players to agree to conducting the evaluation as fear of vested interest may be hindered or due to other ethical objectives. Again, many organisation regard negative findings as hindrance to foster their agenda (Ravillion, 2005;Hulme, 2000 andand Baker, 1999). Despite these apprehension, the benefits of conducting impact evaluation are huge (Karlan & Goldberg, 2006).
To have a proper understand of the right methodology to employ in an impact evaluation, there is the need to have a grasp of the basic concept of impact evaluation. The World Bank (2002) defined impact evaluation on a "systematic identification of the effects on individuals, households, institutions and the environment caused by a given development activity such as a programme or project". Evaluation can be analytical assessment of a programme (OECD, 1999). Thus, an evaluation can take the form of beneficiary assessment, indicator monitoring, public expenditure tracking survey, rapid appraisals, concurrent assessment policy-level assessment and tracer studies (Bloomquist, 2003). It is aimed at assessing programmes performance against explicit counterfactual, such as the situation in the absence of the programme (Ravallion, 2005). Impact evaluation therefore, can be both exante and ex-post. Pitt et al (2003) also evaluated the Bangladesh (BRAC, BRDB, Grameen bank) using the maximum likelihood estimation controlling for endogeity of individual participation and of placement of microfinance programmes. Impact variables being health of men and women (arm measure, body mass under (BMI) and height-for-age). Result shows significantly positive effects of female credit on height-for-age and arm circumference of both men and women. Borrowing by men has either negative or non-significant impact on health of children.

Empirical Literature on Microfinance
Amin et al (2003) examined Bangladesh (ASD, Grameen Bank, BRAC) using non parametric test of stochastic dominance of average monthly consumption or members and nonmembers and maximum likelihood test of micro credit membership on vulnerability, consumption and household characteristics. Findings indicate members are poorer than non-members. Programmes are more successful at reaching the vulnerable. Kaboski and Townsend (2002) tested Thailand (production credit groups, rice banks, women groups, buffalo banks) using two stage LS and MLE test of microfinance impact on asset growth, probability of reduction in consumption in bad years, probability of becoming money lender, probability of starting business and probability of changing job. Separate estimation according to types of MFI. Result were that production credit groups and women groups combined with training and savings have positive impact on asset growth, although rice banks and buffalo banks have negative impacts. Emergency services training and savings help to smooth responses to income shock. Women groups help to reduce reliance on money lenders.

Data and Methodology 3.1 Sources of Data and Description
This study makes use of cross-section data of primary nature. A sample of six hundred and forty (640)  The poverty scores were calibrated as: Least poor = 0-27 code (12), Less poor = (28-45) code (2), average poor = (46-63), code (3), poor = (64-82) code (4) and poorest = (84-100) code (5). High scores are assigned to low level of the poverty indicators. A client is registered as a member to participate in AMJU programmes if his/her poverty score is at least 46 (i.e. average poor to poorest). Thus, programme placement is determined by a "proxymeans test" (assignment of a score to all potential participants as a function of observable characteristics) as often used for targeting anti-poverty programmes in developing countries (Ravallion, 2005).

Sample Selection
The sample was drawn from twenty (20) unions spread across ten (10) branches in five locations. The observable characteristics of interest include: age, sex, previous business experience (in years), loan stage, loan type and location. Data were also collected on marital status, education level, primary business and poverty scores at registration for membership and on completion of loan cycle. In line with Heckman, Ichimura and Todd (1997), these variables are those that influence simultaneously the participation decision as well as the outcome variable. Furthermore, the variables are either fixed over time or are measured before participation. Also as noted by Heckman, Lalonde, and Smith (1999), the data for both the treatment and non-treatment groups are from the same population, a basic requirement for matching. All variables were categorical data, hence reflecting only the direction of change and not the exact magnitude.
In addition to the pipeline comparator, this study also employed propensity score matching to correct for selection bias. Analysis of the characteristics of all clients was used to create the control group. To validate these observables, we carried out a qualitative fieldwork of a sample of clients of AMJU. A propensity function was generated linking client characteristics to the likelihood that a client will access loan from the programme on members. The results of a recent study by Frȍlich (2006) on Gender Wage Gap of College Graduates in the UK showed that the propensity score matching is justified under the same assumption than matching on covariates and Journal of Economics and Sustainable Development www.iiste.org ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online) Vol.11, No.2, 2020 that choice-based sampling can be ignored. In other words, PSM can be applied without conditional independence assumption and on non-independently and identically distributed (iid) data. Hence, PSM allows estimation of mean impacts without arbitrary assumptions about functional forms and error distributions. This facilitates testing for the presence of potentially complex interaction effects.

Model Specification
For an anti-poverty programme, the objective is usually defined in terms of household income or expenditure (on consumption) normalized by a household specific poverty score.
Given that the impact on poverty is known, then set Y = 1 as the outcome with treatment and Y = 0 as the outcome without treatment. Since an individual cannot be in both states, then it is not possible to observe Y = 0 and Y = 1 for the same individual thereby leading to the problem of missing data (Essama-Nssah, 2006;Ravallion, 2005 andWooldridge, 2002). To guard against the possibility of the case where the treatment of one unit affects another's outcome as may be in general equilibrium effect (Heckman et al., 1998), the sample from the population is assumed to be independently and identically distributed (iid). In many cases the outcomes Y = 0 and Y = 1 are binary.

Assessing the Matching Quality
To ensure that the matching procedure is able to balance the distribution of the relevant variables in both the control and treatment group, the quality of matching can be assessed using the Heckman-Hotz (1989) indirect test, standardized Bias, t-test, Joint significance and Pseudo-R 2 and Stratification test.
The Standardized Bias is used the distance in marginal distribution of the X-variables. For each covariate X, this is defined as the difference of sample means in the treated and matched control sub-samples as a percentage of the square root of the average of sample variances in both groups. The standardized bias for before and after matching are given as follows: Where is the mean (variance) in the treatment group before matching and the analogue for the control group. and are the corresponding values for the matched samples. The t-test uses a two-sample t-test to check if there are significant differences in covariate mean for the two groups (Rosebaum & Rubin, 1985). The Pseudo-R 2 shows the extent the regressors X explain the probability of participation. The stratification test is used to t4est after dividing the observations into strata, if within each stratum the distribution of X-variables is the same for both groups.

Estimating the Variance of Treatment Effects
The variances of the treatment effects (ATE and ATET) are not usually easy to compute because of the inclusion of the variance due to estimation of the propensity score (Caliendo & Kopeinig, 2005). However, under unconfoundedness assumption, Haln (1998) estimated the asymptotic variances of ATT and ATET as follows: Where ( $ ) are the conditional outcome variance for the treated (T = 1) and untreated (T = 0) observations. The estimation of the above variances can be carried out by variance approximation by Lechner (2001)  ;.< + $ |/ = 1 + =∑ ?∈{BCD E ? F G 9 : G . !" + -|/ = 0 … … … … … … … … 3.49 Where N1 is the number of matched treated individuals and I J is the number of timers individual j from control group has been used taking into account matching with replacement. The approach assumes homescedasticity of the variances of the outcome variables within treatment and control groups.
To distinguish between population and variances, Abadie and Imbens (2006) estimated the sample-average treatment effect on the treated (SATET) as: .
Depriving a matching variance estimator that does not require additional non-parametric estimation, the variance or SATET is given by: Where M is the number of matches and KM(i) is the number of times unit i is used as a match.

Counterfactual Construction
This study adopted the application of quasi-experimental design because of its ability to resolve problems of endogeneity associated with non-random programme placement and self-selection of members of the treatment group. The matching method using pipeline comparison group was adopted where the control group consisted of those clients who were assessed as poor, registered as members of the microfinance programme but were yet to access loans or had accessed loan the first time but were yet to complete their first loan cycle. In this study, the treatment and control group, have similar observable characteristics as portrayed by the poverty score at point of registration in addition to the demographic characteristics, thus, the issue of selection bias has been reduced. This matching ability of the control group was established using equations (3.37) to 3.44). Galasso and Ravallion (2004) and Chase (2002) affirm that the use of pipeline comparison helps to address the problem of latent heterogeneity. Testing for observable differences between the treatment and non-treatment, Galasso and Ravallion (2004) in their study of Social Protection Programme in Argentina in which the pipeline comparison was adopted, found that the observables including idiosyncratic shocks were well balanced between the two groups. Thus, for this study, the pipeline comparator was employed because there has not been any material change in the criteria for registration as a member of AMJU programmes.

Characteristics of AMJU Clients
In terms of gender targeting, about 82.8% of the union members of AMJU programmes are female members (Table  4.1). This is consistent with AMJU's mission of empowering poor clients who are locked out of institutional credit due to lack of command over land, stocks and other forms of acceptable collaterals.

Source(s): Author's Computation.
About 94.3% of the sampled clients are between 26 and 55 years of age. This age bracket is the bedrock of the economically active population. It reflects good targeting for a microfinance institution whose mission is to Journal of Economics and Sustainable Development www.iiste.org ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online) Vol.11, No.2, 2020 assist the active poor to build viable and sustainable micro-enterprises. Most AMJU clients (about 89%) have a minimum of full primary education. This facilitates better communication between AMJU staff and the clients. Also, it helps clients to quickly understand AMJU's philosophy and enhancing their skills in new business development.

.3 Source(s): Author's Computation.
The majority, about 90.0% of the clients sampled have lived with a partner or have been married. This is a good parameter in group formation as only 5.0% of the clients had not been involved in marital affairs. There is implied sense of responsibilities among clients as the microfinance services are expected to be directed towards the well-being of the families of clients.

Source(s): Author's Computation.
About sixty percent (60%) of AMJU clients are in the urban/semi-urban areas because of high population density. This facilitates group formation at low cost. However, most rural poor are excluded from benefiting from such services because of risk-return considerations. As noted by ADB (2000), most private MFIs are reluctant to invest in financial technology and innovative programmes oriented to the rural poor because the belief is that the market among the poor is limited and externalities will not allow the MFIs to profit from their investments. It appears that prior knowledge of a business is a requirement for eligibility to participate in AMJU programmes. This is probably to guard against fungibility of money and thus ensure that clients use the micro loans for intended purpose (improving their businesses) and thus translate to improvement in the general wellbeing of clients.
Journal of Economics and Sustainable Development www.iiste.org ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online) Vol.11, No.2, 2020   From table 4.7, it is evident that members of the treatment group are clients in their second loan stage or more while the control group consists of clients yet to receive any loan. This enhances the construction of appropriate counterfactual.

Source(s): Author's Computation.
In consonance with AMJU's mission, the targeting tool employed enhanced the recruitment of the poor into its programmes. From the sampled clients (table 4.8), about 98% of AMJU clients are considered at least to be average poor. This recruitment criterion is a major variable in using the propensity score matching methods to establish the adequacy of the non-treatment group.  Vol.11, No.2, 2020 provision/cosmetic shops/kiosk) which is usually a pre-condition for regular repayment programmes of microfinance. These types of small enterprises are usually supported through micro-loans. Also, the homogeneity in the type of businesses facilitates the group lending methodology. This may appear to present a threat to the livelihood of the group due to correlated business risks. However, there is an inherent opportunity as it helps to lower monitoring costs of group members because of their technical familiarity with other members' business activities (Pagura, 2003). The level of poverty reduction (table 4.10) among the treatment group shows that 82.0% of the clients noticed a reduction in their poverty score as a result of accessing loans from the AMJU programmes. The reduction in poverty level cuts across the eligibility criteria (irregular household income, poor nutritional status, unhealthy condition of dwelling place; etc) for recruitment of clients into AMJU programmes. The revelation enhances the attainment of AMJU's mission of targeting the poor. Nonetheless, 4.8% claimed that their level of poverty has worsened while about 12.2% did not notice any change in their poverty status after accessing loan from the programme. A major reasons is that the loan amount is too small to meet their business expansion requirements. It is therefore imperative that producer design should take into consideration the peculiar needs of certain clients instead of providing one-size fit-all products to all clients.

Policy Implications of Findings
The empirical findings of this study have the following policy implications that will help in the implementation of the national microfinance development strategy.
(i) A one-size-fits-all model microfinance delivery mechanism is counter-productive. Instead, MFIs should conduct their operations using a combination of delivery methodology (group or individual) that differentiates among the macroeconomic environment including spatial dispersion of population and other microfinance -driven characteristics (e.g. nature of clients' business and gender concentration). (ii) Policy makers should encourage the conduct of impact evaluation particularly on certain intermediate indicators.
Inter-temporal behavioural responses of participants in a programme are relevant to understanding their impacts (see Ravallion & Chen, 2005). (iii) A major challenge for policy makers is to design policies that promote microfinance practice. Microfinance policy should take into consideration the congruence between commercial objective and poverty outreach. This will ensure that impact assessment is not relegated to the backstage as a result of too much emphasis on institutional sustainability. Also, it is imperative to get this policy presumptuously right from the outset through consultations with all relevant stakeholders in the microfinance industry and pilot testing.

Summary of Findings, Conclusion and Recommendation 5.1 Summary of Findings
The main objective of this study is to determine how effective microfinance policies are in the Nigerian context in reducing poverty of target clients. The Nearest Neighbour, Radius and Kernel Matching Techniques based on propensity scores were used to analyse the effect of client registration in AMJU programmes. The following are summary of findings.
(i) As demonstrated by the increase in the exit rate, AMJU may consider innovations in delivery mechanism particularly the area of evolving individual lending for those clients that have attained some level of stability and relative independence in their businesses. (ii) Also, AMJU needs to review loan sizes since many members expressed dissatisfaction with small loan size. Failure to increase loan size may force some active members to drop out from the Journal of Economics and Sustainable Development www.iiste.org ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online) Vol.11, No.2, 2020 31 programme. It is important to keep in mind that for any financial service to have a lasting impact on poverty reduction, it should be delivered in flexible way and innovative to adapt to the needs of its clients. This product will enhance high repayment rate, engender business continuity and customer loyalty. (iii) In all, there were indicators that clients who have accessed loans (treatment group) are somewhat better off than those that are yet to benefit. The study has also helped to clarify the misconception that propensity score matching (PSM) needed to be modified in the presence of choice-based sampling, over -or under-sampling of treatment group as in Heckman, Ichimura, Smith and Todd (1998) and Heckman, Ichimura and Todd (1997).

Conclusion and Recommendations
One of the areas of interest to these stakeholders is the determination of impact of microfinance. In other words, impact assessment of microfinance intervention became a major issue in the development paradigm. In recent years, researches on impact evaluation produced mixed results due to environmental peculiarities, evaluation method applied by the researchers and operational methodologies by the various microfinance institutions. Nonetheless, despite the conflicting findings, attempts are still being made to ascertain the efficacy of microfinance in delivering the desired promise of poverty reduction. Unfortunately, most of the previous impact assessment focused on Latin America and Asia with little attention to Africa especially Nigeria. Therefore, this study is part of the ongoing efforts in the application of the growing field of microeconometrics in impact evaluation programmes. The study used a successful microfinance institutions, AMJU as a study sample.
The findings in the study reveal that microfinance delivery mechanism such as proper client targeting, appropriate product design, flexible regulatory stance is central to operational methodology. In particular, the study confirmed the assertion that repeat loan is an important feature of microfinance in achieving its poverty reduction objectives.
The following recommendations were made at the end of our study and includes; i. Delivery methodology for microfinance should be tailored after their individual operational strategy and client target. ii.
Feedback mechanism should be built into product delivery to enable impact evaluation of target participants' responses. iii.
Microfinance policies should take cognizance of commercial objective and poverty outreach.