What does the cost disease imply about the public sector?

by on October 9, 2012 at 6:55 am in Economics, Political Science | Permalink

Matt Yglesias has a good post on the recent Steven Pearlstein column.  Here is Matt:

…people need to start paying much more attention to questions of tax efficiency. It’s overwhelmingly likely that we’re going to want the public sector to be a larger share of the economy in 10, 20, 30, 40 years than it is today and we need to find relatively growth-friendly ways to make that happen.

Here is Pearlstein:

    From a political perspective, Baumol’s most important insight is that government spending must grow as a percentage of the economy. Most of the services that are provided by, or financed by government — health care, education, criminal justice, national security, diplomacy, industry regulation, scientific research — are those that suffer most acutely from Baumol’s disease. That’s not because of incompetence or self-interest on the part of public servants or even the socialist instincts of Democratic politicians — it’s in the nature of those activities.

To demand, as Republicans do, that government be held to some historical average as a percentage of the economy stubbornly ignores this reality. It would condemn the country, as John Kenneth Galbraith once put it, to a future of “private affluence and public squalor.”

Let’s for the purposes of discourse take the cost-disease argument, and this classification of sectors, for granted.  I would stress that there are two different ways of measuring the relative size of government in the economy.  The first is as expenditure share, say as a fraction of gdp, and that indeed may well go up because of Baumol’s argument (I’ll return to this).

The second question concerns the real value of outputs from government, as measured from the consumer side.  If government outputs increasingly cost more to produce, should not a substitution effect kick in and lead us to prefer, at the margin, a higher proportion of productivity enhancement-enjoying private sector outputs?  In common parlance, if flat screen TVs become much cheaper, buy more of them.

This implication often receives less stress from cost-disease advocates and you will note that it militates in favor of substituting away from government outputs.

There is a further implication.  What goes up on the expenditure side is the share of any given governmental output in gdp.  If we substitute out of government outputs enough, the total share of government output, as an expenditure fraction of gdp, could go up or down in an optimum.  If the elasticity of demand is sufficiently high, flat screen TVs can become an increasing share of gdp and government social workers a smaller share.  Right now for instance internet commerce is a growing share of gdp as measured in terms of expenditures.

In any case, moving away from expenditure shares and back toward output: if government responds optimally, we end up with government as a smaller share of real output over time, yet at higher costs.  Surely that is a recipe for cynicism, justified or not.

Brandon Berg October 9, 2012 at 7:10 am

That’s not what the Baumol effect is, is it? The Baumol effect is the tendency for the real cost of labor-intensive goods and services to rise as an abundance of capital increases labor productivity in other sectors. It says nothing about expenditures as a percentage of GDP. In fact, Baumol’s logic suggests that, ignoring substitution and income effects, expenditures on these goods and services should remain constant as a percentage of GDP, because the amount of labor required to produce these goods and services remains constant. The increase in real prices comes from the potential output from alternative uses of that labor.

Or am I the one misunderstanding Baumol?

david October 9, 2012 at 8:18 am

If you ignore substitution and income effects, of course nothing changes as a result of a price change. There are no other effects of any price change.

The argument offered by Yglesias is entirely an income effect: non-labour-intensive goods get cheaper, so you get richer in real terms, but spend that increase on the labour-intensive component, thus increasing its share of GDP. You might just end up bidding up the labour-intensive component, a la Baumol, and not getting very much more for that increase – the increase in welfare only coming directly from working less for the non-labour-intensive goods.

Brandon Berg October 9, 2012 at 8:24 am

Yglesias just quotes Pearlstein, who says that this is due to the Baumol effect, not to income effects. And the income effect is a very different claim, namely that government spending increases as a percentage of GDP because it’s doing more, not because it requires a greater share of GDP to maintain a constant level of real output.

david October 9, 2012 at 9:17 am

Um… no? Income effect results when representative real income changes due to the decrease in non-labour-intensive costs. Baumol is what is used to argue that the substitution effect doesn’t outweigh the income effect, so that you don’t buy more of something whose price has fallen. Instead you spend less on it.

Ray Lopez October 9, 2012 at 10:54 am

Spending less on something as the price falls…isn’t that an “inferior good”? So government services are an inferior good? Sounds reasonable to me, on many levels. Once you guys agree on something get back to us…

Ray Lopez October 9, 2012 at 10:59 am

@myself: Maybe I meant a Giffen good/ Veblin good, and perhaps as real income rises demand for an inferior good falls, a slightly different concept, as I think about it.

david October 9, 2012 at 6:13 pm

What the argument says you are spending less on is the non-labour-intensive good, e.g., consumer electronics and other automated goods. You are buying a much better quality of iDevice, but the device gets so much cheaper so fast that increasingly smaller shares of your income are dedicated to it.

‘Inferior’/'normal’ classifications are hard to nail down when quality is changing so fast.

The “isn’t it an inferior good? HAHAHA” argument is the weirdest attempt at mood affiliation I’ve seen, though. It’s just a technical label!

Paul Zrimsek October 9, 2012 at 7:14 am

The 1912 version of Pearlstein: “Baumol’s most important insight is that spending on domestic servants must grow as a percentage of the economy.”

Jim Clay October 9, 2012 at 9:58 am


NVM October 9, 2012 at 7:21 am

Don’t we need to account for the fact that some of these things — like national security, diplomacy, criminal justice, etc. — are things that we simply can’t live without? That is to say, even if government outputs become costlier to produce, we tend to be fairly price insensitive when it comes to those things because we know we need them for society to function. Also, I wonder if less national security, criminal justice, legal work, etc. would eventually diminish productivity in the private sector as we substitute away from necessary public sector activities.

dearieme October 9, 2012 at 3:55 pm

I’d guess that US national security may have been reduced by recent expenditure on”national security”. Attacks on Iraq and Afghanistan may have had that effect.

I doubt that your expensive navy adds all that much to national security (I assume that all surface ships might soon prove obsolete).

Meanwhile your southern border is entirely permeable to anyone who wants to cross it.

And every sign of Romney doing well in the polls must increase the chance of an attack on Iran before the election.

Jay October 9, 2012 at 5:56 pm

There’s actually a huge usage of the Navy going on in the Persian Gulf right now. Not to mention, if something happens and we do “defend Israel,” that will probably require more use of the navy in the mediteranean, the local rivers and seas as well. I’d say that all factors of the military are doing more AND less than perceived. But a solid 80% of this response is a decently biased opinion.

NVM October 9, 2012 at 7:25 am

Brandon — I think public sector expenditures would rise as a percentage of GDP because even if those costs rise at a level that is commensurate with costs in other industries, the public sector activities are not anymore productive so you don’t get the corresponding increase in GDP that you would get as a result of a cost increase in sectors where wages respond to increases in productivity.

Brandon Berg October 9, 2012 at 7:38 am

I don’t think that GDP calculations are that smart. We see that the government is paying its workers more money, so G goes up. We don’t multiply G by some discount factor because its labor productivity isn’t improving as much as it is with C and I.

Brandon Berg October 9, 2012 at 7:48 am

The bottom line is that if it takes 15% of the labor force to provide a certain real level of government output now, increasing labor productivity in other sectors in and of itself isn’t going to push that up to 20%. In fact, any increase in labor productivity in the public sector at all is going to push that number down a bit.

That said, people may respond to this in ways that cause government to grow as a percentage of GDP. For example, if higher real incomes lead to a fall in labor force participation, and the government’s need for labor is constant, that leaves less left over for other sectors, and government increases as a percentage of GDP. An aging population can have the same effect, of course. But neither of these is the Baumol effect, and rising real wages could lead to higher labor force participation if the substitution effect dominates the income effect.

Bill October 9, 2012 at 7:51 am

Brandon, The government purchases services from the private sector as well. Baumol’s point doesn’t depend on whether you purchase the service from a private service contractor or directly from an individual government employee. It is a statement about the service sector.

Brandon Berg October 9, 2012 at 8:07 am

Yes, I was simplifying. Obviously the government can contract things out, but that doesn’t impact my point at all. To be more precise we should say labor-intensive sectors, but I’m addressing the fallacious claim by Perlstein and Yglesias that the Baumol effect means that government spending should increase not just in real terms, but as a percentage of GDP. If the government sector is particularly labor-intensive, Baumol’s logic predicts the first but not the second.

Paul Zrimsek October 9, 2012 at 7:42 am

I don’t believe spending on the things NVM mentions has been increasing even as a proportion of total government spending, let alone the entire economy. The big-ticket items today seem to be things, like cutting Social Securoty checks, that are probably more automated than the economy as a whole.

Slocum October 9, 2012 at 7:57 am

Most of the services that are provided by, or financed by government — health care, education, criminal justice, national security, diplomacy, industry regulation, scientific research — are those that suffer most acutely from Baumol’s disease.

I’m not at all convinced that Baumol’s disease is the main problem in these areas. If you look at forms of medicine that are paid out of pocket (LASIK vision correction, cosmetic surgery, dental implants, for example), costs DO come down. In education, the U.S. has greatly increased money spent on education over recent decades, but chiefly because of administrative bloat and sweetheart deals between politicians and their public sector worker supports (pension spiking is not a form of Baumols). The opportunity for online education to radically reduce costs is obvious, but the state’s near monopoly on education (especially K-12) may stop this in its tracks (or at least delay it). And defense spending? Spending as much money as the next 20 or so nations combined is definitely NOT caused by Baumol’s disease. And, in fact, a high-tech military means less manpower (and shrinking manpower means whatever cost problems we have are not Baumol problems).

Brian Donohue October 9, 2012 at 12:51 pm

FedEx vs. USPS?

The D-man October 9, 2012 at 8:05 pm

Lasik and all the others you mention are elective procedures which have an alternative that costs nothing – do without. What’s the alternative for heart surgery? Diabetes? Emphysema? Arthritis? A broken leg?

FedEx will not deliver a letter to the middle of Alaska for 45 cents.

Dan Cole October 9, 2012 at 7:58 am

I think NVM’s wider point remains valid (if I’m understanding him correctly): some goods have attributes of public goods (whether or not they are entirely pure or impure public goods) or common-pool goods (as the Ostroms might have called them). Private goods may not be adequate substitutes for them, even if the costs of provision of the (partial) public goods rise quite quickly.

NVM October 9, 2012 at 9:08 pm

That’s right, Dan. You stated what I was thinking a lot more succinctly than I did above.

Arthur October 9, 2012 at 7:58 am

Matt thought about that too. Aparently he thinks health care is quite price inelastic:


I thing he got the corners wrong. I think there should be a price elastic row and a price inelastic row, one with servants and smartphones, other with lumber and health care.

But that is what he’s thinking about, the relationship between cost diseases and elasticity.

Brandon Berg October 9, 2012 at 8:18 am

If government outputs increasingly cost more to produce, should not a substitution effect kick in and lead us to prefer, at the margin, a higher proportion of productivity enhancement-enjoying private sector outputs?

On the other hand, the substitution effect is always engaged in a tug-of-war with the income effect. You can only find so many uses for flat-screen TVs, and once you’ve had your fill, expenditures on labor-intensive goods and services become more attractive.

ian October 9, 2012 at 9:41 am

Let’s not limit ourselves to just imagining substitution of private sector goods for public sector services (as the price of those services increase). In the categories that Pearlstein lists we already see the growth of private substitutes with lower cost than their public provided/supported counterparts. Health tourism serves as just one example.

Brandon Berg October 9, 2012 at 10:19 am

It’s not clear to me that medical tourism exerts much of a downward pressure on government health care expenditures. People go overseas to save money when they’re paying out of pocket. When the government’s paying, they don’t care how much it costs.

Brandon Berg October 9, 2012 at 10:20 am

Well, maybe indirectly, insofar as it lowers demand for services from domestic doctors.

joan October 9, 2012 at 8:19 am

If you look and industry sectors by value added from 1975 to 2005 government share of GDP ( excluding transfer payments) has declined from 15.1% to 12.4% of GDP . Most of the decline is probably due the cuts in national security spending at the end of the cold war but if the cost of the other things government does, including public schools, has increased as much as people think total would not have decreased.
However I do not understand how computers increased the share of GDP in the financial sector, I would have expected that they made it more efficient. like they did manufacturing

uffthefluff October 9, 2012 at 5:50 pm

That link does seem to show a nice clear bifurcation between sectors affected by cost disease and those not. The FIRE sector is probably dominated by salespeople as a share of employees which are a prone to catch said disease, but it could just show an increase in successful rent seeking.

Fewer fires, lower crime rates, the end of the Cold War, teacher layoffs – the only major government spending area definitely affected with Baumol’s is health care. All else, including defense, old age pensions and nutrition assistance, seem to be heavily dependent on circumstance and public policy decisions.

We will have robo-nurses and robo-surgeons and maybe eventually AI marketing consultants and AI robo-nurse program(er)s, then all bets are off.

mpowell October 9, 2012 at 11:00 am

It’s just too bad that one party has decided to cynically demagogue this issue into the ground so there is very little chance for any meaningful attempts at tackling these issues in the near future. When one party is just trying to burn everything to the ground, the other party is forced into a desparate defense stance and there is little space for acknowledgement of the need for reform in a government sector (either in the sense of scaling back a policy mission or changing the way it is pursued).

Fred October 9, 2012 at 11:54 am

This is a really interesting, insightful post! Some of the comments are too. Zrimsek makes your point in 20 words or less.

Rich Berger October 9, 2012 at 12:12 pm

” It’s overwhelmingly likely that we’re going to want the public sector to be a larger share of the economy in 10, 20, 30, 40 years than it is today ..”

So sayeth “Matt”. Who are “we”? Not me.

David October 9, 2012 at 12:50 pm

I think Baumol’s disease is something that academics like to discuss, but has little to do with reality … take this (from Wikipedia): Baumol and Bowen pointed out that the same number of musicians are needed to play a Beethoven string quartet today as were needed in the 19th century; that is, the productivity of classical music performance has not increased. On the other hand, real wages of musicians (as well as in all other professions) have increased greatly since the 19th century.

What he misses is that today, I can watch the best musicians in the world perform a Beethoven string quartet on Youtube for nothing! In fact, in the 19th century, very few people (as % of population) could afford to see a string quartet perform Beethoven, today a few billion people can see it for free on Youtube … desease cured?

In fact, this applies to all ‘labor intensive’ activities … think Statin drugs vs. open heart surgery. Think roomba vacuum cleaners vs. house cleaning staff. Heck even basic dishwashers and laundry machines have offset the work performed by these people. Nation defense? Are you kidding me … one super computer today can do more work 10,000 WW2 crypto analysts. Education … has anyone heard of the Khan academy?

To all you academics studying this … I suggest you come down from your ivory towers and see what’s going on in the real world … in the real world, when prices rise, people innovate.

Sometimes innovation does not happen fast enough, especially where teachers and nurses unions fight displacement, or where heavy handed lobbying by AMA or ABA prevent automation. Perhaps you should study the (negative) effects of innovation in industries where there is strong influence of unions or other lobbying groups!

Richard October 11, 2012 at 9:50 am

Absolutely. Baumol-disease advocates take a blinkered view in defining a service by defining away almost all service-related technological progress that generates close substitutes, while emphasizing the role of technologoical advances in the manufacturing sector. The application of Baumol’s disease to education, particularly higher education, is often invoked defensively by those who seek to deny the role of other explanations (an education “bubble” or more probably, the Bennett hypothesis). I suspect the relative attraction of Baumol’s disease to those erecting the barricades lies precisely in the notion that nothing can be done in certain activities to alter the labor/output ratio, so costs and prices simply have to rise ad nauseum, ad infinitum. In this, they will prove sadly mistaken. My main concern is that those embracing this argument are in denial and are, in the process, ignoring ways in which their institutions might participate in change rather than assume (incorrectly) it can never successfully occur. At some point, it will be too late for them to shift gears.

Brian Donohue October 11, 2012 at 12:36 pm

Hmmm…”you have to hire more of us, you have to pay us more, and, oh yeah, don’t expect any fancy productivity increases.”

If I were a public-sector worker, I would hope I’d find this kind of reasoning just a tad…convenient.

Thomas October 9, 2012 at 1:41 pm

I don’t understand why Cost-Disease should result in government expenditures rising as a share of GDP. If we assume that income growth is tied to GDP growth, then income rises by X%, GDP rises by X%, and government spending on wages and goods should also rise by X%. Where does the growth in government spending which exceeds GDP growth come from?

Osman Rahman October 9, 2012 at 3:31 pm

My question is why would substitution effect be such a big deal in here? Some public sector outputs are difficult to substitute by private sector outputs. Replacing the police with private security guards is not cost effective either in a libertarian or a real world.

Scott Linthorst October 9, 2012 at 5:36 pm


I have not read Baumol, so I may be missing some fundamental understanding of his position. However, if Pearlstein’s summary below is accurate, I see two fundamental flaws with the position he takes:

“Despite no gains in productivity, however, the pay of these service workers rises — after all, if it didn’t, over time all those service workers would be lured to the higher-paying goods sector. Moreover, demand for services rises because all those farmers and factory workers want to use their increased income to buy more services. In response to the increased demand and the higher pay, service companies raise their prices.”

1. Why should the pay of service workers rise? They CANNOT be lured to the higher-paying goods sector jobs. As a result of the very productivity improvements which resulted in that sector being higher-paying, there are a finite, shrinking number of these positions.

2. Demand for services in general will rise – but not for the SAME set of services. The relatively wealthy goods-producer only has so much appetite for certain services (e.g., current healthcare services). With increased wealth, s/he will likely pursue different and new services. These new services have to compete with the old sets of services for the total, slightly-increased demand actually putting downward pressure on prices.

If you look at the long-term trend of the cost of services where there is very limited opportunity for productivity improvements, it is NOT true that we see prices growing ever-higher. Look at the history of prices of the world’s oldest profession – prostitution – as an example.

The issue when discussing “healthcare” is that we are actually talking about a set of both products and services that has been rapidly expanding due to radical innovation in healthcare goods and services. If viewed as a single “service” or “good,” this might look like an increase in both demand and cost.

To suggest there is limited opportunity for productivity improvement shows a lack of appreciation for the history of healthcare, and a lack of imagination for the future! New tools (e.g., mobile nursing stations) and techniques (e.g., key-hole surgery) have resulted in sizable efficiency gains – and electronic health records, care management, preventative care, nurse-led clinics, e-enablement/telemedicine, DNA-driven risk scoring, will have similar impacts in the future.

And for “education” I can say the same thing. Isn’t MRU (and TED, on-line lectures and schools, e-books) an example of innovation leading to productivity improvements and lower costs. A teacher can now read “Green Eggs and Ham” in the same 12 minutes – but if you record it and post it to the internet you can transmit it to millions of people instead of one classroom full.

By grouping complex sets of goods and services into over-simplified labels like “healthcare” and “education” and then tying them to a single price (e.g., taxes) one obscures the actual complexity of the market and blocks the ability of prices to be used as signals. This is not Baumol’s cost disease – this is the disease of public sector central planning.

Silas Barta October 9, 2012 at 5:40 pm

The entire Baumol argument is just a complex version of “Gosh, I can’t figure out how this good could be provided except by someone doing it manually. Guess it’ll get more expensive, and we just have to accept that!”

In another century, he might have been telling us how you can never get a person so productive a computing that they can use two slide rules at once. Or in another, that long-distance travel will always require one person driving the horses on the carriage. On in another, that there’s only so much force a single human can exert turning a mill.

Kent Anderson October 9, 2012 at 8:08 pm

It has been my experience that whenever someone says something like: “It’s overwhelmingly likely that we’re going to want..” It almost never transpires. The consensus view fails over time.

FreeDem October 9, 2012 at 8:31 pm

“If government outputs increasingly cost more to produce, should not a substitution effect kick in and lead us to prefer, at the margin, a higher proportion of productivity enhancement-enjoying private sector outputs? In common parlance, if flat screen TVs become much cheaper, buy more of them.”

What would this look like? Perhaps consumers deciding that they don’t want to consume more education and health care, they are happy with a quality of life that doesn’t include college and overindulges in poor lifestyle choices like obesity. Instead, they focus on flat screen TVs and other forms of entertainment.

In other words …

Honey Boo Boo?

chuck martel October 10, 2012 at 2:27 pm

You’re saying that there’s some difference between the two major parties when it comes to government activity? Hahahahahahaha.

john personna October 11, 2012 at 11:48 am

A wonderful discussion, but it is quite depressing that it is so far from the popular sphere.

Comments on this entry are closed.

Previous post:

Next post: