policyengine-uk 2.45.4__py3-none-any.whl → 2.47.3__py3-none-any.whl

policyengine_uk/dynamics/progression.py

@@ -0,0 +1,376 @@
+"""Labour supply progression (intensive margin) dynamics module.
+
+This module handles the intensive margin of labour supply - how people adjust their
+working hours in response to policy changes. It implements the elasticity-based
+methodology from the OBR's labour supply framework.
+
+Reference: https://obr.uk/docs/dlm_uploads/NICS-Cut-Impact-on-Labour-Supply-Note.pdf
+"""
+
+import numpy as np
+import pandas as pd
+from policyengine_uk import Simulation
+
+
+def calculate_derivative(
+    sim: Simulation,
+    target_variable: str = "household_net_income",
+    input_variable: str = "employment_income",
+    year: int = 2025,
+    count_adults: int = 2,
+    delta: float = 1_000,
+) -> np.ndarray:
+    """Calculate the marginal rate of change of target variable with respect to input variable.
+
+    This function computes numerical derivatives by applying small changes to the input
+    variable and measuring the resulting change in the target variable. This is used
+    to estimate marginal tax rates and benefit withdrawal rates.
+
+    Args:
+        sim: PolicyEngine simulation object
+        target_variable: Variable to measure changes in (typically household_net_income)
+        input_variable: Variable to change (typically employment_income)
+        year: Year for calculation
+        count_adults: Number of adults to apply changes to
+        delta: Size of change to apply for derivative calculation (£)
+
+    Returns:
+        Array of marginal rates clipped between 0 and 1
+    """
+    # Get baseline values for input variable and identify adults
+    input_variable_values = sim.calculate(input_variable, year)
+    adult_index = sim.calculate("adult_index")
+    entity_key = sim.tax_benefit_system.variables[input_variable].entity.key
+
+    # Calculate baseline target values
+    original_target_values = sim.calculate(
+        target_variable, year, map_to=entity_key
+    )
+    new_target_values = original_target_values.copy()
+
+    # Apply delta change to each adult sequentially to calculate marginal effects
+    for i in range(count_adults):
+        gets_pay_rise = adult_index == i + 1
+        new_input_variable_values = input_variable_values.copy()
+        new_input_variable_values[gets_pay_rise] += delta
+        sim.reset_calculations()
+        sim.set_input(input_variable, year, new_input_variable_values)
+        new_target_values[gets_pay_rise] = sim.calculate(
+            target_variable, year, map_to=entity_key
+        )[gets_pay_rise]
+
+    # Calculate marginal rate as change in target per unit change in input
+    rel_marginal_wages = (new_target_values - original_target_values) / delta
+
+    # Set non-adult observations to NaN
+    rel_marginal_wages[
+        ~pd.Series(adult_index).isin(range(1, count_adults + 1))
+    ] = np.nan
+
+    # Clip to ensure rates are between 0 and 1 (0% to 100% retention)
+    return rel_marginal_wages.clip(0, 1)
+
+
+def calculate_relative_income_change(
+    sim: Simulation,
+    target_variable: str = "household_net_income",
+    year: int = 2025,
+) -> pd.DataFrame:
+    """Calculate relative change in income between baseline and scenario.
+
+    This function compares the target variable values between the baseline
+    simulation and the reform scenario to measure the income effect of the policy.
+
+    Args:
+        sim: PolicyEngine simulation object (should have baseline attribute)
+        target_variable: Variable to measure changes in
+        year: Year for calculation
+
+    Returns:
+        DataFrame with baseline, scenario, relative change, and absolute change columns
+    """
+    # Get income values from baseline and reform scenarios
+    original_target_values = sim.baseline.calculate(
+        target_variable, year, map_to="person"
+    )
+    reformed_target_values = sim.calculate(
+        target_variable, year, map_to="person"
+    )
+
+    # Calculate relative change, handling division by zero
+    rel_change = (
+        reformed_target_values - original_target_values
+    ) / original_target_values
+    rel_change[original_target_values == 0] = np.nan
+
+    # Clip extreme values and fill NaN with 0
+    rel_changes = rel_change.clip(-1, 1).fillna(0)
+
+    return pd.DataFrame(
+        {
+            "baseline": original_target_values,
+            "scenario": reformed_target_values,
+            "rel_change": rel_changes,
+            "abs_change": reformed_target_values - original_target_values,
+        }
+    )
+
+
+def calculate_derivative_change(
+    sim: Simulation,
+    target_variable: str = "household_net_income",
+    input_variable: str = "employment_income",
+    year: int = 2025,
+    count_adults: int = 1,
+    delta: float = 1_000,
+) -> pd.DataFrame:
+    """Calculate change in marginal rates between baseline and scenario.
+
+    This function computes how marginal tax rates or benefit withdrawal rates
+    change as a result of the policy reform, which drives substitution effects
+    in labour supply responses.
+
+    Args:
+        sim: PolicyEngine simulation object (should have baseline attribute)
+        target_variable: Variable to measure marginal rates for
+        input_variable: Variable to change for derivative calculation
+        year: Year for calculation
+        count_adults: Number of adults to calculate derivatives for
+        delta: Size of change for derivative calculation (£)
+
+    Returns:
+        DataFrame with baseline, scenario, relative change, and absolute change in derivatives
+    """
+    # Calculate marginal rates under baseline and reform scenarios
+    original_deriv = calculate_derivative(
+        sim=sim.baseline,
+        target_variable=target_variable,
+        input_variable=input_variable,
+        year=year,
+        count_adults=count_adults,
+        delta=delta,
+    )
+
+    reformed_deriv = calculate_derivative(
+        sim=sim,
+        target_variable=target_variable,
+        input_variable=input_variable,
+        year=year,
+        count_adults=count_adults,
+        delta=delta,
+    )
+
+    # Calculate relative and absolute changes in marginal rates
+    rel_change = reformed_deriv / original_deriv - 1
+    abs_change = reformed_deriv - original_deriv
+
+    # Clip extreme relative changes to avoid misleading results from small baseline derivatives
+    rel_change = rel_change.clip(-1, 1)
+
+    rel_change[rel_change == np.inf] = 0
+
+    return pd.DataFrame(
+        {
+            "baseline": original_deriv,
+            "scenario": reformed_deriv,
+            "rel_change": rel_change,
+            "abs_change": abs_change,
+        }
+    ).fillna(0)
+
+
+def calculate_labour_substitution_elasticities(
+    sim: Simulation,
+) -> np.ndarray:
+    """Calculate labour supply substitution elasticities by demographic group.
+
+    Uses OBR elasticity estimates to assign substitution elasticities based on
+    gender, marital status, and presence/age of children. These elasticities
+    measure how labour supply responds to changes in marginal tax rates.
+
+    Reference: https://obr.uk/docs/dlm_uploads/NICS-Cut-Impact-on-Labour-Supply-Note.pdf
+
+    Args:
+        sim: PolicyEngine simulation object
+
+    Returns:
+        Array of substitution elasticities for each person
+    """
+    # Get demographic characteristics for elasticity assignment
+    gender = sim.calculate("gender")
+    is_married = sim.calculate("is_married", map_to="person")
+    has_children = sim.calculate("benunit_count_children", map_to="person") > 0
+    youngest_child_age = sim.calculate("youngest_child_age", map_to="person")
+
+    # Initialize elasticity array
+    elasticities = np.zeros(gender.shape, dtype=float)
+
+    # Married or cohabiting women - higher elasticities, especially with young children
+    married_women = (gender == "FEMALE") & is_married
+    elasticities[married_women & ~has_children] = 0.14  # No children
+
+    # Elasticities vary significantly by youngest child's age
+    elasticities[married_women & has_children & (youngest_child_age <= 2)] = (
+        0.301  # 0-2 years
+    )
+    elasticities[
+        married_women
+        & has_children
+        & (youngest_child_age >= 3)
+        & (youngest_child_age <= 4)
+    ] = 0.439  # 3-4 years (highest)
+    elasticities[
+        married_women
+        & has_children
+        & (youngest_child_age >= 5)
+        & (youngest_child_age <= 10)
+    ] = 0.173  # 5-10 years
+    elasticities[married_women & has_children & (youngest_child_age >= 11)] = (
+        0.160  # 11+ years
+    )
+
+    # Lone parents - lower elasticities than married women, reflecting different constraints
+    lone_parents = (gender == "FEMALE") & ~is_married & has_children
+    elasticities[lone_parents & (youngest_child_age <= 4)] = 0.094  # 0-4 years
+    elasticities[
+        lone_parents & (youngest_child_age >= 5) & (youngest_child_age <= 10)
+    ] = 0.128  # 5-10 years
+    elasticities[
+        lone_parents & (youngest_child_age >= 11) & (youngest_child_age <= 18)
+    ] = 0.136  # 11-18 years
+
+    # Men (excluding lone fathers) - moderate, consistent elasticity
+    elasticities[(gender == "MALE") & ~(~is_married & has_children)] = 0.15
+
+    # Single women without children - same as men
+    elasticities[(gender == "FEMALE") & ~is_married & ~has_children] = 0.15
+
+    return elasticities
+
+
+def calculate_labour_net_income_elasticities(
+    sim: Simulation,
+) -> np.ndarray:
+    """Calculate labour supply income elasticities by demographic group.
+
+    Uses OBR elasticity estimates to assign income elasticities based on
+    gender, marital status, and presence/age of children. These elasticities
+    measure how labour supply responds to changes in unearned income.
+
+    Reference: https://obr.uk/docs/dlm_uploads/NICS-Cut-Impact-on-Labour-Supply-Note.pdf
+    Table A2 - Income elasticities
+
+    Args:
+        sim: PolicyEngine simulation object
+
+    Returns:
+        Array of income elasticities for each person (typically negative)
+    """
+    # Get demographic characteristics for elasticity assignment
+    gender = sim.calculate("gender")
+    is_married = sim.calculate("is_married", map_to="person")
+    has_children = sim.calculate("benunit_count_children", map_to="person") > 0
+    youngest_child_age = sim.calculate("youngest_child_age", map_to="person")
+
+    # Initialize elasticity array
+    elasticities = np.zeros(gender.shape, dtype=float)
+
+    # Married or cohabiting women - negative income elasticities (normal good)
+    married_women = (gender == "FEMALE") & is_married
+    elasticities[married_women & ~has_children] = (
+        0.0  # No income effect without children
+    )
+
+    # Stronger negative income effects with younger children
+    elasticities[married_women & has_children & (youngest_child_age <= 2)] = (
+        -0.185
+    )  # 0-2 years
+    elasticities[
+        married_women
+        & has_children
+        & (youngest_child_age >= 3)
+        & (youngest_child_age <= 4)
+    ] = -0.173  # 3-4 years
+    elasticities[
+        married_women
+        & has_children
+        & (youngest_child_age >= 5)
+        & (youngest_child_age <= 10)
+    ] = -0.102  # 5-10 years
+    elasticities[married_women & has_children & (youngest_child_age >= 11)] = (
+        -0.063
+    )  # 11+ years
+
+    # Lone parents - smaller negative income effects than married women
+    lone_parents = (gender == "FEMALE") & ~is_married & has_children
+    elasticities[lone_parents & (youngest_child_age <= 4)] = (
+        -0.037
+    )  # 0-4 years
+    elasticities[
+        lone_parents & (youngest_child_age >= 5) & (youngest_child_age <= 10)
+    ] = -0.075  # 5-10 years
+    elasticities[
+        lone_parents & (youngest_child_age >= 11) & (youngest_child_age <= 18)
+    ] = -0.054  # 11-18 years
+
+    # Men (excluding lone fathers) - small negative income effect
+    elasticities[(gender == "MALE") & ~(~is_married & has_children)] = -0.05
+
+    # Single women without children - same as men
+    elasticities[(gender == "FEMALE") & ~is_married & ~has_children] = -0.05
+
+    return elasticities
+
+
+def calculate_employment_income_change(
+    employment_income: np.ndarray,
+    derivative_changes: pd.DataFrame,
+    income_changes: pd.DataFrame,
+    substitution_elasticities: np.ndarray,
+    income_elasticities: np.ndarray,
+) -> np.ndarray:
+    """Calculate total labour supply response combining substitution and income effects.
+
+    This function implements the Slutsky equation decomposition of labour supply
+    responses into substitution and income effects. The total response is the
+    sum of these two components.
+
+    Args:
+        employment_income: Baseline employment income levels
+        derivative_changes: Changes in marginal rates (substitution effect driver)
+        income_changes: Changes in income levels (income effect driver)
+        substitution_elasticities: Elasticities for substitution effects
+        income_elasticities: Elasticities for income effects
+
+    Returns:
+        Array of employment income changes due to labour supply responses
+    """
+    # Calculate substitution effect: response to changes in marginal rates
+    substitution_response = (
+        employment_income
+        * derivative_changes["wage_rel_change"]
+        * substitution_elasticities
+    )
+
+    # Calculate income effect: response to changes in unearned income
+    income_response = (
+        employment_income
+        * income_changes["income_rel_change"]
+        * income_elasticities
+    )
+
+    # Total labour supply response is sum of substitution and income effects
+    total_response = substitution_response + income_response
+
+    # No response for people with zero employment income
+    total_response[employment_income == 0] = 0
+
+    df = pd.DataFrame(
+        {
+            "substitution_response": substitution_response,
+            "income_response": income_response,
+            "total_response": total_response,
+        }
+    )
+
+    return df.fillna(0)

policyengine_uk/microsimulation.py

@@ -47,7 +47,6 @@ class Microsimulation(Simulation):
         unweighted: bool = False,
     ):
         tracer: SimpleTracer = self.tracer
-
         result = super().calculate(
             variable_name, period, map_to=map_to, decode_enums=decode_enums
         )
@@ -81,3 +80,26 @@ class Microsimulation(Simulation):
             return values
         weights = self.get_weights(variable_names[0], period, map_to=map_to)
         return MicroDataFrame(values, weights=weights)
+
+    def compare(
+        self,
+        other: "Simulation",
+        variables: list[str] = None,
+        period: str = None,
+        change_only: bool = False,
+    ):
+        """Compare two simulations for a specific variable list.
+
+        Args:
+            other: Another Simulation instance to compare against
+            variables: List of variable names to compare. If None, compares all variables.
+
+        Returns:
+            DataFrame with comparison results
+        """
+        df = super().compare(
+            other, variables=variables, period=period, change_only=change_only
+        )
+        return MicroDataFrame(
+            df, weights=self.get_weights(variables[0], period)
+        )

policyengine_uk/parameters/gov/dynamic/obr_labour_supply_assumptions.yaml

@@ -0,0 +1,9 @@
+description: Follow OBR labour supply response assumptions to tax-benefit reforms.
+values:
+  0001-01-01: true
+metadata:
+  unit: bool
+  label: Follow OBR labour supply response assumptions
+  reference:
+    - title: OBR assumptions
+      href: https://obr.uk/docs/dlm_uploads/NICS-Cut-Impact-on-Labour-Supply-Note.pdf