In a post this morning, Tyler Cowen has shown the tip of the iceberg of behind-the-scenes e-mails and speculation about what caused two things we (he, me, Aaron Carroll) noticed yesterday about health care costs in the 1980s. Below I’ll explain the rest.
First, Cowen asked why there was a big spike in U.S. health spending as a fraction of GDP in 1980. You can see it in the chart he posted here. Aaron redid the chart using his OECD data and did not reproduce that spike. So, for now we’re calling that spike an anomaly of the graph Cowen used. We don’t believe there was a huge spike in 1980.
However, I misunderstood what Cowen was asking. Normally when I look at a health care spending graph I don’t wonder a lot where huge increases come from because, well, health care costs are (almost) always hugely increasing and for lots of reasons. So, I immediately began to wonder about the leveling of health care costs (relative to GDP) in the early 1980s that is evident in both Cowen’s and Aaron’s charts (go look, if you haven’t yet). I actually thought it was this leveling that Cowen was interested in.
At first I speculated that the leveling off was due to a shift from inpatient to outpatient care, which began around that time (as Aaron documented here and I showed once before). It was my hypothesis (now proven wrong) that outpatient care could be both more profitable for providers and less costly for payers. That’s what Cowen quoted me writing (via e-mail) in his post.
But Aaron’s graphs don’t suggest a leveling off of costs as they shifted toward more outpatient. So, later I looked up GDP growth in the 80’s and relearned that it was large in 1982-1985 or so. Thus, I think the leveling off of heath care spending per GDP in the early 1980s is a GDP phenomenon, not a health care phenomenon. GDP growth really matters. If only we could keep it high enough to make health care cost growth look reasonable we’d be in fine shape. Good luck.
That explains the data and why none of my reading on the history of U.S. health care has revealed a good reason that costs should have leveled off in the early 80s. They didn’t. (Medicare’s hospital prospective payment system was the big change in the early 1980s but it began in 83 and was phased in over several years. It really didn’t affect spending until later in the decade, 1987 or so.) Moreover, the high GDP growth also explains the leveling of insurance premiums a bit later, despite health cost growth. That’s an underwriting cycle effect. Insurers had made plenty on returns on investment, had sufficient capital, and didn’t need to increase premiums to cover costs. (I used to be skeptical of this, but I’ve seen enough work on the underwriting cycle to believe it happens.)
There, the early 1980s, all explained. Are we done yet?