PolicyEngine US Launches Behavioral Responses

We've expanded our modeling framework to capture how policy changes affect labor supply.

We’re grateful to Arnold Ventures for supporting this model enhancement and research.

When faced with new public policies, Americans adjust their economic behaviors. These policy changes — whether at the federal or state level — set off chain reactions across society. Workers modify their working hours, investors change when to realize their gains, and businesses reshape their practices. Looking at policies through a static lens misses these dynamic human responses to changing incentives. PolicyEngine incorporates two fundamental behavioral responses to policy changes, which we explain in this report:

1. Income effects on labor supply

2. Substitution effects on labor supply

How do we measure behavioral responses to policy changes? The key lies in elasticities, which measure how people respond to policy changes. A larger elasticity signals a stronger behavioral shift, while a smaller one indicates more modest adjustments. PolicyEngine lets users quickly apply elasticities from the Congressional Budget Office or set their own. This report describes these responses and how they affect results in our microsimulation model.

In PolicyEngine, we model how Americans adjust their work hours through two key mechanisms: income and substitution effects. Income effects show how people might choose more leisure time when their income increases. Substitution effects capture how workers respond when their take-home pay for extra hours changes.

How do people adjust their work hours when they become better off? The income elasticity measures this behavior, quantifying how Americans modify their work patterns when their take-home pay changes through taxes and transfers. Following the Congressional Budget Office (CBO)’s central estimate, PolicyEngine sets the default income elasticity at -0.05. In practical terms, a 10% increase in after-tax income results in an average of 0.5% decrease in work.

How much people work often depends on their take-home pay per extra hour. The substitution elasticity measures this relationship — how work habits change when the reward for additional hours shifts. For instance, when higher tax rates or steeper benefit phase-outs reduce the pay from extra hours, workers might cut back on overtime or reduce part-time hours. PolicyEngine implements the Congressional Budget Office’s heterogeneous substitution elasticities, which vary by income level. The elasticities range from 0.31 for the lowest income decile to 0.22 for the highest deciles among primary earners, with secondary earners at 0.32. This means when workers in the lowest income decile keep 10% less of their earnings from each additional hour, they typically reduce their work by 3.1%, while higher-income workers reduce their work by 2.2%. The full range of elasticities by income group is shown in the table below.

To see how these elasticities work in practice, let’s examine how a Massachusetts worker earning $50,000 annually would respond to a potential increase in the state income tax rate from 5% to 6%:

Our Massachusetts worker earns $25.00 per hour, working 2,000 hours annually for $50,000 total earnings. Their net income changes from $40,034 to $39,598 due to an increase in state income tax from 5% to 6%. This $436 reduction represents a 1.09% drop in their current net income ($436 / $40,034 = 0.0109 = 1.09%). The percentage differs from the 1pp tax rate increase because the tax applies to gross income while the percentage change is relative to net income, which is lower due to taxes and deductions.

Under the policy change, their net marginal hourly wage would fall from $18.83 to $18.58; this $0.25 drop is 1% of their $25 nominal wage and represents a 1.3% fall in their net marginal wage. It’s important to note that the change in net income (-1.09%) is smaller than the 1% increase in the state tax rate due to several factors:

• The standard deduction reduces taxable income before the tax rate is applied

• The tax increase only affects state taxes, not federal taxes or other deductions

• The percentage change is calculated against total take-home pay, which includes all adjustments and deductions

To understand these calculations in detail, let's examine how this worker responds to the policy change. Applying the CBO's elasticities, this reform triggers two behavioral responses:

Combining these opposing effects yields a net reduction of 0.271% in work hours. For our Massachusetts worker, this translates to approximately 5.4 fewer work hours (0.271% of 2,000) and $135.5 less in earnings per year (5.4 hours × $25/hour).

While this example shows how behavioral responses affect an individual worker, we can use PolicyEngine’s microsimulation to examine how these responses impact government revenues and distributional effects.

By default, PolicyEngine operates with no behavioral responses. To activate behavioral responses, users can either:

1. Apply CBO’s behavioral responses with one click

2. Set their own elasticities using the same structure as the CBO’s (single income elasticity and substitution elasticities varying by income and primary/secondary)

The fastest way to apply behavioral responses is to toggle “Apply CBO behavioral responses” in the baseline policy section, as shown below.

This sets each of the elasticities per CBO’s table. Alternatively, you can navigate to the Parameters tab and find the elasticity settings under Simulation > Labor supply responses as shown below.

When users input any non-zero elasticities, the model incorporates behavioral responses into its tax-benefit calculations. For each policy reform simulation, the model first calculates how policy changes affect both income and tax rates. Then, it estimates two behavioral changes:

• How changes in total income influence work decisions

• How marginal tax changes affect work hours

Since each person’s earnings can impact household net income differently, the model calculates marginal tax rates for each of the top three earners and applies their labor supply responses separately.

After adjusting how much each worker earns, we apply the microsimulation model again to calculate their new taxes and benefits. For example, if elasticities imply that a reform will increase an earner’s labor supply (and they face a positive marginal tax rate), they will also pay more in tax after the final step.

To demonstrate how behavioral responses affect the broader fiscal impact, let’s examine the revenue implications of raising the Massachusetts state income tax rate from 5% to 6%. Applying CBO’s elasticities, we find that the reform would reduce hours worked by 0.24%. While the income effect raises employment by 1,700 full-time equivalents (since the tax lowers the net income of affected workers), the substitution effect lowers employment by 9,900 full-time equivalents (since affected workers see their net-of-tax wage rate fall), for a net effect of 8,200 FTEs.

In turn, earnings fall 0.27%; this exceeds the hours worked effect because low-income workers, who pay no state tax under current law due to various deductions, exemptions, and non-refundable credits, are unaffected by the reform. Extended visualizations show how these effects vary with income decile.

Finally, the reduced earnings in turn reduce the revenue projection from $3.6 billion without behavioral responses to $3.2 billion with them (an 11% reduction). PolicyEngine estimates the effect of behavioral responses on other outcomes too; for example, without behavioral responses, the tax reform lowers the net income share held by the top 10% by 0.2% without behavioral responses, but 0.3% with them.

In this report, we explain how PolicyEngine models two key behavioral responses:

1. Income effect for labor supply Constant elasticity of -0.05

2. Substitution effect for labor supply Elasticity varying between 0.22 and 0.32 depending on income and whether the worker is primary or secondary within the household

While we have focused on these two core elasticities, PolicyEngine continues to expand its behavioral modeling capabilities; for example, we are now working on adding capital gains responses. The platform’s flexibility allows users to customize behavioral response assumptions, including labor supply elasticities, enabling a more nuanced analysis of public policy impacts.

max ghenis
PolicyEngine's Co-founder and CEO



nikhil woodruff
PolicyEngine's Co-founder and CTO



jason debacker
Associate Professor of Economics at the University of South Carolina



vahid ahmadi
Research Associate at PolicyEngine

© 2025 PolicyEngine. All rights reserved.