Working Papers

Economic hardship can shape health behaviors, including those related to opioids. We estimate the effects of mass layoffs on opioid use and misuse in Denmark with three main contributions on how economic shocks impact health, and for whom. First, we find the clearest evidence that economic conditions affect opioid use: individuals increase consumption by 65%, with evidence of misuse. Second, we disentangle indirect effects: spouses consume 40% more opioids. Third, we connect opioid demand (as we study) to the larger literature on supply, finding evidence that effects of layoffs are stronger in areas with a large underlying availability of opioids.
 
 

The fissuring of production, as through outsourcing, is frequently cited as a driver of changing inequality, but past research has mostly focused on narrow examples of it rather than comprehensively measuring the importance of this channel. We advance this literature by developing a novel algorithm to calculate average pay within a supply chain, regardless of how firm boundaries may split up that production. Using employer-employee matched data combined with firm-to-firm sales data in Chile between 2005 and 2024, we find that changing between-firm inequality is explained not by changes within supply chains (as expected if firms reorganize production through outsourcing), but rather by changes between supply chains (as expected if there are changes in the complementarity of production functions). Past research suggests that outsourcing leads to wage drops; we replicate this result, but reconcile it with our variance decomposition by showing that insourcing, which is rarely studied, is approximately as common as outsourcing and has roughly equal and opposite effects on pay.
 
 

Power laws are common in economics, as in city and firm sizes, and can cause extreme heteroskedasticity. I show that estimators based on observations exhibiting this extreme heteroskedasticity may not be consistent or asymptotically normal and may have unreliable confidence intervals. These problems can occur even without heteroskedasticity if weighted estimators are used. I construct a quasi-maximum likelihood estimator to form more accurate estimates and more reliable inference. This estimator is broadly useful when weighting is considered to improve estimators' precision. Simulations confirm it improves estimation precision and inference, while a replication shows it can lead to substantially different results.
 
Stata command (also available through SSC): regoptwgt
 
 

Some occupations pay substantially higher average wages than others. If workers switch occupations, do their wages increase as much as predicted by differences in averages? We answer this question using a two-way fixed effect framework, with occupation and worker fixed effects, in the United States between 1983 and 2013, finding that workers who move to an occupation paying 10% higher average wages can only expect their own wages to increase by 3.1%. This means that, although occupational average wage differences explain about a third of wage inequality, occupation fixed effects explain only about 3%. Although fixed effects are small relative to differences in average wages, workers do gain an economically significant benefit by moving to certain occupations. We estimate fixed effects precisely; because of their economic significance, we therefore report them individually, along with their correlation with occupational characteristics: higher-skill jobs generally have higher fixed effects. We also show that fixed effects are relatively stable over time.
 
This paper was previously circulated as part of a larger paper titled "What Occupations Do."
 
Data