def test_quantity_response():
"""
Test quantity_response function.
"""
quantity = np.array([1.0] * 10)
res = quantity_response(quantity)
assert np.allclose(res, np.zeros(quantity.shape))
one = np.ones(quantity.shape)
res = quantity_response(quantity,
price_elasticity=-0.2,
aftertax_price1=one,
aftertax_price2=one,
income_elasticity=0.1,
aftertax_income1=one,
aftertax_income2=(one + one))
assert not np.allclose(res, np.zeros(quantity.shape))
def test_labor_response():
"""
Test that labor_response produces the same result as quantity_response
where mtr* = 1 - aftertax_price*, using default earnings/quantity=1.
"""
res_lr = labor_response(substitution_eti=1, mtr1=0.4, mtr2=0.5)
res_qr = quantity_response(price_elasticity=1,
aftertax_price1=0.6,
aftertax_price2=0.5)
assert np.allclose(res_lr, res_qr)
price_elasticity = [-0.1, -0.4]
income_elasticity = [0.1, 0.1]
print('\nResponse in Charitable Giving by Earnings Group')
results = '{:18s}\t{:8.3f}\t{:8.3f}\t{:8.2f}'
colhead = '{:18s}\t{:>8s}\t{:>8s}\t{:>8s}'
print(colhead.format('Earnings Group', 'Num(#M)', 'Resp($B)', 'Resp(%)'))
tot_funits = 0.
tot_response = 0.
tot_baseline = 0.
idx = 0
for grp_interval, grp in gbydf:
funits = grp['s006'].sum() * 1e-6
tot_funits += funits
response = behresp.quantity_response(grp['e19800'],
price_elasticity[idx],
grp['price1'],
grp['price2'],
income_elasticity[idx],
grp['atinc1'],
grp['atinc2'])
grp_response = (response * grp['s006']).sum() * 1e-9
tot_response += grp_response
grp_baseline = (grp['e19800'] * grp['s006']).sum() * 1e-9
tot_baseline += grp_baseline
pct_response = 100. * grp_response / grp_baseline
glabel = '[{:.8g}, {:.8g})'.format(grp_interval.left, grp_interval.right)
print(results.format(glabel, funits, grp_response, pct_response))
idx += 1
pct_response = 100. * tot_response / tot_baseline
print(results.format('ALL', tot_funits, tot_response, pct_response))
vdf['price1'] = 1.0 + mtr1
vdf['price2'] = 1.0 + mtr2
vdf['atinc1'] = calc1.array('aftertax_income')
vdf['atinc2'] = calc2.array('aftertax_income')
# group filing units into earnings groups with different response elasticities
# (note earnings groups are just an example based on no empirical results)
EARNINGS_BINS = [-9e99, 50e3, 9e99] # Two groups: below and above $50,000.
vdf['table_row'] = pd.cut(vdf.e00200, EARNINGS_BINS, right=False).astype(str)
vdf['price_elasticity'] = np.where(vdf.e00200 < EARNINGS_BINS[1], -0.1, -0.4)
vdf['income_elasticity'] = 0.1
# Calculate response based on features of each filing unit.
vdf['response'] = behresp.quantity_response(vdf.e19800, vdf.price_elasticity,
vdf.price1, vdf.price2,
vdf.income_elasticity, vdf.atinc1,
vdf.atinc2)
# Add weighted totals.
# Can also use microdf as mdf.add_weighted_totals(vdf, ['response', 'e19800'])
vdf['e19800_b'] = vdf.s006 * vdf.e19800 / 1e9
vdf['response_b'] = vdf.s006 * vdf.response / 1e9
vdf['funits_m'] = vdf.s006 / 1e6
SUM_VARS = ['funits_m', 'e19800_b', 'response_b']
# Sum weighted total columns for each income group.
grouped = vdf.groupby('table_row')[SUM_VARS].sum()
# Add a total row and make the index a column for printing.
grouped.loc['TOTAL'] = grouped.sum()
grouped.reset_index(inplace=True)