Good policies and the aid squared approach

Get Complete Project Material File(s) Now! »

Previous research

This section will be utilised to attempt to give a summary of the critical findings of earlier research, relating to the effectiveness of foreign aid and to summarising Sweden’s historical contribution to the issue. Also, some general concepts needed to grasp the analysis will be outlined.

Institutions

Within the context of various analytical frameworks, different scholars have stressed the impact of sound institutions on growth, generally defining them as ‘ the social, economic, legal, and political organization of a society’ (Acemoglu & Johnson, 2005, p. 950). Easterly and Levine (2003) find that such institutional factors are critical for sustainable economic development and conversely, that other suggested factors, such as policies, tropics, germs or crops could only affect growth through institutions.
Rodrik, Subramanian and Tribbi (2002) asserts that institutions are the chief explanatory factor for economic growth and that trade and geography are insignificant factors, once in-stitutions are accounted for; however, it is pointed out that trade affects institutions posi-tively. Dollar and Kraay (2003) affirms this view (although they find direct effects on growth from trade) and further argues that countries with better institutions tend to trade more extensively.
The importance of private property rights has been addressed specifically by Hernando de Soto (2000), in the widely referenced work The Mystery of Capital. He argues that poor peo-ple do not necessarily lack resources. In fact, the people of developing countries posses “dead capital”, amounting to about $9.34 trillion, globally. What they do not have, how-ever, is the de jure right to their own possessions in a holistic system of property right pro-tection – meaning that informal possessions cannot be leveraged into capital (i.e. used as collateral when being considered for loans) and hence dramatically decreases investment prospects. (de Soto, 2001)
Further, land that nobody owns will suffer from reduced potential, since nobody will invest in something that could be expropriated at any time by governments, warlords or some al-liance thereof, in the absence of a checks-and-balances system. Formal property will, argues de Soto, bring about benefits such as the networking of people, the protection of transac-tions and the transformation of assets into more fungible ones, providing incentives to use their full potential. (de Soto, 2001)
By this token, it is an intuitive finding that property rights institutions have ‘a first-order ef-fect on income per capita, investment to GDP, the level of credit, and stock market devel-opment’ (Acemoglu & Johnson 2005, p. 975). In light of this fact, it does not seem prob-able that ODA has a high chance of successfully bringing about growth in a state of infe-rior institutional quality.
Douglas North (1981) distinguishes between a “contract theory” and a “predatory theory” of the state (cited in Acemoglu & Johnson, 2005). A recent approach to quantifying institu-tional quality is proposed in Acemoglu, Johnson and Robinson (2001), where European set-tler mortality is suggested as a measurement of the habitability of a new colony. The idea is that the worse the initial living conditions for settlers arriving in a to-be colony, the less at-tractive the spot for a long-term settlement and hence the higher probability for extractive (predatory) institutions to be put in place solely to haul out resources to a third location. This is believed to provide a natural experiment in comparing the two types of institutions.
Extractive institutions would generally not emphasize secure property rights or the en-forcement of contracts; the Belgian colonisation of the Congo – where 240 per 1000 set-tlers perished – is a case in point. Conversely, where living conditions where good, (con-tract) institutions for sustainable living were put in place, as were the cases in Canada, the United States and New Zealand, where settler mortality was low. The negative link between economic growth and settler mortality rates is strong over time. (Acemoglu et al., 2001)
Subsequently, it has been shown that – intuitively – elevated settler mortality rates are asso-ciated with higher macroeconomic instability and more frequent economic crises (Acemo-glu, Johnson, Robinson & Thaicharoen, 2003). The author of this paper wishes to stress that settler mortality offers guidance not only in correlation between variables but also in causality matters. The fact that this variable offers such a good estimator of economic de-velopment lends support to what is concluded in Kaufman and Kraay (2003): institutions may be a prerequisite for growth and not a complement to growth. Easterly and Pack (2004) re-inforces this finding by noting that shortage of financial capital is not what is constraining development in Sub-Saharan Africa. In light of the above, an underlying proposition for this thesis will be that foreign aid will – all else equal – be more effective where institutions initially are reasonably strong.
Thus, there is not much academic controversy about the role of institutions in develop-ment. However, the effects of ODA on institutional quality are at this point in time not en-tirely clear, but there is indeed some evidence to suspect that aid may have perverse effects. As concluded by Rajan and Subramanian (2007), for example, aid inflows may have the in-tuitive effect of rendering a recipient government less dependent on tax revenues.
Moss, Pettersson and Van de Walle (2006) reviews the literature and lend further support to the notion that aid may indeed work counter-productively in this sense, while noting a possible aid-institutions paradox, where governments have less incentives to invest in effec-tive public institutions the more aid it receives. Djankov, Garcia and Reynal -Querol (2005) goes so far as to claiming that aid is a ‘curse’ and finds that for high levels of aid, 75 percent of GDP or more, it poses a serious and long-term threat to development by weakening democratic institutions.
Bottom line here is that governments that receive large percentages of foreign aid may not only lack incentives to be prone to reform, but might even be implicitly encouraged to de-teriorate the situation. Consequently, Bräutigam and Knack (2004) find such relationships between higher levels of foreign aid and the worsening of governmental quality as well as with lower tax effort in Africa; their recommendation is to provide better aid through in-creased selectivity, pointing to the successful case of South Korea, Taiwan and Botswana to prove that foreign aid might indeed work.
Fungibility
In this context, fungibility – the notion that aid frees up resources that can be put to use in other sectors of the economy than otherwise planned – is a vital concept. Pack and Pack (1993) and Svensson (2001) show how fungibility might contribute to a counter-productive resource allocation and how money can be spent contrary to the donor’s intentions. Con-sequently, Feyzioglu, Swaroop and Zhu (1998) lend support to the fungibility concept by noting how earmarked grants tend to decrease domestic spending on education, agriculture and energy. In addition, a functioning democracy is an important determinant for a gov-ernment’s possibility to deliver public services in the first place – even when adjusted for per capita income (Easterly, 2006).
Paul Collier, despite being an aid optimist, estimates that about 40% of Africa’s military ex-penses are de facto financed by foreign aid, even though the aid in question may have been earmarked for other sectors (Collier, 2007). Nilsson (2008) describes how Swedish ODA – through fungibility – has been put to use in ways not intended and even sarcastically re-marks how in the 1980s, Swedish ODA sponsored Marxist school books, written by East German experts in Ethiopia, while funds became available for use in the army.
However, research on this topic is limited. Svensson (2001) notes how fungibility is a hard concept to test for, in part due to lack of relevant data and in part due to lack of knowledge of the counterfactual: how a government would have reacted in the absence of a foreign aid disbursement (Svensson, 2001, p. 6).
A present trend in Swedish ODA is the move from project support based assistance to-wards a larger share of aid devoted to budget support, where recipient countries are given larger degree of autonomy in administering the disbursed funds (Sida, 2008). Indeed, in-creased recipient ownership of funds is a direction in line with the recommendations in the Paris Declaration on the effectiveness of foreign aid, signed by over 100 countries in 2005 (Bour-guignon & Sundberg, 2007). In 2007, Sida decided not to disburse budget support to Bo-livia, Honduras and Nicaragua, due to their alleged insufficient poverty alleviation policies (Sida, 2008).
It is noteworthy how, in 2006, Sweden doled out budget support to eight states; Burkina Faso, Mali, Mozambique, Nicaragua, Rwanda, Tanzania, Uganda and Zambia and how the Swedish government did not follow through on the checks and balances it had imposed on itself in seven out of these eight cases, while at the same time being kept in the dark by Sida, regarding the risks of budget support (Riksrevisionen 2007a; 2007b).
Naturally, the nuisance involved with the fungibility issue is further exacerbated when aid is not at all conditioned, since the concept of budget support allows for no external audit re-garding how the disbursed funds were really used. Riksrevisionen (2007b) points out vari-ous troublesome facts in the Swedish government’s contemporary collaborations regarding this form of disbursements; notably how it has been handled in the cases of Mozambique and Tanzania and how the reports from Sida to the Swedish government neglected to point out the obvious risks involved with this type of development assistance.

On the effectiveness of foreign aid

In 2007, Göran Holmqvist, then general director of Sida, claims that ‘over the past 10 years, some 50 scientific articles have been published by prominent economists [on the ef-fectiveness of foreign aid]. Almost all of these concluded that aid contributes positively to the economic development of [developing] countries’ (Holmqvist, 2007). As reasoned above and as we shall conclude below, this is simply not true. Additionally, as pointed out by Doucouliagos and Paldam (2005), the number of studies alone expressing positive links between aid and growth is not proof that such a relationship exists.
As previously noted, there are diverging views regarding the efficiency of ODA and whether or not development aid has any at all abilities of bringing about growth. Table 3.1 relates to the critical findings in the most well renowned recent literature.
Naturally, various other scholars have implied positive links between aid and growth in their research, albeit without adding new knowledge to the body of empirical research on the overall macroeconomic impact of foreign aid. Normally, those have built upon one or many of the above reports and are hence subjects of the same criticism. Such scholars in-clude Lensink & White (2000); Collier & Dollar (2002) and Addison, Mavrotas, and McGil-livray (2005) . Others have concurred with the critics of the aid-growth research, i.e. Jensen and Paldam (2006), which concludes that neither the virtues of the good policies approach nor the aid squared methodology4 could be generalised and reproduced with different sets of data.
This thesis searches for thoroughly reasoned policy implications that would be of value to aid agencies in a donor nation. Numerous scholars have shown that aid is correlated with growth for a certain set of variables, for some specific years or that some interaction vari-able enters statistically significant with aid for specific periods. However, no scholar (with one possible exception, see below) has been able to incorporate their views in a holistic model useful enough to make general policy implications without at the same time seeing the model in question fundamentally confuted by other scholars. B -D constructs such a model, which is appealing to politicians and the electorate alike5 and had it not been falsified, would have proven a case-in-point. This is the view advocated in Easterly (2003) and Roodman (2007a) among others, all of which exert criticism over scholars’ at times reckless choice of model specification.
There might, however, be one exception, namely Svensson (1999), who asserts the link be-tween aid effectiveness and political rights. It should be noted that this paper explicitly complements the B-D findings and that it has not been under the same amount of scrutiny, due to its relatively small spread. The findings have not been replicated to the knowledge of the author of this thesis. Roodman (2007a) mentions the paper but does not proceed to testing it.
Bearing this in mind, the author of this thesis will be in agreement with Svensson (1999) and argue that aid to democratic states and the rewarding of democratisation is certainly preferable as compared to its alternatives, both in terms of efficiency and on democracy’s own merits, ipso facto. Findings like these might indicate that some of the parameters in the B-D study might in reality have been proxies for democratic governance (Svensson, 2001). Aid to a democracy would at the same time alleviate some of the problems involved with fungibility. Furthermore, a serious famine has never struck within the boundaries of a democratic state with a free press (Sen, 1999).
Other than this, the author does not argue that aid does not work. It is undoubtedly the case that some aid has ended up serving its intended purpose; i.e. financed investment, but the information on the macroeconomic net effects of ODA are largely absent in the litera-ture. This is what Mosley (1987) refers to as the micro-macro paradox of foreign aid: although numerous case-studies have shown that individual projects have been successful on the mi-cro level, tangible results on the macro level have remained elusive6. It is the opinion of the author that aid should be evaluated on its macroeconomic net effects, notably because it might have depressing effects of other parts of the economy than the one directly affected by monetary inflows, as noted in Rajan and Subramanian (2005b).
It is observed from the above table and from the previous outline of arguments that there certainly is no established research consensus to be found on the topic. The author has not come across any peer-reviewed research advocating the effectiveness of foreign aid that has not been directly or indirectly fundamentally disproved by other scholars, if attempted.
Conclusively, Doucouliagos and Paldam (2005) provide a meta-analysis of the aid literature for the past 40 years (accepted for publication in the Journal of Economic Surveys) – a to-tal of 97 studies. They find that aid most probably have not been effective to this point and suggests different reasons why aid pessimistic research may have been marginalised. Also, the researchers deliver a caveat for other scholars: all results produced by model innovation should be treated with caution until the results have been independently replicable. (Dou-couliagos & Paldam, 2005)

History of Swedish ODA activities

As noted above, the author has found that research literature on Swedish ODA with re-spect to the recipient’s form of governance is parsimonious. However, this section will out-line the key findings in literature regarding Swedish ODA flows and past policy regimes.

Theoretical framework

Many scholars cited above concludes that aid should be focused to countries that fulfil some type of criteria. For example, it is suggested that the ‘quality of the recipient – its policies and institutions – is thus a good indicator, albeit not a full determinant, of the re-cipient’s ability and willingness to use aid.’ (Rajan & Subramanian 2005a, p. 15)
Bourguignon and Sundberg (2007) notes how it is important to direct aid to where results can be monitored and that allocation of aid on the basis of performance is a likely path to go down in the future. It is asserted that aid is increasingly disbursed ‘on the basis of coun-try performance that combines governance, general policy environment and some interme-diate or final results’ (Bourguignon & Sundberg, 2007, p. 325). In light of the reasoning put forth in the background section, where it is affirmed that aid might even hamper institu-tional development, this point cannot be overstated. That is, aid could possibly harm insti-tutions but is nevertheless unlikely to fulfil its objectives if institutions are too weak.
It is indeed difficult to suggest useful policy implications since there is no general consen-sus that development aid works in the first place. However, available knowledge on poverty alleviation can indeed serve as a guide for how an environment that best utilises resources should look like. In light of the above, it is the opinion of the author that development aid, if disbursed, should be distributed where results can be monitored to the furthest extent pos-sible. Hence, policymakers should consider the consistencies outlined below. According to the best available research on aid effectiveness and on poverty alleviation, foreign aid should satisfy:
Democratic consistency. Fungibility analysis suggests that a recipient government should be, to the furthest extent possible, accountable for its actions in order to spend disbursed funds righteously. Consequently, some research has indicated that aid to more democ-ratic states might be more effective in bringing about growth (Svensson, 1999).
Institutional consistency. The quality and type of institutions seems to be essential for sus-tainable growth and might well be a prerequisite for it. As discussed above, aid may in fact counteract the development of sound institutions but nevertheless; it is intuitive that it would still make the best contributions, if any, where institutions are initially strong. It is the opinion of the author that completely failed states should not be recipi-ents of Swedish ODA.
The remainder of this thesis will attempt to test whether or not Swedish ODA is disbursed in line with the above conditions. The author will however consider aid that fulfils the con-dition of:
Good policies consistency. Because of its enormous influence and because it has been ac-knowledged by the Swedish government 7, this thesis will also test whether disburse-ments can be considered in line with the Good policies model as outlined in Burnside and Dollar (2000).
Naturally, the ideal situation would be one where Swedish ODA is found consistent with all three criterions.
On a side note, Bourguignon and Sundberg (2007, p. 321) asserts that it is ‘illusory to be-lieve that all interventions can be subject to impact evaluation, and that evaluations will permit to direct the flow of aid exclusively to ‘what works’, as some8 have suggested’. The author of this paper disagrees. As long as the macroeconomic effects of foreign aid remain ambiguous and so long as the impact of aid might even be negative it seems naïve at best to disburse ODA into the unknown.
Some might argue that, given world conditions, it is not realistic to maintain control over all of the projects financed by outflows of Swedish ODA. Those concerns are most proba-bly well founded. Such anxiety, however, does not constitute an argument to disburse ODA where information is absent, but rather not to disburse more funds than can be ac-counted for as long as research on the topic is ambiguous. As noted above, budget support can simply not be traced once it has been absorbed by the budget of the recipient. There-fore, the author argues that budget support, without exceptions, should be disbursed to governments deemed accountable for their actions.

The Svensson framework: democratic consistency

As initially noted, Svensson (1999) adds to the aid-growth literature by suggesting a link in between aid effectiveness and a democratic state of affairs; this finding is a complement to the findings of B-D and is explicitly affirmed by the Swedish government and ODA agen-cies – that is, it might be the single variable that first and foremost merits our attention ana-lytically, even when ignoring the arguably intrinsic value in a democratic development. The caution expressed by Doucouliagos and Paldam (2005) should definitely be taken into ac-count at this point: the results have not been replicated beyond the original study’s time pe-riod or with other sets of data and should be treated with due caution. However, as previ-ously elaborated upon, it is the opinion of the author that aid should be considered qua aid, and that aid to more democratic states is preferable regardless of the robustness of the findings. Not least since aid to such states is minimising the fungibility issue.
The methodology of the original paper is rather straightforward: the government is as-sumed to exert power, but it can only do so while being held accountable to the electorate. Free speech and the possibility of voting an administration out of office, according to this framework, will work to ensure a more efficient use of resources, including foreign aid. Further, it is argued that since donors in reality have small possibilities of influencing re-cipient policy decisions and since aid is fungible, there is a need to address government be-haviour when pondering the foreign aid. (Svensson, 1999)
Measuring democracy is admittedly a difficult endeavour. In the original paper, Svensson (1999) uses the political rights and civil liberties classifications from Freedom House (2007). In this context, one should note how the Freedom House index has been exten-sively criticised for favouring American allies during the cold war. See, for example, Munck and Verkuilen (2002). It is being used here for simplicity, i.e. since it was used in the origi-nal analysis and because observations are available for the whole time period and for virtu-ally every country in the dataset.
If Swedish aid is disbursed in such a way as to satisfy its own mandate to maintain a politi-cal rights oriented approach to development assistance and, indeed, to be in line with the only non-contradicted research regarding aid effectiveness the author of this thesis has found, it should be negatively correlated with the Freedom House classification.

 The Burnside and Dollar framework: good policies consis-tency

As has been shown, B-D approach is perhaps not academically fit to serve as a holistic model of aid, given the data available at this point. However, as previously noted, the good policies approach has been voiced in defence of development assistance by Swedish politi-cians and should therefore be validated in light of what disbursements have been made. Additionally, it is the single most influential case for ODA effectiveness, which on its own merits qualifies it as eligible for testing.
The findings of B-D rest on the analytical approach by Solow (1956) and other contribu-tors to neoclassical growth theory and hence the underlying idea is that aid has the poten-tial of bringing about growth by stimulating domestic savings and investment with decreas-ing returns to capital.
For Swedish ODA to be considered disbursed in accordance with the good policies ap-proach there should be a positive relationship between ODA and trade openness and a negative relationship between ODA and inflation rates as well as with a government deficit.
The good policies approach to foreign aid was constructed by B-D, using trade openness data from Sachs and Warner (1995), which composes data decennially for the 1960’s and 1980’s. This data has subsequently been improved and extended to the 1990’s by Wacziarg and Welch (2008). Ideally, these figures would have been used here, however, this thesis aims to cover more recent years and the author wishes to include as many observations as possible.
Therefore exports and imports as a percentage of GDP from World Bank (2007) are being used in their stead. Furthermore, this enables more detailed investigations, since the new figure is available yearly, whereas the Sachs and Warner index is constant for each decade. There is an obvious problem with this approach, though, namely that the utterly destitute countries often consist of subsistence farmers and, hence, do not trade much. However, as noted by Collier (2007), developing countries overall depend on the manufacturing sector for 80% of its income. It is the belief of the author of this paper that total trade over GDP is the best available estimator of trade openness until the Sachs and Warner (1995) index has been further extended.
The biggest problem with this particular part of the empirical analysis is the availability of data on fiscal prudence. The Government Finance Statistics in IMF (2008b) provides a comprehensive view of cash surpluses and deficits, yearly since 1990 and is analytically use-ful, but unavailability of data will cut off more than half of the original observations in the sample from the analysis and hence tests will be run both with and without this parameter. Even worse, there is adverse selection: when including this variable; the countries missing in the data tend to be the poorest, most corrupt and the ones that merit our attention. For now, this variable will be ignored, but briefly reassessed in the findings section for institu-tional consistency in the empirical analysis part.

Institutional consistency

Following the initial discussion, institutional quality seems to be a prerequisite for sustainable growth, which is the most undisputed previous research finding presented in this thesis. Following the reasoning put forth in de Soto (2000), Dollar and Kraay (2003) and Rodrik et al. (2002) it is the opinion of the author that aid to an inferior institutional environment is, all else equal, more likely to be wasted.
There are numerous ways of measuring institutional quality9. The author has chosen to work with a proxy for secure contract and property rights: the Contract-Intensive Money (CIM) measure proposed Clague, Keefer, Knack and Olson (1999) and defined as the ratio of non-currency money to the total money supply. This reflects general reliance on the banking system; CIM will converge to 1 as public confidence increases. Ideally, CIM will be a good proxy both for institutional quality in general and on private property institutions in particular. CIM has the benefits of being an objective measure and it is widely available. CIM is positively related to investment, income and growth (Clague et al. 1999). Hence, Swedish aid should be positively correlated with the CIM variable to be considered consis-tent with institutional quality.

Table of Contents
Bachelor Thesis within Economics
Kandidatuppsats inom Nationalekonomi
1 Introduction
1.1 Purpose
1.2 Outline
2 Background
2.1 A brief history: Aid effectiveness and poverty alleviation
2.2 Good policies and the aid squared approach
2.3 The role of Swedish ODA
3 Previous research
3.1 Institutions
3.2 Fungibility
3.3 On the effectiveness of foreign aid
3.4 History of Swedish ODA activities
4 Theoretical framework
4.1 The Svensson framework: democratic consistency
4.2 The Burnside and Dollar framework: good policies consistency
4.3 Institutional consistency
5 Model and data
5.1 Data
6 Empirical analysis
6.1 Democratic consistency
6.2 Good policies consistency
6.3 Institutional Consistency
6.4 Extended analysis
7 Conclusions
7.1 Policy recommendations
References
GET THE COMPLETE PROJECT