Add labor supply elasticity age heterogeneity

changelog_entry.yaml

@@ -0,0 +1,4 @@
+- bump: minor
+  changes:
+    added:
+    - Re-add age heterogeneity in labor supply response elasticities using multiplier approach.

policyengine_us/parameters/gov/simulation/labor_supply_responses/elasticities/README.md

@@ -1 +1,206 @@
-# Elasticities
+# Labor Supply Response Elasticities
+
+This directory contains parameters for labor supply response elasticities used in behavioral microsimulation analysis.
+
+## Overview
+
+Labor supply elasticities measure how individuals adjust their work behavior in response to changes in economic incentives:
+
+- **Substitution elasticity**: How labor supply responds to changes in the effective marginal wage rate (after-tax wage)
+- **Income elasticity**: How labor supply responds to changes in disposable income
+
+## Age Heterogeneity: The Multiplier Approach
+
+Both elasticity types support age-based heterogeneity through an **age multiplier** that scales base elasticities for individuals aged 65 and over. This approach is based on empirical research showing that older workers have higher labor supply elasticities than working-age adults.
+
+### Research Findings
+
+- **French (2005)**: Elasticities 3.0x-3.25x higher for age 60 vs age 40 workers
+- **CBO Working Papers (2012-12, 2012-13)**: Frisch elasticity ranges from 0.27 to 0.53 (central: 0.40) for working-age adults
+- **General pattern**: Retirement-age individuals (62-70) have particularly high elasticities, especially on the extensive margin (whether to work at all)
+
+### Why a Multiplier Approach?
+
+The multiplier approach was chosen over separate age-specific parameters because:
+1. **Evidence-based**: Literature consistently shows older workers are more elastic, but doesn't provide income-decile-specific multipliers
+2. **Parsimony**: 13 total parameters instead of 44 (11 base elasticities + 2 age multipliers)
+3. **Transparency**: Users can easily understand and adjust one multiplier value
+4. **Flexibility**: Multiplier can range from 1.0 (no age effect) to 3.0+ (French 2005 finding)
+
+## Substitution Elasticity Structure
+
+The substitution elasticity uses a two-step calculation:
+
+### Step 1: Base Elasticity (by position and decile)
+
+Base elasticities vary by:
+1. **Position**: primary earner vs secondary earner within tax unit
+2. **Decile**: 10 income deciles (for primary earners only)
+
+This creates 11 base elasticity parameters:
+- 10 for primary earners by decile
+- 1 for secondary earners (all deciles)
+
+### Step 2: Age Multiplier
+
+For individuals aged 65 and over, the base elasticity is multiplied by `age_multiplier_65_and_over`.
+
+**Formula**:
+- If age < 65: `elasticity = base_elasticity`
+- If age >= 65: `elasticity = base_elasticity × age_multiplier_65_and_over`
+
+### Global Override
+
+The `all` parameter overrides all base elasticities and age multipliers if set to non-zero.
+
+## Income Elasticity Structure
+
+The income elasticity uses a simpler two-parameter structure:
+
+1. **Base elasticity**: Applied to all working-age individuals (under 65)
+2. **Age multiplier**: Applied to individuals 65 and over
+
+**Formula**:
+- If age < 65: `elasticity = base`
+- If age >= 65: `elasticity = base × age_multiplier_65_and_over`
+
+The `all` parameter overrides base and multiplier if set to non-zero.
+
+## Default Values
+
+- All base elasticities default to 0 (no behavioral response)
+- Age multipliers default to 2.0 (conservative estimate based on literature)
+- Users must explicitly set base elasticity values to enable behavioral responses
+
+## Usage Examples
+
+### Example 1: No behavioral response (default)
+```yaml
+substitution:
+  by_position_and_decile:
+    primary:
+      1: 0
+      2: 0
+      # ... all zeros
+    secondary: 0
+  age_multiplier_65_and_over: 2.0  # Doesn't matter since base is 0
+
+income:
+  base: 0
+  age_multiplier_65_and_over: 2.0  # Doesn't matter since base is 0
+```
+
+Result: Everyone has zero elasticity regardless of age.
+
+### Example 2: Same elasticity for all ages
+```yaml
+substitution:
+  by_position_and_decile:
+    primary:
+      1: 0.31  # CBO-style values
+      2: 0.27
+      # ... other deciles
+    secondary: 0.40
+  age_multiplier_65_and_over: 1.0  # No age effect
+
+income:
+  base: -0.04
+  age_multiplier_65_and_over: 1.0  # No age effect
+```
+
+Result: Same elasticity for everyone regardless of age.
+
+### Example 3: Higher elasticity for elderly (RECOMMENDED)
+```yaml
+substitution:
+  by_position_and_decile:
+    primary:
+      1: 0.31
+      2: 0.27
+      # ... other deciles
+    secondary: 0.40
+  age_multiplier_65_and_over: 2.0  # Conservative estimate
+
+income:
+  base: -0.04
+  age_multiplier_65_and_over: 2.0
+```
+
+Result:
+- Age 40, primary earner, decile 1: elasticity = 0.31
+- Age 70, primary earner, decile 1: elasticity = 0.31 × 2.0 = 0.62
+- Age 40: income elasticity = -0.04
+- Age 70: income elasticity = -0.04 × 2.0 = -0.08
+
+### Example 4: Aggressive multiplier based on French (2005)
+```yaml
+substitution:
+  age_multiplier_65_and_over: 3.0  # Upper end of literature
+
+income:
+  base: -0.04
+  age_multiplier_65_and_over: 3.0
+```
+
+Result:
+- Age 70, primary earner, decile 1 (base 0.31): elasticity = 0.31 × 3.0 = 0.93
+- Age 70: income elasticity = -0.04 × 3.0 = -0.12
+
+## Multiplier Value Guidance
+
+Based on literature review (see `age_heterogeneity_analysis.md`):
+
+| Multiplier | Interpretation | Source |
+|------------|----------------|--------|
+| 1.0 | No age difference | Ignores evidence |
+| 1.5 | Conservative | Lower end of estimates |
+| 2.0 | **Recommended default** | Conservative but evidence-based |
+| 2.5 | Moderate | Midpoint of French (2005) range |
+| 3.0 | Aggressive | French (2005) finding for age 60 vs 40 |
+| 3.25 | Upper bound | Upper end of French (2005) range |
+
+**Note**: The exact multiplier value has uncertainty. We recommend starting with 2.0 and conducting sensitivity analysis with values in the 1.5-3.0 range.
+
+## References
+
+- [French (2005): "The Effects of Health, Wealth, and Wages on Labour Supply and Retirement Behaviour"](https://academic.oup.com/restud/article-abstract/72/2/395/1558553) - Review of Economic Studies 72(2): 395-427
+- [CBO Working Paper 2012-12: "A Review of Recent Research on Labor Supply Elasticities"](https://www.cbo.gov/publication/43675) - McClelland & Mok
+- [CBO Working Paper 2012-13: "Review of Estimates of the Frisch Elasticity of Labor Supply"](https://www.cbo.gov/publication/43676) - Reichling & Whalen
+
+## Implementation Notes
+
+- Age is determined from the `age` variable for the year (`period.this_year`)
+- Age 65 is the cutoff: age < 65 uses base elasticity, age >= 65 applies multiplier
+- Primary earner is defined as the highest earner within the tax unit
+- Earnings deciles are determined using hardcoded markers (TODO: parametrize)
+- Negative total earnings result in zero elasticity (using `max_(earnings, 0)`)
+- Zero base elasticity remains zero even with multiplier (0 × multiplier = 0)
+
+## Technical Details
+
+### How Primary/Secondary Earner is Determined
+
+Within each tax unit, the person with the highest total earnings (employment + self-employment) is designated as the primary earner. All other earners in the unit are secondary earners. This means:
+- Single-person tax units: That person is always primary
+- Multi-person tax units: Only the highest earner gets primary elasticity; others get zero (since secondary earner base elasticities default to 0)
+
+### Earnings Decile Markers
+
+Current hardcoded decile boundaries (TODO: parametrize):
+- Decile 1: $0 - $14,000
+- Decile 2: $14,000 - $28,000
+- Decile 3: $28,000 - $39,000
+- Decile 4: $39,000 - $50,000
+- Decile 5: $50,000 - $61,000
+- Decile 6: $61,000 - $76,000
+- Decile 7: $76,000 - $97,000
+- Decile 8: $97,000 - $138,000
+- Decile 9: $138,000 - $1,726,000
+- Decile 10: $1,726,000+
+
+### Negative Earnings Handling
+
+To prevent negative earnings from causing sign flips in economic responses, the code applies `max_(earnings, 0)` before calculating elasticities. This means:
+- Positive net earnings: Normal elasticity calculation
+- Negative net earnings: Treated as zero earnings, resulting in zero substitution elasticity
+- Income elasticity still applies to individuals with negative earnings

policyengine_us/parameters/gov/simulation/labor_supply_responses/elasticities/income/age_multiplier_over_threshold.yaml

@@ -0,0 +1,11 @@
+description: Multiplier applied to base income elasticity for individuals at or above the age threshold. Research suggests income effects are stronger for older workers. French (2005) shows much higher elasticities near retirement age. A multiplier of 2.0 is conservative. Set to 1.0 for no age difference.
+values:
+  2020-01-01: 2.0
+metadata:
+  unit: /1
+  label: elasticity multiplier for ages at or above threshold
+  reference:
+    - title: "French (2005): The Effects of Health, Wealth, and Wages on Labour Supply and Retirement Behaviour"
+      href: https://academic.oup.com/restud/article-abstract/72/2/395/1558553
+    - title: "CBO Working Paper 2012-12: A Review of Recent Research on Labor Supply Elasticities"
+      href: https://www.cbo.gov/publication/43675

policyengine_us/parameters/gov/simulation/labor_supply_responses/elasticities/income/age_threshold.yaml

@@ -0,0 +1,9 @@
+description: Age at which the age multiplier begins to apply. Individuals at this age or older will have their base elasticity multiplied by the age multiplier. Default is 65 (typical retirement age), but can be adjusted to reflect different retirement patterns or policy scenarios (e.g., 62 for early retirement, 67 for full retirement age).
+values:
+  2020-01-01: 65
+metadata:
+  unit: year
+  label: age threshold for elasticity multiplier
+  reference:
+    - title: "French (2005): The Effects of Health, Wealth, and Wages on Labour Supply and Retirement Behaviour"
+      href: https://academic.oup.com/restud/article-abstract/72/2/395/1558553

policyengine_us/parameters/gov/simulation/labor_supply_responses/elasticities/income.yaml → policyengine_us/parameters/gov/simulation/labor_supply_responses/elasticities/income/all.yaml

@@ -1,4 +1,4 @@
-description: Percent change (of the change in disposable income) in labor supply given a 1% change in disposable income.
+description: Percent change (of the change in disposable income) in labor supply given a 1% change in disposable income. This parameter overrides the base income elasticity and age multiplier if provided.
 values:
   2020-01-01: 0
 metadata:

policyengine_us/parameters/gov/simulation/labor_supply_responses/elasticities/income/base.yaml

@@ -0,0 +1,11 @@
+description: Base income elasticity of labor supply for working-age individuals (under 65). This value is multiplied by the age multiplier for individuals 65 and over. Typically negative, indicating that higher income reduces labor supply.
+values:
+  2020-01-01: 0
+metadata:
+  unit: /1
+  label: base income elasticity below threshold
+  reference:
+    - title: "CBO Working Paper 2012-12: A Review of Recent Research on Labor Supply Elasticities"
+      href: https://www.cbo.gov/publication/43675
+    - title: "CBO Working Paper 2012-13: Review of Estimates of the Frisch Elasticity of Labor Supply"
+      href: https://www.cbo.gov/publication/43676

policyengine_us/parameters/gov/simulation/labor_supply_responses/elasticities/substitution/age_multiplier_over_threshold.yaml

@@ -0,0 +1,11 @@
+description: Multiplier applied to all substitution elasticities for individuals at or above the age threshold. Research shows older workers have higher labor supply elasticities than working-age adults. French (2005) finds elasticities 3x higher for age 60 vs age 40. A multiplier of 2.0 is conservative. Set to 1.0 for no age difference.
+values:
+  2020-01-01: 2.0
+metadata:
+  unit: /1
+  label: elasticity multiplier for ages at or above threshold
+  reference:
+    - title: "French (2005): The Effects of Health, Wealth, and Wages on Labour Supply and Retirement Behaviour"
+      href: https://academic.oup.com/restud/article-abstract/72/2/395/1558553
+    - title: "CBO Working Paper 2012-12: A Review of Recent Research on Labor Supply Elasticities"
+      href: https://www.cbo.gov/publication/43675

policyengine_us/parameters/gov/simulation/labor_supply_responses/elasticities/substitution/age_threshold.yaml

@@ -0,0 +1,9 @@
+description: Age at which the age multiplier begins to apply. Individuals at this age or older will have their base elasticity multiplied by the age multiplier. Default is 65 (typical retirement age), but can be adjusted to reflect different retirement patterns or policy scenarios (e.g., 62 for early retirement, 67 for full retirement age).
+values:
+  2020-01-01: 65
+metadata:
+  unit: year
+  label: age threshold for elasticity multiplier
+  reference:
+    - title: "French (2005): The Effects of Health, Wealth, and Wages on Labour Supply and Retirement Behaviour"
+      href: https://academic.oup.com/restud/article-abstract/72/2/395/1558553

policyengine_us/parameters/gov/simulation/labor_supply_responses/elasticities/substitution/all.yaml

@@ -0,0 +1,6 @@
+description: Percent change (of the change in the effective marginal wage) in labor supply given a 1% change in the effective marginal wage. This parameter overrides all other substitution elasticities if provided.
+values:
+  2020-01-01: 0
+metadata:
+  unit: /1
+  label: substitution elasticity of labor supply