V.4- Regression analysis
Introduced in section IV.3, three hypotheses are derived from the
problem statement.
These hypotheses can be tested using three regression models: the
(1) financial performance regression, (2) social performance regression,
and (3) informal sector regression. This section provides the financial
performance regression analysis, the social performance regression analysis,
and the informal sector regression. It is important to underline that the data
have been analyzed by the Statistical Package for Social Sciences (SPSS), Excel
and Statgraphics software
V.4.1- Financial performance regression analysis
There are three financial performance regression based on the
dependent variables,
namely ROA, ROE and OSS.
? MODEL SUMMARY: ROA in dependent variable
MODEL1: FD, HLAR, COA, AL: ROA Model2: FD, HLAR, COA, AL, CFIR
Model2: FD, HLAR, COA, AL CFGR
|
Model
|
R
|
R
Square
|
Adjusted
R Square
|
Std.
Error of
the
Estimate
|
Change Statistics
|
|
R Square Change
|
F
Change
|
df1
|
df2
|
Sig. F
Change
|
|
1
|
0.673*
|
0.453
|
0.399
|
6.878
|
0.4533
|
8.292
|
4
|
40
|
0.00
|
|
2
|
0.673**
|
0.453
|
0.383
|
6.966
|
0.00
|
0.00
|
1
|
39
|
0.993
|
|
3
|
0.674***
|
0.454
|
0.368
|
7.050
|
0.001
|
0.071
|
1
|
38
|
0.791
|
Tolerance: 0.00
* Predictors: (Constant), FD, HLAR, COA, AL
**Predictors: (Constant), FD, HLAR, COA, AL, CFIR
***Predictors: (Constant), FD, HLAR, COA, AL, CFIR, CFGR
63
Analysis of microfinances' performance and
development of informal institutions in Cameroon
By Djamaman Brice Gaétan
Table10: ANOVA analysis of ROA
regression
|
Model
|
|
Sum of
Squares
|
df
|
Mean Square
|
F
|
Sig.
|
|
1
|
Regression
|
1569.26501
|
4
|
392.316254
|
8.29235437
|
5.7212E-05
|
|
Residual
|
1892.42397
|
40
|
47.3105992
|
|
|
|
Total
|
3461.68898
|
44
|
|
|
|
|
2
|
Regression
|
1569.26831
|
5
|
313.853662
|
6.46806125
|
0.0001819
|
|
Residual
|
1892.42067
|
39
|
48.523607
|
|
|
|
Total
|
3461.68898
|
44
|
|
|
|
|
3
|
Regression
|
1572.8065
|
6
|
262.134417
|
5.27354558
|
0.00049015
|
|
Residual
|
1888.88248
|
38
|
49.7074337
|
|
|
|
Total
|
3461.68898
|
44
|
|
|
|
Table11: ROA regression coefficients
|
Model
|
|
Unstandardized Coefficients
|
Standardized Coefficients
|
t
|
Sig.
|
|
B
|
Std. Error
|
Beta
|
|
1
|
(Constant)
|
6.939
|
1.903
|
|
3.646
|
0.000
|
|
AL
|
0.536
|
0.684
|
2.945
|
0.784
|
0.437
|
|
COA
|
-0.043
|
0.013
|
-0.384
|
-3.238
|
0.002
|
|
HLAR
|
-0.003
|
0.009
|
-0.033
|
-0.285
|
0.776
|
|
FD
|
0.000
|
0.000
|
-3.477
|
-0.926
|
0.360
|
|
2
|
(Constant)
|
6.934
|
1.935
|
|
3.587
|
0.000
|
|
AL
|
0.536
|
0.692
|
2.944
|
0.774
|
0.443
|
|
COA
|
-0.043
|
0.013
|
-0.384
|
-3.184
|
0.002
|
|
HLAR
|
-0.003
|
0.009
|
-0.0337
|
-0.281
|
0.779
|
|
FD
|
0.000
|
0.000
|
-3.477
|
-0.915
|
0.366
|
|
CFIR
|
0.000
|
0.05
|
0.000
|
-0.008
|
0.993
|
|
3
|
(Constant)
|
7.046
|
1.998
|
|
3.526
|
0.001
|
|
AL
|
0.538
|
0.700
|
2.953
|
0.767
|
0.448
|
|
COA
|
-0.043
|
0.014
|
-0.385
|
-3.151
|
0.003
|
|
HLAR
|
-0.003
|
0.009
|
-0.034
|
-0.280
|
0.781
|
|
FD
|
0.000
|
0.000
|
-3.487
|
-0.906
|
0.371
|
|
CFIR
|
0.010
|
0.065
|
0.025
|
0.161
|
0.873
|
|
CFGR
|
-0.009
|
0.034
|
-0.041
|
-0.267
|
0.791
|
64
Analysis of microfinances' performance and
development of informal institutions in Cameroon
By Djamaman Brice Gaétan
Interpretation: the p-value of F test is 0.000.
This means that the overall model is statistically significant under ANOVA
analysis. The R square of all models is less than 0.50. This supposes that
there is a low relationship between independent variables and Return on
Asset.
F test, model 1: observed value 8.292 is greater than empirical
value F= 2.021. Then the regression equation is useful in the estimation of
ROA.
Based on T test, we can also conclude that all variables used in
the regression equation are useful to predict the return on asset.
F test, model 2: observed value 6.468 is greater than empirical
value 0.0150. Then the regression equation is useful in the estimation of ROA.
Consequently social performance variables influence ROA
F test, model 3: observed value 5.273; empirical value 0.0272.
Here, we can also conclude that all variables used in the regression equation
are useful to predict ROA.
As summary, social performance influence the Return on Asset.
Therefore, this can confirm the hypothesis of good management by the MFI. It is
important to underline that the significant level for all F test and T test is
0.05.
? MODEL SUMMARY: ROE in dependent variable
Model1: FD, HLAR, COA, AL
Model2: FD, HLAR, COA, AL, CFIR Model2: FD, HLAR, COA, AL CFGR
|
Model
|
R
|
R
Square
|
Adjusted
R Square
|
Std. Error
of the
Estimate
|
Change Statistics
|
|
R Square
Change
|
F
Change
|
df1
|
df2
|
Sig. F
Change
|
|
1
|
0.234*
|
0.05
|
-0.042
|
52.62214
|
0.055
|
0.564
|
4
|
39
|
0.690
|
|
2
|
0.290**
|
0.08
|
-0.036
|
52.46893
|
0.030
|
1.228
|
1
|
38
|
0.275
|
|
3
|
0.318***
|
0.101
|
-0.044
|
52.67756
|
|
0.700
|
1
|
37
|
0.408
|
* Predictors: (Constant), FD, HLAR, COA, AL
**Predictors: (Constant), FD, HLAR, COA, AL, CFIR
***Predictors: (Constant), FD, HLAR, COA, AL, CFIR, CFGR
65
Analysis of microfinances' performance and
development of informal institutions in Cameroon
By Djamaman Brice Gaétan
Table12: ANOVA OF ROE REGRESSION
|
Model
|
|
Sum of
Squares
|
df
|
Mean
Square
|
F
|
Sig.
|
|
1
|
Regression
|
6252.9397
|
4
|
1563.23492
|
0.56453017
|
0.68979795
|
|
Residual
|
107994.516
|
39
|
2769.09015
|
|
|
|
Total
|
114247.455
|
43
|
|
|
|
|
2
|
Regression
|
9633.87489
|
5
|
1926.77498
|
0.69988475
|
0.62690603
|
|
Residual
|
104613.581
|
38
|
2752.98896
|
|
|
|
Total
|
114247.455
|
43
|
|
|
|
|
3
|
Regression
|
11575.2045
|
6
|
1929.20075
|
0.69522609
|
0.6549512
|
|
Residual
|
102672.251
|
37
|
2774.9257
|
|
|
|
Total
|
114247.455
|
43
|
|
|
|
Table13: ROE regression Coefficients
|
Model
|
|
Unstandardized Coefficients
|
Standardized
Coefficients
|
t
|
Sig.
|
|
B
|
Std. Error
|
Beta
|
|
|
|
1
|
(Constant)
|
24.007
|
15.266
|
|
1.573
|
0.124
|
|
AL
|
6.549
|
5.231
|
6.260
|
1.252
|
0.218
|
|
COA
|
-0.033
|
0.116
|
-0.044
|
-0.280
|
0.781
|
|
HLAR
|
-0.002
|
0.071
|
-0.006
|
-0.035
|
0.972
|
|
FD
|
-0.003
|
0.002
|
-6.129
|
-1.226
|
0.228
|
|
2
|
(Constant)
|
22.314
|
15.298
|
|
1.459
|
0.153
|
|
AL
|
6.602
|
5.216
|
6.311
|
1.266
|
0.213
|
|
COA
|
-0.019
|
0.116
|
-0.025
|
-0.159
|
0.875
|
|
HLAR
|
-0.004
|
0.070
|
-0.009
|
-0.060
|
0.953
|
|
FD
|
-0.003
|
0.002
|
-6.180
|
-1.240
|
0.223
|
|
CFIR
|
0.416
|
0.375
|
0.173
|
1.108
|
0.275
|
|
3
|
(Constant)
|
20.033
|
15.600
|
|
1.284
|
0.207
|
|
AL
|
6.562
|
5.237
|
6.272
|
1.253
|
0.218
|
|
COA
|
-0.019
|
0.117
|
-0.026
|
-0.165
|
0.870
|
|
HLAR
|
-0.004
|
0.071
|
-0.009
|
-0.056
|
0.955
|
|
FD
|
-0.003
|
0.002
|
-6.134
|
-1.226
|
0.228
|
|
CFIR
|
0.160
|
0.485
|
0.067
|
0.330
|
0.743
|
|
CFGR
|
0.210
|
0.251
|
0.168
|
0.836
|
0.408
|
The first model here is not statistically significant because
its F value is lower than the significant level. The R square of all models is
less than 0.50. This supposes that there is a low relationship between
independent variables and Return on Equity.
66
Analysis of microfinances' performance and
development of informal institutions in Cameroon
By Djamaman Brice Gaétan
F test, model 1: observed value 0.564 is less than empirical
value which is 0.690. Thus the regression equation is not useful to predict the
ROE. Consequently social performance variables have no influence on ROE. We can
conclude that this is the hypothesis of mission drift or arbitration.
F test, model 2: as we can observe in ANOVA table of ROE, the
model is statistically significant. The R square of the model is less than
0.50. This implies that that there is a low relationship between independent
variables and Return on Equity. Observed value 0.699 of F test is greater than
the empirical value which is 0.408. We can conclude that all social variables
used in the regression equation are useful to predict ROE.
F test, model 3: under ANOVA table of ROE the model is
statistically significant. We found with the F test that observed value 0.695
is greater than empirical value 0.409. We can conclude that this model is
useful to estimate ROE. Thus social performance variables used in this model
influence the dependent variable ROE (assumption of good management
practice)
? MODEL SUMMARY: OSS in dependent variable
Model1: FD, HLAR, COA, AL
Model2: FD, HLAR, COA, AL, CFIR Model2: FD, HLAR, COA, AL,
CFGR
|
Model
|
R
|
R Square
|
Adjusted
R Square
|
Std.
Error of
the
Estimate
|
Change Statistics
|
|
R Square
Change
|
F Change
|
df1
|
df2
|
Sig. F
Change
|
|
1
|
0.552*
|
0.305
|
0.234
|
88.75586
|
0.30489
|
4.277
|
4
|
39
|
0.006
|
|
2
|
0.553**
|
0.305
|
0.214
|
89.88087
|
0.00054
|
0.030
|
1
|
38
|
0.864
|
|
3
|
0.553***
|
0.306
|
0.193
|
91.07649
|
0.00017
|
0.009
|
1
|
37
|
0.926
|
* Predictors: (Constant), FD, HLAR, COA, AL
**Predictors: (Constant), FD, HLAR, COA, AL, CFIR
***Predictors: (Constant), FD, HLAR, COA, AL, CFIR, CFGR
67
Analysis of microfinances' performance and
development of informal institutions in Cameroon
By Djamaman Brice Gaétan
Table14: ANOVA OF OSS REGRESSION
|
Model
|
|
Sum of
Squares
|
df
|
Mean
Square
|
F
|
Sig.
|
|
1
|
Regression
|
134756.714
|
4
|
33689.1786
|
4.27657748
|
0.00577733
|
|
Residual
|
307226.508
|
39
|
7877.60277
|
|
|
|
Total
|
441983.222
|
43
|
|
|
|
|
2
|
Regression
|
134997.562
|
5
|
26999.5125
|
3.34211531
|
0.01341161
|
|
Residual
|
306985.66
|
38
|
8078.57
|
|
|
|
Total
|
441983.222
|
43
|
|
|
|
|
3
|
Regression
|
135070.9
|
6
|
22511.8167
|
2.71392564
|
0.0275939
|
|
Residual
|
306912.322
|
37
|
8294.92762
|
|
|
|
Total
|
441983.222
|
43
|
|
|
|
Table15: OSS regression coefficients
|
Model
|
Unstandardized Coefficients
|
Standardized
Coefficients
|
t
|
Sig.
|
|
B
|
Std. Error
|
Beta
|
|
|
|
1
|
(Constant)
|
156.438
|
25.749
|
|
6.075
|
0.000
|
|
AL
|
-28.635
|
8.823
|
-13.916
|
-3.246
|
0.002
|
|
COA
|
-0.445
|
0.196
|
-0.307
|
-2.270
|
0.029
|
|
HLAR
|
-0.045
|
0.119
|
-0.050
|
-0.374
|
0.710
|
|
FD
|
0.013
|
0.004
|
13.843
|
3.229
|
0.003
|
|
2
|
(Constant)
|
155.986
|
26.207
|
|
5.952
|
0.000
|
|
AL
|
-28.621
|
8.935
|
-13.910
|
-3.203
|
0.003
|
|
COA
|
-0.441
|
0.200
|
-0.305
|
-2.210
|
0.033
|
|
HLAR
|
-0.045
|
0.121
|
-0.051
|
-0.373
|
0.711
|
|
FD
|
0.013
|
0.004
|
13.836
|
3.187
|
0.003
|
|
CFIR
|
0.111
|
0.643
|
0.024
|
0.173
|
0.864
|
|
3
|
(Constant)
|
155.543
|
26.971
|
|
5.767
|
0.000
|
|
AL
|
-28.629
|
9.054
|
-13.913
|
-3.162
|
0.003
|
|
COA
|
-0.441
|
0.202
|
-0.305
|
-2.182
|
0.036
|
|
HLAR
|
-0.045
|
0.122
|
-0.051
|
-0.368
|
0.715
|
|
FD
|
0.013
|
0.004
|
13.840
|
3.146
|
0.003
|
|
CFIR
|
0.061
|
0.839
|
0.013
|
0.073
|
0.942
|
|
CFGR
|
0.041
|
0.434
|
0.017
|
0.094
|
0.926
|
In the OSS regression, all the models are statistically
significant, because their respective F values (4.276, 3.342, and 2.713) are
greater than their respective significant level (0.005, 0.013 and
68
Analysis of microfinances' performance and
development of informal institutions in Cameroon
By Djamaman Brice Gaétan
0.027). The R square of all the models is less than 0.5; this
supposes that there is a low relationship between independent variables and
Operational Self Sufficiency.
F test, mode l: observed value 4.277 is greater than empirical
value 0.005. Therefore social performance variables used in this model is
useful to estimate OSS
F test, mode 2: observed value 3.342 is greater than empirical
value 0.075. The model is useful to estimate the Operational Self Sufficiency.
Thus all social variables used influence OSS
F test, mode 3: observed value 2.71 is greater than empirical
value 0.108 we can conclude that this model is useful to estimate OSS. Thus
social performance variables used in this model influence the dependent
variable OSS (assumption of good management practice).
| 
