Back of the Envelope: Weak Competition Is Costing Us 250,000 Health Care Jobs


By John Sturm |

Policymakers are quick to consider the costs that uncompetitive markets impose on consumers: monopolies and oligopolies raise prices above the fair, competitive level. But competition policy isn’t just about prices; it’s also about jobs. When a firm raises prices, fewer consumers buy its goods or services. As a result, the firm produces less than it would in a competitive market and therefore needs fewer workers. These effects may play a significant role in U.S. labor markets, one surprisingly overlooked in a political climate so focused on jobs.

So, just how many jobs would exist in more competitive markets? To obtain a ballpark estimate, let’s focus on the U.S. hospital and outpatient care industry, which employs about 12.5 million people. Since 1990, the healthcare industry has been characterized by increased consolidation and rising profits. According to the American Hospital Association’s Annual Survey, hospitals had an average operating margin of 6.4 percent in 2014, and according to Sageworks (, a financial information company, outpatient facilities have been extremely profitable in recent years, with net profit margins around 14 percent.

Any increase in competition would reign in these high levels of profit. Of course, a decrease in profits could result from higher prices and lower quantity or lower prices and higher quantity, but information about the demand for healthcare—how much consumers would buy at different price levels—resolves this uncertainty. One can therefore translate the profit-reducing effect of competition policies into estimates of their impact on healthcare consumption.

A simple model[1] that formalizes this logic leads to the following result: The fractional increase in quantity of health services provided due to a change in the industry-wide profits of inpatient and outpatient facilities is equal to ( (1-P’) / (1-P) )^(-e), where P is the original profit margin, P’ is the new profit margin, and e is the elasticity of demand for health services, or the percentage change in quantity demanded per percentage change in price.

Recent work shows that a 1 percent decrease in the cost of healthcare will lead a typical, insured consumer to consume about 1.1 percent more health services. However, about two-thirds of U.S. healthcare spending corresponds to the small number of individuals who spend more than their out-of-pocket maximum[2] and so are insensitive to prices. We therefore assume that each 1 percent decrease in the cost of healthcare causes one-third of medical spending to increase by 1.1 percent and two-thirds of medical spending to be unchanged, so that total spending increases by 0.37 percent. This implies an elasticity of -0.37 percent.

We consider a scenario in which stronger competition reduces profits of both inpatient and outpatient facilities by half. Since inpatient and outpatient facilities each account for about half of expenditure, this would decrease the average, industry-level profit margin from around 10.2 percent to around 5.1 percent. Plugging into our model, such an increase in competition would result in the production of about 2.1 percent more services.

In theory, medical facilities could increase their output by investing in technology that replaces labor. While such a shift is possible, there is little reason to believe that increased output will accelerate it; the education and health services sector maintained a roughly constant labor share of income between 1987 and 2011, at around 85 percent, despite substantial expansion during that period. Should this relationship continue to hold, a 2.1 percent increase in the output of hospitals and doctors’ offices would correspond to an equal increase in employment, or about 250,000 jobs.

This estimate relies on a variety of assumptions. First, it assumes profits are just the product of quantity and the difference between price and unit cost. A modified model[3] which considers that firms have upfront costs, such as investments in expensive medical equipment, only strengthens the relationship between profits on quantity produced. Second, our simple profit function ignores that the cost of providing medical services may rise as demand increases. Indeed, higher demand could exacerbate the current shortage of doctors, allowing them to extract even higher salaries. Various policies—including expansion of U.S. residency funding, reform of cartel-like licensing boards, and loosening of regulation on nurse duties—could address this issue, increasing the supply of doctors without increasing wages.

Beyond assuming a certain profit structure, this estimate uses an elasticity of demand based on the spending of insured consumers; it ignores that cheaper health services will push down the cost of health insurance, so that more people buy health insurance and therefore consume more health services. This omission results in a more conservative estimate, but it avoids relying on assumptions about the competitiveness of the health insurance industry.

Setting aside the details, this back-of-the-envelope estimate illustrates a few general principles. First, competition policy does not exist solely to protect consumers from price hikes. When goods and services are cheap, consumers demand more of them, so firms employ more workers to produce higher quantities. Second, these effects can be substantial: stronger competition could create 250,000 jobs in healthcare alone. Finally, effective competition policy requires action at multiple nodes in a supply chain, so that efficiency gains flow through to consumers. Such reforms can fix broken markets, increasing efficiency and creating jobs while also—not rather than—redistributing income from wealthy shareholders to consumers and workers.

[1] Suppose net industry profits are given by π = (p – c) Q, where Q is the quantity produced, p is the unit price, and c is the unit cost. The net profit margin is the net industry profit divided by revenues:  P  =  π / (pQ)  =  (p – c) / p  =  1 – c/p

Taking a derivative:  dP / d ln(Q)  =  ( dQ / d ln(Q) ) ( dP / dQ )  =  Q ( c/p^2 ) ( dp/dQ )  =  c / ( p e )  =  (1 – P) / e. So d ln(Q) / dP  =  e / (1 – P)

Integrating,  ln(Q’) – ln(Q)  =  – e ( ln(1 – P’) – ln(1 – P) ). So  Q’/Q  =  ( (1 – P’) / (1 – P) )^-e

[2] We estimate this using summary statistics in Kowalski (2014). For our ballpark estimate, we set aside concerns about the external validity of this sample.

[3] Suppose net industry profits are  π = (p – c) Q – X, where is an upfront cost.

Then P  =  1 – c/p – X / (pQ).

Taking derivatives, we obtain a formula for the sensitivity of profits to quantity produced:

dP / d ln(Q)  =  (1 – P) / e  +  X / (pQ).

Since (1 – P) / e < 0, this implies that upfront costs reduce the sensitivity of profits to quantity produced—i.e. increase the sensitivity of quantity produced to profits—whenever (1 – P) / e  +  X / (pQ) < 0, or equivalently -(1 + e) X < c Q, which holds here, since e > -1.

John Sturm is a Research Associate at the Roosevelt Institute.