To explore The Effect Of Loan Officer Rotation on the likelihood of an applicant acquiring a loan, we utilized dataset from Access Bank Madagascar (ABM). ABM is a commercial micro-credit or financial institution that was formed in October 2006 by Access Holding in corporation with a group of local and international co-investors. The first branch of the bank was opened in February 2007 at Antananarivo, the capital of Madagascar and has since developed rapidly to 26 branches across the country in various towns and villages. The core target of the bank is the concentration on the delivery of credit allied to others, the provision of banking services to micro, small and medium-sized enterprises (MSMEs) in the country.
The dataset comprises all loan/credit applications by clients from January 2007 to December 2012 from 16 branches of ABM. Branches are labeled Branch 1 through to Branch 16. These branches are classified as either rural or urban based on the location. For instance, branches such as branch 8, 9, 10, 13 and 16 are classified as rural. A total of 77,591 loan applications were recorded between the period under consideration. Given that the focus of this paper is to examine the effect of loan officer rotation on the likelihood of a client obtaining a loan, applicants who applied for loan/credit only once are excluded from the sample which reduced the sample from 77,591 to 56,905 observations. Based on the 56,905 observations, the treatment variable of interest, thus a binary indicator equal 1 if an applicant experienced at least a loan officer rotation and 0 if otherwise.
Once the treatment dummy was created and we were able to identify clients who dealt with different loan officers and those who did not, all duplicates were dropped reducing the sample further to 19,855 unique observations. The distribution of observations/clients in the various branches of ABM for the study are shown in Table (1) below. Thus, in our analysis we only had a single observation for each loan applicant. Summary of the key variables contained in the data and utilized in the analysis are presented in Table (2). Of the 19,855 loan applicants, 12,817 respondents were subjected to loan officer rotation whereas 7,038 respondents were not.
Using a unique dataset from Madagascar, we investigate how loan officer rotation affects credit access. We then test whether this effect is similar in both rural and urban centers. A probit model is utilized to investigate our first hypothesis of whether loan officer rotation has a positive and a significant effect on credit access. For the second hypothesis, we employ propensity score matching (PSM) to investigate whether loan officer rotation has similar impact on credit access in rural and urban areas. PSM use observable characteristics to build a statistical comparison group based on the likelihood of participating in a treatment or intervention (Khandker et al. 2009). In this regard, the PSM helps us to identify and classify all clients that accessed credit through more than one loan officer over the period of the study into one group (treatment group). This group of clients is similar in observable characteristics to those clients that accessed credit more than once from the bank through only one loan officer in the same period (control group).
The only difference between the two groups emanates from the rotation of loan officer which is the variable of interest in this case. In respect of the treatment (loan officer rotation), clients are further classified into rural or urban depending on the location of the branch they applied to access the loan. We believe that, per the observable characteristics, and partly driven by the types of loan portfolios provided in these branches, there is no systematic difference between clients accessing credit from similar branch or the same branch apart from the treatment variable (loan officer rotation) in relation to credit access, hence the basics for comparison.