Identifying Sources of Cost Disease

Wednesday June 14, 2017

Cost disease is one of, if not the, most important problem for society to be tackling right now. Cost disease is the notion that many industries have become dramatically more expensive over time, rather than less expensive, as you might expect with technological improvement. To put cost disease in a political frame, I greatly enjoy this quote from Scott Alexander:

Libertarian-minded people keep talking about how there’s too much red tape and the economy is being throttled. And less libertarian-minded people keep interpreting it as not caring about the poor, or not understanding that government has an important role in a civilized society, or as a “dog whistle” for racism, or whatever. I don’t know why more people don’t just come out and say “LOOK, REALLY OUR MAIN PROBLEM IS THAT ALL THE MOST IMPORTANT THINGS COST TEN TIMES AS MUCH AS THEY USED TO FOR NO REASON, PLUS THEY SEEM TO BE GOING DOWN IN QUALITY, AND NOBODY KNOWS WHY, AND WE’RE MOSTLY JUST DESPERATELY FLAILING AROUND LOOKING FOR SOLUTIONS HERE.” State that clearly, and a lot of political debates take on a different light.

For a deep dive into cost disease, I highly recommend his recent blog post.

From my research, cost disease can be found in: real estate, infrastructure, healthcare, education, military, and (arguably) the pharmaceutical industry. To provide a contrast, it's not seen in consumer electronics, energy, manufacturing, or agriculture, to name a few.

However, Alexander (the author) offers only guesses at a solution, not a satisfying answer. A truly satisfying answer has two parts to it. First, where is the money actually, literally going? And second, what market failure prevents these markets from behaving normally, auto-correcting itself, and having competition and innovation which drives down prices and increases quality over time. Let's first explore the first part - where is the money going?

  1. Administrative Bloat. Another economist, Cochrane, assigns most of the blame for cost disease on administrative bloat, per this article. Unfortunately, he provides the assertion but little evidence to back it up. The Commonwealth Fund has some research into this cause in the healthcare industry, where they show administrative costs are both much higher in the United States and are growing. (Article). This indeed makes some sense, but per the article, administrative costs of hospitals are 25%. Cutting that to 0% only cuts total healthcare costs by 25%, interesting but not nearly the 90% reduction to bring costs in line. A more reasonable reduction to 15% (bringing US admin costs in line with rest of world) would only decrease costs by 10%, not close to the desired 90% reduction. Administrative bloat appears to be one factor, but not the predominant one.

  2. Increased utilization. This idea is specific to the healthcare industry, but similar arguments might apply to other industries. Americans in particular have gotten sicker over time (obesity rates being one example of that), and the increased healthcare spending could simply be the population being generally sicker and needing more care. Interestingly, obesity rates started climbing in 1980, the same time that health care costs in US started departing greatly from trends in the rest of the world, per OECD life expectancy vs spending chart:

OECD chart

This thinking is backed up in a 2005 Health Affairs article which found that "for sixteen of [the twenty conditions most responsible for increases in healthcare costs], the rise in treated disease prevalence, rather than a rise in the cost per treated case, accounted for more than half of the growth in health care spending. To take more concrete numbers from the article, "the share of private health care spending attributable to obesity rose from 2 percent to 11.6 percent (from $3.6 billion to $36.5 billion) from 1987 to 2002." Multiplying the percentages implies that private healthcare spending nearly doubled in that period, from $180B to $310B, meaning obesity costs increased five times faster compared to the rest of healthcare (again, caused by number of cases, not the per-case cost).

However, total obesity cost estimates are roughly $200B (source), a mere 6% of $3.2T US annual healthcare spend. Chronic disease in aggregate was $1T, or 38% of total healthcare spend in 20101. Increased utilization caused by chronic disease and general lower health of the population seems to be a major driver of increases in healthcare cost. It doesn't give us a good answer for sources of cost disease in other industries, but hints that we should look for similar "increased spending caused by poor maintenance" patterns in other industries.

EDIT: I grossly under-estimated the cost of obesity here. The $200B figure was from 2009, whereas the $3.2T figure is 2017. In 2009 annual healthcare spending was closer to $1T. This puts the cost of obesity at 20%, in line with other estimates of 20-40%.

  1. Increased reliability of the system. I use reliability in an engineering sense here, meaning decreased risk tolerance for failure. In claiming diminishing returns of healthcare spending, Alexander (and most analysts) point to diminishing returns to Life Expectancy at Birth. However, if most of our healthcare spending is responsible for keeping only a small fraction of people alive, this return on investment wouldn't appear in Life Expectancy. Indeed, a 2012 study from DHHS found exactly that: in 2008 and 2009, 5% of spenders were responsible for 50% of total healthcare spend. "Aside from the fact that such a tiny fraction of the country was responsible for so much of our expenses, it also found that high spenders often repeated from year to year. Those chronically ill patients skewed white and old and were twice as likely to be on public health care as the general population."

    In other words, our healthcare system is supporting more individuals that are more expensive to keep alive, who previously would have passed away. Because these individuals are a minority, they don't appear in average statistics like life expectancy. Alexander brings up this idea in possible explanation 6 from his essay ("Changed our level of risk tolerance"), but I think the evidence above drives the point home. This fact sheet from AHRQ underscores these points, stating "one study found that a small number of conditions accounted for most of the growth in total health care spending between 1987 and 2000—with the top five medical conditions (heart disease, pulmonary disorders, mental disorders, cancer, and trauma) accounting for 31 percent."

Some other factors might be at play, but it does seem like these three factors: administrative bloat, increased utilization, and increased reliability/decreased risk tolerance, provide an answer to the question: where is the money going? Without being too extensive in analysis, you can see similar trends in other industries affected by cost disease.

Looking at education, time in school has gone up without commensurate increases in outcomes. According to "Market Education: The Unknown History," by Andrew Coulson, in 1909-1910, the average American student spent 113 days in school. By 1969-1970 that average had climbed to 161 school days; today that number is approaching 180 days. This means that the high school graduates of today are in class for what amounts to more than four additional school years — at the 180-day school year level — than graduates of 1910. (Source via this article).

As an interesting aside, I'll conjecture that some of the increased utilization and therefore cost increase in education comes from the utilization of primary and secondary as a day-care while both parents work. In other words, increased expenditure on education is a reflection that work which was traditionally not part of GDP or spending (i.e. child-rearing) is now more so reflected in economic output.

We can also see increased reliability/decreased risk tolerance in education through the decreasing race gap in test scores. While average test scores have remained nearly flat, the White-Black gap and the White-Hispanic gap, while still sizable, have decreased anywhere from 36% to 50% in the period from 1973 to 2012, depending on which score you look at. (Source). (It's worth confirming causality here, as my intuition is that minority/majority-minority schools actually get less funding, not more. Perhaps educational outcomes has been de-risked in other ways.)

Market Failure

With this knowledge at hand, it's fair to ask at this point whether cost disease is a market failure at all. If people are getting better, if people at the far end of the bell curve are surviving, then we are getting something for our money, and it is our exposed preference as a society that we make this trade-off. Let's examine several root causes and then return to this question.

With an understanding of where the money is genuinely going in cost disease, let us analyze why the money is going there. In other words, why does competitive or market pressure not drive down costs?

  1. Priceless outcomes/Unbounded demand. I might befall selection bias here, but thinking qualitatively about the industries with cost disease, they have either priceless outcomes, or they have an "arms race" quality to them: healthcare - your life, education - must be smarter than everyone else, military - must be stronger than your opponent. (That said, infrastructure or even real estate doesn't neatly fall in this bucket.) On the other hand, demand in other industries is bounded: a population needs a fixed amount of food from agriculture, growth is catalyzed by a fixed amount of energy. Therefore, the market failure here might be systemic and not a failure so much as a demonstration that human desire for some things is unlimited. However, unbounded demand itself doesn't immediately imply market failure. There has to be a finite supply, otherwise more and more suppliers would flood a market and at minimum drive quality up (even if not prices down). Indeed, when you consider geographic restrictions (e.g. only use a hospital, school, or roads near your home), the supply for cost disease industries seems fixed in some regard. The problem is compounded by the fact that humans often use price as a signal for what they don't understand, driving up cost for these "priceless" outcomes. With finite supply and unbounded demand, the conclusion is the costs in these industries will expand to require available resources (much like Parkinson's Law.
  2. R&D goes to increasing spend over cost-cutting. At risk of sounding conspiratorial, there might be minimal incentive to innovate around cost-cutting. If I'm an scientist, finding a treatment for a disease that currently has no treatments available is: 1. life changing for those affected, 2. life changing for me if I'm personally affected by the disease, and 2. immensely financially rewarding. And especially if these are life-saving treatments, they will be covered by insurance.

    This is a sub-point to the above, but I'm reminded of a pet interest of mine: Vasalgel male birth control. The technology has existed under-developed since the 80's, despite it being a massive advance once it comes on the market, but there being minimal financial incentive to develop it because it is extremely cheap and cuts into recurring revenue from female hormonal birth control. It's now being developed by a non-profit.

    The counter-point is that cost-cutting still has entrepreneurship opportunities in its own right. I think this point is weak- or non-existent, but I'll keep it here in case it sparks an idea for someone else.

  3. Individuals are not the payers. In most industries which befall cost disease, there's a complex interplay between individuals, governments, and the private sector which confound market forces. I envision a sort of uncanny valley here. On one extreme, pure markets work reasonably well. On the other extreme, pure government industries work because government has enough bargaining power to drive down costs (e.g. much less impact of cost disease in countries with single-payer healthcare). Between those extremes, markets fail and government strength fails, leading to over-running costs. This theory explains why cost disease is worst in the US, but it doesn't explain why (secondary) education, probably the most heavily government-run of all cost disease industries mentioned, has some of the worst rates of cost disease.

Of these, the first seems like the strongest reason, exacerbated by the third. Returning to the question of whether cost disease is indeed a market failure, I'm led to conclude, "No." However, that doesn't mean rising costs can't be combatted.

Treatment but no Cure

The unbounded demand leads to following conclusion: even if we optimize delivery, we'll still as a society be hungrier for more spend in these industries. In healthcare, even if we cut US administrative costs by 40% to bring them in line with the rest of the world, we only reduce total cost by 10%, slowing cost growth by a few years. We need to reign in increased utilization with preventative care/maintenance, and we need to reign in costs of increasing reliability.

Combatting increased utilization seems achievable: giving governments the analytical tools to demonstrate that certain expensive decisions (e.g. running a school for 180 days compared to 113) doesn't improve outcomes might lead to policy changes. Giving people better access to healthy food would dramatically lessen chronic conditions and the associated healthcare costs. We just need a cultural shift in government so that they are willing to experiment with such programs.

Combatting costs from increasing reliability/decreasing risk tolerance seem much trickier. It's untenable to say even though we have the technology to save someone, we won't because it's too expensive. In healthcare, a cultural shift from life-saving to palliative care might change this equation. We might also look to better sharing of operational insight to deliver outsized returns on a per-cost basis. In education, are there some schools that have done a particularly good job closing racial achievement gaps with minimal additional spending, and can that knowledge be better shared to districts across the country? To use an example from Alexander's post, can we bring experts that have built subways in Korea to the United States to advise cost-savings on subway projects here?

The discussion on supply and demand in the context of unbounded demand leads to another idea: increase competition on a geographic level. You begin to see this already in healthcare with the explosion of the urgent care industry (although it's unclear this trend will lead to decreased spending on healthcare - one RAND study found the opposite2). It's worth noting that urgent care facilities compete on improved experience, convenience, or availability rather than price or quality of care. If we have better outcomes and satisfaction given fixed costs, then that's certainly a win. While the analysis here indicates we likely won't get cost improvements, market forces should still work effectively to drive improved experiences.

No single one of these ideas jumps out at me as a clear winner. I'll keep thinking.


  1. calculated from direct medical costs from chronic disease taken from the CDC and total healthcare spend in 2010 from the KFF

  2. http://www.rand.org/news/press/2016/03/07.html 


Want to receive similar articles? (No spam, promise!)