Decentralised Loan Repayment Programs and Long Run Health Outcomes

The paper makes the following assumptions to calculate the net cost of the policy:

• Each participating physician receives $120,000 in loan repayment, which is untaxed.
• The analysis covers the period 1995–2017, with 2006 — the midpoint of the rollout window — used as the representative program year for discounting purposes (the applied discounting rate is 3%).
• Health effects are assumed to persist for 4 years following program implementation, based on event-study estimates. That is, if program implementation is in Year 0, the health effects exist for Years 1-4.
• Eligible physicians are assumed to practice for at least 30 years, beginning around age 29–30 upon completing training and continuing through age 60.
• The combined marginal tax rate on physicians’ overall earnings is assumed to be 35%, reflecting a 30% federal rate — averaging the marginal rates applicable to married and single physicians in the sample — plus a 5% state tax rate.

Under these assumptions, the paper calculates the net cost of as the total cost of building the program’s full physician capacity less the fiscal externalities generated by the fully-built program, which are assumed to persist over a 4-year window.

The mechanical cost of the program is 54,234*120,000=$6.51 billion, where 54,234 is the number of physicians estimated to enter the treated counties over the 23-year period. This is treated as a one-time expenditure incurred in the representative year (2006), rather than tracking each treated county’s individual entry date or the actual within-obligation disbursement schedule.

The fiscal externality arising from Medicare reimbursement savings, driven by reduced ER visits and hospital stays, amount to $59.81*13,486,188=$806.6 million/year. Applying a 4-year discount annuity factor of 3.717 (at a 3% discount rate), the present value of these Medicare-related savings is 806.6M*3.717=$3 billion.

The fiscal externality arising from taxation of physicians’ overall earnings results from the program inducing trainees to choose primary care over higher-paying specialties at the outset of their careers, resulting in those physicians earning less (and therefore remitting less income tax) than they otherwise would have. The paper’s estimates suggest that the program results in 9,816 additional eligible primary care entrants over the 23-year period. The average earnings of eligible primary care specialties is $324,968, while the average earnings of ineligible specialties is $425,598 (2019 Doximity Compensation Report). Forgone earnings per induced entrant are therefore $425,598−$324,968=$100,630. Applying the annuity factor over a 30-year career span of a practicing physician, at a 3% discount rate, the present value comes out to be 9,81*100,630*0.35*19.60=$6.78B.

The net government cost is therefore $6.51B – $3.00B + $6.78B= $10.29 billion.

The MVPF then is $10.77B/$10.29B= 1.05.

These should be treated as lower-bound estimates, as the willingness-to-pay calculation captures only mortality reduction and omits the direct welfare value to patients of avoiding preventable hospital stays and ER visits, as well as health benefits accruing to the Medicaid population, both of which are also documented in the paper. While conceptualizing and monetizing the willingness to pay for avoiding these additional hospital-related events is not straightforward, a fuller accounting of these channels would likely push the MVPF higher. 

The paper also considers several alternative scenarios. First, if health effects are assumed to persist for 10 years — matching the horizon over which physician retention holds, rather than the 4-year window currently considered — the MVPF increases to 3.86. Second, if physicians’ own willingness to pay for the program transfer (valued at 50% of the transfer amount, following standard conventions for large induced transfers) is included in the numerator, the MVPF rises to 1.36 under the 4-year persistence window, and 4.37 under the 10-year window.