This version - DROdro.deakin.edu.au/eserv/DU:30051863/feeny-scalingup-2010.pdf · both nationally...

This is the published version: Feeny,SimonandMcGillivray,Mark2010,Scaling‐upforeignaid:willthe'BigPush'work?,AlfredDeakinResearchInstitute,Geelong,Vic.

Available from Deakin Research Online: http://hdl.handle.net/10536/DRO/DU:30051863Reproducedwiththekindpermissionofthecopyrightowner.Copyright:2010,DeakinUniversity,AlfredDeakinResearchInstitute

WORKING PAPER N0 05Scaling-up foreign aid:

Will the ‘Big Push’ work? Simon Feeny and Mark McGillivray

Alfred Deakin Research InstituteDeakin UniversityGeelong Waterfront CampusVictoria 3217 AustraliaTel: +61 3 5227 8028Fax: +61 3 5227 8650

WWW.DEAKIn.EDu.Au/AlfrED-DEAKIn-rEsEArch-InstItutE

WORKING PAPER N0 05Scaling-up foreign aid:

Will the ‘Big Push’ work?Simon Feeny and Mark McGillivray

SERIES EDITOR Kristina Murphy

ALFRED DEAKIN RESEARCH INSTITUTE Deakin University Geelong VIC 3217

AUSTRALIA

ISBN 978-1-921745-04-1

ISSN (online) 1837-7440 ISSN (print) 1837-7432

APRIL 2010

© Alfred Deakin Research Institute, Deakin University

National Library of Australia Cataloguing-in-Publication data: Feeny, S. (Simon). Scaling-up foreign aid: Will the ‘Big Push’ work?

Bibliography

ISBN 978-1-921745-04-1

Foreign Aid. 2. Growth. 3. Poverty. 4. Absorptive capacity. I. Feeny, S. (Simon). II. Alfred Deakin Research Institute. III. Title. (Series: Alfred Deakin Research Institute; Working Paper no. 5).

338.9194

Disclaimer

This article has been written as part of a series of publications issued from the Alfred Deakin Research Institute. The views contained in this article are representative of the author only. The publishing of this article does not constitute an endorsement of or any other expression of opinion by Deakin University. Deakin University does not accept any loss, damage or injury howsoever arising that may result from this article.

discoveryapplication

excellence 3

THE ALFRED DEAKIN RESEARCH INSTITUTE WORKING PAPER SERIESThe Alfred Deakin Research Institute (ADRI) is a specialised research unit that was established at Deakin University in 2009 to generate research that informs public debate and enables government ministers, departments and policy-makers to take action based on evidence.

This series of working papers is designed to bring the research of the Institute to as wide an audience as possible and to promote discussion among researchers, academics and practitioners both nationally and internationally on issues of importance.

The working papers are selected with three criteria in mind: (1) to share knowledge, experience and preliminary findings from research projects; (2) to provide an outlet for policy focused research and discussion papers; and (3) to give ready access to previews of papers destined for publication in academic journals, edited collections, or research monographs.

Founding Series Editor and Coordinator:

Kristina Murphy

Editorial Team:

David Lowe

Mark McGillivray

Jonathan Ritchie

THE ALFRED DEAKIN RESEARCH INSTITUTE WORKING PAPERSNo. 01 Lowe, D. The Colombo Plan and ‘soft’ regionalism in the Asia-Pacific: Australian and

New Zealand cultural diplomacy in the 1950s and 1960s. April 2010.

No. 02 Murphy, K. and Cherney, A. Policing ethnic minority groups with procedural justice: An empirical study. April 2010.

No. 03 Ritchie, J. ‘We need one district government to be set up to replace other district governments’: The beginnings of provincial government in Papua New Guinea. April 2010.

No. 04 Murphy, B. and Murphy, K. ‘The Australian Tax Survey of Tax Scheme Investors’: Survey methodology and preliminary findings for the second stage follow-up survey. April 2010.

No. 05 Feeny, S. and McGillivray, M. Scaling-up foreign aid: Will the ‘Big Push’ work? April 2010.


excellence 5

Scaling-up foreign aid: Will the ‘Big Push’ work?

Simon FeenySChool oF ECoNoMICS, FINANCE ANd MArkETING, rMIT UNIvErSITy

Mark McGillivrayAlFrEd dEAkIN rESEArCh INSTITUTE, dEAkIN UNIvErSITy

A b s t r Ac t

International donors are substantially scaling up aid. At the same time,

they have reservations over how much aid recipient countries can use

effectively. Such concerns are supported by the aid effectiveness literature

which finds that there are limits to the amounts of aid recipients can

efficiently absorb. This paper demonstrates that a ‘big push’ in foreign aid

will not lead to diminishing returns as long as donors get the inter-country

allocation of aid right. This is true even if donors provide aid at levels equal

to the well known target of 0.7 per cent of their Gross National Income.

> mark .mcgi l l ivray@deak in.edu.au

INSTITUTE

NRESEARCHINSTITUTERESEARCH

NRESEARCH

NE

NDEAKIALFRED

6 ALFRED DEAKIN RESEARCH INSTITUTE WORKING PAPER SERIES

I n t r o D u c t I o n

The now rather large aid effectiveness literature suggests that on average, foreign aid is effective at spurring economic growth in recipient countries (see McGillivray et al, 2006 for a recent survey). A consistent finding from the literature, however, is that the aid-growth relationship is subject to diminishing and eventually negative returns (see for example, Hansen and Tarp, 2000, 2001, Lensink and White, 2001, Dalgaard and Hansen, 2001, Hudson and Mosley, 2001, Dalgaard et al, 2004, Roodman, 2004 and Clemens et al, 2004). The finding demonstrates that there limits to the amounts of foreign aid that recipient countries can effectively absorb and, a doubling of aid to Africa (for example) would not double its impact (Collier, 2006).

This issue is of great importance given the dramatic scaling up of foreign aid being embarked upon by donors. Official Development Assistance (ODA), increased to an estimated record high of US$106 billion in 2005 before falling slightly to about US$104 billion in 2006. ODA is expected to increase further to US$130 billion by 2010 (OECD, 2006a). In line with much of the aid effectiveness research, the international community has reservations that some countries might struggle to use additional aid flows effectively. It is imperative that the effectiveness of aid in promoting growth and reducing poverty is not impaired by over-aiding recipients relative to their absorptive capacities.

Current orthodoxy on this issue within research and policy circles is that the inter-recipient allocation of aid should be such that it maximises its positive development impact in recipient countries. An efficient inter-recipient aid allocation is one that equalizes the marginal impact of aid across all recipients. In the unlikely event of donor budgets being sufficiently large, the total impact of aid is maximised globally if the marginal impact in all countries falls to zero. This is a basic message from the well-known Collier and Dollar (2001, 2002) ‘poverty efficient’ aid allocation model.

It is clear that the level of foreign aid that maximises per capita income growth for a given aid budget is of huge importance to aid donors, if foreign aid is to be allocated in a manner to maximise its impact on growth, poverty reduction and progress towards the achievement of the Millennium Development Goals (MDGs). This paper uses the results from existing aid effectiveness studies to estimate the level of aid which maximises the rate of per capita income growth in recipient countries. Ideally aid effectiveness studies would reveal (directly) the relationship between aid and poverty reduction. Unfortunately, this is not the case since data relating to poverty are unreliable and not available for most countries. Estimates of the level of aid at which its incremental impact on recipient country growth diminishes vary, depending inter alia on the time period and sample of countries under consideration, but it typically seems that this occurs at around 20 percent of recipient GDP. The paper terms this level of aid as the ‘growth efficient’ level. This paper examines the levels of aid which recipients currently receive and compares these levels to the growth efficient level. While recognising that donors have political and other motives and considerations in adjusting aid flows to developing countries, the paper examines the gains in per capita income growth that could be achieved if aid was allocated in accordance with a growth efficient aid allocation strategy subject to various world aid budgets.

In adopting this approach, it becomes clear that some countries are seemingly over-aided. 16 countries received, on average, aid at levels which exceed the growth efficient level during the period 2002-05. With the rise in global aid budgets many more countries are likely to be added to this list. If donors are concerned with using aid to maximise growth, they either need to cut back aid to these countries or very quickly work to increase their absorptive capacities. However, the main finding from this paper is that donors do not need to be concerned with diminishing returns to aid as long as they allocate aid according to a growth efficient strategy. This is true even if donors provide aid at levels equal to the well known target of 0.7 per cent of their Gross National Income.


excellence 7

This is true with a doubling of aid and even if donors provide aid at levels equal to the well known target of 0.7 per cent of their Gross National Income (GNI). While this is clearly not an argument for not building capacity in developing countries, it is an argument for getting the inter-country allocation of aid right.

The remainder of the paper is organised as follows. Section II reviews the approach of the empirical aid effectiveness literature and examines the relationship between aid and per capita income growth. In Section III, the paper discusses the rise in global aid budgets. In Section IV, the paper obtains empirical estimates of growth efficient levels of aid and compares these levels to actual allocations in 2005. Section V concludes with the policy implications arising from the research. The appendix to this paper extends the analysis undertaken in Section II by examining the behavioural relationship between aid and the level of per capita income (in addition to growth in per capita income).

the relationship between foreign aid and economic growth

The aid effectiveness has come a long way during the last decade, with researchers benefitting from larger datasets and more advanced econometric techniques. After many decades of ambiguous findings, virtually all studies during the course of the last ten years, have found that growth would be lower in the absence of aid.1 While this overall message from the literature is clear, disagreements over the environments in which aid works best remain and the process by which aid drives growth is still unclear.

Recent aid effectiveness studies have commonly allowed for a non-linear relationship between foreign aid and growth in their empirical models. They typically estimate the following equation:

á â â â ì2 /1 2 3 1,...,i i i i ig a a Z i n= + + + + = (1)

where gi is per capita GDP growth in aid receiving country i, ai is aid relative to GDP to that country, Zi is a vector of additional variables, μi is a residual, α is a constant and β1, β2 and β3 are coefficients. The vector of additional variables (Z) typically includes the initial level of GDP per capita, measures of ethnic fractionalization, the number of assassinations, institutional quality, macroeconomic policy indicators and region dummies. The expected signs of β1 and β2 are positive and negative, respectively. The hypothesized relationship between aid and growth described by (1), given these expectations, is diagrammatically represented in Figure 1 below.

The figure depicts the expected relationship between aid provided to a recipient (ai) and the growth in the recipient’s per capita income (gi). The incremental impact of foreign aid on growth is positive at all levels of aid up to ai*. This level is interpreted in the literature as the saturation level of aid, which corresponds to the unconstrained optimal level of aid according the poverty efficient approach. This paper terms this level the growth efficient level of foreign aid. The negative incremental impact on growth beyond this level of aid is rightly associated with absorptive capacity constraints within recipient countries. It is obvious from Figure 1 that if donors want to maximize gi, they should only provide aid up to and not beyond ai* in all countries. 2

1 Rajan and Subramanian (2005) provide an exception. They find little robust evidence of a positive (or negative) relationship between aid and growth. This represents a minority finding with over 50 other recent studies concluding that aid is effective at spurring economic growth. 2 Figure 1 takes the same form as the well-known Kuznets curve, representing the relationship between development and inequality. Kuznets is sometimes criticised for drawing the relationship from cross-sectional data rather than from time-series data. It is important to note that the relationship in Figure 1 is drawn from cross-sectional time-series (panel) data. Studies typically span the period from the late 1970s to 2001, with levels of aid to GDP remaining fairly constant over time. Although some studies have examined links between aid and growth using time series data for individual countries (see, for example, Gounder, 2001 and Feeny, 2006). No evidence of diminishing returns between aid and growth is presented in these studies despite the strong theoretical grounds for the relationship. An explanation is that there is insufficient variation in aid flows over time around the threshold to empirically capture the relationship.

INSTITUTE


NRESEARCH

NE

NDEAKIALFRED


There are a number of explanations underlying the diminishing returns relationship. The broad literature on aid points to many deficiencies in the delivery of aid which could restrict the capacity of recipients to productively absorb additional inflows. High levels of aid place a huge administrative burden on recipients with public sector officials facing excessive negotiation, management and reporting requirements. This is particularly true in the presence of a high level of donor proliferation. In the presence of many donors operating in many different sectors, aid projects and programmes are often uncoordinated and different reporting procedures will have to be adhered to. Other explanations relate to diminishing returns to investment/capital (which aid often funds) and Dutch disease effects whereby high levels of aid have an adverse impact on the export competitiveness of developing countries by causing an appreciation of the local currency.

FIgure 1: the IncrementAl ImpAct oF ForeIgn AID on economIc growth

From equation (1) it follows that:

(2)

The appendix to this paper examines the behavioural relationship between aid and income levels in addition to income growth. However, at this point it is important to recognise that the analysis relates to the impact of aid on economic growth rather than the overarching objective of foreign aid: which is poverty reduction. The paper notes that aid can influence many of the determinants of poverty and it follows that the level of aid that maximises its impact on poverty reduction is not necessarily that which maximises its impact on economic growth. Donors might, for example, want to target particularly poor groups in certain countries. They might also want to support particular public expenditures that have a pro-poor orientation. Such interventions might do a lot for poverty reduction, but might not be optimal in terms of current growth at the national level. As noted in the introduction, a paucity of data has prevented extensive examinations of the relationship between aid and poverty. Alternatively, donors might want to establish preconditions for growth and poverty reduction in periods later than the current one, on the grounds that growth in the short run might not be sustained over the long run. They might, for example, want to prevent a country sliding back into armed conflict. Donors might also want to rebuild physical and human infrastructure in countries in which these variables have reached a critically low level.

ai*

gi

ai0

Growth efficient aid

- 5 -

From equation (1) it follows that:

( )1

2

β

2β*ia

−= (2)

The appendix to this paper examines the behavioural relationship between aid

and income levels in addition to income growth. However, at this point it is important to

recognise that the analysis relates to the impact of aid on economic growth rather than

the overarching objective of foreign aid: which is poverty reduction. The paper notes that

aid can influence many of the determinants of poverty and it follows that the level of aid

that maximises its impact on poverty reduction is not necessarily that which maximises

its impact on economic growth. Donors might, for example, want to target particularly

poor groups in certain countries. They might also want to support particular public

expenditures that have a pro-poor orientation. Such interventions might do a lot for

poverty reduction, but might not be optimal in terms of current growth at the national

level. As noted in the introduction, a paucity of data has prevented extensive

examinations of the relationship between aid and poverty. Alternatively, donors might

want to establish preconditions for growth and poverty reduction in periods later than the

current one, on the grounds that growth in the short run might not be sustained over the

long run. They might, for example, want to prevent a country sliding back into armed

conflict. Donors might also want to rebuild physical and human infrastructure in

countries in which these variables have reached a critically low level.

These recognitions provide recipient country-specific cases for allocating aid in

excess of the growth efficient levels. Donors might have information on the relationship

between aid and growth that is recipient country-specific. Recognising that the growth

efficient levels depicted above are derived from average behavioural relationships, drawn

from cross-country data, this provides a powerful, additional case for deviations from

them. Given, however, that any deviation from growth efficient levels represents an

opportunity cost in terms of forgone current growth, donors would need to examine

closely any large deviations if they are concerned with growth, seeking to ascertain

whether they can be justified on valid poverty reduction criteria.

Empirical research consistently finds that the relationship between aid and

growth is U-shaped, with β1 and β2 being significantly different from zero and being

positive and negatively, respectively. The actual level of ai* is obviously of huge

importance to aid donors to ensure that their aid is effective. Table 1 provides the

estimates of ai* from well-cited existing aid effectiveness studies.


excellence 9

These recognitions provide recipient country-specific cases for allocating aid in excess of the growth efficient levels. Donors might have information on the relationship between aid and growth that is recipient country-specific. Recognising that the growth efficient levels depicted above are derived from average behavioural relationships, drawn from cross-country data, this provides a powerful, additional case for deviations from them. Given, however, that any deviation from growth efficient levels represents an opportunity cost in terms of forgone current growth, donors would need to examine closely any large deviations if they are concerned with growth, seeking to ascertain whether they can be justified on valid poverty reduction criteria.

Empirical research consistently finds that the relationship between aid and growth is U-shaped, with β1 and β2 being significantly different from zero and being positive and negatively, respectively. The actual level of ai* is obviously of huge importance to aid donors to ensure that their aid is effective. Table 1 provides the estimates of ai* from well-cited existing aid effectiveness studies.

tAble 1: estImAtes oF growth AnD Income eFFIcIent AID From prevIous stuDIes

Study β1 β2 ai*

Durbarry et al (1998) 0.119 -0.001 41.9

Hansen and Tarp (2000) 0.207 -0.004 24.0

Hansen and Tarp (2001) 0.199 -0.606 17.0

Lensink and White (2001) 0.174 -0.002 49.6

Dalgaard and Hansen (2001) 1.340 -0.127 5.3

Hudson and Mosley (2001) 0.815 -0.056 6.7

Dalgaard et al (2004) 1.910 -0.115 4.8

Roodman (2004) 0.385 -0.010 19.3

Clemens et al (2004) 0.960 -0.059 18.1

Average 20.74

Notes: Coefficients are based on average coefficients reported in the studies. Estimates of ai* are based on total foreign aid and are adjusted if the study includes an aid interaction term. Dalgaard and Hansen (2001) and Dalgaard et al (2004) use Effective Development Assistance (EDA) as the measure of aid. EDA is the sum of aid grants and the grant equivalent of aid loans. All other studies use Official Development Assistance (ODA). Further details are provided in Table A1 in the appendix.

Table 1 shows considerable variation in the estimates of ai*. This is not surprising given the different datasets, time periods and econometric techniques used by researchers. For the analysis undertaken in this paper we take an average. There is no reason to assume that one study has more accurately estimated the diminishing returns relationship than other studies. The average value of ai*, the level of aid to GDP which maximizes growth is 20.74 per cent.3 The lowest estimates of ai* are from studies using EDA as the measure of aid rather than the traditional measure (ODA). Note that if these studies are omitted from the analysis the average estimate of ai* is higher and even fewer countries would be receiving levels of aid in excess of the growth efficient levels. The upper and lower bounds of a 95 per cent confidence interval around our estimate of ai* are 15.3 and 26.2 per cent. The implications of this range in estimates are discussed further below.

3 A similar estimate of ai* is obtained if we take a trimmed mean, omitting the largest and smallest values. These values are closest to those reported by Hansen and Tarp (2000). Therefore the coefficients used in calculating the impact of aid on growth are taken from one of their regressions: β1 = 0.165 and β2=-0.004. The lowest estimates of ai* are from studies using EDA as the measure of aid. Note that if these studies are omitted from our analysis the average estimate of ai* is higher and even fewer countries would be receiving levels of aid in excess of the growth efficient levels.

INSTITUTE


NRESEARCH

NE

NDEAKIALFRED


global aid budgets

While international aid donors have started scaling up their aid programs, there is some uncertainty over the level that global flows will increase to. At the International Conference on Financing for Development held in Monterrey, Mexico in March 2002, a number of governments reasserted the commitment for donor countries to achieve an ODA to GNI target of 0.7 per cent. This level of aid has been a long standing development target. The United Nations Millennium Project calculates that if countries follow through on their commitments to increase aid to this level, the MDGs could be achieved by 2015. However, while some donors have already exceeded the 0.7 target other donors have failed to fully commit to the target despite pledging to increase their level of aid.

Figure 2 plots the level of ODA in real US dollar amounts provided by members of the OECD’s Development Assistance Committee (DAC) since 1990. The figure shows that although aid flows fell slightly during the first half of the 1990s, since 1997, the level of ODA has steadily increased to almost $80 billion in 2004. Further, the OECD projects that ODA is likely to amount to $130 billion in 2010 (based on recent pledges by donors). This implies that aid flows will have more than doubled since 1997. Foreign aid flows from non-OECD members are also likely to increase substantially during the next decade. This paper undertakes a number of simulations to examine whether the scaling up of aid will induce diminishing returns and hamper its effectiveness.

FIgure 2: oDA From the oecD countrIes 1990 to 2010

Source: OECD (2006a)

Inter-country allocation of foreign aid

Table A2 in the appendix to the paper shows the difference between the level of aid (on average) that countries received during the period 2002-05 and ai*. It also provides the impact of aid on growth and amount of additional growth that could be achieved if the level of aid was increased/decreased to ai*. On average, countries receive a level of aid which accounts for 8.9 per cent of GDP, a level much lower than the growth efficient level. The vast majority of countries listed in Table A2 can receive substantially larger amounts of aid without negative growth returns setting in. However, 16 countries receive aid at levels higher than ai*. Ten of these countries are located in Africa, four

- 7 -

International Conference on Financing for Development held in Monterrey, Mexico in

March 2002, a number of governments reasserted the commitment for donor countries

to achieve an ODA to GNI target of 0.7 per cent. This level of aid has been a long

standing development target. The United Nations Millennium Project calculates that if

countries follow through on their commitments to increase aid to this level, the MDGs

could be achieved by 2015. However, while some donors have already exceeded the

0.7 target other donors have failed to fully commit to the target despite pledging to

increase their level of aid.

Figure 2 plots the level of ODA in real US dollar amounts provided by

members of the OECD’s Development Assistance Committee (DAC) since 1990. The

figure shows that although aid flows fell slightly during the first half of the 1990s,

since 1997, the level of ODA has steadily increased to almost $80 billion in 2004.

Further, the OECD projects that ODA is likely to amount to $130 billion in 2010

(based on recent pledges by donors). This implies that aid flows will have more than

doubled since 1997. Foreign aid flows from non-OECD members are also likely to

increase substantially during the next decade. This paper undertakes a number of

simulations to examine whether the scaling up of aid will induce diminishing returns

and hamper its effectiveness.

Figure 2: ODA from the OECD countries 1990 to 2010

Source: OECD (2006a)

0

20

40

60

80

100

120

140

1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010

OD

A (

20

04

US

$ b

illi

on

)


excellence 11

in the Pacific and two in Asia. Two of these countries (Malawi and Rwanda) would no longer be classified as ‘over-aided’ if we used the upper estimate of ai* within the 95 per cent confidence interval while a further ten countries would be included if we used the lower estimate.

Table A2 also indicates that when aid is provided at growth efficient levels, its marginal impact on economic growth is 1.7 per cent. The incremental impact of aid on growth would be an average 1.1 per cent higher if aid allocations were equal to ai*. The largest gains would be achieved by scaling back aid to those countries with levels currently exceeding ai*.

It is possible to execute some simple simulations of the growth impact of increasing foreign aid. All simulations assume that the inter-country allocation of foreign aid remains the same. If we assume that developing country GDP remained unchanged but doubled the amount of foreign aid to each recipient (from 2002-05 levels) then a total of 41 countries would receive levels of aid exceeding growth efficient aid. Results are provided in column 8 of Table A2. Note that increasing aid to this level implies a larger total aid budget than the US$130 billion project by the OECD for 2010. Moreover, if we assume that all DAC donors provided 0.7 per cent of GNI in aid and allocated the same proportion of this aid to each developing country recipient, a total of 49 countries could be deemed over-aided4. A summary of these scenarios and how many countries will become over-aided are provided in Table 2 below. The implication of using the lower and upper bound estimates of a* are also provided. Clearly, unless donors consider absorptive capacity constraints when scaling up, diminishing returns to aid will be experienced in a large number of countries. The table also shows how large aid flows would be if they were allocated according to a growth efficient strategy and scaled up to account for 20 per cent of GDP in all developing countries.

tAble 2: scAlIng up ForeIgn AID – key results

Global aid budget Value of ODA (US$, billions)

Number of over-aided countries (ai> ai*)

ODA- 2002-05 levelLower bound ai*Upper bound ai*

78.5 162614

Doubling aid (from 2002-05 level)Lower bound ai*Upper bound ai*

157 414930

ODA 0.7% of DAC GNILower bound ai*Upper bound ai*

207 496242

ODA if all recipients receive ai*

ODA if all recipients receive ai* (excluding China and India)

1,466

922

-

-

The clear message from this analysis is that aid flows can increase dramatically with large returns in terms of growth and poverty reduction. In fact aid flows can increase to levels far in excess of those implied by an outright doubling of aid, or where it accounts for 0.7 per cent of donor GNI and still not necessarily lead to diminishing returns. ODA would need to increase to US$1,466 billion to account for 20 per cent of GDP in all developing countries. The figure falls to US$922 billion if China and India are excluded from the calculation. This figure is still far higher than the levels of aid currently being called for by pressure groups, think-tanks and international organisations.

4 DAC countries provided, an average of 0.265 per cent of their GNI in ODA during the period 2002-05.

INSTITUTE


NRESEARCH

NE

NDEAKIALFRED


The paper recognises such levels are unpractical and that donors should not strive to scale up aid to such an extent in many countries. However, undertaking this exercise does indicate that aid flows could increase dramatically without leading to diminishing returns in recipient countries if they were allocated according to a growth efficient strategy. This is true even if we use our lower confidence bound estimate of ai* as the growth efficient level of foreign aid.

conclusion and policy recommendations

This paper indicates that aid budgets can increase dramatically without inducing diminishing returns on per capita income growth. Therefore a ‘big push’ in foreign aid levels can lead to important increases in economic growth and reductions in poverty. This is one of the main points of the paper. The other is of the importance of getting the inter-recipient allocation of aid right. The paper demonstrates that aid flows can increase to levels far more than any currently being proposed and still not induce diminishing returns if it is allocated across recipients up to the point where its marginal impact on growth is zero. However, this should not be used as an argument against donors increasing the efficiency of foreign by working towards the Paris Declaration principles and improving the absorptive capacity of aid recipients.

Specifically, using findings from well-cited existing studies, the paper estimates that the level of aid which maximises per capita income growth in recipient countries is when it accounts for 20.7 per cent of a recipient’s GDP. This level of aid is termed the growth efficient level. The paper identifies 16 developing countries which received average aid levels which exceed the growth efficient level during the period 2202-05. Donors should be aware that reducing aid to these recipients could yield higher marginal impact on per capita income growth. This needs be supplemented with country–specific information and the current environment in which the aid program operates. If countries are clearly receiving levels of foreign aid that cannot be reasonably justified, donors could reduce the level of aid that they disburse to the recipient or use their aid to quickly increase the capacity of the recipient.5

The paper notes, however, that the estimates is provides should only be used as a rough guide for aid donors. Econometric models estimate relationships that apply on average across countries. It is also true that different types of foreign aid are likely to have different impacts and different thresholds with regard to economic growth. A better knowledge of these impacts will help inform donors of the correct mix when scaling up foreign aid amounts.

5 For example, with the exception of the Marshall Islands, all countries which received aid at levels in excess of the income efficient level in 2004, have recently experienced periods of civil unrest. Larger amounts of aid could be reasonably justified in these circumstances.


excellence 13

r e F e r e n c e s

Clemens, M., S. Radelet and R. Bhavani (2004), “Counting Chickens When They Hatch: The Short-Term Effect of Aid on Growth”, Center for Global Development Working Paper Number 44.

Collier, P. (2006), “Is Aid Oil? An Analysis of Whether Africa Can Absorb More Aid”, World development, Vol.34, No.9, pp.1482-1497.

Collier, P. and D. Dollar (2001), “Can the World Cut Poverty in Half? How Policy Reform and Effective Aid Can Meet International Development Goals”, World development, Vol.29, No.111, pp.1787-1802.

Collier, P and D. Dollar (2002), “Aid Allocation and Poverty Reduction”, European Economic review, Vol.46, pp.1475-1500.

Dalgaard, C-J and H Hansen (2001), “On Aid, Growth and Good Policies”, Journal of development Studies, Vol.37, No.6, pp.17–35.

Dalgaard, C-J, H. Hansen and F. Tarp (2004), “On the Empirics of Foreign Aid and Growth”, Economic Journal, Vol.114, pp.191-216.

Durbarry, R., Gemmell, N., Greenaway, D. (1998). New Evidence on the Impact of Foreign Aid on Economic Growth. CREDIT Research Paper No. 98/9. Nottingham: University of Nottingham.

Feeny, S. (2006), “Aid and Growth in Papua New Guinea”, Journal of development Studies, Vol.41, No.6, pp.1092-1117.

Gounder, R. (2001), “Aid-Growth Nexus: Empirical Evidence from Fiji”, Applied Economics, Vol.33, pp.1009-1019.

Hansen, H and F Tarp (2000), “Aid Effectiveness Disputed”, Journal of International development, Vol.12, pp.375–398.

Hansen, H and F Tarp (2001), “Aid and Growth Regressions”, Journal of development Economics, Vol.64, pp.547–570.

Hudson, J. and P. Mosley (2001), “Aid, Policies and Growth: In Search of the Holy Grail”, Journal of International development, Vol.13, pp.1023-1038.

Lensink, R and H White (2001), “Are There Negative Returns to Aid?”, Journal of development Studies, Vol.37, No.6, pp.42–64.

McGillivray, M, Feeny, S, Hermes, N. and R. Lensink (2006), “Controversies Over the Impact of Development Aid: It Works; It Doesn’t; It Can, But It Depends….”, Journal of International development, Vol.18, pp.1031-1050.

Organisation for Economic Co-operation and Development (OECD) (2006a), dACNews: Ideas on Aid, April, Paris, OECD.

Organisation for Economic Co-operation and Development (OECD) (2006b), International development Statistics on-line, Paris, OECD.

Rajan, R. and A. Subramanian (2005), Aid and Growth: What does the Cross-Country Evidence really Show?, IMF Working Paper No.127, International Monetary Fund, Washington.

Roodman, D. (2004), The Anarchy of Numbers: Aid, development and Cross-country Empirics, CGD Working Paper No. 32, July 2004, Centre for Global Development, Washington, D.C.

World Bank (2006), World development Indicators, World Bank, Washington.

INSTITUTE


NRESEARCH

NE

NDEAKIALFRED


A p p e n D I x

A key issue is whether the relationship depicted in Figure 1 is between foreign aid and short or long run per capita income growth. Theoretical growth models predict that foreign aid can impact on both short and long run growth. Early growth models (for example Harrod-Domar, 1946, Solow Swan, 1956) predict that aid impacts on short run growth by increasing the rate of savings. This will lead to an increase in the steady state income level. However these models (as well as the more recent endogenous growth models such as Romer, 1986, Lucas, 1988) recognise that foreign aid can impact on long run growth if it impacts on the level of technology or productivity. For example if foreign aid increases human capital, includes technically superior imports and learning by doing externalities, it can increase the long run growth rate.

The issue has important implications regarding the impact of foreign aid on income levels as well as on income growth. While the relationship between aid and per capita income growth is important, what is more important for reducing the number of people living in poverty (for given poverty-income elasticities), is to maximize the level of GDP per capita within country i. If the relationship between aid and long run economic growth is assumed to that depicted in Figure 1, the corresponding relationship between aid and the level of per capita income (yi) is provided in the second half of Figure A1 below. The level of aid that maximises per capita income is ai**. The paper terms this level of aid as the income efficient level. At this level the impact of aid on the rate of growth approaches zero. Clearly, while the incremental efficiency of aid on growth falls continually for all levels greater than ai*, aid remains effective, increasing the level of per capita GDP, up to ai**.

It is evident from Figure A1 that the income efficient amount of aid is twice the growth efficient amount. The important implication from this analysis is that if the estimated aid growth relationships provided in the literature can be interpreted as depicting a long-run equilibrium, donors wanting to maximize per capita incomes in recipient countries are able to provide far higher levels of foreign aid than previously thought. Donors should quickly increase aid to recipients up to the level that maximizes growth (ai*) and then steadily and incrementally increase it up to the level that maximizes the level of GDP per capita (ai**). Rapidly increasing aid levels to ai** is unlikely to work. It is important for donors to gradually increase the level of aid for aid recipients to experience all of the benefits from aid.


excellence 15

- 14 -

Figure A1: Aid, Economic Growth and Per Capita Income

T

T

From equation (1) recall that:

( )1

2

β

2β*ia

−= (3)

it follows that:

1

2

β

β**ia

−= (4)

ai*

yi

aiai**

0

ai*

gi

aiai**

0

Growth efficient aid

Incomeefficient aid

INSTITUTE


NRESEARCH

NE

NDEAKIALFRED


- 15 -

Table A1: Detailed Estimates of Growth and Income Efficient Aid from Previous Studies

Study and Estimation Technique β1 β2 ai*Durbarry et al (1998)Table 2: Two-way fixed effects model (OLS) 0.176 -0.002 44.8Table 3: GLS with regional dummies (GLS) 0.101 -0.001 41.0Appendix 1 Table 2: One-way fixed effects model (OLS) 0.105 -0.001 41.0Appendix 3 Table 2: Two-way random effects (GLS) 0.094 -0.001 40.6Average 0.119 -0.001 41.9Hansen and Tarp (2000)Table 4, Column 3 (OLS) 0.166 -0.003 27.7Table 4, Column 4 (OLS) 0.165 -0.004 20.6Table 4, Column 5 (OLS) 0.182 -0.004 22.8Appendix B Column 3 (IV) 0.236 -0.005 23.6Appendix B Column 4 (IV) 0.230 -0.005 23.0Appendix B Column 5 (IV) 0.265 -0.005 26.5Average 0.207 -0.004 24.0Hansen and Tarp (2001)Table 1, Column 1.1 (IV) 0.238 -0.754 15.8Table 1, Column 1.2 (IV) 0.241 -0.763 15.8Table 3, Column 3.1 fixed effects (OLS) 0.117 -0.301 19.4Average 0.199 -0.606 17.0Lensink and White (2001)Table 10, extreme bounds analysis (OLS) 0.174 -0.002 49.6Average 0.174 -0.002 49.6Dalgaard and Hansen (2001)Table 4, Column 7 (IV) 1.327 -0.126 5.3Table 4, Column 8 (IV) 1.352 -0.127 5.3Average 1.340 -0.127 5.3Hudson and Mosley (2001)Table 4, Column 1, (2SLS) 0.541 -0.039 6.8Table 4, Column 2, (WLS) 1.155 -0.068 8.5Table 4, Column 3, (2SLS) 0.950 -0.058 6.5Table 4, Column 4, (2SLS) 0.612 -0.060 5.1Average 0.815 -0.056 6.7Dalgaard et al (2004)Table 1, Column 3 (2SLS) 1.350 -0.130 5.2Table 1, Column 6 (2SLS) 2.470 -0.100 4.3Average 1.910 -0.115 4.8Roodman (2004)Table 4: Reproduction pf Hansen and Tarp (2001) (2SLS) 0.250 -0.010 12.5Table 12, With population included (2SLS) 0.520 -0.010 26.0Average 0.385 -0.010 19.3Clemens et al (2004)Table 4, Column 5 (2SLS) 0.960 -0.059 18.1Average 0.960 -0.059 18.1


excellence 17

-16

-

Ta

ble

A2

: Im

pa

cts

of

Gro

wth

Eff

icie

nt

Aid

Cou

ntr

yO

DA

/GD

P (

200

2-0

5)G

row

th

effi

cien

t ai

dG

row

th e

ffic

ien

t -

actu

alG

row

th im

pac

t of

cu

rren

t ai

d

Gro

wth

imp

act

of

grow

th e

ffic

ien

t ai

d

Gro

wth

gai

n

from

gro

wth

ef

fici

ent

aid

Scen

ario

1

Dou

blin

g A

id:

OD

A/G

DP

Scen

ario

2 D

AC

0

.7%

GN

I:

OD

A/G

DP

Sao

Tom

e an

d P

rin

cip

e52

.320

.7-3

1.6

-2.3

1.7

4.0

104.

513

7.9

Tim

or-L

este

52.1

20.7

-31.

4-2

.31.

74.

010

4.2

137.

5

Mic

ron

esia

, Fed

. Sts

.46

.020

.7-2

5.3

-0.9

1.7

2.6

92.0

121.

5

Mar

shal

l Isl

and

s42

.620

.7-2

1.9

-0.2

1.7

1.9

85.

211

2.4

Bu

run

di

41.9

20.7

-21.

2-0

.11.

71.

88

3.8

110

.6

Eri

trea

41.3

20.7

-20

.60

.01.

71.

78

2.6

109.

0

Con

go, D

em. R

ep.

41.1

20.7

-20

.40

.01.

71.

78

2.3

108

.6

Solo

mon

Isl

and

s39

.920

.7-1

9.2

0.2

1.7

1.5

79.8

105.

4

Afg

han

ista

n35

.920

.7-1

5.2

0.8

1.7

0.9

71.8

94.8

Gu

inea

-Bis

sau

35.8

20.7

-15.

10

.81.

70

.971

.694

.5

Sier

ra L

eon

e32

.520

.7-1

1.8

1.1

1.7

0.6

64.9

85.

7

Kir

ibat

i31

.620

.7-1

0.9

1.2

1.7

0.5

63.3

83.

5

Lib

eria

29.8

20.7

-9.1

1.4

1.7

0.3

59.6

78.6

Moz

ambi

que

26.9

20.7

-6.2

1.5

1.7

0.2

53.8

71.1

Mal

awi

25.7

20.7

-5.0

1.6

1.7

0.1

51.4

67.8

Rw

and

a23

.720

.7-3

.01.

71.

70

.047

.462

.5

Pal

au19

.220

.71.

51.

71.

70

.038

.350

.6

Nic

arag

ua

19.0

20.7

1.7

1.7

1.7

0.0

37.9

50.1

Eth

iop

ia18

.420

.72.

31.

71.

70

.036

.748

.5

Mau

rita

nia

16.3

20.7

4.4

1.6

1.7

0.1

32.7

43.2

Cap

e V

erd

e16

.120

.74.

61.

61.

70

.132

.242

.5

Mad

agas

car

16.0

20.7

4.7

1.6

1.7

0.1

32.0

42.3

Nig

er15

.920

.74.

81.

61.

70

.131

.942

.1

Zam

bia

15.9

20.7

4.8

1.6

1.7

0.1

31.8

42.0

Mon

goli

a15

.820

.74.

91.

61.

70

.131

.741

.8

Gam

bia,

Th

e15

.420

.75.

31.

61.

70

.130

.940

.7

Uga

nd

a14

.820

.75.

91.

61.

70

.129

.539

.0

INSTITUTE


NRESEARCH

NE

NDEAKIALFRED


-17

-

Cou

ntr

yO

DA

/GD

P (

200

2-0

5)G

row

th

effi

cien

t ai

dG

row

th e

ffic

ien

t -

actu

alG

row

th im

pac

t of

cu

rren

t ai

d

Gro

wth

imp

act

of

grow

th e

ffic

ien

t ai

d

Gro

wth

gai

n

from

gro

wth

ef

fici

ent

aid

Scen

ario

1

Dou

blin

g A

id:

OD

A/G

DP

Scen

ario

2 D

AC

0

.7%

GN

I:

OD

A/G

DP

Tan

zan

ia14

.320

.76.

41.

51.

70

.228

.537

.6

Ton

ga14

.220

.76.

51.

51.

70

.228

.537

.6

Gu

yan

a13

.920

.76.

81.

51.

70

.227

.836

.7

Bu

rkin

a F

aso

13.0

20.7

7.7

1.5

1.7

0.2

25.9

34.2

Mal

i12

.720

.78

.01.

41.

70

.325

.433

.5

Lao

PD

R12

.320

.78

.41.

41.

70

.324

.632

.4

Gh

ana

12.3

20.7

8.4

1.4

1.7

0.3

24.5

32.3

Van

uat

u11

.720

.79.

01.

41.

70

.323

.531

.0

Bh

uta

n11

.720

.79.

01.

41.

70

.323

.330

.8

Dji

bou

ti11

.620

.79.

11.

41.

70

.323

.130

.5

Kyr

gyz

Rep

ubl

ic11

.220

.79.

51.

31.

70

.422

.429

.5

Taj

ikis

tan

11.2

20.7

9.5

1.3

1.7

0.4

22.3

29.4

Sam

oa11

.020

.79.

71.

31.

70

.421

.928

.9

Con

go, R

ep.

10.6

20.7

10.1

1.3

1.7

0.4

21.1

27.9

Cam

bod

ia9.

920

.710

.81.

21.

70

.519

.926

.2

Sen

egal

9.7

20.7

11.0

1.2

1.7

0.5

19.3

25.5

Bol

ivia

8.7

20.7

12.0

1.1

1.7

0.6

17.4

23.0

Ben

in8

.520

.712

.21.

11.

70

.616

.922

.4

Ch

ad8

.220

.712

.51.

11.

70

.616

.321

.5

Com

oros

8.1

20.7

12.6

1.1

1.7

0.6

16.2

21.4

Dom

inic

a8

.020

.712

.71.

11.

70

.616

.021

.1

Hai

ti7.

920

.712

.81.

11.

70

.615

.820

.9

Hon

du

ras

7.5

20.7

13.2

1.0

1.7

0.7

15.1

19.9

Iraq

7.5

20.7

13.2

1.0

1.7

0.7

14.9

19.7

Arm

enia

7.2

20.7

13.5

1.0

1.7

0.7

14.5

19.1

Les

oth

o7.

220

.713

.51.

01.

70

.714

.419

.0

Bos

nia

an

d H

erze

govi

na

7.0

20.7

13.7

1.0

1.7

0.7

14.1

18.6

Jord

an6.

920

.713

.80

.91.

70

.813

.718

.1


excellence 19

-18

-

Cou

ntr

yO

DA

/GD

P (

200

2-0

5)G

row

th

effi

cien

t ai

dG

row

th e

ffic

ien

t -

actu

alG

row

th im

pac

t of

cu

rren

t ai

d

Gro

wth

imp

act

of

grow

th e

ffic

ien

t ai

d

Gro

wth

gai

n

from

gro

wth

ef

fici

ent

aid

Scen

ario

1

Dou

blin

g A

id:

OD

A/G

DP

Scen

ario

2 D

AC

0

.7%

GN

I:

OD

A/G

DP

Gu

inea

6.7

20.7

14.0

0.9

1.7

0.8

13.4

17.7

Nep

al6.

620

.714

.10

.91.

70

.813

.217

.4

Cen

tral

Afr

ican

Rep

ub

lic

6.4

20.7

14.3

0.9

1.7

0.8

12.9

17.0

Serb

ia a

nd

Mon

ten

egro

6.4

20.7

14.3

0.9

1.7

0.8

12.8

16.9

Mol

dov

a6.

220

.714

.50

.91.

70

.812

.516

.5

Pap

ua

New

Gu

inea

6.2

20.7

14.5

0.9

1.7

0.8

12.4

16.4

Geo

rgia

6.1

20.7

14.6

0.9

1.7

0.8

12.3

16.2

Mac

edon

ia, F

YR

5.2

20.7

15.5

0.8

1.7

0.9

10.4

13.8

Mal

div

es5.

020

.715

.70

.71.

71.

09.

913

.1

Alb

ania

4.9

20.7

15.8

0.7

1.7

1.0

9.9

13.0

Cam

eroo

n4.

820

.715

.90

.71.

71.

09.

612

.7

Sud

an4.

620

.716

.10

.71.

71.

09.

212

.2

Gre

nad

a4.

620

.716

.10

.71.

71.

09.

112

.0

Vie

tnam

3.9

20.7

16.8

0.6

1.7

1.1

7.9

10.4

Ken

ya3.

820

.716

.90

.61.

71.

17.

59.

9

Sri L

anka

3.5

20.7

17.2

0.5

1.7

1.2

7.0

9.2

Tog

o3.

420

.717

.30

.51.

71.

26.

89.

0

An

gola

3.2

20.7

17.5

0.5

1.7

1.2

6.4

8.4

Aze

rbai

jan

3.0

20.7

17.7

0.5

1.7

1.2

6.0

8.0

Nam

ibia

3.0

20.7

17.7

0.5

1.7

1.2

5.9

7.8

Yem

en, R

ep.

2.9

20.7

17.8

0.4

1.7

1.3

5.7

7.6

Cot

e d

'Ivo

ire

2.8

20.7

17.9

0.4

1.7

1.3

5.6

7.4

Nig

eria

2.8

20.7

17.9

0.4

1.7

1.3

5.5

7.3

Zim

babw

e2.

520

.718

.20

.41.

71.

35.

06.

5

Ban

glad

esh

2.3

20.7

18.4

0.4

1.7

1.3

4.7

6.2

Fij

i2.

320

.718

.40

.41.

71.

34.

66.

1

Suri

nam

e2.

020

.718

.70

.31.

71.

44.

05.

3

St. K

itts

an

d N

evis

2.0

20.7

18.7

0.3

1.7

1.4

4.0

5.3

INSTITUTE


NRESEARCH

NE

NDEAKIALFRED


-19

-

Cou

ntr

yO

DA

/GD

P (

200

2-0

5)G

row

th

effi

cien

t ai

dG

row

th e

ffic

ien

t -

actu

alG

row

th im

pac

t of

cu

rren

t ai

d

Gro

wth

imp

act

of

grow

th e

ffic

ien

t ai

d

Gro

wth

gai

n

from

gro

wth

ef

fici

ent

aid

Scen

ario

1

Dou

blin

g A

id:

OD

A/G

DP

Scen

ario

2 D

AC

0

.7%

GN

I:

OD

A/G

DP

Uzb

ekis

tan

1.8

20.7

18.9

0.3

1.7

1.4

3.5

4.6

Pak

ista

n1.

720

.719

.00

.31.

71.

43.

54.

6

Seyc

hel

les

1.7

20.7

19.0

0.3

1.7

1.4

3.3

4.4

St. V

ince

nt

and

th

e G

ren

adin

es1.

620

.719

.10

.31.

71.

43.

34.

3

Swaz

ilan

d1.

520

.719

.20

.21.

71.

53.

03.

9

Leb

anon

1.4

20.7

19.3

0.2

1.7

1.5

2.9

3.8

Egy

pt,

Ara

b R

ep.

1.4

20.7

19.3

0.2

1.7

1.5

2.7

3.6

El S

alva

dor

1.4

20.7

19.3

0.2

1.7

1.5

2.7

3.6

Bel

ize

1.3

20.7

19.4

0.2

1.7

1.5

2.7

3.5

Mor

occo

1.3

20.7

19.4

0.2

1.7

1.5

2.6

3.5

St. L

uci

a1.

320

.719

.40

.21.

71.

52.

53.

3

Tu

nis

ia1.

220

.719

.50

.21.

71.

52.

53.

3

Equ

ator

ial G

uin

ea1.

020

.719

.70

.21.

71.

51.

92.

5

Gu

atem

ala

0.9

20.7

19.8

0.1

1.7

1.6

1.8

2.4

An

tigu

a an

d B

arbu

da

0.9

20.7

19.8

0.1

1.7

1.6

1.8

2.3

Par

agu

ay0

.820

.719

.90

.11.

71.

61.

62.

2

Per

u0

.720

.720

.00

.11.

71.

61.

41.

8

Ph

ilip

pin

es0

.720

.720

.00

.11.

71.

61.

31.

8

Ecu

ador

0.6

20.7

20.1

0.1

1.7

1.6

1.2

1.6

Kaz

akh

stan

0.6

20.7

20.1

0.1

1.7

1.6

1.2

1.6

Col

ombi

a0

.620

.720

.10

.11.

71.

61.

21.

6

Gab

on0

.620

.720

.10

.11.

71.

61.

21.

5

Ind

ones

ia0

.620

.720

.10

.11.

71.

61.

21.

5

Bot

swan

a0

.520

.720

.20

.11.

71.

61.

11.

4

Tu

rkm

enis

tan

0.5

20.7

20.2

0.1

1.7

1.6

1.1

1.4

Bar

bad

os0

.520

.720

.20

.11.

71.

60

.91.

2

Dom

inic

an R

epu

bli

c0

.420

.720

.30

.11.

71.

60

.91.

2

Jam

aica

0.4

20.7

20.3

0.1

1.7

1.6

0.8

1.1


excellence 21

-20

-

Cou

ntr

yO

DA

/GD

P (

200

2-0

5)G

row

th

effi

cien

t ai

dG

row

th e

ffic

ien

t -

actu

alG

row

th im

pac

t of

cu

rren

t ai

d

Gro

wth

imp

act

of

grow

th e

ffic

ien

t ai

d

Gro

wth

gai

n

from

gro

wth

ef

fici

ent

aid

Scen

ario

1

Dou

blin

g A

id:

OD

A/G

DP

Scen

ario

2 D

AC

0

.7%

GN

I:

OD

A/G

DP

Syri

an A

rab

Rep

ubl

ic0

.420

.720

.30

.11.

71.

60

.81.

1

Alg

eria

0.4

20.7

20.3

0.1

1.7

1.6

0.8

1.1

Cro

atia

0.4

20.7

20.3

0.1

1.7

1.6

0.8

1.0

Mau

riti

us

0.4

20.7

20.3

0.1

1.7

1.6

0.7

0.9

Sou

th A

fric

a0

.320

.720

.40

.11.

71.

60

.70

.9

Om

an0

.320

.720

.40

.11.

71.

60

.60

.8

Ind

ia0

.220

.720

.50

.01.

71.

70

.40

.5

Pan

ama

0.2

20.7

20.5

0.0

1.7

1.7

0.3

0.5

Ukr

ain

e0

.220

.720

.50

.01.

71.

70

.30

.5

Uru

guay

0.1

20.7

20.6

0.0

1.7

1.7

0.2

0.3

Tu

rkey

0.1

20.7

20.6

0.0

1.7

1.7

0.2

0.3

Mal

aysi

a0

.120

.720

.60

.01.

71.

70

.20

.3

Cos

ta R

ica

0.1

20.7

20.6

0.0

1.7

1.7

0.2

0.3

Iran

, Isl

amic

Rep

.0

.120

.720

.60

.01.

71.

70

.20

.2

Ch

ina

0.1

20.7

20.6

0.0

1.7

1.7

0.2

0.2

Ch

ile

0.1

20.7

20.6

0.0

1.7

1.7

0.2

0.2

Arg

enti

na

0.1

20.7

20.6

0.0

1.7

1.7

0.1

0.2

Bel

aru

s0

.120

.720

.60

.01.

71.

70

.10

.2

Ven

ezu

ela,

RB

0.1

20.7

20.6

0.0

1.7

1.7

0.1

0.1

Bra

zil

0.0

20.7

20.7

0.0

1.7

1.7

0.1

0.1

Lib

ya0

.020

.720

.70

.01.

71.

70

.00

.1

Mex

ico

0.0

20.7

20.7

0.0

1.7

1.7

0.0

0.1

Sau

di A

rabi

a0

.020

.720

.70

.01.

71.

70

.00

.0

Tri

nid

ad a

nd

Tob

ago

0.0

20.7

20.7

0.0

1.7

1.7

-0.1

-0.1

Th

aila

nd

-0.1

20.7

20.8

0.0

1.7

1.7

-0.3

-0.4

Sou

rce:

OE

CD

(20

06

b)

and

au

thor

s’ o

wn

cal

cula

tion

s

This version - DROdro.deakin.edu.au/eserv/DU:30051863/feeny-scalingup-2010.pdf · both nationally...

Documents

Transcript of This version - DROdro.deakin.edu.au/eserv/DU:30051863/feeny-scalingup-2010.pdf · both nationally...