Replicating Ashraf, Karlan and Yin (2006)

Ashraf, N., D. Karlan, and W. Yin. 2006. “Tying Odysseus to the Mast: Evidence From a Commitment Savings Product in the Philippines.” The Quarterly Journal of Economics 121 (2): 635–672. https://doi.org/10.1162/qjec.2006.121.2.635.

The following is based on the replication files for this paper (available from Harvard Dataverse). This is written in a jupyter notebook using the ipystata python library. The following reproduces only parts of the analysis from the do files: SEED.QJEtable.impact.do and SEED.QJEtables.takeup.do


In [4]:
import ipystata

Observations about the datasets

These are the original datasets. They have non-obvious names:

  • seedanalysis_011204_1.dta: -- Six month impacts

  • seedanalysis_011204_080404_1.dta: -- Twelve month impacts

  • seedanalysis_080404_1.dta: -- not sure yet..

The dataset has 3153 observations but the baseline survey appears to have been administered to only 1777 (if call==1).


In [7]:
%%stata
cd G:\GC\Dev-II\notebooks
use AKY06\seedanalysis_011204_080404_1_v11.dta

tab group
tab group if treatment !=.


G:\GC\Dev-II\notebooks

      group |      Freq.     Percent        Cum.
------------+-----------------------------------
          C |        809       25.66       25.66
          M |        776       24.61       50.27
          T |      1,568       49.73      100.00
------------+-----------------------------------
      Total |      3,153      100.00

      group |      Freq.     Percent        Cum.
------------+-----------------------------------
          C |        469       26.39       26.39
          M |        466       26.22       52.62
          T |        842       47.38      100.00
------------+-----------------------------------
      Total |      1,777      100.00

Table 6: Impact

OLS 6 months

Note that the 6 months data is in the file: AKY06\seedanalysis_011204_1.dta


In [8]:
%%stata
use AKY06\seedanalysis_011204_1_v11.dta
xi: reg balchange treatment marketing, robust
xi: reg balchange treatment if (treatment == 1 | marketing == 1), robust


Linear regression                               Number of obs     =      1,777
                                                F(2, 1774)        =       2.90
                                                Prob > F          =     0.0553
                                                R-squared         =     0.0018
                                                Root MSE          =     2293.2

------------------------------------------------------------------------------
             |               Robust
   balchange |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   treatment |   234.6785   101.7479     2.31   0.021     35.12004     434.237
   marketing |   184.8506   146.9815     1.26   0.209    -103.4246    473.1257
       _cons |   40.62573   61.67617     0.66   0.510    -80.33988    161.5913
------------------------------------------------------------------------------

Linear regression                               Number of obs     =      1,308
                                                F(1, 1306)        =       0.10
                                                Prob > F          =     0.7495
                                                R-squared         =     0.0001
                                                Root MSE          =     2550.2

------------------------------------------------------------------------------
             |               Robust
   balchange |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   treatment |   49.82794    156.027     0.32   0.750    -256.2631    355.9189
       _cons |   225.4763   133.4046     1.69   0.091     -36.2344     487.187
------------------------------------------------------------------------------

OLS 12 months

Note that the 12 months data is in the file: AKY06\seedanalysis_011204_080404_1.dta


In [4]:
%%stata
use AKY06\seedanalysis_011204_080404_1_v11.dta

xi: reg balchange treatment marketing, robust

xi: reg balchange treatment if (treatment == 1 | marketing == 1), robust


Linear regression                               Number of obs     =      1,777
                                                F(2, 1774)        =       1.43
                                                Prob > F          =     0.2398
                                                R-squared         =     0.0016
                                                Root MSE          =       4528

------------------------------------------------------------------------------
             |               Robust
   balchange |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   treatment |   411.4664   244.0205     1.69   0.092    -67.13149    890.0644
   marketing |    123.891   153.4397     0.81   0.420    -177.0506    424.8327
       _cons |     65.183   124.2152     0.52   0.600    -178.4406    308.8066
------------------------------------------------------------------------------

Linear regression                               Number of obs     =      1,308
                                                F(1, 1306)        =       1.58
                                                Prob > F          =     0.2085
                                                R-squared         =     0.0008
                                                Root MSE          =     5025.5

------------------------------------------------------------------------------
             |               Robust
   balchange |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   treatment |   287.5754   228.5226     1.26   0.208    -160.7361    735.8869
       _cons |    189.074   90.07236     2.10   0.036     12.37169    365.7764

In the analysis that follows e(sample)==1 restricts the calculations to the sample that was used in the lsat regression and _result(3) refers to the mean of hh_inc


In [5]:
%%stata

xi: reg balchange treatment marketing, robust
    
**economic impact #1: % of hh_inc 
summ hh_inc if e(sample)==1
disp _b[treatment] / (_result(3)*10000)

**economic impact #2: % of prior balance
summ totbal if e(sample)==1
disp _b[treatment] / (_result(3)/100)


Linear regression                               Number of obs     =      1,777
                                                F(2, 1774)        =       1.43
                                                Prob > F          =     0.2398
                                                R-squared         =     0.0016
                                                Root MSE          =       4528

------------------------------------------------------------------------------
             |               Robust
   balchange |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   treatment |   411.4664   244.0205     1.69   0.092    -67.13149    890.0644
   marketing |    123.891   153.4397     0.81   0.420    -177.0506    424.8327
       _cons |     65.183   124.2152     0.52   0.600    -178.4406    308.8066
------------------------------------------------------------------------------
. **economic impact #1: % of hh_inc 

    Variable |        Obs        Mean    Std. Dev.       Min        Max
-------------+---------------------------------------------------------
      hh_inc |      1,777    1.506908    1.953011          0    34.2623
.02730535

    Variable |        Obs        Mean    Std. Dev.       Min        Max
-------------+---------------------------------------------------------
      totbal |      1,777    50912.48    50489.83          0     223836
.80818389

Probit regressions


In [6]:
%%stata
xi: dprobit frac_change_00 treatment marketing, robust
xi: dprobit frac_change_00 treatment if (treatment == 1 | marketing == 1), robust
xi: dprobit frac_change_20 treatment marketing, robust
xi: dprobit frac_change_20 treatment  if (treatment == 1 | marketing == 1), robust


Iteration 0:   log pseudolikelihood = -1072.4045
Iteration 1:   log pseudolikelihood = -1064.6708
Iteration 2:   log pseudolikelihood = -1064.6667

Probit regression, reporting marginal effects           Number of obs =   1777
                                                        Wald chi2(2)  =  15.34
                                                        Prob > chi2   = 0.0005
Log pseudolikelihood = -1064.6667                       Pseudo R2     = 0.0072

------------------------------------------------------------------------------
         |               Robust
frac~_00 |      dF/dx   Std. Err.      z    P>|z|     x-bar  [    95% C.I.   ]
---------+--------------------------------------------------------------------
treatm~t*|   .1021901   .0266729     3.82   0.000   .473832   .049912  .154468
market~g*|   .0482741   .0314653     1.56   0.119    .26224  -.013397  .109945
---------+--------------------------------------------------------------------
  obs. P |   .2915025
 pred. P |    .290007  (at x-bar)
------------------------------------------------------------------------------
(*) dF/dx is for discrete change of dummy variable from 0 to 1
    z and P>|z| correspond to the test of the underlying coefficient being 0

Iteration 0:   log pseudolikelihood = -812.57798
Iteration 1:   log pseudolikelihood =  -810.3887
Iteration 2:   log pseudolikelihood =  -810.3884

Probit regression, reporting marginal effects           Number of obs =   1308
                                                        Wald chi2(1)  =   4.35
                                                        Prob > chi2   = 0.0369
Log pseudolikelihood =  -810.3884                       Pseudo R2     = 0.0027

------------------------------------------------------------------------------
         |               Robust
frac~_00 |      dF/dx   Std. Err.      z    P>|z|     x-bar  [    95% C.I.   ]
---------+--------------------------------------------------------------------
treatm~t*|   .0557175   .0263389     2.09   0.037   .643731   .004094  .107341
---------+--------------------------------------------------------------------
  obs. P |   .3126911
 pred. P |   .3121752  (at x-bar)
------------------------------------------------------------------------------
(*) dF/dx is for discrete change of dummy variable from 0 to 1
    z and P>|z| correspond to the test of the underlying coefficient being 0

Iteration 0:   log pseudolikelihood =  -784.0733
Iteration 1:   log pseudolikelihood = -772.39775
Iteration 2:   log pseudolikelihood = -772.35865
Iteration 3:   log pseudolikelihood = -772.35865

Probit regression, reporting marginal effects           Number of obs =   1777
                                                        Wald chi2(2)  =  22.93
                                                        Prob > chi2   = 0.0000
Log pseudolikelihood = -772.35865                       Pseudo R2     = 0.0149

------------------------------------------------------------------------------
         |               Robust
frac_~20 |      dF/dx   Std. Err.      z    P>|z|     x-bar  [    95% C.I.   ]
---------+--------------------------------------------------------------------
treatm~t*|   .1006771   .0223461     4.51   0.000   .473832   .056879  .144475
market~g*|   .0405021   .0274247     1.53   0.126    .26224  -.013249  .094253
---------+--------------------------------------------------------------------
  obs. P |   .1609454
 pred. P |   .1572024  (at x-bar)
------------------------------------------------------------------------------
(*) dF/dx is for discrete change of dummy variable from 0 to 1
    z and P>|z| correspond to the test of the underlying coefficient being 0

Iteration 0:   log pseudolikelihood = -617.43102
Iteration 1:   log pseudolikelihood =  -613.2009
Iteration 2:   log pseudolikelihood = -613.19501

Probit regression, reporting marginal effects           Number of obs =   1308
                                                        Wald chi2(1)  =   8.31
                                                        Prob > chi2   = 0.0039
Log pseudolikelihood = -613.19501                       Pseudo R2     = 0.0069

------------------------------------------------------------------------------
         |               Robust
frac_~20 |      dF/dx   Std. Err.      z    P>|z|     x-bar  [    95% C.I.   ]
---------+--------------------------------------------------------------------
treatm~t*|   .0636029   .0212161     2.88   0.004   .643731    .02202  .105186
---------+--------------------------------------------------------------------
  obs. P |   .1804281
 pred. P |   .1786767  (at x-bar)
------------------------------------------------------------------------------
(*) dF/dx is for discrete change of dummy variable from 0 to 1
    z and P>|z| correspond to the test of the underlying coefficient being 0

Quantile Regressions

6 months


In [7]:
%%stata
use AKY06\seedanalysis_011204_080404_1_v11.dta
sqreg balchange treatment marketing, q(.1 .2 .3 .4 .5 .6 .7 .8 .9)

sqreg balchange treatment if (treatment == 1 | marketing == 1), q(.1 .2 .3 .4 .5 .6 .7 .8 .9)


(fitting base model)

Bootstrap replications (20)
----+--- 1 ---+--- 2 ---+--- 3 ---+--- 4 ---+--- 5 
....x.x....x..x.....x....

Simultaneous quantile regression                    Number of obs =      1,777
  bootstrap(20) SEs                                 .10 Pseudo R2 =     0.0089
                                                    .20 Pseudo R2 =     0.0004
                                                    .30 Pseudo R2 =     0.0029
                                                    .40 Pseudo R2 =     0.0011
                                                    .50 Pseudo R2 =     0.0012
                                                    .60 Pseudo R2 =     0.0006
                                                    .70 Pseudo R2 =     0.0002
                                                    .80 Pseudo R2 =     0.0015
                                                    .90 Pseudo R2 =     0.0051

------------------------------------------------------------------------------
             |              Bootstrap
   balchange |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
q10          |
   treatment |     317.49   96.41809     3.29   0.001      128.385     506.595
   marketing |     327.15   106.2085     3.08   0.002      118.843     535.457
       _cons |    -828.09   89.08566    -9.30   0.000    -1002.814   -653.3661
-------------+----------------------------------------------------------------
q20          |
   treatment |         20   24.16008     0.83   0.408    -27.38521    67.38521
   marketing |         20   24.86448     0.80   0.421    -28.76677    68.76677
       _cons |       -300   22.99448   -13.05   0.000    -345.0991   -254.9009
-------------+----------------------------------------------------------------
q30          |
   treatment |     107.03   15.97238     6.70   0.000     75.70331    138.3567
   marketing |      100.9    31.1462     3.24   0.001     39.81289    161.9871
       _cons |    -276.99   12.32701   -22.47   0.000     -301.167    -252.813
-------------+----------------------------------------------------------------
q40          |
   treatment |   42.50999   18.72339     2.27   0.023     5.787762    79.23223
   marketing |   29.60999   17.30305     1.71   0.087    -4.326526    63.54651
       _cons |    -148.98   16.03131    -9.29   0.000    -180.4222   -117.5378
-------------+----------------------------------------------------------------
q50          |
   treatment |         62   12.94691     4.79   0.000      36.6072     87.3928
   marketing |      21.58   19.00313     1.14   0.256    -15.69088    58.85089
       _cons |     -98.22   9.810769   -10.01   0.000    -117.4619   -78.97812
-------------+----------------------------------------------------------------
q60          |
   treatment |      37.62    10.6566     3.53   0.000     16.71919    58.52081
   marketing |      22.59   23.21491     0.97   0.331    -22.94144    68.12144
       _cons |     -37.62    10.6566    -3.53   0.000    -58.52081   -16.71919
-------------+----------------------------------------------------------------
q70          |
   treatment |       6.55   2.694475     2.43   0.015     1.265321    11.83468
   marketing |   2.92e-13   1.094166     0.00   1.000     -2.14599     2.14599
       _cons |  -2.97e-13   1.16e-13    -2.57   0.010    -5.24e-13   -7.00e-14
-------------+----------------------------------------------------------------
q80          |
   treatment |      65.79   15.76751     4.17   0.000     34.86515    96.71485
   marketing |       4.02   6.252162     0.64   0.520    -8.242379    16.28238
       _cons |       6.85   1.559019     4.39   0.000     3.792292    9.907708
-------------+----------------------------------------------------------------
q90          |
   treatment |     437.23   148.7802     2.94   0.003      145.427     729.033
   marketing |     265.06   176.7912     1.50   0.134    -81.68094    611.8009
       _cons |      107.4   52.46879     2.05   0.041     4.492847    210.3072
------------------------------------------------------------------------------
(fitting base model)

Bootstrap replications (20)
----+--- 1 ---+--- 2 ---+--- 3 ---+--- 4 ---+--- 5 
xxxxxx.xx...xxx..xxxxxxxxx.x.x.xxxxxxxxxx..xxxxxxx.xxxxxxxxxx.xxxxxxxxx.xxx.xxx..xxxx..x.

Simultaneous quantile regression                    Number of obs =      1,308
  bootstrap(20) SEs                                 .10 Pseudo R2 =     0.0001
                                                    .20 Pseudo R2 =    -0.0000
                                                    .30 Pseudo R2 =     0.0000
                                                    .40 Pseudo R2 =     0.0003
                                                    .50 Pseudo R2 =     0.0006
                                                    .60 Pseudo R2 =     0.0001
                                                    .70 Pseudo R2 =     0.0001
                                                    .80 Pseudo R2 =     0.0012
                                                    .90 Pseudo R2 =     0.0006

------------------------------------------------------------------------------
             |              Bootstrap
   balchange |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
q10          |
   treatment |  -9.660004   105.8526    -0.09   0.927    -217.3198    197.9998
       _cons |    -500.94   105.6118    -4.74   0.000    -708.1273   -293.7527
-------------+----------------------------------------------------------------
q20          |
   treatment |          0   14.62935     0.00   1.000     -28.6996     28.6996
       _cons |       -280   10.26516   -27.28   0.000     -300.138    -259.862
-------------+----------------------------------------------------------------
q30          |
   treatment |    6.12999   32.19983     0.19   0.849    -57.03907    69.29905
       _cons |    -176.09   21.26357    -8.28   0.000    -217.8045   -134.3755
-------------+----------------------------------------------------------------
q40          |
   treatment |       12.9   14.01399     0.92   0.357    -14.59238    40.39239
       _cons |    -119.37   9.211683   -12.96   0.000    -137.4413   -101.2987
-------------+----------------------------------------------------------------
q50          |
   treatment |      40.42   16.54314     2.44   0.015     7.965957    72.87404
       _cons |     -76.64   12.82142    -5.98   0.000    -101.7928   -51.48717
-------------+----------------------------------------------------------------
q60          |
   treatment |      15.03     12.593     1.19   0.233    -9.674723    39.73472
       _cons |     -15.03     12.593    -1.19   0.233    -39.73472    9.674723
-------------+----------------------------------------------------------------
q70          |
   treatment |       6.55   3.169129     2.07   0.039     .3328592    12.76714
       _cons |   1.04e-13   .3108714     0.00   1.000     -.609862     .609862
-------------+----------------------------------------------------------------
q80          |
   treatment |      61.77   17.17601     3.60   0.000     28.07442    95.46558
       _cons |      10.87   4.421678     2.46   0.014      2.19563    19.54437
-------------+----------------------------------------------------------------
q90          |
   treatment |     172.17   192.5683     0.89   0.371    -205.6071    549.9471
       _cons |     372.46   164.7767     2.26   0.024     49.20397     695.716

12 months


In [8]:
%%stata
use AKY06\seedanalysis_011204_080404_v11.dta, clear
sqreg balchange treatment marketing, q(.1 .2 .3 .4 .5 .6 .7 .8 .9)
sqreg balchange treatment if (treatment == 1 | marketing == 1), q(.1 .2 .3 .4 .5 .6 .7 .8 .9)


file AKY06\seedanalysis_011204_080404_v11.dta not found
r(601);
(fitting base model)

Bootstrap replications (20)
----+--- 1 ---+--- 2 ---+--- 3 ---+--- 4 ---+--- 5 
....x..x...x.....xx.....xx.

Simultaneous quantile regression                    Number of obs =      1,777
  bootstrap(20) SEs                                 .10 Pseudo R2 =     0.0089
                                                    .20 Pseudo R2 =     0.0004
                                                    .30 Pseudo R2 =     0.0029
                                                    .40 Pseudo R2 =     0.0011
                                                    .50 Pseudo R2 =     0.0012
                                                    .60 Pseudo R2 =     0.0006
                                                    .70 Pseudo R2 =     0.0002
                                                    .80 Pseudo R2 =     0.0015
                                                    .90 Pseudo R2 =     0.0051

------------------------------------------------------------------------------
             |              Bootstrap
   balchange |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
q10          |
   treatment |     317.49   81.95606     3.87   0.000     156.7494    478.2306
   marketing |     327.15   75.29605     4.34   0.000     179.4717    474.8283
       _cons |    -828.09   79.19658   -10.46   0.000    -983.4184   -672.7616
-------------+----------------------------------------------------------------
q20          |
   treatment |         20   35.28501     0.57   0.571    -49.20457    89.20457
   marketing |         20   38.20277     0.52   0.601    -54.92716    94.92716
       _cons |       -300   35.68941    -8.41   0.000    -369.9977   -230.0023
-------------+----------------------------------------------------------------
q30          |
   treatment |     107.03   18.42875     5.81   0.000     70.88564    143.1743
   marketing |      100.9   19.00898     5.31   0.000     63.61764    138.1823
       _cons |    -276.99   15.21966   -18.20   0.000    -306.8403   -247.1396
-------------+----------------------------------------------------------------
q40          |
   treatment |   42.50999   9.363725     4.54   0.000      24.1449    60.87509
   marketing |   29.60999   11.60699     2.55   0.011     6.845176    52.37481
       _cons |    -148.98   9.653964   -15.43   0.000    -167.9143   -130.0457
-------------+----------------------------------------------------------------
q50          |
   treatment |         62   15.82094     3.92   0.000     30.97037    93.02963
   marketing |      21.58    20.5556     1.05   0.294    -18.73574    61.89574
       _cons |     -98.22    15.6571    -6.27   0.000    -128.9283   -67.51171
-------------+----------------------------------------------------------------
q60          |
   treatment |      37.62   13.34395     2.82   0.005     11.44848    63.79152
   marketing |      22.59   19.36071     1.17   0.243    -15.38219    60.56219
       _cons |     -37.62   13.34395    -2.82   0.005    -63.79152   -11.44848
-------------+----------------------------------------------------------------
q70          |
   treatment |       6.55    3.06789     2.14   0.033     .5329406    12.56706
   marketing |   2.92e-13   1.276839     0.00   1.000    -2.504266    2.504266
       _cons |  -2.97e-13   1.10e-13    -2.69   0.007    -5.14e-13   -8.06e-14
-------------+----------------------------------------------------------------
q80          |
   treatment |      65.79   20.26571     3.25   0.001     26.04281    105.5372
   marketing |       4.02   6.550422     0.61   0.539    -8.827356    16.86736
       _cons |       6.85    3.14329     2.18   0.029     .6850583    13.01494
-------------+----------------------------------------------------------------
q90          |
   treatment |     437.23    99.8697     4.38   0.000     241.3553    633.1047
   marketing |     265.06   211.3404     1.25   0.210    -149.4423    679.5623
       _cons |      107.4   56.54647     1.90   0.058    -3.504705    218.3047
------------------------------------------------------------------------------
(fitting base model)

Bootstrap replications (20)
----+--- 1 ---+--- 2 ---+--- 3 ---+--- 4 ---+--- 5 
xxxxxxx.xx.xxxxxxxxx.xxxx.xxxxxxxxxxx.xxxxxx..xxxxxxxxxxxx.xxx.xxxxxxxxx.xxxxxxxx..xxxx.xxx.xxxxxxxxxxx.xx.xxx.xxxx.xxx.xxx.

Simultaneous quantile regression                    Number of obs =      1,308
  bootstrap(20) SEs                                 .10 Pseudo R2 =     0.0001
                                                    .20 Pseudo R2 =    -0.0000
                                                    .30 Pseudo R2 =     0.0000
                                                    .40 Pseudo R2 =     0.0003
                                                    .50 Pseudo R2 =     0.0006
                                                    .60 Pseudo R2 =     0.0001
                                                    .70 Pseudo R2 =     0.0001
                                                    .80 Pseudo R2 =     0.0012
                                                    .90 Pseudo R2 =     0.0006

------------------------------------------------------------------------------
             |              Bootstrap
   balchange |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
q10          |
   treatment |  -9.660004   64.97214    -0.15   0.882    -137.1212    117.8012
       _cons |    -500.94   54.30732    -9.22   0.000    -607.4791   -394.4009
-------------+----------------------------------------------------------------
q20          |
   treatment |          0   22.41162     0.00   1.000    -43.96671    43.96671
       _cons |       -280   4.786988   -58.49   0.000     -289.391    -270.609
-------------+----------------------------------------------------------------
q30          |
   treatment |    6.12999   33.11795     0.19   0.853    -58.84021    71.10019
       _cons |    -176.09   21.44269    -8.21   0.000    -218.1559   -134.0241
-------------+----------------------------------------------------------------
q40          |
   treatment |       12.9   12.24727     1.05   0.292    -11.12647    36.92648
       _cons |    -119.37   8.031628   -14.86   0.000    -135.1263   -103.6137
-------------+----------------------------------------------------------------
q50          |
   treatment |      40.42   15.55394     2.60   0.009     9.906553    70.93344
       _cons |     -76.64   10.85084    -7.06   0.000    -97.92698   -55.35301
-------------+----------------------------------------------------------------
q60          |
   treatment |      15.03   12.87917     1.17   0.243    -10.23613    40.29613
       _cons |     -15.03   12.87917    -1.17   0.243    -40.29613    10.23613
-------------+----------------------------------------------------------------
q70          |
   treatment |       6.55   2.234516     2.93   0.003     2.166367    10.93363
       _cons |   1.04e-13   .2258429     0.00   1.000    -.4430545    .4430545
-------------+----------------------------------------------------------------
q80          |
   treatment |      61.77   22.97972     2.69   0.007     16.68879    106.8512
       _cons |      10.87   8.091951     1.34   0.179    -5.004645    26.74464
-------------+----------------------------------------------------------------
q90          |
   treatment |     172.17   245.2228     0.70   0.483    -308.9036    653.2436
       _cons |     372.46   225.9348     1.65   0.099    -70.77494    815.6949

Table VIII: ITT effect of subgroups

12 months


In [9]:
%%stata
xi: reg balchange treatment marketing female female_treat, robust 
xi: reg balchange treatment marketing active active_treat, robust
xi: reg balchange treatment marketing edhi edhi_treat, robust
xi: reg balchange treatment marketing hi_hh_inc hi_hh_inc_treat, robust
xi: reg balchange treatment marketing hyper_mon_new2 hyper_mon_new2_treat, robust
xi: reg balchange treatment marketing silly_mon_new2 silly_mon_new2_treat, robust


Linear regression                               Number of obs     =      1,777
                                                F(4, 1772)        =       1.98
                                                Prob > F          =     0.0948
                                                R-squared         =     0.0022
                                                Root MSE          =     4529.2

------------------------------------------------------------------------------
             |               Robust
   balchange |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   treatment |   680.2893     420.26     1.62   0.106     -143.968    1504.547
   marketing |   137.2044   150.0914     0.91   0.361    -157.1704    431.5792
      female |   192.9631   135.0963     1.43   0.153    -72.00175    457.9279
female_treat |  -443.4216    483.559    -0.92   0.359    -1391.828    504.9845
       _cons |  -53.72176   93.64086    -0.57   0.566    -237.3799    129.9364
------------------------------------------------------------------------------

Linear regression                               Number of obs     =      1,777
                                                F(4, 1772)        =       8.72
                                                Prob > F          =     0.0000
                                                R-squared         =     0.0040
                                                Root MSE          =     4524.9

------------------------------------------------------------------------------
             |               Robust
   balchange |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   treatment |   676.3477   327.5403     2.06   0.039     33.94163    1318.754
   marketing |   122.4113   152.3797     0.80   0.422    -176.4515    421.2741
      active |   637.8621   204.6197     3.12   0.002     236.5407    1039.183
active_treat |  -738.1954   393.8328    -1.87   0.061    -1510.621    34.23035
       _cons |   -164.665   81.52621    -2.02   0.044    -324.5626   -4.767302
------------------------------------------------------------------------------

Linear regression                               Number of obs     =      1,777
                                                F(4, 1772)        =       1.57
                                                Prob > F          =     0.1785
                                                R-squared         =     0.0018
                                                Root MSE          =       4530

------------------------------------------------------------------------------
             |               Robust
   balchange |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   treatment |   247.7803   362.0502     0.68   0.494    -462.3101    957.8708
   marketing |   122.8684   154.1362     0.80   0.425    -179.4395    425.1764
        edhi |  -145.0309   166.6162    -0.87   0.384    -471.8158     181.754
  edhi_treat |   279.7698   448.2779     0.62   0.533    -599.4393    1158.979
       _cons |   148.0578   200.4277     0.74   0.460    -245.0418    541.1574
------------------------------------------------------------------------------

Linear regression                               Number of obs     =      1,777
                                                F(4, 1772)        =       2.05
                                                Prob > F          =     0.0849
                                                R-squared         =     0.0019
                                                Root MSE          =     4529.9

---------------------------------------------------------------------------------
                |               Robust
      balchange |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
      treatment |   464.2607   271.0703     1.71   0.087    -67.39051     995.912
      marketing |   131.9819   150.2833     0.88   0.380    -162.7692     426.733
      hi_hh_inc |   193.5094    153.943     1.26   0.209    -108.4197    495.4384
hi_hh_inc_treat |  -106.6214   444.0923    -0.24   0.810    -977.6211    764.3784
          _cons |  -32.60318   84.76645    -0.38   0.701    -198.8559    133.6496
---------------------------------------------------------------------------------

Linear regression                               Number of obs     =      1,774
                                                F(4, 1769)        =       0.89
                                                Prob > F          =     0.4683
                                                R-squared         =     0.0018
                                                Root MSE          =     4533.8

--------------------------------------------------------------------------------------
                     |               Robust
           balchange |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
---------------------+----------------------------------------------------------------
           treatment |   344.6337     290.47     1.19   0.236    -225.0667    914.3342
           marketing |    126.032   153.4896     0.82   0.412     -175.008     427.072
      hyper_mon_new2 |  -28.40704   132.3356    -0.21   0.830    -287.9576    231.1436
hyper_mon_new2_treat |   243.8663   470.7961     0.52   0.605    -679.5089    1167.242
               _cons |   72.63304   142.1695     0.51   0.609    -206.2049     351.471
--------------------------------------------------------------------------------------

Linear regression                               Number of obs     =      1,774
                                                F(4, 1769)        =       2.09
                                                Prob > F          =     0.0800
                                                R-squared         =     0.0023
                                                Root MSE          =     4532.8

--------------------------------------------------------------------------------------
                     |               Robust
           balchange |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
---------------------+----------------------------------------------------------------
           treatment |   516.7935    261.952     1.97   0.049     3.025574    1030.561
           marketing |    127.571   151.9509     0.84   0.401    -170.4513    425.5932
      silly_mon_new2 |   284.8332   353.4208     0.81   0.420    -408.3331    977.9995
silly_mon_new2_treat |  -633.5807   448.5194    -1.41   0.158    -1513.264     246.103
               _cons |   15.99006   80.69286     0.20   0.843    -142.2733    174.2534

There is a lot more to the impact results in the original do file.... but for now the analysis here turns to the takeup questions:

TAKEUP


In [15]:
%%stata

use AKY06\seedanalysis_011204_080404_1_v11.dta, clear
gen control = 1 if group == "C"
replace control = 0 if control ==.
macro define GB "GBloan GBloandefault"
macro define fullset "married edhi numhh unemployed age $GB hh_inc hh_inc2"
macro define fullset_noGB "married edhi numhh unemployed age hh_inc hh_inc2"

gen dormant_new = 1- active
drop dormant
rename dormant_new dormant
replace totbal = totbal/100
replace newtotbal = newtotbal/100
gen dist_GB = dbutuan if butuan ==1
replace dist_GB = dampayon if ampayon == 1
destring pop, ignore(",") replace
gen brgy_penetration = no_clients /pop
bysort brgy: egen sd_totbal = sd(totbal)
bysort brgy: egen mean_totbal = mean(totbal)


(2344 missing values generated)
(2344 real changes made)
(2839 real changes made)
(2193 real changes made)
(762 missing values generated)
(762 real changes made)
pop: characters , removed; replaced as int

In [7]:
%%stata
tab active if call==1,summ(totbal)


            |  Summary of Client savings balance
     Active |             (hundreds)
    account |        Mean   Std. Dev.       Freq.
------------+------------------------------------
          0 |   483.12239   498.52538        1145
          1 |    556.2335    513.2936         632
------------+------------------------------------
      Total |   509.12477   504.89831        1777

Table 1: Savings Goals


In [8]:
%%stata
tab goal_category if seedtakeup==1
tab goal_type if seedtakeup==1
tab box if seedtakeup==1


                    Savings goal |      Freq.     Percent        Cum.
---------------------------------+-----------------------------------
Agricultural Financing/Investing |          4        2.04        2.04
            Capital for Business |         20       10.20       12.24
Christmas/Birthday/Celebration/G |         95       48.47       60.71
                       Education |         41       20.92       81.63
House/Lot construction and purch |         20       10.20       91.84
                         Medical |          1        0.51       92.35
  Personal Needs/Future Expenses |          3        1.53       93.88
Purchase or Maintenance of Machi |          8        4.08       97.96
                 Vacation/Travel |          4        2.04      100.00
---------------------------------+-----------------------------------
                           Total |        196      100.00

    Type of |
    savings |
       goal |      Freq.     Percent        Cum.
------------+-----------------------------------
     amount |         62       30.69       30.69
       date |        140       69.31      100.00
------------+-----------------------------------
      Total |        202      100.00

     Bought |
  ganansiya |
        box |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |         35       17.33       17.33
          1 |        167       82.67      100.00
------------+-----------------------------------
      Total |        202      100.00

Table II: Summary Statistics


In [24]:
%%stata
tab call


                 Result of call |      Freq.     Percent        Cum.
--------------------------------+-----------------------------------
                      completed |      1,777      100.00      100.00
--------------------------------+-----------------------------------
                          Total |      1,777      100.00

Panel A


In [22]:
%%stata
tabstat totbal active if call==1, by(group) stats(mean sem)
reg totbal treatment control if call ==1
reg active treatment control if call ==1
sort brgy
tabstat dist_GB brgy_penetration sd_totbal mean_totbal pop if call==1, by(group) stats(mean sem)
reg dist_GB treatment control if call ==1
reg brgy_penetration treatment control if call ==1
reg sd_totbal treatment control if call ==1
reg mean_totbal treatment control if call ==1
reg pop treatment control if call ==1


Summary statistics: mean, se(mean)
  by categories of: group 

 group |    totbal    active
-------+--------------------
     C |  530.7378  .3603412
       |  23.38712  .0221926
-------+--------------------
     M |  499.0084  .3626609
       |  23.50222  .0222951
-------+--------------------
     T |   502.685  .3491686
       |  17.33115  .0164382
-------+--------------------
 Total |  509.1248  .3556556
       |  11.97734  .0113593
----------------------------

      Source |       SS           df       MS      Number of obs   =     1,777
-------------+----------------------------------   F(2, 1774)      =      0.59
       Model |  301690.551         2  150845.275   Prob > F        =    0.5536
    Residual |   452440317     1,774  255039.638   R-squared       =    0.0007
-------------+----------------------------------   Adj R-squared   =   -0.0005
       Total |   452742008     1,776  254922.302   Root MSE        =    505.01

------------------------------------------------------------------------------
      totbal |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   treatment |   3.676564   29.15807     0.13   0.900    -53.51121    60.86434
     control |   31.72941   33.03165     0.96   0.337    -33.05563    96.51446
       _cons |   499.0084   23.39434    21.33   0.000      453.125    544.8918
------------------------------------------------------------------------------

      Source |       SS           df       MS      Number of obs   =     1,777
-------------+----------------------------------   F(2, 1774)      =      0.15
       Model |   .06859733         2  .034298665   Prob > F        =    0.8612
    Residual |  407.157064     1,774  .229513565   R-squared       =    0.0002
-------------+----------------------------------   Adj R-squared   =   -0.0010
       Total |  407.225661     1,776  .229293728   Root MSE        =    .47908

------------------------------------------------------------------------------
      active |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   treatment |  -.0134923   .0276604    -0.49   0.626    -.0677428    .0407582
     control |  -.0023198   .0313351    -0.07   0.941    -.0637773    .0591377
       _cons |   .3626609   .0221928    16.34   0.000     .3191342    .4061876
------------------------------------------------------------------------------

Summary statistics: mean, se(mean)
  by categories of: group 

 group |   dist_GB  brgy_p~n  sd_tot~l  mean_t~l       pop
-------+--------------------------------------------------
     C |  21.86567  .0219365  487.1321  473.3466  5853.949
       |  .8419475  .0004179  3.507831   3.74361  213.4831
-------+--------------------------------------------------
     M |  23.22961  .0216379  491.3003  476.9744  5708.208
       |  .8874667  .0004076  3.449334  3.716019  203.4358
-------+--------------------------------------------------
     T |  22.70784  .0219269  488.0445  476.0261    5729.5
       |  .6718419   .000292  2.442733  2.599624    152.98
-------+--------------------------------------------------
 Total |   22.6224  .0218536  488.6575  475.5676  5756.762
       |  .4525451  .0002066  1.735807  1.854791  106.1334
----------------------------------------------------------

      Source |       SS           df       MS      Number of obs   =     1,777
-------------+----------------------------------   F(2, 1774)      =      0.61
       Model |  446.531678         2  223.265839   Prob > F        =    0.5417
    Residual |  645883.097     1,774  364.082918   R-squared       =    0.0007
-------------+----------------------------------   Adj R-squared   =   -0.0004
       Total |  646329.629     1,776   363.92434   Root MSE        =    19.081

------------------------------------------------------------------------------
     dist_GB |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   treatment |  -.5217753   1.101679    -0.47   0.636      -2.6825     1.63895
     control |  -1.363942   1.248034    -1.09   0.275    -3.811715    1.083831
       _cons |   23.22961   .8839083    26.28   0.000       21.496    24.96322
------------------------------------------------------------------------------

      Source |       SS           df       MS      Number of obs   =     1,777
-------------+----------------------------------   F(2, 1774)      =      0.19
       Model |   .00002944         2   .00001472   Prob > F        =    0.8238
    Residual |  .134701232     1,774  .000075931   R-squared       =    0.0002
-------------+----------------------------------   Adj R-squared   =   -0.0009
       Total |  .134730672     1,776  .000075862   Root MSE        =    .00871

------------------------------------------------------------------------------
brgy_penet~n |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   treatment |    .000289   .0005031     0.57   0.566    -.0006977    .0012758
     control |   .0002987   .0005699     0.52   0.600    -.0008192    .0014165
       _cons |   .0216379   .0004037    53.60   0.000     .0208462    .0224296
------------------------------------------------------------------------------

      Source |       SS           df       MS      Number of obs   =     1,777
-------------+----------------------------------   F(2, 1774)      =      0.44
       Model |  4662.35264         2  2331.17632   Prob > F        =    0.6473
    Residual |  9504305.63     1,774  5357.55672   R-squared       =    0.0005
-------------+----------------------------------   Adj R-squared   =   -0.0006
       Total |  9508967.98     1,776  5354.14864   Root MSE        =    73.195

------------------------------------------------------------------------------
   sd_totbal |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   treatment |  -3.255741   4.226085    -0.77   0.441    -11.54437    5.032889
     control |  -4.168199   4.787511    -0.87   0.384    -13.55796    5.221557
       _cons |   491.3003   3.390708   144.90   0.000     484.6501    497.9505
------------------------------------------------------------------------------

      Source |       SS           df       MS      Number of obs   =     1,777
-------------+----------------------------------   F(2, 1774)      =      0.28
       Model |  3412.62158         2  1706.31079   Prob > F        =    0.7567
    Residual |  10853846.7     1,774  6118.29015   R-squared       =    0.0003
-------------+----------------------------------   Adj R-squared   =   -0.0008
       Total |  10857259.4     1,776  6113.32171   Root MSE        =    78.219

------------------------------------------------------------------------------
 mean_totbal |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   treatment |  -.9482933   4.516166    -0.21   0.834    -9.805859    7.909273
     control |  -3.627729   5.116129    -0.71   0.478      -13.662    6.406545
       _cons |   476.9744   3.623448   131.64   0.000     469.8677     484.081
------------------------------------------------------------------------------

      Source |       SS           df       MS      Number of obs   =     1,777
-------------+----------------------------------   F(2, 1774)      =      0.15
       Model |  6154208.23         2  3077104.11   Prob > F        =    0.8576
    Residual |  3.5543e+10     1,774  20035747.1   R-squared       =    0.0002
-------------+----------------------------------   Adj R-squared   =   -0.0010
       Total |  3.5550e+10     1,776  20016649.5   Root MSE        =    4476.1

------------------------------------------------------------------------------
         pop |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   treatment |   21.29185   258.4388     0.08   0.934    -485.5846    528.1683
     control |   145.7407   292.7718     0.50   0.619    -428.4732    719.9546
       _cons |   5708.208   207.3528    27.53   0.000     5301.527     6114.89

Panel B


In [17]:
%%stata
tabstat yearsed female age impatient_mon01 hyper_mon_new2 if call==1, by(group) stats(mean sem) 
reg yearsed treatment control if call ==1
reg female treatment control if call ==1
reg age treatment control if call ==1
reg impatient_mon01 treatment control if call ==1
reg hyper_mon_new2 treatment control if call ==1


Summary statistics: mean, se(mean)
  by categories of: group 

 group |   yearsed    female       age  impa~n01  h~n_new2
-------+--------------------------------------------------
     C |  18.19403  .6162047  42.05117  .8081023  .2622601
       |  .1373271  .0224796  .5943071  .0393999  .0203327
-------+--------------------------------------------------
     M |  17.91845  .5472103  42.87124  .8903226  .2758621
       |   .144761  .0230834   .658222  .0409272  .0207714
-------+--------------------------------------------------
     T |  18.22209  .5985748  42.10808  .8690476  .2782402
       |  .1052178   .016903  .4578191  .0296984   .015462
-------+--------------------------------------------------
 Total |  18.13506   .589758  42.29319  .8585118  .2733935
       |   .072415  .0116717   .318447  .0205285   .010585
----------------------------------------------------------

      Source |       SS       df       MS              Number of obs =    1777
-------------+------------------------------           F(  2,  1774) =    1.60
       Model |  29.8721265     2  14.9360633           Prob > F      =  0.2014
    Residual |  16519.7137  1774  9.31212722           R-squared     =  0.0018
-------------+------------------------------           Adj R-squared =  0.0007
       Total |  16549.5858  1776  9.31846048           Root MSE      =  3.0516

------------------------------------------------------------------------------
     yearsed |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   treatment |   .3036353   .1761892     1.72   0.085     -.041925    .6491956
     control |   .2755749   .1995956     1.38   0.168    -.1158923    .6670421
       _cons |   17.91845   .1413616   126.76   0.000      17.6412    18.19571
------------------------------------------------------------------------------

      Source |       SS       df       MS              Number of obs =    1777
-------------+------------------------------           F(  2,  1774) =    2.56
       Model |  1.23708842     2  .618544211           Prob > F      =  0.0776
    Residual |  428.696508  1774  .241655303           R-squared     =  0.0029
-------------+------------------------------           Adj R-squared =  0.0018
       Total |  429.933596  1776  .242079727           Root MSE      =  .49158

------------------------------------------------------------------------------
      female |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   treatment |   .0513645   .0283827     1.81   0.071    -.0043024    .1070315
     control |   .0689944   .0321532     2.15   0.032     .0059322    .1320566
       _cons |   .5472103   .0227722    24.03   0.000     .5025471    .5918735
------------------------------------------------------------------------------

      Source |       SS       df       MS              Number of obs =    1777
-------------+------------------------------           F(  2,  1774) =    0.59
       Model |  212.035992     2  106.017996           Prob > F      =  0.5555
    Residual |  319828.212  1774  180.286478           R-squared     =  0.0007
-------------+------------------------------           Adj R-squared = -0.0005
       Total |  320040.248  1776  180.202842           Root MSE      =  13.427

------------------------------------------------------------------------------
         age |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   treatment |  -.7631686   .7752405    -0.98   0.325    -2.283649    .7573122
     control |  -.8200719   .8782295    -0.93   0.351    -2.542545    .9024014
       _cons |   42.87124   .6219975    68.93   0.000     41.65132    44.09117
------------------------------------------------------------------------------

      Source |       SS       df       MS              Number of obs =    1774
-------------+------------------------------           F(  2,  1771) =    1.17
       Model |  1.75557046     2  .877785228           Prob > F      =  0.3093
    Residual |   1323.7309  1771  .747448278           R-squared     =  0.0013
-------------+------------------------------           Adj R-squared =  0.0002
       Total |  1325.48647  1773  .747595302           Root MSE      =  .86455

------------------------------------------------------------------------------
impatien~n01 |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   treatment |   -.021275   .0499723    -0.43   0.670    -.1192859     .076736
     control |  -.0822202   .0565785    -1.45   0.146    -.1931878    .0287473
       _cons |   .8903226   .0400926    22.21   0.000     .8116888    .9689564
------------------------------------------------------------------------------

      Source |       SS       df       MS              Number of obs =    1774
-------------+------------------------------           F(  2,  1771) =    0.20
       Model |  .080716446     2  .040358223           Prob > F      =  0.8164
    Residual |  352.323455  1771  .198940404           R-squared     =  0.0002
-------------+------------------------------           Adj R-squared = -0.0009
       Total |  352.404171  1773  .198761518           Root MSE      =  .44603

------------------------------------------------------------------------------
hyper_mon~w2 |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   treatment |   .0023781   .0257935     0.09   0.927    -.0482107     .052967
     control |  -.0136019    .029205    -0.47   0.641    -.0708818    .0436779
       _cons |   .2758621   .0207063    13.32   0.000     .2352507    .3164734

Table III: Time preference questions

0 = patient, 1 = somewhat impatient, 2=most impatient


In [23]:
%%stata
tab impatient_mon01 impatient_mon67 if call==1


 Impatient |
 Now w/r/t |   Impatient Later w/r/t money
     money |         0          1          2 |     Total
-----------+---------------------------------+----------
         0 |       606        126         73 |       805 
         1 |       206        146         59 |       411 
         2 |       154         93        299 |       546 
-----------+---------------------------------+----------
     Total |       966        365        431 |     1,762 

Table V: Determinants of Takeup


In [24]:
%%stata
gen female_hyper_mon_new2 = female * hyper_mon_new2


(1379 missing values generated)

In [27]:
%%stata
xi: dprobit seedtakeup hyper_mon_new2 female impatient_200p250_01 impatient_250p200_01 impatient_200p250_67 impatient_250p200_67 fem_married $fullset fem_frac_veryown_inc_0_25 fem_frac_veryown_inc_25_50 fem_frac_veryown_inc_50_75 fem_frac_veryown_inc_75_100 frac_veryown_inc_0_25 frac_veryown_inc_25_50 frac_veryown_inc_50_75 frac_veryown_inc_75_100 active if group=="T" & call==1 & reached==1, robust


> _50 fem_frac_veryown_inc_50_75 fem_frac_veryown_inc_75_100 frac_veryown_inc_0_25 frac_veryown_inc_25_50 frac_veryown_inc_50_75 frac_veryown_inc_75_100 active if group=="T" & call==1 & reached==1, rob
> ust

Iteration 0:   log pseudolikelihood = -425.65043
Iteration 1:   log pseudolikelihood = -408.47662
Iteration 2:   log pseudolikelihood = -407.69152
Iteration 3:   log pseudolikelihood =   -407.141
Iteration 4:   log pseudolikelihood = -407.06088
Iteration 5:   log pseudolikelihood = -407.05886
Iteration 6:   log pseudolikelihood = -407.05886

Probit regression, reporting marginal effects           Number of obs =    715
                                                        Wald chi2(25) =  35.68
                                                        Prob > chi2   = 0.0767
Log pseudolikelihood = -407.05886                       Pseudo R2     = 0.0437

------------------------------------------------------------------------------
         |               Robust
seedta~p |      dF/dx   Std. Err.      z    P>|z|     x-bar  [    95% C.I.   ]
---------+--------------------------------------------------------------------
h~n_new2*|   .1248659   .0667069     1.94   0.053   .278322  -.005877  .255609
  female*|    .098883   .1372349     0.70   0.482        .6  -.170093  .367858
im~50_01*|  -.0300081   .0496469    -0.59   0.552   .241958  -.127314  .067298
i~200_01*|   .0764479   .0718961     1.07   0.285   .444755  -.064466  .217362
im~50_67*|   .0968518   .0646814     1.56   0.119   .206993  -.029922  .223625
i~200_67*|   .0150482    .063883     0.24   0.814   .545455   -.11016  .140257
fem_ma~d*|  -.1132497   .0905445    -1.22   0.222   .423776  -.290714  .064214
 married*|   .0489946   .0770504     0.62   0.536   .765035  -.102021  .200011
    edhi*|   .0834128   .0383418     2.14   0.032    .59021   .008264  .158561
   numhh |   .0001891   .0076993     0.02   0.980   5.58462  -.014901   .01528
unempl~d*|   .0404135   .1091283     0.38   0.703   .033566  -.173474  .254301
     age |  -.0019155   .0013788    -1.39   0.165   42.6531  -.004618  .000787
  GBloan*|  -.0141822   .0363595    -0.39   0.697   .502098  -.085445  .057081
GBloan~t*|  -.0317615    .072292    -0.43   0.670   .055944  -.173451  .109928
  hh_inc |   .0486918   .0307147     1.56   0.118   1.55682  -.011508  .108892
 hh_inc2 |   -.007847    .004361    -1.76   0.078   5.82637  -.016394    .0007
fem_f~25*|   .0153704   .1822196     0.09   0.932   .051748  -.341773  .372514
fem~5_50*|   .0480785     .16932     0.29   0.770    .14965  -.283783   .37994
fem_f~75*|   .1349672   .1818598     0.79   0.431   .114685  -.221472  .491406
fe~5_100*|   .0182208   .1546866     0.12   0.905   .174825  -.284959  .321401
frac_~25*|  -.0106804   .1540169    -0.07   0.945   .095105  -.312548  .291187
fra~5_50*|  -.0472313   .1407564    -0.33   0.745   .218182  -.323109  .228646
frac_~75*|  -.0340687   .1388783    -0.24   0.810   .202797  -.306265  .238128
fr~5_100*|   .0247978    .142089     0.18   0.861   .355245  -.253692  .303287
  active*|  -.0360562   .0341531    -1.04   0.296   .353846  -.102995  .030883
---------+--------------------------------------------------------------------
  obs. P |   .2825175
 pred. P |   .2647361  (at x-bar)
------------------------------------------------------------------------------
(*) dF/dx is for discrete change of dummy variable from 0 to 1
    z and P>|z| correspond to the test of the underlying coefficient being 0

More Exploratory


In [29]:
%%stata
tab seedtakeup if call==1


  SEED take |
         up |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |      1,575       88.63       88.63
          1 |        202       11.37      100.00
------------+-----------------------------------
      Total |      1,777      100.00

In [ ]:
%%stata
tab impatient_mon01
tab active impatient_mon01,row col

In [11]:
%%stata --graph
tab zerototbal if call==1
summ howlongopen
hist howlongopen


 zerototbal |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |      1,623       91.33       91.33
          1 |        154        8.67      100.00
------------+-----------------------------------
      Total |      1,777      100.00

    Variable |       Obs        Mean    Std. Dev.       Min        Max
-------------+--------------------------------------------------------
 howlongopen |       212    340.9764    48.90105         84        404
(bin=14, start=84, width=22.857143)
Out[11]:

In [12]:
%%stata
tab active if call==1,summ( nonseedbalchange)
summ nonseedbalchange,detail
tab active active_treat if call==1


     Active |     Summary of nonseedbalchange
    account |        Mean   Std. Dev.       Freq.
------------+------------------------------------
          0 |   105.73035   4714.6168        1145
          1 |   370.66207   2928.8638         632
------------+------------------------------------
      Total |    199.9548   4169.1327        1777

                      nonseedbalchange
-------------------------------------------------------------
      Percentiles      Smallest
 1%     -1506.14       -4541.77
 5%      -947.57       -2168.88
10%      -500.94        -2138.5       Obs                3152
25%     -255.515       -2114.25       Sum of Wgt.        3152

50%       -81.16                      Mean           133.8924
                        Largest       Std. Dev.      3287.174
75%            0       30921.14
90%       115.06       52460.47       Variance       1.08e+07
95%       870.87       100067.1       Skewness       24.70557
99%      5866.63         114001       Kurtosis       757.9191

    Active |     active_treat
   account |         0          1 |     Total
-----------+----------------------+----------
         0 |     1,145          0 |     1,145 
         1 |       338        294 |       632 
-----------+----------------------+----------
     Total |     1,483        294 |     1,777 

In [13]:
%%stata --graph
hist balchange_total if balchange <2500
count if balchange_total==0
tab savedathome


(bin=32, start=-2168.8799, width=145.59624)
  201

savedathome |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |      1,426       80.25       80.25
          1 |        351       19.75      100.00
------------+-----------------------------------
      Total |      1,777      100.00
Out[13]:

In [14]:
%%stata
tab active if call==1, summ(time_acct_tot)
tab active,summ (housedebt)


     Active |      Summary of time_acct_tot
    account |        Mean   Std. Dev.       Freq.
------------+------------------------------------
          0 |   19.213974   650.15954        1145
          1 |   15.822785   397.77864         632
------------+------------------------------------
      Total |   18.007878     573.153        1777

            |   Summary of total household debt
            |  (amount owed on dwelling, loans,
     Active |     agricultural assets and re
    account |        Mean   Std. Dev.       Freq.
------------+------------------------------------
          0 |   18896.648    75069.07        2125
          1 |   12595.804   48363.039        1028
------------+------------------------------------
      Total |   16842.329    67588.91        3153

In [32]:
%%stata
tab active seedtakeup if call==1
tab active seedtakeup if call==1,summ(balchange)


    Active |     SEED take up
   account |         0          1 |     Total
-----------+----------------------+----------
         0 |     1,009        136 |     1,145 
         1 |       566         66 |       632 
-----------+----------------------+----------
     Total |     1,575        202 |     1,777 


                Means, Standard Deviations and Frequencies
                  of Change in total savings held at bank

    Active |    SEED take up
   account |         0          1 |     Total
-----------+----------------------+----------
         0 | 94.116509  908.01279 | 190.78891
           | 4834.5791  7223.3173 | 5178.9086
           |      1009        136 |      1145
-----------+----------------------+----------
         1 | 379.61849  1313.6483 | 477.15959
           | 2968.6202  3238.7227 | 3008.8488
           |       566         66 |       632
-----------+----------------------+----------
     Total | 196.71595  1040.5472 | 292.63824
           | 4260.3395  6202.5993 | 4529.0155
           |      1575        202 |      1777

In [33]:
%%stata
tab group active,summ(balchange)


                Means, Standard Deviations and Frequencies
                  of Change in total savings held at bank

           |   Active account
     group |         0          1 |     Total
-----------+----------------------+----------
         C | -81.54299  301.87018 | 43.575771
           | 1244.8123  3496.8758 | 2248.5158
           |       545        264 |       809
-----------+----------------------+----------
         M |-18.270314  478.89511 | 150.86844
           | 1339.7333  2709.9326 |  1931.619
           |       512        264 |       776
-----------+----------------------+----------
         T | 225.59401  386.63116 | 276.94514
           | 5360.6638  1981.0306 | 4563.1934
           |      1068        500 |      1568
-----------+----------------------+----------
     Total | 88.065447  388.55799 | 186.03764
           | 3909.5162  2631.1481 | 3546.1206
           |      2125       1028 |      3153

Decile plots

This isn't quite the Decile plots on page 661 but close


In [34]:
%%stata
xtile decile = balchange if call==1, n(10)
tab decile group if call==1, summ(balchange)


                Means, Standard Deviations and Frequencies
                  of Change in total savings held at bank

        10 |
 quantiles |
        of |             group
balchange  |         C          M          T |     Total
-----------+---------------------------------+----------
         1 |-1120.4232 -994.66139 -986.56178 |-1035.6207
           | 449.23924  353.19363  346.42548 | 390.14235
           |        63         43         73 |       179
-----------+---------------------------------+----------
         2 | -336.8518 -340.81136  -369.2864 |-352.69417
           | 83.117129  82.913023  90.475814 | 87.534972
           |        72         59        111 |       242
-----------+---------------------------------+----------
         3 |-240.94806    -232.93 -225.79961 |-231.92044
           | 29.195496  31.587574  28.978035 | 30.185034
           |        31         32         51 |       114
-----------+---------------------------------+----------
         4 |-146.92694 -145.79947 -152.07352 |-148.62831
           | 20.957726  21.480777  20.236842 | 20.920627
           |        49         57         71 |       177
-----------+---------------------------------+----------
         5 |-95.722433 -94.705294 -94.852584 |-94.991977
           | 15.786802  16.028026  16.594359 | 16.179985
           |        37         51         89 |       177
-----------+---------------------------------+----------
         6 |-17.134444 -16.453053 -14.004251 |-15.546676
           | 22.666709  21.527046  19.618369 | 21.028635
           |       108         95        167 |       370
-----------+---------------------------------+----------
         8 | 10.502083  9.0546808  9.6755881 | 9.7399386
           | 5.5756922  5.8252881  5.5605988 | 5.6349909
           |        48         47         68 |       163
-----------+---------------------------------+----------
         9 |   127.853  93.227714   102.9846 | 105.25742
           |  86.67977  69.590768    70.4752 | 73.637829
           |        30         35        113 |       178
-----------+---------------------------------+----------
        10 | 4552.6512  3605.4385  5405.5254 | 4778.1628
           | 9386.6839  4902.8496  17029.412 | 13541.666
           |        31         47         99 |       177
-----------+---------------------------------+----------
     Total | 65.183004  189.07403  476.64944 | 292.63824
           |  2690.654   1944.996  6093.2376 | 4529.0155
           |       469        466        842 |      1777

In [ ]: