The Millennium Development Goals, Aid Targets, And The Costs Of Over-Expectations
| Author | A. Clemens/J. Moss/J. Kenny |
| Position | Research Fellows at the Center for Global Development/Senior Economist |
| Pages | 18 |
Page 58
In September 2000 at the United Nations, 147 presidents, prime ministers, and monarchs-the largest-ever gathering of heads of state-unanimously adopted the Millennium Declaration, committing themselves to a series of international development objectives to be reached by 2015.1 Known since 2001 as the Millennium Development Goals ("MDGs"), these eight goals are widely cited as the primary yardstick against which advances in international development efforts are to be judged.2 The official MDG website says that the goals "have been commonly accepted as a framework for measuring development progress."3
After the MDGs were established, a priority became calculating what kinds of resources and actions would be necessary to reach them. There are now several "costing" studies, estimating how much money would be required to reach the goals. In addition to other conditions, such as higher economic growth and the improved economic policies, these studies have concluded that something in the range of $40-70 billion in extra resources each year will be necessary. Fifty billion dollars is the most commonly cited figure for new annual aid requirements.
Nearly five years after establishing the MDGs, it appears that the global goal of halving poverty may soon be reached because of rapid progress by the two population giants of India and China.4 However, it appears almost certain that the majority of developing countries will not meet the country-level poverty targets set by the Millennium Declaration, nor many of the other goals. Of the 47 African countries, 42 are considered "off track" for at least half of the targets and twelve are "off track" for all targets. Meeting the goals for the majority of country indicators would require more than doubling the rate of progress.5 For instance, Barbara Bruns, Alain Mingat, and Ramahatra Rakotomalala estimate that 86 out of 155 countries are at risk of not achieving the goal of universal primary education.6 Twenty- seven of these countries are not even expected to break the fifty percent completion threshold by 2015. These forecasts exclude the sixteen developing countries with no data-all of which are likely to have extremely low indicators. In the 2003 Human Development Report, the United Nations Development Programme estimates that, on current rates of progress, sub- Saharan Africa would not meet the hunger, primary education, and child mortality targets for at least another century.7
This apparently bleak state of affairs is already leading to complaints that the rich countries are not living up to their end of the MDG bargain.8 The eighth goal commits rich countries to a global partnership for development, wherein they promise to allow greater trade access, reduce debts, and increase aid. Although there has been substantial progress in debt reduction, rich country trade policies have not significantly changed to be more favorable to developing countries, notably on agricultural market reform. Furthermore, the estimated levels of mobilized resources required have not been forthcoming from donors. Official development assistance ("ODA") from the main international donors9 totaled $53 billion in 1999, and this figure rose to just $57 billion in 2002 and $79 billion in 2004, far from the doubling of aid called for by a range of costing studies. If many of the MDG targets are formally missed in 2015, will we be able to point to the failure of donors to provide resources as the main culprit?
This article discusses the links in the chain of causality between increased aid flows and attaining the MDGs, and suggests that, due to several caveats, the MDGs are unlikely to be reached regardless of new aid flows. The article examines several specific goals to highlight some of these issues. In each case, it appears that there are limits on the role for increased financing to accelerate trends, and many countries are very likely to miss the MDG targets, regardless of rapid progress. There may, in fact, be costs to over-expectations and the manufacture of "failures" to meet the MDGs.
From the beginning, the MDGs were interpreted as the need for greater donor financing.10 The 1990 OECD DAC development policy review, which gave birth to the MDGs, stated bluntly, "Development costs money . . . the high-income countries need to supply more aid."11 The United Nation's Monterrey Consensus proclaimed that "a substantial increase in [ODA] and other resources will be required if developing countries are to achieve the internationally agreed development goals."12 Page 59
Several efforts have been made to estimate the cost to achieve the MDGs. A 2001 study led by former Mexican President Ernesto Zedillo and known as the "High-level Panel on Financing for Development"13 estimated that reaching the MDGs would require an additional $50 billion per year in ODA, plus three billion dollars more in humanitarian aid, and about fifteen billion dollars more for "global public goods." This brings the total of additional resources to at least $68 billion, or slightly more than double the current levels of aid. A second widely cited study by Shantayanan Devarajan, Margaret J. Miller, and Eric V. Swanson of the World Bank uses two estimates for costing the MDGs.14 First, it calculates the additional resources necessary to increase economic growth so as to reduce income poverty. For this it suggests a "financing gap" of $54-62 billion each year. Second, it estimates the cost of meeting specific goals in health, education, and environment by using country-specific unit costs and then multiplying by the uncovered population. Through this method they find that $35-75 billion more per year is needed. Based on these two methods, this article concludes that ODA increases in the range of $40-70 billion are required. Again, this is an approximate doubling of official aid flows, roughly confirming the Zedillo estimate. A series of other papers have also tried to quantify the costs of meeting the MDGs for particular regions, countries, or for meeting an individual goal. Of note are the various estimates for costing universal primary education or Education for All ("EFA"), including nine billion dollars per year15 and between ten and fifteen billion dollars,16respectively.
Most of these costing studies are careful to clarify that the resulting estimates are imprecise and that several caveats apply to their conclusions. The two most commonly cited assumptions, and perhaps the most important, are: (1) the policy environment within developing countries considerably improves; and (2) current bottlenecks and capacity constraints are substantially relieved. Regarding policies, the Zedillo study, for example, assumes that recipients are doing "what's necessary" to improve policies. Similarly, Alain Mingat, Ramahatra Rakotomalala, and Jee-Peng Tan, in looking at the costs associated with reaching universal primary education for 33 African countries, state that "the implicit assumption is that countries would reform their education sector policies as needed to ensure that resources are used to offer quality services in a cost-efficient manner."17 The Devarajan study also explicitly leaves aside the question of developing countries' capacity to spend aid effectively.
Despite the careful qualifications included in many of these studies, many in the policy, advocacy, and media world have inappropriately focused attention on the bottom line figure: $50 billion more in aid is needed to achieve the MDGs. The misuse and misinterpretation of the costing studies has added to the impression that resources and aid flows are the critical or even sole determinant of development outcomes.
The first MDG is to halve the 1990 poverty headcount by 2015. On a global scale, this goal is very likely to be reached, almost entirely because of poverty reduction in India and China.18 At the same time, the majority of countries appear unlikely to halve poverty by 2015. In addition, it is doubtful that increased aid will sufficiently accelerate growth rates to meet the poverty reduction target. Most of the costing studies use the "financing gap" to calculate the additional aid required for meeting growth targets, but this approach is problematic and raises further doubts about the estimates.
Economic growth is central to the poverty reduction goal because it is the only source of increased income for the poor that can be (comparatively) rapidly achieved. Poor people in developing countries can become wealthier either through receiving a greater share of existing national income (redistribution of wealth from rich to poor) or a similar share of a greater national income (equitable economic growth). However, it is historically very rare to see rapid changes in income inequality (up or down) over time, and so those countries that have achieved rapid and substantial poverty reduction have done so mainly through economic growth.19 To achieve meaningful poverty reduction, economic growth rates will have to accelerate in the countries where the poor reside.
The World Bank suggests that the typical African economy will need to grow on average at least seven percent for the next fifteen years in order to halve poverty rates.20 This compares to an average regional growth rate of just 2.4 percent for the past fifteen years. High rates of growth are unusual for Africa as well as for the world as a whole. Between 1985 and 2000 only five countries managed to sustain a seven percent growth average.21
Figure 1 reveals the stark contrast between UN goals and performance in least developed countries ("LDC") growth, measured by the Penn World Table. The UN General Assembly resolutions declaring the second and third "development decades" -the 1970s and 1980s-gave explicit goals for average real annual growth in GDP and GDP per capita in developing countries: 3.5 percent and 4.5 percent, respectively. In both cases, population growth was assumed to be 2.5 percent per year.
The UN declarations for the first and fourth "development decades"-the 1960s and the 1990s-only give targets for GDP growth, but we can approximate the implied GDP per capita growth rate by assuming roughly the same rate of population growth of 2.5 percent. This means that the First Development Decade goal of five percent GDP growth implies roughly 2.5 percent in per capita growth, and likewise implies that the Fourth Development Decade goal of seven percent in GDP growth implies 4.5 percent in per capita GDP growth. Comparing all of these targets to actual LDC performance in the latter three decades of the twentieth century shows an arresting pattern. Every decade or two since the 1960s, the UN has increased its goal for developing country growth by one percent while growth in LDCs has not changed much. This "goal inflation" has, with the MDGs, arrived at the point where expectations of LDCs lie at the very extreme of the distribution. Decades after the first round of goals, we still do not know how to turn Zambia into Botswana, nor how to turn Laos into Korea.
[FIGURE IS NOT INCLUDED]Page 60
The expectation that unusually rapid growth rates might now be achieved more widely is based on two assumptions: (1) that policy changes will foster growth; and (2) that increased aid in the presence of those good policies will catalyze faster growth. There are problems with both of these assumptions, however. Further caution is required regarding the link between increased donor assistance and higher economic growth. This assumption underlies all of the costing studies that use the financing gap model for estimating how much aid will be needed to reach certain growth targets. These estimates start from a measurement of poverty-income elasticity and current growth rates, which suggests a "growth gap"-the rate at which the economy must grow to see the desired reduction in the poverty headcount-or in the case of reaching the poverty MDG, to halve the poverty ratio by 2015. This approach then uses the incremental capital output ratio ("ICOR") to calculate the level of investment required to reach the growth levels, and then subtracts domestic savings to get the external financing gap-or the amount of required aid. The problem is that, in practice, the financing gap model does not appear to work.22 As one example, William Easterly demonstrates in a 1999 study that had the financing gap approach worked as expected over the period 1960-1994, Zambia's per capita income would have been $20,000, or 33 times the actual figure of about $600.23
One recent and more positive contribution to the aid and growth literature is the work of Craig Burnside and David Dollar.24 This highly influential study has been used to make the case that aid can lead to growth under the right circumstances- including the policy environment assumed by many of the costing studies. The results of the Burnside and Dollar study appear to be somewhat fragile-the results tend to weaken when changing the time period, adding new country data, or altering the definitions of "aid" or "good policies."25 Robert Cassen's conclusion some ten years ago still seems to hold:
Inter-country statistical analyses do not show anything conclusive-positive or negative-about the impact of aid on growth. Given the enormous variety of countries and types of aid this is not surprising.... If such a relationship does not emerge overall, it only shows the unex- citing conclusion that aid may or may not be strongly related to growth, depending on circumstances.26
More recently, Michael Clemens, Steve Radelet, and Rikhil Bhavnani have found striking evidence of a positive average effect of development aid on growth, once the type of aid analyzed is matched to the time horizon of its expected growth effect.27 While this effect is clear on average, it varies greatly by country and its magnitude is limited to a certain range-two reasons to question whether even unlimited aid could cause a particular high level of growth in any given country.
It seems plausible to assume that the relationship between aid and growth in the presence of good policies holds, at least under some circumstances. Does this suggest that significant increases in aid are likely to help meet the poverty MDG? The answer is still uncertain. This is because most low-income countries with high levels of poverty either have poor policies and weak institutions (and thus are assumed to be unable to use additional aid effectively) or already receive considerable amounts of external assistance. This second factor may be a problem, because even work that accepts a link between aid and growth finds that, above a certain level of aid, the relationship begins to break down.28 Many poor countries thought to have "good policies" already receive substantial aid. Many top-performing countries-such as Ghana, Ethiopia, Uganda, Nicaragua, Honduras, Burkina Faso, and Tanzania-receive aid flows well above ten percent of GDP. Were total ODA levels to be doubled, as called for in the costing studies, the countries that are perhaps best able to absorb large aid increases are India and China, which currently receive minimal aid (0.36 percent and 0.13 percent of GDP, respectively). However, these two countries are both considered "on track."29 The MDGs do not change the oft-noted irony of aid: those that need it most are frequently the ones least able to use it effectively.
We turn now to the social sector MDGs. There is already a large literature on the complex relationships between conditions, interventions, and outcomes, but this appears to have been somewhat neglected in public discussion of the MDGs. For example, most health or education variables are quite closely related to income.30 However, studies have shown that over time, progress on these indicators is not correlated with the rate of growth in that country.31 Given this, it may be difficult to considerably accelerate progress through policy changes or alteration in resources.
Adding to the complexity of the causal chain between expenditures and outcomes is the fact that certain sector interventions can have impacts on other sector outcomes. Michael Kremer and Edward Miguel, for instance, found that de-worming programs had a strongly positive impact on school attendance in Kenya.32 Gustav Ranis and Frances Stewart, who found that health expenditures appeared to have little or no impact on life expectancy, suggest that increased female primary enrollment did.33
All of this suggests that: (1) additional aid may not be the most important factor in improving social outcomes; (2) sectoral distribution of aid to maximize progress on any particular social Page 61 MDG is not clear; (3) unit-cost approaches utilized in costing studies may be dramatically misleading; and (4) "best practices" may not be easily exportable because they are dependent on a range of determining factors that may be difficult to replicate. Of course, some of the costing study authors have suggested such problems themselves, but this has not stopped the widespread misinterpretation of their work. To better illustrate these arguments and the weaknesses in the costing approach to the MDGs, we turn now to the specific social sector goals and the historical record, focusing mainly on the second MDG of achieving universal primary education and, briefly, touching on the other goals.
Over the decades, rates of primary enrollment and completion have risen nearly everywhere-even in many of the very poorest countries-and they have risen at remarkably uniform rates. Figure 2 gives an overview of what happened to enrollments between 1960 and 2000. The figure answers the following question: If we were to make the assumption that the growth of primary enrollment across time in all countries followed the same pattern, based on one particular S- shaped curve (or "logistic" curve), what would that curve look like? The figure takes the path that each country followed during those four decades, and lines up each country individually so that, if following that single curve, it would have hit the fifty percent enrollment mark in the same year, that is, "adjusted year" zero.
Two things jump out from the graph. First, our assumption is not that bad; countries' idiosyncratic paths from low to high enrollment cluster remarkably closely around a single S-curve, whose slope at the halfway mark (or "transition speed") is about 0.04. There is variation around the curve, but remarkably little given that the countries there include tropical and arctic, rich and poor, socialist and capitalist, war-torn and peaceful. Second, the slope of the curve, or the "typical" transition speed from low to high enrollment, is low compared to growth rates required to meet the MDG target. While the extension of mass schooling in the latter half of the twentieth century was vast, it also took a long time. If it started at fifty percent enrollment, the typical country, whether rich or poor, would have risen to 70 percent after 22.3 years, 80 percent after 36.4 years, and 90 percent after 57.7 years.
Figure 3 shows that for the 90 countries for which we have data-representing a very broad range of wealth and institutional, political, and geographic conditions-no country has a transition speed above 0.13, the rate required to increase enrollment from 80 percent to 95 percent over a fifteen year period. There are 38 countries in this data set that have 2000 enrollment rates below 80 percent. For these countries, and others for which we do not have data, reaching 95 percent enrollment by 2015 (still short of the MDG target) will require historically unprecedented rates of progress.
It appears unlikely that many countries will even manage to approach the 0.13 transition rate. One reason for, or at least a related phenomenon to, the strong historical determinism of primary enrollment growth rates is the strong relationship between parental primary completion and child enrollment. This correlates with more than ten times the amount of cross-country variation in transition speeds than does education spending. This suggests a significant "demand side" element to primary education, with wealthier, educated parents far more willing to send their children to school. Deon Filmer estimates that even if all rural people in a sample of 22 countries lived right next door to a school, enrollment rates would only increase from an average of 49.8 percent to 53.1 percent-suggesting the dominance of "demand-side constraints."34 Across countries, there is no significant relationship between public spending per child on education and the primary school completion rate, once income is controlled for.35
Given that education expenditures do not appear to be a particularly strong historical determinant of enrollment, that not all countries will grow rapidly, and that it is very hard, in a fifteen year period, to dramatically increase parental primary completion rates, it is unlikely that many countries will achieve high transition speeds. Indeed, even most "exceptions to the rule" or "best practices" of rapid increases are, sadly, related to unique circumstances, considerable reductions in quality, or questionable data.36
[FIGURE IS NOT INCLUDED]Page 62
Turning to gender equality in education, Figure 4 shows the path taken by the female-to-male ratio in gross primary enrollments for all 168 countries in the UNESCO database since 1950. The figure is constructed in a manner identical to that of Figure 2. Once again we see that the assumption that all countries follow the same S-shaped path to gender parity is not strictly true, but neither is it far from the mark. The typical country has taken a long time to reach gender parity in primary school enrollment. The shape of the curve in Figure 4 suggests, for example, that a country whose ratio of girls' gross primary enrollment to boys' is 0.8 typically takes 28 years to reach a ratio of 0.95. In 2000, seventeen countries had a ratio of less than 0.8. Nevertheless, the great majority of developing countries are already fairly close to meeting the MDG target of gender equality in education.37 A substantial majority of countries will likely reach this particular goal by 2015. The use of static unit costs in education estimates, as used in the costing studies, is therefore likely to ignore perhaps the most important determinants of enrollment.
In the fourth MDG, governments have committed themselves to reducing child mortality by two-thirds between 1990 and 2015. Assume (as we did with the education goals) that every country follows the exact same S-curve, in this case towards zero infant mortality, but (again) that each country goes through this transition at a different time.
These assumptions once more appear reasonably robust. Figure 5 shows what happens if we take data for 35 rich and poor countries covering roughly the twentieth century, assume that the historically highest level of infant mortality was 350 infant deaths per 10,000 live births, and then horizontally align all the series so that every country passes through fifty percent of the maximum-that is, 175-in the same year ("adjusted year" zero). The lessons are familiar from the previous two sections. Immediately we see that: (1) there is remarkably little diversity in the rate at which this has occurred; and (2) the typical experience of a country in the twentieth century was that this transition was slow as compared to the MDG target decline. The slope of the S-curve running through the middle of the cloud, representing the experience of the "typical" country among these 35, is -0.0339 at the inflection point. That means that a country typical of this group, if it started out at one hundred infant deaths per ten thousand live births, would take 40 years to decrease this level by two-thirds. This figure is relatively insensitive to the starting level.
Nor has the story dramatically changed in the late twentieth century, despite technology and economic advances. Although such advances allowed for far lower infant mortality rates essentially everywhere, the speed at which countries made the transition from high to low infant mortality did not change significantly. We will now restrict the sample to the years 1980-2000, and consider all 176 countries for which the World Development Indicators provide data. The S-curve will show that a typical country in this group, starting from an infant mortality of one hundred, would take 42 years to lower this figure by two-thirds.38
This suggests that the forces that primarily determine the speed of this transition go beyond public health policy and inputs-a conclusion supported by a number of different studies. It appears that one of the reasons that interventions often do not have the desired outcome is, as Lant Pritchett showed with the relative ineffectiveness of family planning aid,39 that supply- side responses do not address the demand components that are affected by broader social and economic changes. Income inequality, literacy, ethnic composition, and religion are all factors slow to change, at least within a fifteen-year perspective.
None of this is to say that specific public health interventions and large injections of inputs purchased by aid cannot affect infant mortality; obviously they can. Lucia Hanmer, Robert Lensick, and Howard White suggest that while income per capita, education, and gender inequality are all robust determinants in explaining infant and child mortality across countries at a single time, some health spending (particularly on immunization programs) can also have a significant impact.40 Rather, the question is whether the rapid increase in the mortality transition rate needed to meet the fourth MDG is accessible to even the wisest and best-funded policy interventions.
[FIGURES ARE NOT INCLUDED]Page 63
It is not that we cannot imagine a scenario where historically unprecedented progress is made towards the MDGs. If effective preventative and treatment interventions available for preventing childhood mortality became ubiquitous, the number of under-age-five deaths worldwide might fall by as much as 63 percent. For instance, one recent cross-country analysis suggests that, in countries with good governance, additional health spending and aid financing can have an impact on health outcomes.41 Yet this same study suggests how difficult it would be to meet the health MDGs in Africa, and why the MDGs at their current levels are over-reaching. First, the average quality of African government institutions, as measured by the World Bank's Country Policy and Institutional Assessment, would have to leapfrog to one standard deviation above the mean global score (from significantly below the average today). Then government health expenditures as a percentage of GDP would have to reach as much as 16.5 percent - an implausible level greater than that spent by any country in the world.
It may be that progress in meeting some of the other MDGs- perhaps in halting the spread of HIV/AIDS, reversing the incidence of malaria, and halving the proportion of people without sustainable access to drinking water and sanitation-could in fact be rapid enough to meet the MDG timetable. This appears possible because a small number of technological advances, or a significant increase in investment in a particular infrastructure, could have a major effect on these areas in ways not obvious in the other goals. The creation of a malaria vaccine, for example, would have a monumental impact toward meeting the sixth MDG,42 but it is difficult to imagine a similar single technological advance that would significantly impact education outcomes.
Growing concern that the MDGs will not be achieved should not obscure the bigger picture: development progress in terms of social indicators has been occurring at unprecedented levels throughout the great majority of the world's population over the past thirty or more years. For example, the divergence in life expectancies between rich and poor countries, which had been occurring since perhaps as early as the fifteenth century, has been dramatically reversed in the second half of the last century. Between 1950 and 1999, the population-weighted average life expectancy has risen from 51 to 69 years while the population-weighted standard deviation has fallen from thirteen to seven years.43 Data on infant survival suggests a similar performance. In the second half of the twentieth century, average global literacy increased from 52 percent to 81 percent, while the weighted standard deviation dropped from 38 percent to seventeen percent. Turning to female literacy as a percentage of male literacy, over the 1970-2000 period, the global average ratio has improved from 59 percent to 80 percent. This reflects a dramatic long-term improvement in social indicators even for countries that have seen limited economic growth. For example, average life expectancy for countries at $300 GDP per capita in 1999 is slightly higher than that for countries with a GDP per capita of $3,000 in 1870 (in constant dollars). In other words, it now takes only one tenth the income to achieve the same life expectancy as it did 130 years ago.44
It is hard to view this progress as anything other than a dramatic success. Even if divergence continues "big time" with regard to income,45 other quality of life indicators suggest historically unprecedented improvement. There are, of course, significant clouds on the horizon-the AIDS pandemic is having a particularly dramatic impact in Sub-Saharan Africa, where life expectancy in the region as a whole has declined in recent years, and is likely to level off only in 2010.46 Nonetheless, it is not clear why we should expect progress to halt more broadly.
A continuation of the progress that has characterized the typical developing country of the last fifty years will, by and large, leave countries missing the MDGs in 2015, yet still outperforming the historical trajectories of now-developed countries. In Figure 6, for example, we see the trajectory of primary school enrollment for Burkina Faso. The country is on a trend well above the typical country since 1960 and even further ahead of the typical rich-country transition in the nineteenth century. (The same statement is true of Mali, Senegal, Madagascar, and Nicaragua, among others.) Surely this is not unambitious performance. Despite this success, however, Burkina Faso will fail to meet the MDG target. It is perhaps worth asking whether the success or the target should be questioned.
[FIGURE IS NOT INCLUDED]Page 64
There is a long history of international goal setting for development. For instance, a steady succession of international conferences since the 1940s has declared universal primary education achievable within a short-time period and pledged to make the necessary investments to do so.47 Nevertheless, countries all over the world continue to undergo the slow adjustments of their societies and economies that have allowed more children to go to school. But in setting unrealistic goals and claiming they can in fact be universally met, the MDGs may run the risk of creating a climate of inaccurate pessimism about both development and aid.
The MDGs can be understood in two ways. One interpretation is to take the specific goals literally, accept them as the real targets of the development community, and take the costing study estimates as reflective of the amount of aid needed to reach those goals. This view sees the MDGs as an important mechanism for raising aid flows and ensuring accountability for donor promises. This literal interpretation of the MDGs and misreading of the costing studies can lead to the belief that huge aid flows can quickly produce epochal change in an array of development indicators across broad regions. Perhaps such outcomes will be achieved, particularly regarding the goals in areas such as water and sanitation, but, as we have seen, the historical evidence suggests it is unlikely that the majority of goals will be reached by the majority of countries.
The determinants of the outcomes embodied by the MDGs are in fact complex. Time itself clearly is an important factor but is not accounted for by universal time-bound goals. Many of the available interventions in terms of policy reforms or increases in resources are supply-side responses. These interventions do little to increase demand, which is linked to longer-term social and economic changes. This does not mean that poor people do not desire better standards of living; rather the range of incentives faced by many poor people lead them to make choices that might contradict the outcomes represented in the MDGs. This is one reason why outcomes seem to change only slowly, and also suggests that there may be a limited potential role for aid in meeting extremely ambitious, universal, time-bound goals. This does not claim in any sense that aid is unimportant or ineffective, but only that aid cannot by itself deliver the MDGs.
A second understanding of the MDGs is a more nuanced view-that the goals are a symbol of the kinds of outcomes toward which the development community should strive. Similarly, new aid flows are considered just one of several necessary conditions for progress on development indicators. This second interpretation makes the MDGs a tool rather than a practical target. Goals generate discussion, focus attention, and help assign accountability for leaders' pledges. There can be little doubt that the MDGs helped galvanize the aid community and reverse the aid declines after the end of the Cold War. The United States, the European Union, Canada, Norway, and others made promises of substantial aid increases at Monterrey in 2002, a result doubtlessly influenced by the MDG negotiations two years earlier.
In spite of the notable benefits of the MDGs, even when taken as symbolic rather than literal, there has been almost no discussion so far of potential costs of the specific form taken by these goals. These potential costs take two distinct forms: unreasonable expectations about what is likely to be achieved within a short time period, and unreasonable expectations about the role of aid in the development process.
The specific targets of the MDGs have set up many countries for unavoidable "failure." Some governments pursuing wise policies and historically encouraging progress on development indicators could be weakened or de-legitimized by the label of "failure" in 2015. The MDGs confuse interpretation of their performance with universal, time-bound targets that for many countries are, in practice, impossible to reach. Costing studies, by positing that such goals are attainable and asking merely for the resource inputs, contribute to the illusion that the goals are attainable for all countries. Even if most development practitioners know this is not true, they must recognize that the expectations of many have been raised.
Another potential downside is the possibility for adding to donor fatigue and distracting recipient countries from much- needed domestic reforms. If donors provide additional tens of billions of dollars in aid per year sometime in the next few years, and if subsequently many of the goals are still not met, this will provide ammunition to interest groups in rich countries seeking to give up on development assistance. Developing countries will undoubtedly need many decades of sustained assistance-as Korea, Botswana, and other eventual successes have received-and this must not be interrupted by declarations of failure in 2015. However, if the increase in aid does not materialize, a failure to reach the MDGs may help legitimize leaders in the developing world who pursue policies that are anathema to economic growth. "What else could we do," they will ask, "when the rich countries broke their promises?" The ensuing finger pointing could also undermine constituencies throughout the developing world for necessarily slow but essential reforms toward transparency, accountability, rule of law, and meritocracy.
Moving forward, the donor community should accept that it is not feasible for the majority of countries to reach the majority of the MDGs. Similarly, the costing studies should not be invoked as evidence that we can simply purchase outcomes with more assistance. The studies themselves make no such claims, and history shows this is highly unlikely to be true. Instead, the MDGs should be presented as useful benchmarks that publicly bring out the stark contrast between the world we want and the world we have, and cause us to redouble our search for points of intervention to close the gap.
The donor community might also consider ways of institutionalizing the recognition of development success. The government of Burkina Faso, for example, should be supported and lauded by the international community for raising school enrollments much faster than most poor and now-rich Page 65 countries did in the past, rather than criticized and delegitimized because primary enrollment is less than fifty percent. Country-specific benchmarks can help signal when interventions of some kind are necessary, and they can also provide markers for progress along the way, given a country's circumstances. Instead of focusing time and energy on compiling lists of "off-track" countries, effort should be spent on compiling lists of countries that are "on-track" or better after taking account of their particular circumstances and historical trends. Some kind of institutionalized response by the international community would thus redefine Burkina Faso as the educational success it has been. This would not in any way endorse the fact that half of Burkinabe children still do not enroll in school, nor imply in any way that schooling is not their right. It would, however, bring important international pressure to bear in support of those who for decades have been working to get Burkinabe children into school at a rate faster than many far richer countries have managed.
Lastly, future international development goals might avoid some of these pitfalls. The next round of goals should: (1) be country-specific and flexible, more like today's International Development Association targets; (2) take historical performance into account; (3) focus more on intermediate targets than outcomes; and (4) be considered benchmarks to spur action in cases where assistance is not working, rather than technically feasible goals.
This last point is worth emphasizing: it is useful to know that a country is raising school enrollments more slowly than the historically typical rate. It can give political support to constituencies in that country seeking changes in national policies, spur donors to intervene, and support change through financial and other means. But this is much different from the effects of declaring that it is feasible for a country to raise enrollments at five times the historically typical rates. Country-specific benchmarks carry the benefits of goal setting without the potential downsides of universal goals. This suggests that goal setting at the global level should be bottom-up rather than top-down-that is, the world targets should start from country goals and then aggregate up, rather than setting global goals and then estimating what countries would need to do to achieve them.
Indeed, for the next round of goals, the donor community might consider avoiding global-level costing studies, especially for outcomes known to be only tenuously linked to financial inputs. Rough back-of the-envelope estimates can potentially be useful for identifying the hypothetical scale of resources and also for some limited supply-side interventions. Yet the widespread misinterpretation of the studies suggests that, however narrowly conceived by the authors, misuse appears difficult to avoid. A more direct approach might be to advocate cost-specific interventions and link them to intermediate indicators rather than outcomes; for example, costing an immunization program rather than child mortality. Calculating financing gaps and unit costs for final outcomes appear to merely create more illusion than illumination.
It is worth stressing the caveats attached to our analysis. None of this is to argue against aid or that goal setting is per se counterproductive. Aid has clearly been an important part of developmental progress for many countries. Perhaps aid levels should increase by $50 billion, but not with the expectation that this will cause the MDGs to be met. Similarly, goals should indeed be set to enhance accountability and allocational efficiency, but goals must take history and context into account or potentially risk malign irrelevance.
Perhaps most significantly, we have based most of our argument on historical precedent. History can be a fickle guide to the future. To take two recent development trends as an example, the spread of the Internet has been more rapid than the spread of the mobile phone, which was in turn more rapid than the television, which was in turn more rapid than the fixed telephone. And the spread of democratic institutions in developing countries over the past fifteen years would have been poorly predicted based on a trend of declining democratic freedoms over the thirty years previously. It may be (and we hope it is) the case that policies will improve, that the environment for the effective utilization of aid becomes friendlier, and that technology and policy trends combine to allow historically unprecedented levels of progress across the broad range of development that is encompassed by the MDGs. Even if that is not the case, many countries will reach at least some of the MDG targets. More importantly, it is quite probable that the significant rate of improvement that we are already seeing in developing countries will continue in the next fifteen years, enhancing the lives of billions worldwide.
----------------------------------
ENDNOTES :
[1] United Nations Millennium Declaration, G.A. Res. 55/2, UN Doc. A/RES/55/2 (Sept. 8, 2000).
[2] In December 2000, the UN General Assembly requested that Secretary General Kofi Annan prepare a "road map" of how to achieve the targets of the Millennium Declaration to which the leaders at the Millennium Summit had committed in September of that year (G.A. Res. A/RES/55/162, 18). Annan's response, issued in September 2001 as the Road Map towards the Implementation of the United Nations Millennium Declaration, proposed the eight Millennium Development Goals ("MDGs") in their final form and by that name (U.N. Doc. A/56/326), drawing only on elements to which 147 heads of state or government had directly agreed at the 2000 summit.
[3] UNITED NATIONS, MILLENNIUM DEVELOPMENT GOALS, at http://www.un.org/millenniumgoals (last visited Nov. 2, 2005).
[4] The leaders at the Millennium Summit committed only to halving global poverty, regardless of what happens in any given country or region. They never agreed that each country or region would individually halve poverty by 2015, though the first MDG has often been given this latter interpretation. The UN's Human Development Report 2003 (pp. 198-202) and the World Bank's World Development Report 2004 (pp. 254-255), for example, both track individual countries' progress towards halving national poverty by 2015 as indicators of progress towards the first MDG.
[5] GABRIEL CARCELES, BIRDER FREDRIKSEN, AND PATRICK WATT, WORLD BANK AFRICA REGION HUMAN DEVELOPMENT WORKING PAPER SERIES, CAN SUB-SAHARAN AFRICA REACH THE INTERNATIONAL TARGETS FOR HUMAN DEVELOPMENT? (2001). 6 BARBARA BRUNS, ALAIN MINGAT, AND RAMAHATRA RAKOTOMALALA, WORLD BANK, ACHIEVING UNIVERSAL PRIMARY EDUCATION BY 2015: A CHANCE FOR EVERY CHILD (2003).
[7] HUMAN DEVELOPMENT REPORT, UNITED NATIONS DEVELOPMENT PROGRAMME (2003).
[8] See, e.g., Andrew Balls, Donors fail on education funding, FIN. TIMES, Mar. 29, 2004.
[9] "Donors" is used here to signify the 23 members of the Development Assistance Committee ("DAC") of the Organization for Economic Cooperation and Development ("OECD"). All Official Development Assistance ("ODA") data comes from the DAC.
[10] Since 1970 the UN has had an explicit goal to raise ODA to 0.7 percent of gross national income. The origin of the aid target was the 1969 Pearson Commission report, later accepted by the DAC and then endorsed by the UN General Assembly in Oct. 1970.
[11] OECD, SHAPING THE 21ST CENTURY: THE CONTRIBUTION OF DEVELOPMENT COOPERATION (1986), available at http://www.paris21.org/betterworld (last visited Nov. 2, 2005).
[12] Int'l Conference on Fin. & Dev., Monterrey, Mexico, Mar. 18-22, 2002, Report, UN Doc. A/CONF.198/11.
[13] International Conference on Financing and Development, Monterrey, Mexico, Mar. 18-22, 2002, Report of the High Level Panel on Financing for Development, available at http://www.un.org/reports/financing (last visited Nov. 2, 2005).
[14] SHANTAYANAN DEVARAJAN, MARGARET J. MILLER, AND ERIC V. SWANSON, WORLD BANK POLICY RESEARCH WORKING PAPER 2819, GOALS FOR DEVELOPMENT: HISTORY, PROSPECTS AND COSTS (2002) [hereinafter GOALS FOR DEVELOPMENT].
[15] ENRIQUE DELAMONICA, SANTOSH MEHROTRA, AND JAN VANDEMOORTELE, UNICEF INNOCENTI WORKING PAPER, IS EFA AFFORDABLE? ESTIMATING THE GLOBAL MINIMUM COST OF "EDUCATION FOR ALL" 87 (2001).
[16] GOALS FOR DEVELOPMENT, supra note 14.
[17] ALAIN MINGAT, RAMAHATRA RAKOTOMALALA, AND JEE-PENG TAN, WORLD BANK, FINANCING EDUCATION FOR ALL BY 2015: SIMULATIONS FOR 33 AFRICAN COUNTRIES ix (2002).
[18] Surjit Bhalla claims that poverty has already been halved, Imagine There's No Country: Poverty, Inequality, and Growth in the Era of Globalization, (INT'L INST. OF ECON. 2002); the World Bank disputes this, WORLD BANK, WORLD DEVELOPMENT REPORT (2004) [hereinafter WORLD DEVELOPMENT REPORT].
[19] See DAVID DOLLAR AND AART KRAAY, WORLD BANK POLICY RESEARCH WORKING PAPER 2587, GROWTH IS GOOD FOR THE POOR (2001).
[20] Ricardo Gottschalk, Growth and Poverty Reduction in Developing Countries: How Much External Financing Will Be Needed in the New Century? (INSTIT. OF DEV. STUDIES, U. OF SUSSEX, 2000)(estimating an even higher required rate of 8.2 percent for sub-Saharan Africa and 10.2 percent for Latin America); see also WORLD BANK, STRATEGIC FRAMEWORK FOR ASSISTANCE TO AFRICA: IDA AND THE EMERGING PARTNERSHIP MODEL (2004).
[21] Calculated from WORLD BANK, WORLD DEVELOPMENT INDICATORS (2004).
[22] William Easterly, The Ghost of Financing Gap: Testing the Growth Model Used in the International Financial Institutions, 60 J. of Dev. Econ. 423-438 (1999) [hereinafter Easterly 1999].
[23] WILLIAM EASTERLY, THE ELUSIVE QUEST FOR GROWTH 43 (MIT Press, 2001).
[24] Craig Burnside and David Dollar, Aid, Policies, and Growth, 90 AM. ECON. REV. 847-868 (2000).
[25] See e.g. William Easterly, Ross Levine, and David Roodman, New Data, New Doubts: Revisiting 'Aid, Policies, and Growth' (Center for Global Dev., Working Paper 26, 2003); H. Hansen and F. Tarp, Aid and Growth Regressions, 40 J. of DEV. STUD. 64(2) (2001), at 101-118; R. Lensink and H. White, Are there Negative Returns to Aid?, J. of DEV. STUD. 37(6) (2001), at 42-65; P. Guillamont and L. Chauvet, Aid and Performance: A Reassessment, J. of Dev. Stud. 37(6) (2001), at 66-92.
[26] ROBERT CASSEN, DOES AID WORK? (Oxford Univ. Press, 1994).
[27] Michael Clemons, Steven Radelet, and Rikhil Bhavnani, Counting chickens when they hatch: The short-term effect of aid on growth (CTR FOR GLOBAL DEV., Working Paper 44, 2004).
[28] Paul Collier and David Dollar, World Bank Policy Research Working Paper 2403, CAN THE WORLD CUT POVERTY IN HALF? HOW POVERTY REFORM AND EFFECTIVE AID CAN MEET INTERNATIONAL DEVELOPMENT GOALS (2000). However, some suggest that, using the coefficients from Collier and Dollar, the $50 billion additional aid they suggest could be apportioned only hitting this binding constraint in two countries. Furthermore, if aid were better delivered by donors, this binding constraint might be even higher. See Shantayanan Devarajan, William R. Easterly, and Howard Pack, Low Investment is not the Constraint on African Development (Center for Global Dev., Working Paper No. 13, 2002).
[29] Ray Marcelo, India Opts to Decline Aid from All but Six Countries, FIN. TIMES, July 8, 2003.
[30] Lant Pritchett & Lawrence H. Summers, Wealthier is Healthier, J. OF HUM. RESOURCES 31(4) (1996), at 841-868.
[31] See Easterly 1999, supra 22; Even if one accepts a close link between improvements in social sector performance and income growth, the rates of income growth required to meet MDG social targets are historically unprecedented - using cross-country variations in child mortality and income in 2000 to calculate elasticities suggests that low income countries would have to grow at 6.7 percent a year to reduce mortality by two- thirds in 2015. WORLD DEVELOPMENT REPORT, supra note 18.
[32] Michael Kremer & Edward Miguel, Worms: Education and Health Externalities in Kenya (Nat'l Bureau of Econ. Res., Working Paper No. 8481, 2001).
[33] Gustav Ranis & Frances Stewart, Growth and Human Development: Comparative Latin American experience, 39 Developing Economies 333- 365 (2001).
[34] DEON FILMER, WORLD BANK, SCHOOL AVAILABILITY AND SCHOOL PARTICIPATION IN 21 DEVELOPING COUNTRIES (2004).
[35] WORLD DEVELOPMENT REPORT, supra note 18.
[36] Michael Clemens, The Long Walk to School: International education goals in historical perspective (Center for Global Dev., Working Paper 37, 2004) [hereinafter The Long Walk to School].
[37] In 2000, the World Development Indicators 2004 lists data for 131 countries on the ratio of girls to boys in primary and secondary education. Of those countries in that year, only ten remained below the level of 80 girls per 100 boys. Supra note 21.
[38] This figure is not strictly comparable with the above figure for the whole twentieth century, since the sample of countries is quite different and the post-1980 figure includes far more poor countries.
[39] Lant Pritchett, Desired Fertility and the Impact of Population Policies, POPULATION & DEV. REV. 31(4) (1994), at 1-55.
[40] Lucia Hanmer, Robert Lensink, and Howard White, Infant and Child Mortality in Developing Countries: Analysing the Data fro Robust Determinants, J. OF DEV. STUD. 40(1) (2003), at 101-118.
[41] WORLD BANK, THE MILLENNIUM DEVELOPMENT GOALS IN HEALTH: RISING TO THE CHALLENGE (2003).
[42] But see McCarthy, F. Desmond, H. Wolf, and Y. Wu, The Growth Costs of Malaria (Nat'l Bureau of Econ. Research, Working Paper 7541, 2000). Looking at determinants of malaria infection rates, McCarthy et al. calculate that weather, latitude, income, poverty and inequality account between them for 4.9 times the variation in infection rates than does access to rural health care; even though this does suggest some role for government expenditure and aid, see also Anne Mills & Sam Shillcutt, Copenhagen Consensus Challenge Paper on Communicable Diseases (Copenhagen Consensus Challenge Paper, 2004).
[43] Charles Kenny, Why Are We Worried About Income? Everything that Matters is Converging, World Development (2005) [forthcoming].
[44] Id.
[45] Lant Pritchett, Divergence, Big Time, J. OF ECON. PERSPECTIVES 11(3) (1997), at 3-17.
[46] Richard Easterlin, The Worldwide Standard of Living since 1800, J. OF ECON. PERSPECTIVES 14(1) (2000), at 7-26.
[47] The Long Walk to School, supra note 36.
