Lant Pritchett

My 2022 paper on national development showed that the fourfold transformation of high productivity (economic growth), state capability, a responsive government, and social cohesiveness (equal treatment of all citizens) was empirically necessary and empirically sufficient for high levels of human wellbeing.

Knowing that this four-fold transformation or “getting to Denmark” reliably leads to good outcomes for wellbeing does not, however, say how to make progress. Some people respond, “well, ‘national development’ would be great, but ‘we’ (whoever ‘we’ that is) don’t know how to do that, so let’s stick to what ‘we’ do know how to do, like implement projects or programs.” Fair point. Knowing that faster than light travel would allow you to make a fortune betting on sports and stocks doesn’t make it possible, or even make a research program on how to achieve faster than light travel interesting as our best available theories say it is impossible.

But economist’s “best available theories” do not say accelerated economic growth is impossible, and cannot, as growth accelerations happen frequently (Hausmann, Pritchett, Rodrik 2005, Pritchett, Sen, Kar and Raihan 2016, Gootjes, de Haan, Stamm, and Yu 2024) and the sustained rapid growth episodes have transformed countries–and the growth accelerations in the population giants of China and India, have transformed the world.

Moreover, the legitimate and deserved criticism of the growth regressions as a research method and of the practice of “one size fits all” structural adjustment recommendations and the Washington Consensus was acknowledged and (mostly) absorbed (e.g. World Bank 2005). This led to new paths in the research into the question of how to produce more relevant and reliable recommendations based on growth diagnostics (e.g. Hausmann, Rodrik, Velasco 2004 ), analytic narratives incorporating context and politics (Rodrik 2003, Pritchett, Sen and Werker 2017), new approaches to structural transformation (e.g. Hausmann and Hidalgo 2009), and more sophisticated analysis of episodes to examine the contingent connections between policy reform and growth episodes (e.g. Peruzzi and Terzi 2021 ).

And recent decades (until at least COVID) have actually been quite good for growth in the developing countries and there is evidence that after decades (centuries) of divergence, big time (Pritchett 1997) there has been convergence (poorer countries growing faster than richer countries (e.g. Patel, Sandefur, and Subramanian 2021, Kremer, Willis and You 2021).

Recent growth raises the question of the joint dynamics of the elements of national development. That is, while many countries are experiencing rapid growth, and hence are “getting to Denmark” in levels of GDP per capita, is this growth accompanied by progress in state capability, responsive governments, and greater social cohesion and equal treatment (rule of law)? An interesting case is Bangladesh. For some time now people have been celebrating the “Bangladesh Miracle” of sustained rapid economic growth (and improvements in other indicators as well). But this was accompanied by concerns about the “Bangladesh Paradox”–that economic growth was rapid but accompanied by stagnating (or deteriorating) conditions of governance and “institutions”–and Raihan, Bourguignon and Salam 2024 asked “Is the Bangladesh Paradox Sustainable?” The events of August 2024 revealed that the existing political regime could not be sustained.

The theme of the Asian Development Bank Institute’s annual conference for 2024 was “Can Asian Economies Forge a High-Income Future and Avoid Burn Out?” For that conference I wrote the attached paper: “Exits from the Four-lane Highway to National Development: What are the Risks to Sustained Economic Growth?” The paper is mainly descriptive (with lots of graphs) and examines the association between the recent growth rates of GDP per capita and the growth rates of a measure of State Capability. The main empirical results are that: (a) having low levels of state capability is not (much) of an obstacle to country’s beginning episodes of rapid growth, (b) economic growth and state capability do not appear to reliably grow together as many countries have been experiencing rapid growth for more than two decades with stagnating or deteriorating state capability, and (c) if one looks for a “high income future” based on extrapolating recent past performance one has to believe that countries will be able to achieve high income at levels of state capability far, far, lower than the state capability of the now high income countries. Whether that is really feasible and how this growing tension will be resolved are, to me, super-interesting questions without much current research (theory or empirics).

The two other things along this line of the joint dynamics of national development are not even papers yet, but speeches given in December 2024.

One was a speech to the 3rd annual Pathways to Development conference at LUMS in Lahore Pakistan, which as given remotely on December 17th. As the theme of the conference was “Governance and Inclusion” the title of my talk was “Governance and Inclusion: “End of History” or “Hell in a Handbasket?” This plays on the theme that in the aftermath of the end of the Soviet Union Francis Fukuyama argued this was the “end of history” (in a quasi-Hegelian meaning of “end” and “history) as the contest over alternative systems that dominated the 20th century was effectively over. This might suggest that “getting to Denmark” would then be the common outcome. But the triumph of (managed) capitalism, strong bureaucracies, and liberal democracy was a tad pre-mature. Lots of countries have had only one of those, or none. The video of the presentation is here and the power point is attached.

I gave a speech more focused on the “hell in a handbasket” theme, using data to look at the joint evolution of GDP per capita, state capability, and other measures of governance. This was at the XKDR Forum in Mumbai India on December 10, 2024. In this presentation I lay out the data on the joint evolution of GDP per capita and governance indicators (state capability and measures of “responsive state”) and then explore specific mechanisms that allow for rapid growth with deteriorating governance.

These three products (paper and two speeches with new empirical results) are just trying to: (a) lay out the questions about the joint dynamics of the elements of national development, (b) provide some descriptive facts with the data we have, and (c) tentatively positive some models (or elements of models) that allow for the range of what we have observed over recent decades, which is examples of very rapid growth but with weak, and deteriorating, governance (on the usual measures). I am just scratching at the surface of these deep questions so far.

As a contribution to a Growth Summit in Washington DC on October 16 and 17 2024 I wrote a synthesis of recent empirical work of mine articulating that economic growth (and more broadly national development) is central to achieving goals for improving human wellbeing. I show that across an array of potential goals: (i) reducing poverty, (ii) improving the ‘basics” of material well, and (iii) general indicators of social progress.

This draws on, but also improves on and substantively expands on, four previous works (some published, some just posted): (a) “Economic Growth in Five Figures“, (b) “Randomizing Development: Method or Madness?” (which contains empirical analysis of the connection between headcount poverty and median income/consumption), (c) a new(ish) still unpublished paper on basics of human material wellbeing and GDP per capita, and (d) a paper on “national development” and measures of “social progress” (both in scare quotes as the words refer to specific concepts and measures) “National Development Delivers.”

Two things are new, or at least much better articulated in this paper than in the previous works separately.

First, I make clearly the case for “inclusive enough” growth. That is, previous work on the relationship between economic growth, taken as the growth in aggregate GDP (or consumption) per capita, and other indicators of wellbeing is that this approach “ignored” inequality in income, either in the level or in “growth incidence.” This paper takes the approach of building concerns about inequality directly into how the distribution income and growth incidence affects measures of wellbeing. So, just as an example, if one is concerned about child mortality as an important indicator of wellbeing (as it is) and one worries that “growth” isn’t necessarily connected to improvements because growth of the average may be driven by growth of “the rich” (e.g. “growth incidence” is pro-rich) then the elasticity (responsiveness) of child mortality to the actual pattern of growth incidence can be calculated as the interaction of the growth incidence (which, say, decile, grew faster or slower) and the elasticity of child mortality to growth in income/consumption at the level of income of each decile. This integrates direct concerns with important normative indicators (and so acknowledges economic growth as a means to an end) with concerns about inequality using empirically estimable relationships between levels of income and indicators of wellbeing. This approach can therefore document that the incidence of growth matters (so that pro-poor growth incidence has bigger impact than pro-rich growth) but can also document by how much it is better so that one can measure the impact of wellbeing of growth that is, while “pro-rich” nevertheless “inclusive-enough” to have powerful impacts on non-economic measures of wellbeing and social progress.

Second, this paper also makes three points about the case “against” economic growth. The case is often made (although often this is just implicit) that economic growth isn’t necessary for improving wellbeing and, that is OK because there are other project or programmatic approaches than can achieve goals for, say, poverty reduction. One, in general this claim just isn’t true. That is, while there might be cost-effective programs for reducing poverty this doesn’t imply that these programs can be scaled by a country to achieve any given desired reduction in country level poverty. The paper shows that there are no cases in which countries achieve very low levels of poverty at very low levels of income and there is very little variation in poverty rates among countries with similar median consumption. Growth in median consumption is an empirically necessary condition for large progress in poverty reduction. Two, the debate about economic growth and programmatic approaches often poses a false dichotomy, arguing that proponents for sustained rapid economic growth are somehow necessarily opposed to effective action to, say, reduce poverty or improve health or improve education or expand publicly provided sanitation. This just isn’t true. Most proponents of taking the actions needed to accelerate economic growth are happy with a “growth plus effective public action” approach. Three, often implicit in the debate about accelerating economic growth versus, say, programmatic approaches or reducing inequality is an assumption that achieving accelerated economic growth is hard or impossible or exactly how to accelerate growth is unknown or even unknowable but that in the same country conditions reducing inequality or implementing effective health improving programs is easy (or at least easier). This is possible, but it cannot be generally true as many countries have managed to achieve sustained rapid poverty reducing and wellbeing improving economic growth while not managing to reduce inequality or implement scaled effective public action in many domains.

I have presented this paper at recent seminars at Georgetown University School of Foreign Service, Harvard Kennedy School, and London School of Economics. Attached is the version of the presentation of the paper I gave at the Dean’s Dialogue series at LSE on October 10, 2024.

This is the title of a new column by the lead economic columnist for the Financial Times Martin Wolf that reflects some of my papers in discussing a challenge faced by all rich industrial countries: ageing. Ageing populations are leading these economically advanced countries into ratios of the labor force to historically unprecedented low levels.

The attached recent paper of mine (which is still a working draft and, as such, has not yet been refereed or published) documents the implications of the standard UN Population Division demographic projections for 31 rich industrial countries. Using the UN World Population Projections “zero migration” scenario and the ILO data on labor force participation rates I show that in the absence of migration and with constant labor force participation rates (by age and sex) by 2050 most of those ageing, rich industrial countries will move into ratios of labor force to population over 65 that will create massive economic and fiscal challenges. For instance, this ratio in Spain falls to 1–one worker for every person over 65 and in Italy it falls to .88 and the Europe wide average falls to 1.34.

This is the sense in which “immigration” (which I use in quotes because I want to analytically distinguish between modes of labor mobility) is “essential.” I don’t think anyone believes that anything like Spain’s existing social contract for its elderly population can be sustained by one worker for every person over 65 (until as recently as 1980 this ratio was above 3 in every country).

The paper emphasizes that the very pressures that make “immigration” essential also make it “impossible” politically. If countries were to make up for the smaller native born labor force through standard “pathway to citizenship” immigration this would to ratios of foreign born to native born among citizens about three times as high as that of the USA at its peak in the “open borders” of the early 20th century. It is hard to look at the recent elections in for the European Parliament or the ongoing 2024 Presidential election in the USA can conclude there is a political climate conducive to ratios of immigrants (who, of demographic necessity will have to come from the Global South in the aggregate) many-fold higher than the current levels.

I argue that a potentially solution to the tension that “immigration” is both essential and impossible is to shift to acknowledging that the question: “Who is allowed to legally reside and work in our country?” has not just two but three possible answers. One answer is “those who we (current citizens and voters) see as the future of us–those admitted on a direct (if perhaps lengthy and contingent) pathway to citizenship.” Another current answer is “those we allow as movers of distress (refugees, asylum seekers).” A third answer, which is already present in at least some form in nearly all countries is “Those who are allowed to reside and work in our country on a contractually time-limited basis.” The paper (together with an earlier companion paper focused on the politics) argues that a massive expansion of rotational labor mobility is a politically possible and administratively pragmatic.

And, and the underlying motivation of my engagement as a development economist, is that the calculations in the paper suggest that the annual gains to workers (and by extension their families and loved ones) in the Global South, workers who would otherwise lack attractive employment options, of large scale rotational labor mobility are in the trillions of dollars annually. The potential gains to development objectives from labor mobility are orders of magnitude bigger than the entire aid industry.

I was invited to give a keynote speech as part of a 10 speaker series as part of the celebration of the 75th year of CEPAL. I was invited with a specific title, which I pretty much stuck to, and which is the title of the attached longish essay that was the substance behind the speech (and which will by published in the CEPAL review along with the other essay).

The video of the speech is in this link, one main aspect of is a long and flattering introduction by the Executive Secretary of CEPAL, José Manuel Salazar-Xirinachs, which, naturally, I would encourage everyone to watch.

The paper covered some old ground I have gone over many times (the fact that schooling has expanded massively but the quality of learning in schools varies massively across countries), goes over some old ground I have not revisited in a while (the “where has all the education gone” and “does learning to add up add up” papers about the lack of correlation of “schooling capital” growth and GDP per capita (or per worker).

It adds two new things.

One, drawing mainly on new cross-national data estimating a cardinal measure of learning of a typical enrolled youth, both in the World Bank’s Harmonized Test Scores and the new estimates of Gust, Hanushek and Woessmann 2023, I tried something that almost never works, which is to run a simple OLS regression on GDP per capita with schooling, average learning, and an interaction term. The interactive specification always made much more sense to me than a “horse race” of including schooling and learning separately was it seems that adding a worker with a given level of S should contribute more if they have more learning and conversely a higher level of learning should contribute more the more youth are getting it. Anyway, I was kind of surprised that simple regression gave sensible results (and robust across the two measures of learning, which is not so surprising given they are highly correlated).

The other new thing is some speculations on my part of both how the PISA-like learning estimates can be so low and why that probably matters for the contribution of schooled youth to output. The reason is that while these scores are a single number (e.g. Chile’s score on mathematics in PISA is X), this is a single number with (at least) two dimensions, like area is length times height. The two conceptual dimensions of an assessment score on a domain are “coverage” (e.g. across different sub-domains of mathematics) and “depth” (the extent to which the assessed individual have more than just a rote or purely procedural understanding). This in part helps understand “how can kids have attended 9 years of schooling and yet score so low?” if the assessment probes for depth of understanding whereas the child’s learning has just pushed through coverage and emphasized a merely rote, memorized, or procedural understanding.

The second aspect is that if schooling just emphasized a rote/memorized/procedural understanding this implies that when youth, once out of school, encounter a problem for which a conceptual understanding would help them produce better answers and solutions and judgments about concrete problems their schooling hasn’t really equipped them for this.

I want to make this last point because, as an economist who was worked on issues of education and development for a long time now I find that when economists say things like “a countries economic growth/level of productivity is associated with its scores on a PISA-like assessment of mathematics” we can be heard (caricatured) as saying three things I am not saying.

One, that “education” is, or has to be, instrumentally justified by its impact on economic measures only. No, of course not, education has lots of justifications and lots of (potential) positive impacts on human lives.

Two, that economists in pointing out a connection of economic contribution to a score on assessment therefore are “recommending” and certain type of approach to the process of teaching and learning because we economists are focused on test scores. Again, no, of course not. In particular, we economists are often accused of wanting “drill and kill” or “back to basics” approaches to education, but this is definitely not what I am emphasizing. I am emphasizing teaching and learning practices in schools that lead to the depth of conceptual understanding necessary to produce valued and use competencies in adults.

Three, that economists in pointing out a connection between, say, mathematics scores and economic output must think that lots of jobs require workers to use, say, algebra or trigonometry or calculus (and point out that is ridiculous). But again, no, of course not. Yes understanding mathematics per se is important for lots of careers and professions. But the main point is that lots of people have to reason their way to correct judgments when faced with new and non-routine situations and that this kind of in-context application of skills is a hugely important capability for youth to acquire. But this emphasizes more “depth” of understanding and ability to apply knowledge and skills than any memorized formula or factual information.

The “Washington Consensus” started from John Williamson’s 1989 attempt to briefly summarize some things that were driving the development agenda and then became, even in the pre-“meme” or “viral” era a two word viral meme that has elicited strong reactions ever since. With more bravery than I have, LSE professors Andres Velasco, Tim Besley, and Ricky Burdett are attempting to construct a new, better, more sophisticated statement of where thinking about improving the human condition is, calling this a “London Consensus“.

As I spent eight years as the head of a large research project, Research on Improving Systems of Education (RISE) they asked me to do a chapter on where was the “consensus” about basic education in the developing world relative to the Washington Consensus. So, in the usual way of global discourse, boil down an enormous topic (or even the over 500 works produced by RISE) into 10,000 words. While I wrote lots of things for RISE (papers, notes (e.g. on “evidence” and the challenges of internal and external validity), presentations, synthesis reviews on key topics (e.g. framework for system accountability, teachers, politics of learning)), even including co-authoring a RISE “policy brochure” that tried to boil down the action implications of RISE into acceptably brief nuggets, this is my first post-RISE writing, where I am just me, Lant Pritchett, not representing anyone or anything.

The overall direct of the shift in consensus is clear, but what follows is more contested.

Williamson’s summary of the Washington Consensus was (as advertised) just the consensus of the day, both among economists and educationists (e.g. the Jomtien Declaration on Education for All)–roughly that “governments should spend money to expand access to schooling to reach universal completion of primary (and basic, maybe secondary) schooling in order to expand human capital.” This consensus is not longer particularly relevant in three regards.

One, it is a victim of its own success as, since so many countries have expanded schooling so much the remaining incremental gain possible from expansion of access/enrollment/attainment alone for building human capital is pretty limited.

Two, if we take one aspect of “human capital” to be the school acquired cognitive skills that school curriculums aim to teach and which assessments of learning on reading, mathematics, science (and other topics) measure then the “learning crisis” is now front and center. Arithmetically the stock of cognitive skills a youth has at age 15 (or 17 or 19 or 23)–but 15 is the PISA assessment age–is just the number of years times the (net) gain in the stock per year and the net gain in stock per year from school is the number of years enrolled in school times the (net) gain per year of schooling. The “learning crisis” is the realization that in many countries the “learning profile” (the description of stock of cognitive skills by year of schooling) is show shallow (so little learning per year of schooling) that even if youth are enrolled at age 5 and stay in school until age 15 they are still far from what are widely regarded as the skills need to cope successfully as adults in the world today (and, even less so, of the future they will live in).

{This is a topic I have been writing on for a long time (e.g. 2006 (with Deon Filmer and Amer Hasan), 2013) with more and better evidence being generated on stocks of learning over time (among the many, there are four key recent additions to the body of knowledge about learning outcomes are the PISA-D (analyzed in Pritchett and Viarengo 2023), ASER’s work on assessing youth skills in rural India in “Beyond Basics” (2023), the use of DHS data to create information on trends on learning over long periods in Le Nestour, Moscoviz, and Sandefur (2022), and the efforts to amalgamate the existing assessments into a single measure for (nearly) all countries by the World Bank in the Human Capital Index (N. Angrist et al 2021) and separately by Gust, Hanushek and Woessmann (2022 ).}

Three, the idea that government “spend” was a sufficient statistic for learning or creating human capital or that “spend more” is, in and of itself, useful policy advice has been completely discredited through evidence across countries, across regions, across sectors, and over time. McKinsey’s latest report (2024) on improving education “Spark and Sustain” (2024) opens with Exhibit 1 that shows that there are countries with public expenditures per student between $2,000-$4,000 (Vietnam) and over $14,000 per student (USA) achieving roughly the same results and among countries spending between $2,000 and $4,000 the learning outcomes range from “below poor” (Dominican Republic, South Africa), to “Poor” (Brazil) to “Fair” (China) to “Good” (Turkey, Vietnam).

If one concedes that “access” has been successful but one needs “every child in school and learning” and that “spend” alone may be necessary but is far from sufficient for improving learning outcomes, where is a new consensus?

The first element is that while it is an obvious truism that for learning outcomes to improve there must be “better” spending on teaching and learning practices (and materials) that are “evidence based” and “cost effective” this is insufficient as a guide to action. This “proximate determinant” of learning approach that “recommends” doing this or that particular action (e.g. hire teachers this way, use textbooks of this type, generate and use this kind of information on learning) has to acknowledge that the existing status quo and its proximate determinant outputs and outcomes are the result of the way the system of education now operates. To scale better teaching and learning practices and processes one needs those efforts to be embedded in a system that allows, facilitates, and rewards that. So the first element of consensus is that if we want large, sustained, improvements over time in learning outcomes one needs to change the system so that it is geared to, and coherent for, achieving that goal.

Beyond that, what exactly are the ways to change education systems such they endogenously produce the dynamics of accelerated learning, I recommend five broad “principles” for action. But I will be the first to admit these are broad principles and how to achieve the granular instantiation of those principles in practices in any given country (or regional) context is far from known (and hence any “consensus” on particulars is pre-mature).

Attached is the text where it stands now.

Attached is a first draft of a paper that is pair with another paper in progress.

This paper articulates why the combination of (a) “poverty reduction is the objective of development” and (b) the poverty line used to define and measure global poverty is (or any update of) the WDR 1990 ‘dollar a day’ poverty line is no longer a viable stance for a development organization or actor.

The basic point is that ‘dollar a day’ was always an unreasonably penurious poverty line and now, with the continued economic growth (even where not particularly rapid) nearly all developing countries have reached a point where there are very few ‘dollar a day’ poor. For instance, in the latest World Bank data Pakistan is reported as having only 4.9 percent of its population in the ‘dollar a day’ (which is now P$2.15) or “extreme” poverty. This implies that development actions that produced broad-based benefits in Pakistan would not benefit the “global poor” as only 1 in 20 Pakistanis is “poor.” This is just ridiculous (or perhaps beyond ridiculous, I am not quite sure).

The companion paper, which is underway and co-authored with Martina Viarengo examines the question “what is a plausible global upper bound poverty line?” Whereas ‘dollar a day’ is the answer to the question ‘what is the lowest a global poverty line could be?’, if one is going to use a range of poverty lines (as we argue that one must to have poverty reduction and development be synonymous) then one needs a range, a lower bound (‘dollar a day’) and an upper bound. This paper should be finished soon (in the academic sense of ‘soon’).

There is a very nice blog from June 2023 on the interactions between climate change and poverty on the World Bank Open Data site (based on also very nice recent research papers by the authors (working paper and published version). However, their opening paragraph contains a statement that is, kind of, on the face of it, correct, but, at the same time, can create confusion.

“Eradicating extreme poverty and stopping global warming can only be tackled together: reducing poverty without considering carbon emissions is a self-defeating strategy, as climate change impacts will threaten hardly won development gains. But despite rapid progress to decarbonize the world economy, reducing poverty by increasing people’s consumption today requires increasing carbon emissions, since economic systems in most developing countries still rely on fossil fuel energy. And the eradication of extreme poverty is so urgent that it cannot be delayed, especially to fix a problem that has largely been created by the richest among us.”

The potential confusion here is that this paragraph can be read as implying that for a given country addressing their own poverty that the problems of poverty and carbon emissions must be tackled together. In a direct causal sense, this is just obviously false. That is, take a country with currently high levels of extreme poverty, say, Ethiopia at 27 percent or Malawi at 70 percent. If one is concerned about their levels of extreme poverty to 2050 then it should be clear that, to first order, nothing about Malawi’s own carbon emissions between now and 2050 will have anything to do with the consequences of climate change for Malawi and its poverty. This is because Malawi’s own incremental emissions between now and 2050 under any growth scenario for Malawi are so small relative to the global total stock of carbon in the atmosphere in 2050 that the consequences for Malawi (or any other country) are vanishingly small. This is for three reasons, which are three deep features that make the challenge of climate change so very challenging.

One, carbon stays in the atmosphere a very, very long time and hence much of the relevant stock in 2050 has already been determined by emissions that have already happened. (This feature is in sharp contrast to some, but not all, other environmental challenges where the natural processes can remove pollution such that if the flow goes down the stock can go down.)

Two, Malawi by being both very poor and very small in population has very small total emissions and hence incremental emissions under various feasible growth paths are also very small. So, according to the figures reported in Our World in Data the emissions in 2021 in Malawi were 23.85 million tonnes whereas in the USA they were 5.93 billion tonnes. This means even if Malawi’s emissions were to grow by 50 percent by 2050 this would be the equivalent of a .2 percent (one fifth of one percent) reduction in US emissions. And China’s 2021 emissions are 13.71 billion tonnes.

Three, climate change in a global phenomena and depends on the total global stock and hence all countries are affected by the totals, irrespective of their own emissions. So nothing Malawi does about carbon emissions will have any first order effect on Malawi’s damages from climate change compared to about anything China or the US or Europe or India do.

I think there are three important reasons to make this clear.

First, this aspect of the “sustainability” of economic growth is very different from considerations where the feedback loops are at the country level and hence the default concern was that a given country was using up its own stock of natural resources (e.g. groundwater, forests, fertile soil, fisheries, oil, minerals, etc.) or creating its own environmental damages (e.g. air pollution, water pollution) that these would put a check on future economic production or wellbeing. But these feedbacks loops were national (or perhaps regional) and hence the question was whether the processes of economic growth and wellbeing gains could be sustained or were ignoring the “true” total stock of productive assets of the country, including their natural resources. In this respect climate change is very different as Malawi’s growth/wellbeing prospects may well suffer and become “unsustainable” due to climate change, but this will be because of the past and future actions of others, not (to first order) because of anything Malawi does about its own emissions. (And “Malawi” is just a place holder for “small to medium sized developing country with high extreme poverty). So, from the point of view of a given country there may be much more pressing environmental sustainability issues actually under its more direct policy control than carbon emissions. And, from a country view the actions to mitigate the negative consequences of climate change on their wellbeing might also be much, much, more important than controlling their emissions.

Second, there are enormous legitimate political pressures to dissemble about this. That is, controlling climate change requires many countries to cooperate in lowering their emissions relative to their BAU baseline. But these reductions are a “global public good” and every country understands that their efforts alone are ineffective. So many countries may reasonably say: “I will only cooperate to reduce emissions if all other countries cooperate.” This creates a pressure to get all countries to commit to action on their own emissions. This in turn creates pressures to tell countries: “The consequences for your country of climate change are going to be very bad so you should, as a country, commit to actions to reduce your carbon emissions, at the very least for any given amount of economic growth (e.g. carbon intensity) even if not in absolute terms.” But, while that statement is rhetorical powerful, everyone, not least developing country political leaders and policy makers, can see that it is logically flawed and false, or, at the very least, leaves out the key feedback mechanism working through global politics: “…if you fail to commit to climate actions this may cause other countries, whose emissions in total will really make a difference to climate change, to not take action and these emissions of other countries could make climate change consequences worse in your country.” Improving Malawi’s carbon emissions will only benefit climate conditions in Malawi through an indirect channel of political transmission.

Third, this is especially important for the World Bank, which is seeking to make the “global public good” of climate change more important in its agenda. But this is going to necessitate a tricky change for the World Bank. Previously, if the World Bank (via IDA) was doing projects in Malawi it was usually the case that the costs and benefits of the project were justified exclusively by the costs and benefits to Malawi as, in the end, the World Bank is a Bank and it makes a loan (even if, on IDA terms this has a huge grant element relative to market costs) to Malawi. But if the World Bank is making a loan to Malawi that affects carbon emissions one has to be clear whether this is justified as actions by Malawi based on the costs and benefits just for Malawians or whether part of the justification is that Malawi is making a contribution to the “global public good.” And if the answer is that the actions are contributing to the “global public good” with resources borrowed from a global development agency like the World Bank, what is the appropriate allocation of those incremental costs between the future citizens of Malawi (who have to repay the loan) and the rest of the global population which enjoys the benefits of Malawi’s actions. The tensions here are obvious and the stakes are high and lack of clarity here can undermine in the long-run the World Bank’s legitimacy.

Interestingly, after this introduction claiming “reducing poverty without considering carbon emissions is a self-defeating strategy” the blog, based on real research and actual numbers, makes exactly the opposite point. The incremental carbon emissions to reduce extreme poverty are inconsequentially small. By their calculations the incremental carbon emissions to “eliminate” (drive to 3 percent or less in every country) extreme poverty at historical relationships of GDP growth to poverty and GDP growth to emissions was only 4.9 percent of 2019 levels–which is roughly the amount global emissions have been increasing every three years since 2020. So, however negative the consequences of climate change are in 2050 it almost certainly will not be because countries grew at a pace needed to reduce extreme poverty. Rather what will determine climate change (including on countries with extreme poverty) is whether the countries responsible for nearly all historical and current emissions–and who do not have any appreciable extent of “extreme poverty”–did or did not reduce their emissions. T

This is related to a presentation about “sustained” versus “sustainable” development for a group in Pakistan in 2022, in which I make this point about climate change without the benefit of the more accurate and concrete numbers this recent research provides.

I have been contacted by journalists at times as a “critic” of RCTs. This worries me. My wife, who is a wise woman, has been encouraging me to stop writing and talking about RCTs because I will be seen as a cranky old kook who doesn’t “get it” and doesn’t have anything better to do than nit pick about the work of others.

This blog post is just to clarify that I am “for” things and spend most of my time and research and writing promoting those things. In a sports metaphor, I try and spend most of time playing “offence”–but at times, unfortunately, the best offence is a good defense and in a game like chess, you cannot just pursue your own strategy or you can end up checkmated.

Overall what I am actively “for” has been pretty constant. What I am “for” is that more and more people on the planet have the same opportunities that I have had and the same access to prosperity and safety and security and order and decent schools that I had growing up in the USA in the 1970s as a child of middle class white parents.  The expansion of opportunities for people to live a life of their choosing is how I think of the “development” agenda. And, as an economist, I am more about what can be done at the system level to shape the choices people have rather than imagining that I know better and should nudge people about the choices they make.

I regard the “national development” as an instrumental path to raising human wellbeing. National development is a four-fold transformation of countries to have (i) a more productive economy, (ii) a more capable state, (iii) a more responsive polity, and (iv) a more equal treatment of all citizens.

Here are four agendas in research and practice that I have been actively working on (and writing papers about) over the last 20 years or so.

Rapid and sustained growth in broad based labor productivity. I could say “economic growth” but this often raises hackles needlessly as people assume that “economic growth” must always mean “GDP” as currently measured. Like all professional economists I know: (i) don’t regard GDP as a direct measure of wellbeing at all, (ii) acknowledge the many limitations of GDP in measuring and valuing “true” economic output (and in keeping track of “wealth”) and (iii) am more than happy to put more weight on the gains in income/consumption of poorer people than richer people. But, at the same time, GDP per capita turns out to be a handy and available proxy.

Rapid and sustained economic growth is empirically necessary and empirically sufficient for achieving nearly all goals in improving human material wellbeing (including (i) the reduction of standard measures of poverty, (ii) any aggregate of the basics of material wellbeing (e.g. health, education, water and sanitation, child malnutrition), and (iii) (together with state capability) broad based indicators of human wellbeing, including (but not limited to) measures like the Social Progress Index that are exclusively based on non-money metric measures of wellbeing.

Higher levels of state capability. In addition to growth, a second transformation is having high state capability, which I define as having public sector organizations capable of effective implementation of the laws, regulations, policies, programs and projects that advance the legitimate goals of these organizations. In common sense terms this is having police forces that create order and security, education systems that equip kids with the skills and knowledge and capabilities they need, tax agencies that collect taxes without corruption, agencies that produce reliable infrastructure services (e.g. water, power, roads), etc. (And this definition does not resolve any particular “public vs private” debate as the “make vs buy” question of whether the direct producers of services should be public or private organizations does not imply there is no need for “state capability” as contracting out to or regulation of private providers requires state capability).

In “National Development Delivers” I show that both growth and state capability are strongly associated with cross-national human wellbeing measures and the strength of the relationship differs with (i) level of income (growth is more important at lower levels of income than at high levels), (ii) how “basic” the indicator is, with economic growth more important for more basic to wellbeing, and (iii) state capability is more important the more important effective collective action is to achieving good outcomes (e.g. the more it is a “public good”)).

The Building State Capability project at Harvard Kennedy School’s Center for International Development points a useful way forward on this agenda.

Effective education for all. I spent the last 8 years as the Research Director of RISE (Research on Improving Systems of Education) (the project recently ended, as planned, on March 41, 2023). This is part of a larger agenda built around the twin ideas that (i) while countries have been very successful at expanding schooling many countries have very low (and in many cases declining) levels of learning per year of schooling so “schooling” goals are being met while the true “education” objectives of schools are far from being achieved and (ii) accelerating progress in learning is going to require not just “project” tweaks inside “business as usual” but pretty thorough-going “system” reform.

Labor mobility. Given that “national development” doesn’t always happen and even when it happens it often takes a long time, this means that today (and into the foreseeable future) there will be large gaps in the productivity of the same individuals across places and hence there are, at the margin, massive income gains to allowing people to move from low productivity places to high productivity places. I am currently working to promote “more and better” pathways for people to move to opportunity via LaMP (Labor Mobility Partnerships).

I am not arguing these are the only aspects of development that are pressing or needed, there are lots of huge and important elements of the national development agenda that I have not been actively working on (e.g. infrastructure, energy, agriculture, urbanization, gender, law and order/policing, climate change, etc.). Mine are just the four that through my contingent life/professional trajectory I have ended up working on through a some combination of interest, opportunity, and an assessment these topics were “important, neglected, tractable.” And these have kept me very busy and productive as over the last 20 years or so I have written (mostly with co-authors) books, journal articles, papers, blogs, policy briefs on these topics.

Playing defense. That said, in order to promote the national development agenda (both broadly and in the specific domains in which I was active) I have played some defense. From my point of view, a major problem with development organizations and funding is that over the last 30 years or so there have been constant efforts to “define development down“, that is, shift the agenda away from “national development” (that sees the challenge as equal opportunity for people across the planet and hence has expansive, long-term, ambitious goals) to “kinky development” (here) that looks to narrow development to “charity work” by advocating just “low bar” goals in a few sectors (here). Development funding (both official and philanthropic) from the “North” or “West” (neither of which are of course geographically literal as the “North” includes Australia and the “West” includes Japan) has had a tendency, driven by their own domestic politics and needs, not the concerns in developing countries (here ), to shy away from the hard slog of the four-fold transformations and instead look to fund specific (often cocooned from systems and implemented by NGOs to bypass states) project “interventions” that are “effective” and “attributable.” So I have written papers (and blogs and speeches and etc.) arguing: (i) that “dollar a day” poverty is an obscenely low standard (here and here) and “dollar a day” (or other penurious poverty goals) cannot be goals around which a countries can build its development agenda (here); (ii) the Millennium Development Goal for “completing primary school” while ignoring any measure of learning of skills or capabilities was misguided (here and here–and then everywhere in RISE); (iii) that “kinky development” was not a development agenda that met the legitimate and pressing goals and ambitions of the governments and people in “the South” (here and here).

I think it is obvious to most observers that “national development” is the big agenda and “kinky development” is the small agenda. Moreover, it is also pretty clear that even within the “kinky development” agenda the kind of evidence that the “randomista” movement can, even in principle, generate is just one (small?) part of promoting and implementing effective progress even within that limited kinky agenda. So, while this (faith based) movement has been a relatively important part of academic development economics in the West, it is, at best, literally a footnote to the actual development experience.

As a thought experiment, the 13 developing countries with populations over 90 million people, which together account for over three quarters of the developing world population, are: China, India, Indonesia, Pakistan, Brazil, Nigeria, Bangladesh, Mexico, Ethiopia, the Philippines, Egypt, Vietnam, DR Congo. Each of these countries has an interesting, often turbulent, recent economic history, some with amazing success at improving the material wellbeing of their populations, some with mixed results, some catastrophic. Ask yourself: would it be plausible to write a recent history of each these countries, and even a recent economic history, or even of their recent “development” experience without any mention of RCTs or the generation of “rigorous evidence” about specific interventions?

I am for national development, which has an array of important elements within it. Over recent years I have (mostly) been doing research and working on four topics: economic growth, state capability, basic education, and labor mobility. As part of being a proponent of national development and of key issues within that, I have played some defense against the temptations of “kinky development” and, within that, spent some quite small amount of my time trying to play down expectations for what the very visible and very popular randomista agenda could really deliver in practice, on a number of fronts (here and here and here). But this does not make me a “critic”–much less an “enemy” or “opponent” of RCTs–this just makes me a consistent proponent of the effective promotion of national development as a pathway to higher human wellbeing.

Here is what seems like a pretty simple question. You have done an RCT in country X and produced a consistent estimate of the treatment effect of intervention I on outcome O. I am in country Y and have a simple OLS estimate of the partial correlation of I on O. How much should I move my priors from being centered on the OLS estimate from my country to the “rigorous” treatment effect estimate from country X? (And one could extend this to having done N RCTs in N countries that were not country Y).

Not only does this seem like a simple question, it seems like a pretty important question as without an answer to this question one cannot make any claims about the benefit-cost calculations of doing RCT research. If the scope of applicability of the finding isn’t known then, at best, one can only apply the rigorous treatment effect estimate to exactly the conditions in which it was generated–and hence the benefit/cost is likely to be very low (unless the country/program is massive scale).

Unfortunately, the answer to this question is not simple, the “intuitive” answer has very bad properties, and empirically, just using OLS from Y and ignoring the estimate that was rigorous in country X can produce better predictions.

1. the answer is not simple as an RCT, in principle (and often in practice) produces an OLS estimate and an RCT estimate and hence produces an estimate of the bias in OLS for the true treatment effect in X. This means there are two distinct pieces of “rigorous evidence”: the treatment effect estimate and the OLS bias estimate. Given an existing distribution of OLS estimates it is likely that these will suggest moving the OLS estimate in some countries, Y, in different directions–the treatment effect estimate will suggest the country Y treatment effect should be bigger but the OLS bias estimate will suggest country Y treatment effect should be smaller.
2. If one says the seemingly intuitive “move the treatment effect estimate for Y to the estimate for X (or to the average of the estimates for the N countries)” this (likely) implies (i) some countries revise their TE estimate upward from OLS and others downward and (ii) that the variance across countries in the true treatment effect goes to zero, even when the OLS estimates have large variance. Neither of these make any sense.
3. The Root Mean Square Error (RMSE) of the “collapse onto the rigorous estimates” prediction is, in now three empirical examples, larger than just using OLS country by country because the “internal validity” problem solved by RCTs is just so much smaller than the “external validity” problem created with the heterogeneity across countries in the “true” treatment effects.

(That was a new introduction to the following blog)

The attached paper (submitted to a special issue of Review of Development Economics) is a case in point. You would think that, after all the intellectual and financial resources that have gone into RCTs and into the creation of “systematic reviews” that aggregate the “rigorous evidence” there would be a sensible and empirically validated answer to the question: “How should be beliefs about the impact of actions X on outcome Y in my country context, call it C, (LATE(X,Y,C) change in response to a rigorous study (or systematic review of rigorous studies) from other countries/contexts?” But there just isn’t.

And it is easy to point out that things you might think sensible could be said, like: “Beliefs about LATE(X,Y,C) should move from existing the existing non-rigorous estimates in context C towards the findings from rigorous studies” don’t pass muster as being even logically coherent. As Justin Sandefur and I pointed out some years ago since the true LATE(X,Y,C) can be decomposed into the non-rigorous estimate in C and the bias in that estimate, if there is heterogeneity (variance) in the non-rigorous estimates across contexts (and there is) then this generically implies that in response to a systematic review some countries should shift their beliefs towards a larger LATE and some countries towards a lower LATE, which implies that the bias in those cases has a different sign. So any generic advice about adopting the “rigorous evidence” essentially demands people adopt beliefs about bias that are a wildly implausible set of measure zero. That is bad science.

Slightly harder is to point out that adopting as the prediction of LATE(X,Y,C) the systematic review point estimate does empirically worse that just using OLS(X,Y,C) as the prediction of LATE(X,Y,C). This point is stunning. If the heterogeneity in the true LATE across contexts is large relative to the bias in non-rigorous estimation methods then it is the case the most naïve possible thing is actually better than the supposedly new, better, cooler, more “sciency” approach of doing some RCTs in some countries and then aggregating those in a systematic review.

That this is so is harder to point out for two reasons. One, because there are so few RCTs that are even moderately comparable it is hard to have enough estimates of the “true” context specific LATE to create a variance in predictions. Two, because systematic reviews (and the underlying papers) tend to ignore the existing non-rigorous estimates altogether so the question of “how much better?” cannot be answered.

The attached paper solves these problems by using two sources that have comparable estimates of a “raw”, an “OLS” and a “LATE” for the same quantity for a larg(ish) number of countries. The LATE is the Oster estimate using the standard assumptions.

For 42 developing countries there is a Raw, OLS, and LATE(Oster) estimate of the wage gain for a typical low-education level worker moving from their home country to the USA from Clemens, Montenegro and Pritchett (2019).

For 29 developing countries there is a Raw, OLS, and LATE(Oster) estimate of the private sector learning premium from Patel and Sandefur (2020).

Once one has data like this, the rest is easy, just simple arithmetic (hence the temptation of paper arbitrage): compare the Root Mean Square Error (RMSE) of (i) using the average of the LATE estimates (the “systematic review”) to predict the LATE in each country or (ii) use the OLS from each country to estimate its LATE.

As expected, the answer depends on the ratio of the variance of the LATE to the typical bias in OLS. For wage gains the variance is huge and the bias modest so using context specific OLS is twice as good as using the “rigorous evidence.” For the private sector learning premium the variance is modest and the bias substantial (selectivity into private schools is large) and hence OLS and the “rigorous evidence” do about the same in RMSE.

So, for about 25 years now there has been a major advocacy movement selling people on the notion that doing RCTs about specific interventions in specific contexts and then aggregating these was going to lead to “evidence based” policies based on “rigorous evidence” and that would lead to better development outcomes. But there has never been any evidence these claims were true. Moreover, they always seemed pretty improbable and inconsistent with our “best available” models of development phenomena, that suggested that contextual variance in policy outcomes was likely to be very large.

These claims about external validity are not a picayune detail as without some clear idea about the scope of reliability of the results across contexts it is impossible to claim any piece of research, and especially expensive research, like RCTs, are a cost-effective investment.

I was invited to give a talk at George Mason University’s Mercatus Center and the series is about economics but also politics and economics. So I decided to give a seminar on a new paper I have just started working on which is the argument that development economics has been suffering from a “Gresham’s Law” (bad money drives out good) in which bad economics (poverty analysis with low bar poverty lines) has driven out good economics (the use of (inequality averse) social welfare functions. The paper is going to articulate why poverty with low bar poverty lines when used as a development objective or to evaluate policies/programs inevitably makes key mistakes and, moreover, doesn’t satisfy a basic “golden rule” of treating others like you would like to be treated. I argue that low bar poverty lines are both bad economics and also just plain bad–morally bad.

Here is an (ex-post edited) version of the slides that articulates the four key analytical mistakes an analysis using low bar poverty lines makes (all of which are avoided in standard economic welfare measures).