The controversy over the supply and demand of doctors is not dying down. The government concluded that as of 2040 at least 5,015 doctors will be lacking, but the medical community offered the opposite analysis that more than 10,000 doctors could be oversupplied at the same point in time.
The key is which formula is used to envision the future.
At a joint-planning seminar titled "Problems and alternatives in the government's physician workforce projections" held at the Korean Medical Association Organization hall in Yongsan-gu, Seoul, on the 13th, the Korean Medical Association Organization's Research Institute for Healthcare Policy said, "Before increasing medical school enrollment, we need to rebuild the projection structure," and presented an alternative workforce projection model.
◇ Instead of "number of licenses," "amount of labor"… Reassessing the number of doctors with FTE
The core proposed by the association's research institute is projections based on FTE (full-time equivalent), which reflects doctors' actual workload. The idea is to look at doctors not by "how many" but by "how much they work."
The association set the average annual working hours per doctor at 2,302.6 hours. It recalculated the number of active doctors by converting 2,080 hours per year, based on 40 hours per week, into 1 FTE. As a result, the number of active doctors on an FTE basis was estimated at 154,601 in 2035 and 164,959 in 2040.
Medical demand was aligned to the same standard. Considering that medical aid beneficiaries use more medical services than National Health Insurance subscribers, which could inflate overall demand, the focus was on National Health Insurance utilization. The workload weights for inpatient and outpatient care applied the same 3.9-to-1 ratio used by the Physician Workforce Projection Committee, and a composite scenario was set that reflects changes in the future medical environment and health care policy.
As a result, the number of doctors needed on an FTE basis came to 142,844 in 2035 and 150,275 in 2040. When the composite scenario is applied, the required numbers decrease further to 140,634 and 146,992, respectively.
In sum, the calculation shows that in 2035 there could be an oversupply of at least 11,757 to as many as 13,967 doctors, and in 2040 an oversupply of at least 14,684 to as many as 17,967.
◇ Same future, different math… Where did they diverge?
The association's alternative projections directly overturn the conclusions of the government and the projection committee. The medical community believes this gap stems from differences in the assumptions and methodology that run through the projections.
A representative example is the method of calculating inpatient and outpatient workload. The projection committee, citing limits in data availability, used the ratio of inpatient to outpatient medical expenses (3.9 to 1) as workload weights.
On this, Park Jeong-hun, a senior researcher at the association's research institute, noted, "If you calculate workload based on medical expenses, you run into the problem that the high costs of tests and equipment included in surgeries or procedures during hospitalization are counted as doctors' labor."
The ARIMA (autoregressive integrated moving average) model used for projecting medical demand also came under scrutiny. The model predicts future demand based on time-series trends in past utilization, but structurally assumes that medical utilization increases endlessly.
Although adjustments were made by sex and age, criticism also arose that by using "total medical utilization" rather than "per-capita medical utilization," the projections failed to fully reflect the natural population decline that has already begun. In fact, the committee's data show that for men ages 60-64, inpatient days per person rise from five in 2024 to nine in 2050, and outpatient days rise from 16 to 34, more than doubling.
The period of data use was also cited as a problem. The longer long-term utilization data are applied as is, the steeper the increase becomes. In reality, the growth rate in inpatient days by type of institution averaged 11.8% from 2004 to 2010, but slowed sharply to 1.9% from 2010 to 2023.
Assumptions about "future changes in the medical environment" included in the committee's scenarios also became a focal point. The committee assumed that with AI adoption, physician productivity improves by 6% while working days decrease by 5%. Park said, "A reduction in working days is a matter of social consensus that the government cannot force," adding, "Applying it simultaneously in combination with a productivity-improvement scenario is conceptually unscientific."
◇ "Rather than rushing to conclusions, redo the math first"
The medical community argues that the pace toward the conclusion of expanding medical school enrollment should be moderated first.
Park said, "We should take sufficient time to build micro-level data sources and, based on them, project the physician workforce," adding, "We need to move away from relying on limited data and shift to a structure that reflects doctors' actual workload (FTE) and changes in productivity."
Considering that building FTE-related data takes time, a short-term alternative proposed was to use the procedure-specific physician workload included in the relative value scale.
They also argued that the projection methodology itself should be changed. Park explained, "Countries such as the United States and the Netherlands, which regularly project physician supply and demand, use micro-simulation models rather than ARIMA," adding, "Estimating demand and supply in FTE units while reflecting diverse variables is far more realistic."
Park said, "If you apply a single assumption without fully considering policy changes and shifts in the medical environment that could realistically occur, other distortions are bound to recur," adding, "The debate over the number of doctors is not ultimately about the numbers, but about how you design the structure of the calculation."