On Household Size and Economic Convergence

A few days ago, one of my papers was accepted for publication at the Scottish Journal of Political Economy (working paper version here). Co-authored with Vadim Kufenko and Klaus Prettner, this paper makes a simple point which I think should be heeded by economists: household size matter. To be fair, economists are aware of this when they study inequality or poverty. After all, the point is pretty straightforward: larger households command economies of scale so that each dollar goes further than in smaller households. As such, adjustments are necessary to make households comparable.

Yet, economists seem to forget it when times come to consider paths of economic growth and convergence across countries. In the paper, we try to remedy this flaw. We do so because there was a wide heterogeneity of household size throughout history – even within more homogeneous clubs such as the countries composing the OECD.  If we admit, as the economists who study poverty and inequality do, that income per person adjusted for household size is preferable to income per person, then we must recognize that our figures of income per capita will misstate the actual differences between countries. In addition, if households grew homogeneously smaller over a long period of time, figures of income per capita will overstate the actual improvements in living standards. As such, we argue there is value in modifying the figures to reflect changing household sizes.

For OECD countries, we find that the adjusted income figures increased a third less than the unadjusted per capita figures (see table below). This suggests a more modest growth trend. In addition, we also find that up to the structural break in variations between countries (NDLR: divergence between OECD countries increased to around 1950) there was more divergence with the adjusted figures than with the unadjusted figures (see figure below). We also find that since the break point, there has been less convergence than previously estimated.

While the paper is presented as a note, the point is simple and suggests that those who study convergence between regions or countries should consider the role of demography more carefully in their work.



Empire effects : the case of shipping

I have been trying, for some time now, to circle an issue that we can consider to be a cousin of the emerging “state capacity” literature (see Mark Koyama’s amazing summary here). This cousin is the literature on “empire effects” (here and here for examples).

The core of the “empire effect” claim is that empires provide global order which we can consider as a public good. A colorful image would be the British Navy roaming the seas in the 19th century which meant increased protection for trade. This is why it is a parent of the state capacity argument in the sense that the latter concept refers (broadly) to the ability of a state to administer the realm within its boundaries. The empire effect is merely the extension of these boundaries.

I still have reservations about the nuances/limitations of state capacity as an argument to explain economic growth. After all, the true question is not how states consolidate, but how they create constraints on rulers to not abuse the consolidated powers (which in turn generates room for growth). But, it is easy to heavily question its parent: the empire effect.

This is what I am trying to do in a recent paper on the effects of empire on shipping productivity between 1760 and 1860.

Shipping is one of the industry that is most likely to be affected by large empires – positively or negatively. Indeed, the argument for empire effects is that they protect trade. As such, the British navy in the 19th century protected trade and probably helped the shipping industry become more productive. But, achieving empire comes at a cost. For example, the British navy needed to grow very large in size and it had to employ inputs from the private sector thus crowding-it out. In a way, if a security effect from empire emerged as a benefit, there must have been a cost. The cost we wish to highlight is the crowding-out one.

In the paper (written with Jari Eloranta of Appalachian State University and Vadim Kufenko of University of Hohenheim), I argue that, using the productivity of the Canadian shipping industry which was protected by the British Navy, the security effect from a large navy was smaller than the crowding-out from high-levels of expenditures on the navy.

While it is still a working paper which we are trying to expand and improve, our point is that what allowed the productivity of the Canadian shipping industry (which was protected by Britain) to soar was that the British Navy grew smaller in absolute terms. While the growth of the relative strength of the British Navy did bolster productivity in some of our tests, the fact that the navy was much smaller was the “thing in the mix that did the trick”.  In other words, the empire effect is just the effect of a ramping-down in military being presented as something else than it truly is (at least partly).

That’s our core point. We are still trying to improve it and (as such) comments are welcomed.

The Antebellum Puzzle, Anthropometric History and Quebec

I recently gave an interview on Economics Detective Radio with Garrett Petersen to talk about my forthcoming article in Economics & Human Biology (with Vadim Kufenko and Alex Arsenault Morin). In the interview, I explain why anthropometric history is important to our understanding of living standards, their evolution and short-term trade-offs in economic history. The interview is below, but you should subscribe to Garrett’s podcast as he is well on his way to becoming a serious competitor to EconTalk with the bonus that he does lots of economic history.

Podcast link (download).

The Heights of French-Canadian Convicts, 1780 to 1830

A few days ago, it was confirmed that my article with Vadim Kufenko and Alex Arsenault Morin on the heights of French-Canadians between 1780 and 1830 was accepted for publication in Economics and Human Biology. In that paper, we try to introduce French-Canadians before 1850 to the anthropometric history literature by using the records of the prison of Quebec City. Stature is an important measure of living standards. As it is heavily related to other aspects of health outcomes, it is a strong measure of biological living standards. More importantly, there are moments in history when material living standards and biological living standards move in opposite directions (in the long-run, this is not the case).

We find three key results. The first is that the French-Canadians grew shorter throughout the era when living standards did not increase importantly (and were very volatile). This puts them at odds from other places in North America where increases in stature were experienced up until the 1820s. Furthermore, stature stops falling around 1820 when economic growth picked up. This places the French-Canadians in a unique category in North America since it seems unlikely that they experienced a strong version of the antebellum puzzle (decline in stature with increases in material living standards which is what the US experienced). The second key result is that the French-Canadians are the shortest in North America, shorter even than Black Americans in slavery. However, they are considerably taller than most (save Argentinians) Latin Americans. More importantly, they are considerably taller than their counterparts in France. The third key result is related to the second key result. Today, French-Canadians are noticeably shorter than other Canadians. However, the gap was more important in the late 19th century and early 20th century. Pegged as a “striking exception” within Canada, we do not know when it actually started. Thanks to our work, we know that this was true as far back at the early 19th century.

The working paper (dramatically different than the accepted version) is here and I am posting key results in tables and figures below.  Moreover, I will be talking about anthropometric history and economic history with Garrett Petersen of Economics Detective Radio this Tuesday (I do not know when the podcast will be made available, but you should subscribe to that show anyways).



A hidden cost of the war on drugs

AI just completed another paper (this time with my longtime partner in crime Vadim Kufenko) where we question an hypothesis advanced by Samuel Bowles regarding the cost of inequality. In the process, we proposed an alternative explanation which has implications for the evaluation of the war on drugs.

In recent years, Samuel Bowles (2012) has advanced a theory (well-embedded within neoclassical theoretical elements while remaining elegantly simple) whereby inequality increases distrust which in turn magnifies agency problems. This forces firms to expend more resources on supervision and protection which means an expansion of the “guard labor force” (or supervisory labor force). Basically, he argues there is an over-provision of security and supervision. That is the cost of inequality which Bowles presents as a coordination failure. We propose an alternative explanation for the size of the guard and supervisory labor forces.

Our alternative is that there can be over-provision of security and supervision, but this could also be the result of a government failure. We argue that the war on drugs leads to institutional decay and lower levels of trust which, in turn, force private actors to deploy resources to supervise workers and protect themselves. Basically, efforts at prohibiting illicit substances require that limited policing resources be spread more thinly which may force private actors to expend more resources on security for themselves (thus creating an overprovision of security). This represents a form of state failure, especially if the attempts at policing these illicit substances increase the level of crime to which populations are vulnerable. To counteract this, private actors invest more in protection and supervision.

Using some of the work of Jeffrey Miron and Katherine Waldock, we show that increases in the intensity of prohibition enforcement efforts (measured in dollars per capita) have significant effects on the demand for guard labor. Given that guards represent roughly 1 million individuals in the US labor market, that is not a negligible outcome. We find that a one standard deviation increase in the level of drug enforcement efforts increases the ratio of guards to the population by somewhere between 12.92% and 13.91% (which is the equivalent of roughly 100,000 workers).

While our paper concentrated on proposing an alternative to the argument advanced by Bowles regarding the cost of inequality, we (more or less accidentally) measured a hidden cost from the war on drugs. The insecurity (increased crime rates and spillovers from illegal markets into formal markets) brought forth by drug prohibition  forces an over-provision of security and supervision (our supervision measure which includes workers that supervise other workers were smaller than with the security guard measure).

Basically, a hidden (private cost) of the war on drugs is that we must reallocate resources that we could have used otherwise. Its a little like when I say that it is meaningless to compare healthcare expenditures to GDP in Canada and the United States because Canadians assume costs in a hidden manner through rationing. Waiting lists in Canada are longer than in the US. The cost is lost wages and enduring pain and that cost will not appear in measures of expenditures to GDP. The war on drugs works the same way. There is a fiscal cost (expenditures dedicated to it and the taxes that we must impose), there is a crime cost (destruction of lives and property) and there is a reallocation cost of privately providing security which is hard to measure.

*The paper is available here. 

Household size and growth since 1870 (albeit in Canada)

Two days ago, I posted something on how much we were estimating growth since the 1950s. While organizing another research paper that I am trying to finish, I realized that I could make a follow-up to this based on previous research of mine.

A few months ago, I published (alongside Vadim Kufenko and Klaus Prettner) a short note in Economics Bulletin where we showed that the large differences in household size in Canada that existed up to 1975 led many to overestimate the level of differences between provinces. Moreover, we pointed out that because household size were converging at the same time as incomes, we argued that the rate of convergence from 1945 onwards was slightly overestimated. That paper convinced us to do the same between all the OECD countries (we are assembling the data right now).  But this was an argument about variance, what if we simply plot the “per capita” income of Canada with the “per adult equivalent” income of Canada since 1870.

By using the Maddison dataset combined with the data from my article, it took me a few seconds to get the graph below. What is important to notice in this graph is that, incomes per adult equivalent (measured in 1990 Geary-Kheamis dollars) have increased 40% less than incomes per person. Since adult equivalents are a better measure of living standards (because you capture the economies of scale associated with household size), we can easily say that we have been underestimating the level of improvement in Canada (it is still substantial however).


Malthusian pressures (as outcome of rent-seeking)

Nearly a week ago, I intervened in a debate between Anton Howes of King’s College London whose work I have been secretly following  (I say “secretly” because as an alumnus of the London School of Economics, I am not allowed to show respect for someone of King’s College) and Pseudoerasmus (whose identity is unknown but whose posts are always very erudite and of high quality – let’s hope I did not just write that about an alumnus of King’s College). Both bloggers are heavily involved in my first field of interest – economic history.

The debate concerned the “Smithian” counter-effect to “Malthusian pressures”. The latter concept refers to the idea that, absent technological innovation,  population growth will lead to declining per capita as a result of marginally declining returns. The former refers to the advantages of larger populations: economies of scale, more scope for specialization and market integration thanks to density. Now, let me state outright that I think people misunderstand Malthusian pressures and the Smithian counter-effect.

My point of is that both the “Smithian counter-effect” and “Malthusian pressures” are merely symptoms of rent-seeking or coordination failures. In the presence of strong rent-seeking by actors seeking to reduce competition, the Smithian counter-effect wavers and Malthus has the upper hand. Either through de-specialization, thinner of markets, shifting to labor-intensive technologies, market disintegration and lower economies of scale, rent-seeking diminishes the A in a classical Cobb-Douglas function of Total Factor Productivity (Y=AKL). This insight is derived from my reading of the article by Lewis Davis in the Journal of Economic Behavior and Organization which contends that “scale effects” (another name for a slight variant of the “Smithian counter-effect) are determined by transaction costs which are in turn determined by institutions. If institutions tend to favor rent-seeking, they will increase the likelihood of coordination failure. It is only then that coordination failures will lead to “Malthusian pressures” with little “Smithian counter-effect”. Institutions whose rules discourage rent-seeking will allow markets to better coordinate resource use so as to maximize the strength of the “Smithian counter-effect” while minimizing the dismal Malthusian pressures.

In essence, I don’t see the issue as one of demography, but as one of institutions, public choice and governance. I am not alone in seeing it this way (Julian Simon, Jane Jacobs and Ester Boserup have documented this well before I did). Why the divergence?

This is because many individuals misunderstand what “Malthusian pressures” are. In an article I published in the Journal of Population Research, me and Vadim Kufenko summarize the Malthusian model as a “general equilibrium model”. In the long run, there is an equilibrium level of population with a given technological setting. In short-run, however, population responds to variation in real wages. Higher real wages from a “temporary” positive real shock will lead to more babies. However, once the shock fades, population will adapt through two checks: the preventive check and the positive check. The preventive check refers to households delaying family formation. This may be expressed through later marriage ages, planned sexual activities, contraception, longer stays in the parental household and greater spacing between births. The positive check refers to the impact of mortality increasing to force the population back to equilibrium level. These checks return to the long-term equilibrium. Hence, when people think of “Malthusian pressures”, they think of population growth continuing unchecked with scarce ressources. But the “Malthusian model” is basically a general equilibrium model of population under fixed technology. In that model, there are no pressures since the equilibrium rates of births and deaths are constant (at equilibrium).

However, with my viewpoint, the equilibrium levels move frequently as a result of institutional regimes. They determine the level of deaths and births. “Poor” institutions will lead to more frequent coordination failures which may cause, for a time, population to be above equilibrium – forcing an adjustment. “Poor” institutions would also lead to an inability to respond to a change in constraints (i.e. the immediate environment) by being rigid or stuck with path-depedency problems which would also imply the need for an adjustment.  “Good” institutions will allow “the Smithian counter-effect” to intervene through arbitrage across markets to smooth the effect of local shocks, a greater scope for specialization etc.

My best case for illustration is a working paper I have with Vadim Kufenko (University of Hohenheim) and Alex Arsenault Morin (HEC Montréal) where we argue that population pressures as exhibited by the very high levels of infant mortality rates in mid-19th century Quebec were the result of institutional regimes. The system of land tenure for the vast majority of the population of Quebec was “seigneurial” and implied numerous regressive transfers and monopoly rights for landlords. This system was also associated with numerous restrictions on mobility which limited the ability of peasants to defect and move. However, a minority of the population (but a growing one) lived under a different institution which did not impose such restrictions, duties and monopolies. In these areas, infant mortality was considerably lower. We find that, adjusting for land quality and other factors, infant mortality was lower in these areas for most age groups. Hence, we argued that what was long considered as “Malthusian pressures” were in fact “institutional pressures”.

Hence, when I hear people saying that there are problems linked to “growing population”, I hear “because institutions make this a problem” (i.e. rent seeking).