Encuesta dirigida a los Docentes

The CI variables used emphasise bank traditional performance measures as well as non-financial factors such as transparent information reporting, customer satisfaction, sustainable growth and productivity (active clients), portfolio quality, and benchmarking critical performance to ensure good financial health of commercialising MFIs. This was due to the revelation that non-financial measures are better predictors of a firm’s long-run performance, and they help managers monitor and assess their firm’s progress towards strategic goals and objectives (Hussain, & Hoque, 2002). The nine performance indices, as composed by the 15 financial variables, reflect dimensions of interest to potential lenders and investors. These dimensions combined into the composite index were used in gauging the probability of success in tapping the financial markets. The performance indices were transformed into a single commercial financing rating score (CFR score) that is sensitive to differences in performance of an MFI with respect to its attractiveness to commercial lenders. The purpose of this process was to capture the complexity that goes into determining commercial viability of an MFI given diversity of success factors across countries in Africa. This single score defined as an index includes various effects of successful commercialisation and ability to attract commercial capital.

The index is constructed through a scoring process of the 15 criterion measures -m 1-15 (financial

ratio variables) grouped in the nine indices (see Appendix B for details). The CI consists of both a weight and a CFR component for each of the nine performance indices (Hendricks & Singhal, 2001; Laittinen, 2002). The measures taken together are intended to pay attention to and/or control the conditions specified by each of the performance indices. These conditions ensure sufficient success for good performance of a microfinance institution, and thereby attraction of commercial funding. However, intake of commercial capital needs to be controlled since heavy debt load can hurt an organisation. High leverage affects the probability of its default, as large amounts of debt increases the MFI’s interest charges and poses strain on cash flow (Barrios & Blanco, 2003; Berger et al., 1995). The index thus has inbuilt internal measures to ward off potential risk of high indebtedness.

The index was modeled using time series data of three years (between 1998-2003). It uses two years’ (for example 2002-2003) data for the development of the measure of future success and one prior-year’s (say 2001) financial information for predicting a two-year success in commercialisation (Laittinen, 2002; Pille & Paradi, 2002; Kolari et al., 2002). The CI is therefore a measure of future success in commercialisation, and is an indicator of future access to commercial financing opportunities measured in commercial financing rating scores (CFR). The CI balances access to leveraged financing with critical performance in microfinance business. The index defines degree of commercial orientation and informs management of the likelihood of success should the MFI decide to seek commercial funding. The index values are obtained by the following formula for CFR scores:

CI6 789 2002 2003 ∑F >? $@ABCDE

GH ,I ...(3.5)

That is; JDG, DGL, DGM, DGN, DGO, DGP, DGQ, DGR, DGES,I

Where >GI = Index of successful commercialisation for the (DG)with performance indices for the mj

the criteria measure. The nine financial performance measures in the index are equally weighted except for the LMR measure DG of effective access to funding, which has a higher weighting of 4

CFRs. Also, weights for the years 2002 and 2003 are the same, each with a weight of 1. The index assesses each MFI in the sample if the needed measurement criteria (critical performance for tapping commercial funding) for the performance indices have been met. If the m-criteria have been met, an increased attractiveness is identified by a simple addition procedure20. Finally, the two-year successful commercialisation prediction index is obtained by summing up the resulting CFR scores for the nine performance indices.

The CIij index scores are measured in CFR and scales from 0 to 25, whereby the maximum possible scores are 25. Higher CFR scores indicate likelihood of successful commercialisation. The median score (M) under this scale is 13 CFR scores and this is the critical value for the binary classification. The index was also conceptualised as a linear function of cumulative CFR scores for performance indices 1 to 9 minus the median; to arrive at normalised >6 789. This was

specified as

> 789 ∑EGH>? DGI , ...(3.6)

The median score was then normalised to zero to get a better visualisation of the binary classification, so that if index exceeds zero, an MFI is classified as successful. The index therefore

reflects the ease with which an MFI can tap capital from the wider financial market system, while maintaining performance sufficient for business excellence in microfinance. Thinking of the model index this way facilitated a more clear interpretation and exposition of the outcomes of the CI prediction index. Assuming a normal distribution of CFR scores, the classification can be illustrated as in Figure 3.1 below

Figure 3.1: Classification on CI scale

The CI measure of success defined as above was used to segment the sample MFIs into categories of successful and less successful. Classification was based on the index values (or CFR scores), with the cut-off being the critical value of 0 or median score of 13 CFRs. The sample comprises 103 MFIs across Africa that had completed three-year time series financial data between 1998 and 2003.

For each MFI in the sample, both Total CFR scores and CI values were generated. Higher scores indicate a higher likelihood of success while lower scores indicate high dependence on donations. CFR scores centred at the median show that, if the index exceeds zero, the MFI has high probability of success with access to commercial funding and in adapting a commercialisation strategy. The binary classification indicates those classified as successful coded as ‘1’, while those scoring less than 13 CFRs (or index values < 0) grouped as less successful and were coded ‘0’.

The binary classification for this measure of success resulted in 45 successful and 58 less successful MFIs.

Group of Successful MFIs Less successful MFIs

CI values

////or 0

Degree of Success >>>>>>>

3.3.5.1 Estimating the rating rule

The firms in the entire sample were classified into two groups for the two measures of success as explained above. As per the procedure in logistic modeling, the dependent variable, successful commercialisation, was converted into a dichotomous variable comprising those institutions more successful coded (1) and those that were less successful coded (0) for both two sets of success measures (Liu & Lee, 1997; Kennedy, 2001; Laittinen, 2002). Estimation of the binary variables (LMA and CI) was according to maximum likelihood. Future success in commercialisation, as measured for two years, was predicted by prior year one (2001) data using logistic regression analysis. Thus, if effective, the CI and the LMA will provide a useful commercial rating tool for preliminary screening of potential successful commercial MFIs.

The purposes of this logistic analysis was to estimate the conditional probability that an MFI belongs to the category of commercialising institutions, identify significant predictors, and to test the effectiveness of the models in classifying the sample of 103 firms. The choice of this statistical analysis was because the data set contained binary variables and it is said to be suitable where data is not normally distributed as opposed to conventional discriminant analysis (Laittinen, 2002; Kolari et al., 2002; Kennedy, 1998). It also allows for tests of overall fit of a model.

In the logistic classification model, the variable (y) refers to MFIs that are successful in commercialisation, and the probability of being successful is estimated by DBA TCC CU@@CCVUT AB W 1. This in turn implies that the probability of an MFI belonging to

the less successful category is:

XBA TCC CU@@CCVUT AB W 0 XBA 1 X W 1

The logic of discriminant analysis is formulated by the linear rating rule, namely classifying an MFI with characteristics given by the explanatory variables ( 9_{, ...}, 9_F) to category у equals 1 or 0, if

the conditions are met. The logistic regression model estimated by the method of maximum likelihood can be formulated as follows (Laittinen, 2002, Kolari et al., 2002, Kennedy, 1998):

X W 1

Z[\]

where: ^ _ 9 L9L . . . F9F

W = the dichotomous dependent variable, successful commercialisation

Ρ у 1 the conditional probability of an MFI being classified as successful or less successful a7 are the independent variables from 2001

b an intercept term

F the parameters for the logistic regression coefficients for predictor variables ( 9, ..., 9F) the quantity 2.1828+, the base of natural logarithms