Author: admin

Individual actions that receive praise and social approbation are often those were the individual doesn’t act in ways consistent with their narrow self-interest but sacrifices self-interest for the public good.

A jargon economists use for one type of that public spirited action is “internalizing externalities.” An “externality” is something I do that imposes costs or creates benefits for others but which isn’t reflected in the costs and benefits I personally face. So I could litter and just throw trash on the ground rather than find a waste bin and that would be the privately optimally action given costs and benefits if I didn’t factor in the well-being of others. But my litter reduces enjoyment for others and I should expect social disapproval (“hey, jerk, don’t litter”) if I don’t internalize the externality.

Lots of actions, like not using single use plastic bags or using paper straws or not eating beef or driving a Prius are reflections that the full cost of our consumption decisions on the environment are not factored into the prices we face. When the price of carbon is too low, people want to internalize those externalities and want those actions to be acknowledged with positive social approval.

But for people living in rich countries most people are not considering at all their behavior towards what is the largest price distortion they face, by far. All rich OECD countries impose massive barriers to the mobility of labor. This means the domestic price of labor that rich country citizens face vastly overstates the global cost of low to medium skill labor. Estimates I have done with Michael Clemens and Claudio Montenegro on the Place Premium suggests that in the USA the domestic cost of low skill labor is roughly four to five times higher than its true availability cost to the USA.

That means that if you live in an OECD country, while your current consumption might have too many plastic bags or hamburgers and laundered towels relative to their true cost that factors in environmental damage (externalities not reflected in prices), that is trivially small compared to the fact that your consumption, by reflecting prices distorted by policy choices and not true scarcity values, has dramatically too little unskilled labor.

Here is an example of the odd economics this creates. Here is a automated check out machine at a convenience store in the UK charging me for use of plastic bags. Every economist; left, right, center, agrees that prices should reflect their total, all-in, scarcity value (including externalities, such as environmental damage, that might be under-priced by a free market). But I am being charged this fee for a plastic bag by a machine that has been invented and is being used because a distortion in markets that is larger than the under pricing of plastic bags, by orders of magnitude. Unskilled labor in the UK doesn’t reflect the true global scarcity of labor, it reflects massive policy based distortions to the mobility of labor in the UK that create local prices and availability of labor that do not reflect true economic costs and induce people to use far too little labor and use cheap capital and technology instead.

Using too little labor is bad because far and away the largest proximate cause of poverty is low wages for low skilled labor. Since raw labor is far and away the most important asset of the world’s poor anything that drives down the demand for labor creates a negative externality by raising poverty. So just as using too many plastic bags harms the environment more than its benefits, using too little labor relative to its true scarcity and productivity increases poverty more than its benefits in economizing on labor.

A second example, which gets all of the issues of private and public costs and incentives so badly muddled it is tragi-comic. I went to Norway awhile back and the hotel I stayed in had this door hanger. Given the very high cost of labor in Norway the hotel has powerful incentives to induce its customers to not have their rooms made up as this is likely the largest single element of their marginal cost of providing a room. So they are, in their private interest given costs, trying to convince their customers to voluntarily forgo having their rooms cleaned as it increases their profits. I understand that.

The hotel is trying to convince customers to increase hotel profits is by convincing the customer they are more virtuous if they don’t have their room cleaned because (a) “it saves our planet” and (b) they will donate to victims of human trafficking.

First, the real virtuous thing to do if you want your consumption to reflect externalities is to insist that the hotel use the unskilled labor to clean your room every day (maybe twice a day) because labor is not globally scarce and the hotel should do the virtuous thing and use more unskilled labor reflecting is global scarcity not distorted domestic price.

Second, while not having one’s towels laundered might have some externalities if water and energy are under-priced I suspect these distortions are trivially small compared to actual costs, as water in Norway is priced and electricity in Norway is priced and these prices to the hotel may already reflect true environmental scarcity (particularly if electricity is from non-carbon using sources). So the idea this “saves the planet”–as opposed to “saves on the hotel’s expenses”–is a bit surreal.

Third, and this is truly where my head starts to spin, is the connection with human trafficking. So, Norway imposes an effective ban on low skill migration. This means that the wages for someone in Norway to clean your room is ten times higher than any income from employment a typical person could get in a poor African country. At these wage differentials, that actually reflect true productivity differentials, that is the hypothetical African worker’s marginal product in Norway really does justify the high wage, there is enormous pressure for people to get into Norway, legally or illegally. This desire to move to engage in perfectly ordinary labor–like cleaning rooms–is thwarted by Norway’s policy of restricting labor movement into Norway. This creates demands for “human trafficking” as people want to avoid the restrictions Norway has placed on their opportunities. So this hotel wants you to collaborate with them in driving down global demand for unskilled labor and in maintaining the distortions that make labor costly in Norway by not using unskilled labor to clean your room–and then make you feel good about that terrible choice by donating money to the victims of the illegal and extortionate and abusive industry those distortions create.

This could not be more economically wrong headed (except for about the hotel’s profits) and yet is accompanied by an encouragement to make the customer feel better about themselves.

Jesus once criticized people for “straining out the gnat but swallowing the camel.” People in the OECD today are busy using paper straws to strain out the environmental externality gnat while swallowing the the labor reducing effects of policy distorted labor costs, that are orders of magnitude larger, whole.

I completed a paper (ao-authored with Farah Hani) for the Oxford Research Encyclopedia: Economics and Finance on the economics of international migration. A lot of this was review of the wage differentials (like the Place Premium) but there were three new things.

First, this was the first time I had looked at the Gallup Survey data on whether people, if they were given the choice, would like to move permanently to another country and if so, which country. They then calculate the “net migration change” in the world if everyone who wanted to move did move to their first choice for all population and for youth. The graph is a scatter plot of that with the estimates of labor compensation per worker from the Penn World Tables 9.1 and is divided by terciles of wages (in ‘000s). The oil-rich countries (e.g. Kuwait, Saudi Arabia, etc.) are excluded. To make the graph more readable I compress some of the smaller high attractiveness countries, like Canada whose youth population would increase 343 percent. The median youth population loss in the bottom tercile is 29 percent and the median gain for highest tercile of wage countries is 111 percent–and the USA is right on the median. The middle tercile has very high variance as some regional destinations (e.g. Botswana, Panama) gain alot and some would lose more than 40 percent (Tunisia, Peru).

The second new thing to this paper was to revisit why labor mobility across countries has not been a bigger part of the development agenda (whereas labor mobility within countries, rural to urban, has always been seen as a part of the development process). Besides the obvious political explanation (rich countries didn’t want it to be) there is the point that if one’s growth theory had rapid convergence of TFP (as would be the case if TFP were primarily driven by easily portable codifiable knowledge of the type that “knowledge as an externality”) then accumulated factors (capital and human capital) should move to (newly) high TFP and low factor abundance places. However, and this is one of the biggest facts we learned about development from accumulating data and experiences with growth over time, TFP did not converge. Rather in most regions of the world TFP growth was slower than in the OECD. A lack of TFP convergence puts labor mobility back on the table in a big unavoidable way.

The third was a single paper but it is one of the most striking and borderline macabre economics papers I have seen. It is a “field experiment” by Bah and Batista about the expressed desires of youth in rural Gambia to move illegally to Europe. 46 percent express a desire to migrate illegally to Europe even though they estimate the risk of death is very high. When provided in the field experiment with the “correct” estimate that the trip involved a 1 in 5 chance in death the expressed desire to attempt the journey went up. The fact that about half of rural youth say they would risk a 1 in 5 chance of death to migrate illegally to Europe speaks volumes.

Yesterday I read”Why Chimpanzees Can’t Learn Language and Only Humans Can” by Herbert Terrace. His argument is that chimps (or any other animal) cannot learning language not because they cannot learn grammar (which was key to the Skinner-Chomsky debate about language) but because they cannot learn words. Terrace argues that animals cannot learn words because they lack two key features that humans gain as very small babies that are precursors to language: intersubjectivity and joint attention. He argues these arise mental features from the uniquely human feature of extended helpless of the infant and the resultant cradling of the key caregiver.

I often being teaching about economics and public policy by talking about baby ontology. Even toddlers have a pretty clear “two kind of thing” division of the world: agents and stuff. If I am trying to understand “why” something happened (in part to make it work for my purposes) I divide the world into agents who have purposes and stuff that has properties. Human beings can mostly be successful through their entire lives–even as top professionals like doctors or engineers–with this baby ontology. If I can successful understand agents and their purposes and stuff and its properties I can manage my world.

But baby ontology is not enough to do economics or public policy or social science. There are other ontological categories besides agents and stuff and they do matter for outcomes. But the tendency to rely on baby ontology, which after all, we have been perfecting since we were months old, is massive and we tend to “understand” complex systems by anthropomorphism. James Carville, a political adviser of Bill Clinton famously said ” I used to think if there was reincarnation, I wanted to come back as the president or the pope or a .400 baseball hitter. But now I want to come back as the bond market. You can intimidate everybody. ” This is insightful because governments really are powerfully constrained by the response to their actions mediated via bond markets, but it is funny because we all recognize the category mistake. The bond market is not an agent. We can imagine being reincarnated as a baseball player or a cat or a horse or a bat (with apologies to Nagel) as we regard those all as agents but not as “the bond market” as, ontologically, the bond market is neither agent nor stuff.

The key insight of Adam Smith, formally modeled in Arrow-Hahn-Debreu general equilibrium models and the two welfare theorems, is that an “economy” is an ontological something that has properties, really interesting properties, that are not the result of the purpose of any agent in the economy. Under very strict and very “unrealistic” formal conditions one can show the “economy” has an equilibrium (which itself is truly stunning and not “common sense”) and moreover this equilibrium has the property of being Pareto-optimal (no agent can be made better off without making another agent worse off). Even prior to the debate about how useful these theorems are for thinking about real policies in real economies, a conceptual contribution of these formal models is a “proof by construction” that complex systems have interesting properties (no one doubts that Pareto-optimal is an interesting property) and that these are emergent properties of the system–the way in which the agents’ interactions are structured–and not the result of the purpose of any given agent.

Of course it is a testament to the power of baby ontology that when economists, like Adam Smith, try and explain that baby ontology does not exhaust the modes of causal explanation he reverts to the metaphor of an “invisible hand” and explains that the agents “as if guided by an invisible hand” are led to produce good outcomes they did not intend. But this anthropormophizing metaphor leads to a false understanding. It leads people to say, “oh, I get it, there is an agent called a “market” and it has purposes and accomplishes its purposes by using its hand.” The main point about the invisible hand is not that it is invisible but that it is not a hand. The “market” is not an agent. The “market” is a complex system that patterns the possible interactions among agents and the market has properties but the market has no purposes or intentions.

Believe or not, all of this is to set up the question of what does it mean to explain “poverty” and in particular, what would an explanation of the causes of cross-national differences in the level of, and pace of reduction in, poverty look like? The official World Bank data on “dollar a day” (PPP $1.90 with inflation) poverty say that Zambia’s headcount poverty rate in 1981 was 47.3 percent and was 57.4 percent in 2015 and that Vietnam’s was 78.2 percent in 1981 and 2.2 percent in 2015.

Baby ontology explanations of the level and change in national poverty rates are just not even really plausible candidates. That is, explaining poverty as a function of the purposes of agents or their ability to adequately pursue those purposes or as the properties of the stuff in Vietnam and Zambia are going to have a very hard time explaining in any sensible way they Vietnam went from so much higher to so much lower. Did Vietnamese people suddenly change their purposes? Suddenly change their individual capabilities to pursue those purposes?

While one might, just might, explain why this person is poor and that person isn’t poor with baby ontology to understand why some countries (or regions) have high levels of poverty and why some countries have been successful in reducing poverty and other have not almost certainly going to have to rely on analysis of the emergent properties of systems, which can be better or worse at creating conditions that allow people to take actions that pursue their purposes.

More simply put, poverty is not the consequence of poor people, poor people are the consequence of poor places and poor places are the consequence of poor “systems” (like markets, capable organizations, political structures, norms) that allow people to interact and cooperate to pursue their purposes. The systems can change over time in ways that unleash human potential, create prosperity, and (essentially) eliminate poverty very rapidly.

However, in the hurly burly of public opinion (either mass or elite) system explanations are at an enormous disadvantage as nearly every person is a practiced expert in baby ontology and almost no one is expert in thinking about systems. When your child says “Daddy/Mommy tell me a story” they want a narrative about agents and their purposes and that is how any person with common sense understands and explains their world. People really want a clear narrative where the “solution” to poverty is that good person X does action Y for poor person Z. But those stories are wrong. At the large scale poverty for person Z is reduced when Z lives in a productive place, full stop, and productive places are nearly all about systems and emergent properties of ontological categories that are neither agent nor stuff.

I wrote a book in 2006 Let Their People Come that suggested that development friendly, rights respecting, rotational, occupational (e.g. low/medium skill) labor mobility was an under-used tool in the field of development.

Needless to say, one of the many things that the financial crisis of 2008 did was eliminate any discussion of the need for labor mobility in tight and tightening labor markets in the OECD countries.

However, I think labor mobility is back on the agenda. Bryan Caplan’s new book is a graphic non-fiction book (in which I have a small cameo) that stakes out, self-consciously, the far edge of the Overton window by arguing for open borders. The Economist had a recent Special Report on immigration issues by Robert Guest. Jason De Parle has a recent great book A Good Provider is One Who Leaves which is an excellent account of one Philippine family’s experience with taking jobs abroad (and afloat).

A theme of my book was that there is gridlock in labor mobility as irresistible forces meet immovable ideas. One of the irresistible forces for greater labor mobility in the OECD is the inversion of the demographic pyramid which implies that there just will not be enough workers in the future, particularly for labor taxes to sustain the social security and other social protection schemes. This force, of increased absolute numbers of elderly and yet absolutely fewer workers, is now 13 years closer than it was in 2006 and shows no sign of abating. Even Japan, a country that has traditionally been very resistant to any migration, has recently been signing agreements with countries to allow more labor to enter.

Since the 2006, because of the research of Michael Clemens and others, we have better evidence about the magnitude of the Place Premium (the wages gains to movement from a low productivity place to a high productivity place) and about the impact in the US of labor schemes.

The big question in my mind is how to create the political conditions, both globally and nationally, for “more and better” rotational labor mobility. My conjecture is that in order for this type of mobility, which is full of risks of exploitation of workers, work well one needs to create a “good industry” to be an “industry for good.” That is, the facilitation of rotational labor mobility from identification and recruitment in country to protection while abroad to planned return is itself a potentially large industry. Just as airlines and airports have made international air travel safe, an association that looked to make the “people who move people” a safe and law abiding and non-exploitative industry by bringing together in a pluri-lateral (not “global”) fashion governments, sending country industry, and using sectors to promote an upward pressure for more and better mobility might just work. I am currently working on a very small project (LaMP–Labor Mobility Partnerships) to explore the feasibility of this idea.

Here is a presentation I gave in Nepal at the invitation of my ex-students on this topic.

This article was finished in June 2019 for a collection of essays about RCTs (including a piece by Angus Deaton) and a really excellent piece on the rhetoric of Poor Economics (which argues the book is indeed poor economics but fantastic rhetoric in the Aristolean sense of merging logos, pathos, ethos) so is not at all a response to the Nobel Prize and all the hoopla. But, it turns out in a pre-written rebuttal to the arguments Banerjee and Duflo make in their recent Foreign Affairs article.

Turns out, as expected, we agree on a lot of things, like that getting growth going is the best thing to do. We also agree that their argument for RCTs hinges on two things.

First, it hinges on a really striking pessimism about the prospects for facilitating more rapid economic growth in any given country. They make a big deal out of that we economists cannot be exactly sure about what to do to promote growth (beyond obvious basics that have mostly been done) but, as I point out in the article, making modestly better decisions about big things is still far more important than great decisions about small things.

Second, it hinges on optimism about the extent with which the micro knowledge that RCTs generate can or will scale. Their argument is: “We don’t know how to do growth so lets invest in education and build courts that work.”

But there have been far more successful instances of accelerating economic growth in ways that led to sustained poverty reduction that there have been successful instances of going beyond expanding schooling (which most countries did and are doing) to improving learning outcomes. Some preliminary calculations with DHS data about literacy and schooling suggest that of out about 50 countries with data only 1 has had sustained success in improving literacy acquisition per year of schooling. So the idea that “we” (who?) don’t know how to accelerate growth (though, by the way, a fair number of countries have) but we do know how to “invest in education” (at scale) is not at all obvious on the face of it.

Similarly, I have done work on what we call “state capability” (of the type that would include “making the courts work” and again, at least at the aggregate level of the standard measures of state capability there is much less success on this than there is on growth–over half of countries have, by these standard measures, deteriorating state capability and most are making very slow progress. Again, the idea that “we” don’t know how to accelerate growth but do know how to improve the functioning of the courts is kind of hard given that there are many instances of the former and very few of the latter.

Spent an interesting couple of days in Paris with the chapter authors of a proposed book on RCTs, mostly with those amenable to the correct technical claims but skeptical of the outlandish claims that have been made. One of the authors is using the Gartner Hype Cycle as the organizing frame, thinking maybe we are headed to a relatively high “productivity plateau” after peak hype.

My paper argues that the RCT are not the disease but a symptom. The disease is the shift of focus from “national development” to “kinky development” or from doing and promoting developing to just trying to use limited, targeted programs to mitigate the worst consequences of the lack of development. It is pretty obvious that if one is asking questions about how countries successfully complete the four-fold transformation of development (productive economies, responsive polities, capable administration, and equalized social relations) the parts of that agenda for which RCTs can, even in principle, provide reliable (much less “rigorous”) guidance is tiny. It is only when one creates the illusion that “development” is about low-bar, sector specific targets that one can even imagine the design of “programmatic” interventions is at all central (and even then it is highly dubious). So RCTs are the research fad as a handmaiden of the larger shift to the kinky among North/Western agencies, which itself is a political shift due to their politics, having little or nothing to do with the facts of development or the interests of developing countries.

Attached is a very early draft of my chapter, will be substantially revised.

I spoke at the 25th Anniversary conference of the ERF (Economic Research Forum)–a consortium of economists from the Middle East/North Africa in Kuwait on March 12th. I spoke on the fact that the largest policy induced price distortion in the global economy (and probably in the history of mankind) is that the US price of low/medium skill labor is 5 to 10 times the world opportunity cost (supply). This makes labor artificially scarce.

This has the consequence that we live in the perverse world in which the scarcest resources on the planet–the very high opportunity cost scientific, technical, and entrepreneurial talent–is devoting itself to inventing devices to economize on, and hence drive down the demand for, low skilled labor.

This makes the pressing social, political, and economic challenge for most developing countries of finding good jobs for their youth bulge of (by global standards) low skill labor harder.

The massive distortions in the price and availability of low skill labor in rich countries is causing the world’s geniuses to devote their efforts to making things worse for the world’s worse off. There isn’t even yet a global discussion on how to stop this bias in the the pattern of technological change.

I am reading (parts of) Hume’s History of England (jet lag and all that) and he raises “damning with damning praise” to new heights in his discussion of Cromwell, who he clearly detests everything about but must acknowledge Cromwell’s amazing success at achieving his aims.

“This artful and audacious conspirator had conducted himself in the parliament with such profound dissimulation, with such refined hypocrisy, that he had long deceived those who, being themselves very dexterous practitioners in the same arts, should naturally have entertained the more suspicion against others.”

And the volume is full of such wonderful, old fashioned, sentences.

Sam Hickey and Naomi Hossain launched a new book with Oxford University Press today: The Politics of Education in Developing Countries: From Schooling to Learning. The book is case studies of the politics of education in six developing countries using a new theoretical framework, also used by Brian Levy (who contributes a chapter here) in his (edited) book on the politics of education in South Africa (The Politics of Governance of Basic Education: A Tale of Two South African Provinces). I will have more to say about Brian’s book in a week or two (I am in the middle of a review essay). According to the OUP website this book “Includes critical commentaries from leading scholars in the field”–one of those is my chapter. I could tell you what I said, but then you wouldn’t read it.

Here is a number to remember:  .994 (and not because 994 is the country telephone code for Azerbaijan).

The measure of poverty most commonly used by the World Bank is the “headcount”: the proportion of people below a poverty line, a fixed level of income or consumption expenditures (CEX) per capita.  The Foster, Greer and Thorbecke (1986) measures of poverty are weighted sums of people from a given distribution of consumption expenditures (or income), and the headcount is the simple case where the weights are equal for each person, irrespective of how far from the poverty line their CEX is.

This leads to a simple question:  “How much of the observed variation in headcount poverty rates across countries is due to variation in the median of the distribution of consumption expenditures?”  

The answer, shown in Figure 1, is (roughly) “All of it.”  The R-Squared of the median (with various powers to account for non-linearity) for explaining headcount poverty for the three poverty lines is:

• $5.50 per day, R-Squared=.988, correlation(poverty, predicted)=.994
• $3.20 per day, R-Squared=.988, correlation(poverty, predicted)=.994
• $1.90 per day, R-Squared=.983, correlation(poverty, predicted)=.991

The simple correlation between the actual $3.20/day or $5.50/day headcount poverty rate and headcount poverty as predicted using only the median of the country distribution is .994 and for $1.90 it is .991.  These are about as high a correlation as real world data can produce. 

Figure 1:  Headcount poverty rates are extremely highly associated with median consumption expenditures

Source:  Author’s calculations based on data from:  “PovcalNet: the on-line tool for poverty measurement developed by the Development Research Group of the World Bank”

The regression uses the 389 country/time observations from the World Bank data that are based on consumption expenditures (not income) and recent data (not distant extrapolations). Since headcount poverty is a partial integral of a distribution of consumption expenditures the relationship between the median and headcount has to be non-linear.  I use powers of the median from -2 to 5 to allow for flexible non-linearity.

Nancy Birdsall and Christian Meyer have argued that for development issues “The Median is the Message.”  For headcount poverty they are completely right.  The answer to the question:  “Why does a country at a given time have headcount poverty rate it does?” is “Because of its median of consumption expenditure.” Pretty much full stop.   Conditional on the median, any and all other factors or variables can explain at most 1.2 percent of the variation in country headcount poverty rates (maybe 0, but at most 1.2).  

You might be saying, “Lant, why are you making such a big deal of this correlation?”  Well, thanks for asking. 

This very tight correlation is not built in.  One can usefully decompose (as many have done) the difference in poverty rates comparing two distributions (between countries or over time) into three elements on the (mostly accurate) assumption the distribution is log-normal (that is, the natural log of consumption expenditures is distributed as a Gaussian normal distribution):

1. Differences in the central tendency of the log-normal distribution,
2. Differences in the variance of the log-normal distribution and
3. Differences in the distribution below the poverty line being more or less favorable to poverty than would be expected of a log-normal of a given central tendency and variance (as the log-normal is a two-parameter distribution this forces an exact shape).

How much of the variance in headcount poverty is an empirical fact that depends on the actual distributions of consumption expenditures across countries and the fact that (2) and (3) account for 1.2 percent of the variance could have been otherwise, it is not cooked into definitions.  In fact, one can easily imagine policies or programs that would bring up the lower tail, and hence reduce poverty, much more than the log-normal would predict.  So it is a striking finding that both differences in variance (inequality) with the assumption of log-normality and deviations from log-normality below the poverty line together account for at most 1.2 percent of observed variation in poverty.

Included in the country/time varying factors whose variation in the observed data cannot explain more than 1.2 percent of the observed variation in headcount poverty rates are things like: “budget (government or other) devoted to anti-poverty programs” and “efficacy of the design of anti-poverty programs” or “whether the country’s anti-poverty programs are ‘evidence based’” or, for that matter, any interaction of those factors, like:  “whether a country devoted budget to well-designed anti-poverty programs based on evidence.”  The median explains nearly all variation in poverty across countries with no reference to targeted programs of any kind: not micro-credit, not conditional cash transfers, not chickens, not livelihood programs, nothing that claims to impact poverty without changing the median.

Given the amount of time, energy, intellectual firepower, academic publication, and advocacy that go into discussions of anti-poverty programs one might think they are a large part of the “solution” to global poverty.  But they just have not been.  If your median consumption expenditure went up then your headcount poverty went down and nothing else that any country has done besides that seems to be very important in explaining poverty reduction.

The relative importance of growth of median consumption expenditures versus “all else” can be illustrated with two different poverty lines at two different levels of income, the “extreme poverty” penurious poverty line that the World Bank often uses (but which I think is fundamentally illegitimate as it is too low) and the “$5.50/day” line (which I think is still too low).

The predicted level of $1.90/day “extreme poverty” for a country at the average income of the lowest quartile of countries is 72.2%.  If its poverty rate were better by one standard deviation of the residual conditional on the median it would fall to only 68.6 percent.  Even if it had the best poverty conditional on its median for any country in the bottom quartile it would fall by about 10 percentage points to 62.7 percent.  In contrast, if that country grew by two percent a year for 20 years (which is roughly the average growth in the post WWII era) poverty would fall to 35.9 percent—about in half.  If it grew at 4 percent per capita for 20 years (this is about one standard deviation above the average growth of 2 percent) predicted headcount poverty would fall to 12.1%.  Sustaining rapid growth starting from a low median consumption expenditures reduces poverty 50 percentage points more than having the best observed poverty conditional on the low median.  With sustained growth half the population moves out of extreme poverty compared to 10 percent even the best observed poverty with stagnant income (and just to be clear, the data here don’t tell what accounts for these observed low poverty rates conditional on the median).

Figure 2:  Even getting to the best headcount poverty for a given median expenditures versus the average produces a small gain relative to the poverty reduction from sustained growth of the median

I do the same exercise with the second quartile of income and the $5.50/day poverty line, with roughly the same results.  Even the best performance for poverty conditional on median produces gains much, much, smaller than the gains in poverty reduction from sustained rapid growth.

The results of these regressions are just facts about the world and do not directly reveal causal structures.  In particular, there may well be cost effective poverty reducing programs that merit support by governments and/or philanthropists.  The cross-national correlations can only speak to what have been the correlates of poverty, not what could be.  But, while one doesn’t want to over-interpret facts, neither does one want to under-fact interpretations of very specific and particular empirical findings about specific programs either. 

The next time you hear the phrase “solve global poverty” remember the number: .994.  If what follows “solve global poverty” isn’t about raising median consumption expenditures a very good question is: “why not?”

(This blog is titled “Part I” because I plan a Part II that does a bit more on the technical issues of these types of decompositions and a Part III that discusses a bit the broader implications.  But, unlike the Lord of the Rings (in which all three were filmed at the same time) these are not yet written and the future is unpredictable).