Wednesday, 31 August 2016

What I've been reading (vol 9.) Economic history: Malthus and beyond

After the very intriguing part 1 covering two fascinating books whose contribution extends across a few scientific fields, this one lays out two books with a more contemporary historical emphasis. 

Clark, Gregory (2011) A Farewell to Alms: A Brief Economic History of the World. Princeton University Press

Clark’s book is first and foremost a wonderful economic history textbook. Although it carries in its title the word “brief” it represents quite a comprehensive overview of the patterns before, during and after the Industrial Revolution that can be accredited with having created the modern world. It is a good book, full of empirical data and a very detailed overview of the facts behind the patterns responsible for modern development. It is thus more a scientific work that a popular one (closer to Diamond’s Guns, Germs, and Steel, than Harari’s Sapiens).

The book is separated into three parts, each important in painting the picture of why and how the Industrial Revolution changed things 200 years ago. It begins by defining a simple model of the Malthusian economy – the state of the world since the beginning of the Agricultural Revolution 8000 years B.C. until 1800 A.D. A world that was, according to Hobbes, “solitary, poor, nasty, brutish, and short” (interestingly Hobbes wrote this in 1651 referring to the hunter-gatherer societies of the distant past, not knowing that his own living standards were hardly any better, even with technological progress of that age). The second part of the book looks at how the Industrial Revolution unfolded, what made it possible, and what were its long-lasting consequences on the significantly improved living standards we today take for granted. The final part focuses on its not so great consequences and what is known today as the Great Divergence, a phenomenon of rapidly increasing inequality between the world’s richest and poorest nations. The book is therefore not only a companion in economic history, but it also addresses the topic of economic development and between-country inequality. 

Clark summarizes his book into a single chart (I've used it already in an earlier text) that shows the levels of income per person during the three historical periods he examined. The most striking is the conclusion that for almost 10,000 years (actually even more considering the anatomically correct Homo Sapiens hunter-gathers originating from 100,000 B.C.) the global economy as well as its society was constrained by something called the Malthusian trap. For over 10,000 years living standards of the vast majority of the population haven’t really improved at all. Even the wealthiest of the people in 1800 weren’t significantly better off (in terms of things like life expectancy, height, calorie intake, etc.) than their foraging ancestors 10,000 years ago.
The quality of life was virtually unchanged: life expectancy was between 30 to 35 years (due to high levels of death during infancy or before the age of 16); quality of diet and exposure to diseases were even worse than during hunter-gatherer times (diet was worse in general for settled agricultural societies, and large population densities generated a fertile ground for the spread of diseases), stature was roughly the same, and work life was much easier during hunter-gatherer times. Furthermore hunter-gatherer societies were mostly egalitarian, whereas the world after the agricultural revolution was characterized by massive inequality. Most importantly social mobility actually went downwards as the rich had more surviving children than the poor (4-5 compared with 2), meaning that the offsprings of wealthier individuals had to descend down the social hierarchy as only the eldest son (or daughter) inherited the family fortune. 

The Industrial Revolution changed all that in a mere 200 years. Income per person simply skyrocketed, life expectancy doubled, people could earn, buy and own incomparably more than ever before, societies became much more dynamic, nutrition improved, more calories were consumed, people grew taller, and working hours declined making it possible to discuss things like work-life balance. But it wasn’t like that for all societies across the globe. In fact, many nations became poorer in the aftermath of the Industrial Revolution and were generally worse off than before it. As Clark says: “There walk the earth now both the richest people who ever lived and the poorest”.

The book is therefore interested in answering why was technological advancement so slow during pre-Industrial times; what was the reason behind incomes rising so rapidly after 1800; and why didn’t all nations experience this breakthrough? To give an answer, one first needs to understand the logic of the Malthusian economy and see how it evolved into creating the underlying foundations for the onset of the Industrial Revolution. This is arguably the best part of the book as it lays out a theoretical and empirical case for how the Malthusian economy operated. In a nutshell this is it: when birth rates exceed death rates, population growth increases, income will decline, as will material living standards (more people per fixed amount of land). As long as income is higher than its subsistence level, population growth will go up, and income will continue to fall. The only way to stop this decline of living standards is to push income back to its subsistence level equating the birth rate to the death rate. So the forces of the Malthusian trap will always push the economy back to its subsistence level of income as any increase in birth rates which leads to higher population growth will hurt living standards. In order to increase living standards and life expectancy in the Malthusian world, one had to limit birth rates or increase death rates. Which is why any technological advancement had no particular effect in the Malthusian world. Any technological advancement that did in fact occur throughout the old and middle ages only led to higher population and had no lasting impact on living standards: “In the preindustrial world sporadic technological advance produced people, not wealth”. It was a truly natural economy system, similar to the one in the animal kingdom. The book goes on to detail a whole plethora of empirical arguments to support this theoretical conclusion, all of which seem very convincing. 

Shortcomings: revolution within the Malthusian model? 

The biggest criticism I have of the book is that it all too easily dismisses the institutional argument and by using a very illogical set of assumptions as to what causes economic growth. This was particularly evident in chapter 8, which transcends over to chapter 9, where he presents his main reasons as to why the Industrial Revolution had occurred. He tries to overturn the argument that the Malthusian era had no incentives for economic growth, by presenting a series of data on taxation, price stability, public debt, security of property, and even social mobility. To my surprise he cites preindustrial England as a country with beneficial factors for accumulating economic growth based on very low tax rates or low levels of public debt. This is a very flawed argumentation given that preindustrial societies had no or very little public goods necessary to levy taxes for. Only democratization, that occurred in England gradually after the Glorious Revolution, increased the demand for government, and hence its spending levels, its taxation levels, and its public debt. The argument of price stability is even more absurd as comparing pre and post Industrial Revolution inflation is meaningless. For one thing, the post Industrial Revolution world is characterized by a significantly different financial architecture, and many more societies operating on these modern principles, implying that its rates of inflation and purchasing power are incomparable. More importantly, low inflation in pre-industrial times is a consequence of zero to minimum economic growth. Comparing the two is not only wrong, it is unacceptable and it is a clear example of endogeneity (most notably, the omitted variable bias problem). Finally the argument on social mobility is contradictory to the one expressed in an earlier chapter (ch.6). He states that peasants moved up the social hierarchy and how this was a sign of a growth-enhancing mechanism, but the entire previous chapter goes on about how the Malthusian trap implied persistent downward social mobility among the wealthier with more children. Out of the entire book this part is the most problematic of all – the great economic historian that Clark is, I am surprised he fails to account for the role of finance (credit) in pre- and post-Industrial world development, as well as the fact that there was no economic growth in pre-industrial times making modern economic indicators incomparable to the past ones.

Furthermore, Clark discusses four profound features on the economy within the Malthusian era that created the momentum for the Industrial Revolution to start by 1800s: 1) Interest rates fell to modern low levels by 1800 (again a huge comparison problem; supply was scarce and demand failed to materialize due to a lack of belief in the future), 2) literacy and numeracy became the norm (this did increase probably due to rewarding incentives for investment in skills as the author claims), 3) working hours rose to modern levels (this happened long before the Industrial Revolution), and 4) there was a decline in interpersonal violence. All of these changes gave rise to the middle classes within societies which later became the proponents of change.  

He claims that enhanced production of knowledge capital generated benefits throughout the economy, primarily through increased efficiency. But why did this happen in 1800s and not before? Why hasn’t any society been able to maintain modern rates of production growth over any significant time period? What was the change that broke this status quo and generated the Industrial Revolution? His partial answer is that this was all a result of a gradual, millennium-long evolutionary change, started after the arrival of first institutionally stable societies of the ancient era, and then these sets of institutions interacted to change the human culture. He cites millennia of stability that despite Malthusian constraints rewarded effort and hard work, fertility limitation, and encouraged the development of cultural forms (like time preference and family formation) that facilitated modern growth. The gradual adaptation essentially developed a more economically oriented mindset, and it affected not only England but all other preindustrial societies at about the same time. In fact Clark even goes so far to claim that the long Malthusian era had a profound biological impact through better adaptation to the modern world. However, little evidence (if any at all) is given to back up this biological argument. In other words, Clark states that the Malthusian world shaped the population to be more inclined towards accepting a capitalist system

I find this entire argumentation difficult to believe. It seems to rest upon the assumption that the Industrial Revolution was inevitable and that it was only a matter of time before the “wheel of history” turns in that direction. It actually pegs the Industrial Revolution as a necessary and an anticipated outcome of the Malthusian model. The biggest problem with this assumption is explaining why 1800s, and not before or later? Yet he does offer his explanation of four aforementioned factors that happen to coincide with exactly this period and have resulted in greater economic growth.

But just as the author correctly notes in one footnote on pg. 205 that close correlation between two variables must be that one causes the other, or that there is a single independent cause for both, so do I believe that all of the mentioned factors instead of causing higher economic growth, have had a series of common independent causes => the most important one being the rise of credit and the modern financial system. In addition there were also the following factors: the rise of demand for democratization started during the Glorious Revolution, the rising power of mercantilist merchants which have started to join the aristocracy and therefore demanded greater political clout (the demand for which can even be traced back to the Magna Charta), the Scientific Revolution of the 16th century and the great discoveries (in fact the great discoveries led to mercantilism which more than any other era provided certain underlying conditions for the Industrial Revolution), and finally, just like with the origin of life, chance played an important role as well. It was a series of political events that made England perfectly placed to take advantage of the Industrial Revolution. Had Netherlands not entered complacency after its ascent in the 17th century, and had Parliament lost its battle against James II., the Industrial Revolution would not have happened in England. Perhaps it would later, but surely not when it did.

What is therefore entirely lacking in Clark’s analysis is the political factor. He fails to take into consideration the extremely important political conditions that drove investment, innovation, and migration decisions throughout human history. Wars tend to be a more important explanation of altered human behaviour than any corresponding economic factor such as interest rates or working hours. 

Diamond, Jared and Robinson, James (eds.) (2010) Natural Experiments of History. Belknap Harvard, Cambridge, MA

This fascinating collection of articles edited by two great scholars Jared Diamond and James Robinson provides the reader with a great guidebook on how to perform natural experiments in the social sciences, or to be more precise, to use historical perturbations as settings for scientific experiments. One of the reasons I decided to read it, apart from the obvious (the authors and the title, duh!) was its message that doing proper science is not just limited to the natural sciences, laboratories, and highly quantifiable data, but that it can also be done using so-called “natural experiments”. Natural experiments imply observing two systems that are similar to a large extent (either by historical background, geography, culture, its people, etc.), but are different in the key factor that the researcher wishes to focus on. However, unlike laboratory experiments natural experiments in history aren’t triggered artificially so that we may observe how the outcome phases out by careful observation. They depend on a historical event that caused an initial difference that later unfolded and led to significantly divergent outcomes. Proving that a historical event triggered the divergence in outcomes (inferring a causal relationship) is obviously the hardest part of natural experiment studies.

The most common examples of natural experiments of history are South and North Korea, and West and East Germany. Same nation, same history, same culture, same heritage, same geography, same diseases, same people – but different economic fortunes, all as a result of an initial divergence (a post-war split). A similar example can be the divergence in outcomes between two countries living on one island – Haiti and the Dominican Republic (Diamond covers this case in Chapter 4 of the book actually). I called upon these and other examples as well, one of which was from Acemoglu and Robinson’s 2012 hit book Why Nations Fail – the city of Nogales, on the US-Mexican border. Same story – one city divided by a border, same people, same history, same climate and geography (this is on a small scale so the differences are really miniscule). One part however has much better living standards and economic outcomes – the US part of Nogales. The difference between them? Institutions. Same as with the two Germanys and the two Koreas.

But this book goes beyond these wide-known examples and delivers something more. It presents eight cases of natural experiments through history that give the reader a very detailed overview of how these should be performed. It even finishes with a step-by-step guide on what practitioners of history and other branches of social sciences should pay attention to while attempting to utilize a natural experiment. Quite useful.

As I mentioned, the book is organized into eight articles. The first four are presented from a standard historical, non-quantitative, narrative approach, while the other four are quantitative statistical studies more familiar to other social science departments, not so much to the history scholars (apart from cliometrics of course – praise to them!). Naturally, I was more interested in the quantitative studies. Don’t get me wrong, I like a good narrative as much as the next guy, but I too fall under the impression that for serious science there must be some numbers involved (this doesn’t make me one of those model-everything freaks, I just tend to be more suspicious towards a narrative if it isn’t backed by quantifiable and testable data).

Anyway, the first study, done by Patrick Kirch, looks at three Pacific islands colonized by ancestral people and why history unfolded differently in each of them. The second, done by James Belich, observes several frontier societies (like the US West, or Australia, or Siberia) and how they achieved different outcomes despite having awfully similar initial cycles of development (population growth, imports boom, subsequent bust leading to bankruptcies and declining growth, and a rescue in terms of exports). The third study, by Stephen Haber, looks at the origins of banking systems in the US, Mexico and Brazil, explaining why they developed differently in each of these countries. The fourth, done by Diamond, observes the aforementioned differences between Haiti and the Dominican Republic (the differences can be traced to different colonization patterns – Haiti by the French, Dominica by the Spanish), while the fifth, also by Diamond (first quantitative), looks at the environmental causes of deforestation in pre-colonization times of 81 Pacific islands. 

It gets more interesting from here. The sixth study quantifies the relationship between African slave trade and Africa’s current stages of development. The author, Nathan Nunn proves (read his QJE paper) that the differences in slave exploitation across Africa caused African countries to develop differently in the subsequent four centuries. In other words, former slave exporting countries are poorer than non-slave exporting countries, and it was the slave trade that caused the initial divergence. Without it, Africa’s poorest countries would not have been as poor as they are now. They would be the average of the developing world. The seventh study, done by Banerjee and Iyer, looks at the difference in British colonial rule in India based on different land tenure systems imposed by the British. Areas that were put under control of the landlords today lag behind in the provision of public goods like schools and roads, compared to areas where land ownership rights were more decentralized. The final study, done by Acemoglu, Cantoni, Johnson and Robinson tests the impact of institutional reforms brought upon Germany’s territories after the invasion of the French Revolutionary armies and later Napoleon. In every territory they conquered the French brought massive institutional changes (introducing the Code de Civil, abolishing guilds, initiating land reform and abolishing serfdom, and emancipating religious minorities – in this case, the Jews) that were aimed at repealing the old feudalistic regime and set the stage for the origination of capitalism. The authors stress that areas which accepted these changes, and crucially - haven’t repealed them after the French retracted, embraced the Industrial Revolution which set them on a positive growth and development trajectory over the next century. Those that repealed the French reforms and re-introduced the old regime institutions resisted change and had consequentially worse off economic outcomes.

A quick guide on how to test natural experiments 

So what are the lessons a social scientist must infer from a book about how to do natural experiments? First the author must resolve the issue of selection of perturbed areas over the non-perturbed ones (distinguishing between the treatment group and the control group). For example, was the reason why Napoleon attacked the particular German areas because of its economic potential (which could have affected its long-term trajectory), or was it due to something else (in other words – was their selection random with respect to the outcome we are observing)? The answer is clearly the former – Napoleon attacked those areas due to proximity and a strategic advantage in combat, not because of their better economic potential over other areas. This issue is the most important one in determining whether or not a natural experiment satisfies all the conditions of an experiment.

The second condition is that the outcome could be delayed by decades or even centuries (as the examples of African slave trade, or Haiti and Dominican Republic tell us). This is normal for societal changes – it takes a long time for an institutional reform to fully yield its desired outcomes. More time than the voters are prepared to wait, actually, which is why such reforms are difficult to engage in today’s democracies. In addition to the outcomes being delayed, they could also be indirect (as the Acemoglu et al paper shows – these reforms by themselves did not cause better outcomes, they just made it easier for the industrial revolution to be introduced).

The third lesson is to be careful about omitted variable bias (when other, unobserved variables are actually affecting the outcome, not the ones we think – this means that the error term in the equation would be correlated with the dependent variable, yielding a biased estimate and implying wrong conclusions of our study). Another is to worry about potential measurement errors, and more importantly – reverse causality. Essentially, an experimental setup should help the researcher get rid of these concerns, which is why defining the experiment is the most difficult job for a social scientist.

Finally there are also a few good points on complexity vs clarity (don’t throw in too many variables in the pot), on the difficulty of measuring some social phenomena, and on precisely quantifying the variable of interest. None of these is ever straightforward in natural experiments, but that’s what makes them challenging in the first place.

No comments:

Post a Comment