Why the Correct Answer to “What Should we do About Slowing Growth?” is “Nothing.”
There has been a lot of agonizing lately about slowing economic growth. Some of it understandably laments the slow recovery from the Great Recession, but the more interesting part concerns long-term growth prospects. Brad DeLong has recently written a thorough review of the stagnation literature under the title “Is Economic Growth Getting Harder? If so, why, and what can we do about it?”
DeLong is a little less pessimistic than some of the others. He thinks the growth of real GDP per capita will slow only by three-tenths of a percentage point from its historic norm of 2 percent. But, more importantly, he suggests an entirely different way of thinking about growth policy—one that changes the question from “What can we do about slow growth?” to What should we do about it?” If we follow DeLong’s reasoning to its logical conclusion, I think the correct answer is that we should do nothing.
The rest of this post will highlight a couple of DeLong’s key themes and flesh out some of his generalizations with relevant data.
GDP is the wrong indicator
The first point is that real GDP per capita, or per worker, is the wrong indicator. As DeLong puts it, “What we are really interested in is material well-being, what people want and need, and whether they actually obtain it. We are interested in material production as a springboard to human flourishing.”
The point is hardly original, but it is important. Let me show why by using data from the Social Progress Index (SPI), which is the latest and most ambitious effort to attach specific numbers to the concept of “human flourishing.” The SPI is especially appropriate for our purposes because it explicitly treats GDP as an input to human welfare and asks how well various economies transform raw GDP into outputs such as health, education, rights, and opportunities.
Our first chart plots GDP per capita against the SPI for two measures of GDP per capita, one stated in U.S. dollars at market exchange rates and the other in dollars that are adjusted for purchasing power parity (PPP), using the newest data from the World Bank’s International Comparison Program. The PPP adjustment takes into account differences in the cost of living among countries, so we might expect PPP-adjusted GDP per capita to correlate more strongly with the SPI than GDP measured at market rates. In practice, though, the difference in fit is small. The point that represents the United States is close to the trend line for both measures.
Three features stand out.
- First, the positive slope of the scatter plots shows that material production, as measured by GDP per capita, really is a springboard to human flourishing, as measured by the SPI. The overall correlation between the SPI and the log of GDP per capita is .88 for market exchange rates and 0.91 for purchasing power parity. The biggest outliers on the downside—wealthy countries with SPI scores far below the trendline—are Kuwait, the United Arab Emirates, and Saudi Arabia. (I discussed some possible reasons in an earlier post on the curse of riches.) The biggest outlier on the upside is New Zealand, which manages to achieve the top SPI score even though its PPP-adjusted real GDP per capita of $31,000 is only twenty-fourth in the world and just two-thirds that of the United States.
- Second, the relationship of GDP to the SPI is far from linear. For poor countries, even modest gains in GDP have a big payoff in terms of the quality of life. As income increases, the curve flattens out. After reaching a GDP per capita of $25,000 measured by PPP, improvements in the SPI from further GDP growth alone are modest. (Portugal and Slovakia are just above the $25,000 line, Estonia and Russia just below it.)
- Third, the relationship of GDP to SPI not only flattens as countries grow wealthier, it becomes significantly less tight. For countries below the $25,000 threshold, the correlation between GDP and the SPI is .83 but it is just .60 for countries above it. If we interpret those correlation coefficients in terms of R2s, we could say that for the poorer countries, GDP “explains” about two-thirds of differences in SPIs, but for the wealthier countries, only about one third.
We can explore the relationship between GDP and human flourishing further by breaking the SPI down into three main sub-indexes:
- Basic human needs, which includes measures of nutrition, maternal mortality, access to electricity, sanitation, availability of affordable housing, level of violent crime, and traffic deaths, along with several related indicators.
- Foundations of wellbeing, which covers adult literacy, school enrollment, access to information (mobile phone subscriptions, internet users, press freedom), life expectancy, pollution, biodiversity, and other indicators.
- Opportunity, which includes indicators of political rights, personal freedoms, property rights, low corruption, tolerance, access to advanced education, and related issues.
The next chart plots each of the three sub-indexes separately against GDP per capita at PPP. The general pattern is the same for each of them as for the SPI as a whole: Each sub-index rises with increasing GDP; the payoff to added GDP decreases as countries get richer; and the correlation between GDP and noneconomic indicators gets looser as income rises.
This chart has two additional features that are not evident from looking at the SPI as a whole:
- The United States, which lies close to the trendline for the SPI in our first chart, turns out to perform below the trend in converting GDP to satisfaction of “basic human needs” and “foundations of wellbeing,” but well above the trend for the “opportunity” sub-index.
- Exponential trendlines, as used in the first chart, provide a reasonably good fit for “foundations of wellbeing” and “opportunity,” but not for “basic human needs.” For that sub-index, a quadratic trendline provides a better fit. The satisfaction of basic needs peaks at a GDP per capita just over $40,000 and falls thereafter. The best-performing country on the basic needs sub-index is Denmark, which ranks eleventh in terms of per capita GDP in our sample of 127 countries. Japan, which ranks twentieth in GDP, is only one index point below Denmark.
Let’s turn now to another of DeLong’s key points. He is concerned that the secular stagnation literature is addressing the wrong questions when it comes to policy. “The focus on real GDP growth and its possible—or likely—slowing,” he says, “is a setup to panic us into making policy decisions we really do not want to make.” “Whether we should tradeoff potential growth for other objectives,” he continues, “is not something that depends on how fast growth is.” Whether we think the underlying trend of growth is slowing or not, policy should not be focusing on growth for growth’s sake.
That certainly makes sense to me. To see why, let’s zoom in on a small inset from our second chart:
What we’re looking at here is the position of the US economy in terms of real GDP per capita at PPP (horizontal axis) and the sub-index of the SPI called “foundations of wellbeing,” (vertical axis), as indicated by the red points and red trendline. The United States is below the trendline—an underperformer. Consider two policy strategies for improving the situation.
Strategy A would consist of an all-out effort to accelerate growth. If it succeeded in doing that without worsening the efficiency with which the US economy converts GDP inputs into wellbeing outputs, the policy would move us parallel to the red trendline, eventually passing close to the point that represents the United Arab Emirates. That would entail roughly a 25 percent increase in real GDP per capita (measured at PPP) and an increase in our “foundations of wellbeing” score from 75.96 to 76.74. If Strategy A managed to increase the rate of growth from DeLong’s forecast of 1.7 percent to the 2 percent historical norm, we would catch up with today’s UAE in just 12 years (Presumably, the future UAE would have moved on by then.)
Strategy B would aim for Australia rather than for the UAE. Australia is not a superstar in the wellbeing sweepstakes. It ranks just eighteenth out of the 127 countries in our sample, but that is better than the United States, which ranks thirty-seventh. Superstar or not, the important thing is that Australia does better than the United States at converting raw GDP into human welfare. Its “foundations of wellbeing” score is 80.27 compared to 75.96 for the United States despite the fact that its per capita GDP at PPP is 17 percent lower.
A look at the components of the “foundations of wellbeing” sub-index shows that Strategy B would require policy initiatives in areas like improving the quality of elementary and secondary education, upgrading the Internet, and cutting air pollution. All those things would require resources, which would mean either raising taxes, or diverting private investment away from building shopping malls to building schools, or some combination of the two. It might be a good idea to throw in a carbon tax to cut pollution while raising revenue for other parts of the package.
Critics might say that the cost of those initiatives would slow the growth rate. Maybe or maybe not—some components of the package, such as improved education and better information infrastructure, could actually enhance growth. Still, for the sake of discussion, suppose that Strategy B cut the growth of per capita GDP to just 1 percent a year, and that it took 12 years for the program to raise the US economy’s efficiency in converting GDP to “foundations of wellbeing” to the Australian level. Even so, the United States would reach the trendline at a point where it was still ahead of Australia in terms of GDP per capita.
So here is the choice: After 12 years of Strategy A, which focuses on growth, we would gain slightly less than one point on the “foundations of wellbeing” index and catch up with the UAE. After 12 years of Strategy B, which focuses on directly on wellbeing, we would add five points to the same index and catch up with Australia, with room to spare. Which outcome is more attractive? You can take your choice, but I would definitely choose Strategy B.
The bottom line
This brings us back to our original question of what we should do about a possible slowdown in GDP growth. The answer is, “Nothing.”
In that respect, I agree completely with DeLong, who says that policies that make sense if underlying GDP growth is 3 percent probably still make sense if underlying GDP growth is 1 percent, while policies that don’t make sense at 1 percent still don’t make sense if underlying GDP per worker growth is 3 percent.
Let me make three caveats:
- First, my position is not anti-growth. Rather, it takes a laissez fair stance toward growth. I am not among those who think we can save the planet only if we stop the economic machine in its tracks. Advances in technology, along with growth of human and physical capital, make it possible to invest more in all kinds of welfare –enhancing activities, including, if we choose to do so, protecting and restoring the natural world. It is simply untrue that stopping growth is the only way to save the environment.
- Second, saying that we should focus on welfare-enhancing policies leaves a lot of room for discussion about which policies best enhance welfare. The Social Progress Index is a grab bag of every good thing its developers could think of. Clearly, peoples’ priorities differ. Some would prefer a focus on health, others on education, and still others on personal and political freedoms. The only point I insist on is that the proper guide to policy is some rationally weighted concept of human welfare rather than the mindless assumption that faster GDP growth will automatically take care of everything.
- Third, I would emphasize that the do-nothing conclusion of this post applies to the long-term growth trend, not to short-term cyclical movements. Whatever the long-term trend, we still need intelligent countercyclical policy for the short run. When the economy is operating below its potential, as the US economy has in recent years, we should take prudent measures to accelerate the return to normal and cheer when those policies succeed, just as we should take prudent measures to avoid overheating during a boom.
The bottom line is that policies that improve human welfare always make sense. Policies that sacrifice human welfare to achieve growth for its own sake never do. Yes, there is a positive correlation between growth of real GDP per capita and broad measures of human welfare, but the relationship is much weaker for rich countries than for poor ones. For high-income countries, strategies that focus on improving the efficiency of transformation of GDP into wellbeing are far more attractive than policies that focus on growth and hope that wellbeing will take care of itself. So, I say, go for welfare-enhancing policies and let growth come out however it may.
15 Responses to “Why the Correct Answer to “What Should we do About Slowing Growth?” is “Nothing.””
An interesting column as always. I look forward to reading DeLong's original.
Picking up on your second caveat, an issue that has been on my mind as workforce participation dwindles and social programs fill gaps is: what happens to a society when significant portions of it don't contribute to the care and maintenance of the "commons" through obvious mechanisms like work? Part of what makes a society cohesive is shared goals and efforts to reach those goals. Automation and off shoring can keep GDP moving (so long as people have money with which to consume) but will other weaknesses appear in the social fabric that may be more difficult to mend? Maybe time for me to reread The End of Work.
Falling labor force participation is an important issue in this context, but each of its components needs separate treatment. For example:
It would not be appropriate, in my view, to respond to the part of falling LFP that is due to aging (probably the largest part) with policies that try to force the rest of us to work harder than ever, but we should look critically at things that make it hard for people to save for retirement
Another big chunk of falling LFP is due to our broken disability system, which is misused as a form of long-term unemployment insurance, but has the terrible side effect that once you get on disability, it is almost impossible to get off. Fixing that would be a real welfare-enhancing policy initiative.
Still another large piece of falling LFP is due to younger people spending more years in school. That may be a change that both improves welfare and helps maintain GDP growth.
I agree with the general line of thought, though I observe that various forces in the economy are very motivated to see things that add up to GDP growth. So while GDP as a number may not matter to good policy, the policies that make good GDP as a number matter a great deal to politics.
All that said…. I have 2, uh, "issues"
1. I think it misleading to compare the US with other countries, most with much smaller populations. How does the US do versus the EU? Or, how do, oh, Washington state or Iowa do in the charts? Saying that New Zealand does a good job converting GDP to SPI is less exciting when you find it's population appears to be smaller than Chicago.
2. There will per force be bias in the measurement of SPI – so one might claim that the right choose an insurance company is a freedom, but also that the right to government funded universal care is a form of basic well being – can't be both at once. That does't make the argument, or especially the line of thought go away. But it does means that "rational" and "fair" may not really apply.
Thanks for your views. Some comments:
1. You say it is misleading to compare the US with other countries, most with smaller populations. I'm not quite sure what you are getting at here. I take it you noticed that all my GDP numbers are per capita, so it can't be just that you have in mind. Evidently, you are suggesting that there are diseconomies of scale in providing good economic policy. Maybe there are. For example, some people have proposed that factors like cultural homogeneity, higher urbanization, politics closer to the people, and so on make countries more responsive. Also, in the US, many people seem to think that it is generally good, where possible, to devolve government functions from Washington to the states. I have also seen this argument being made by advocates of Scottish and Catalan independence. I would be interested in your further thoughts on this theme. Maybe I will do a post on it sometime.
2. You note that there are different ideas of what should go into an SPI or how the items should be weighted. I fully agree. Even the three sub-indexes I mention are too broad. You should check the list at the SPI website. For example, the third category, opportunity–the one where the US does best–is itself a very mixed bag. It includes things like tolerance for gays, religious minorities, and immigrants. Yet there are countries like, say, Russia or Saudi Arabia, where these features are not only not recognized as virtues, but are, as a matter of official policy, treated in a negative light.
I hadn't really considered the broken disability system and your point is well taken – and pretty disturbing.
I'm still looking for convincing evidence that the preponderance of workers leaving the workforce are doing so because of an active wish to retire rather than retirement forced by inability to find work. I'm not even entirely convinced that this demographic accounts for the preponderance of non-workers but accept it provisionally.
I hate to sound pessimistic because I'm not, really. But many young people seem to be staying in school because of poor job prospects and will exit university with very expensive credentials that may not prove to be good investments.
I've been working my way through Piketty and have been thinking more and more about what real increases in per capita GDP going forward means and the ways it might be achieved and sustained – and if that is an unalloyed good thing. If any readers here have suggestions for further reading in this pursuit, I'd love to see suggestions.
"can't be both at once."
Perhaps it can. Imagine a model along the lines of Medicare where a universal single-payer plan covers basic care and supplemental insurance in a variety of flavors (and price points) covers catastrophic and extended care. This attempts to balance the utilitarian impulse to, as you say, ensure basic well-being while recognizing that health care is a scarce resource faced with essentially insatiable demand.
Some of the best research I have seen on labor force participation issues has come out of the Atlanta Fed. Here are two links worth looking at:
Many thanks. JT
[…] came across an interesting post by Ed Dolan, on what we should do about slowing growth in the U.S. His answer is “Nothing”, and he […]
Ed – neat post. As a growth economist, I like to think about this in terms of the marginal utility of goods and services (GDP) versus the marginal utility of intangibles (some items in the SPI). I just put up a blog entry inspired by your post: http://growthecon.wordpress.com/2014/05/29/who-ca…
I would write a longer email, but I cannot find your email address. I will try to explain what I mean here, I think (obviously) that it is worth thinking about when comparing policies.
You list "cultural homogeneity, urbanization" and other issues. I think ethnic structure and simple geography should be added, as well as certain aspects of recent history.
Likewise, we compare say Germany or France to the whole US, but neither of them have any of the problems of scale and heterogenity the US has. It's observed history Germans and French don't want a combined capital, yet it is vastly farther from DC to Phoenix than from Berlin to Paris. The climate and geography are more different. Yet we often compare the whole US to just Germany, just France, just the UK.
I claim it makes more sense to compare, say, Chicago to Denmark than to compare the whole US to Denmark.
Why should this matter? I think population scale alone and geographic scale alone matter, especially when multiplied across the hetergenous religious, economic, ethnic, etc. structures of the US.
As a further, it may more practically inform US policy to compare Chicago or Nevada to various countries and come away with "well, here's an argument that Houstan's policies are better than those of San Francisco for issue X" Can probably get further with that than claiming that "well Sweden does this better than the whole US".
Put another shorter way – National boundaries are largely accidents of history, why do we think they are the good determinants of spheres of policy to study? And given that on-the-ground effects of policy in the US are always a mix of federal/state/local, why do we think that federal policy alone is the right thing to compare?
The combined effects of at least some parts of federal policy are different in say Alabama than Minnesota. I'm not persuaded that these effect differentials are so pronouced throughout France or Sweden.
Thanks, I invite all readers to follow up on Vollrath's post as linked in his comment. If you do so, you can also read my comment on his post and add your further comments either here or there.
First, for the the benefit of BW and all readers, you can e-mail me via this contact:
If you forget the URL, you can always find it by clicking the "contact" button on the page that lists the posts for me blog (not as easy to find as it should be).
As for the substance of your remarks, I agree that there are some purposes for which comparison of US states with European countries would be appropriate. For example, I remember not long ago getting into a discussion with an American ex-pat living in Sweden, who was telling me how much better many indicators of the good life were in Sweden (he was talking about things like education, crime, and health). I suggested that the real comparison ought to be between those indicators for Sweden, and those indicators for the State of Minnesota, which has a large population of descendants of Scandinavian immigrants. I never did the numbers, but my intuition tells me that many of them would be closer to the Swedish norm than to the average for all of the US. Sounds like an idea for a future post.
Thank you for interesting perspective. Valuable material to ponder about for policy makers.
I don't think there is a trade off between SPI and GDP. Resources not spent on military purposes but instead e.g. on education has no effect on GDP while may have substantial positive effect on SPI in case sub-index related to education needs were downsized before.
I perceive relation between nation's SPI and GDP as an indicator of economy structure efficiency, namely how well-structured is economy in terms of social expectations. Relative high SPI rank and low GDP would mean the best 'return on economy", i.e. for every dollar spent social satisfaction is highest.
So countries with relative high GDP and low SPI may be inhabited mostly with onery, frustrated owners of BMW cars and luxury yachts. This community will have realtive high score in monetized wealth and terribly low in terms of unmonetized wealth.
I don't think we should do nothing with slowing growth. I'd rather recommend to implement full employment policy as it would increase both SPI and GDP. SPI could be a fantastic indicator for a policy maker were currently unemployed human resources should be directed. Does it make sense ?