Health care prices do not play the role most believe

As I mentioned in two (long) posts on how comprehensive measures of household consumption or disposable income explain high US national health expenditures (NHE) rather well, I believe US health expenditures are overwhelming accounted for by higher volume of health consumption (the quantities of goods and services consumed in the health care sector).  I reject the popular notion that idiosyncratically high prices are what causes the US to spend so much more than other OECD countries and I reject the notion that rising health prices explain why US health expenditures have grown so much faster than income over the past few decades (at least).

This does not mean that prices play no role whatsoever in international comparisons of NHE.  It does not mean, for instance, that if we could somehow force U.S. physicians, nurses, and various other workers in the health care sector to work for the real remuneration paid to their counterparts in, say, Brazil with equal productivity that the price of medical care, especially more labor intensive categories, would not plummet (presumably real NHE could decrease if volume were hold constant…. not a very safe assumption imo).  It means that the reason health expenditures have accounted for an increasing share of our consumption (or income) over time has little to do with prices increasing relative to income and a great deal to do with the volume of health goods & services consumed rising at a faster rate than overall consumption (or income) per capita.

animated_nhe_consumption_share_by_country

Much the same goes for international price comparisons in cross-sectional analysis.  The cost of an internationally equivalent basket of health care (both goods and services) rises a little faster than the cost of a broader, economy-wide/GDP, basket of goods and services with rising income levels, but it does not rise all that much faster with respect to income (actually somewhat less overall in the long run).

In order to significantly explain the presumably large US residual in NHE per capita (obviously I don’t quite agree with this perspective) there must be a large residual in the actual relative price levels; we simply do not find any evidence of this with the best available evidence.   With better data US residual is either slightly above trend (by no means an “outlier”) or below trend, depending on the precise predictor (e.g., GDP, AIC, average wages, etc) and the basket of goods and services employed.  In no case do I find evidence of a large departure from trend in any broad, highly impactful, index of health price levels that would suggest that high US NHE isn’t overwhelmingly explained by high volumes of (real) health care consumption.  More generally, the cross-sectional and time series data are actually quite consistent with each other, i.e., the issues that drive (US) expenditures up in time series are quite consistent with and related to the what we observe in cross sectional analysis between countries.


First, a quick and dirty cross-sectional comparison published by the OECD using their latest data (not yet available online, sadly).

screenshot_2097.png

screenshot_2096.png

screenshot_2112.png
source

According to this data US health expenditures are ~2.4 times (~240) greater than the OECD average using economy-wide PPPs whereas they are  “only” ~2.1 times (~210) greater if we use the more appropriate, carefully constructed, health PPPs to convert US domestic expenditures into internationally comparable units (roughly the volume of health we actually consume per capita).

The relative differences between the two US PPPs are quite modest.   A given amount of US dollars within the US only goes a little further if spent throughout the entire economy (GDP) than if spent on health if we compare the differences in how far these expenditures go throughout the OECD.   My eyeball estimate of these figures suggests health is only about 14% more expensive according to the differences in the volumes they derived there.  By way of reference the real volumes of GDP per capita was about 37% higher and AIC about 47% higher in the US in 2014 (note: if you look carefully at the plot, figure 4, you might also notice there is a pretty strong correlation between relative price levels and how wealthy these countries are).

Note: Sector specific PPPs for 2014, 2011, 2008, and 2005 from OECD if you’re interested in checking this out for yourself.  In no case will you find US is 100% more expensive overall as implied by a handful of shoddy, but widely cited, statistics.


I can estimate this more precisely and do further analysis using the 2014 PPPs.

rcafdm_0717_01_health_price_by_aic.png

There is at least modest correlation between overall health prices and AIC.  Though I am not the only one that finds AIC a more useful way to understand price levels and volume of health expenditures than GDP (see this plot, for example), we still obtain fairly similar results with GDP.

rcafdm_0717_02_health_price_by_gdp.png

 


These relationships tighten up considerably if we restrict the PPP to hospital services (which lean more on domestic labor than the broader health PPP)

rcafdm_0717_03_hospital_services_price_by_aic.png

Here, as with the overall health PPP, US is actually below trend.  The price differences appear somewhat more consistent with the handful of sloppily produced, yet oft cited,  prices produced for surgical procedures by IFHP insofar as you might be able to see how the OECD estimates could be fairly accurate and the IFHP may not be exactly fabricated (insofar as there are several very real methodological issues with their methods, their comps are almost always poorer countries, potential for biased selection, etc etc).

By comparing these to the GDP (economy-wide) PPP, rather than to, say, individual consumption or AIC, I am likely inflating the GDP correlation some relative to AIC, but for completeness:

rcafdm_0717_04_hospital_services_price_by_gdp.png

US is modestly above trend for GDP, though hardly what I would call an outlier.   You might also note that Canada is similarly above trend despite the fact that they are one of the few countries with single payer.  Given potential issues with measurement error, unmeasured differences in quality,  likely a not insubstantial amount of volatility year-to-year, and so on I would not get terribly excited about any modest residual here, let alone be willing to confidently ascribe it to differences between health systems.

It is abundantly clear that hospital services are quite a bit more dear in rich countries than poor countries even though the differences in the actual overall health PPPs are quite modest (the weights attached to them and rising productivity in the less labor intensive areas likely explains this).   This is quite consistent with research on the consequences of Baumol’s effect on health expenditures.

screenshot_2100.png

source

The (ball-park) central estimate seems to be somewhere around 0.6, which, if correct, implies that health prices only grow about 60% as fast as productivity (or equivalent) increases, which implies that if volume per capita were held roughly equal we would see the health share declining relative to income levels (price * volume = expenditure).

Health prices can still grow faster than other price levels and contribute to overall inflation even while not (usually) growing faster than income, remuneration, etc.

screenshot_2101.png

source

 

Hashing out Baumol

Although cost disease does not, in and of itself, drive NHE faster than income growth, it is still important to understand insofar as (1) more labor intensive areas of health care are likely considerably more subject to said effect than the overall index suggests  (2) if the true Baumol variable were actually ~zero, if productivity gains in health equaled that of the rest of the economy, the elasticity of NHE with respect to income might be somewhat smaller (i.e, assuming volume doesn’t increase correspondingly in said counterfactual!) and (3) it is likely that most of the stable price variances we see between countries in these more labor intensive categories are a product differences in income levels, not to so much to differences in health policy (most of the relevant market distortions are shared quite widely across reasonably developed countries).

Potential differences in quality aside, it is very likely the reason less developed countries (e.g., Russia, Mexico, etc) generally have relatively “cheap” health care (especially hospital services) is because they are less productive overall, real wages are lower, and ~all goods are relatively more expensive as compared to services, not because these less developed countries have better health systems that lead to greater efficiency or because they have more monopsony powers to dramatically reduce the cost of relevant factor inputs.

rcafdm_0717_05_total_services_and_goods_by_gdp.png

In higher income countries, like the US, overall economic productivity is much higher, workers command higher wages, and skilled and motivated labor is particularly scarce.   Services are generally not tradable and tend not to enjoy the same sorts of gains in productivity as seen in other areas of the economy (manufacturing, agriculture, etc) so they tend to rise with respect to goods in both the spatial (towards high income countries) and temporal (as countries move towards higher income levels) dimension.

We see these same patterns play out in more disaggregated expenditure categories.

Like education:

rcafdm_0717_06_education_by_gdp.png

Also note that the US’s high level of expenditures aren’t likely to be explained solely by high tertiary costs.  We find conservative estimates of per pupil expenditures (not price levels) for primary and secondary schools track rather well with GDP despite the fact that it’s unlikely to be well explained by higher quality of teachers in rich countries.

Collective services (e.g., police, courts, etc):

rcafdm_0717_07_collective_services_by_gdp.png

Consumer services:

rcafdm_0717_11_consumer_services_by_gdp.png

It is also likely some related issues have been found in construction costs and other areas that are predominantly non-tradable and/or fundamentally cannot match the productivity gains found in other areas of the economy.

These differences in service costs are also quite well correlated with each other.

For instance, hospital services vs education:

rcafdm_0717_12_hospital_services_by_education.png

and to, much much lesser degree, overall health expenditures (which leans much more heavily on tradable & manufactured goods etc than the hospital services subset)

rcafdm_0717_13_health_by_education.png

In short:

  1. much of the observed differences in health prices are likely to be explained by non-health system factors
  2. it correlates quite well with national income and consumption levels
  3. there are sound theoretical reasons to expect this
  4. the US residual is not especially large (very much contrary to popular belief)
  5. the impact of prices differences on real expenditures is generally vastly over-rated (it only marginally reduces the differences in volumes associated with income levels)

Domestic time series

The limited role of price found in health PPPs and the like between countries or in OECD analyses is also consistent with domestic time series studies in the US and other countries.  For instance between 1940 and 1990 estimates of the increase in per capita health expenditure attributable to medical price inflation range from 0 to 22% (a period in which income more than doubled).

screenshot_2072.png

source

Likewise since 1970 nominal GDP per capita has grown at least as fast as the PCE health index, which is regarded by experts in this area as being amongst the best price indices for health prices.

screenshot_2103.png

source

If this index is presumed to be reasonably representative of true (overall) NHE prices more broadly this means that price increases do not explain the large long run increase in NHE relative to GDP in this span and that the overall long run increase is almost entirely attributable to increases in volume.   In reality there is good reason to believe these price indices miss real quality improvements in treatment and that treatment prices tend to overstate overall NHE costs in the long term (productivity improvements in other areas that sway NHE), i.e., the actual volume increase is probably somewhat larger still.

I index from 1970 in the plot above to make a point.  Up until late 70s/early 80s US NHE was comparable to the NHE seen in other rich OECD countries, the plot above, however, suggests the price of health care in the US didn’t suddenly shoot up relative to incomes.  In other words, changes in price are unlikely to explain it whereas changes in consumption and disposable income largely do  (the US clearly diverged more starkly from the rest of the OECD in both of these dimensions in this same time period and both are exceptionally strong predictors of NHE).  You would need to have some strange beliefs about the prices that must have existed in the US in the 70s to try to explain cross-sectional differences through prices if these price indices are reasonably accurate, as I think they are.

There are also some recent studies looking at disaggregated data on private claims:

screenshot_2104.png

screenshot_2105.png

source

Most categories saw little to no price increase and overall there was effectively zero net increase in prices.


This is not contradictory

To be perfectly clear: there is no contradiction in saying that overall health care prices have risen much faster than broader price indices and that the reason we spend more on medicine relative to our incomes is because we actually consume much more health care in real terms. This entirely consistent with Baumol’s effect and what Baumol’s variable greater than zero and less than one implies (see the 0.6 estimate in the OECD report above).  That is, it implies that the price of said goods and services rise causally in response to greater productivity in other areas of the economy (rising prevailing wage levels without corresponding gains in the sector’s productivity) but less than the actual wage increase.

aei_mark_perry_analysis_of_various_price_indices.png

source

As we have gotten much more efficient at producing manufactured goods, food, etc domestically and have enjoyed the benefits of cheaper labor for tradable goods (and some services– see foreign call centers, transcription services, etc) via “free trade” those remaining categories of expenditure, like health and education, which cannot or do not exploit automation, labor in non-rich countries, etc to nearly the same degree will appear to shoot up in relative terms.   This, of course, does not imply that they have truly become less efficient (with the same real factor inputs) or that the relative change in price level causes us to spend more relative to good measures of national income (note: the cash remuneration observed in common wage measures is very much incomplete insofar as it tends to exclude benefits, which have clearly risen considerably,  not to mention profits/investment income and the like, which also contribute to AIC, GDP, etc).

So while it is generally true that we could buy a considerably larger basket of other goods and services (e.g., electronics, clothing, food, etc) for what it costs to buy the equivalent basket of health care in 2017 as compared to 1970, it is generally not true that the overall, long run, rate of growth in health price levels has exceed the average rate of productivity growth, GDP per capita, AIC per capita, etc (although this gets increasingly blurry as we approach an index/measure than lean much more heavily on domestic labor inputs)  or, in and of itself, i.e., without increase in volume, could explain why overall health care costs are rising faster than income.


Health workforce

I believe in multiple lines of evidence so, to this end, I will present some data that points to higher volumes and suggests that price (especially via higher medical wages) is not the issue people think it is.

For instance, between 1965 and 2011 the health sector grew from ~3% to ~11% of the non-agricultural civilian workforce, a more than a 3.5 times increase in its share (and likely more of the workforce overall).

screenshot_2106.png

source

Clearly this isn’t just the roughly same number of providers demanding ever more money for their services.  This is consistent with rising volumes of health care: more providers of various sorts, more technicians, more para-professionals, etc providing, administering, or otherwise facilitating the delivery of health care at various levels.  Is it theoretically possible these people aren’t employed productively and that we suddenly forgot how to run a reasonably efficient system?  Maybe, but that strikes me as highly unlikely (especially when we’re likely witnessing similar patterns in other developed countries).


Physicians wages explain little

Likewise, while quite a few people allege physicians are the culprit for (supposedly) high US prices (presumably due to uniquely restricted supply), increasing physician remuneration is not likely to explain much in cross-sectional or, especially, domestic time series analysis.

screenshot_2107.png

screenshot_2108.png

source

Or, over a longer period (albeit from blog):

screenshot_2113.png

source

And the ratio of physician remuneration relative to average wages is not that much higher than seen in other developed countries:

screenshot_2110.png

screenshot_2109.png

source

Keep in mind that both the US remuneration levels are for the self-employed (salaried rates tend to be quite a bit lower in the US and other developed countries due to differences in responsibilities, hours, etc) and total physician remuneration only accounts for something like eight or ten percent of health care costs.  Even if US physicians worked for the wages seen in, say, the UK, setting aside all the other differences that contribute to this, the implied amount this would save isn’t exactly overwhelming.


On proxies for volume

Curiously some people make the claim that because the US doesn’t have more physicians, physician consultations, hospital beds, etc on a per capita basis and has shorter hospital stays that that must mean that it cannot be explained by volume (quantities) and that price must therefore play a major role.  I spoke to these issues in a prior blog post, but, briefly, what they neglect to mention is that these measures are, for the most part, weakly to even negatively associated with NHE amongst OECD countries.  If they can’t explain more than a few percent of the variance, many coefficients in the “wrong” direction and/or insignificant, there’s little reason to believe these variables say what they think about the US system.  It is just not a good representation of what drives modern health care costs (maybe 50 years ago the correlation would be much higher).  Moreover, if one looks at a much broader (albeit still woefully incomplete) array of rates of procedures, treatments, diagnostics, and other health system characteristics the US behavior is quite consistent with what we find in other high income countries (whether or not these patterns amount to “good” medicine or is “worth it” is another question), but we can at least say that high income, high NHE countries tend to behave like this, increasingly so as they get richer.  It is also not unreasonable to think these characteristics (greater use of technology, diagnostics, cutting edge treatment, etc) would be causally associated with higher levels of health expenditures.

comparison_between_scaled_NHE_and_correlation_by_year


Evidence pertaining to a handful of other (tangential) arguments

Some people argue the US healthcare is so profoundly different than other countries that this must somehow explain stuff.  While I can’t speak to every argument with sufficient length and detail, here are some statistics which suggest these sorts of differences are not all they are cracked up to be and that the US system generally looks much more like the rest the OECD than many people appreciate (especially on the delivery end).

Health finance

The voluntary/private/out-of-pocket share of NHE isn’t that much larger than many others and these don’t correlate with overall NHE significantly.

screenshot_2115.png

source

Net cost of health insurance

The net cost of (private) health insurance, the difference between what insurance companies collect and what they pay out in benefits, has averaged around 5-6% of NHE in recent history.

screenshot_2116.png

source

This very much includes the cost of processing claims, anti-fraud efforts, contracting with providers, coordination of care (kaiser etc), investing, marketing, executive salaries,  and so on and so forth.  Much of this is clearly needed under any reasonable system (some of the ‘savings’ in lower administration costs under Medicare, for example, also lead to higher rates of fraud and waste — not necessarily a ‘win’ for society).

Per beneficiary costs of private and public plans

Both Medicaid and Medicaid programs spend substantially more per beneficiary than private plans.

cms_per_beneficiary_cost_estimates

source

Obviously these population differ in substantive ways (age & income/poverty status), but no one could reasonably claim to call them cheap in the ways that matter.  They are still clearly much more expensive than what we see in lower income countries for equivalent populations.

Hospitals

The for-profit share of hospital beds in the US is quite modest and not much larger than many other OECD countries:

screenshot_2114.png

Though some might argue non-profit hospitals aren’t publics and that there may be some principal-agent problems, these sorts of issues don’t necessarily disappear in public institutions (not to mention other problems can arise) and we can at least say that the non-profit distinction is an important one insofar as actual profit motive is concerned.

Concentration and persistence

Though many suggest the allocation of health expenditures across society is abnormal, the best available evidence suggests the US looks much more the rest of the developed world as a function of age, end-of-life status, health status, income levels, and so on.

screenshot_1734

2014_medical_expenditure_panel_data_by_income


Conclusion

The better data suggests price has been massively over-emphasized and the role of volume has been given short shrift in both public and academic forums as explanation for high and rising health care expenditures.   Demonstrable errors have been allowed to go unchecked for some time even in exact sciences like (applied) physics.  It surely doesn’t help that this area is highly politicized, the data have not been that good, and it takes a composite of a good number of data points with sound methodology to generate a truly representative sample to draw robust conclusions (only in recent years are the health PPPs reliable enough to take very seriously).  Idiosyncratically (very) high US prices has become received wisdom and even honest, intelligent people that must know better (because they’ve actually reviewed enough of the pertinent the data) dodge the issue.  I have seen numerous cases where objective quantitive results of the researchers’ own work rather soundly contradicts casual assertions made and many other cases where inconvenient data like this are simply truncated from plots and allowed to go without comment (though accessible in tables, appendices, etc for those with sufficient interest and attention to the topic)…. likely because the researchers efforts are not especially concerned with US prices specifically and they would rather avoid getting mired in controversy.

It seems likely people are wildly overpromising the changes any realistic reform of our health care systems could bring, both with respect to the long term trajectory of health care costs and the so-called health outcomes.  I suspect our high and rising consumption of health care is largely a function of national wealth (+/- modest differences in propensity to consume, fiscal policy, etc) and I see little evidence that suggest that the different systems found throughout the OECD, let alone the average one, will produce much in the way of sustained savings.  As in, even if we manage to marginally reduce health prices through use of greater monopsony powers, it’s not without some tradeoffs (e.g., reduced R&D incentives, reduced dynamism, etc) and it’s quite likely to spur further health consumption (not necessarily a significant net reduction in expenditure or a net gain in health outcomes).  Though I certainly believe the systems found in several other OECD countries work reasonably well for them and have some advantages (and disadvantages), it’s just not the black and white issue people make it out to be where we’re simply leaving piles of cash on the table for no real reason and that all we need to do to capture it is adopt any other system.

Advertisements

3 thoughts on “Health care prices do not play the role most believe

  1. What about the fact that Utah spends less on Healthcare per capita than Canada and Arizona just a little bit more? That makes it look to me like healthcare spending in the USA could be much lower than it is. It also looks to me like it is state regulation that makes health care spending so high in soem states. See: http://www.kff.org/other/state-indicator/health-spending-per-capita/?currentTimeframe=0&sortModel=%7B%22colId%22:%22Location%22,%22sort%22:%22asc%22%7D

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s