This section reviews the intellectual underpinning of the studies on finance needs that were undertaken in the MDG context and the methodologies used in order to calculate them. The review reveals a range of unrealistic assumptions that often underpinned these studies that were largely ignored in making the case to scale up ODA in order to meet the MDGs. One assumption that we consider critically throughout the Report is that finance - and ODA in particular - will automatically (with fixed multipliers) lead to development outcomes without considering other finance flows, policies, or required structural changes. This chapter also discusses cases where modelling studies have improved upon or learned from the past.
2.1.1 The intellectual underpinning
of the MDG approach to finance needs studies
The studies on finance needs tended to regard ODA as a means to fill a finance gap similar to the
‘take-off’ and ‘big push’ literature developed by economists such as Arthur Lewis (1954), Walt Rostow (1960) and Paul Rosentein-Rodan (1943, 1961), with the underlying idea that ODA could help to
‘unlock’ growth in developing countries. In this vein, the Harrod–Domar model was extensively used to calculate the finance required to achieve the MDGs, in particular the eradication of extreme poverty. In this line of thinking, the lack of capital accumulation and investment is a key constraint to economic growth. External finance, in particular ODA, can launch a ‘take-off in self-sustained growth’.
Drawing on this, Chenery and Strout (1966) developed the two-gap model used to justify ODA and still widely used in estimating finance needs. This model identifies two constraints to economic development that could be addressed by ODA: the first is between import requirements
2.1
A major objective of the MDGs was to broaden the development discourse beyond the focus on economic growth. Consequently, they served to push ODA away from the broader emphasis on economic growth even though the overall approach was still aid-centred. The debate on aid effectiveness became a lively issue, as embodied by two economists: Jeffrey Sachs supporting the argument for increased aid based on the ‘big push’ argument, and William Easterly criticising the MDGs’ ‘one-size fits-all’ and comprehensive approach to development, in particular the over- reliance on ODA. Debates on the MDGs have too often been caricatured as opposing ‘growth’,
‘aid’ and ‘governance’ (Vandemoortele, 2011), although all three seemed to be components of the same package.
2.1.2 Methodological steps needed in finance needs studies
Many studies estimate finance needs and gaps either by ‘sector’ (e.g. health, education or the environment) and/or by objectives (such as the MDGs whether globally or at the country level).
We have reviewed these and there has been a proliferation of global estimates. As Box 2.1 describes, there have also been different types of country-level studies to estimate the cost of achieving the MDGs. We refer to these studies throughout the chapter and in particular to country-based MAMS modelling commissioned for this Report.
Estimates of finance needs and gaps are based on underlying assumptions and vary widely according to the context and definition of a target. They also depend on the policy context and on how efficiently existing financial resources are used. The term
‘cost’ refers to the volume of finance as opposed to ‘needs’. This suggests that those supplying the finance make interventions, with the implicit assumption that they will be involved in seeking and providing resources to achieve a target. Embedded in the literature on the finance gap for the MDGs is the assumption that ODA will fill it. Talking about
financial need offers more flexibility regarding the sources that could fill the gap.
According to the UNTT a finance gap represents the ‘difference between the current situation and a desired situation’ in relation to a goal or target, or the difference between the available and required finance to meet a specific objective (UNTT, 2013a: 33). The definition of a gap also depends on what is assumed to be available.
Some consider that available resources refer only to domestic resources while others (e.g. OECD, 2011) include external financing – another source of inconsistency across studies.
Studies on finance needs take important methodological steps in five areas: choosing a target: mapping the means by which it can be reached; choosing a scenario; choosing an estimation method; and other considerations.
While the resulting approach can lead to new insights, each step also comes with a set of problems that need to be addressed if the underlying assumptions are unrealistic. It further explains why these studies are not strictly comparable.
There have been three approaches to costing the MDGs at the country level:
UNDP country studies: In the early 2000s the United Nations Development Programme (UNDP) piloted country-level costing exercises in Cameroon, Malawi, Uganda, Tanzania and the Philippines. The models focused on six MDGs: income poverty, primary education, child mortality, maternal health, HIV/AIDS and water. The methodology used differed across targets and countries and identified key interventions for each objective.
The Millennium Project: In 2005, the Millennium Project (directed by Jeffrey Sachs for the United Nations) published Investing in Development: A Practical Plan to Achieve the Millennium Development Goals. It was based on a number of country case studies to identify major ‘interventions’ required to achieve all eight MDGs in each of these countries. Expert task forces developed ‘MDG needs assessments’, compiling lists of technical interventions and associated investment plans to attain the MDGs. In education, for example, the list includes providing more schools and teachers, ending tuition fees, and providing books and uniforms. Countries included Bangladesh, Bolivia, Cambodia, Ghana, Malawi, Tanzania and Uganda. Local counterparts collected information on the unit costs of the interventions. The linear ‘scale up’ of interventions and investments in each sector was summed in order to estimate resource requirements and develop a financing strategy. The study did not consider policy or institutional reforms.
World Bank: In the early 2000s, a World Bank project focused on the following countries: Benin, Burkina Faso, Ethiopia, Madagascar, Mali, Mauritania, Tanzania and Uganda; Bangladesh, India, Indonesia, Pakistan and Vietnam; Bolivia, Honduras; Albania and Kyrgyz Republic. Its approach gave priority to the macroeconomic policy objectives (such as the containment of inflation, budget deficits and current account deficits) emphasised in the respective Poverty Reduction Strategy Papers (PRSPs), and asked how, given these priorities, the MDGs could best be achieved.
The World Bank also developed a CGE modelling approach. The MAMS models (MAquettes for MDG Simulations) developed by World Bank researchers (Bourguignon et al., 2008) and since used extensively by UNDESA, provide a general equilibrium framework for countries to simulate the effect of different financing sources (e.g. ODA grants, foreign borrowing, and domestic taxation) on different MDGs. The models also take into account the effects of progress in one MDG on progress in others and include issues of absorptive capacity related to large finance inflows.
Sources: Bourguignon et al. (2008); Millennium Project (2005); Reddy and Heuty (2006); Sánchez et al. (2013)
Box 2.1 | Approaches to costing the MDGs at the country level
2.1.2.1 | Choice of target
The first step is to identify a relevant target, e.g.
specific MDG targets such as food security or access to safe water. While it is important to use appropriate, common descriptions of a target, this is often not the case. Moreover, different studies may cover a similar target but include different countries, again making most of them not strictly comparable.
2.1.2.2 | The choice of means by which to reach a target
A second step is to identify the means by which to reach the target. Often, there are several
strategies and means available. Depending on country characteristics, there could be various options for how to attain the same target, each with a different cost. For example, some might emphasise the potential of organic agriculture to reduce poverty, while others highlight the advantages of investing in chemical inputs or domestic seed banks. Some suggest promoting school enrolment by providing mid-day meals, others by reducing the distance pupils have to travel to school. Often such strategies could be complementary.
As described in Box 2.1, the ‘Millennium Project’
(2005) mapped out the major policy steps and
investments necessary to achieve the MDGs in selected countries. Researchers identified a set of key interventions to support the achievement of the targets. For each target and country the methodology was to first examine the existing gap and its geographical distribution, in particular distinguishing between urban and rural areas.
This made it possible to identify the investment needs and costs of the interventions by which to reach the target.
Unfortunately, many models of MDG finance needs offer only one option, even if is not the most efficient, rather than considering a range of alternative means and approaches. This is particularly the case for global studies that examine the achievement of MDG1, where ODA is linked to aggregate investment, aggregated growth and then aggregated poverty, thus losing valuable detail on different and more efficient ways to reduce poverty. There are exceptions, as we shall discuss in this Report, most notably those studies that use MAMS CGE modelling (see e.g. Box 2.4, Box 2.5 and Box 2.6), and also some global estimates that differentiate among different sectors (IFPRI, 2008).
2.1.2.3 | Choosing a scenario
Most early analyses are based on status quo or
‘all other things being equal’ scenarios. This can be unrealistic, however, and shocks can have a large impact. Reddy and Heuty (2006) point to the impact of the AIDS epidemic on health targets, which severely affected estimates of finance needs. Further, the literature on climate change highlights the fact that various projections and sensitivity analyses yield very different finance needs. For example, infrastructural needs differ markedly depending on whether climate change is included in the scenarios.
The choice of scenarios relates to the assumptions underpinning the chosen methodology. For example, estimates based on the ‘back of an envelope’ and descriptive econometric models
use past experience to define indicators or multipliers in order to predict the future. This means that they might not take account of dynamics over time. Other models, such as CGEs, allow for building dynamic scenarios to account for degrees of shock on various trends.
An example of this is the World Bank’s application of CGE models, initially designed to evaluate the distance to be travelled in order to reach the MDGs (Bourguignon et al., 2008). While CGE models are calibrated according to past experience, they can introduce new parameters, although the estimates rely on an accurate choice of parameters and assumptions.
The IPCC-SRES (IPCC-SRES, 2000: 23) defines scenarios as follows:
Scenarios are images of the future, or alternative futures. They are neither predictions nor forecasts.
Rather, each scenario is one alternative image of how the future might unfold. A set of scenarios assists in the understanding of possible future developments of complex systems. Some systems, those that are well understood and for which complete information is available, can be modelled with some certainty, as is frequently the case in the physical sciences, and their future states predicted. However, many physical and social systems are poorly understood, and information on the relevant variables is so incomplete that they can be appreciated only through intuition and are best communicated by images and stories. Prediction is not possible in such cases. (IPCC-SRES, 2000:23)
2.1.2.4 | Choosing an estimation method Three main methods have been used to estimate the finance necessary to reach the MDGs:
1
Unit-cost-based analyses or ‘back of an envelope’ estimates: An average unit cost of action is identified in relation to the means selected to reach the target. It is then multiplied to reach the size of the targeted population. Most such studies are country-specific and sector- specific and rely on the availability of micro-data.
An example of using unit costs is Delamonica
et al. (2001), who divide countries’ current expenditure on primary education by the number of pupils in order to obtain a cost per pupil. This
‘unit cost’ is then multiplied by the incremental number of children who need to attend primary school in order to meet MDG2 by 2015. The Millennium Project (2005) uses this approach, producing an aggregate estimate of the cost of meeting the MDGs at the country level, based on the preliminary needs assessments carried out in Bangladesh, Cambodia, Ghana, Tanzania and Uganda.
Reddy and Heuty (2006) highlight the lack of consistency regarding the concept of unit cost in the financial estimates for reaching MDGs.
Unit costs may change, for example, in relation to potential economies or diseconomies of scale over time and across countries. Marginal costs can change and exogenous factors such as the development of new technology or institutions can also have a major influence on the cost of achieving the objective. Unit costs also vary by location, casting doubt on the possibility of aggregating detailed unit costs for certain locations into national or regional, let alone global, unit costs.
The problem posed by using unit costs can also be seen by comparing them across a range of studies for the same country. In the case of education, this shows that a major source of variation in estimates of finance needs lies in the estimated cost per pupil, which could differ by a factor of five even in the same country (Uganda) at a similar time: $13 (UNICEF, 1998), $27.50 (World Bank, 2003), $46 (EPRC, 2001) and $63 (Millennium Project, 2005).
2
Growth models, most of which are backed by the standard Harrod–Domar model, estimate a target (e.g. growth rates) and, by making assumptions about several macro trends based on historical evidence, calculate the finance need ‘backwards’.
A number of studies use this theoretical framework to assess the resources needed to
achieve the level of growth that would, in theory, make it possible to achieve the MDGs. One of the most cited examples is Devarajan et al. (2002), but more recent analyses such as Atisophon et al.
(2011) and OECD (2011) also adopt this approach.
Section 2.1.1 sets out a critique of this two-gap model. In addition, each individual link can be criticised. For example, Reddy and Heuty (2006) highlight the implausibility of using the same growth to poverty elasticity’s across all countries, given that the poverty elasticity of growth varies by country and over time, depending in part on economic structures, complementary policies and institutions.
3
CGEs such as the MAMS models (see Box 2.1) developed by World Bank researchers (Bourguignon et al., 2008), and used extensively by UNDESA (Sánchez et al., 2013), are macro models that include a cost-minimising government as the main agent acting on different sectors (health, education etc.). CGE models aim to reconcile the standard micro-based needs assessment with macroeconomic modelling. They capture micro–macro spillovers via fluctuations in wages. Moreover, by introducing intermediate goods that can be bought on foreign markets, these models allow for the effects of exchange- rate fluctuations. An important feature is the inclusion of decreasing marginal returns on additional government spending, meant to capture the ‘absorptive capacity’ threshold in a more satisfactory way than in other costing methodologies. The MAMS methodology focuses primarily on the education and health- related MDGs. It also takes into account the fact that there may be cross-sectoral spillovers, on which the effectiveness of government spending depends: for example, spending on education becomes more effective if at the same time there is an improvement in health conditions, because absenteeism is reduced, or if infrastructure improves, and vice versa. This ‘joint production’
of MDGs is not incorporated in the global sectoral estimates that use unit-cost approaches based on
simply adding up separate sector estimates. This suggests that the country-level MAMS models improve on the earlier estimates and constitute an important point of learning.
2.1.2.5 | Other conceptual issues across all models
While models have become more sophisticated, the estimates share common conceptual and implementation limitations. One issue relates to data reliability and robustness of estimations, especially in data-intensive models such as MAMS, but also in more straightforward calculations such as ICOR growth models. All these methodologies rely on estimates of unit costs, which are sensitive to data-collection issues, and the models’
predictive capacity relies on the relative fit of existing data to reality. This may be problematic in the context of countries where the accuracy of data, especially at the aggregate economic level, may be far from comprehensive or reliable.
Nonetheless, estimates at the country level tend to be sounder than estimates at the global level.
Another issue relates to aggregation, double counting, and the consideration of spillover effects across the targets. Interdependency among targets, double counting, and trade-offs are only partially addressed by the most recent models, such as CGEs. The literature on cost includes various scales of analysis, from one target in a sub-sector in a single country to multiple global targets. As highlighted by Devarajan et al. (2002), the estimated finance gap to achieve MDG1 should not be added to sectoral estimates of other MDGs but should be compared, since they are two estimates of the additional global ODA necessary to achieve the MDGs, based on the underlying assumption that achieving MDGs 2–7 is one way to achieve MDG1.
In addition to these concerns, the existing models have not really taken into account the role of ‘soft’ skills, such as good governance and institutions, in defining the relative effectiveness
of finance, which is crucial to costing. Widespread corruption, for example, may jeopardise the effectiveness of additional ODA, and limit the country’s ability to mobilise domestic resources or to undertake fiscal reform. These variables, which authors as Devarajan et al. (2002) stress are crucial to the achievement of development goals, are not taken into account by existing models, other than through relative ‘productivity’ parameters or residuals in econometric estimations. This key criticism is developed in this Report and relates to the importance of complementary policies (addressed in Chapter 5).
Those issues, and particularly the issue of double counting, spillovers and institutional environment, are often mentioned in relation to the scale of the cost estimates. There is general consensus on the limits of costing methodologies at the global level, with a call for more country-focused analyses because these address some of the shortcomings. A more disaggregated level of analysis makes it possible to tackle concerns about country-specific capacity and needs, although it is still subject to many other caveats. A country- level approach does, however, make it easier to identify accurate targets and pathways to reach them, and makes it possible to take into account the institutional environment, thus providing more refined estimates of unit and marginal costs and benefits of action. Recently, a number of MAMS applications have built country-level models (e.g.
Sánchez et al., 2010).
In conclusion, researchers sometimes (have to) make unrealistic assumptions and follow methodologies that may be subject to criticism, but could still be improved upon by examining the issues in greater depth and giving more detailed consideration to the role of context for the effective use of finance. There has been learning in the development of country-level MAMS modelling, which offers some advances over global estimations. In particular MAMS models can take into account country-level specificities;
address the linkages across individual MDGs; and
compare different types of financing for the same objective, e.g. domestic versus foreign. Although MAMS models offer a promising route, they have not to date fully assessed the importance of context (Box 2.4 describes two commissioned MAMS studies that aim to do this: Levin 2015a, 2015b). Better data and more disaggregation might solve some of the problems associated with the finance needs studies, but would not address the need for an alternative FFD vision that explicitly considers the policy context and the role of finance for the enablers of sustainable development transformation – a Finance and Policy Framework for Development (FPFD).
Three MAMS modelling studies were commissioned for this Report to consider the role of finance and policy in the context of the MDGs. MAMS modelling for Tanzania and Bangladesh simulates whether it would be possible to achieve the MDGs in an extended timeframe (2025 and 2021 respectively) (commissioned modelling papers: Levin, 2015a, 2015b; Box 2.5). MAMS modelling for Moldova simulates infrastructural investment (commissioned modelling paper; Kinnunen, 2015). It finds that market-related reforms will lead to greater benefits for the population and inclusive growth when public measures are taken to redistribute the gains of growth (see Chapter 6). Modelling was also undertaken on the impacts of Basel III implementation, quantitative easing and tackling tax evasion, using the NiGEM Model (commissioned background paper; Fic, 2015) and on the beneficial interaction between government effectiveness and the effectiveness of ODA and FDI flows, using the International Futures Model (see Box 4.3; commissioned modelling paper, Lenhardt, 2015). The results clearly demonstrate the importance of the policy context and are further explored below and in Chapter 4.