Take up the White Man's burden—
In patience to abide,
To veil the threat of terror
And check the show of pride;
By open speech and simple,
An hundred times made plain,
To seek another's profit
And work another's gain.

Take up the White Man's burden—
The savage wars of peace—
Fill full the mouth of Famine,
And bid the sickness cease.
—Rudyard Kipling, The White Man's Burden, 1899

In surmising the post-war history of development aid, we observe two tragedies. The first is frequently eulogised by political leaders: this is the fact that interventions such as anti-malarials and bed nets cost almost nothing and are still desperately needed in parts of the third world. The second is rarely mentioned: that the West has spent $2.3 trillion on overseas aid over the last five decades¹⁾ and yet these basic needs remain. If this second tragedy was more commonly discussed, then there might be a greater recognition that the problem lies more in the nature and modalities of aid than in a lack of resources. “This book is about that second tragedy.”²⁾

Easterly distinguishes two approaches to overseas aid, planning and searching. The essential difference is that planning is a top-down approach and searching a bottom-up. Planning starts with top-level political will and an overarching goal, for instance the Millennium Development Goals. It begins with an attractive political vision. It is a politician's response to the feeling that “something must be done”. Massive resources are mobilised, large-scale organisation is set up to allocate and disburse those resources. It is the application of utopian social engineering to the field of overseas aid. It typically fosters very weak forms of feedback (information struggles to battle its way back up the enormous hierarchy from poor rural recipient through half a dozen layers of bureaucracy back to the planner himself) and little or no accountability (the planner will not still be in office by the time the results of his intervention are in, and in any case, the realisation that a ten- or twenty-year plan has failed is uselessly blunt information as far as holding planners accountable is concerned). Easterly sees the theme of planning as uniting a coherent history of the West's approach to the third world, spanning centuries.

Searchers, on the other hand, operate from the ground up. Searching starts with observation of the problems and constraints facing poor people, and with experimentation with micro-level solutions whether they be organisational or technological. Searching is an individual's response to the feeling that “something must be done”, or rather, that “I must do something”. It begins with no resources, and seeks the minimum necessary resources as and when they are needed. It is the application of entrepreneurial initiative to the field of overseas aid. It integrates and supports existing or nascent solutions developed by poor people themselves. It is typically highly responsive to feedback and naturally holds those responsible to account, within a useful timeframe that makes clear the different levels of success of projects competing for resources. Easterly sees searching as essentially the same as the successful process not only of Western business but Western politics, in which 'political entrepreneurs' seek out policies that will win voters and enact them in a piecemeal fashion — rather than designing a grand utopian plan.

It is worth noting that, generally speaking, it is only possible to get away with planning when the planner is answerable to a population that isn't being planned. Thus Western voters select searchers to represent them, but elect planners to work on foreign economies, since planners can deliver broad, ambitious, superficial promises, which are much more attractive to Western voters than the piecemeal support of a hodgepodge of existing micro-schemes that essentially amount to no more than a continuation of existing processes.

There is a simplistic legend that has survived since at least 1950, which states that poor countries are locked into poverty because they do not have enough income above consumption to invest. The legend further claims that aid can finance that investment for a short period until takeoff, the moment when investment and accumulation have raised income to the point at which people can afford to save as well as consume, even after taking into account whatever level of uncertainty and income shock they are subjected to. The legend implies that a large volume of aid should be targeted at the poorest countries, and that this will lift them out of poverty. Jeffrey Sachs and the Millennium Challenge Corporation (MCC) are the latest advocates.

Unfortunately, this legend is at best not supported and at worst contradicted by available evidence. Firstly, comparisons of the growth of poor and rich countries do not reveal a level of poverty (on the country level) at which growth is low. Poor countries do have lower growth under some formulations, but this is because they have worse governments — when controls for quality of governance are added the difference between rich and poor disappears. Further, many countries wander up and down the ladder (countries are continually 'escaping' this poverty trap whilst others are returning to it). Researchers have repeatedly attempted to find a relationship between aid and growth, and although their occasional findings are widely publicised by the aid community, the successful study is usually subsequently repeated with a new dataset only to find that the positive relationship disappears.

This chapter is both complex in its treatment of various problems in specification and testing, and incomplete in its practice of saying no more than two sentences about any given study. But it is clear that available research has produced very little clear knowledge about the relationship between aid and growth. Research which is lined up in support of a new Big Push has been hand-picked from a messy and esoteric debate — whilst this research in itself is perfectly scientific, the cherry-picking only of studies that happen to support a particular conclusion and the disregarding of the rest is certainly not. When looked at in detail, instead of picking out a particular paragraph or two, often those few studies employed to support the Big Push agenda fail to do so for one or other technical reason.

Development economists are on much safer ground when assessing micro interventions, such as a new healthcare programme. By measuring specific target outcomes, using randomised selection and a control group, the assessment of individual activities is well within our power. One of the problems with the Planner's Big Push approach is that it is extremely difficult to appraise in a scientific way: it is an act of faith that a large increase in aid will lead to growth. Interventions that can easily be appraised in a scientific way ought to be favoured.

[F]ree markets work, but free-market reforms often don't. —p53

Whilst the market is fundamentally a good system for handling most production and exchange, imposing free market practices from above is surprisingly difficult. Markets can function badly for a wide range of reasons, and detailed knowledge of local conditions is required to navigate the various potential pitfalls — which simply cannot be done in the context of sweeping, large-scale reform from above. The emergence of a successful market can be seen as a gradual progression in which market participants seek locally viable solutions to a range of standard economic problems. In the typical case, a top-down effort to fast-track this process leads to the imposition of a set of formal rules that fail due to the planners' lack of understanding of local conditions, and actually make things worse by undermining a nascent but potentially successfully evolving informal system.

The free market depends on the bottom-up emergence of complex institutions and social norms that are difficult for outsiders to understand, much less change. —p54

Easterly briefly details the history of disastrous failure of structural reform programmes, pushed heartily by the IMF and World Bank since around 1980, in particular in the transition economies of Eastern Europe, as well as in Latin America and Africa.

He briefly reiterates the strength of the market as a means of enabling specialisation and exchange, domestically and internationally, but uses an example to explain the notion of dynamic comparative advantage.³⁾ Financial markets also play an important role in allocating investment efficiently, rather than forcing people who have money to invest it in their own activities, whether it is an efficient use of the money or not.

The most important explanation in the chapter runs through the bottom-up problems with markets, and gives examples of ways in which solutions evolve in ad hoc ways in different developing economies.

One of the most fundamental problems is that of trust. Easterly notes that social attitudes towards trusting strangers are strongly correlated with material wealth: it is harder to trust people in poor economies. But many solutions to the problem of trusting people in economic exchanges involve transaction costs that are too high for the small-scale transactions of developing economies: the courts, for instance, or 'Better Business Bureaus'. Transactions may simply not happen if the costs of making sure the other side honours the contract are higher than the value of the transaction itself. Courts are often unreliable in developing countries anyway, favouring the more financially powerful party in most disputes. Credit reporting agencies require formal information which doesn't exist (people need driver's licenses, titles to their land, phone numbers, and a paper trail in general).

The problem is particularly acute in non-instantaneous transactions, where payment is delayed for some reason, which may be inevitable if the work involved on one side is spread out but delivery happens in a discrete moment (because either the manufacturer incurs costs before he's been paid or the customer pays before he's received the goods — either solution requires trust from at least one side).

In many societies family businesses, and even racial castes emerge in a particular business or industry as trust-networks who share information about their members, enabling members to be 'ex-communicated' if they cheat on a contract. This can be seen as a 'multilateral punishment' strategy, which is far more effective than bilateral punishment but nevertheless remains informal.

One ethnic group is usually prominent in business in a poor society. In pre-industrial Europe, it was the Jews. In East Africa, it's the Indians… In West Africa, it is the Lebanese. In southern Africa, it is whites and Indians. Among indigenous African groups, often one dominates trading — the Bamileke in Cameroon, the Luba in the [DRC], the Hausa in West Africa, the Igbo in Nigeria, and the Serahule in the Gambia. In Southeast Asia, the overseas Chinese… play this role. —p73

In many environments, it is prohibitively costly to protect one's own property. An activity that would otherwise be productive may not be worthwhile if the final product would merely be stolen, and the cost of defending it may be so large as to make the project unprofitable. This is closely related to the cheating problem, and again there is evidence that in richer societies, people behave in a less predatory manner, and so even in the absence of formal systems to protect property rights, less investment is needed in defending property.

In developing economies, property rights can be defended informally through self-protection groups, whereby a group based on a family or tribe collectively defend their property from other groups. The Mafia is essentially a more powerful protection group, and illustrates the potential problems with all such solutions. Even when they are an effective solution to establish property rights, they are very difficult to dismantle when the economy reaches a point at which they are no longer efficient.

Elsewhere, authorities such as “warlords, clan leaders, semi-feudal landholders, tribal chiefs, and village headmen often dispense justice”.⁴⁾

This illustrates that bottom-up problem-solving doesn't always lead to 'attractive outcomes'. But formal Western systems of property rights evolved gradually through similar processes, and the most important feature of a formal system if it is to succeed is that it reflects and builds upon existing informal systems and values, rather than creating a different, parallel, competing system of property ownership.

When a formal system, such as land titling, is introduced from above without detailed knowledge of the systems already in place to recognise land rights, the result too often is that the two systems (the new formal and the old informal) compete with one another, during a period of uncertainty during which each system is enforced, by different means, to a greater or lesser degree. In many cases the informal system may get more usage out of the land by combining more subtle forms of land rights (for instance a mixture of collective ownership during part of the year with individual use during another part, or else one group of people having the right to one use of a piece of land whilst an individual or a different group can use it for something else), and replacing this with a simplistic concept of individual ownership may reduce the efficiency of land usage, since participants cannot afford the transaction costs of establishing such a complex contractual relationship (and enforcing it) through the nascent formal system, which they don't understand anyway. In many practical cases, the new formal system is largely ignored. It may be ignored for a period, permitting the two systems to become out of synch with one another, before state power will be called in to defend the formal against the informal, illegitimately ignoring the economic practice of the community. The uncertainty over land tenure introduced by the dual system may in itself create a disincentive to invest in property.

What looks like opportunistic behaviour could be the mingling of private property with traditional values, which place obligations to kin above those to strangers or banks. By imposing land titling in such complex social customs, “private property rights” may actually increase the insecurity of land tenure rather than decrease it. —p85

Land titling is a common feature of top-down solutions, and yet the transaction costs involved may simply be too large to be an efficient solution, particularly where appropriate informal systems of land ownership are already functioning.

As legal practitioner Wade Channell summarised the legal reform experience of Eastern Europe after 1990: “It is hard to imagine any rule of law aid specialist pursuing law reform in his or her own country in this fashion. If I assembled half a dozen recognised European or US specialists to redraft the US Code of Judicial Ethics and then tried to get it passed by the US Congress with little or no input on the proposed draft from congressional committees, the judiciary, the bar, business interest, law schools, or other stakeholders, I would be looking for a new career rather quickly. Based on many current practices, however, that career could easily be found abroad 'helping' transition countries with the same process.”
—p83, citing Wade Channell, 2005, Lessons Not Learned: Problems with Western Aid for Law Reform in Postcommunist Countries, Carnegie Endowment for International Peace, Democracy and Rule of Law Project, No 57, May 2005, p6.

Formal Western systems of private property and the legal system of enforcement of that private property are good systems that work well, but they are good systems that work well because they evolved slowly in a piecemeal way from existing practice, and were not imposed from above. In this connection, Easterly comes down firmly in favour of the English common law system relative to the Continental civil law system for its ability to adapt and take into consideration new cases.

As the great American jurist Oliver Wendell Holmes said, “It is the merit of the common law that it decides the case first and determines the principles afterwards.” —p85

He notes also that the countries that created the civil law system, France and Spain, apply it far less rigidly than the countries that inherited from them (the French and Spanish ex-colonies), to the latter's detriment.

This chapter, on the role of 'democracy' or 'good governance' more generally, in many ways closely mirrors the previous one on free markets. Just as the previous chapter concluded that free markets were good, but free market reforms usually failed, so this chapter argues that democracy is good, but that attempts to impose or encourage better governance are riddled with intractable problems and, according to historical evidence, fail.

One gut instinct that many people have about the poverty of nations is probably close to the mark: it's all politics. —p101

Many countries have long-standing, deep-routed political problems, and so far aid agencies' attempts to address them (or, more usually, to ignore them completely whilst attempting to implement generic, off-the-shelf policies) demonstrate a dangerous lack of awareness of local context and doom their interventions to failure. It is important to have an understanding of the most common political problems, and the humility to appreciate that centuries-old divides cannot be quickly remedied.

Planners opt for one of two camps on bad government. One camp (associated with the US government, the WB and IMF) says that poor-country governments are awful and the West should get tough with the bad governments — force them to change in return for aid. The other camp (associated with the UN and Jeffrey Sachs) says that poor-country governments are not so bad and that countries should be free to determine their own development strategies. However, this artificially restricts the debate. It may be true that poor-country governments are bad, and it may be just as true that Western attempts to change them have been fruitless. Continuing my subliminal quest for the most politically unappealing truths, this chapter considers what to do if both statements are true. —p101

Good governance can address many of the market failures discussed in the previous chapter. Impartial police forces and courts can support contracts, efficient spending on infrastructure and education can foster growth. Research offers strong support that this reasoning is borne out by experience. But there are many reasons that democracies may be difficult to establish quickly:

1. Cheating. As with markets, what is thought of as 'democracy' is an intricate set of rules and customs that are slowly developed from the bottom up to prevent cheating, ensure fair play and provide adequate 'checks and balances' against misuse. Cheating covers election-fixing, bribery of the courts or police, intimidation of voters and theft from the treasury.
2. The democratic paradox makes the distinction between democracy and mere majority rule, arguing that true democracy must also incorporate protection for minority rights, dissent, political protest and so on. Without these rights there may be the temptation for the majority to vote property from the minority or even to disenfranchise that minority. In practice, this is a particular problem in an ethnically heterogeneous society, or one with a small elite with disproportionate wealth. In such situations, there is an incentive for government to use demagoguery, for instance, to draw attention away from its own theft or inaction.
3. Democracy and oligarchy. The transition from oligarchy to greater political participation is fraught with difficulties. From an economic standpoint, oligarchy may provide for better economic growth than democracy in the short or medium term, although in the longer term it tends to be less able to react to changing economic trends and shift production to take advantage of new opportunities. Oligarchic power can be particularly robust when an economy is based on a large resource endowment, which provides the oligarchy with a strong incentive to maintain power and easy access to the resources necessary to do so.⁵⁾ In many cases, a transition to a more participatory system involves a tacit negotiation between the oligarchy and poorer classes, in which greater political participation is granted in order to assuage calls for revolution. Where this negotiation breaks down, a convoluted and mutually destructive series of revolutions, coups and countercoups ensues. The transition is likely to be smoother in industrial economies (where the oligarchy finds it hard to appropriate wealth and to control the increasingly skilled urban population) and in cases in which the oligarchy manages to secure strong guarantees of its property rights, as was the case in the UK and US.⁶⁾

There are good governments in the tropics, and there are economic success stories based on them. Botswana is amongst the best examples: despite having no natural resources on independence, but with the assistance of a large flow of development aid (averaging 16 per cent of income), long-term growth was strong, even after a discovery of diamonds, which was well-managed by government.

However, this is not the trend. Corruption, on average across the African continent, is not improving, unlike many more tangible development goals. And no less disturbing, there is more evidence to suggest that large aid inflows make governments worse than better. The IFIs have a terrible history of providing enormous amounts of money to all but the very worst governments, with no tangible political or economic improvement even over decades.

Who got the most standbys [a type of short-term loan] from the IMF over the last half century? The answer is Haiti, with twenty-two. And not just Haiti, but the Duvalier family (Papa Doc and Baby Doc), under whom Haiti got twenty of the twenty two standbys from 1957 to 1986. —p130

Donors, particularly large donors, deal exclusively or almost exclusively with governments. Diplomacy (and perhaps the need to appear to be succeeding) stifles honest criticism. The typical statement by donors is that the situation is “poor but improving”.

There has been a recent flurry of interest within the IFIs towards “stakeholder participation” in their schemes, particularly in the evolution from 'structural adjustment programmes' to their successors, 'poverty reduction strategy papers' (PRSPs). IFIs insist that the latter are prepared with extensive “stakeholder consultation” in which civil society is consulted. Unfortunately, the entire exercise is decisively undermined by the next step: if the PRSP conforms closely to what the IFIs would have written in a SAP it is accepted; if not, it is rejected. Naturally, the result is that if governments want the money that is on offer, they have to make sure that any stakeholder consultation they do somehow comes up with all of the “right answers”.

An even more recent fad has been the concept of 'peer review,' in which African governments enforce standards of governance and reform on one another. Not only is this a typical example of a scheme that could never, ever be acceptable to the donor agencies' own governments (imagine the French government reviewing the British), it also misses the basic idea that democracy means being answerable to one's own people, not to foreign governments.

Overall, donors take almost no notice of how bad a government is in determining how much aid to provide — and they should.

The governments rated as among the worst tenth among developing countries in terms of corruption spent an average of 20 per cent of the time in IMF agreements, which is significantly less than the average of 41 per cent of the time for the rest of the sample…So the IMF does show some willingness to lend less to the most awful governments. Unfortunately, once governments get out of the worst tenth, there is no further tendency to penalise bad governments. For example, governments rated among the second-worst tenth on democracy and corruption are no less likely to spend time in IMF programmes than the rest of the sample. —p134–5

Bad governments should be screened out, and not offered such assistance, which merely strengthens them and helps to support them in power anyway — and certainly won't reach the poor. Instead, in countries with particularly poor governance, experiments should begin as to how aid can be provided through programmes that are not channelled through government.

This chapter's reading of the evidence suggests dropping the obsession with always working through the government. However, let's hold fast to opposing shock therapy and universal blueprints, even for the country of foreign aid. Any of these changes should be tried in a gradual, piecemeal, experimental way, and the answers will be different in different countries and different sectors…

The principle is non-intervention. Don't reward bad governments by working through them, but don't try to boss them around or overthrow them either. The status quo of both donors and gangsters badly needs some work. —p137–8

The extent to which aid agencies respond to the needs of the poor depends on how much the poor can provide information to, and control the actions of, aid agencies. In market systems, suppliers can only survive by finding what the consumer wants and by satisfying that need. In well-functioning democratic systems, the same rules apply: voters can apply pressure on political representatives, who in turn can exert power over bureaucracies to deliver the services that the voters need. Aid does not work like this. Many of the reasons that it fails to deliver what the poor need can be explained in terms of economic principal-agent analysis (which considers a situation in which a 'principal' employs an 'agent' to do something on the principal's behalf, in particular looking at what can go wrong if the principal and agent have differing interests and preferences, as they typically do). Some principal-agent problems in aid:

• The wrong principal controls the agent. This is almost ubiquitous: aid agencies are primarily controlled by political representatives in rich countries, and indirectly by rich voters, rather than the poor constituents that they claim to serve. To make matters worse, they have to work closely with (and to some extent satisfy) a recipient country government who may have an entirely different set of interests to both the poor and to political representatives of the rich.
• Partly as a consequence, the agent's success is unobservable, at least to rich politicians and voters. They are aware of the plans and goals of aid agencies, and to some extent of conferences and reports that are covered by the press, but it is very difficult for them to get good information about how big an impact a given aid agency is having on the lives of the poor.
• The principal has broad, vague objectives which resist measurement and make it much more difficult to observe whether any individual agent is being successful. For instance, it would be much easier to determine whether an agent (aid agency) was successful if its mandate was to 'build and maintain adequate roads in northern Tanzania' than if its goals are to 'reduce global poverty and increase pro-poor economic growth'.
• The principal has multiple objectives and most aid agencies do not specialise at all in which they address. As Easterly illustrates in detail, the actual situation is much more chaotic than described in the previous paragraph, with individual agencies publishing reports crammed with literally hundreds of goals and subgoals across all conceivable areas of human endeavour.
• This means that agents' goals overlap; many agents work on each goal, with the result that if a goal is not met it is unclear who failed, and it is always possible for everybody to blame it on a 'coordination failure' which ensures that nobody is any more to blame than anybody else.
• Some agents have many principals, especially the multilateral institutions, and to an even more extreme extent the UN institutions, who are ultimately controlled by an extremely disparate collection of 191 countries. The basic problem with having multiple principals is that any failure to satisfy one principal can be blamed on efforts to satisfy another principal.

Nevertheless, it is important to appreciate that despite these problems, some aid agencies do successfully meet worthwhile goals, and also to note that the division between the successes and failures broadly support this explanation for why aid so often fails.

Aid efforts have generally been more successful in areas with clear, measurable outcomes, such as immunisation programmes (where immunisation and child mortality rates can be measured easily), school enrolment and provision of water and sanitation — as opposed to, say, attempts to reduce 'corruption'. Programmes are also more successful where there exist specialised international organisations rather than generalists, for example the World Health Organisation (WHO) has a clear delineated mandate, and to some extent the IMF is also relatively successful within its narrow sphere of operations.

A particular problem of the unobservability of aid outcomes is that the donor community has tended to focus on inputs as a proxy, which is disastrous for determining the efficiency and effectiveness of aid, because performance seems to 'improve' the more inputs an agency is able to successfully waste. It is also striking that for at least the last century almost every commentator has demanded that aid flows be doubled as soon as possible.

“an increase in the per capita national incomes cannot be brought about without… a sum of money… of about $3 billion a year” —UN Expert Group, 1951

“the current flow of ODA… is only half the modest target prescribed by the internationally accepted United Nations Strategy” —Robert McNamara, World Bank President, 1973

“If we are serious about… meeting multilateral development goals we have all signed on to, we must double ODA from its current level of about $50 billion a year.” —James Wolfensohn, World Bank President, 2001⁷⁾

It is also striking to note the consistency with which aid agencies have been promoting donor coordination as desirable since the 1950s. Basic analysis of the principal-agent problem shows that where aid agencies act together, collective responsibility ensures that none of them will be held to account for their joint failures. Where a single agency acts alone on a programme, then that agency can be held to account for the outcome, and if agencies were to specialise to the extent of a single agency taking exclusive responsibility for a sector, then it would be far clearer which agencies were failing.

Another effect of agencies' need to conduct high-visibility projects that appear to Western politicians to be effective is the preference for new-build projects rather than routine maintenance. This is justified by agencies on the grounds of 'sustainability' — the project will only be sustainable if the government takes over maintenance once the road or the school is built. Unfortunately, decades of history prove that this handover to government almost invariably fails. In effect, attempts to make projects 'sustainable' usually guarantee that they will not be sustained.

Here is one way to make aid work better: aid donors should just bite the bullet and permanently fund road maintenance, textbooks, drugs for clinics, and other operating costs of development projects. Politically dysfunctional governments that don't do maintenance can concentrate on other things. —p167

Another result of the problem that the 'wrong principal' directs aid agencies work is 'tied aid', in which rich country donors require that some part of development aid be spent on its own countries' exports. This lowers the value of aid to the recipient because it limits what it can be spent on and from whom. The US is the worst offender by this metric, requiring that aid be spent on products from American firms for about three quarters of its aid. Technical assistance is usually even worse, as many rich countries insist that their own nationals be employed as technical experts, which also ensures that a large portion of money devoted to technical assistance programmes never leave the country that pays for them.

Aid agencies are also attentive to the need to reward political allies of the rich countries with aid. The frequency with which a recipient country votes with the donor in the UN, and whether the recipient is an ex-colony of the donor, affects how much aid that country gets. After September 11, 2001, agencies gave new aid to allies in the war on terror, such as Central Asia, Pakistan, and Turkey. —p169

Examining different evaluation mechanisms is one way to address this problem. The IMF and World Bank both have units which evaluate their own agencies' programmes, which is a step in the right direction, but the independence of these units is questionable. They, and others, need truly independent randomised evaluation of their projects.

The currently fashionable focus on 'participation' and 'consultation of stakeholders' in planning process is depressingly farcical. Global plans such as the Millennium Development Goals (MDGs) are formulated and translated by an international bureaucracy to regional and national-level planning. Somewhere at the end is an insistence that stakeholders must be consulted and the entire programme must be nationally owned, with a total incapacity to understand what the idea of a recipient-driven aid effort might mean.

There is notable variation amongst aid agencies. The US aid agency USAID

brazenly states its objective is to further “the foreign policy goals of the United States.” —p176

DfID (the UK agency) is notably more committed to independent evaluation of its aid projects than most other aid agencies. At the other extreme, the UN system contains much of the most absurd conference-, report- and meeting-driven time-wasting, and various examples of their nonsense are cited.⁸⁾

To be fair, there is incomprehensible language [like that generated by various UN agencies on a continual basis] also in private-sector documents, such as investment prospectuses or engineering designs. The difference is that in private-sector documents, the jargon actually has some meaning to specialists. In UN documents, the jargon has no substantive content for anyone. —p176

The IMF is the most powerful Western agency for “dealing” with many poor countries. It makes short-term loans (to be repaid within two to four years) to countries facing fiscal crises, and negotiates changes in government policy designed to increase the government's ability to repay (such as tax increases and spending cuts).

The IMF has had some significant successes. Easterly cites South Korea and Thailand in the 1980s, Mexico in 1994-5 and the East Asian financial crisis in 1997-8. In addition, the research department provides high-quality briefing documents to developing country governments.

However, IMF attempts to make adjustments in poorer economies have had little success. IMF assistance is only accepted due to the typical (and understandable) myopia of governments faced by urgent financial crises.⁹⁾

The IMF assists by providing foreign currency (US dollars) to insolvent or near-insolvent governments. Governments need this foreign exchange to provide in exchange for local currency. When a government spends more than it earns, then either:

• people choose to hold this extra money (they would not be expected to unless something structural has changed, such as economic growth),
• prices rise (as this extra money chases the same quantity of goods within the domestic economy), or
• people attempt to exchange the extra currency for foreign exchange (for instance so as to buy imported goods).

The currency provided by the IMF is designed to enable to government to respond to the last request, in order to prevent some form of banking crisis. Inflation is undesirable in itself, but if sustained it also provides an incentive for people to exchange local currency (which is losing value) for foreign currency (which is not).

Although there is a certain logic in the basic idea of the IMF getting involved with 'improving' government policy so as to improve the likelihood that its loans will be repaid, in practice the IMF's interference is inappropriate and ineffective. The conditionality it attaches to loans are extremely detailed and extensive, but the IMF often has a low awareness of domestic political pressures. The result is the 'IMF riot', by now a common feature of intervention by the institution. It is often sparked by some particularly unpopular policy that the IMF has forced the government to adopt without an understanding of the likely consequences, such as the removal of subsidies on fuel or basic foodstuffs.

In the first nine months of 2000 alone, there were demonstrations against IMF programmes in Argentina, Bolivia, Brazil, Colombia, Costa Rica, Honduras, Kenya, Malawi, Nigeria and Zambia [in addition to repeated rioting and coup attempts in Ecuador]. —p191-2

But there are more serious consequences of IMF intervention than mere rioting. Examining all eight worldwide cases of “state failure of collapse”, Easterly finds that the IMF had been heavily involved in seven of those countries in the decade preceding the collapse.

Statistically, spending a lot of time under an IMF programme is associated with a higher risk of state collapse… In retrospect, it would have been better if the IMF were not involved at all in these cases. —p192

Another major reason that the IMF shouldn't get too closely involved in negotiating government policy is that it has terrible information on what is happening in country. The IMF relies on statistics that are commonly highly inaccurate (there are often discrepancies of 100 per cent merely between two simultaneous IMF measures of the same variable), and the models the IMF tries to employ are influenced by a large number of variables that the IMF is unable to measure at all.

In March 2003, IMF staff put Mali's GDP growth in 2001 at 1.5 per cent. By August 2003, it had raised the 2001 number to 3.5 per cent. Just five months later, in January 2004, IMF staff now put Malian growth in 2001 at 13.3 per cent! This is not to say the IMF is incompetent at statistics; it is just that any statistics are very shaky in very poor countries. —p196

Setting targets for these measures in the face of this inaccuracy is often farcical. Despite the inaccuracy of available data, governments must act within the real constraints — but it is impossible to predict where these constraints lie with existing data. In such cases micromanagement is futile, and the government must be trusted to do whatever it sees fit in order to sustain the key outcomes that it relies on for solvency (such as foreign currency reserves and inflation).

The statistics show that failure to repay IMF loans does not make it more difficult for debtor governments to obtain further loans from the IMF. Nor do 'bad government policies' make further loans less likely.¹⁰⁾ In more extreme cases, the World Bank is often drafted in to provide additional loans to help governments repay previous loans. The IMF has a certain incentive to act in this way — whilst it may not protect the IMF's principal in the long run, it does prevent it having to admit mistakes in the short term.

This repeated-lending cycle reached an unsustainable crisis during the 1990s, at which point many of the poorest developing countries had clearly reached a point at which their debts were unserviceable. After 1996, the IMF's response was to label certain countries 'Heavily Indebted Poor Countries' (or HIPCs). Countries qualified largely by agreeing to much the same conditionalities as they had previously as part of structural adjustment programmes (SAPs). The HIPC initiative was supposed to be a once-and-for-all solution that came with optimistic projections for buoyant growth, which again largely failed to materialise for much the same reasons that the original loans had failed — necessitating a second round of forgiveness in 1999.

This is [the] general pattern: the growth in programme countries fell short of the IMF's own targets. On average for IMF programmes in the 1990s, the target GDP growth was 4 per cent, but actual growth was only 2 per cent. Since population growth was also about 2 per cent, this meant that the actual growth of income per person was close to zero. —p204

The 1999 cancellation, in turn, proved itself to be insufficient and a further $40 billion of debt relief, representing 100 per cent of the debt of 18 countries, followed in 2005. By now, forgiveness has reached a point at which loans to low-income countries are no longer at all credible. For this reason, it makes sense to provide finance only in the form of grants.

The IMF's role as a provider of short-term finance to economies facing crises, particularly in emerging markets, is a valuable one, and the global economy needs an institution to fulfil this function. However, the role it has invented for itself since 1980 of providing development loans, particularly to the poorest countries, is ill-conceived, has failed miserably, and should be abandoned. This practice is becoming ever more absurd, with the IMF's new fascination with getting popular endorsement for its programmes from the populations of partner countries — because this 'local design' of programmes has not changed its insistence on its own exhaustive list of conditionalities for a loan to be approved. Where the outcomes of popular consultations suggest designs that don't incorporate traditional IMF conditionalities, loans are not approved.

We will tell you what to do, as well as promise you that you are doing it of your own free will. —p206

The IMF must:

1. simplify its dealings with poor countries so as to reduce its intrusion in matters of domestic politics;
2. withdraw from the poorest countries, where its role is simply inappropriate — the IMF should leave these countries to the aid agencies and concentrate in developing markets;
3. get rid of conditionality: the only important criterion is whether a country repays the loan (and if it does not, the IMF can sanction it by withholding future loans — since the IMF has an agreement with its debtors that it is always paid first, then the IMF has a lot of leverage in keeping other creditors out if it refuses to lend (compare with the multilateral punishment strategy discussed in Chapter 3)).

Ultimately, the IMF needs to return to its original mandate of financial stabilisation and abandon its failed attempts as a development partner.

Health interventions in development are normally relatively successful. The outcomes are easy to measure: levels of infection and mortality. The goal of saving lives is politically uncontroversial both domestically and in the West. Often, individual agencies will assume responsibility for a particular threat within a particular territory, providing ideal conditions for strong accountability for success.

These favourable conditions have translated into successes such as: the virtual eradication of measles in southern Africa between 1996 and 2000 through immunisation and vaccine; radical reduction in diarrhoea-related deaths in Egypt in the 1980s through oral rehydration therapy; the virtual elimination of polio from Latin America from 1985; drastic reduction in the leading cause of blindness, trachoma, in Morocco from 1997; drastic reductions in maternal death in Sri Lanka; reductions in TB in China during the 1990s; the elimination of river blindness in West Africa from 1974; etc, etc.

It is because of these many successes that the catastrophic failures in aid agencies' attempts to tackle the HIV/AIDS pandemic, both historic and present, are particularly galling. By 1983 many voices had already identified the African risks if serious action was not taken; by the late 1980s there was a strong consensus among experts both that drastic action was urgently required, and in projecting the scale of the consequences if nothing was done. Nevertheless, the most minimal interventions were made by all aid donors until the issue became popular in the Western media in the late 1990s, with the number of infections reaching approximately 40 million by the turn of the century.

The first tragedy was the failure of aid agencies to take swift action. The reason for inaction in the late 80s and early 90s had a few simple causes. Firstly, no single agency took responsibility for the consequences of the pandemic. Many aid agencies could have acted to avert the crisis, and a sincere campaign by any one of them would have been sufficient to seriously mitigate the consequences. However, action on this scale would have been 'costly' for any single agency, because the threat, although well understood by all health experts, was not a popular cause in the Western media. Easterly proposes a model: when one agency acts, or one department of an agency acts, then they alone bear the costs of action whereas all agencies or departments share the accolades, since it is difficult to observe who was most effective in achieving those results (and everybody was at least pretending to do something). In this model, when only one agency is in a position to act it will, but as the number of agencies with responsibility for this field rises, the probability of action by any one agency decreases.¹¹⁾

The second tragedy is that aid agencies' current strategy towards fighting both the virus, as well as health threats more broadly, is grossly inefficient. One of the basic principles of good health management is that limited resources — and resources available to combat health problems are always limited, as much so in the West as in Africa — should be spent on the cheapest available means of saving and extending life. This is categorically not the approach that is currently being taken.

In the context of fighting HIV/AIDS, there has been a sustained lack of research into the relative cost-effectiveness of different approaches to prevention and cure. Nevertheless, it is now clear that preventative measures such as education, condom distribution, targeting of particularly high-risk groups such as prostitutes and simple drug interventions such as a one-off dose of nevirapine during labour to prevent transmission to newborns, are many (often hundreds) of times more cost-effective than adult treatment with antiretrovirals.

The…WHO report…states that money spent on educating prostitutes saves between one thousand and one hundred times more lives than the same amount of money spent on antiretroviral treatment. —p227

Looking at the context more broadly, there are a large variety of health interventions in other diseases that are just as fatal as HIV/AIDS that also cost far less than antiretroviral treatment, that are currently being neglected in favour of antiretrovirals.

Bush's cut in other health spending [of $100 million when announcing a $15 billion intervention against AIDS] was particularly unfortunate when two and a half times as many Africans die from other preventable diseases than die from AIDS…include[ing] malaria, [TB], diarrhoea and others. Worldwide, in 2002 there were 15.6 million deaths from these causes, as opposed to 2.8 million deaths from AIDS. —p222

At the time of writing, first-line antiretrovirals cost approximately US$304 per year, having rapidly reduced in price from 30 times that due to competition from generics pharmaceutical firms (under exemptions from TRIPS that have now expired). However, treatment is notoriously complex and difficult to administer, even in the West. Drug administration needs close observation by well-trained medical staff to ensure that the patient conforms to the regimen (otherwise the virus will generate tolerance to the drugs and resistant strains will quickly develop — even in the West often 20-40 per cent of patients fail to conform to their regimen) and also to make subtle changes to the cocktail of drugs to combat any extreme side-effects. This medical support costs in the region of US$1200 per patient per year and unlike the cost of drugs there is little scope for this to ever be reduced. Moreover, first-line antiretrovirals are typically effective for only 3-5 years before the patient develops full-blown AIDS and requires drugs that are too expensive to be considered in the African context (and will not be subject to the same competition from generics manufacturers due to the expiry of the TRIPS waiver). In Brazil, this average period is as low as 14 months. In comparison, many interventions against threats such as measles and malaria require drugs that cost only a few dollars administered once or infrequently that can be managed by low-skilled programme officers with minimal health training. Current health aid is insufficient to cover all of these costs, despite recent increases associated with the AIDS pandemic.

The medicines that cure TB cost about ten dollars per case… Worldwide, three million people die a year because they are not fully vaccinated [against diseases for which vaccines exist], even though vaccines cost only pennies per dose… A full course of treatment for a child suffering even from drug-resistant malaria costs only about one dollar… Vietnam, a relatively poor country, reduced deaths from malaria by 97 per cent from 1991-97 with a campaign that included bed nets and antimalarial drugs. —p223

There are a few additional annoyances particular to the US, such as conditions imposed on American aid by the far right: that only a fifth of its spending go to prevention, and that of this one-third be devoted to abstinence-only programmes (which have been shown to have no effect other than to reduce condom use by US research). They have essentially prevented any of the money being spent on programmes that target prostitutes, so as to avoid promoting sinful sexual behaviour.

This short-sighted focus on treatment rather than cure has been supported by a disparate coalition of groups in the West, ranging from the American Christian right (who refuse to subsidise condoms, work with prostitutes or generally discuss sex in any detail) through to activist groups such as ACT UP. Easterly suggests that the campaign to provide treatment plays particularly well in the Western media because the story even has a villain — the drug companies that tried so hard to prevent the price of treatment falling (and have been much more successful with more recently developed drugs).

It is the job of health economists to research the cost of intervening against different health threats and calculate the most cost-effective use of available funds, and in the context of the AIDS crisis they have entirely failed in this duty.

We do know that Colonel Qaddafi has been and will continue to be a destabilising force in the region, so nothing would surprise us, and we do know that Sudan is…Sudan is…Sudan is…one of those countries in that region of Africa.
—President Reagan, explaining American support for the northern Sudanese government despite its subjugation of the south in a press conference, cited p266

This is a long and detailed chapter, summarising various histories of African and Asian colonies during colonisation, colonial rule and decolonisation. However, brushing over this evidence, the essential argument of the chapter is straightforward.

Parts of the American elite are increasingly viewing new imperialist projects favourably, after several decades during which imperialism had universally been considered exploitative and malign. In 2004, the US State Department admitted that it had prepared detailed plans for the reconstruction of up to 25 states which were not yet in conflict.

The [US Office of the Coordinator for Reconstruction and Stabilisation's] mandate is not to rebuild any old states…but to create “democratic and market-oriented” ones…Sometimes rebuilding, [office head Carlos Pascual] explained, means “tearing the old apart.”
—Naomi Klein in The Nation, 2nd May 2005, cited p238

These sections of the US elite believe that, in countries where economic development and good governance have failed to take root, it would be possible for the US military to take control of the territory, and then for a collaboration of the military and civilian reconstruction teams (supported by USAID, the IMF and WB) to establish new democratic institutions that will bring peace, democracy and economic prosperity under a US-style market system. Outside the government, this vision is supported by the likes of Niall Fergusson, a Harvard historian who states that there is

such a thing as liberal imperialism and that on balance it was a good thing…in many cases of economic 'backwardness,' a liberal empire can do better than a nation-state. —cited on p239

However, this new intellectual fashion is wrong — European colonialism did not encourage economic development, nor good governance. This was due not only to the material self-interest of the colonisers (as is often argued today by those calling for a new, benevolent, imperialism) — “there were humanitarian instincts at work during colonialism similar to those in today's nation-building (just as there are some self-interested objectives today).”¹²⁾ Instead, much of the failure of colonialism came down to the incompetence of the planners who implemented it, which was inevitable due to their approach of designing colonial systems centrally without meticulous regard for local conditions. This is implicitly the approach advocated by the new imperial elite today. The results will be similar to those of classical imperialism.

The main problems with imperialism are summarised as follows:

1. Local power structures were poorly understood by the colonisers who lacked the resources or interest to learn about them. With interest in ensuring obedience to the colonial power rather than good governance, the European powers generally stripped away existing checks and balances on the exercise of despotic power in Africa, establishing the autocratic rule of blacks by blacks that in many areas has continued until today.
2. Many interventions aimed at economic development (especially in agriculture) were arrogant and ignorant, ignoring accumulated local knowledge to establish poorly suited crops which failed, or with an undue preference for cash crops over food crops.
3. Decolonisation was handled with similar incompetence, in particular the rapid establishment of arbitrary nation-states without the human resources to be sustainable were hampered further by three main mistakes:
   1. “the West gave territory to one group that a different group already believed in possessed” (thinking particularly of Palestine);
   2. “the West drew boundary lines splitting an ethnic group into two or more parts across nations;” (for example the Pathans¹³⁾, divided between Afghanistan and Pakistan);
   3. “the West combined into a single nation two or more groups that were historical enemies”¹⁴⁾ (for example the enmity between groups in northern and southern Sudan).

Two original statistical tests are also provided to back up these conclusions. The first is a comparison of economic development between countries that had never been colonised ('non-colonies') and 'non-settled colonies' — European colonies in which migrants from Europe had not largely displaced indigenous peoples (so this would exclude the US, Canada, Australia, etc).

The non-colonies had more rapid increases in secondary education from 1960 to 2001. Growth per capita from 1950 to 2001 was 1.7 [per cent] higher in the non-colonies than the non-settlement colonies, a huge difference for a fifty-one year period. By 2001, income was 2.4 times higher in the non-colonies than in the former non-settlement colonies. —p250

No attempt is made to control for any characteristics of countries that might have made them more likely to become colonies in the first place, although it is plausible that such characteristics might exist. The non-colonies varied widely in their economic growth (highly successful countries like Japan, Thailand and China are included in this group). Also, it is conceded that the only African non-colony in the sample, Ethiopia, fared similarly to the non-settled colonies, so it would appear from the detail given here that the difference between the two groups could, for instance, also be explained as the difference between Africa and Asia.

The second analysis attempts to show that “countries with artificial borders do worse on economic development.”¹⁵⁾ Two measures of border-artificiality are used:

1. the percentage of the population that belongs to ethnic groups split between countries (whilst controlling for ethnic heterogeneity per se); and more exotically
2. “a mathematical measure of how wiggly or straight”¹⁶⁾ the borders of each country are, based on the assumption that poorly considered borders designed by bureaucrats are more likely to be straight lines than those that arise organically.

One or both measures of border-artificiality were shown to be associated with worse performance in democracy, government service delivery, rule of law, corruption, infant mortality, literacy, childhood immunisation and access to clean water in the present.

Western intervention in the government of the Rest, whether during colonisation or decolonisation, has been on the far side of unhelpful. The West should learn from its colonial history when it indulges neo-imperialist fantasies. They didn't work before and they won't work now. —p268

This chapter stands out from the general tone of the book in several ways. The arguments are almost entirely political rather than economic.^{17) 18)} The assumed audience appears to be the centre or centre-right of the American media; consequently assertions unusual in that context are argued with care and at length whereas other quite remarkable and unfounded statements are made without any attempt at justification. Often these would strike a British or European audience as lazy or absurd.¹⁹⁾ Quite generally, evidence and argument is patchier and open to alternative explanations than elsewhere in the book.²⁰⁾ These factors make it more difficult to strike an appropriate balance between recounting Easterly's arguments and drawing attention to their flaws or providing counterarguments. As the most politically tenuous area of the book, it would be one of the best to read in its original form and analyse for oneself, as one of the hardest to fairly summarise and critique here.

Easterly sees military interventions of the type carried out by the US, particularly since World War II (but also prior) as similar in nature (but even more extreme) to failed interventions by aid agencies:

[M]ilitary intervention is too perfect an example of what this book argues you should not do — have the West operate on other societies with virtually no feedback or accountability. The military is even more insulated from the interests of the poor than aid agencies are. —p274 (emphasis original)

Further, interventionists ignore local conditions and apply the same confused approach as aid agencies to “dealing with gangsters”. The same collective responsibility problems apply, since Western countries can blame one another and the UN for failures, whilst the UN blames member countries on the few occasions that it is diplomatically possible to do so — and none of these actors bear any cost for mistakes.

Most of the analysis in this chapter focuses on the Cold War era. Easterly contends that the Cold War did not substantially differ from the periods before or afterwards.

Today advocates of Western military intervention see it as trying to introduce democratic capitalism. In the Cold War, by contrast, the Americans tried to convince third world nations that a better system than communism was…democratic capitalism. In the bad old days of the Cold War, Americans embraced some dictators as allies. In today's war on terror, the Americans embrace some dictators as allies. The various military interventions of the US even involve some of the same people… —p275

He chooses to examine Cold War conflicts because more information is available about their long-term consequences, compared with more recent interventions.

Easterly begins with a statistical analysis of a number of the conflicts he has in mind: Vietnam, Cambodia, support for the mujahideen in Afghanistan, Guatemala, Korea, Iran, Liberia, Ethiopia, Somalia, El Salvador, Nicaragua, Zaire and Angola. He finds that in 2004, the typical nation involved in these conflicts was in the bottom 15 per cent on democracy, bottom 18 per cent on the rule of law, and the bottom 22 per cent on economic freedom. He then admits that the selection bias involved renders this finding meaningless, but then appeals:

[R]emember, Americans say we won the Cold War. Whose victory is it when most of the poor countries where (and allegedly, on whose behalf) the Americans fought the Cold War are still in such bad shape? —p276

He then turns to more detailed histories of Nicaragua and Angola, highlighting in detail the reasons he believes that American intervention hampered good governance and economic stability, then blithely asserts that “the other side” were, of course, just as bad — but we can't really do anything about that and shouldn't try.

The continuity of this policy — and its harmful results — is illustrated with reference to:

1. pre-war interventions in Latin America and the Caribbean, with particular focus on Haiti, and
2. the failure of post-Cold War UN “interventions” in Bosnia, Somalia and Rwanda (it may be difficult to see how Rwanda can be characterised as a case of 'too much intervention', but the argument shifts here to conclude that “the peacekeeping system…could not get right the decisions about whether and how to intervene”).²¹⁾

In conclusion, wars should not be prevented by outside intervention, because a stable peace can only be enforced once one side has developed a decisive military advantage, which provides the capability to project enough power to sustain a lasting peace. Where foreign forces such as the UN prevent hostilities before this happens, then the country will be paralysed by peace negotiations that are doomed to fail since it will be impossible to convince all factions that they could not achieve more by force than through talks. Quantitative evidence is presented:

UN interventions produced a stable peace [10 years without the resumption of war] only a quarter of the time. With no UN intervention, a stable peace resulted nearly half of the time. —p293

(Again, the potential selection bias that the UN might be more likely to intervene in particularly intractable or long-standing wars, or those that appear unlikely to resolve themselves rapidly, is not mentioned.)

Next it is claimed that logically, anybody that favours any humanitarian intervention must favour infinite intervention, “given the ubiquity of human rights violations” — in other words, it is impossible to determine that one humanitarian crisis is more acute than another.

Finally, Easterly may be trying to back down from this position partially, suggesting that “rescuing innocent civilians from murderous attacks” is in some way an entirely different proposition to humanitarian intervention, implying that it may be both viable and beneficial. He offers no clue how to distinguish the two different cases — recall that he classes the Rwandan genocide as a situation in which intervention was inevitably doomed to failure and should not have been attempted.

This chapter briefly outlines the 'success stories' of Japan, China, Hong Kong, Singapore, India, Turkey, Botswana and Chile and claims in each case that this economic success came from local 'Searchers'. They sought solutions to their economic challenges from abroad at times, but imported only those ideas that were appropriate and often localised them. Western assistance in each of these cases had no impact; in cases such as Hong Kong, Singapore and Botswana that involved European colonisation, the countries were fortunate that the coloniser was either unable or too lazy to have its usual, malign impact. India suffered under the British empire, but after it had gotten over a couple of decades of interventionist ideology that Britain left behind, it managed to successfully move on and achieve independent economic success on its own terms. There is no mention of US intervention in Chile, even though the Allende-Pinochet transition is explicitly discussed — presumably it can't have been proper interference in this instance, since economic growth followed.

It is easier to search for solutions to your own problems than to those of others. —p303

The great bulk of development success in the Rest comes from self-reliant, exploratory efforts, and the borrowing of ideas, institutions, and technology from the West when it suits the Rest to do so. —p318

The strategies employed by these success stories are markedly varied. We should take from this that the adaptation of policies and institutions to local conditions would seem to be much more important than deploying the 'perfect strategy' in some abstract, generalised sense. The “market for feedback and accountability” seems to be involved in all of these stories, but they do not follow anything close to an idealised free-market model.

Independence from Western interference clearly isn't a panacea in itself; development can easily be hampered by autocratic institutions and corruption, and many economies have not yet found their own particular path to development success. And Western assistance can play a part in helping these countries to find that path, provided that it is “suitably humbled and chastened by the experience of the past”.²²⁾

[W]hat would the World Bank and IMF shock therapists advise if foreign aid were a country? They would probably abolish state ownership, do rapid privatisation and downsizing, let the market work, and put an end to bureaucratic central planning. I would favour a more gradual approach, of piecemeal reforms for the troubled country of foreign aid. —p321

As ought to be clear from the body of the book, recommendations for change focus on the process of finding context-specific policies which can be shown to make an incremental improvement, rather than any particular set of policies (or 'utopian grand plan').

If you want to aid the poor, then

1. Have aid agents individually accountable for individual, feasible areas for action that help poor people lift themselves up.
2. Let those agents search for what works, based on past experience in their area.
3. Experiment, based on the results of the search.
4. Evaluate, based on feedback from the intended beneficiaries and scientific testing.
5. Reward success and penalise failure. Get more money to interventions that are working, and take money away from interventions that are not working. Each aid agent should explore and specialise further in the direction of what they prove good at doing.
6. Make sure incentives in (5) are strong enough to do more of what works, then repeat step (3). If action fails, make sure incentives in (5) are strong enough to send the agent back to step (1). If the agent keeps failing, get a new one.

It's so obvious, I'm embarrassed even to lay it out. But it's worth laying out only because it is the opposite of the present Western effort to transform the rest. —p333-4

A number of examples of interventions that have been subjected to systematic evaluation against control conditions are described, in an attempt to illustrate that it is nontrivial to predict which a priori 'good ideas' produce results and which fail for reasons that may or may not be easy to discover. This illustrates the need for robust independent evaluation mechanisms.

Potential pitfalls are flagged. The first is that a redoubled focus on observable outcomes is likely to increase the current bias towards projects with the most observable outcomes.

It is hard to think of any incentive system that is going to successfully reward invisible effort toward producing invisible outcomes. Give it up. —p324

Another is that making aid agencies more accountable for results might push them towards tackling 'easier' problems, ini which they can be confident that they will achieve their goals. These may well be the sort of tasks that national governments would have addressed themselves in the absence of the agencies, perhaps encouraging the worst kind of fungibility.

This “fungibility” of aid money is something that aid analysts worry about, but perhaps is less of a worry when there are so many things failing. —p325

Two examples of possible mechanisms to improve accountability and feedback are detailed, one existing and one postulated. The first is the established website GlobalGiving, a bulletin board for aid projects in which agencies, individuals or firms post project proposals and donors of all kinds select those they wish to support. The second is a voucher scheme in which individuals and/or communities are given vouchers that they can spend with aid agencies operating in their area: people on individual interventions such as vaccinations or food supplements and communities on larger projects such as roads and wells. Such a scheme would give a voice to the recipients of aid, enabling them to 'vote' for the interventions that they most want or need. Agencies receiving vouchers could then exchange them for cash from donors, giving arms-length donors better information about which interventions the poor were really demanding. This process also favours agencies that achieve the same outcomes as others but at lower unit cost. Again, it's emphasised that possibilities such as these, no matter how good they sound, should be tested on a small scale and independently evaluated before being considered useful in any particular context. Incremental improvement requires not only new ideas but proper evaluation of those ideas to determine if, when and where they work.

Discard your patronising confidence that you know how to solve other people's problems better than they do. Don't try to fix governments or societies. Don't invade other countries, or send arms to one of the brutal armies in a civil war. End conditionality. Stop wasting our time with summits and frameworks. Give up on sweeping and naïve institutional reform schemes. The aim should be to make individuals better off, not to transform governments or societies…

Remember, aid cannot achieve the end of poverty. Only home-grown development based on the dynamism of individuals and firms in free markets can do that. —p322