Do outreach approaches differ between Self-Help Group-Bank Linkage and Microfinance Institution-based microfinance? Evidences from Indian states

Abstract

In India, currently, two prominent modes of microfinance operations are in vogue—Self-Help Group-Bank linkage programme and private Microfinance Institutions. However, their penetration across states is not uniform. As there are two approaches to microfinance outreach, namely ‘financial systems approach’ and ‘poverty lending approach’ (Robinson in The microfinance revolution: sustainable finance for the poor, World Bank, Washington, 2001), in Indian context, this study examines whether the approach to outreach is common to both the models or there exists divergence. The study observes the persistence of interregional divergence in microfinance penetration with larger concentration of loan portfolios and client base in the southern regions of the country. There is empirical support to the possibility of mission drift as microfinance providers tend to prefer relatively developed states. Despite some indications of poverty lending, the financial sustainability appears to be the key factor behind the skewed regional distribution of both the microfinance providers, hence warranting a robust regulatory system to overcome the problem.

Appendix

Construction of infrastructure development index

In the present study, following Raychaudhuri and Haldar 2009), the following steps were followed to construct the Index. First, we converted the raw data into normalized form to make the infrastructure indicators unit free. In order to do this, maximum and minimum values in a particular infrastructure indicator (X) were identified. Then, we applied the following formula to attain the normalized values for the particular indicator.

$${\text{NV}}_{ij} = \frac{{{\text{Observed}}\;X_{ij } - {\text{Minimum}}\;X_{ij} }}{{{\text{Maximum}}\;X_{ij} - {\text{Minimum}}\;X_{ij} }}$$

${\text{NV}}_{ij}$ is the normalized value, $i$ is the $i^{th}$ observation for $j^{th}$ state. The normalized values are having values 0–1. Table 7 presents the normalized values of different infrastructure indicators for all the states for a particular year (2008). Similar estimations were carried out for the remaining years.

Table 7 Normalized values of different infrastructure indicators

Second, we ran the Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy test and Bartlett’s test of sphericity to check whether the taken sample size was adequate and considered infrastructure indicators were having sufficient commonality. The test statistics for KMO test (0.579) and the Bartlett’s test of sphericity (p value of 0.001) qualified the preliminary conditions for carrying out PCA (Table 8).

Table 8 Kaiser–Meyer–Olkin and Bartlett’s test

Rotation method: varimax with Kaiser normalization

While running PCA, we identified eigenvalues that were more than one. As an example, we present above the case of Andhra Pradesh for which we found that extraction of two components explaining about 74% of the variance (Table 9). This is shown in the rotational component matrix as in Table 10. In order to estimate the weights of different indicators in the index, we multiplied first eigenvalue with the first extracted component, and second eigenvalue with the second extracted component. For this purpose, we considered mod values. The two values were summed up to obtain the weight of each variable in the infrastructure development index. For example, the weight of RL was estimated as follows:

$$W_{\text{RL}} = 2.188*0.627 + 1.499*0.012 = 1.390$$

Table 9 Principal component analysis

Table 10 Rotated component matrix

Similarly, the weights of all other indicators were calculated (Table 11).

Table 11 Weights of various indicators infrastructure

The following formula is applied to construct the index for each state:

$$I = \frac{{\left\{ {\mathop \sum \nolimits_{i = 1} X_{i} *W_{i} } \right\}}}{{\mathop \sum \nolimits_{i = 1} W_{i} }}$$

where $I$ is the index, $X_{i}$ is the normalized value for the $i^{th}$ indicator, and $W_{i}$ is the factor loading for the $i^{th}$ indicator. The following is an example of the IF index for Andhra Pradesh (AP).

State	$W_{\text{RL}}$ = 1.390	$W_{\text{VE}}$ = 1.296	$W_{\text{PO}}$ = 1.631	$W_{\text{SC}} = 1.529$	$W_{\text{HB}}$ = 1.496	IF index
AP	0.108 * 1.390	1.000 * 1.296	0.332 * 1.631	0.215 * 1.529	0.097 * 1.496	2.461/7.342 = 0.335

1. The total weights for all the indicators are 7.342
2. In the same manner, the IF indices for the rest of the states were estimated