There is little (good) evidence that aid systematically harms political institutions
Sometimes in discussions of foreign aid or charity I see people raise the point that aid might do good directly, such as by providing a health service, but that it might cause harm indirectly, for example by allowing an incompetent or corrupt state to continue existing without being forced to become better by harsh economic realities. These arguments come up in conversation, and also in books by the likes of Angus Deaton or Larry Temkin. Recently, Martha Nussbaum brought up these concerns. They’re worth taking seriously. Thankfully, political scientists and economists have in fact looked at them (somewhat) seriously.
In this post I’m going to tell you a tiny bit about me, explain the institutional criticism of aid in a bit more detail, and then explain the results of two papers testing this relationship. I’ll also explain a tiny bit about the limitations of this kind of work. The upshot of all this is that we have very little strong evidence that aid systematically harms political institutions. My best personal guess is that while aid can sometimes have medium-sized positive or negative effects on politics in recipient countries in specific cases, on average right now and in the recent past it has very small effects in either direction.
All About Me
I’m a political scientist by training and most of my teaching is in a development studies department. When I was considering grad school, I was really interested in political accountability and the ways that “easy money” like oil can distort political relationships. My reading of recent history suggested to me that getting oil before having representative institutions meant locking in autocratic rule.[1] I thought about studying oil states but then somewhat unrelatedly I tried living in Cairo for about half a year and found it quite challenging culturally—so I figured studying oil states was probably not going to work for me. My next idea was thinking about aid. In a lot of ways it seemed similar to oil: it was “easy money” for governments that let them provide goods or services to their citizens without taxing them. In many very poor countries it was also a large flow in terms of government expenditure or GDP. I wrote my big undergrad paper on this. While doing so I read the work of Deborah Brautigam on this topic and then I went to do my PhD with her.
Aid and institutions in theory
The “aid harms institutions” story isn’t dumb. It makes internal sense, and the early (cross-sectional) evidence that we had on it sometimes suggested harm. There are so many ways that aid could harm institutions or governance in recipient countries. It could do so by acting like oil. This primarily means allowing governments to exist absent taxation. If citizens aren’t taxed, so the theory goes, then maybe they won’t demand representation or results from government in the same way. And if governments don’t have to collect tax, then maybe they won’t do state building things like building up good ways of gathering information on everybody. If we look at European states, you can easily tell a story where states felt the acute need for more money (often for fighting wars) and this set off a chain of events that built strong states and created demands for state accountability to at least some segment of the population.
Aid could also harm institutions in more mundane ways. For example, donors might want to hire local staff and might pay well by local standards. This seems like a clear positive, but if enough donors do this they might poach all of the best people from government bureaucracies. Donors might also want lots of reporting to make sure money is well spent. Again this seems good, but with so many donors all wanting this kind of reporting it can be a large drain on the time and attention of a limited number of already overworked bureaucrats in aid-receiving countries.
However, aid also isn’t oil. Donors care about how recipients act in a way that buyers of oil do not. Aid is often explicitly or implicitly tied to recipients doing or not doing certain things. It’s easy to tell a story where donors do things that help improve institutions or governance, like donors withholding aid to Kenya in the early 1990s and pushing the country towards democratization.
Testing aid’s effect on institutions
Early work testing the effects of aid on institutions tended to find a negative relationship. I read some of this work while doing my undergrad paper. Looking back, however, a lot of that work is not up to answering our question. One common problem is usually that it was cross-sectional: it is statistical work that has data on many counties at one point in time. You can think of this like a scatter plot, where aid is on the x-axis and some measure of good governance is on the y-axis. When you run these regressions you will often see a correlation where places with more aid have worse governance. However, it pretty unclear if we should interpret this as saying that aid causes problems in governance or aid goes to places with worse governance. Fire engines suspiciously appear near houses with fires.
The authors of this work understood that this was a problem and tried to deal with it in various ways (control variables, IV methods), but for reasons I won’t get into here these methods are pretty unreliable in this context and have rightly fallen out of favour. It’s just very hard to nail down causation in this context.
Some more recent papers come at this question using panel data, which means that they have data on many countries over many years. This lets one follow the trajectory of a country over time, as aid to it rises and falls. This is still far from something like a randomized control trial, but it’s usually better than using a cross-section. Two papers that do this are “Aid Is Not Oil: Donor Utility, Heterogeneous Aid, and the Aid-Democratization Relationship” by Sarah Blodgett Bermeo and “Does foreign aid harm political institutions?” (pdf) by Sam Jones and Finn Tarp.
Sarah’s paper examines how aid influences democratization in non-democratic states. I like it because she also examines some past research and shows how the authors arrived at results that conflict with hers.[2] The basic takeaway is that she finds that during the Cold War aid did make autocracies less likely to democratize, but that this effect is mostly gone in the post-Cold War period. The only exception is that if a country is strategically really important then maybe it can still use aid in this way. I’m saying maybe because it really hinges on whether or not you think her interaction terms have sufficient power to detect these effects (I’m skeptical). Nevertheless, the basic takeaway is that in the post-Cold War period aid doesn’t seem to do anything to institutions.
Sam and Finn’s paper looks at the effect of aid on measures of governance. I won’t discuss all of their empirical approaches but they do some cross-sectional and some panel analyses examining the effect of aid on measures of democracy, the number of veto players over political decisions, the level of executive constraints, a political terror scale, and a measure of judicial independence. Their data runs from 1983-2010. They tend to find small positive effects of aid on political institutions.
Should you believe either of these results?
A litte?
The authors did their best and it’s an improvement on the work that came before it, but because we have nothing like a clean experiment there is a lot of room for skepticism.[3] However, any skepticism one has of these sort of results should also be applied to the claims of Deaton, which rely on no systematic analysis at all. I’ve been pretty disheartened to see people like Larry Temkin or Martha Nussbaum defer to Deaton on this when his analysis rests on the kind of loose hunches that I had as an undergrad. The hunches are reasonable, but we also should temper our hunches with at least a passing glance at empirical evidence.
I think the main takeaway here is that while the evidence that we have isn’t very good (because the question is so hard to study), what evidence we do have suggests that aid in general likely does not systematically help or hinder governance in recipient countries (or that the effects are too small for us to reliably pick them up, in which case they probably don’t matter much).
- ^
Yes, I know this isn’t some novel insight.
- ^
In part, the answer is ridiculous p-hacking.
- ^
I’ve obviously only covered a small amount of the research on this question, but this post is informed by my reading of most of it. If you want a nice recent paper on this question that also provides a good entry point to the recent literature, see “Foreign aid, oil revenues, and political accountability: Evidence from six experiments in Ghana and Uganda.”
Another case study I’d throw out is Somaliland.[1] Somaliland is the top bit of the horn of Africa, usually labeled as part of the better-known Somalia.
It’s one of those fascinating corners of the world which is in the grey area between country and not.[2] Its government exercises sovereign authority within its borders, it fields an army and enforces its borders, it issues and backs its own currency; compared to Somalia, Somaliland is far more stable and democratic. However, it is only recognized as a sovereign state by Taiwan (not a UN state itself), and has never been officially recognized by the UN or any international organization.
Due to this isolation,
So there’s a country which received practically no ODA (US$100,000 total between 1991 and 2016) and flourished, next to Somalia, which received more than US$13,000,000,000[3] and practically regressed. The border between Somalia and Somaliland[4] is the development practitioner’s version of the Korean peninsula nightlight picture.
These are, of course, just two countries, being assessed in a period featuring genocide, civil war, ethnic conflict, and extreme drought and famine. In other words, not very generalizable, especially compared to the panel data in the papers Ryan cites. But it is further along the spectrum towards random assignment.
In fact, I would call the Somali/land case quasi-random. The reason Somalia receives aid and Somaliland doesn’t is largely a matter of historical contingency.[5] Now, if only I had another thirty such cases, we’d be ready to run some numbers.
But — it’s just a case study. So just how successful has Somaliland been? Most of its success (relative to Somalia) is in civil order. Somaliland has not experienced large-scale violence since late 1996, unlike Somalia, where it’s easier to measure the months which don’t see civil war than the months that do. Somaliland looks better on economic indicators, but not wildly so. Because international organizations don’t operate (much) in Somaliland, we don’t have good health indicators, but what do do have, too, look marginally better than their southern neighbor.
I wouldn’t say Somaliland has been successful because of a lack of aid, and Phillips doesn’t go quite that far either.[6] One connection I see between aid and later prospering is agricultural. Somaliland is pastoral; Somalia is agrarian. Further, there had been less investment in Somaliland’s agricultural and domestic infrastructure, due to the dictator’s ethnic preferences.
In 1991, Siyad Barre, Somalia’s dictator for 22 years, was ousted, sparking the Somali Civil War which continues to today. Somalia was already the largest per capita aid recipient in Africa, and this had greatly weakened agricultural self-sufficiency, especially in the south. This key point isn’t really controversial: it’s consensus among academics and donors that the Somali famine was largely avoidable through better aid policy.
But other than that, what can we conclude from the lack of aid in Somaliland?
I see it as the opposite side of the Afghanistan coin which another commenter brought up. Much like Afghanistan, aid went wrong in Somalia. Really, egregiously, please-just-leave-them-alone wrong. This isn’t to say aid-done-well was impossible in either case. But maybe these were both situations where getting-it-right was so a priori difficult that it wasn’t worth the try. In Somaliland, we have the control group that we didn’t have in Afghanistan. Not through any purposeful decision, in fact through callous neglect, the international community did Somaliland a favor by just… staying away.[7]
There’s a ton of nuance which I’m missing here — what does it mean to “stay away”, how can we tell when a local or national situation is too complex, how to disaggregate “aid” to the useful and dangerous. If anything, Somaliland is a plea for modesty. As Ryan writes, aid “has very small effects” on average. But that average hides some disasters (and presumably, some miracles).
Final note: in the last few years, Somaliland’s economy has opened up to foreign investment, mostly due to its very strong geographic position and lack of pirates. The UAE is building US$250 million highway from the Berbera port in Somaliland into Ethiopia, and an Emirati company was granted a half-billion USD concession to develop the port. These have not yet come with political recognition from any of the Gulf States, but they’re much-needed external dollars. Somaliland is a fascinating and beautiful country, and I hope to visit in the next few years. Even if its history of isolation had helped it remain stable, there’s no good reason to keep its population of five million separate from the world.
This comment is based largely on Sarah Phillips’ book “When there was no aid: war and peace in Somaliland” and conversations with two friends, one Somalilander and the other a polisci professor. In fact, this comment isn’t really that related to this post at all, I just finished this book and talked to these friends and thought it’d be a fun thing to write.
The two books on this grey area I can recommend are Josh Keating’s “Invisible Countries” (highly recommend) and “An Atlas of Countries that Don’t Exist” (meh). For a fun short Wikipedia read with colorful maps, see here.
This of course includes support for military and peace-building processes during and following the civil war, tho not direct military assistance. ODA isn’t always exactly what we care about. Somaliland number is from Phillips; Somalia number is from the World Bank.
Tho, it is a messy border. There’s another quasi-state in between the two, Puntland, and the borders are constantly shifting.
This footnote was originally longer than the entire comment, so I’ll cut it short. Somaliland was independent and internationally recognized for five days in 1960, just after their independence from Britain. Then they voluntarily united with their neighbor, forming Somalia. Almost immediately, the national government was dominated by the southern clans, and the once and future Somaliland spent the next thirty years fighting to regain their independence. They succeeded in 1991. What I meant by “contingent” is that a) the initial union was a priori unlikely, b) the union was unusually structured to make the south the stronger partner, and c) the 1991 collapse of Somalia put Somaliland in an unusually weak position to negotiate with the international community.
Phillips’ explanation is super interesting but not really in the scope here. She proposes that Somaliland has developed a unique “independence discourse” which discourages violence and encourages inter-ethnic cooperation. In this discourse, Somaliland is exceptional precisely because of its autonomy from international structures, and finding post-conflict and post-genocide peace without external support justifies their country’s existence. The subtle and counterintuitive bit of Phillips’ narrative is that Somaliland’s weak institutions actually strengthen this peace. Violating the peace is “a choice which, if taken, is unlikely to be contained by the country’s governance institutions and therefore must not be chosen.” I love this argument. It’s very “exactly because arf not arf!”.
This is the most pessimistic paragraph I’ve ever written about aid, and part of me wants to walk it back and say, “ok, but maybe we could’ve limited aid to the un-misappropriable, like maternal mortality interventions”, but no. I think I stand by it.
note that a large portion of Somaliland appears occupied by rebels at the moment. But other than that it has indeed been much more peaceful.
https://en.m.wikipedia.org/wiki/2023_Las_Anod_conflict#:~:text=The%202023%20Las%20Anod%20conflict,violent%20crackdown%20on%20civil%20protests.
https://en.m.wikipedia.org/wiki/Somali_Civil_War
Yeah, the conflict in Laascaanood is a bit of a damper. But the rebels control less maybe 15% of the country’s land area, and ~5% of its population.[1] Further, Somaliland has never really asserted its sovereignty over the city,[2] and it’s not particularly important.[3] It wasn’t clear in Phillips why Somaliland attempted to include the Sool region in their secession from Somalia, as it voted against the constitution in a referendum. This current flare-up is a continuation of the (longer border conflict with Puntland)[https://en.wikipedia.org/wiki/Puntland–Somaliland_dispute].
I’m generally confused by this conflict. My main thought, different from what I wrote above, is that it’s an indicator that the Isaaq majority is more willing to assert stronger political authority, weakening the clan-based power sharing structure.
These are really rough guesses. Would be happy to see good sources.
And neither did Puntland: “In many respects, Laasaanood seems to be part of the Puntland state of Somalia. [...] In Garoowe it becomes clear that Laascaanood is perceived as the political periphery and people there are not fully trusted by officials in the capital of Puntland.” (Hoehne, 104)
I think it would be the seventh or eighth largest city in Somaliland, in a largely un-urbanized country.
Totally agree that it’s a fascinating case. Thanks for this!
This is very interesting!
I would love for you to take a look at the specific case of Afghanistan. It is political and there are much more bigger factors at play than aid (the effects of war, American politics, Soviet politics, regional politics). However, an argument that makes sense to me is that the Taliban had significant popular support because the institutions were so inefficient most people did not like the previous government, and these were institutions sustained by aid (their budget was 75% aid or so according to World Bank).
Sorry for going slightly off-topic, by I wanted to share this which is only related to the initial part of your post:
I am one of those people sustaining aid can be worse than no aid, but not just from the perspective of political institutions. I have a couple of assumptions based on my experience, that put together make me thing most aid has a negative outcome.
Most programs financed through aid and implemented by states and NGOs have a very modest effect, no effect, or a negative effect (usually also modest).
Most of these programs are implemented through institutions that implement systematic corruption. Non-profits and government units that are de facto for profit teams but can’t say it, so they have to collect this profit through corrupt processes.
Executing corrupt processes takes time, skill, and effort. You need to know how to fabricate financial reports and documentation, how to falsify beneficiary data, how to create the right paper trail, how to create an extortion network. To do it very well, you have to involve several people carefully, often your whole team, discuss it abundantly, and plan everything. Sometimes your donor or supervisor finds out or kind of finds out and you have to fabricate more explanations or defend yourself. It takes time. So for most of the professionals in these teams, corruption is a significant time sink, maybe 50% of their daily work gets dedicated to activities related directly or partially to corruption, maybe 25%, but for some people it could be more than 50%.
In many developing countries these institutions are very relevant economic actors and they attract some of the best minds in the country. In most developing countries there are not enough jobs in the private sector, and everyone practices entrepreneurship but wants to step out of it. Charities and government are a major employer when put together. In many cases, they pay top salaries. The worse the situation in the country, the more humanitarian and development funding, and the more significant these institutions are.
So we have an economic activity that does not have any significant positive effect (not when all budgets are aggregated) and where a substantial portion of the time is spent un a useless or harmful skill (it is not good for anyone that people learn to do this, maybe a small percentage can eventually be re-oriented to fraud prevention, but we don’t need this number of people learning all this) and the best people in the workforce are doing these “useless” tasks every day.
If there was less aid funding, there would be less of such corrupt teams, and people would be forced to start businesses, work for the private sector, or to emigrate and send back money, all seem better options for the good of the country.
Claims 1 and 2 are not hunches, I have worked in these type of programs at all levels and I have seen enough institutions to say that if this is not true then I might have had an extremely unusual sample in front of me. I have worked for close to 10 years in Cameroon, Nigeria, and Afghanistan and every day I had to work with, against, and around these type of teams.
Quick thought on the tangent, which I’d also love to hear more thoughts on from other people.
I’m skeptical that corruption is a big obstacle to growth and development. Measurement and historical comparisons are tricky here, but corruption seems to be a pervasive feature across many societies.
Even the United States had its local political machines and share of bribery before the Progressive Movement in the 1920s tried to filter it out. And conventional wisdom credits the Industrial Revolution (of the 19th century before the US reduced its corruption) with our modern wealth.
I suspect if we applied our same concern of corruption to currently-developed countries to their past, we’d find they (1) would fare just as bad and (2) had their development periods before they dealt with the corruption
The political scientist Yuen Yuen Ang has some great work addressing this neighborhood of intuitions. Her view is basically that “corruption” is decomposable into a several distinct types of phenomena, and some of these can be growth-promoting (as in China during the period between, approximately, Deng and Hu), whereas others can be fairly extreme impediments to growth.
Thanks so much for this! I don’t know why I ever thought about decomposing the idea of corruption but it seems like a really obvious framework now that you’ve mentioned it. Hoping to give that a read sometime.
Happy to recommend her work highly.
Thanks Geoffrey it’s and interesting discussion. I have mixed thoughts about this. There’s a great section in the book ” bad Samaritan’s” by the awesome economist Ha Joon Chang which makes this argument very well.
Thanks for the comments!
Off the top: I’m not an expert on Afghanistan and it wouldn’t be overly surprising to me if we could find specific times in specific countries when aid did affect politics. Maybe post-invasion Afghanistan is one. All that said, my personal bet would be that aid just isn’t doing much in Afghanistan.
Now if the question is “does aid work well in Afghanistan?” then I’d guess the answer is “no.” I fully believe that politics can interact with aid to make aid more or less effective, especially in the sense that aid to very badly governed places might do very little. However, that isn’t the question. The question here is “would Afghanistan be better off without aid?” and while I’m open to the answer being “yes” I imagine that most of the problems are larger and more serious, and that aid offers only a very minor push in any direction. And of course this goes quadruple for Cameroon or Nigeria, where aid is a sideshow compared to the other money in the system.
I think it is very difficult to argue that aid “didn’t work well” in Afghanistan when you look at any education or health metric. According to UNICEF there was a 90% increase in child malnutrition in the year from June 2021-2022, capturing the period following the collapse of the state and the majority of aid projects (number is partially inflated by the expansion of UNICEF programming to cover gaps left by other actors).
I don’t think it is controversial that there was too much money and way too much corruption in Afghanistan. Obviously, the state-building project failed. But “aid” in that environment covered such an extreme range of activity. I’ll be honest, it’s a sensitive personal topic, having lived there and given everything that since happened, but it does bother me to see “aid” as a whole written off when it includes such diverse activities as addressing child malnutrition and funding unaccountable quasi-police forces.
One concern that asymmetrically affects this discussion is measurement error. Democracy or institutional quality or any such impact is measured with a lot of error, so estimates of the impact of X on democracy/institutional quality are going to be super noisy and it’s going to be harder to reject the null of no impact—even if there is a real negative impact. Is that something that’s received a lot of attention?
In general, it’s harder to put stock in “studies have not found evidence of X” than “studies have found evidence of X” since everything is measured with error, so rejecting nulls is harder than failing to reject them.
Good question. It’s worth recalling that Sam and Finn’s JDE actually finds small positive (significant) effects of aid on many governance outcomes. I’m not sure I actually believe that those positive effects exist, but it’s important to see that my claims above don’t hinge on over interpreting (imprecise) nulls. Also I realize you know this, but for others it’s good to remember that classical measurement error in the DV will increase noise but will not introduce bias.
The article in footnote 3 is also an example of other work I (in the interest of brevity) didn’t summarize but have read that lends support to my claims re: aid and governance by testing one mechanism thought to link aid and governance in proper experiments (this is another related one based on qualitative work: https://rdcu.be/b4aTu), so my claims don’t rest purely on country-year panels. If some people DM me and say that they want a longer summary covering this stuff, I might try to find the time to do it.
One of the problems that I observe in this conversation is just the meaning of the word “Aid”. It seems like in some cases this can refer to directly supporting a government’s budget, while in other cases it could refer to a foreign NGO directly administering a program. Should we expect such diverse interventions to have equivalent risk of corruption or institutional effects? To me they seem quite different.
Fair.
I used the term to mean ODA, as did all of the authors that I cited. Basically, this would capture grants and concessional (below market rate) loans that are aimed at promoting economic development or welfare (so not military aid). From a donor’s POV, it includes money that it gives to multilateral organizations like the WB that then pass the grant on as well as classic bilateral money. From the recipient’s POV, it includes money from bilateral donors and from multilaterals. It does not include non-concessional loans or private charity. Money can be ODA even if the government doesn’t administer the aid. This is sometimes called “bypass aid.” ODA would also include things like project aid (money for a specific project, e.g. to build a road) or programme aid aimed at sectors or general budget support.
I hope that helps!
Thanks for the quick and detailed response!
Mapping this language onto the EA landscape, it sounds like most GiveWell-recommended or GiveWell-style charities, programs, or grants would not be considered ODA or “Aid”, do you agree with that?
That is correct. I think it is pretty uncontentious to say that the people who study this in general worry less about these private charity-type interventions than they do about ODA.
This surprises me, since on-budget ODA works through government and government accountability frameworks, whereas off-budget and totally private spending does not. Is this perhaps just because private philanthropy in international development is just very small (and therefore less risky) in dollar terms compared to ODA?
In the domestic context, excessive private philanthropic funding can be seen as undermining democracy and weakening the state, in relation to education policy for example.
I’m thinking for example of the arguments made in “Winner Takes All: The Elite Charade of Changing the World” by Anand Giridharadas—a book which I really didn’t like and don’t personally endorse but makes some valid points, which I think are commonly held.
My claiming it’s uncontentious is based on working in this research area and talking to lots of researchers about it. When asked, most say they’re less worried about charity than ODA causing these sort of governance issues. Now I get that your question is “why?” and my answer here is more tentative, because I don’t know what is going on in their heads.
I do think “size of flow” is a big part of it. I’d guess “large flow” is a necessary but not sufficient condition for governance issues, and absent something like a big GiveDirectly UBI-type thing the size of charity flows is often not that large compared to e.g. recipient government budgets.
In terms of the theory, I honestly just think our theory is pretty weak. We’ve often expected flows to cause harm when it looks like they didn’t. I don’t want to say theory isn’t important here, but I think we should be at least as cautious about theory as we are about empirics (very). Maybe it’s worth pointing out that my title was that we don’t have good evidence for harm, which I strongly stand by, not that we have good evidence that these flows are benign (we don’t). This is just a very hard area to study.
Thank you, Ian, for asking the question that was in the back of my mind while I was reading this well-written and accessible post by ryancbriggs. I think it would be nice if the OP could add this caveat (that the evidence concerns a specific type of aid), since I assume some of the people reading this post in the EA forum will possibly update unjustly against aid recommended by, for instance, GiveWell.
Executive summary: The evidence that foreign aid harms political institutions in recipient countries is weak; recent studies find small or no effects.
Key points:
The “aid harms institutions” argument makes theoretical sense but early empirical support was flawed.
Recent studies using panel data tend to find small or no effects of aid on measures of democracy, governance, etc.
The evidence is imperfect but better than anecdotal claims; suggests aid likely does not systematically help or hurt institutions.
Effects are probably small in either direction or studies would pick them up more clearly.
Skepticism is warranted, but should also apply to claims of harm, which lack systematic analysis.
Bottom line: little strong evidence aid harms institutions on average, though context may matter.
This comment was auto-generated by the EA Forum Team. Feel free to point out issues with this summary by replying to the comment, and contact us if you have feedback.
Thanks for sharing this. I like seeing this kind of stuff on the EA Forum.
By “this kind of stuff” I roughly mean “correcting commonly held perceptions that are false, and providing not-too-much and not-to-little context.” The falsehood of these perceptions might be obvious to someone who has dug into the topic a bit, but there are a large number of topics that I never explored beyond the superficial level.