Economic policy in poor countries

And thanks for your reply!  I hope that you are now satisfied, since the issue is being discussed :)  More seriously I really am glad you brought it to the fore again, because it deserves it, and I'm being [sincerely] critical only because I take it seriously and respect it.

Re the FP report: so did it find anything promising or not? My reading (could be wrong, obviously, so let me know) was that they/you didn't in fact find anything to point to; that they/you believe such a thing does exist; but that it would take a lot of time and effort to find it. This is not especially encouraging, given that between them and the experts and you and Lant, at least one person-year has been spent on this already. If all you want is more discussion / research, as I said upfront I agree 100%. If you want to convince me that it is likely to succeed, you need to point to something more than intuition (in part because even the numbers you have thrown around are somewhat suspect).

There still seems to be an apples-to-oranges comparison here. You can criticize specific RCTs, sometimes validly and sometimes not (I'm happy to share the 10+ page report I wrote for GiveWell which on the whole validated the deworming results, and David Roodman's extensive research came to a similar conclusion). But it's not helpful to compare a single intervention (note that GiveWell has multiple charities with similarly high estimated ROIs) to "whatever it is that caused [those countries] to grow" which neither you nor the best macroeconomists seem to uncontroversially agree on even in retrospect. My claim is that if deworming had been added to their policy mix (and other aspects had adjusted however they adjusted), that would have been a good thing. 

Sure you can run an RCT on allowing foreign investment: subsidize it in some regions and not others. Same for immigration: encourage it in some places (or times) and not others. Same for road pricing. These won't pick up anything that is truly nationally systemic (and for that reason I agree full-scale tax reform is more difficult, although many taxes can be varied locally), but it doesn't have to be perfect for us to learn a lot. It also doesn't have to directly involve a "policy that you might  enforce in the real world" - it just has to have useful implications for such policies. Some of these might be hard and take time and involve theory; so is figuring out what causes growth and how to make it happen.

The ICRIER case that you and Lant and FP point to is simply wrong. You suggested in the comments below that >30% of the economics profession is working in the RCT space; this is wrong for development econ (where it is less than 10%) and even further from the truth in other sub-fields. You say in your response to me that "EAs do not support a randomista approach to the rich world" - obviously I can't speak for EAs in general, but I have worked on RCTs in the US (here and here and forthcoming here), RCTs in Europe (here and here), and for what it's worth non-RCTs in the US (here and here and here) as well. Eva Vivalt, another EA-affiliated economist, is also working on randomized cash transfers in the US. David Reinstein is working on RCTs of charitable giving. 

Purely anecdotally, I have found 'randomistas' to be highly practical, perhaps in part because they work more often in the field. I think most of them are happy to apply the technique anywhere it can help. I also think most of them admit that it isn't always the best approach, whereas I rarely see Lant or Angus Deaton saying the same about RCTs (perhaps because they incorrectly believe the method can only be applied to second- or third-order domains? then again they are very smart people so I have some trouble accepting that explanation). 

And finally back to the main question of this particular post. I'm personally not convinced that US policy work will necessarily get close to the GiveWell bar. Whether it does or not, I do think it's nontrivially possible that developing-world policy wouldn't beat it by an order of magnitude or more. As has been noted by others here I think, policy work often requires local contextual knowledge and connections, which for many poor countries would first have to be built up by the EA community. Most such work will be country specific, and most such countries (with some important exceptions, like Nigeria and India) have an order of magnitude lower population than the US. Most such countries have less stable governments, so any policy changes won't last as long in expectation. Lots of factors come into play; it's not so simple as saying 100x and that you personally don't see how it couldn't be true. Don't get me wrong: it might be true! In fact I do suspect that policy work, like other interventions, will in general have a higher return in the global south. But it's not a priori obvious.

Thanks for the reply. I'll have to answer it... it was supposed to be short, I really didn't have the time, but then I started enjoying it. But I have a TL;DR.
TL;DR: I guess we’re not understanding each other very well, as you seem to be responding to other people (unfortunately, because I’m a big fan). I don’t see why you categorize me as a skeptic. I think we actually agree (i) RCT shouldn’t be the main path for dev-eco researchers, and (ii) there should be more research focused on developing countries. But: (iii) development economics is more complex than you make it look like, (iv) I keep the discussion on dev-eco research apart from the discussion on which cause areas to choose. Also, please, take a charitable look at my answer to your point (5)... but I don’t see how you can talk about Rwanda’s growth after 94 without talking about genocide and war.

• It is the second most upvoted post in EA Forum history, which suggests that lots of people in the community found the post persuasive

Congrats 😉 on the other hand, like many people who commented on the subject, I upvoted your post back then, but I don’t find your case against randomistas so persuasive; I don’t upvote only posts I agree with – but if this is sending the wrong signals, perhaps I should start down-voting more.

• he has a PhD in economics from MIT, has been a professor of development at Harvard and Oxford, and worked at the World Bank for 15 years

Actually, I think your constant emphasis on Pritchett (as if the argument for dev-eco depended on him) is one of the weakest points of the post.

• Some of the most prominent economists in the world (Rodrik, Acemoglu, Deaton, Cowen, Caplan) have endorsed similar conclusions.

Also, five economists who disagree a lot about how development can be achieved. For instance, Acemoglu emphasizes the role of culture and institutions (but I think Melissa Dell research makes an even better case for this), but then Hausmann, Pritchett and Rodrik (2004) say, in the same article, that “Political-regime changes” and “economic reform” are “significant predictors of growth acceleration,” even though “the vast majority of growth accelerations are unrelated to standard determinants such as political change and economic reform, and most instances of economic reform do not produce growth accelerations.”

Also, Rodrik (my favorite one among them) claimed globalization (with free capital flows), national sovereignty and democracy are incompatible, defended an industrial policy approach to growth for a while, but admitted, in an interview with Tyler Cowen in 2015 (I recommend it), that the window for policies based on East Asia growth is likely gone; also in the same interview, he predicted Brazil would return to growing faster than India again – which didn’t happen (I mean, it’s sad for me, a Brazilian, to here it today).

In spite of this, no-one in the EA community has ever published a response to these arguments

I’m not sure what’s the problem here and what sort of response you’re expecting. But as mentioned before, you also received a lot of comments, many of them discussing some of your main points; and you cite Founder’s Pledge analysis of the area, which concludes (very persuasively, in my opinion) that “We’re not continuing this project because we think identifying funding opportunities would be too costly for us and for the organisations under consideration.”

Perhaps you’re victims of your own success, since searching for “randomista” in the EA Forum mostly leads to references to your post – but it shows people are actually taking it into account; I think it might provide some sort of middle ground between short-term and longterm causes). But, yeah, the only thing that seems as deep as your original post is K. Sarek’s long summary of Ogden’s “RCTs in Development Economics.”

However, I must add that I think EA is more often identified with randomista economics by outsiders – precisely because RCTs in economics provide a useful rhetorical advice to distinguish EA from other approaches to philanthropy.

Turning to the object-level, I take it that you are making two arguments here - (1) scepticism about growth economics, (2) neglectedness. I have several responses.

I’m sorry, but calling my position scepticism is getting too close to a strawman – an unfortunately common move among us philosophers, as if it’s impossible to engage with someone without putting a tag on them. For instance, you don’t call someone who says “people disagree about cancer” a skeptic. I explicitly said economic development is very important, and I don’t deny we know a lot about it, nor that people should research it (quite the opposite); I’m just claiming that arriving to specific recommendations on growth, at the required scale, is way harder than you make it look like (particularly if we restrict the discussion to a small sample of economists, in a limited period of time). If I'm skeptical about anything, it is that we will figure out something like a feasible way to, e.g., make Haiti grow like Botswana did, or even to ensure the latter keeps its current trajectory for decades.

Also, I suspect there’s a widespread Dunning-Kruger effect in discussions about economics, where experts usually express themselves with way less confidence than the average amateur. So, I’m not saying we’re in the dark here – actually, we are more likely being obfuscated by too much noisy light. Thus, sorry, but I think your “replies to skepticism” are answering the wrong man (unfortunately, I’d like it to be me, as I’m a big fan of your work), and I’m afraid we’re not quite understanding each other.

On the other hand, I feel compelled to answer in full, as you do provide me with some examples I can target:

1. It is not possible to sustain the position that we know how to make policy-led improvements to economic welfare in the US, but not in Kenya

Actually, I don’t see why this is impossible.

1. We may know how to improve A’s health (because she’s kinda healthy, or has a well-known condition), but not B’s – because he has a very peculiar condition, or because you never studied a non-WASP person like B, or because he has diabetes + allergies + cancer + covid-19 etc. and anything we do has a significant risk of causing harm (because of, e.g., noise). Nothing here implies that we only know “one of two things about how to boost” health – actually, precisely because we know a lot about health we can recognize we’re uncertain.
2. Actually, development economists often consider it quite obvious that the advice you would give to poor countries (such as “improve capital formation”) is quite different from what you would focus on with rich countries (“get back to investing on innovation, lest others will leave you behind”).
3. Again, if most of our research is about developed countries and mostly focus on their trajectories (“see what developed /OECD countries have done more”), we’re very likely neglecting risks of selection bias and path dependencies. I can’t stress this enough: instead of thinking about Kenya in terms of what I know from US, I’d start studying why Cape Verde and Mauritius (hmmm… bad example: small insular nations), or Botswana and Gabon are better (in terms of HDI) than Kenya.

However, I must remark: I think we actually agree economic researchers should focus more on developing countries, right?

2) For example, Indian farmers are forced to sell their products to state middlemen at a guaranteed minimum price. According to Wikipedia, foreign supermarkets are banned in many Indian states. The time taken to start a business ranges from 1 day in New Zealand to 230 in Venezuela.

I agree these examples show problems that could be fixed with simple recommendations from economics. Do you think there’s any initiative which is likely to lead to these intended outcomes? What do you think its cost-effectiveness could be? I'd gladly know more about it.

But notice they're very different issues, and they're more like symptoms than causes of the underlying conditions affecting these nations. I don’t think the problem here is anything like a lack of economic literacy, or advocacy for the right economic policies – it’s more things like inadequate equilibria.

3) The experience of the world since 1950 makes it extremely unlikely that we don't know anything about growth. Economists and states were prominently trying to increase development from 1950, and development did in fact increase. Indeed, there was more progress on all subjective and objective measures of wellbeing than all prior human history combined. It might be true that this was an enormous coincidence, but at the very least someone actually needs to sit down and explain why it was.

I think this remark is surprising, because it totally ignores my point on the Nurkse v. Hirschman debate, which is my very example of how these discussions are still unsettled. And even now, we still don’t really know, e.g, what kickstarted the Industrial Revolution in England, but we do know economists back then were wrong about many things – even though economies have increased exponentially since then.

4) I don't think there's any point in answering (4), as it seems premised on the accusation of broad skepticism. I agree that finding out, e.g., how to make everyone grow like Botswana would be huge. Except that perhaps you're framing the problem in a way that underestimates the costs of obtaining new information here. Again, development economics has been a hot topic for a long time.

5) Suppose you were a philanthropist with the opportunity to influence policy in Rwanda in 1994. Would you focus on encouraging them to do some RCTs of social programmes, or would you try as hard as possible to get them to do whatever it is that they did that led to sustained economic growth in Rwanda of 9% since 1995?

I’d first try to avoid another Rwandan genocide and the following Congo Wars, which are among the worst things that have happened in the world in the last 30 years. Seriously.

My point is not the rhetorical effect of “how could you forget that?”, but that any explanation of Rwanda’s growth after 1994 that fails to take it into account is likely flawed. WorldBank’s overview of the country refers it explicitly, for instance.

6) I will answer this separately. It’ll take longer, and, unfortunately, I’m not paid to do this – quite the opposite, perhaps.

7) If our knowledge of growth is incomplete, I don't think it helps that a substantial chunk of the economics profession (>30%?) have started to work on RCT-evaluable things, which do not attempt to make a non-trivial dent on GDP per capita.

 Well, this is the debate I actually want to have – and I guess you do, too. Dufflo and Banerjee response to this would be (in my interpretation from Good Economics for Hard Times: Better Answers to Our Biggest Problems) that we know more about how to make progress in this area (RCT), and that researchers can learn useful stuff with that (which might provide positive feedback loops); besides, we already had, and will keep having, public policies that will need to be oriented by research.

They have a point, but I agree with you (as I said before) this is likely not the best approach to development – as it is more about redistributing resources than creating them (i.e., growth), and it still leaves a huge blank in our theory of dev-eco. But this is about the roads economic researchers are trailing, not about EA recommendations.

8) This scared me a bit, too. But you have to read into the whole context. They're emphasizing the need to change consumption patterns (particularly in developed countries), because energy-efficiency programmes failed to deliver results, and so recognizing there's a trade-off here (and pending towards lowering emissions).  They are not against growth. But I got scared: what if Cowen's "Crusonia plants" are getting harder to find?

Neglectedness

I’d like to see someone investigate it more thoroughly. I agree US policy is not neglected. But I guess neglectedness is more about opportunity costs: how much do you expect the counterfactual effect of more input on this area to be? What will it add? Will it crowd-in or crowd-out additional investment?

(You know, this is analogous to a very controversial point on the role of the State in dev-eco since at least Nurkse: will a state intervention here crowd-out or crowd-in private investments?)

Finally, summarising my point: some of the best minds in economics are already taking dev-eco seriously, funded by governments and wealthy organizations, and things are still quite uncertain when it comes to feasible particular recommendations – as my future answer concerning my own country (I’ll see what I can say, despite a small but positive risk of suffering future persecution - sorry, but paranoia is starting to spread around) might exemplify. But I agree with you that sending 30% of economic researchers to do RCTs on things like transfers is likely a wrong move.

Off topic, but I didn't realise you'd left Founders Pledge. May I ask what you're up to now?

+1