Read on Twitter
More NBER Live Tweeting -- @GottliebEcon preventing on healthcare spending and physician earnings. (Sorry I skipped Maura's excellent paper because people kept sliding into my Zoom DMs).
This paper is about how payments from government policy pass-through to physician earnings. Point is: PF always asks about how taxes and transfers impact inequality. But the real impact of govt. policy on inequality comes from how docs are reimbursed!
They link physician income (from tax filings) w/ govt. reimbursement as well as work hours. Physicians get wage income, but also salary and business income so you need to be careful about how you measure earnings.
Basic fact 1: physician income has a mean of $343,600. It's very steep in your 30s, peaks around age 50. Overall mean is about 100K higher than survey data suggested, only 68% of income is reported on W-2, 30% comes from business income.
Half of physicians are in the top 2% of income quantiles and only 1/4 are not in the top 5% of income quantiles. Physicians are rich. Neurosurgeons earn a ridiculous amount, about 5 times family practice physicians, peaking at $1 million / yr around age 50.
Incidentally, a motivation that Josh didn't give for this paper but that is implicit -- if you want to know why the US spends so much more on healthcare than other countries, physician wages are an important part of that story.
The highest earning states for physicians are not the highest earning states for lawyers. Physicians earn the most in the middle of the country.
Okay, now let's turn to the government's roll in all of this. Physicians offer in smaller firms and often residual claimants of firm earnings. This paper looks at the Medicaid Fee Bump in 2013-2014. Runs a diff-in-diff comparing PCPs in different specialties.
Rather, not comparing different PCPs but PCPs vs. medicine sub-specialists. After the Medicaid fee bump, the relative income of PCPs increased by 5%.
The estimates imply that the extra income for PCPs is half of federal spending on the fee bump. In other words, the incidence of the fee bump on physicians is about 50%.
The effect comes entirely from privately employed physicians, not physicians employed by the government. The pass-through is about three times as large for self-employed physicians relative to other privately employed physicians.
Large firms, or firms diversified among specialties share risks. Small firms with only one type of specialist experience the biggest bumps.
Incidentally, a paper I love on this topic is: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4511963/
Back to this paper: physicians are near the top of the income distribution and govt. payments have a large impact on physician incomes. Physician earnings are about 9% of total health expenditures.
How much could be saved by reducing physician spending? If we think of lawyers as a reasonable outside option, physicians earn substantially more, but also work about 5 more hours per week.
In PDV terms accounting for training, physicians earn about 3 million more over their life. 1.2 million accounted for by working more hours. Tuition is negligible compared to this. But 1.7 million in addition income is unexplained.
Are physicians just higher "ability" than lawyers? Or maybe lawyers just have such a pleasant just that they earn less due to compensating variation? Seems not all that likely.
If we cut all physician incomes to the lawyer average (adjusting for hours worked), we'd say $59B, 1.8% of health spending. If we cut specialities with a positive residual conditional on hours and training down to the regression line, we save $23B.
This is an excellent descriptive paper with fascinating new data. It raises many questions. In my view, the foremost among them is -- what bad stuff, if anything, might happen if we pay physicians less?
This question needs to be considered in the context that current physician wages are not first-best labor market equilibria, but heavily influenced by govt. payments and other distortions.
Before editorializing more, let me tweet about Doug Staiger's discussion. Point #1: this is unbelievable data. For the first time, we know what doctors actually earn on a national basis (a lot!) and don't have to rely on survey data.
Tax data solves the underreporting of business income which comes from survey data (e.g. American Community Survey). I wonder if similar issues arise in other countries, or if data is more comprehensive in places with single-payer (my guess).
Doug points out there is a huge bug in the CPS data on number of physicians that he's been harassing them to fix for months and that they still haven't fixed (i.e. in contrast to this paper which uses administrative rather than survey data).
This paper suggests govt. has a lot of price-setting / monopsony power in the short-run. Is this a more competitive labor market in the long-run? Doug suggests using income shock as a first stage to infer the long-run supply elasticity.
Doug has a minute or two left, "REALLY??"
Doug does some new analysis with the @GottliebEcon data -- Medicare physician spending per enrollee is pretty highly correlated with average physician income.
Now for a second discussion from @omzidar . BTW the original paper is joint w/ Maria Polyakova, Kevin Rinz, Hugh Shiplett, Victoria Udalova.
@omzidar points out that among the top earning privately owned firms, physicians are at the top of the list. Emphasizes that we should think about supply and demand (this is going to be a tough sell at the National Bureau of Economic Research).
Starts by thinking about the physical asset market. There is a rental market and an asset market. What are the key elasticities and quantities to understand this market? Stock of physician capital, depreciation, new flows (residents, immigrants, retraining).
Now want demand and supply shifters for the demand and supply of local physician services. Medicaid fees. What else? I got a little confused thinking about this but definitely agree with Owen that this is the direction to head to think about counterfactuals.