Pursuit of Truthiness

my gut tells me I know economics

We Need Better Experts- and a Better Public

leave a comment »

Experts are often wrong. Non-experts are even more often wrong.

These two statements are both true and important but it is hard to keep both of them in mind at the same time. Experts and members of the public both misunderstand this, and each ‘side’ can err by being either too humble or too arrogant:

ExpertFailureModes

COVID has provided many examples of all 4 failures. The expert institutions that were supposed to handle this best, the WHO and CDC, kept being arrogantly wrong thoughout the first quarter of 2020- saying that there was no human-to-human transmission, that travel bans were unnecessary, that masks don’t work, and in the case of CDC producing a botched test kit while preventing the use of alternative tests. Throughout this time, much of the public, and even experts outside the institutions, humbly deferred.

At the same time, many people who had figured out what was really happening were too humble to speak out, or too scared, or simply didn’t think of it. In the second quarter of 2020, expert institutions gradually figured things out, but much of the public still arrogantly dismisses their advice.

There are a few reasons this isn’t easy to get right.

Generalizability- Sure, in early 2020 on COVID you would have been better off listening to the Bay Area technologists I follow on Twitter than to the WHO and CDC- but is that generally the best way to get health advice? How many people are capable of listening to both groups and evaluating their specific arguments to figure out who is right?

Bias- experts really are biased, but it isn’t always obvious in which directions this matters, and members of the public have their own biases. For instance, are the public health agencies overstating the risk of COVID in order to get more funding and power, or understating it so as not to embarrass the governments that fund them? Both sounds plausible in the abstract, and you could say either in order to justify your own biases. In this case I do think they understated risks due to political pressure, but its also possible they simply weren’t as smart as we, or they, thought they were.

Expert, compared to who?- For quasi-experts, its important to keep in mind what your audience knows and where their information will come from if not from you. Many economists get annoyed and don’t answer when regular people ask them about the stock market- “I’m not an expert, that’s not what economics is really about”- but I think this is a bad response. We’re not true experts, but we should know much more than the average person, and be glad for the opportunity to share the basics (save through your 401(k) and put it in low-fee diversified index funds) and point them to the true experts. On the other hand, if you think your quasi-expertise means you can beat the market day-trading you’re likely to have a bad time.

How can members of the public know who are the real relevant experts? This is often far from obvious. Should I trust epidemiologists at the CDC more or less than those at universities? When evaluating potential treatments should I most trust virologists, epidemiologists, medical doctors, or someone else? If I want to know how COVID will affect the economy should I ask epidmiologists, economists, or someone else? If economists, who or which subfield? For forecasting COVID cases, is the relevant expertise domain knowledge like epidemiology or is it general forecasting ability?

Getting experts to be less biased and to have the appropriate level of confidence in their abilities and predictions is vital. So is getting members of the public to know who the relevant experts are, and how much to defer to their judgement in various situations- yet none of this is explicitly taught.

A parting thought, paraphrasing Garett Jones‘ twist on an old William Buckley quip- “I’d rather be governed by the first 2,000 names in the Boston phone book than by the faculty of Harvard, but I’d rather be governed by the faculty of MIT than either”

Related reading:

Inadequate Equilibria, Eliezer Yudkowsy is all about this issue: “the single most obvious notion that correct contrarians grasp, and that people who have vastly overestimated their own competence don’t realize: It takes far less work to identify the correct expert in a pre-existing dispute between experts, than to make an original contribution to any field that is remotely healthy.”

“inside a civilization that is often tremendously broken on a systemic level, finding a contrarian expert seeming to shine against an untrustworthy background is nowhere remotely near as difficult as becoming that expert yourself. It’s the difference between picking which of four runners is most likely to win a fifty-kilometer race, and winning a fifty-kilometer race yourself. Distinguishing a correct contrarian isn’t easy in absolute terms. You are still trying to be better than the mainstream in deciding who to trust. For many people, yes, an attempt to identify contrarian experts ends with them trusting faith healers over traditional medicine. But it’s still in the range of things that amateurs can do with a reasonable effort, if they’ve picked up on unusually good epistemology from one source or another.”

Myth of the Rational Voter, Brian Caplan: Argues that the public is rationally ignorant of issues (like public policy and politics) where more knowledge would not personally benefit them. But also shows that experts have systematically different political views than the general public, and showcases a statistical method to “correct” for this and show  what economists would think on various policy issues if their income level and general political views matched that of a typical person.

The Lancet on Medicare for All: Optimistic Yet Depressing

leave a comment »

This Lancet article gives an optimistic but plausible take on what would happen to health spending if we expanded the current version of Medicare to cover all Americans: a 13% drop in health spending.

Why Optimistic

The big problem I see in the article is that it claims to be evaluating what would happen if we passed the Sanders Medicare for All proposal, which involves changing current Medicare in a way that would make it much more expensive- removing coinsurance and deductibles. Removing cost-sharing would increase utilization for most everyone, not just people who are currently uninsured or “underinsured” (which they count at a combined 79 million); studies on Medigap insurance suggest this could raise spending 33% relative to standard Medicare. The Lancet study ignores this, but it would turn their predicted 13% spending reduction into a nearly 20% spending increase.

Why Depressing

Still, I like having a study like this around, as it gives a clear presentation of what one version of Medicare for All might look like and where various savings and costs might come from. The depressing part is that even if all of their optimistic projections are true, and if a giant reform like Medicare for All actually happened, America would still have by far the highest health care spending in the world, with at-best average results. The latest OECD data shows the US spending 16.9% of GDP on healthcare, with Switzerland in a distant second place spending 12.2%. If Medicare for All, the biggest reform of the US health care system since at least 1965 and possibly ever, actually worked as predicted here, we would be spending 14.7% of GDP on health care- still far more than Switzerland or anywhere else (which makes sense as Medicare is costly by international standards). The article doesn’t put their mortality estimates into life expectancy terms, but the 68000 fewer annual deaths they predict represents a 2.4% drop; a 2.4% increase in life expectancy would bring us from 78.6 years to 80.5 years, still on the low end for rich countries.

This shows how our problems are bigger than just insurance and health financing, but even ‘just’ fixing the financing part (either with M4A or any other major reform) is unlikely.

Why It Won’t Happen

The authors argue that Medicare for all is politically feasible because, under their somewhat optimistic assumptions, the average American saves some money, while the average doctor and hospital roughly breaks even (Medicare generally pays less than private insurance but more than Medicaid and charity care), with the savings coming from lower administrative costs, lower fraud, and lower pharmaceutical prices. But I think this misreads the politics; while the Median Voter Theorem can be a good first approximation, the US political system tends to favor the well-off, special interests, and those who trying to maintain the status quo rather than change it. The paper itself makes clear who some of the losers from Medicare for All would be: private insurers, pharmaceutical companies, doctors and hospitals with a higher-than-average share of privately insured patients, and most people who spend a less-than-average share of their income on health care (which higher-income people tend to do; they also vote at higher rates). Further, “936,000 administrative positions and 746,600 positions in the health-care insurance industry are estimated to become redundant”. Many influential groups would suffer concentrated costs that would motivate dedicated lobbying effort against, while the largest benefits would be concentrated among the ~10% of Americans who are uninsured, who tend to be relatively poor. The ACA was a much smaller reform that did much more to buy off or otherwise placate most of these interest groups, and it still barely passed and lead to a huge political backlash, a lesson that at least some major relevant political figures took to heart.

Written by James Bailey

February 18, 2020 at 2:03 pm

Which Countries Have Single-Payer?

with 2 comments

This seems like the kind of question that should have a straightforward, generally agreed-upon answer. But right now, it doesn’t.

I can only find a single attempt to answer it, from a random blogger whose original post describing the methods seems to have disappeared. This should really be a job for a big respected organization like The WHO or OECD, but barring that, I can be the next random blogger to update the answer.

There are two challenges to getting a good answer to “which countries have single-payer?”: choosing a good operational definition of “single-payer”, and then researching every country’s health-care system to see if it qualifies.

First, the definition. Literal single-payer would mean there is a single entity, presumably a national government, that pays 100% of health care expenses; but this would imply no countries currently have single-payer, since every country has some out-of-pocket spending:

OECD Chart: Health spending, Out-of-pocket, % of health spending, Annual, 2016

At the other extreme, a broad definition would count any country with universal government-sponsored insurance, even if it covers so little that most spending still comes from individuals or supplemental private insurance. Other important questions are whether any private insurance can exist, and whether all spending has to go through the national government or if many states/provinces operating their own insurance systems still counts as “single”.

Some day I will state 5+ alternate definitions of single-payer and do the deep institutional research to say whether each of the 188+ countries qualifies based on each of the 5+ definitions, the sort of deep dive that was done for 12 countries here. But today, you aren’t getting the thorough answer, you’re getting the quick answer, because today I decided its ridiculous that no one has provided any real answer.

So my working definition of single-payer, which doesn’t make the most sense but has the great virtue that it can be calculated quickly from existing data, is:

Single-payer: A country has a “single-payer” health system if at least 75% of health spending is done through government.

Why 75%? I originally thought 90% made sense, then looked at the WHO health expenditure data and saw that this would only count 6 countries. In the archetypical single-payer system, the United Kingdom, only 79% of expenditure is from government, and so 75% is the lowest round-is number that will count the UK. Even this doesn’t count Canada (69%), but I can’t bring myself to put the bar below 70. If you want to make your own map with a different cutoff, the Excel file with 2016 government spending as a % of all health spending is here.

Single-Payer Health Care Map

SinglePayerMap

Source: Calculated from 2016 WHO Global Health Expenditure Database

Update: The WHO numbers for government spending do not include compulsory social insurance programs that many people would consider part of government spending (see table 7.2 here for an explanation of these and other spending categories; basically “social insurance” is tied to individual contributions to a fund, like in Medicare, though in the case of German sickness funds this is really stretching the ‘single’ in single-payer). If I add “social insurance” and “government schemes” together, several more countries hit 75%+ of spending:

Alternate Single-Payer Map (Includes Social Insurance)

SinglePayerCountingSocial

Source: Calculated from 2016 WHO Global Health Expenditure Database

It’s worth repeating one more time, these are preliminary numbers made using a definition that prioritizes ease of calculation over accuracy, and I encourage others to make their own maps, especially if you have the time or resources to study the institutional details of every country.

Written by James Bailey

October 16, 2019 at 10:14 am

Who gets health insurance from their employer?

I’ll be writing at some other sites as I do a deep dive into data on health insurance and health care that might interest a wider audience.

The first post in this series, on what kinds of people get employer health insurance and what kinds of jobs offer it, is up at Medium.

 

Written by James Bailey

July 26, 2019 at 3:50 pm

Why US Health Spending is So High: The Beginnings of an Answer

with 2 comments

The US spends twice as much on health care as the average developed country, and three times as much as a percentage of GDP as we did in 1965, but our life expectancy is below average for a developed country. These facts are a regular refrain for health economists like me, and are often given as a motivation for our research. But one year ago I realized I had no good answer for why US health spending is so high, and no one else seemed to either. I’m still not close to finishing a book about it, but I’m finally starting to feel like I have the beginnings of an answer.

First, that even by developed country standards we are quite rich, and richer countries spend more on care.

Second, the biggest single mechanism driving this higher spending seems to be the particular design of Medicare reimbursement, especially for outpatient care. That’s the high spending.

Why don’t we seem to get much for it? Here, I think the thing is simply that life expectancy is not primarily about medical care, and that the factors driving out life expectancy down are mostly or entirely outside the medical system (obesity, overdoses, accidents, violence, et c).

In fact it is possible, though not firmly established, that our medical system really is the best in the world; that if Americans got the English or Japanese medical system our life expectancy would be even shorter, and that if the English or Japanese got the American medical system their life expectancy would be even longer. The world-beating 86 year life expectancy of Asian-Americans, a group that gets US medical care while mostly avoiding the worst of US health behavior, is suggestive of this.

Though, even if the US medical system is the best in the world, I still think it is unlikely that quality is ahead of other countries by anything like as much as spending is. We might spend a lot more and only have a slightly better quality medical system because of classic diminishing returns, or inefficiency, or because part of our high spending is on developing and implementing new technology that everyone else quickly adopts once it is cheap.

In short: it shouldn’t seem mysterious that a very rich country, especially one with insurance designed like Medicare, spends a lot on health care; and it shouldn’t seem mysterious that this high spending is unable to fully overcome the life-expectancy-dropping effects of the American lifestyle.

What does still seem mysterious to me, but I hope won’t in another year: To what extent can high private spending be explained by the influence of Medicare? Does any other rich country have an insurance design resembling Medicare Part B, and if so why has it not yielded similarly high spending? Where exactly does our higher spending go to? In particular, how much is price vs quantity vs quality today vs new tech dev vs pure inefficiency? How would the US medical system compare to other countries in a ranking of value added (contribution to life expectancy)- rather than the usual rankings that measure only inputs or outcomes?

Written by James Bailey

April 23, 2019 at 7:14 am

US Health Spending Growth: Its All Outpatient

with one comment

I previously pointed out that while we spend twice the OECD average on health care overall, we spend three times the OECD average on outpatient care.

Now I want to show you the history and give one theory that explains why. First, we’ve become much more of an extreme outlier on outpatient spending since 1987 (when OECD data on the US first starts distinguishing between inpatient and outpatient spending), as our outpatient spending has nearly doubled as a % of GDP while other countries are mostly flat:

Source: customized chart using stats.oecd.org

What is going on here? One big hint is that over the same time period, our inpatient spending is basically flat as a % of GDP, and remains in line with other countries:

Source: customized chart using stats.oecd.org

There’s a lot going on but if I had to pick a single explanation, it would be the design of Medicare and its 1983 Prospective Payment reform. Basically, the way Medicare was originally designed (cost-based reimbursement that gives providers a big incentive to do more) led to huge growth in spending following the introduction of Medicare in 1965.

The 1983 Prospective Payment reform reduced the incentives to over-treat and led to slower spending growth- but it only applied to Medicare Part A, which covers inpatient treatment. Despite a 1992 reform that changed prices around, for Medicare Part B (which covers outpatient care) reimbursements still seem to work on the same essential principle as in 1965- the more treatments you do and the more costly the treatments you do, the more you get paid. Certainly Part B has been the fastest-growing part of Medicare spending since the 1980s, and its direct contribution to spending together with spillover effects on pricing and reimbursement norms for private insurance could explain most of our spending increase since the 1980s.

Still looking for annual data that breaks out Part A vs B back to 1965

 

 

Written by James Bailey

April 22, 2019 at 5:20 pm

What Happened to U.S. Health Spending After 1980? It Slowed Down.

with one comment

In a NYT Upshot post titled Medical Mystery: Something Happened to U.S. Health Spending After 1980, Austin Frakt shows that US health spending pulled ahead of other countries in the 1980’s, and tries to explain this by looking for other ways the US was exceptional in the ’80s.

OECD

But looking at the RATE of spending growth rather than the absolute amount of spending tells a very different story. The US got to the top of health spending by rapid growth between 1960 and 1980; US spending growth actually slowed in the 1980s, and has slowed further since.

Average annual spending growth during 1960’s: 8.3%

Average annual spending growth during 1970’s: 6.0%

Average annual spending growth 1980’s: 5.4%

Real NHEA increase

So what happened before 1980? My main theory is, Medicare. Started in 1965 with cost-based reimbursement that incentivized hospitals to spend more, transitioned to a reformed payment scheme (Part A Prospective Payment) starting in 1983.

Average annual spending growth 1965-1983: 6.3%

Average annual spending growth 1984-2016: 4.1%

Numbers are my calculations from NHEA data and CPI. The most obvious counter to the Medicare theory is clear in the graph above: growth was even faster in the early ’60s before Medicare. Of course, I’m far from the first to point out the potential importance of Medicare or the 1983 reform. See Amy Finkelstein’s papers for the causal arguments (Introduction of Medicare, 1983 Reform), or this paper (summary) from CMS researchers that gives the best overview of historical US spending I’ve seen. I particularly like their table of US health spending ‘eras’:

US Health Spending Eras

US Health Spending Since 1960

with 2 comments

As a health economist, I’m legally obligated to remind you that US health spending has more than tripled as a proportion of GDP since 1960. But while the “proportion of GDP” statistic is the one we share most often, I’m not sure its the most interesting. Total US Real GDP per capita has also tripled since 1960, implying that real US health spending per capita is actually 9 times the level of 1960.

The National Health Expenditure Accounts report that health spending was $10,739 per American in 2017. Their historical data goes back to 1960. Oddly, they don’t provide real inflation-adjusted spending figures themselves, so I calculated them using historical NHEA data and the CPI; you can get my spreadsheet here. I estimate that real per capita health spending in 1960 was $1212, just over 1/9th of current spending.

The sources of health spending have also changed dramatically since 1960, as you can see in these charts I made from NHEA data:

1960 spending pie

2017 spending pie

Spending has shifted from direct payments out of the pockets of patients/consumers, and toward insurance, especially public insurance (Medicare and Medicaid). Real out-of-pocket spending per American has merely doubled since 1960, while private insurance spending is 14 times its previous level, and public insurance is 58 times higher.

Why has spending increased so much? More coming soon…

Written by James Bailey

April 9, 2019 at 4:12 pm

Is the US Healthcare system responsible for our low life expectancy?

with 2 comments

No. For life expectancy in developed countries, individual differences matter more than health systems.

The US is known for having below-average life expectancy for a developed country. This is true, but all generalizations about the US are wrong in important ways because the country is so huge and diverse (can I claim this as Bailey’s law? Or did someone else in our huge, diverse country already take credit?).

First there is huge variation across states, with average life expectancy ranging from at totally respectable 81.3 years in Hawaii to 75 years in Mississippi (putting it in the tier of much poorer countries like Iran, Brazil, or Sri Lanka).

Its common to blame low US life expectancy on “the US healthcare system”, but this can’t be right. Are the 12+ states with life expectancy of 80+ years not part of the real US health care system? Health policy is largely set at the federal level. Perhaps the exceptions are important, or for some other reason (more/better doctors?) the healthcare system really is better in those states. There is no “US healthcare system”, but rather 50+ US healthcare systems. Possible.

But there is also huge variation by race– from 74.6 years for African-Americans to 86.5 years for Asian-Americans. If Asian-Americans were their own country, they would have by far the longest life expectancy of any country in the world (currently Japan leads at 83.7). African-Americans, by contrast, would be down in the tier of middle-income countries, just as Mississippi was (though their life expectancy is still well above that of any sub-saharan African country, including relatively prosperous ones like South Africa and Botswana). Its also not a simple story of income differences- note that average Latino life expectancy in the US is 82.8 years, 3.9 years above whites (the Hispanic Paradox).

When we take advantage of the race-by-state cross-tabulations, we can really start to tell which stories about life expectancy have more merit. Asian-Americans maintain world-class life expectancies even while living in states near the bottom of the US rankings, like Alabama and Louisiana. If the problem were really about the health systems in those states being much worse than average, this would be hard to explain. One explanation I’ve heard from several other health economists is that the US healthcare system is actually good for most people, but bad for the ~20% who are poor and/or uninsured. This 20% does dramatically worse, skewing our overall average life expectancy downwards; and they are more concentrated in certain states, which explains the state-to-state variation. This sounds plausible and mostly matches the state variation, though it would mean the Hispanic Paradox remains a paradox; also, in Massachusetts blacks as well as Hispanics and Asians now outlive whites.

What explanations remain? Differences in health behaviors (e.g. diet, exercise, drug and alcohol use, medication adherence, accidents) are looking good. Genetics are possible too. Both could also explain why women live five years longer than men. My ad-hoc analysis here matches up with a lot of the more serious life expectancy research, for instance the research behind this tool you can use to calculate your personal life expectancy. Note how almost all of their questions are about your daily behaviors, and almost none are about interactions with the medical system.

Written by James Bailey

February 16, 2019 at 3:56 pm

The ACA and Entrepreneurship

leave a comment »

In a newly published paperDhaval Dave and I tried to assess how the Affordable Care Act affected entrepreneurship. Many people have been speculating about this, but very few have actually tried to measure it. Part of the reason for this is that the appropriate data only recently became available; most earlier work, including my dissertation, had to focus on the relatively minor parts of the ACA that started earlier, while the biggest provisions didn’t kick in until 2014.

So, did the ACA help or hinder entrepreneurship? In short, it probably helped, but with huge heterogeneity. In our paper we compare entrepreneurship rates for people in their early 60’s (who might benefit from the availability of individual insurance through the ACA) with a “control group” of people in their late 60’s (who are eligible for Medicare and presumably less affected by the ACA). We find that the ACA led to a 3-4% increase in self-employment for people in their early 60’s.

So, should we take that estimate and extrapolate it to everyone, and use that to imply that the ACA created a million businesses? I don’t think so. I expect that the biggest positive effect of the ACA was for exactly the group we studied- people in their early ’60s. Both because they are old enough to have substantial average health costs and health insurance premiums  (so they will factor health insurance into their decisions more strongly than younger people) and because of the community rating provisions of the ACA (which generally reduced individual premiums for older people while raising them for younger people). The other new papers on the ACA and entrepreneurship use identification strategies that let them study adults of most ages, and they generally find no effect or smaller positive effects.

If I do more work on the ACA specifically, it will be to probe the effects on other subgroups that I expect to have particularly positive effects (e.g. less healthy people) or negative effects (e..g young, health people). Alternatively, we can start to look to post-ACA data and assess to what extent entrepreneurship lock remains a problem even after the full ACA implementation.

Written by James Bailey

September 25, 2018 at 2:04 pm