Should we abolish foreign aid?

Altruism is the cardinal virtue – but closely followed by scepticism. This essay is a study of these two forces in the economics of official aid.

Following the hysterics of 2005 – Bono, Gleneagles, the call for 'double aid' – suspicion of aid is again encroaching, as it has done in cycles throughout the modern aid period. 1 A number of academic works have questioned the value of aid; Gibson et al note that ‘hundreds of researchers have identified hundreds of problems’ with the aid process - an array of dismaying hypocrisies, frictions, and leakages. 2

Even sympathetic macro syntheses conclude that aid is not necessary for development, nor obviously sufficient.34 (This does not mean that it has been without net use – to ascertain that, we would have to investigate the $3.4 trillion of aid spent since 1960, at an exhausting level of granularity. 5)

Some economists go further, arguing that aid is not only wasted (that is, zero effect), but is actively perverse. Donald Kommers speaks for many when he says:

Aid workers... are good people committed to caring for the victims of poverty, disease, violence, and war. But their caring is selective, their advocacy misleading, their impact negligible, and their independence politically compromised. Worse, they are sometimes unknowing accomplices in the evils they seek to eradicate. 6

Call the position of ending ODA transfers to the developing world, aid abolitionism. It earned lip-service from some world leaders in the early C21st 789, after its original statement in the 1950s by Peter Bauer (who was himself following C19th arguments about the abolition of welfare in general by Malthus). 10

Of course abolitionism is contested in the strongest terms. Jeffrey Sachs is the exemplar of the new "Big Push" for development aid:

The inattention and neglect of our policy leaders lull us to believe casually that nothing more can be done … a small fraction of that [military] money, if it were directed at development approaches, could save millions of lives and set entire regions on a path of economic growth. 11

See also the Liberian president, Ellen Johnson-Sirleaf:

Critics say that African economies are shrinking, that poverty is rising – and that failing aid is the culprit. But this argument is at least a decade out of date... Reducing aid would slow private sector growth, stall poverty reduction & undermine peace and stability in countries that are struggling to become part of the global economy. 12

Figure 1 sketches the ‘Great Aid Debate’ between these opposed views, reflecting the curious alliance of neoliberals and neo-Marxists under abolitionism:

(This diagram is post-hoc and illustrative, and shouldn't be treated as evidence per se.)

Owen Barder divides development goals into two: Transformation goals and Solidarity goals. 24

Transformation is the classic Rostovian macroeconomic project: the acceleration of deep economic and institutional change in LDCs, until they achieve take-off growth, and thereby independence.

Solidarity is the impulse to ease underdevelopment now: e.g. food and commodity donations, services offered in warzones, palliative medicine.

On this view, the development process is long and hard, and one role for outsiders is to enable people to live better lives while this process is happening in their country. 25

It shades into emergency/humanitarian aid and takes us out of our present scope. (A better name for solidarity aid might be "marginal" aid: to me, 'solidarity' connotes a purely symbolic, materially ineffectual gesture of help.) Aid has aimed at both, but only succeeds demonstrably on the latter type.

A third, ignoble goal involves giving aid to boost donor industries, or to gain soft geopolitical power over recipients. Another is the signalling of one's wealth and virtue by giving aid. The only real problem with the latter two is that they tend to be very ineffective, since a metric other than 'impact' is being targeted.

• Inclusions: I am talking about ‘official aid’, government-to-government assistance. I include non-DAC aid where I can, though the data for it are even sparser than usual.
  
  I am aiming at detecting structural, 'transformative' effects of aid on the economic capacity of recipients.



• Exclusions: I do not treat emergency aid – widely considered to the only option for its particular goals, and so above abolition.13 Discussion of aid reform theory is limited to a sketch. China is such an exceptional LDC that I try to exclude it from aggregates. I also do not consider abolitionisms motivated by political ‘particularism’ (positions that weight some group intrinsically more than others): the abolitionists’ arguments are nominally free of geography, nationality, or similar illiberal criteria.

Apart from discussing Toby Ord's important argument about smallpox, I mostly don't cover the many, many incremental gains that didn't involve capital-d Development (where Development can be glossed as: full industrialisation, universal quality of life, graduation from aid, and democratic flourishing).

• Impact: A buzzword for the absolute effect of some project or program. Call the general impact (the summary of impact sought by cross-country regressions), the ‘macro impact of aid’.


• Effectiveness: Aid efficiency: impact per unit resource.


• Development: The most fraught definition. A heuristic is: “rapid improvement in capabilities and quality of life across a whole society”. It is about more than economic productivity, but the term is very often used to mean economic development, i.e. GDP growth, or total factor productivity growth. I use the 2010 HDI scale as a leaky proxy.


• Human Development Index (HDI): From 2011, a measure of basic development, incorporating equal weights of life expectancy (LE), literacy/education (EI), and standards of living (II) in the following simple function: $$ HDI = \sqrt[\leftroot{-1}\uproot{2}\scriptstyle 3]{ LE \times EI \times II } $$ 18


• Development aid: International transfer of funds, nominally to stimulate development.
  
  A general technical definition is difficult, owing to the appropriate discount rate for the value of loans relative to the value of grants, and contentious ‘disguised’ aid elements (e.g. costs of accepting refugees, trade-tariff easing, state-owned enterprise investments, or military support).


• Humanitarian aid: Ultra-short-term disaster relief (immediate transfer of goods and services). This is not development aid.


• Official aid (ODA): Public transfers like Official Development Assistance (ODA). “Flows of official financing with the promotion of economic development of developing countries as the main objective, which are concessional in character with a grant element of at least 25 percent (using a fixed 10 percent rate of discount).” 16
  
  The loan component of ‘aid’ must be stressed, since even at concessionary interest rates it gives rise to serious side-effects. Technically, the 'ODA' term only applies to countries in the Development Assistance Committee (DAC), but I use 'ODA' to include non-DAC aid (e.g. China and Saudi Arabia). To avoid arbitrariness in adding up total ODA, all real ‘flows’ of resources should be considered. Official aid is thus financial assistance and technical assistance and debt relief and commodity transfers. I exclude military assistance and transfers to NICs where possible.


• Less developed countries (LDCs): Poor nations. Membership varies according to the metric used; I use the term to cover the bottom two quartiles on the HDI (currently, those below 0.698). The IMF currently lists 150 LDCs, comprising 70% of the world’s population. 14


• Least developed countries (LLDC): The poorest subset of LDCs – the ‘bottom billion’ people. The UN uses three criteria for an LLDC: “a low-income criterion… of GNI per capita (under $750 for inclusion); a human resource weakness criterion; and an economic vulnerability criterion…” 15 People in these comprise 13% of the world’s population.


• Newly industrialised countries (NICs): Countries with strong growth rates, and usually ballooning trade. I use the term for the 2nd quartile (currently, those above 0.698 but below 0.793 in HDI).


• More Developed Countries (MDCs): The top quartile of countries on the HDI (currently those above 0.793). More usually taken to be those countries with GDP above $12,275 at 2010 prices. (I adopt the euphemistic term ‘MDC’ only to emphasise the relativity and incompleteness of their superiority.)


• Development Assistance Committee (DAC): A committee for tracking government aid. Rougly "Western aid".


• Technical assistance (TA): Transfer of skilled labour, services or training. Supposedly ‘exporting knowledge’. Often in the form of economists, logisticians, managers and teachers, without whom aid is minimally useful. Forms perhaps 30% of total ODA, and has done throughout the modern aid period. 17

Figure 2 gives a breakdown of recent official aid. Note that bilateral official aid still constitutes 68% of total flows. 19

Non-DAC countries do not report much, and it is difficult to estimate proportions for them, but we do know they gave something like $11 - 41 billion on top of the above; they thus represent between 8 and 31% of all ODA. 20 NGO aid has increased rapidly in recent years, reaching $24 bn, or 15% of total aid. 21

A note for perspective: we often exaggerate aid’s proportion of the fiscal makeup of LDCs. Even at its proportional peak in 1968, aid still only contributed 10% of the GDP of recipient LDCs. 22 Fig.3 illustrates the recent dwarfing of aid by other flows (themselves still dwarfed by domestic activity):

Given some common moral intuitions, we cannot ignore absolute poverty. In the abstract:

1. Moral universalism: “neither our distance from a preventable evil nor the number of other people who, in respect to that evil, are in the same situation as we are, lessens our obligation to mitigate or prevent that evil.” 26
2. Underdevelopment is an evil.
3. So if underdevelopment can be alleviated, it should be.
4. Aid alleviates underdevelopment.
5. Conclusion: We should give aid to alleviate underdevelopment.

Ascertaining premise 4 has been the (unsuccessful) focus of generations of developmentalists. Abolitionists dispute it, and ask instead “Would development benefit if aid were stopped?”

Notice that, even if premise 4 is false, it does not follow that we should not give aid; it vitiates only the most ambitious Rostovian aid. Discharging our obligations in this case would instead involve solidarity programs (like health) and the opening of borders, to allow access to economic opportunity where it already exists.

Who must justify themselves here, the aid pusher or the abolitionist?

Since aid is always a cost to donors, aid could be said to ‘fail’ if it harms or if it has no net effect.

But, as Riddell puts it, the prima facie nature of aid gives us a reasonably strong prior probability in favour of aid:

...aid is something which is added: it is additional to what the recipient currently has... The more reasonable default position on the impact of aid is that it should be viewed as helpful, unless it can be shown not to be beneficial, rather than (as frequently occurs at present) it is assumed to be detrimental unless proved to have been effective. 27

Transformation aid has usually been justified in terms of the balance of payments of recipients – poverty is seen as primarily a lack of capital, so aid is intended to fill ‘gaps’, stimulating growth, and thus springing populations from poverty ‘traps’.

LDCs are thought to be unable to develop independently as a result of environmental shocks; international power structures; market failures; or government failures (like chronic war).

Aid is intended to improve growth via capital accumulation. The classic mechanism runs:

Aid -> capital accumulation -> growth -> income -> saving -> investment -> take-off growth

28

The 'two-gap' analysis argues that one of two constraints will always be limiting growth: either the savings gap (the shortfall of domestic savings from the level of investment required for independent growth); or the foreign-exchange gap (the deficit of export revenues from the value of imports needed for take-off development). Where I is total investment, F the volume of capital inflows, and s the marginal propensity to save and Y is annual national income, the savings gap is: $$ I \leq F + sY $$ And where m_1 is the marginal import share, m_2 is the marginal propensity to import, and EX is export volume, the foreign-exchange gap is: $$ F \geq (m_1 - m_2) \,I + m_2 Y $$ (Even though F, E and Y are held exogenous, note that the two gaps are not independent.) 29 If there is a poverty trap, aid will lead to efficiency gains and the return there will be a surplus; if not, the aid will merely redistribute. Apparent traps are seen in many facets of the life of the poor: Sachs’ nutritional traps 30; Taylor’s fiscal and inflation gaps 31, Collier’s conflict and landlock traps 32 and Birdsall’s ‘institutional trap’. 33 These are general rationales for aiding civil society, supplying post-conflict regions, building infrastructure, and encouraging governance reform.

Finally: in a weak sense of 'necessity' there is an obvious necessity for aid. Call a given aid package a practical necessity if it meets three criteria:

1. It is required by a recipient;
2. It could promote development;
3. It cannot be effectively obtained from other sources.

These situations often arise in LDCs, this is a fine prima facie template to apply to particular cases. Aid, then, can be minimally construed as the last resort of development.

(The above model of economic growth is far from current, but it represents past beliefs well, and is a good clear position to react against later.)

To say that increases in (microeconomic) welfare are merely about 'solidarity' is to downplay the scale of our success on the nonstructural front.

Toby Ord has a neat argument placing a very positive lower bound on the impact of foreign aid. This bound actually balances all aid spending as cost-effective, even given unrealistically pessimistic assumptions. Consider just the smallpox eradication program, and imagine that all other aid spending ever had no net effect at all.

1. Say that smallpox eradication was the only good to come of aid.
2. We have spent $2.7 tn on foreign aid.
3. Eradicating smallpox has saved at least 60 million lives. 321
4. Then it cost an average $45,000 to save one life via aid.
5. The de facto price cap, in the rich world, for preventing one death is about $7m. 322
6. So aid per se has been cost-effective on average, even given the most pessimistic assumptions: it saved lives at 1/150 of the going rich world rate.

(The total cost of eradicating smallpox was actually about $400 million and, quite besides the misery and death it has averted, it continues to save hundreds of millions of dollars a year in obviated healthcare costs and counterfactual lost productivity. 333 It is hard to overemphasise what an incredible project it was and is.)

I find this persuasive: persuasive enough to answer the headline question above as "No, we shouldn't: keep vertical health interventions at the very least". But this essay focusses narrowly on long-term macroeconomic transformation (starting a self-sustaining engine of autonomy and growth), rather than on (object-level, microeconomic) welfare. Stop reading now if you're not as hopeful about that.

Peter Bauer maintained his caustic, heterodox view on aid for forty years - that is, over the whole modern aid period. In 1996 he continued, as he had begun in the 50s:

...official government-to-government subsidies ought to be terminated or at least drastically curtailed. This seems impractical, partly because of the momentum of existing commitments, partly because of the extremely powerful and articulate interest groups behind the policy.” 34

Instead of investment gaps or poverty traps, Bauer attributes economic underdevelopment to either a lack of the ‘social determinants of development’, or to government intervention – which he often equates with government failure.

Bauer's philosophical critique of aid

1. Aid as falsely axiomatic. Bauer’s early work was written in an incredibly uncritical context: the standard theory and practice was more or less, “any aid is beneficial to any country”. Bauer argued that this view was endemic in C20th aid, as seen even in its name – since to call aid ‘aid’ is to beg the question: its name should be determined by its effect on recipients, not by the nominal intentions of donors.
   
   According to him, it was also common to elide between ‘increasing’ (the volume of) aid and ‘improving’ aid. The two-gap take-off model of growth (outlined above) was a commonplace: aid was held to be sufficient in itself to transform LDC economies.
   
   Since the larger part of this essay is an analysis of the dozens of mechanisms that undermine aid impacts; since the sufficiency of aid has never been demonstrated, Bauer’s attack on the ‘axiomatic approach’, as opposed to the new empirical approach, was important and prescient.
   
   
2. Against narrow conceptions of development. For much of the mid-C20th, 'development' was read as 'increase in material wealth or productive capacity for material wealth'. Bauer is broader: he attributes underdevelopment to missing ‘social determinants of development’ (by which he means economic rationality, rule of law, norms against corruption, background trust, and a pro-market, entrepreneurial culture). His stress on the ‘social determinants’ of development amounts to the claim that LDC populations are culturally backward: as he nicely puts it, “Material advance requires a modernisation of the mind.” 46
   
   Elsewhere he prefigures Amartya Sen’s capability approach to development: “I regard the extension of the range of choice, that is, an increase in the range of effective alternatives open to people, as the principal objective and criterion of economic development.” 36
   
   
3. Aid as perverse politicisation. Bauer holds that poverty is generally a reflection of government failure. This is an extrapolation from the Econ 101 principle of perfect competition:
   
   

   [Market] restrictive measures affect adversely both the allocation of resources and the growth of resources. 35

   Bauer regarded ODA as politically dangerous because it centralises economic power in the government and removes negative feedback against bad policies. Adding resources independent of the tax base plausibly leads to a corrosion of democratic accountability; removing the connection between policy and results means that development is prevented from occurring by aid, and that the apparent gains are lost is when aid is withdrawn. This process is key for Bauer for more than efficiency reasons. He frequently emphasises that 'responsibility' is as much a constituent of development as is wealth; and “In this sense aid pauperises those it purports to assist.” 37
   
   

Bauer's technical critique of aid

4. Against material poverty traps. Bauer identifies an apparent paradox – if poverty alone traps regions in poverty, then no country could ever have developed, since all countries began poor. Syntheses of aid effects tend now to conclude that aid is not a necessary condition of development, nor clearly sufficient. 3940
   
   He goes so far as to deny that volume of investible funds is a determinant of development at all: “What has to be remembered and emphasised is that having capital is the result of successful economic performance, not its precondition.” 41 There is some contemporary support for this. 42 Most extraordinarily, in a fit of Classical economics, Bauer explains the low domestic savings rates in LDCs as a rational reflection of the lack of opportunities for investing the savings locally. 43
   
   
5. Adverse political selection. Bauer further claims that aid causes poverty traps, by rewarding destructive policy and propping up incompetent governments: “By maintaining a minimum level of consumption, the subsidies avert total collapse and conceal from the population the worst effects of destructive policies. These subsidies also suggest external endorsement of damaging policies.” 38
   
   
6. Adverse economic selection: Bauer argues that aid fuels the need for itself, systematically sustaining badly-managed economies. The mechanisms he suggests this occurs by are:
   
   1) ‘Dutch disease’, i.e. damage to domestic industry by unstable financial inflows. 44 A current-account surplus leads currency appreciation which leads to uncompetitive exports which leads to the decline of domestic industry – which leads to increased aid flows to address the stagnation;
   
   2) the ‘Samaritan’s Dilemma’. Intuitively, aid poses a moral hazard to recipients: with the expectation of help independent of results, individuals are less incentivised to be economically rational and entrepreneurial, and recipient government budgets are less constrained by prudence and negative feedback. The latter licences chronic inefficiency due to the lack of competitive pressure to invest properly, there is no automatic pressure towards the efficient use of donated capital. At minimum we can agree with Bauer that development involves both poverty reduction and autonomy promotion.



7. Aid as rent-seeking.
   
   
8. Recipient misuse. The latter problems enable allocations with blatantly poor connection to development. The demarcation between misuse and merely inefficient use is not sharp (for instance, what exactly constitutes a ‘white elephant’, grossly inefficient public project? And the ubiquitous bias towards urban and elite uses of aid could be classed ‘misuse’). Fig.4 gives an idea of the extent – perhaps a fifth of all aid – and confirms Bauer’s insistence that capital inflows are misused – though not perhaps his explanation for the misuse.

Against Bauer

Happily, many of the above imply falsifiable claims. This isn't a criticism of Bauer; instead it is a mark of his virtue as a scholar.

Part of his cultural thesis is contradicted by the strenuous activity found among the poor by economic field research: for instance, the average (non-agricultural) self-employment rate in LDCs is 40%. 47

His point also predates the recent rapid uptake, and emergent uses, of technology (e.g. the growth of mobile use in Africa – from 3% subscription in 2000 to 73% in 2011. 48 Also, most strikingly, the use of mobile phone credit as a liquid asset in increasing areas of Africa). 49

Here is a crude heuristic for measuring the cultural quality of an economy (call it a perseverance index, PI). Where BC is the annual rate of business creation in the country, W is size of workforce, E is the country's rank on the Ease of Doing Business index:

$$ \text{Formal
perseverance} = \frac{ BC }{ W } \times \log{(E)} $$ And where S is the self-employment rate in the country, and Y / W is national income per capita: $$ \text{Informal perseverance} = \frac{ S }{ Y / W } \times \log{(E)} $$ $$ \text{Perseverance} = (\text{Formal} + \text{Informal}) $$ This measure attempts to represent 'two main constraints' on LDC entrepreneurs – ‘ease of doing business’ (well-represented by the World Bank’s index of the same name 50) and low capital. Fig.5 shows some values for this (totally uncalibrated!) measure:



Bauer’s aversion to government intervention also falls short, theoretically and empirically. Theoretically, he entirely neglects ‘second-best’ welfare economics, which gives a conditional rationale for intervention in a wide variety of contexts. As early as 1956, Lipsey and Lancaster derived a result within general equilibrium, the ‘Theorem of the Second Best’. Informally: “If a market is not perfectly competitive, then all bets are off; the second-most competitive market is not necessarily more efficient than the third-most, etc.” 53

Since no market has ever satisfied the neoclassical optimality conditions, this demotes the theoretical conclusion we may draw from general equilbrium to: “In any given real market, increasing competition might increase efficiency, depending on the interaction of its frictions”. Bauer had little respect for formal economics, but this result is ubiquitous and uncontroversial in even neoliberal discussions. 54 The economic experience of most of the world contradicts Bauer's universal prohibition on fruitful state intervention. (For instance, China, South Korea, and Japan stand out as uncontroversial successes of industrial policy.) 55

Regarding the corruption of political processes by aid flows, Bauer omits to mentio the ubiquitous malign influence of private capital on democratic life, in LDCs and beyond.

The final test of Bauer’s claims must be empirical and specific. Bauer's work was formulated in an extremely data-poor time; it is possible that his claims stood before, given the available data. I cover the present evidence for his claims in "", below.

Onto two contemporary Bauerians: William Easterly and Dambisa Moyo.

"The aid campaign against diseases in Africa... is likely the single biggest success story in the history of aid to Africa... The well-known Kremer and Miguel paper showed a strong effect of deworming on worm infection rates in a district in Kenya, which reflected not only direct effects on children receiving the drugs but also surprisingly strong externalities to others in the same school or nearby schools...

Breastfeeding, immunization against diarrheal diseases, micronutrient supplementation and oral rehydration therapy (ORT) have all been found to work in randomized trails in the fight against diarrhea..."

Who is this wild-eyed aid optimist? Oops, it's me.

The point is that even those of us labeled as "aid critics" do not believe aid has been a universal failure. If we give you aid agencies grief on failures, it is because we have seen some successes, and we would like to see more!

The Burden

William Easterly’s critique is significant and well-founded, but with fits of extremism. (Amartya Sen describes Easterly’s White Man’s Burden as a “shotgun summary” of the history of aid. 56) Easterly's scepticism is thoroughgoing :

the problem of making of making poor countries rich was much more difficult than we thought. It is much easier to describe the problems facing poor countries than it is to come up with workable solutions to their poverty. 57

Many of his themes – a focus on incentives, rejection of top-down intervention, a Hayekian view of society – overlap with Bauer’s, in recent work Easterly lets slip nuanced views which make it difficult to call him an abolitionist. Most clearly: “The aid agencies need tough love from the critics – who must not abolish them but must pressure them so that aid actually reaches the poor.” 58

So, though he is actually a reformer, I include him here because his ideas still entail a radical reorganisation of foreign aid. (He would abolish the largest category of intervention, the 'Poverty Reduction Strategy Papers' and all such development Plans.)

His organising distinction is to split developmentalists into the (paternal and aloof) ‘Planners’ and the (piecemeal and entrepreneurial) ‘Searchers’:

In foreign aid, Planners announce good intentions but don’t motivate anyone to carry them out; Searchers find things that work and get some reward. Planners raise expectations but take no responsibility for meeting them; Searchers accept responsibility for their actions. Planners determine what to supply; Searchers find out what is in demand… Planners never hear whether the planned got what it needed; Searchers find out whether the customer is satisfied. 59

Like Bauer, Easterly’s prescription is a call for more capitalists. But Easterly is more critical than Bauer (and to be fair, has much better data); Easterly can accommodate public choice problems where Bauer’s focus was on individuals.

• Aid has evaluative impunity. Accountability from donors and aid agencies, and independent evaluations of the effectiveness of aid efforts. This is for aid effectiveness:
  
  

  We still believe that more aid will reach the poor the more people are watching aid, but, as we’ve always known, there’s a lot more to development than aid. 60

  The other prong is to attack the incredibly low quantity and quality of impact evaluation in aid; reviewing 127 evaluations, the ILO found that just two of them had internal validity (i.e. could distinguish program impacts from external factors). 61 Project self-evaluation is the only common data form. Only three of the perhaps 100 official aid agencies complete regular, systematic overviews of their work (UK’s DFID, Sweden’s Sida and the Netherlands’ NL-Aid). 62 This apparent indifference to actual effects is damning of current aid.
  
  
• Ghost of Financing Gap. Easterly also presents strong evidence against the two-gap model: “the Financing Gap growth model fails all theoretical checks and empirical tests this paper performs upon it. There is no theoretical or empirical justification for assuming a short-run proportional relationship between growth and ‘investment requirements.’” 63 Note that this general trend does not weaken the practical necessity of aid in particular circumstances.

Criticising Easterly

• Admirably empirical as he is, Easterly still overstates. For instance, he portrays the aid effort as heartbreaking squander:
  
  

  The West spent $2.3 trillion and still had not managed to get twelve-cent medicines to prevent half of all malaria deaths… four-dollar bed nets to poor families… three dollars to each new mother to prevent five million child deaths… It’s a tragedy that so much well-meaning compassion did not bring these results for needy people. 64

  But this doesn’t stand up. Firstly, regarding ‘generosity’: dividing $2.3tn over 50 years, coming from an average 750 million donors per year, the contribution to aid per rich person was just $61 per year, or 0.3% of total income. 65
  
  Moreover, this amounts to around $15 per recipient per year – a sum which is unsurprisingly ineffective at ending poverty. (Given the constant intentional mistargeting of aid, to the non-poor, neither is it right to take the full $2.3tn figure as the operative one.)
  
  Easterly also confounds matters by citing the above number without a reference class. How much should it have cost to eliminate malaria, or to stimulate take-off growth in an industrialising economy? Easterly’s marginal approach rules that this last question is irrelevant (because that's not how economies develop), but answers to the others are scarcely clearer.
  
  



• There are at least two tolerable reasons for the historical lack of impact evaluations of aid programs. Oxfam America found that the costs of running random controlled trials (RCTs) on ‘Saving for Change’, its microfinance project in West Africa, would match those of the program itself, and so cancelled the evaluation in the name of impact. 66
  
  This is certainly shortsighted (an excessive time-discount preference) in a development agency, but not obviously dishonest. Secondly, RCTs can cause selection bias; they draw attention away from factors that are harder to operationalise (for instance ‘culture’ in institutions) – which may constitute much of meaningful development.



• Easterly is in the odd position of accepting the strong evidence that many global health interventions have been extraordinarily cost-effective -
  
  

  There are well known and striking donor success stories, like the elimination of smallpox, the near-eradication of river blindness and Guinea worm, the spread of oral rehydration therapy for treating infant diarrheal diseases, DDT campaigns against malarial mosquitoes (although later halted for environmental reasons), and the success of WHO vaccination programs against measles and other childhood diseases. The aid campaign against diseases in Africa (known as vertical health programs, see discussion below) is likely the single biggest success story in the history of aid to Africa. 666

  - without following through to Ord's conclusion, of an acceptable lower bound for aid overall. This inconsistency makes sense if we consider that we still want to disincentivise ineffectual aid, even if 'aid' overall is a net positive because of some excellent health programs.
  
  
• Finally, Easterly rejects ‘plans’ too indiscriminately. There are areas – health, education, sanitation, water, environmental protection – where positive externalities, plant requirements and complementarities are so high that co-ordinated and top-down action has comparative advantage. In LDCs, which tend to lack capital for both halves of public-private partnership schemes, and also the syndicate 'strategic investment agreements' that sometimes suffice for very large HDC infrastructure projects.
  
  A collection of atomised searchers cannot harness co-ordination benefits or economies of scale, which minimises their impact on some tasks that we can well-define. One of the biggest barriers to aid effectiveness is the existing level of fragmentation of efforts; Easterly’s decentralising proposal would deepen this.

Easterly’s call for efficiency is humane and constructive. But advocating marketisation and 'narrowing' of aid means turning away from some of LDCs' more important problems: promotion of good governance – on the basis of notoriously weak data.

Aid has been and continues to be an unmitigated political, economic, and humanitarian disaster for most parts of the developing world.

...[Moyo] seems to mistake absence of proof for proof of absence—and to confuse failure to prove benefit with proof of harm.

In 2009, Dambisa Moyo published a strident analysis of ODA to Africa, Dead Aid. The book is dedicated to Peter Bauer – but she absorbed his caustic tone without his analytical power. She presents a shock therapy thought-experiment:

What if, one by one, African countries each received a phone call (agreed upon by all their major aid donors – the World Bank, Western countries etc), telling them that in exactly five years the aid taps would be shut off, permanently? 69

She claims that aid has clearly failed, by increasing poverty; and that there are obvious alternatives for African development. I here address these claims and what she calls ‘alternatives’ to aid (which are actually complements).

To be able to condemn the whole aid industry, Moyo is highly selective about evidence, and poses a false dichotomy between aid and other development measures. Barder:

She just asserts that aid causes corruption, bottlenecks, losses of competitiveness and erosion of accountability… Moyo does not support any of this with any evidence, and – more alarmingly – she misrepresents the academic literature to pretend that it supports her conclusions. 70

The evidence presented for the claim that 'aid has clearly failed' is a correlation lifted from Easterly:

Though striking, she does nothing to control for reverse causality; Fig.6 is only the impression of evidence. Moyo points out aid is grossly mistargeted and misused, but, as Adekeye Adebajo points out, this is not an argument against aid: “She notes that 85% of aid is diverted to other purposes, which suggests that it is the abuses of aid — rather than aid itself — that is the main problem.” 72

As regards the claim that aid has increased poverty: there simply is not enough evidence to support this. Moreover, by attributing latter-day African poverty primarily to aid, Moyo ignores or plays down factors with actually documented links. Adebajo enumerates them: “...structural issues from colonialism, external debt, and the global distribution of resources, as well as conflicts, poor governance, lack of a productive base, and failures of economic integration, have all contributed much more to poverty than aid.” 73 Moyo’s positive prescriptions are:

1. Free the markets. The book is the mirror of Jeffrey Sachs’ hubris – and is in fact more utopian than his The End of Poverty. She claims that “Development is not a mystery; each of the elements of the Dead Aid proposal has been tried and tested and yielded success – and governments and policymakers know it”. 74 She effectively advocates shock therapy – a policy which the historical record and her fellow aid sceptic Easterly both countermand – and moreover that LDCs undergo it without the benefit of any added resources with which to transform the economy. 75
2. More microfinance. In an overview of microfinance in Bangladesh, one long-time advocate, Jonathan Morduch admits that, “Strikingly, 30 years into the microfinance movement we have little solid evidence that it improves the lives of clients in measurable ways.” 76 The book excessively operationalises development metrics, apparently under the assumption that real effectiveness can be reduced to the volume of participants and cost-effectiveness. For instance: her primary evidence for the success of microfinance is high repayment rates. This is a poor measure of actual development impact; we cannot assume, as she does, that repayment is a real proxy for overcoming dependence, since repayment has very often been enforced by shame and intimidation. 77
3. Exaggerate macrofinance. Even writing in the midst of a global financial crisis, Moyo overstated the availability of bond issues and credit, as Collier noted: > Moyo has been unlucky in her timing… The opportunity for African governments to raise money on international markets has evaporated even more rapidly than it opened around four years ago. The global financial crisis has drastically reduced investor appetites for risk: for example, the government of Kenya had planned to raise $500m through an international bond issue, but that is now out of the question. 78 Even were it available, Moyo’s appeal to commercial credit entails a vast increase in debt for the world’s poorest – “in effect advocating sub-sub-sub-prime lending. We have learned the risks to sub-prime lending in the rich world. Do we really want to engage full scale in extending the predictable risk in the poorest parts of the planet?” 79 Moreover, global interest rates have been near-zero in the period 2009-2012. With continuing quantitative easing, this trend should not last in rich-world economies: current monetary expansion will force rising interest rates. Global credit could then become too expensive for most LDCs, and heavily indebted poor countries (HIPCs) will risk crisis (as seen in 1982, when the US Federal Reserve raised US interest rates and caused a decade-long debt crisis in much of Latin America 80; Asia in 1997 81; Argentina in 2001 82; and most of the world in 2008).
   
   Nor is the bond market necessarily suitable for funding social programs, which tend to have a long gestation periods, or social returns generated in other units than financial profit. Results from Ira Saltz suggest that portfolio investment follows growth, and does not lead it. 83 If so, this makes Moyo’s advice totally inapplicable, to HIPCs at least.
   
   

Other prescriptions – encouraging FDI, an end to protectionism in MDCs, an end to IFI structural adjustments – are commendable. But they are not aid substitutes, and aid can help redress the damage that even they do.

Moyo is correct: aid cannot develop LDCs alone. But no economist – not even Jeffrey Sachs – anymore says it would. Moyo’s bowdlerised economics aside, the process by which development can in general be caused is still extremely mysterious to us.

A few people have a bed for the night
For a night the wind is kept from them
The snow meant for them falls on the roadway
But it won’t change the world
It won’t improve relations among men
It will not shorten the age of exploitation.

Some postcolonial critics argue that aid is more or less espionage: a circumspect method for internationalising capital and Westernising the poor. (In this framework, the former is said to be a bad thing.) Teresa Hayter’s passionate attack on some political dynamics in C20th aid led me to call this topic 'abolitionism' in the first place; she sees aid as a disguised, leveraged control of LDC politics and resources.

She recently reiterated her position:

Some aid projects may in themselves be useful. Others may be problematic: foreign so-called experts, however well-meaning, often get it wrong. But, above all, the policies which governments have to adopt to get the money are deeply damaging to the interests of the poor. I believe, therefore, that official government aid, on balance, does more harm than good to the poor of the Third World. 85

‘The Creation of World Poverty’. Hayter emphasises the role of the colonial legacy and the rules of international trade in the subjection and stagnation of LDCs. This argument is not empty: colonial extraction occurred in many cases, like the Belgian extraction colonies, the whole slave trade, or the UK’s opium wars. 86 Colonists also created most of the arbitrary, centralised, and abusable states in which the poorest quartile live.

Some contemporary empirical support for the accusation of neo-colonialism:

1. The total volume of aid is dwarfed by the 'shadow costs' of MDC protectionism. To take one large subsidy: the EU’s Common Agricultural Policy costs LDCs $15bn a year, by artificially suppressing the market price of EU crops. 87
2. Aid has constantly been mistargeted for politically expediency. For instance, Israel, one of the most developed countries in the world, is the recipient of by far the most US aid, historically. 88
3. Contemporary surges in Chinese investment in, and aid to, Africa are raising concerns about the manipulation of government and local law. 89 There is little reason to think that Western finance is less concerning (except where it is directly supporting recipient democracy).
4. Excess profit extraction (through transfer pricing and tax evasion) is customary of multinational corporations operating in LDCs. 90
5. Excluding China, the world poverty headcount has not declined significantly over the last 20 years, despite apparently large flows of aid. 91

Why does Hayter say this is?

• Tying. Tied aid (aid resources which commercially serve the donor) seems straightforwardly neo-imperialist. Though in most cases it is merely inefficient and venal (tied aid is subject to 20% discounting in good measures of ODA 92), imports replace domestic products, which may further harm domestic industries. Food aid is a severe instance of tying suppressing local prices and depressing output. 93 The proportion tied remains incredibly high – perhaps 60% of ODA in 2006. 94



• Conditionality and imperialism. Aid is seen by Hayter and others as a denial of LDCs’ national autonomy. This is anti-development in a less quantitative sense. As Arturo Escobar has it:
  
  

  The infantilization of the Third World was integral to development as a ‘secular theory of salvation’.” 95

  The salient point to draw from this slightly vague accusation, is the IFIs’ record of giving 'false paradigm' economic advice to LDCs. 96 Christian Aid puts the total cost of structural adjustment (that aid given in order to facilitate rapid marketization reform) to Sub-Saharan Africa at $272bn. 97 Moreover, aside from slightly less emphasis on extreme privatisation, the IMF’s conditions have not changed substantially since the 1990s. 98
  
  

Criticising Hayter

The main thing to note is that many of these criticisms are not criticisms of aid: they are instead aimed at the general political economy of the modern world. The hypocrisy of MDCs – protectionism for themselves, but free trade for the poor; 'socialism for the rich'; unsafe policy manipulations in LDCs; and the unchecked brain-drain of the brightest away from LDCs – does not speak against aid unless we take a very stern, Kantian ideal of international action – that the impure motives that are doubtlessly involved in donor countries preclude practical benefits for recipients.

Furthermore, some of her key claims are false. Not all colonies were extractive; Bauer’s historical study of rubber in West Africa and agriculture in Zimbabwe shows that at least in places, colonial forces created domestic capital and employment, and passed much of it intact to the post-colonial population, which subsequently generally failed to develop. 99

Tied aid is now illegal in the UK, and similar motions are being considered throughout the DAC countries. 100 Hayter also duplicates Moyo’s simplistic correlation error, as regards the headcount of poverty relative to the increasing aid level.

Finally, the economic experience of autarkic and ‘de-linked’ countries speaks against her positive prescription: to just do without aid. Hayter’s analysis relied on the existence of a supportive international Socialist network to replace capitalist aid. 101 Even granting the dubious notion that Soviet or Chinese aid would have been less manipulative or dependence-forming, every economy in the world is now deeply interdependent, and the available evidence does not disqualify aid as a possible economic route out of dependence.

Hayter forces us to consider the politics of aid. Large parts of the aid industry have primarily served donors. But when evaluating aid as a future policy option, it's important to consider the tractability to reform of the hypocritical and inefficient parts: future aid could be different. I rely on the following chapter to suggest that large parts are tractable.

Direct effects

Fragmentation and bureaucratic deadweight. In theory, where aid is ‘fragmented’ (delivered by many donors or in many small projects), a large, revenue-consuming domestic bureaucracy is required to report on the aid. In practice, there is evidence suggests that fragmentation does impair the quality of LDC 102 , but no detailed analysis is yet available.

Institutional effects

“The Aid Curse”. In theory, any pool of expenditure independent of the tax base (and so democratic accountability) can “curse”; that is, cause institutional damage, such as increased corruption, rent-seeking and autocracy. (This was first identified in natural resource-rich LDCs.) In practice, these effects are not exhibited across recipients in the post-Cold War period. 103 Sarah Bermeo explains this by the fall in fully-fungible ‘budget support’ aid: “Contrary to previous studies, I show that it is donor intent, and not an unintended “curse” associated with increased financial flows, that influences this relationship.” 104

The Samaritan’s Dilemma. In theory, aid can lead to moral hazard on a grand scale, because the anticipation of aid diminishes responsible budgeting and self-directed development work, creating, perhaps, dependency and chronic inefficiency. In practice, dependency is extraordinarily hard to operationalise. Aid/GDP or Aid/budget ratios are too detached from the ground-level problem to indicate very much.

False-paradigm conditionality. In theory, aid given conditional on policy changes can lead to IFI economists causing large-scale economic damage, owing to the misapplication of models or lack of local knowledge and analysis. In practice, the evidence is that this was prevalent and remains worryingly high; Riddell demurs giving a quantitative damage estimation, in terms of cost or scale, but implies its costs were and are severe. 105

Distorted incentives. In theory, imports such as tied aid can replace domestic products, harming domestic industries. They can also undermine local markets by simple supply and demand logic: increased supply depresses prices, weakening incentives to produce. In practice, this has been seen – severely so with regards to (now largely discredited) food aid. 106

Macroeconomic effects

Does aid boost growth? Four problems preclude tidy summary of the net effect:

1. Weak data. Riddell emphasises the partial and provisional nature of all cross-country and much individual project data. The gaps are staggering: the largest failing of donors to date has perhaps been the simple failure to study actual outcomes of giving. 107 Pace Hayter, it seems likely that most abuses would stop were donors to look more closely.


2. Vast cross-country variation. The marginal benefit of aid varies enormously across target nations, sectors of nations, and by volume. David Roodman’s index discounts a dollar of aid to Iraq to 22¢ per dollar (due to Iraq’s aid saturation and endemic corruption), while aid to Ghana counts 97¢ in the dollar (due to its high poverty and adequate governance). 108


3. Enormous confounding shocks to LDCs. The developing world underwent vast political, ecological, and economic violence over the modern aid period. The Centre for Systemic Peace notes that of the 348 ‘major episodes of political violence’ since World War II, 339 occurred primarily in LDCs. 109 79% of large-scale catastrophes occurred in countries receiving aid, as did 90% of the severe droughts. 110 The burden of the AIDS epidemic falls disproportionately on the poor. These latter natural events prohibit the apparent causality exploited by Moyo, that aid is correlated with disaster.


4. Contradicting overviews. One famous paper by Burnside and Dollar (2000) found a good correlation given “good fiscal, monetary and trade policies”; their conclusion became received wisdom for a time. 111 However, Easterly et al (2003) ran their method with broader data, and undermined it. 112 Clemens et al (2004) 113 found a strong short-term aid-growth link regardless of policies; Rajan & Subramanian (2005, 2008) nulled the link. 114 There are other antagonistic pairs of papers. 115116117118 As a result, David Roodman discards cross-country regression altogether:
   
   

   On balance, the quantitative approach to exploring grand questions about aid effectiveness, which began 40 years ago, was worth trying... given the limited and noisy data available, its effects on growth in particular probably cannot be detected. 119

Other perverse macroeconomic effects:

5. Fungibility. Aid intended to increase spending in crucial social and economic sectors can instead substitute for spending recipients would have undertaken in those sectors anyway. If resources are then used unproductively, the gains are lost (and can have a perverse effect on domestic savings). One analysis claims that 70% of all aid is potentially fungible. 120 In practice, again, this has been seen on a smaller scale – the actual substitution is very likely closer to 20%. 121
   
   Furthermore, fungibility is not necessarily inefficient, and can in fact make efficiency gains, owing to better local information about how resources should be used. (This could explain why ‘good governance’ is associated with aid effectiveness: one part of good governance is better information about the jurisdiction.)
   
   Most worryingly, Collier and Hoeffler find that around 11% of aid is able to leak into military spending. 122 This problem at least is tractable: Collier and Hoeffler themselves note that “Donors appear to be fully successful in preventing aid from leaking into military expenditure”, eliminating each new leakage over time. 123
   
   
6. Dutch Disease. In practice: Adam (2005) surveys available cases of currency-depreciation caused by sudden financial inflows, and finds little evidence of aid-to-depreciation correlations. 124 Barder is also unworried: “Any reduction in output because of an appreciation of the real exchange rate is more than offset by the additional consumption and investment that the aid finances.” 125 Barder stresses that Dutch Disease is manageable if policy is free to adapt, and if aid is timetabled – since it is the volatility of flows, rather than inflows per se that give rise to it.

We face enormous uncertainties (even more than usual for macroeconomics); Table 1 is an attempt to duly weight the above using expected value and a third 'tractability' estimate.

• ‘Magnitude’ is the prima facie impact of aid on development;
• ‘Frequency’ how often the effect actually manifests;
• ‘Tractability’ how easy the effect is to improve.

Positive integers represent support for aid; negative for abolition. Data is poor for almost all factors; exclamations represent extremely unreliable values (error above ± 20%). Frequency data is invariably weak, since they stem from incomplete overviews of the phenomenon. To be frank, the values are less important than the structure of this assessment:



[Work in progress]

126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156

One basic test of robustness would be to run the above excluding the global health outlier from aid. (This is in a sense unfair to the aid boosters, since health is a primary determinant of human capital and thus growth. But it could be an interesting way to save abolitionism.)

Table 1 does not allow us to assert that aid ‘works’ in the transformative sense; the evidence is too weak. But it does entail that abolitionism is not warranted given the evidence: the evidence speaks clearly against net harm, and that there is some evidence that aid can be improved.

If the abolitionists outlined above are guilty of anything, it is only overstatement, and not the heartlessness or ideological blindness that we might have expected from anti-aid economists. Aid has been ripe for criticism. Easterly’s quip – that “critics are better for foreign aid than apologists” – is substantially correct. 157 But, sloganeering further, abolitionists have opted for the wrong prong of “end it or mend it”: we are instead compelled to mend the abuses, difficulties, and ambiguities of aid.

Bauer was the first to identify abiding problems with aid, but his extreme neoliberalism is misleading about aid’s actual prospects. Easterly – not really an abolitionist, despite his bark – fared better, but downplays the upside of C20th aid. Moyo repeats Bauer in simplified form; cherry-picked as it is, her case is more easily deflected. Hayter’s principled attack on aid was seen to focus on motives rather than impact, with her policy prescriptions based in a departed world order.

All of them avoided discussing NGO aid – perhaps due to its relative efficiency, or its historical unimportance. My analysis suggested that most strong conceptual arguments against aid are let down by weak evidence. Even so, the many historical failures of aid warrant present-day watchfulness and scepticism.

Bauer is right to insist that aid workers be abolitionists in the long-term: aid must aim to make itself unnecessary. While aid is neither necessary nor sufficient to develop LDCs; while, to date, development aid has not worked efficiently or without side-effects, there seems to be reason to continue.

Aid should not be abolished, in part because the insights of the abolitionists have revealed a way forward for it.