Todd Morris

We study the workplace spillovers of a Dutch policy that progressively raised the statutory retirement age and nearly tripled employment at targeted ages. Using high-frequency register data, we estimate firms’ responses to individual retirement delays of 3–13 months. Consistently, our estimates show firms delay hiring around these events, and coworkers receive fewer career-progression opportunities via negative effects on hours worked and wages. Combined, these spillovers fully offset the additional hours worked by older workers within the firms, disproportionately curb career progression for young workers and women, and approximately triple the policy’s ratio of welfare costs to fiscal savings.

We study the welfare implications of employment protection for older workers, exploiting recent bans on mandatory retirement across Canadian provinces. Using linked employer-employee tax data, we show that the bans cause large and similar reductions in job separation rates and retirement hazards at age 65, with further reductions at higher ages. The effects vary substantially across industries and firms, and around two-fifths of the adjustments occur between ban announcement and implementation dates. We find no evidence that the demand for older workers falls, but the welfare effects are mediated by spillovers on savings behavior, workplace injuries, and spousal retirement timing.

We show that a targeted law can modify teens’ risky behavior. We examine the effects of an Australian intervention banning first-year drivers from driving late at night with multiple peers, which had accounted for one-fifth of their traffic fatalities. Using data on individual drivers linked to crash outcomes, we find the reform more than halves targeted crashes, casualties and deaths. There are large positive spillovers through lower crashes earlier in the evening and beyond the first year, suggesting broad and persistent declines in high-risk driving. Overall, the targeted intervention delivers gains comparable to harsher restrictions that delay teen driving.

Many governments offer tax concessions for retirement contributions. In this paper, we show that income responses are crucial for understanding their effectiveness in raising retirement savings and alleviating the fiscal pressures of population aging. Using tax register data, we study large changes in caps on tax-favored contributions to individual retirement accounts in Australia. We find that higher caps increase retirement contributions considerably, with around two-thirds of this response financed by increases in earned income. The resulting gain in income tax revenue offsets the fiscal loss from tax concessions, highlighting the importance of taking income and labor supply responses into account.

As governments try to contain rising expenditure on retirement pensions by increasing eligibility ages, there are concerns that such reforms disproportionately affect poorer households. Using detailed longitudinal data, I examine this trade-off in the context of an Australian reform that increased women’s pension-eligibility age from 60 to 65. While this reform significantly reduced government spending on women at affected ages, the negative effects on household incomes were concentrated among poorer households. These unequal impacts meant that the reform temporarily increased relative poverty rates among affected women by around 4 percentage points and inequality measures by 6 to 19 percent.

I revisit the labor supply effects of a major Australian reform that increased women’s pension age from 60 to 65. Atalay and Barrett (RESTAT 2015) studied these effects using repeated household surveys and a differences-in-differences design in which male cohorts form the comparison group. They estimate that the reform increased female labor force participation by 12 percentage points. Using earlier data, I show that the parallel-trends assumption did not hold before the reform because of a strong female-specific trend in participation rates across the relevant cohorts. Accounting for this trend, the estimated effect on female participation falls by two-thirds and becomes statistically insignificant at conventional levels.