The Economic Value of Eliminating Diseases

We estimate the causal effect of 334 different types of health shocks on a set of economic and health outcomes: medical expenses, mortality, disability, labor market participation, labor earnings, and the need for nursing home care. To this end, we examine detailed data on 6.9 million people diagnosed by medical specialists between 2013 and 2017. We address the empirical challenge that latent variables (e.g. life style) could be correlated both with the treatment status (diagnosis) and the outcome variables: for each individual diagnosed in a specific year, we consider as the counterfactual an individual that was not diagnosed in that year but will be in subsequent years. If some latent variables (e.g. life style) are related to a disease, the timing of the occurrence is shown to be largely random. We quantify the benefits of (hypothetically) ``eliminating'' diseases and demonstrate that they have different consequences for people of different social strata. Our results reveal substantial heterogeneity in the welfare loss by types of disease for different types of people. For instance, the (equally weighted) average effect of health shocks on labor market participation is decreasing in permanent income and people with lower permanent income are more likely to be declared fully disabled. Furthermore, conditional on staying in the labor market, the drop in earnings following a health shock is more severe for people in the low permanent income group. In addition, we find that the average causal effect of a health shock increases with age for each of the above economic and health outcomes. We then incorporate the estimates into a life-cycle model to estimate the welfare gains associated with curing diseases. By means of three examples of very different diseases we illustrate that the willingness to pay to eliminate a disease varies significantly. For instance, a 25-year old woman in the highest permanent income group would pay 25 times more to eliminate the risk of lung cancer than a woman in the lowest income class. The former would only pay 5 times as much to eliminate herniated disc problems. A man in the highest income class would pay less than half than his female counterpart to eliminate these diseases. We also point out the potential implications of such results for the financing of medical research