What Are the Long-Term Effects of Cash Assistance on Adult Earnings and Intergenerational Outcomes?

Price and Song (2024) examine whether temporary cash assistance has long-term effects on adult earnings and children’s outcomes. They use data from the Seattle-Denver Income Maintenance Experiment, linked to Social Security administrative records over roughly 40 years. They find that adults who received cash assistance earned about $1,800 less annually (a 7.4% decline) and were 6.3 percentage points more likely to apply for disability benefits (about a 20% increase). These effects are concentrated near retirement age. In contrast, they find no statistically significant effects on children’s earnings, disability use, or broader life outcomes.

Why This Article Was Selected for The Policy Scientist

This article addresses a central policy question: whether direct cash transfers produce lasting economic effects for recipients and their children. This issue is especially salient given renewed interest in guaranteed income policies, pandemic-era transfers, and persistent inequality. Price and Song build on a long-standing literature on income support but extend it by examining outcomes over four decades, a horizon rarely observed. Their use of a large randomized controlled trial—the Seattle-Denver Income Maintenance Experiment—substantially strengthens causal inference, and the linkage to administrative data enhances measurement quality.

Full Citation and Link to Article

Price, D. J., & Song, J. (forthcoming). The long-term effects of cash assistance. American Economic Journal: Economic Policy. https://www.aeaweb.org/articles?id=10.1257/pol.20220166

Central Research Question

This article examines whether temporary cash assistance generates persistent, long-term effects on economic behavior and well-being. Specifically, Price and Song ask whether providing several years of unconditional or quasi-unconditional income support to low- and middle-income families alters adult labor supply, earnings trajectories, disability program participation, and related outcomes decades after the intervention ends. A secondary question concerns intergenerational effects: whether children exposed to increased household income and reduced parental labor supply experience measurable differences in adult earnings or reliance on government programs. The study is motivated by a broader policy concern—whether cash transfers meaningfully reduce poverty in a durable way or instead produce unintended behavioral responses that offset initial gains.

Previous Literature

The article situates itself within three strands of research. First is the literature on income maintenance experiments (IMEs), particularly the negative income tax trials conducted in the United States during the 1960s and 1970s. These studies established that cash transfers reduce labor supply in the short run, primarily through reductions in hours worked. However, they largely failed to track participants beyond a few years after program completion, leaving long-term consequences unresolved.

Second is a broader literature on income effects and labor supply, including studies of lottery winnings and other exogenous income shocks. These studies generally find modest reductions in labor supply, though the magnitude of long-run responses remains contested.

Third is the intergenerational mobility literature examining whether increases in parental income improve children’s long-term outcomes. While some studies identify positive effects—particularly when income increases occur early in childhood—others find limited or no impacts. Price and Song contribute by integrating these strands, providing rare long-term causal evidence on both adult and child outcomes following a temporary income intervention.

Data

The analysis relies on the Seattle-Denver Income Maintenance Experiment (SIME/DIME), a large-scale randomized controlled trial conducted in the 1970s. Approximately 4,800 families were randomly assigned to receive a guaranteed minimum income for three or five years, with varying benefit levels and tax-back rates on earned income. This design provides a strong basis for causal inference.

The key innovation of the study is the linkage of experimental records to administrative data from the Social Security Administration (SSA), covering the period from 1978 through 2013. These records include annual earnings, disability insurance applications and awards, and mortality. The authors also incorporate state-level data on marriage and divorce outcomes. Because the original experimental data lack standard identifiers, the authors develop a novel matching algorithm based on family birth patterns and name frequencies, ultimately linking approximately 45% of adult participants and 59% of children to administrative records.

The resulting dataset spans roughly four decades, allowing the authors to observe outcomes across the full lifecycle, including retirement behavior. The use of administrative data substantially reduces concerns about attrition and measurement error, though the partial match rate introduces potential concerns about external validity.

Methods

The study employs a straightforward but rigorous empirical strategy grounded in randomized assignment. Treatment effects are estimated using least squares regressions comparing outcomes for individuals assigned to the cash assistance program with those assigned to the control group. The models include controls for stratification variables used in the original experimental design—such as site, race, family structure, and baseline income—as well as demographic characteristics.

For outcomes observed over time, such as annual earnings, the authors estimate panel regressions with year fixed effects, allowing them to trace treatment effects across the lifecycle. Standard errors are clustered at the family level to account for within-family correlation.

The randomized controlled trial design is the central methodological strength, ensuring that estimated differences can be interpreted causally under standard assumptions. The authors complement this design with extensive robustness checks, including alternative specifications, subgroup analyses, and validation exercises addressing potential biases introduced by the matching process. While the analysis relies primarily on reduced-form estimation rather than structural modeling, the causal identification strategy is strong and transparent.

Findings/Size Effects

The results reveal substantial and persistent effects on adult outcomes. Individuals assigned to receive cash assistance earn approximately $1,800 less per year in the post-experimental period, representing a 7.4% reduction relative to the control group mean. This decline is driven largely by a reduction in labor force participation rather than lower wages conditional on working. Treated individuals are also 3.3 percentage points less likely to work in a given year.

The effects are concentrated later in life, particularly near retirement age. Treatment induces earlier exit from the labor force, with individuals retiring nearly one year earlier on average. In addition, treated individuals are 6.3 percentage points more likely to apply for disability benefits, a roughly 20% increase relative to the baseline application rate. However, there is no statistically significant increase in disability benefit receipt, suggesting that marginal applicants are less likely to qualify.

When aggregated over the lifecycle, the earnings losses are substantial. The authors estimate that each dollar of additional government transfers during the experiment reduces lifetime earnings by more than three dollars, reflecting both short-term and long-term behavioral responses.

In contrast, the analysis finds no meaningful effects on children’s long-term outcomes. There are no statistically significant differences in adult earnings, disability participation, or major demographic outcomes for children exposed to the treatment. The estimates are sufficiently precise to rule out moderate-sized effects, indicating that the income gains experienced during childhood did not translate into measurable improvements in adult economic status.

Additional analyses suggest that the adult effects are not driven by changes in wealth accumulation, preferences for leisure, or wage rates. Instead, the evidence points toward changes in occupational trajectories. Treated individuals appear to shift into occupations requiring less education and cognitive skill, which may contribute to earlier retirement and reduced long-term labor supply.

Conclusion

The findings indicate that temporary cash assistance can produce durable changes in adult economic behavior, even decades after the intervention ends. These effects manifest primarily through reduced labor force attachment and earlier retirement, rather than through persistent wage declines. At the same time, the absence of detectable intergenerational effects suggests that increased parental income, at least in this context, does not systematically improve children’s long-term economic outcomes.

The study’s contribution lies in its combination of experimental design and long-term follow-up, addressing a gap in the literature on the persistence of income transfer effects. While the historical setting and program design limit direct generalization to contemporary policies, the results provide a benchmark for evaluating the potential long-run consequences of cash-based interventions.