Fix EITC returns are incorrectly calibrated #97

changelog_entry.yaml

@@ -1,5 +1,4 @@
 - bump: minor
  changes:
- added:
- - EITC calibration by child counts.
- - 10% dropout during weight calibration.
+ fixed:
+ - EITC calibration.

policyengine_us_data/datasets/cps/cps.py

@@ -54,6 +54,8 @@ def generate(self):
  raw_data.close()
  self.save_dataset(cps)
 
+ add_takeup(self)
+
 
 def add_rent(self, cps: h5py.File, person: DataFrame, household: DataFrame):
  cps["tenure_type"] = household.H_TENURE.map(
@@ -114,6 +116,30 @@ def add_rent(self, cps: h5py.File, person: DataFrame, household: DataFrame):
  cps["real_estate_taxes"][mask] = imputed_values["real_estate_taxes"]
 
 
+def add_takeup(self):
+ data = self.load_dataset()
+
+ from policyengine_us import system, Microsimulation
+
+ baseline = Microsimulation(dataset=self)
+ parameters = baseline.tax_benefit_system.parameters(self.time_period)
+ generator = np.random.default_rng(seed=100)
+
+ snap_takeup_rate = parameters.gov.usda.snap.takeup_rate
+ data["takes_up_snap_if_eligible"] = (
+ generator.random(len(data["spm_unit_id"])) < snap_takeup_rate
+ )
+
+ eitc_takeup_rates = parameters.gov.irs.credits.eitc.takeup
+ eitc_child_count = baseline.calculate("eitc_child_count").values
+ eitc_takeup_rate = eitc_takeup_rates.calc(eitc_child_count)
+ data["takes_up_eitc"] = (
+ generator.random(len(data["tax_unit_id"])) < eitc_takeup_rate
+ )
+
+ self.save_dataset(data)
+
+
 def uprate_cps_data(data, from_period, to_period):
  uprating = create_policyengine_uprating_factors_table()
  for variable in uprating.index.unique():

policyengine_us_data/utils/loss.py

@@ -208,8 +208,16 @@ def build_loss_matrix(dataset: type, time_period):
  )
  eitc_eligible_children = sim.calculate("eitc_child_count").values
  eitc = sim.calculate("eitc").values
+ if row["count_children"] < 2:
+ meets_child_criteria = (
+ eitc_eligible_children == row["count_children"]
+ )
+ else:
+ meets_child_criteria = (
+ eitc_eligible_children >= row["count_children"]
+ )
  loss_matrix[returns_label] = sim.map_result(
- (eitc > 0) * (eitc_eligible_children == row["count_children"]),
+ (eitc > 0) * meets_child_criteria,
  "tax_unit",
  "household",
  )
@@ -219,7 +227,7 @@ def build_loss_matrix(dataset: type, time_period):
  f"irs/eitc/spending/count_children_{row['count_children']}"
  )
  loss_matrix[spending_label] = sim.map_result(
- eitc * (eitc_eligible_children == row["count_children"]),
+ eitc * meets_child_criteria,
  "tax_unit",
  "household",
  )