Estimating the Blood Supply Elasticity: Evidence from a Large Scale Benefit Scheme

In all industrialized countries, the market for blood donation is regulated by a zero nominal price, which results in supply and demand imbalances. Aiming to reduce such imbalances, public policies have to use non-price incentives such as gifts as tokens of appreciation, waiver of user fees at government facilities, and promotion campaigns such as blood drives (which are on-site blood collection events). 

In Portugal, starting from 2003, the government strictly enforced the collection of user fees, but regular blood donors received a waiver of such fees. However, from 2012, emergency care no longer qualified for the user-fee waiver. In other words, regular blood donors experienced a potential benefit from 2003, but this was reduced in 2012. I study the effect of these benefit changes on blood donations.

A donor’s decision depends on potential benefits, convenience of donation, and social customs. My study uses information in my data sets on all three components to estimate the blood supply elasticity. First, the changes in user-fee waiver have been well document. Second, blood drives made donation more convenient, and I have information of all drives in the same time period. Third, the social setting at which blood drives occurred (schools, churches, shopping mall) can be a proxy of social norms, and I have such data.

 User fees changed over time. These changes resulted in changes in benefits of being a regular blood donor. I use these variations to estimate the elasticity of blood supply with respect to benefits. I estimate supply elasticity by a difference-in-difference linear regression, and used county-level user fee variations to identify the elasticity.

 My actual implementation uses an instrumental variable estimation method. The number of blood drives and blood donation are determined endogenously. Hence, an estimation of blood supply elasticity with blood drives as an explanatory variable suffers from an endogeneity problem. However, an estimation of elasticity without blood drives as an explanatory variable suffers from an omitted variable bias.

 I solve this problem by instrumenting for blood drives. First, for each county, over the entire sample period, I calculate the proportion of drives that happened over weekends. Second, for each county and for each month, I calculate the total number of public holidays and weekend days. The instrument is the product of these two constructed variables. Furthermore, I estimate the elasticity of blood drives with respect to the user fee in a first-stage regression.

 A 1% increase in user fees increases blood donations by 0.052%. With blood drives omitted in the regression, the estimated elasticity would become only 0.04%. Moreover, the number of blood drives decreases as user fees increase. User fees and blood drives are substitutes for the promotion of donations.

 I present other results. Changes in user fees affect returning donors most. User fees have a stronger effect on donations at blood drives than at donation centers. Finally, men and women have similar supply elasticities, and young donors do not respond to user fees.