MTRs on capital income

[jdebacker]

When I compute the weighted average marginal tax rate on capital income in get_micro_data.py using the pitSmallData.csv, I'm getting a weighted average marginal tax rate on capital income that exceeds 100% for most tax payers.

Looking into the underlying data, no filer has an MTR for any capital income source above 36%:

In [25]: df[['STCG 1 MTR', 'STCG 2 MTR', 'STCG AMT']].describe()
Out[25]:
         STCG 1 MTR    STCG 2 MTR      STCG AMT
count  10000.000000  10000.000000  10000.000000
mean       0.143129      0.286303      0.191846
std        0.044887      0.089743      0.120481
min        0.000000      0.000000      0.000000
25%        0.156000      0.312000      0.052000
50%        0.156000      0.312000      0.208000
75%        0.156000      0.312000      0.312000
max        0.179400      0.358800      0.358800

In [26]: df.keys()
Out[26]:
Index(['STCG 1 MTR', 'STCG 2 MTR', 'STCG AMT', 'LT CG 1', 'LT CG 2',
       'Not Horse'],
      dtype='object')

In [27]: df[['LT CG 1', 'LT CG 2', 'Not Horse']].describe()
Out[27]:
            LT CG 1       LT CG 2  Not Horse
count  10000.000000  10000.000000    10000.0
mean       0.095404      0.190853        0.0
std        0.029944      0.059836        0.0
min        0.000000      0.000000        0.0
25%        0.104000      0.208000        0.0
50%        0.104000      0.208000        0.0
75%        0.104000      0.208000        0.0
max        0.119600      0.239200        0.0

There must be an issue computing this weighted average. It should be fixed as it's preventing any subsequent analysis after tax functions are estimated.

[rickecon]

This is an Indian taxcalc issue that needs to be resolved by an update to the taxcalc repository. @sebastiansajie.

[jdebacker]

@rickecon This is not a taxcalc issue. It has to do with something in the weighted average computed in ogindia.get_micro_data.py. I'm looking into it. But from the results posted above, the rates direction from taxcalc look reasonable.

[jdebacker]

This issue was addressed with PR #20