Merge pull request #23 from masterismail/DistributionalImpact

[1] - added Distributional impact by income

policyengine/economic_impact/distributional_impact/by_income_decile/average/average.py

@@ -0,0 +1,37 @@
+from policyengine.economic_impact.base_metric_calculator import BaseMetricCalculator
+from policyengine_uk import Microsimulation
+from microdf import MicroDataFrame, MicroSeries
+
+
+class Average(BaseMetricCalculator):
+    def __init__(self, baseline: Microsimulation, reformed: Microsimulation, default_period: int = 2024) -> None:
+        super().__init__(baseline, reformed, default_period)
+        self.baseline = baseline
+        self.reformed = reformed
+
+    def calculate(self):
+
+        baseline_income = MicroSeries(self.baseline.calculate("household_net_income"), weights = self.baseline.calculate("household_weight"))
+        reform_income = MicroSeries(self.reformed.calculate("household_net_income") , weights = baseline_income.weights)
+
+        decile = self.baseline.calculate("household_income_decile")
+        income_change = reform_income - baseline_income
+
+
+        avg_income_change_by_decile = (
+                income_change.groupby(decile).sum()
+                / baseline_income.groupby(decile).count()
+            )
+
+
+        avg_decile_dict = avg_income_change_by_decile.to_dict()
+        result = dict(
+
+        average={int(k): v for k, v in avg_decile_dict.items() if k > 0},
+         )
+
+
+        return {
+            "average": result["average"],
+
+        }

policyengine/economic_impact/distributional_impact/by_income_decile/relative/relative.py

@@ -0,0 +1,38 @@
+from policyengine.economic_impact.base_metric_calculator import BaseMetricCalculator
+from policyengine_uk import Microsimulation
+from microdf import MicroDataFrame, MicroSeries
+
+
+class Relative(BaseMetricCalculator):
+    def __init__(self, baseline: Microsimulation, reformed: Microsimulation, default_period: int = 2024) -> None:
+        super().__init__(baseline, reformed, default_period)
+        self.baseline = baseline
+        self.reformed = reformed
+
+    def calculate(self):
+
+        baseline_income = MicroSeries(self.baseline.calculate("household_net_income"), weights = self.baseline.calculate("household_weight"))
+        reform_income = MicroSeries(self.reformed.calculate("household_net_income") , weights = baseline_income.weights)
+
+        decile = self.baseline.calculate("household_income_decile")
+        income_change = reform_income - baseline_income
+
+
+        rel_income_change_by_decile = (
+        income_change.groupby(decile).sum()
+        / baseline_income.groupby(decile).sum()
+        )
+
+
+        rel_decile_dict = rel_income_change_by_decile.to_dict()
+
+        result = dict(
+        relative={int(k): v for k, v in rel_decile_dict.items() if k > 0}
+
+         )
+
+
+        return {
+            "relative": result["relative"],
+
+        }

policyengine/economic_impact/economic_impact.py

@@ -36,6 +36,9 @@
     PensionCredit
 )
 
+from .distributional_impact.by_income_decile.average.average import Average
+from .distributional_impact.by_income_decile.relative.relative import Relative
+
 from .budgetary_impact.overall.overall import (
     BudgetaryImpact, 
     BenefitSpendingImpact, 
@@ -108,9 +111,10 @@ def __init__(self, reform: dict, country: str) -> None:
             "poverty/deep/male": DeepMalePoverty(self.baseline, self.reformed),
             "poverty/deep/female": DeepFemalePoverty(self.baseline, self.reformed),
             "poverty/deep/gender/all": DeepGenderAllPoverty(self.baseline, self.reformed),
+            "distributional/by_income/average": Average(self.baseline, self.reformed),
+            "distributional/by_income/relative": Relative(self.baseline, self.reformed),
             "winners_and_losers/by_income_decile": ByIncomeDecile(self.baseline, self.reformed),
             "winners_and_losers/by_wealth_decile": ByWealthDecile(self.baseline, self.reformed),
-
         }
 
     def _get_simulation_class(self) -> type:

policyengine/tests/economic_impact/distributional_impact/by_income/average/average.yaml

@@ -0,0 +1,19 @@
+# Regular poverty by age
+- test_by_income_average:
+    reform:
+      gov.hmrc.income_tax.rates.uk[0].rate:
+        "2024-01-01.2100-12-31": 0.55
+    country: uk
+    expected:
+      average:
+        1: -288.4419422028393
+        2: -936.7849231542535
+        3: -1655.8066942572789
+        4: -3040.3849413280304
+        5: -5082.727697456715
+        6: -8483.137124053015
+        7: -11510.447500088267
+        8: -14567.684605663524
+        9: -17960.33204429504
+        10: -22070.767989875396
+        11: -13195.0

policyengine/tests/economic_impact/distributional_impact/by_income/average/test_average.py

@@ -0,0 +1,38 @@
+import pytest
+import yaml
+import os
+from policyengine import EconomicImpact
+
+def assert_dict_approx_equal(actual, expected, tolerance=1e-4):
+    assert set(actual.keys()) == set(expected.keys()), f"Keys don't match. Actual: {actual.keys()}, Expected: {expected.keys()}"
+    for key in expected:
+        if isinstance(expected[key], dict):
+            assert_dict_approx_equal(actual[key], expected[key], tolerance)
+        else:
+            assert abs(actual[key] - expected[key]) < tolerance, f"Key {key}: expected {expected[key]}, got {actual[key]}"
+
+yaml_file_path = "policyengine/tests/economic_impact/distributional_impact/by_income/average/average.yaml"
+
+# Check if the file exists
+if not os.path.exists(yaml_file_path):
+    raise FileNotFoundError(f"The YAML file does not exist at: {yaml_file_path}")
+
+with open(yaml_file_path, 'r') as file:
+    test_cases = yaml.safe_load(file)
+
+@pytest.mark.parametrize("test_case", test_cases)
+def test_economic_impact(test_case):
+    test_name = list(test_case.keys())[0]
+    test_data = test_case[test_name]
+
+    economic_impact = EconomicImpact(test_data['reform'], test_data['country'])
+
+    if 'average' in test_name:
+        result = economic_impact.calculate("distributional/by_income/average")
+    else:
+        pytest.fail(f"Unknown test case: {test_name}")
+
+    assert_dict_approx_equal(result, test_data['expected'])
+
+if __name__ == "__main__":
+    pytest.main([__file__])

policyengine/tests/economic_impact/distributional_impact/by_income/relative/relative.yaml

@@ -0,0 +1,19 @@
+# Regular poverty by age
+- test_by_income_average:
+    reform:
+      gov.hmrc.income_tax.rates.uk[0].rate:
+        "2024-01-01.2100-12-31": 0.55
+    country: uk
+    expected:
+      relative:
+          1: -0.026265868982450032
+          2: -0.05321454070727573
+          3: -0.07293061898874185
+          4: -0.1046825029478255
+          5: -0.1407466605531786
+          6: -0.19129333726780962
+          7: -0.2185085314786923
+          8: -0.22920465056330844
+          9: -0.21991139800089637
+          10: -0.12790076021888797
+          11: -0.007019584080998393

policyengine/tests/economic_impact/distributional_impact/by_income/relative/test_relative.py

@@ -0,0 +1,38 @@
+import pytest
+import yaml
+import os
+from policyengine import EconomicImpact
+
+def assert_dict_approx_equal(actual, expected, tolerance=1e-4):
+    assert set(actual.keys()) == set(expected.keys()), f"Keys don't match. Actual: {actual.keys()}, Expected: {expected.keys()}"
+    for key in expected:
+        if isinstance(expected[key], dict):
+            assert_dict_approx_equal(actual[key], expected[key], tolerance)
+        else:
+            assert abs(actual[key] - expected[key]) < tolerance, f"Key {key}: expected {expected[key]}, got {actual[key]}"
+
+yaml_file_path = "policyengine/tests/economic_impact/distributional_impact/by_income/relative/relative.yaml"
+
+# Check if the file exists
+if not os.path.exists(yaml_file_path):
+    raise FileNotFoundError(f"The YAML file does not exist at: {yaml_file_path}")
+
+with open(yaml_file_path, 'r') as file:
+    test_cases = yaml.safe_load(file)
+
+@pytest.mark.parametrize("test_case", test_cases)
+def test_economic_impact(test_case):
+    test_name = list(test_case.keys())[0]
+    test_data = test_case[test_name]
+
+    economic_impact = EconomicImpact(test_data['reform'], test_data['country'])
+
+    if 'average' in test_name:
+        result = economic_impact.calculate("distributional/by_income/relative")
+    else:
+        pytest.fail(f"Unknown test case: {test_name}")
+
+    assert_dict_approx_equal(result, test_data['expected'])
+
+if __name__ == "__main__":
+    pytest.main([__file__])