Tag Archives: economic growth

Growth in France and the US: The Bottom 90% Have Done Better in France

Posted on by

France vs US, 1980-2012, GDP per capita overall and of bottom 90%

A.  Introduction

Conservative media and conservative politicians in the US have looked down on France over the last decade (particularly after France refused to join the US in the Iraq war, and then turned out to be right), arguing that France is a stagnant, socialist state, with an economy being left behind by a dynamic US.  They have pointed to faster overall growth in the US over the last several decades, and average incomes that were higher in the US to start and then became proportionately even higher as time went on.

GDP per capita has indeed grown faster in the US than it has in France over the last several decades.  Over the period of 1980 to 2007 (the most recent cyclical peak, before the economic collapse in the last year of the Bush administration from which neither the US nor France has as yet fully recovered), GDP per capita grew at an annual average rate of 2.0% in the US and only 1.5% in France.

But GDP per capita reflects an average covering everyone.  As has been discussed in this blog (see here and here), the distribution of income became markedly worse in the US since around 1980, when Reagan was elected and began to implement the “Reagan Revolution”.  The rich in the US have done extremely well since 1980, while the not-so-rich have not.  Thus while overall GDP per capita has grown by more in the US than in France, one does not know from just this whether that has also been the case for the bulk of the population.

In fact it turns out not to be the case.  The bottom 90%, which includes everyone from the poor up through the middle classes to at least the bottom end of the upper middle classes, have done better in France than in the US.

B.  Growth in GDP per Capita in France vs. the US:  Overall and the Bottom 90%

The graph at the top of this post shows GDP per capita from 1980 to 2012 for both the US and France.  The figures come from the Total Economy Database (TED database) of the Conference Board, and are expressed in terms of 2012 constant prices, in dollars, with the conversion from French currency to US dollars done in terms of Purchasing Power Parity (PPP) of 2005.  PPP exchange rates provide conversions based on the prices in two respective countries of some basket of goods.  They provide a measure of real living standards.  Conversions based on market exchange rates can be misleading as those rates will vary moment to moment based on financial market conditions, and also do not take into account the prices of goods which are not traded internationally.

Real GDP per capita (for the entire population) rose for both the US and France over this period, and by proportionately somewhat more in the US than in France.  These incomes are shown in the top two lines in the graph above, with the US in black and France in blue.  GDP per capita in France was 83% of the US value in 1980, and fell to 72% of the US by 2012.

But the story is quite different if one instead focuses on the bottom 90%.  The GDP per person of those in the bottom 90% of the US and in France are presented in the lower two lines of the graph above.  The figures were calculated using the distribution data provided in the World Top Incomes Database, assembled by Thomas Piketty, Emmanuel Saez, and others, applied to the GDP and population figures from the TED database.  The US distribution data extends to 2012, but the French data only reaches 2009 in what is available currently.

The Piketty – Saez distribution data is drawn from information provided in national income tax returns, and hence is based on incomes as defined for tax purposes in the respective countries.  Thus they are not strictly comparable across countries.  Nor is taxable income the same as GDP, even though GDP (sometimes referred to as National Income) reflects a broad concept of what constitutes income at a national level.  But for the moment (the direction of some adjustments will be discussed below), distributing GDP according to income shares of taxable income is a good starting point.

Based on this, incomes (as measured as a share of GDP, and then per person in the group) of the bottom 90% in France were 88% of the US level in 1980.  But this then grew to 98% of the US level by 2007, before backing off some in the downturn.  That is, the real income of the bottom 90%, expressed purely in GDP per person, rose in France over this period from substantially less than that for the US in 1980, to very close to the average US income of that group by 2007.  And since one is talking about 90% of the population, that is all those other than the well-off and rich, this is not an insignificant group.

C.  Most of the US Income Growth Went to the Top 10%

Figures on the growth of the different groups, and their distributional shares, show what happened:

France US
GDP per Capita, Rate of Growth, 1980-2007
  Overall 1.5% 2.0%
  Bottom 90% 1.4% 1.0%
Share of GDP, 1980
  Top 10% 31% 35%
  Bottom 90% 69% 65%
Share of GDP, 2007
  Top 10% 33% 50%
  Bottom 90% 67% 50%
Share of Increment of GDP Growth, 1980-2007
  Top 10% 36% 62%
  Bottom 90% 64% 38%

As noted before, overall GDP per capita grew at a faster average rate in the US than in France over this period:  2.0% annually in the US vs. 1.5% in France.  But for the bottom 90%, GDP per capita (for the group) grew at a rate of only 1.0% in the US while in France it grew at a rate of 1.4% per year.  The French rate for the bottom 90% was almost the same as the overall average rate for everyone there, while in the US the rate of income growth for the bottom 90% was only half as much as for the overall average.

Following from this, income shares did not vary much over the 1980 to 2007 period in France.  That is, all groups shared similarly in growth in France.  In contrast, the top 10% in the US enjoyed a disproportionate share of the income growth, leaving the bottom 90% behind.

In 1980 in France, the top 10% received 31% of the income generated in the economy and the bottom 90% received 69%.  With perfect equality, the top 10% would have had 10% and the bottom 90% would have had 90%, but there is no perfect equality.  The US distribution in 1980 was somewhat more unequal than in France, but not by much.  In 1980, the top 10% received 35% of national income, while the bottom 90% received 65%.

This then changed markedly after 1980.  Of the increment in GDP from growth over the 1980 to 2007 period, the top 10% received 36% in France (somewhat above their initial 31% share, but not by that much), while the bottom 90% received 64%.  The pattern in the US was almost exactly the reverse:  The top 10% in the US received fully 62% of the increment in GDP, while the bottom 90% received only 38%.  As a result of this disproportionate share of income growth, the top 10% in the US increased their overall share of national income from 35% in 1980 to 50% in 2007.  Distribution became far more unequal in the US over this period, while in France it did not.

The data continue to 2012 for the US, but the results are the same within roundoff.  That is, the top 10% received 62% again of the increment of GDP between 1980 and 2012 while the bottom 90% only received 38%.  For France the data continue to 2009, but again the results are the same as for 1980 to 2007, within roundoff.

With this deterioration in distribution, the bottom 90% in the US saw their income grow at only half the rate for the economy as a whole.  The top 10% received most (62%) of the growth in GDP over this period.  In France, in contrast, the bottom 90% received close to a proportionate share of the income growth.  For those who make up the first 90%, economic performance and improvement in outcomes were better in France than in the US.  Only the top 10% fared better in the US.

D.  Other Factors Affecting Living Standards:  Social Services and Leisure Time

In absolute terms, even with the faster growth of real incomes of the bottom 90% in France relative to the US over this period, the bottom 90% in France came close to but were still a bit below US income levels in 2007.  They reached 98% of US income levels in that year, and then fell back some (in relative terms) with the start of the 2008 downturn.

But the calculations discussed above were based on applying distributional shares from tax return data to GDP figures.  For income earning comparisons, this is reasonable.  But living standards includes more than cash earnings.  In particular, one should take into account the impact on living standards of social services and leisure time.

Social services include services provided by or through the government, which are distributed to the population either equally or with a higher share going to the poorer elements in society.  An example of a service distributed equally would be health care services.  In France government supported health care services (largely provided via private providers such as doctors and hospitals) are made available to the entire population.  Since individual health care needs are largely similar for all, one would expect that the bottom 90% would receive approximately 90% of the benefit from such services, while the top 10% would receive about 10%.  If anything, the poor might receive a higher share, as their health conditions will on average likely be worse (and might account for why they are poor).  For other social services, such as housing allowances or unemployment compensation, more than 90% will likely accrue to the bottom 90%.

Taking such services into account, the bottom 90% in France will be receiving more than the 67% share of income (in 2007) seen in tax return data.  How much more I cannot calculate as I do not have the data.  The direction of change would be the same in the US.  However, one would expect a much lower impact in the US than in France because social services provided by or through the government are much more limited in the US than in France.  While Medicare provides similar health care as one finds in France, Medicare in the US is limited to those over 65, while government supported health care in France goes to the entire population.  And the social safety net, focussed on the poor and middle classes, is much more limited in the US than in France.

In addition, economists recognize that GDP per capita is a only crude measure of living standards as it does not take into account how many hours each individual must work to obtain that income.  Your living standard is higher if you can earn the same income but work fewer hours as someone else to receive that income, as the remaining time can be spent on leisure.  And there is nothing irrational to choose to work 10% fewer hours a year, say, even though your annual income would then be 10% less.  The work / leisure tradeoff is a choice to be made.

GDP per capita may often be the best measure available due to lack of data on working hours, but for the US and France such data are available (and are provided in the TED database referred to previously).  One can then calculate GDP per hour of work instead of GDP per capita, both overall and (using the same distributional data as above) for the bottom 90%.  The resulting graph for 1980 to 2012 is as follows:

France vs US, 1980-2012, GDP per hour overall and of bottom 90% (Autosaved)

By this measure, overall GDP per hour of work in France was similar to that of the US in the 1990s, but somewhat less before and after.  Overall GDP per capita was always higher in the US over this full period (the top graph in this post), and by a substantial 20% (in 1980) to 38% (in 2012).  Yet GDP per hour worked never varied by so much, and indeed in some years was slightly higher in France than in the US.

But for the bottom 90%, income received per hour of work has been far better in France than in the US since 1983.  By 2007, GDP per hour worked was 30% higher in France than in the US for the bottom 90%.  This is not a small difference.  French workers are productive, and take part of their higher productivity per hour in more annual leisure time than their US counterparts do.

E.  Summary and Conclusions

The French economic record has been much criticized by conservative media and politicians in the US, with France seen as a stagnant, socialist, state.  Overall GDP per capita has indeed grown faster in recent decades in the US than in France, averaging 2.0% per annum in the US vs. a rate of 1.5% in France.  While such a difference in rates might appear to be small, it compounds over time.

But the picture is quite different if one focusses on the bottom 90%.  This is not a small segment of the population, but rather everyone from the poor up to all but the quite well off.  Growth in average real income of this group was substantially faster in France than in the US since 1980.  While overall growth was faster in the US than in France, most of this income growth went to the top 10% in the US, while the gains were shared more equally in France.

Furthermore, when one takes into account social services, which are more equally distributed than taxable income and which are much more important in France than in the US, as well as leisure time, the real living standards of the bottom 90% have not only grown faster in France, but have substantially surpassed that of the US.

For those other than those fortunate enough to be in the top 10%, living standards are now higher, and have improved by more in recent decades, in France than in the US.

The Rate of Economic Growth and the Budget Gap: Returning to the Long-Term Average Growth Rate Would Eliminate It

Posted on by

Long Run US GDP per Capita Growth (1870-2088) in logarithms

Larry Summers published an op-ed yesterday (appearing in Reuters, the Financial Times, the Washington Post, and probably elsewhere) in which he makes the important point that the current budget impasse is focussed on the wrong issues.  The discussion, at least as publicly expressed, has focussed on what is seen as needed to deal with the fiscal deficit and the resulting public debt.  Even the Republican attempt to end ObamaCare was ostensibly about cutting the government deficit (even though the CBO concluded that the opposite would happen, as they found that the ObamaCare reforms will reduce the deficit, rather than increase it).

Yet this focus on near term and projected budget deficits is misguided.  As Summers notes, under current policies the public debt to GDP ratio is falling, and is projected to continue to fall into the 2020s.  The recently issued Long-Term CBO budget projections indicate that while the debt ratio would then start to rise (primarily driven by expected higher health care costs), there is a good deal of uncertainty in those projections.

Specifically the CBO figures show that it would not take much, in terms of either higher revenues or lower spending, to keep the public debt to GDP ratio flat.  Higher revenues or lower spending or some combination of the two, of 0.8% of GDP over the next 25 years or 1.7% of GDP over the next 75 years, would suffice.  This is consistent with an earlier post on this blog, which showed that if the Bush tax cuts had not been extended for almost all households, the projected debt to GDP ratio in the CBO numbers would fall rapidly.

But projections of revenues or of spending are highly uncertain.  Projected health care spending has been coming down steadily in recent years, for example, in part due to the slow economy, but also in part as a result of the efficiency gains and cost reductions that the ObamaCare reforms are leading to.  With these lower costs, the CBO has been steadily reducing the projected costs to the government budget from Medicare, Medicaid, and other such health programs.  In the recent CBO report, for example, the projections of government spending on health care programs in the 2030s were reduced by 0.5% of GDP from what the CBO had projected just one year earlier.  Going back further, the CBO projections for government spending on health care in 2035 were over 1% of GDP lower in the projections recently issued than in the projections published in June 2010.

This should not be interpreted as a criticism of CBO.  Their projections are probably the best available.  Rather, the point is that these projections are inherently hard to do, and the uncertainty surrounding them should not be ignored.  Yet the politicians often ignore precisely that.

Furthermore and perhaps most importantly, the projected budget deficits and resulting public debt to GDP ratios depend critically on the rate of growth of the economy.  The CBO uses a fairly detailed and reasonable model to project this (based on projected labor force growth, investment in capital, and productivity growth).  However, it is probably even more difficult to project GDP than to project future spending levels and tax revenues for any given level of future GDP.  But Summers notes the critical sensitivity of the projected future deficits to the projected growth in GDP.  He states “Data from the CBO imply that an increase of just 0.2 percent in annual growth would entirely eliminate the projected long-term budget gap”.

One can calculate from the data made available with the CBO report their projected growth of real per capita GDP.  For 2013 to 2088, it comes to 1.60%  year.  A previous post on this blog noted the remarkable constancy of the rate of growth of real per capita GDP since at least 1870 of 1.9% a year (or 1.87% a year at two digit precision).  That earlier post noted that real per capita GDP in the US, despite large annual variations and even decade long deviations (such as during the Great Depression, and then during World War II), has always returned to a path of 1.87% growth since at least 1870.  That path even did not shift when there were even substantial deviations, such as during the Great Depression.  Rather, the economy always returned to the same, previous, path, and not one shifted up or down.

This is truly remarkable, and no one really knows the reason.  The path can be seen as a trend growth of capacity (based on labor available and capital invested, coupled with the technology of the time), but why this should path should have grown at 1.87% a year in the late 1800s; in the early, mid and late 1900s; and all the way into the 21st century, is not known.

Since we do not know why the economy has always returned to this one path, we need to be careful in looking forward.  Still, it is noteworthy that the CBO projections imply that the economy will now slow, to just 1.60% real per capita GDP growth over the next 75 years.  This CBO path is substantially lower than the path of 1.87% growth that has ruled for the last 140 years in the US.

The graph at the top of this post shows the path of GDP per capita projected by the CBO (which one should note is a year by year projection, which just averages out to 1.60% per year over the full period), along with an extension of the 1.87% path that has ruled since at least the 1870s.   The graph is adapted from my earlier post (although now converted to prices of 2005 whereas the earlier one was in prices of 1990; this does not affect the rates of growth).  It is expressed in logarithms, since in logarithms a constant rate of growth is a straight line.

It is not clear why there should be this deceleration to 1.60% from the 1.87% rate of growth the economy has followed over the last 140 years.  Mechanically, one can ascribe the deceleration to what the CBO assumes for the rate of growth of technological progress.  But projecting growth in technology over a 75 year period is basically impossible, as the CBO notes.

The deceleration over the next 75 years has a very important implication, however.  The CBO found in its sensitivity analysis that a rate of growth that is just 0.2% faster will suffice to close the budgetary gap, even if one does not take any new measures to raise revenues or cut government spending.  Hence a return to the previous historical growth path of 1.87% a year from the 1.60% rate the CBO projects, or a difference of 0.27%, will more than suffice to close the budgetary gap.

The policy implication is that with such sensitivity to the growth in GDP, we should be focussed on measures to raise growth, rather than short term budgetary measures that will act to reduce growth.  The economy has suffered from government austerity since 2010, which has held back growth.  Government has been cutting spending, thus undermining demand in an economy with high unemployment and close to zero interest rates, where more labor is not employed and more is not produced because the resulting products could not then be sold due to the lack of demand.  As an earlier post on this blog noted, if government spending had been allowed to grow simply at the historic average rate (and even more so if it had been allowed to grow as it had under Reagan), the US would by now be back at full employment.

Over the medium term, Summers notes that both conservatives and liberals agree that growth should be raised, and on the types of measures which should help this.  More investment, both public and private, is required rather than less.  Research and development, both public and private, is important.  More effective education is also required.

I would agree with all of these.  But to be honest, since we do not really understand why the economy always returned to the 1.87% growth path over the last 140 years, it would not be correct to say we can be sure such measures will be effective.  However, what we can say with confidence is that measures that hinder the recovery of the economy, as the government spending cuts have been doing, will certainly hurt.

The Big Squeeze on Government: Consequences of Baumol’s Cost Disease

Posted on by

Government Share of GDP and Baumol's Disease, 1952 to 2012

A.  Introduction

A point on which all agree, whether conservative or liberal, Republican or Democrat, is that the cost of government keeps rising.  Whether it is the cost of building new roads or new military jet fighters, or the cost of schools or health services, the cost now is much more than in the past.  And this is not simply general inflation.  The cost of government services has risen at a significantly faster pace than general inflation.

This is true.  But what is not generally recognized if the fundamental cause, nor the implications as we as a nation have struggled to maintain government services.  The fundamental cause is not waste and corruption, nor lazy government workers.  Rather, it lies in the nature of the goods and services used for the public services the government provides.

This blog post will first review the facts on what has happened to expenditures on government goods and services (which for brevity, will hereafter often simply be referred to as government goods) over the past 60 years.  The 60 year period is taken so as to encompass most of the post-World War II period, but to begin once the numbers had stabilized from the very high levels during the war and the immediate post-war fluctuations.

The post will then review the fundamental cause, drawing on the work that has come to be called “Baumol’s Cost Disease”.  The post will discuss how this applies to the government sector, and the implications.

B.  The Share of Government Expenditures in GDP

The share of government spending in GDP has declined over the last 60 years, from almost 25% of GDP in 1952 to less than 20% in 2012, a fall of a fifth.  It is shown as the blue line in the graph at the top of this post.  [Note:  The definition of “government spending” used here is for government as it appears in the GDP demand accounts.  It includes all level of government – federal, state, and local – but only includes direct government spending on goods and services.  Hence it excludes government transfers payments, such as for Social Security or farm subsidies.  Transfer payments are spent by those receiving the funds.]

A fall of a fifth is a significant reduction in the government share.  But it does not show the true extent of the fall, as such GDP share calculations are based on the prices of each given year.  One also needs to know how much one received in real terms for what was spent, and this depends on how prices have changed.

The GDP accounts issued by the BEA do include estimates of the changes in the relative prices of the different components of the GDP accounts.  Over the 60 years from 1952 to 2012, the GDP deflator (the index of inflation for all the goods and services making up GDP) rose at an annual average rate of 3.3%.  Over this same period, the deflator for government spending rose at a somewhat higher rate of 4.1% a year.  This might seem to be only modestly higher, and for a short period it would be.  But compounded over 60 years, this difference in inflation rates cumulates to a difference of 58% in the prices of goods and services used for government vs. goods and services used in overall GDP.

With this relative price change, it now (in 2012) on average costs 58% more (compared to 1952) to produce goods to be used for government expenditures, than it does to produce goods for overall GDP.  Since GDP is also our income (i.e. what we as a nation receive for what we produce), it takes a higher share of our income today to buy the same real goods used for government expenditures as it would have at the relative prices of 60 years ago.

Put another way, to get the same real goods used for government expenditure in 2012 as one would have gotten at the relative prices of 60 years ago, one will now have to spend 58% more.  Or if one spends the same dollar amount adjusted for general (GDP) inflation, one will receive only 1/1.58 = 63% as much.

This impact is huge.  We are indeed receiving far less now in government services for a given dollar expenditure than we would have at the relative prices of 1952.

One way to view this is to ask how much would we have spent as a share of GDP in 1952, for the same real level of goods used for government in that year, if the prices then were instead the relative prices we had in 2012.  The result is the red line in the graph at the top of this blog.  The same goods used for government in 1952 would, at the prices of 2012, have been equivalent to expenditures of 39% of GDP in that year.

Over time, this red line then fell.  It fell in part because the share of GDP used for government (in the contemporaneous prices of each year) was reduced over time and by a fifth by 2012, but more importantly also because the relative prices of the goods used for government provided services rose by 58% over the period.  The government share fell until it reached just 19 1/2% of GDP in 2012.  That is, correcting for the fact that prices of goods used for government were rising (relative to other prices) over time, in addition to the cut-back in the share at contemporaneous prices by a fifth, real government expenditures in terms of GDP share were only half as much in 2012 (19 1/2% of GDP) as what they were in 1952 (39% of GDP).

This fall by half is huge, and explains why we seem to get less and less from our government expenditures (whether on roads, or for military equipment, or in schooling), even though government spending as a share of GDP only fell by a fifth when measured in the current prices of each year.

Another way to look at this would be to ask what government spending would be now, if it had been allowed to grow over the 60 years at the same pace as GDP grew.  If there had been no relative price change, then at equal rates of growth the share of government in GDP would not have changed.  But with the relative price changes, a higher share of GDP would have been spent on government to provide such a bundle of goods for government. The fact that we spent less than that is a measure of how much government spending has been squeezed.
The result is the green curve in the graph at the top of this post.  It shows what would have been spent on goods used for government in each year, if government spending had grown at the same rate as GDP in that year, and valued in the prices of each year.  From 1952 to 2012, real GDP grew at an average rate of 3.1% a year (note this is total, not per capita, GDP).  Real government spending grew only at a rate of 1.9% a year over that period.  Cumulated over 60 years, the difference in growth rates meant that GDP grew by a total of twice as much as government did.  And the goods used for government provided services would have totaled 39% of GDP in 2012 at this constant growth share, or double the 19 1/2% of GDP actually spent in 2012.
[Note:  It is not a coincidence that the 39% of GDP in 2012 on the green line is the same as the 39% of GDP in 1952 on the red line.  The red line figure shows what the spending would have been at the government share in 1952 but at 2012 prices.  The green line figure projects forward this same 1952 share, leaving it unchanged relative to GDP in real terms, and in 2012 shows what this share would then have been at 2012 prices.  But the paths between 1952 and 2012 will differ, and are not mirror images.]
Whichever way one looks at it, this reduction by half in the government share is a huge squeeze on public services.  It goes a long way to explaining why our roads are so much more inadequate now compared to decades ago, with extreme congestion and poor repair.  It explains why our state universities are charging so much more in tuition, while state support has declined.  It explains why what we receive today in public services simply is not what it used to be.  But why have these costs of goods for public use risen so much faster than the cost of other goods?

 

C.  Baumol’s Cost Disease

What has come to be called Baumol’s Cost Disease (or sometimes simply Baumol’s Disease) was developed by William J. Baumol (then Professor of Economics at Princeton), together with William G. Bowen (then also Professor of Economics at Princeton, and later President of Princeton) in the mid-1960s.  They were engaged in a research project looking at why the cost of tickets to live performances of the fine arts had to rise continually, at rates above the general inflation rate, and yet still could not keep up with costs.  A recent re-statement of the issue (but with a particular focus on health care), is provided in the 2012 book by Baumol and others, titled “The Cost Disease:  Why Computers Get Cheaper and Health Care Doesn’t”.

In a nutshell, the fundamental cause of the cost problem is that labor productivity, while perhaps rising in all sectors, will not rise as fast in some sectors as in others.  The sectors where labor productivity rises relatively less fast will face increasing costs, as labor in such sectors will need to be paid more, due to competition for such labor from those sectors where productivity is rising faster.  Yet those sectors with the relatively slower productivity growth will not be able to offset that rising cost of labor with a rate of productivity growth that is as high as that enjoyed by the other sectors.  If we still want or need what the sectors with the relatively slower productivity growth produce, we will need to pay a higher relative price to cover those higher costs.

This is clear in the example of the performing arts.  A Mozart string quartet that required four performers 20 minutes to play in 1780, still required four performers 20 minutes to play in 1966, or in 2013.  Their productivity has not grown at all in over two centuries. Such performers could instead be employed in other sectors, and paid at increasing rates over time there, as labor productivity rose in those other sectors.  If they are going to be employed still to perform Mozart, they will need to be paid more, even though their productivity in playing a Mozart string quartet has not risen in centuries.

Baumol’s Cost Disease will arise whenever productivity growth in the sector being examined is less than productivity growth in the rest of the economy.  There has been a good deal of discussion recently of the implications of this in health (as for example in the Baumol “The Cost Disease” book cited above), as well as in education (explaining why university tuition has steadily risen at a pace greater than general inflation).  But it applies generally for sectors where the goods produced are labor intensive or hand crafted.

Much of the goods and services used by government for the services it provides are of this nature.  Roads, for example, are custom made for the specific site; military jet fighters (and most high tech military equipment generally) are made by highly skilled technicians in small batches of a perhaps a few dozen a year; elementary school teachers teach in classes that are similar in size now as they were 60 years ago; public health workers need to examine patients one on one; and public safety workers (police, firemen, prison guards, and other security workers) provide what they do by their direct presence; and so on.  Teachers, health care workers, and public safety workers, plus military personnel and postal workers facing similar issues, account for most public sector employees (keep in mind we are referring to all levels of government in this note).  And by its nature, the work of those in general public administration (“bureaucrats”) is also highly labor intensive.

It should be emphasized that productivity growth in the provision of government services has not been zero or negligible.  There have been efficiency gains in the government sector.  But the important point for Baumol’s Cost Disease is not that the productivity growth in that sector is zero, but rather that it is simply something less than the rate of productivity growth in other sectors.

And the nature of what government provides makes it impossible to match the productivity growth rates that one has seen most spectacularly in goods such as microchips and hence computers, but more generally in manufacturing and agriculture.  Government services, like many services, have had improvements in productivity, but at rates that simply cannot match the pace of productivity growth possible elsewhere.

Hence, because of Baumol’s Cost Disease the relative price of government services should be expected to go up over time.  This is precisely what has been observed.  There is no reason to attribute this rise in the relative price to allegations of corruption or lazy government workers.  It is of course possible that corruption and lazy workers exist, but for this to have caused the rise in the relative price over time one would need to make the case that corruption and lazy workers are not only worse now than before, but that they have become steadily worse over time.  There is no evidence that supports this.

It is also important to note that while the relative price of government services has risen over time in the past, with this also expected to continue going forward, this does not imply that we as a society will be unable to afford the government services at the higher relative price.  Labor productivity is growing, in the government sector as well as in the rest of the economy, and hence the cost in terms of labor time of the goods of government as well as this cost in the rest of the economy are both getting cheaper.  Hence we can afford to devote a higher share of GDP to government services over time, if we so choose, as the relative cost of government services rises.  And since what government provides, whether in education and health services, or infrastructure, or security and national defense, are all important, we should want to ensure they are adequately provided.

There is therefore nothing wrong for the share of government in GDP to rise over time, as Baumol’s Cost Disease will predict will happen if the services government provides are important.  They would need to be paid for, through higher taxes, but as the society grows richer from the productivity growth in both the government and non-government sectors, we can afford this.  The only problems that arise come from not recognizing this.

D.  Implications, and Conclusion

Since the implications of Baumol’s Cost Disease for government services has generally not been recognized, there are indeed problems.  There has been a tremendous squeeze on government, leading to government services that are an embarrassment for a rich country.  As the numbers above indicate, we are now spending only half as much on government in real terms as we would have had government been allowed to grow at the same pace as GDP since 1952.

Note that this is not an argument that government spending should have been twice as much in 2012 as it was.  This would have matched the real share that it was in 1952, and therefore is an indicator of how much government has been squeezed over this period, but the 1952 benchmark is arbitrary.  And with the 58% higher relative price for government goods over this period, it would be rational to try to scale back on the expenditures for the now more expensive goods.  But cutting back by half is extreme. Rather, the argument made here is that one should be making a well-considered decision at any point in time on whether particular government expenditures (whether for education, or for police, or for military jets) are worthwhile at the price of the time.  If so, one should do it.  But one should not subject total government expenditures to some arbitrary cap, and say that expenditures under that cap are fine while expenditures over that cap should not be allowed.  Since the higher prices over time (due to Baumol’s Cost Disease) reflect differential but still positive productivity growth rates, we can afford those higher government expenditures if we so choose.

Unfortunately, much of the budget discussion in recent years has focussed precisely on setting some fixed cap on government expenditures as a share of GDP.  There have been calls for such a cap directly, or indirectly by saying government revenues should be set at some cap as a share of GDP and that there should then also be a balanced budget (or a budget surplus).

For example, the Bowles-Simpson budget plan called for federal government revenues to be capped at 21% of GDP, with expenditures then set to match this.  The Paul Ryan budget plan called for federal revenues to be capped at 19% of GDP, with expenditures reduced to meet this and then to fall even further.  [Note that both of these figures are for total federal government expenditures, including transfers.  The figures in the graph at the top of this post are for government direct spending only, excluding transfers, but for federal, state, as well as local government.]

Understanding the underlying dynamics resulting from Baumol’s Cost Disease shows how misguided such constant share of GDP targets are.  They ignore that a growing economy, with a growing population, will need to be supported by growing government services.  Given the nature of government services, one cannot expect the rate of productivity growth  in government to match that enjoyed elsewhere in the economy.  There is nothing wrong with that, and does not necessarily reflect a lack of innovation or skill.  Some goods are simply more labor intensive than others, and productivity growth will generally be less for such goods.

By Baumol’s Cost Disease we can see that then the prices of the goods used for government will rise relative to others, and that if we still wish to obtain such goods, we will need to pay more.  The GDP share will rise, but we can afford it as productivity is rising in all the sectors.  They are simply rising at different rates.