Abstract: Health shocks impose substantial economic costs on individuals, but within-household spillovers on unaffected spouses are less well understood. Spouses can increase work to replace lost income (“the added worker effect”) or reduce work to provide care, and gender norms may play an important role in these responses. I study how serious health shocks affect household labour supply using administrative data linking hospital records with tax returns for Canadian married couples from 2004 to 2019. Exploiting variation in the timing of cancer diagnoses, strokes, and acute hospitalizations, I find asymmetric spousal responses by gender. Following a partner’s health shock, wives reduce earnings by 4 percent in the first year, with losses persisting at 3.5 percent through year five. Husbands reduce earnings by 1.5 percent initially but recover to baseline within three years. The gender gap widens with shock severity. For strokes, which generate intensive caregiving needs, wives’ earnings fall by 7 percent while husbands’ earnings increase by 5 percent. The probability of permanent labour force exit rises by 2.0 percentage points for wives compared to 1.6 percentage points for husbands. These patterns emerge despite universal healthcare coverage that eliminates medical cost concerns. The results reveal that when formal long-term care is limited, households respond to binding caregiving constraints through gender specialization, with women bearing disproportionate and lasting economic costs. 

Funded by: Resources for the Future - Resilient Energy Economics Initiative 

Abstract: The final cleanup of an oil or gas well is an expensive endeavor, resulting in many yet-to-be-remediated wells scattered across North America.  Recent initiatives have devoted billions of dollars in federal funds for the cleanup of these wells.  In this paper, we evaluate the impact of a program in Alberta, Canada, that directed some of the federal funding for cleanup toward investment in Indigenous firms and communities. The program would cover up to 50% of the cost to clean up inactive wells, but in certain phases, would cover 100% of the cost if an Indigenous-owned company were contracted to do the cleanup. In one phase, the program specifically earmarked $100 million for closure work on First Nations reserves or Métis Settlements. 

We show that the added incentives increased well cleanup on First Nations reserves and also led to the creation of new Indigenous-owned companies. During the phases of added subsidies for Indigenous firms, almost half of the funding allocations went to Indigenous-owned companies. Upon removal of the added subsidies, we see a reduction in the hiring of Indigenous-owned companies. Our results shed light on industrial policies aiming to mitigate environmental risks and promote inclusive economic growth. 

Specific role in the paper: Author of one replication report.

Abstract: Trust in science is paramount for advancing knowledge. To ensure thistrust, scientific research must adhere to transparent processes and yield robust find-ings. The present study pushes the boundaries of understanding research reliabilityby mass reproducing and replicating claims from 110 papers in leading economicand political science outlets. The analysis, involving computational reproducibilitychecks and robustness assessments, reveals several patterns. First, we uncover a highrate (over 85%) of fully reproducible results. Second, excluding very minor errorslike missing packages or paths, we uncover coding errors forabout 25% of studies,with some studies containing multiple errors. Third, we test the robustness of theresults to 5,494 re-analyses. We find a robustness, reproducibility and replicabilityrate of 71%. Robustness reproducibility rates are relatively higher for re-analysesusing new data and lower for re-analyses involving changingthe definition of the de-pendent variable or the sample. Fourth, 52% of reanalysis effect size estimates aresmaller than the original published effects, and the averagestatisticalsignificanceof re-analysis is 77% of the original. Last, we rely on a many-analysts approach toanswer eight additional research questions. Our findings underscore the need forcontinued efforts to enhance the transparency and reliability of scientific research