Did We Evolve the Capacity for Sustained Growth?

NOTE: The Growth Economics Blog has moved sites. Click here to find this post at the new site.

I posted a few pieces (here and here) recently on genetics and growth. The Economist even picked up on Justin Cook’s work on lactose tolerance and development. Justin’s work on both lactose and the HLA system are about very specific genes, while the other research I mentioned is about genetic heritability of certain behaviors associated with growth, without specifying any particular genes.

There is another line of research on evolution and growth pioneered by Oded Galor and Omer Moav. They propose that natural selection over different types of individuals could have led to the onset of sustained economic growth. In particular, they focus on selection over preferences for the quantity and quality of kids. This is very much the second kind of research I mentioned above; it does not identify some specific gene that matters for growth, it suggests a mechanism through which selection could have operated. The original paper is linked here, but they have a nice summary article here that explains the logic without all the math.

Let’s be careful about terminology here. Evolution in general requires both mutation and natural selection. GM is really about natural selection, not mutation. They take as given the presence of two types of people in the population. “Rabbits” like to have large families, but do not invest much in their kid’s human capital. “Elephants” have a few kids, but invest a lot in those kids. Their theory is about the proportions of those types change over time due to economic forces, and eventually how a rising prevalence of Elephants leads to a speed-up in technological change. Yes, at some point there must have been a mutation that led to the differentiation between the types, but we can think of that as happening well back in history. They don’t propose that some mutation occurred at some specific year or a specific place to make this all work.

How does the underlying logic work? In the early Malthusian period, with very low income per capita, the Elephants actually have the evolutionary advantage. Why? In the Malthusian world, everyone is so poor that higher income leads to higher fertility no matter your type. Each Elephant kid has high human capital, and thus relatively high fertility compared to Rabbits. So the proportion of Elephants tends to increase in the population. And a higher proportion of Elephants means that average human capital is rising over time.

As the human capital rises, so does the pace of technological progress. At first this doesn’t do much, as the growth of technology is not sufficient to overcome the force of Malthusian population pressure. But eventually there is high enough human capital that technological change happens so rapidly that people reach the upper limit on fertility rates, and choose to spend any additional income on increasing their kids human capital rather than having more kids. This is the tipping point where human capital and technological change go into a virtuous cycle. Higher human capital leads to higher technological change, which leads to higher human capital, etc.. etc.. and you have sustained growth. Once this occurs, the relationship of income and fertility flips to become negative – the richer you are the fewer kids you have, just the opposite of the Malthusian period. This flip in sign is not unique to their explanation based on natural selection, the same type of flip is central to the general unified growth model in Galor and Weil.

After this transition point, the evolutionary advantage also flips to Rabbits. Why? Because the fertility rates decline with income, and as Elephants are richer due to their human capital, they have fewer kids than Rabbits. So Rabbits begin taking up a larger and larger proportion of the population. But everyone is already relatively rich, so this doesn’t mean that human capital levels are low generally. There is sufficient human capital to sustain technological progress.

Do we know if this exact mechanism is what generated sustained growth? No. To establish that you’d have to identify the precise genes that govern preferences for quantity/quality of kids and show that they varied within the population over time in a manner consistent with the GM model. But there are little bits and pieces of circumstantial evidence that work for GM. Greg Clark’s Farewell to Alms documents his research showing that in fact richer families tended to have more kids in pre-Industrial Revolution England. This fits with the selection mechanism proposed by GM. Similarly, Galor and Marc Klemp have a working paper out on the reproductive success of families in 17th and 18th century Quebec (a place and time with particularly detailed records), and the data shows that it was families with moderate fertility rates that actually had the most kids in subsequent generations, not those with the higher fertility rates. Again, it fits the selection mechanism proposed by GM for the Malthusian era.

Note that even if it isn’t true genetic differences in preferences for quantity/quality, you still need to have selection working for population composition to matter for sustained growth. Let’s say that quantity/quality preferences are purely cultural, passed on from parents to kids imperfectly but with some fidelity over time. Then the GM mechanism could still hold up, but it would be the cultural spread of preferences for high quality that generated the take-off, not the spread of specific genes.

There are reasons to be skeptical about this explanation, just as you should be skeptical about any hypothesis. But don’t dismiss it on the basis that natural selection moves far too slowly for this to have mattered for human populations. Galor and Moav have a number of very telling examples regarding the speed of selection within populations over just a few generations. The classic story is peppered moths during the Industrial Revolution. Peppered moths tend to be white, with little black spots on them – hence the name. But there are black varieties. With the rise of coal in the UK black moths became far more prevalent, as they were harder to spot for predators against the blackened sides of buildings. Within a few years the population jumped from predominantly white to predominantly black. And then flipped back to white when clean air regulations came into force. Given the variation in the population already exists, natural selection can take place very quickly to change population composition. So imagining that human population composition could change substantially over hundreds or thousands of years is reasonable.

Last, does GM mean that generating growth in poor countries is doomed to failure because their genetic composition is “wrong”? No. GM is a story about the rise of sustained growth at the global level. Suggesting that poor countries need to get their genetic mix right in order to grow is like suggesting that they need to adopt steam engines and telegraphs before they can step up to gas engines and mobile phones. The question of how to catch up to the frontier is an entirely different question than explaining how we got a frontier in the first place.

Tuesday Growth Links

NOTE: The Growth Economics Blog has moved sites. Click here to find this post at the new site.

Non-economics note: I finished the Souther Reach Trilogy (Annihilation, Authority , and Acceptance) by Jeff VanderMeer. I had been worried about a “Lost” situation, where the story doesn’t tie off nicely at the end, and….the books do not tie off nicely. BUT the writing is hypnotic, and the stories weave together so nicely, that I still recommend the books. If you cannot handle books without neat answers to their mysteries (and I normally cannot) then you should be careful of these books. But if you can give that kind of story a chance, these are a great place to start.

On to links that have been piling up over the last two weeks:

• Great visualization of the spread of mega-cities over time. When you go to the site, notice that large cities appear first in relatively rich places (Europe, North America) and then slowly spread across the rest of world. Now, most mega-city growth is in particularly poor countries. Remi Jedwab and I have a paper we are working on right now regarding this rise of poor mega-cities. We link it to the change in mortality rates within cities after World War II. Historically, cities were deadly, and their growth was muted by the awful conditions. But with the epidemiological transition, cities became in many ways healthier than rural areas, meaning explosions of population growth, which ultimately let congestion outweigh the positive agglomeration effects of cities.
• Actual data on the effect of robots! VoxEU post by Guy Michaels and Georg Graetz. They build a new dataset of information on industrial robots in use in 17 countries (OECD) by sector. They find that robot use is associated with higher labor productivity, wages, and total factor productivity, but no effect on labor’s share of output. They also find that robot use lowers the employment of low-skilled workers, and only a marginal effect on medium-skilled workers. They are studying industrial robots, and not necessarily the idealized general-purpose robot that seems to be the big worry of some, so their results are not immediately applicable to the future. But finally studying this in the data is a huge step. Original paper is located here.
• Tim Harford on Luddites and their modern equivalents. A nice explainer of how Luddites were not anti-technology, and did not think that technology would result in aggregate loses of jobs. They were worried about losing their market power as skilled artisans.
• This was floating around a lot on my Twitter feed. There was a severe bottleneck in the Y chromosome some time around 4-8 thousand years ago. What does that mean? It means the diversity of Y chromosomes in the human population dropped remarkably in that period, indicating that a relatively small number of men were fathering most children. The diversity of Y chromosomes recovers in most areas in the centuries that follow, indicating that more men are having children. So was it the onset of settled agriculture that led to this bottleneck, with a few males at the top of early agricultural civilizations able to dominate the pool of available females? Was early agriculture as bad for living standards, as has been suggested, so that many men were not healthy enough to have kids or have kids that survived?
• Dated (from three weeks ago) but excellent post by Cardiff Garcia on the long lags between technology introduction and the effects on labor markets, using “Engel’s pause” in the 19th century to illustrate. Strongest point made here is that we have so few points of evidence regarding the effect of massive technology shifts on labor markets that trying to say anything firm about robots, AI, or anything else is almost impossible.
• Tim Taylor on Paul Rubin on the mis-use of the idea of competition. The economy involves far more cooperation (implicit or explicit) than we like to give it credit for. Perfect competition is a non-existent, theoretical construct that is useful when writing models and you want to avoid talking about irrelevant things. But that doesn’t mean it is how things actually work, or how they should work. It definitely isn’t true that perfect competition would necessarily make an economy richer in the long run.
• Many poor people in developing countries are poor in part because they live on really poor agricultural land. About 1.3 billion people are on what Edward Barbier and Jacob Hochard term less-favored agricultural land. Which of course leads to the big question regarding development. Is it better (or even possible) to improve that land, or is it better (or even possible) to get those people to leave those less-favored areas? If it’s the former, then your concern is more with technology and input provision (fertilizer, etc..). If it’s the latter, then your concern is more with property rights and compensating people who have to adapt to additional farmers showing up in their favored areas.
• A good economic smack-down on trying to use market-based arguments against paying college athletes.
• Kindle-to-Evernote script. Simple Python script that will suck up your Kindle highlights when you plug it into your computer and e-mail them to your Evernote account, organized nicely by book. Small tweaking necessary/possible to get things to your liking. But I cannot tell you how much I love this script. I want to give it a great big hug for saving me from having to go to my Amazon account all the time for notes.

Great Britain and Laissez Not-so-Faire Economics

NOTE: The Growth Economics Blog has moved sites. Click here to find this post at the new site.

I recently finished State, Economy, and the Great Divergence by Peer Vries. It’s a comparison of the activities of the state in Great Britain and China in the period running up to and including the Industrial Revolution, roughly 1650-1850.

Vries critiques the standard view on the role of the state and the divergence between these two places, encapsulating that view in the following:

In the Smithian interpretation of British economic history, that fits in quite neatly with the Whig interpretation of Britains overall history, the primacy of Britain and its industrialization are by and large regarded as the culmination of a long process in which Britains economy increasingly became characterized by free and fair competition and in which government increasingly tended to behave according to Smithian logics.
….
For those who endorse them, the predicament of imperial China, that it did not industrialize, has always been quite easy to explain. They only need to refer to the fact that China was characterized by some kind of oriental despotism. This notion has a long pedigree whose beginnings can be traced back at least to Marco Polo.

The alternative that Vries proposes is that China is far more “Smithian” than Great Britain in this period, in the sense that it operated a very hands-off government that mainly served to provide some subsistence insurance to its population, while Great Britain had a relatively large, intrusive, and active government managing its economy and actively interfering in the process of industrialization. With regards to the idea that Great Britain enjoyed a meaningful advantage in institutions (re: property rights) after the Glorious Revolution, Vries has this to say:

I, moreover, see no concrete direct links between changes in property rights and the emergence of modern economic growth during industrialization in Britain, or rather I do not see any major changes in that respect just before and during take off. In several respects property rights in Britain after 1688 were not better protected, as a strengthened central government had acquired more power to interfere with them on the basis of national interest. More in general, one has to realize that, as will be discussed later on, the history of Western Europe was not exactly lacking examples of expropriation and that well protected, entrenched property rights including patents can also be an obstacle to growth.

Vries then spends a good portion of the rest of the book laying out the evidence on government expenditures, taxes, employment, and transfer payments to support the idea that Great Britain had a much more intrusive state than China in this period.

I’ll leave you to the book for the full details, but here are some essential highlights. Taxes per capita in Great Britain were approximately 20 times higher than in China. As a percent of GDP, the figure depends on exactly your preferred source for GDP data, but taxes were again much higher in Great Britain (3-5 times higher depending on the measure). Further, taxes were rising in both per capita and percent of GDP terms over this entire period in Great Britain, while they were essentially flat in China. Finally, the government in China never ran deficits in this entire period. If you are familiar with the history of Great Britain, then you know that government debt as a percent of GDP was essentially zero in 1689, right after the Glorious Revolution. From there it rose steadily, reaching a peak of almost 250% of GDP after the Napoleonic wars. It wasn’t until after 1850 that debt fell back below 100% of GDP. In terms of the number of government officials, Vries cites data that China had between 20,000 and 30,000 civil servants in the 18th century. Great Britain had an equal amount, for a population roughly 30 times smaller. Great Britain spent a much larger fraction of GDP on welfare and poor relief than China ever did.

Drawing on the excellent War, Wine, and Taxes by John Nye, Vries also talks about the attitude of Britain towards free trade:

In the 1820s, for example, the average tariff rate for imported manufactured goods was between 45 and 55 per cent. It was only after 1850, and even then only quite temporarily, that Britain really became a free-trading nation. Overall, its tariffs in the first half of the nineteenth century were so high, higher for example than in France, and continued to be high for so long that any explanation of the first industrial revolution by reference to the existence or emergence of a free-trade economy is extremely improbable. When Britains economy took off, the country definitely was not a free trader in matters of international trade.

Compared to Britain, China was much closer to a free trade nation, declining to interfere or promote imports or exports actively.

Vries wraps up his argument with

This book maintains that the historical evidence now is so heavily in favour of industrial and military policies successfully encouraging long-term economic development in England, admittedly through far more complex means than simply setting tariffs to encourage domestic manufactures, that the burden of proof falls on neoclassical economics, not on the historic record.

Mercantilism, as practiced throughout this period in Great Britain, was not simply a fascination with collecting gold. The British government actively looked to strengthen manufacturing (of imported raw materials) and used military and naval power to open markets with that purpose in mind. To do this it taxed heavily, borrowed heavily, and spent heavily.

What to make of this? There is no necessary link between strict laissez-faire policies and growth. The first industrial nation in the world was anything but laissez-faire, and it intervened far more deeply into its economy than China, which functioned in some sense as the idealized “night watchman” state of Adam Smith. There is little to no evidence that government “just getting out of the way” leads to development. The interventions Great Britain did make certainly resulted in massive monopoly rents to small groups of people at times. So let’s not go overboard in the other direction and conclude that massive state interventions are necessary or optimal. But it is valuable knowing just how un-laissez-faire Britain was during this period.

Why did Britain take off even with all this government interference? Vries doesn’t say this explicitly, but I think his answer is partly that large-scale industrialization has big fixed costs. I want two things before I undertake big fixed investments: a large market and low risk. The British government used the high taxes to fund a military that could ensure large markets around the world, and could ensure that those markets remained open so I could earn enough to pay off my fixed cost. That military (directly or by proxy) could also actively ensure that other markets did not develop competitive industries, again ensuring that I could earn enough to make the fixed costs worth it. Without the market size and low risk, maybe British capitalists are not willing to create the large-scale industries that drove the IR. In that sense, the large size of government was necessary to the industrialization of Britain.

Plows were the Robots of the 13th Century

NOTE: The Growth Economics Blog has moved sites. Click here to find this post at the new site.

Jury duty this morning, which meant lots of quiet reading time and in the end no *actual* jury duty (yeah for settlements!).

I am reading Rural Economy and Country Life in the Medieval West, by Georges Duby. I came across the following description of how the development of improved harnesses and plows in the Medieval period displaced a large fraction of rural labor (p. 116):

On the other hand, manual laborers without draught animals underwent no technical progress and sustained no rise in yields: on the contrary there was a relative fall in their living conditions…..That the increased value of farming equipment strengthened the hold of the wealthy over the peasantry cannot be denied….Everywhere the lord maintained his authority over his men by helping them to acquire livestock or by threatening them with its confiscation. When in some provinces in the thirteenth century servitude was born anew and flourished, it was the need to acquire agricultural equipment, efficient though costly, which led poorer peasants to bind themselves into dependence. The same needs held them in servitude, for although they had the right to decamp….they could do so only…by giving up their plough animals. In fact because of this, agricultural growth appears to have been a very powerful agent of social differentiation.

A couple of things struck me about the passage. First, the analysis of the disruption caused by the introduction of a new technology embodied in capital goods (plows, harnesses, and horses) sounds similar to some worries regarding the introduction of robots. With capital owned by only a few, those without capital become dependent on the wealthy and have their living standards driven down. Second, innovation favors those with the skills to work with the new technology. Skilled ploughmen – who only got that way by having a team of horses and a plough to begin with – were the high human capital workers of their day.

Mainly, though, it is just an interesting example of how the same issues with innovation, technology, and displacement have been occurring forever. The question of what happens when robots are plentiful is not a question unique to robots, it is a question about how we adapt to disruptive technology. The evidence suggests that whoever owns the technology or the capital associated with it will use it as leverage over those who do not, just like always.

By the way, I think the lady next to me in the jury room would have looked less shocked if I had told her I was reading a porn magazine.

Genetic Origins of Economic Development

NOTE: The Growth Economics Blog has moved sites. Click here to find this post at the new site.

I recently posted about the genetic component of savings behavior. The paper I reviewed there said that one could account for about 1/3 of variation in savings behavior by appealing to genetic differences. Whatever the authors of this study found (rightly or wrongly), they did not identify the gene(s) for savings. They identified the proportion of savings behavior that is correlated with some as-yet-unknown set of genes.

This is not atypical for a paper on economic or social outcomes and genetics. The findings support the idea that “genetics” explain some proportion of behavior, but this does not mean that we know the specific genes involved.

An entirely different kind of study is one where the researcher looks at a specific gene(s), with a known biological function, and examines whether this has a social or economic influence. I’m going to highlight two papers by Justin Cook, who has undertaken exactly this kind of research on genes and economic development.

Justin’s first paper is on disease resistance and development. There is a human leukocyte antigen (HLA) system, which is determined by a set of 239 genes. The HLA system identifies foreign pathogens so that your immune system can kill them. Within populations, there is a lot diversity in this system. That is, people vary in their alleles in the HLA system. At the population level, this is good, because this means that even if I cannot identify the pathogen (and hence die a horrific death), *your* body can identify it and survive to live another day. Populations that are very uniform in the HLA system are thus more susceptible to disease, as one bad bug (or mutation of that bug) can kill them off more effectively. So a lot of heterogeneity in the HLA system in your population is good for surviving diseases, as a population.

You can measure the HLA variation at ethnic-group levels, and then roll this up into HLA variation at country-group levels based on their underlying ethnic composition. This is what Justin does, and then looks at how life expectancy or mortality are related to it. Sure enough, Justin finds that in 1960 there is a significant relationship of HLA heterozygosity (i.e. variation in HLA alleles) and life expectancy across countries. But as you go forward in time, the relationship weakens. By 1990 the relationship has half the estimated strength, and by 2010 only one-fifth. Further, by 2010 the relationship is no longer statistically significant.

There are a couple of interesting implications of this result for thinking about genetics and development. First, it shows that genetics are not fate. Yes, having low HLA variation in a country was bad for life expectancy in 1960, but with the advent of the epidemiological transition after WWII, the effect starts to fall. With antibiotics, vaccinations, public health measures, etc.., the underlying HLA variation matters less and less for life expectancy.

Second, prior to the epidemiological transition, genetics could have played a (statistically) significant role in variation in living standards. Justin shows that HLA variation (which is good) is positively related to the years since the Neolithic revolution in your underlying population, and also positively related to the number of potential domesticable animals in your underlying population. Longer exposure to agriculture and animals generated benefits in dealing with disease, presumably because the populations were exposed longer and to more pathogens. (By “underlying population” I mean the ancestry-adjusted composition of your population today – so the US HLA variation depends mainly on European exposure to diseases). Thus places that had longer histories of civilization, by building up variation in HLA, would have enjoyed higher life expectancies and (assuming that living longer is good), higher living standards. You could spin this out further to speculate that places with higher life expectancies had greater incentives to invest in human capital and achieve even more gains in living standards historically.

The second paper is on lactose tolerance and development. Simply put, if you can digest milk, then you have an additional source of nutrition that lactose-intolerant people do not have. It changes the productivity of dairy-producing animals, making them a better investment. But no other mammal, and the vast majority of humans, do not produce lactase (the enzyme to break down lactose) beyond weaning from breast milk. At some point in time a sub-population of humans acquired a mutation that allowed them to keep producing lactase beyond weaning, meaning they could continue to consume dairy and use the nutrition available.

Justin backs out the ethnic composition of countries in 1500 (you can do this by using data on migration flows and known ethnic groups). He can then look at lactose tolerance in countries in 1500 by using the existing lactose tolerance of ethnic groups (which is presumed to not have changed much in 500 years). He finds that population density in 1500 is highly related to lactose tolerance in the population. This holds up even after you throw a lot of other controls into the specifications, including continent dummies – which is important in establishing that this is not just a proxy for some broader Asia/Europe difference.

Lactose tolerance acted like a Malthusian productivity boost, raising population density in 1500. Did this have long-run consequences for living standards? Maybe. Places that were densely population in 1500 tend to be relatively rich today, even if you control for their contemporary lactose tolerance levels. So through that channel, lactose tolerance may have helped push up living standards today. The story here would be something about dense populations having greater capacity for innovation, or density indicating broader potential for productivity increases.

I think what Justin’s papers show is that a useful way of thinking about genetics and development is in the sense of budget constraints. Gene(s) change the relative price of different activities or goods, which can alter social and/or economic outcomes, without implying that they make one person or population superior. People who can drink milk without getting sick are not making better decisions than people who cannot, they simply are less constrained in their budget set. Genes, in this sense, are just like geography, which creates different relative prices for populations in different areas. This is different than saying that genes “determine” behavior (e.g. a “patience” or “savings” gene) and that this creates variation in how people respond to an identical set of constraints.

Forecasting Future Growth

NOTE: The Growth Economics Blog has moved sites. Click here to find this post at the new site.

In a post a few days ago I made a distinction between looking at “measured growth over a few decades” and “the long-run growth rate of GDP”. The point was that while we can calculate the former, the latter is something that we have to be more speculative about.

As part of that post, I dropped in a comment about Robert Gordon’s work on the long-run growth rate of GDP. I said “He also argues that g will fall due to us running out of useful things to innovate on”, where “g” is the growth rate of output per worker. That is a far too glib characterization of Gordon’s work. So, first, my apologies to Bob for an unfair comment.

So what *does* Bob say about future growth? He’s got several recent working papers out on the subject that you can read to get his full story. See here, here, here, and here. Pulling from the latest paper (the first link), here is his prediction:

Future growth will be 1.3 percent per annum for labor productivity in the total economy, 0.9 percent for output per capita, 0.4 percent for real income per capita of the bottom 99 percent of the income distribution, and 0.2 percent for the real disposable income of that group.

For comparison, note that growth in output per capita from 1891 to 2007 was approximately 2.0 percent per year. So his projection is that growth will be a little under half (0.9 compared to 2.0) of this long-run historical rate.

Mechanically, where is this projection coming from? Output per capita is Y/N = Y/H x H/N, where Y is output, N is the number of people, and H is the number of hours worked. Slow growth in output per capita (Y/N) must come from either slow growth in output per hour (Y/H) or slow growth in hours per capita (H/N), or both.

Gordon talks about how demographics, institutional issues in the labor market, and the slowdown in education will limit growth of H/N (and to some extent Y/H). As I mentioned in the prior post, I don’t know that these are terribly controversial. He also touches on inequality and globalization as factors affecting not only these growth rates, but the growth rate of income for different parts of the distribution. That’s outside of what I want to talk about here, so let’s leave it aside.

What does Gordon assume about growth in productivity, meaning growth in Y/H? Gordon does not assume that growth in Y/H will fall in the future (which is what my comment implied). He assumes that growth in Y/H will stay the same as the average rate from 2004-2013, about 1.3% per year.

That is a reasonable thing to do in the absence of any alternative evidence. Given the above figure (which Bob kindly provided), you could even argue that 1.3% is a particularly optimistic choice, as from 2004 to 2013 the growth rate of productivity has been moving down. So Bob is not making a particularly pessimistic claim about future productivity growth, he is guessing it will remain at 1.3%.

Assuming that the growth rate of Y/H will continue at the current average rate could turn out to be wrong. Here is a another figure from his recent paper, on the average growth rate of Y/H in different periods.

If in 1972 you assumed the growth rate of Y/H would be 2.36% per year going forward, you were wrong. If in 1996 you assumed the growth rate of Y/H would be 1.38% per year going forward, you were wrong. If in 2004 you assumed the growth rate of Y/H would be 2.54% per year going forward, you were wrong. So it is quite possible that assuming the growth rate of Y/H will be 1.3% per year going forward is wrong too.

But that doesn’t mean Bob Gordon is wrong for choosing to use 1.3% in his projections. I think Bob’s criticism of the techno-optimistic literature is accurate. Examples of the cool things that you can do with a 3D printer (or driverless cars, etc..) do not constitute evidence that growth in Y/H *will* rise in the future. In the absence of any compelling evidence that productivity growth is accelerating, using the existing rate is reasonable.

I think the right way to think about this is that Bob has chosen a decent null hypothesis that g = 1.3%. If the techno-optimists are right, then it should show up in the next few years in the data on productivity growth. Give me a couple of years of 3% or 4% productivity growth, and maybe I’ll reject the null. But from the perspective of today, I cannot.

Where I disagree with Bob is on the probability that things like 3D printing, big data, robots, and medical innovation could lead to a period of accelerated growth in productivity similar to the 1990s. I think Bob’s papers (see section 7 of the most recent one) make it clear that he sees this probability as low (0-5%?). I personally think it is slightly higher (25-30%?), but less than what Brynjolfsson-McAfee would guess (80-90%?).

Genetic Factors in Savings Behavior

NOTE: The Growth Economics Blog has moved sites. Click here to find this post at the new site.

There is a recent article by Henrik Cronqvist and Stephan Siegel on the origins of savings behavior (published in JPE, but link is for working paper). They use the Swedish Twin Registry, which gives them data on roughly 15,000 twins, and link that to the deep Swedish data on income, savings, employment, and other information. They use this to examine whether savings behavior has a genetic component. Essentially, they are asking whether genetically similar people (twins) have similar savings behaviors. Figuring this out is hard, as twins share not just genes but also share home environments.

To get around this, Cronqvist and Siegel use the differences between identical and fraternal twins to their advantage. Here is the basic idea. If genes matter for savings behavior, then identical twins should have a higher correlation of their savings behavior than fraternal twins because fraternal share (on average) 50% of their DNA while identical twins share 100%. On the other hand, twins of either type will experience similar environmental factors (i.e. parenting). That is, the assumption is that fraternal twins share 100% of the common environment, just like identical twins, and not just 50%.

You have to be careful. Savings behavior can be correlated across twins at 100%, and yet that doesn’t mean that genes matter. It may mean that two individuals raised in a similar environment share similar attitudes towards savings. So the absolute level of correlation is not important, but the pattern between identical and fraternal twins is. It is by comparing the correlations within the two groups that allow the authors to draw out the importance of genetics.

Here’s a crude first look at their data:

You can see that identical twins do in fact have higher correlations in their savings rates than fraternal twins. Much of the remainder of the paper is confirming that this figure holds up with various controls included. Perhaps not surprisingly, it does hold up. You can argue with their exact measure of savings (changes in net worth divided by disposable income), but it is a measure used in other papers, and they are not trying to compare across countries so definitional issues in the dataset are less problematic.

The end result is that roughly 1/3 of variation in savings behavior can be accounted for by genetics (a little higher than this for men, and a little less for women). As an example, if you pulled two pairs of identical twins out of the population, you might find that Alice and Agnes saved 15% and 18% of their income, while Bob and Bubba saved 10% and 11%, respectively. About one-third of the difference in average savings (17.5% versus 10.5%) is due to genetic differences between the A girls and the B boys. The A family presumably has alleles that code to more patience on the “savings gene”, while the B family has alleles that code to less patience.

Maybe as interesting as the 1/3 number is that the share attributed to common family experience is essentially zero. Their paper supports a “nature” over “nurture” view on savings behavior. For completeness, the remaining 2/3 of variation in savings behavior is purely idiosyncratic. That is, 2/3 of Alice and Agnes’s higher saving rate is simply a result of Alice being Alice and Agnes being Agnes.

Do we know what or where “the savings gene” is? No. It is almost certainly not even a single gene, but rather some complex set of genes that combine to determine savings behavior. But what Cronqvist and Siegel establish is that it is reasonable to suspect that this complex set of genes actually exists.

From a growth perspective, research that examines heterogeneity in individual behaviors within economies is often useful in thinking about heterogeneity across countries. This is particularly true when you realize that much of the cross-country variation in economic development is driven by the composition of country’s population.

The Cronqvist and Siegel paper cannot tell us whether there are true genetic differences in savings behavior *between* different populations. The genetic variation in savings behavior within Sweden might be similar to genetic variation in savings behavior within Burundi, or Nepal, or Peru. But it opens up the possibility that there could be some genetic variation in savings behavior between countries. If there is a set of genes that code for savings (or patience, or long-run planning, or whatever) then it is certainly theoretically possible that populations vary as well.

Given the relative importance of population composition in accounting for differences in living standards, we cannot dismiss the idea that there is a genetic component involved. Note that this doesn’t mean that high-saving or low-saving populations are biologically different, any more than blue eyed populations and brown-eyed populations are biologically different. That is, high-savings populations are not super-patient mutants (who would make the worst X-men ever). They have a gene expression that may lead to higher savings rates.

There are starting to dribble into the research world studies that look at actual genetic differences across populations and the implication of those for economic development. We are no where close to a thorough accounting of the role of genetic variation in explaining development, but it is beginning to look as if we should accept that there is a meaningful role for it.

Has the Long-run Growth Rate Changed?

NOTE: The Growth Economics Blog has moved sites. Click here to find this post at the new site.

My actual job bothered to intrude on my life over the last week, so I’ve got a bit of material stored up for the blog. Today, I’m going to hit on a definitional issue that creates lots of problems in talking about growth. I see it all the time in my undergraduate course, and it is my fault for not being clearer.

If I ask you “Has the long-run growth rate of the U.S. declined?”, the answer depends crucially on what I mean by “long-run growth rate”. I think of there as being two distinct definitions.

• The measured growth rate of GDP over a long period of time: The measured long-run growth rate of GDP from 1985 to 2015 is . Note that here the measurement does not have to take place using only past data. We could calculate the expected measured growth rate of GDP from 2015 to 2035 as . Measured growth rate depends on the actual path (or expected actual path) of GDP.
• The underlying trend growth of potential GDP: This is the sum of the trend growth rate of potential output per worker (we typically call this ) and the trend growth rate of the number of workers (which we’ll call ).

The two ways of thinking about long-run growth inform each other. If I want to calculate the measured growth rate of GDP from 2015 to 2035, then I need some way to guess what GDP in 2035 will be, and this probably depends on my estimate of the underlying trend growth rate.

On the other hand, while there are theoretical avenues to deciding on the underlying trend growth rate (through , , or both), we often look back at the measured growth rate over long periods of time to help us figure trend growth (particularly for ).

Despite that, telling me that one of the definitions of the long-run growth rate has fallen does not necessarily inform me about the other. Let’s take the work of Robert Gordon as an example. It is about the underlying trend growth rate. Gordon argues that is going to fall in the next few decades as the US economy ages and hence the growth in number of workers will slow. He also argues that will fall due to us running out of useful things to innovate on. (I find the argument regarding strong and the argument regarding completely unpersuasive. But read the paper, your mileage may vary.)

Now, is Gordon right? Data on the measured long-run growth rate of GDP does not tell me. It is entirely possible that relatively slow measured growth from around 2000 to 2015 reflects some kind of extended cyclical downturn but that and remain just where they were in the 1990s. I’ve talked about this before, but statistically speaking it will be decades before we can even hope to fail to reject Gordon’s hypothesis using measured long-run growth rates.

This brings me back to some current research that I posted about recently. Juan Antolin-Diaz, Thomas Drechsel, and Ivan Petrella have a recent paper that finds “a significant decline in long-run output growth in the United States”. [My interpretation of their results was not quite right in that post. The authors e-mailed with me and cleared things up. Let’s see if I can get things straight here.] Their paper is about the measured growth rate of long-run GDP. They don’t do anything as crude as I suggested above, but after controlling for the common factors in other economic data series with GDP (etc.. etc..) they find that the long-run measured growth rate of GDP has declined over time from 2000 to 2014. Around 2011 they find that the long-run measured growth rate is so low that they can reject that this is just a statistical anomaly driven by business cycle effects.

What does this mean? It means that growth has been particularly low so far in the 21st century. So, yes, the “long-run measured growth rate of GDP has declined” in the U.S., according to the available evidence.

The fact that Antolin-Diaz, Drechsel, and Petrella find a lower measured growth rate similar to the CBO’s projected growth rate of GDP over the next decade does not tell us that or (or both) are lower. It tells us that it is possible to reverse engineer the CBO’s assumptions about and using existing data.

But this does not necessarily mean that the underlying trend growth rate of GDP has actually changed. If you want to establish that or changed, then there is no retrospective GDP data that can prove your point. Fundamentally, predictions about and are guesses. Perhaps educated guesses, but guesses.