economics Paradigm Challenge

Turns out the economic 'cost' of diabetes on the workforce has been wildly overestimated for the last 30 years.

March 17, 2026

Original Paper

Bounding the Economic Loss from Diabetes with Imperfect Instruments

Matthew E. Kahn

SSRN · 6307883

The Takeaway

While data consistently shows people with diabetes work 10% less, this study uses a new statistical model to show the disease itself isn't the primary cause. It reveals that the same early-life conditions that lead to diabetes also lead to unemployment, meaning the disease is more a marker of pre-existing disadvantage than a direct cause of productivity loss.

From the abstract

Roughly 25% of older Americans have diabetes. OLS estimates indicate that diabetes is associated with a roughly ten percentage point reduction in employment and this effect has been constant for decades. Estimating its causal effect is challenging because diabetes is correlated with health behaviors and early-life conditions that also shape labor market outcomes. Using thirty years of data from the Health and Retirement Study (HRS), I implement an imperfect instrumental variables (IIV) approach