-
Notifications
You must be signed in to change notification settings - Fork 10
/
generate_baseline.py
207 lines (183 loc) · 8.13 KB
/
generate_baseline.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
"""
The code in this file calculates and saves the various adjustment factors
and DataFrames necessary for the calculations in Business-Taxation.
We need to produce the following objects:
adjfactors.csv
pass-through shares
"""
import copy
import numpy as np
import pandas as pd
import scipy.optimize
from biztax.data import Data
from biztax.asset import Asset
from biztax.debt import Debt
from biztax.policy import Policy
# Specify single Data object (for convenience)
data1 = Data()
"""
Section 1. Calculation of the adjustment parameters
"""
def calcAMTparams2():
"""
Calculates the adjustment factors for the AMT and PYMTC
"""
# Grab historical data
hist_data = copy.deepcopy(data1.historical_combined)
taxinc = np.array(hist_data['taxinc'])
amt = np.array(hist_data['amt'])
stock13 = 26.0
A13 = 4.196871
tau_a = 0.2
tau_c = 0.347
pi0 = 0.865
pi1 = 0.494
gamma = (A13 * (2. - pi0 - pi1) /
(stock13 * (1. - pi1) + A13 * (1. - pi0 - pi1)))
A_over_TI = sum([amt[i] / taxinc[i] for i in range(16)]) / 16.
# Calculate solution to AMT parameter
def amterr(lam):
# Squared difference between actual AMT/TaxInc ratio vs. predicted
ATI_pred = tau_a / lam * np.exp(-lam * (tau_c / tau_a - 1))
err = (ATI_pred - A_over_TI)**2
return err
lamf = scipy.optimize.minimize_scalar(amterr,
bounds=(0.001, 100),
method='bounded').x
theta = np.exp(-lamf * (tau_c / tau_a - 1))
return (lamf, theta, gamma, pi1, pi0)
def calcWAvgTaxRate(year):
"""
Calculates the weighted average statutory corporate tax rate
in all OECD countries in a given year.
"""
assert year in range(1995, 2028)
year = min(year, 2016)
gdp_list = np.asarray(data1.ftc_gdp_data[str(year)])
taxrate_list = np.asarray(data1.ftc_taxrates_data[str(year)])
# remove observations with missing data
taxrate_list2 = np.where(np.isnan(taxrate_list), 0, taxrate_list)
gdp_list2 = np.where(np.isnan(taxrate_list), 0, gdp_list)
avgrate = sum(taxrate_list2 * gdp_list2) / sum(gdp_list2)
return avgrate
def calcFTCAdjustment():
"""
Calculates the adjustment factor for the FTC.
"""
ftc_actual = np.asarray(data1.ftc_other_data['F'][:19])
profits = np.asarray(data1.ftc_other_data['C_total'][:19])
profits_d = np.asarray(data1.ftc_other_data['C_domestic'][:19])
profits_f = profits - profits_d
tax_f = []
for i in range(1995, 2014):
tax_f.append(calcWAvgTaxRate(i))
ftc_gross = profits_f * tax_f / 100.
adjfactor = sum(ftc_actual / ftc_gross) / 19.
return adjfactor
def calcIDAdjustment(corp, eta=0.4):
"""
Calculates the adjustment factors for the corporate and noncorporate
debt and interest.
eta: retirement rate of existing debt
"""
policy = Policy()
policy_params_df = policy.parameters_dataframe()
# Create Asset object
asset = Asset(policy_params_df, corp)
asset.calc_all()
# Get asset forecast
forecast = asset.get_forecast()
# Create Debt object
debt = Debt(policy_params_df, forecast, corp=corp)
debt.calc_all()
# Get unscaled net interest deduction
NID_gross = debt.NID[38:54]
# Get net interest deduction from historical IRS data
if corp:
NID_irs = np.array(data1.debt_data_corp['NID_IRS'])[38:54]
else:
NID_irs = np.array(data1.debt_data_noncorp['ID_Scorp'][38:54] +
data1.debt_data_noncorp['ID_sp'][38:54] +
data1.debt_data_noncorp['ID_partner'][38:54])
NID_scale = sum(NID_irs / NID_gross) / 16.0 # 16 = 54 - 38
return NID_scale
# Calculate the adjustment and dynamic parameters for AMT & PYMTC
all_amt_params = calcAMTparams2()
# Calculate the FTC adjustment parameters
adjfactor_ftc_corp = calcFTCAdjustment()
"""
Section 2. Calculation of pass-through shares
Note: All shares are estimated for 2013.
"""
# Total depreciation
totaldep = (data1.partner_data['dep_total'][19] +
data1.Scorp_data['dep_total'][18] +
data1.sp_data['dep_total'][16])
# Depreciation shares for S corporations (by income status)
depshare_scorp_posinc = data1.Scorp_data['dep_posinc'][18] / totaldep
depshare_scorp_neginc = (data1.Scorp_data['dep_total'][18] / totaldep
- depshare_scorp_posinc)
# Depreciation shares for sole proprietorships (by income status)
depshare_sp_posinc = data1.sp_data['dep_posinc'][16] / totaldep
depshare_sp_neginc = (data1.sp_data['dep_total'][16] / totaldep
- depshare_sp_posinc)
# Depreciation shares for partnerships (by income status)
depshare_partner_posinc = data1.partner_data['dep_posinc'][19] / totaldep
depshare_partner_neginc = (data1.partner_data['dep_total'][19] / totaldep
- depshare_partner_posinc)
# Total net interest deduction, excluding finance sector and holding companies
totalint_exfin = (data1.partner_data['intpaid_total'][19]
+ data1.Scorp_data['intpaid_total'][18]
+ data1.sp_data['mortintpaid'][16]
+ data1.sp_data['otherintpaid'][16]
- data1.partner_data['intpaid_fin_total'][19]
- data1.Scorp_data['intpaid_fin'][18]
- data1.sp_data['mortintpaid_fin'][16]
- data1.sp_data['otherintpaid_fin'][16])
# Net interest share for S corporations (by income status)
intshare_scorp_posinc = ((data1.Scorp_data['intpaid_posinc'][18]
- data1.Scorp_data['intpaid_fin_posinc'][18])
/ totalint_exfin)
intshare_scorp_neginc = ((data1.Scorp_data['intpaid_total'][18]
- data1.Scorp_data['intpaid_fin'][18])
/ totalint_exfin - intshare_scorp_posinc)
# Net interest share for sole proprietorships (by income status)
intshare_sp_posinc = ((data1.sp_data['mortintpaid_posinc'][16]
+ data1.sp_data['otherintpaid_posinc'][16]
- data1.sp_data['mortintpaid_fin_posinc'][16]
- data1.sp_data['otherintpaid_fin_posinc'][16])
/ totalint_exfin)
intshare_sp_neginc = ((data1.sp_data['mortintpaid'][16]
+ data1.sp_data['otherintpaid'][16]
- data1.sp_data['mortintpaid_fin'][16]
- data1.sp_data['otherintpaid_fin'][16]) /
totalint_exfin - intshare_sp_posinc)
intshare_partner_posinc = ((data1.partner_data['intpaid_posinc'][19]
- data1.partner_data['intpaid_fin_posinc'][19])
/ totalint_exfin)
intshare_partner_neginc = ((data1.partner_data['intpaid_total'][19]
- data1.partner_data['intpaid_fin_total'][19])
/ totalint_exfin - intshare_partner_posinc)
# Save the adjustment factors and pass-through shares
adj_factors = {'param_amt': all_amt_params[0],
'amt_frac': all_amt_params[1],
'userate_pymtc': all_amt_params[2],
'trans_amt1': all_amt_params[3],
'trans_amt0': all_amt_params[4],
'ftc': adjfactor_ftc_corp}
passthru_factors = {'dep_scorp_pos': depshare_scorp_posinc,
'dep_scorp_neg': depshare_scorp_neginc,
'dep_sp_pos': depshare_sp_posinc,
'dep_sp_neg': depshare_sp_neginc,
'dep_part_pos': depshare_partner_posinc,
'dep_part_neg': depshare_partner_neginc,
'int_scorp_pos': intshare_scorp_posinc,
'int_scorp_neg': intshare_scorp_neginc,
'int_sp_pos': intshare_sp_posinc,
'int_sp_neg': intshare_sp_neginc,
'int_part_pos': intshare_partner_posinc,
'int_part_neg': intshare_partner_neginc}
df_adjf = pd.DataFrame({k: [adj_factors[k]] for k in adj_factors})
df_adjf.to_csv('biztax/adjfactors.csv', index=False)
df_pts = pd.DataFrame({k: [passthru_factors[k]] for k in passthru_factors})
df_pts.to_csv('biztax/passthru_shares.csv', index=False)