کمک و رفتار مالی بخش دولتی در ایالت های ورشکسته

This paper looks at interactions between foreign aid and the public sector in developing countries, especially those considered to be fragile or failing states. A model is proposed which employs actual budgetary appropriations and revenue estimates (rather than estimated target variables) and allows for asymmetric preferences. Variants of the model are estimated using time-series data for Papua New Guinea (PNG). PNG is classified as a fragile state by the international community owing to perceived policy and institutional inadequacies. Results obtained suggest that foreign aid increases consumption and investment expenditures and decreases tax revenues and the level of borrowing.

The international community has acquired lofty expectations of foreign development aid in recent years, especially since the adoption of the Millennium Development Goals (MDGs) at the United Nations (UN) Millennium Summit in September 2000. The MDGs aim inter alia to halve world-wide the number of people living in extreme income poverty by 2015. Part of the strategy to achieve or at least work towards the MDGs is a substantial scaling up of aid flows by donor nations (UN Millennium Project, 2005). This is clearly evident in aid statistics, with global Official Development Assistance (ODA) rising from $US69 billion in 2003 to around $US106 billion in 2005. ODA is expected to rise to $US130 billion in 2010 (OECD, 2006). While there is much optimism in the donor community about the impact of these flows, it has grave concerns about the effectiveness of aid to countries it classifies as ‘fragile states’.1 The impact of aid on growth and poverty reduction and the ability to efficiently use additional inflows is thought to be substantially lower in these countries compared to other recipients. Donors rightly insist that unless aid can be made to work better in fragile or failing states (the term preferred by and subsequently used in this paper), the intended developmental dividend from these increased flows will not be observed, and the world-wide achievement of the MDGs will not be possible in the foreseeable future, let alone by the agreed target of 2015.2Collier (2007) argues that finding initiatives to help the bottom billion (about three quartes of which live in fragile or failing states) is one of the key challenges facing the world in the twenty-first century. While the countries belonging to the donor community failing state group are diverse in many respects, they all share two common characteristics. The first is that they are desperately poor. According to one of the two classification criteria currently used by the donor community, 46 countries were classified as fragile (failing) states in 2004.3 Roughly one-third of the world's population living in extreme income poverty resides in these states. This equates to 340 million people. Of the estimated 10.8 million children that died before their fifth birthday in 2002, just over 40% lived in these 46 states (Branchflower et al., 2004). The second characteristic is that they all have critically substandard public policies, poorly performing public institutions, or both. Donors actually use a measure of policy quality and institutional performance – the World Bank's Country Policy and Institutional Assessment (CPIA) – to deem a country fragile. It is indeed on the basis of this specific characteristic that donors primarily believe that these countries use aid poorly.4 Concerns over the likely impact of aid to failing states are broadly consistent with points long recognised in the research literature: recipient country public sectors are key players in determining the effectiveness of aid and that the broader impacts of these inflows are mediated by the manner in which public sectors use aid inflows.5 The second of these points has been addressed in a research literature dating back to the 1970s that looks at the impact of aid on various categories of public sector revenue and expenditure. This research has its origins in the seminal work of Heller (1975), which went largely unnoticed in the literature on aid for a number of years. Gang and Khan (1991), in an important study, changed this state of affairs by applying Heller's model to time-series data for India.6Chishti and Hasan, 1992, Khan and Hoshino, 1992, Binh and McGillivray, 1993, Gang and Khan, 1993, Gang and Khan, 1999, Khan, 1994, McGillivray, 1994, White, 1994, Otim, 1996, Franco-Rodriguez et al., 1998, McGillivray and Ahmed, 1999, Franco-Rodriguez, 2000, McGillivray, 2000, Mavrotas, 2002, Mavrotas, 2005, McGillivray and Ouattara, 2005, Mavrotas and Ouattara, 2006, Ouattara, 2006a, Ouattara, 2006b, Feeny, 2006 and Feeny, 2007 have all followed Gang and Khan's lead with their own contributions to the ‘fiscal response’ literature, a term attributed to White (1992).7 The fiscal response literature has advanced both conceptually and econometrically as a result of the preceding studies. Advances include more theoretically sound and institutionally realistic utility function and constraint equation specifications, basing conclusions and policy recommendations on both reduced-form and structural equation parameter estimates, endogenising aid, analysing alternative aid disaggregations, using time-series data and the reporting of constraint and structural equation parameters consistent with the underlying theory. While these advances are important, it is widely acknowledged that there is still plenty of room for further improvement in both the design and application of fiscal response models (McGillivray and Morrissey, 2001a). In particular, no study has been able to address satisfactorily the two deficiencies widely thought to be the most serious in the fiscal response literature: the reliance on crudely estimated revenue and expenditure targets and the use of a utility function premised on perfectly symmetric policymaker preferences.8 Failing states have also been very much under-represented in the fiscal response literature, although this is understandable given that conditions in these countries are such that their failings include the reporting of requisite data. Only two of the above-cited studies, McGillivray and Ouattara, 2005 and Mavrotas and Ouattara, 2006, have looked specifically at a country classified as a failing state. Both, however, looked at the case of Côte d'Iviore.9 More knowledge of the fiscal response to aid inflows in these countries is crucial, arguably more so than in other aid recipients. This paper addresses both of the just-mentioned deficiencies. It builds on the work of Gang and Khan (1999) through its consideration of asymmetric preferences. It develops a generic fiscal response model that allows for asymmetric preferences and estimates this model using actual expenditure appropriations and revenue estimates instead of target estimates. It reveals that the problem with using estimated target values is far more serious than previously understood in the literature, and has in all likelihood led to seriously misleading policy recommendations. A specific form of this model is applied to Papua New Guinean time-series data for the period 1969 to 2000. This model has been largely based on fieldwork conducted in Papua New Guinea (PNG). Estimates of structural and reduced-form parameters are provided, based on alternative disaggregations of aid inflows. While the modelling approach of this paper has a general appeal to the analysis of aid impact, a focus on PNG is interesting in its own right, for three reasons. First, PNG is unlike most other failing states in that the country has published annual data since the early 1960s on various categories of planned and actual revenues and expenditures. These data permit the estimation of a suitably specified fiscal response model. This in turn provides the potential for insights into the relationship between aid and various fiscal aggregates that are simply not possible for most failing states, and many other developing nations as well. Second, the level of official aid to PNG is particularly large. It has averaged 35% of government revenue since 1960, reaching as much as 60% in some years, and is much greater in magnitude than private foreign inflows (GPNG, 1962–2002; OECD, 2005). No other South Pacific country has received as much aid since 1960 as PNG (OECD, 2005). In per capita terms, aid to PNG has averaged US$90 since 1975 — nine times the average per capita receipts of all developing countries over the last ten years (OECD, 2005; UNDP, 1993–2003). Not only does this level of aid suggest that it is likely to have had observable impacts, but that the outlay of donor taxpayer funds warrants an evaluation of these impacts. Third, PNG has experienced increasingly difficult times in recent decades (Windybank and Manning, 2003). PNG in many respects resembles the stereotypical failing, although still functioning, developing country. Achieving independence in the mid-1970s, PNG has experienced or exhibited frequent changes of government, a civil war, a military that has threatened revolt on a number of occasions, seemingly endemic corruption, a rapidly growing HIV/AIDS problem, low levels of personal security and a very poorly performing economy according to most if not all criteria. Some studies assert that foreign aid has actually contributed to aspects of PNG's decline (Hughes, 2003; Windybank and Manning, 2003). While there is little scientific evidence to support this claim, it is widely recognised that the PNG public sector has limited capacities to perform a number of functions, including the efficient application of aid inflows. This is reflected in it being classified by the international donor community as a fragile state, based on it being rated in the bottom two quintiles of CPIA scores (World Bank, 2003; Jones et al., 2004; McGillivray, 2005). In a further, somewhat radical and controversial response to the deteriorating position in the South Pacific's largest country, Australia, the lead PNG donor, has recently placed its own civil servants in various PNG government departments to help strengthen the level of economic and public sector management (AusAID, 2004). The case for a formal evaluation of the impact of aid on public sector fiscal aggregates in PNG appears stronger, given this background. Moreover, as a seemingly typical failing state, knowledge of the fiscal impact of aid to PNG might provide useful insights into aid effectiveness in other such states. This paper is structured as follows. Section 2 develops a general fiscal response model that both builds on past developments in, and addresses remaining inadequacies of, the fiscal response literature. Included in Section 2 is a discussion of the relevance of this model to failing states. Section 3 outlines various versions of the fiscal response model that will subsequently be estimated using PNG data. Section 4 discusses data and estimation issues, prior to reporting results from estimating these versions of the model for PNG. Section 4 also draws inferences from these results for the broader impacts of aid to PNG, as well as comparing them with results from fiscal studies of other recipient countries, seeking to ascertain how PNG rates in a broader international context. Section 5 concludes, identifying implications for policy, both in PNG and other failing states.

Aid flows are an important component of the public sector budgets of most developing countries. A greater understanding of the interactions between aid and various categories of public sector expenditure and revenue is required if aid is to contribute further to economic growth and the achievement of the internationally agreed MDGs. This paper has sought to contribute to this understanding in two ways. The first was to further improve the aid-fiscal behaviour literature by addressing two key, widely acknowledged weaknesses in fiscal response models. This involved building a model that allows for asymmetric fiscal decision maker preferences and uses expenditure appropriations and revenue estimates rather than vaguely defined and statistically generated targets. The second contribution was to apply this model to Papua New Guinea (PNG), using 1969 to 2000 annual time-series data. As outlined above, PNG is a particularly interesting case study. It is a country that has experienced many economic, political and social difficulties, despite receiving relatively large amounts of aid. It has recently been classified as a fragile state by the international donor community, in part based on perceptions that it has a weak and ineffective public sector. It is also a country that is generally considered to have not used aid especially effectively, and the Australian government (the largest donor of aid to PNG) has responded to this and the general state of affairs in PNG in a rather radical manner by placing its own civil servants in key positions in the PNG's public sector. The results obtained from applying the fiscal response model to PNG paint a picture of a poorly performing public sector that has not used aid and other revenues especially well and worse than many other countries examined in the literature. Consistently high proportions of taxes and other recurrent revenue, aid and non-aid borrowing are allocated to consumption expenditure and there is evidence of aid loans financing unanticipated shortfalls in the recurrent budget and of aid displacing other recurrent revenue. There were some pleasing results, with aid being positively associated with fixed investment and non-aid borrowing but the overall picture of the effectiveness of aid to PNG is relatively poor. This is not at all to say that aid has failed PNG, in that it has failed to contribute to growth and by assumption poverty reduction. Indeed, there is evidence from the aid-growth literature that growth in PNG would be lower in the absence of aid. But it does suggest that the manner in which the PNG public sector has used aid has limited the overall effectiveness of these inflows in promoting growth and more importantly reducing poverty. To these extents there would appear to be empirical support for the radical Australian intervention mentioned at the outset of this paper. Time will, of course, be the ultimate judge as to its success. Finally, some comments on the impact of aid on public sector fiscal aggregates in failing states in general and for donor policy are warranted. Here we rely not only on the results reported in this study, but also on the findings from other studies examining aid and the public sector in failing states, cited above. These comments relate to failing states that are similar to PNG in this regard, and not to those without any functioning public sector. Clearly, the results reported in this paper and the other studies will not be relevant to such states. These results suggest that failing states are characterised by using non-aid borrowing to fund recurrent budgets. Findings also suggest that these states use aid as a substitute for taxation and other recurrent revenue: these revenues decrease in response to aid inflows. Donors need to be conscious of both outcomes in the design and delivery of aid programs to failing states. The former we interpret as the result of unexpected shortfalls in revenues. Such behaviour could be justified as a response to a revenue decline if it means that essential expenditures such as the delivery of key services are maintained. However, in the longer term, this justification would appear to be difficult to support. Public sector borrowing needs to be repaid, increasing the level of debt and a country's future debt service obligations. Given the political instability and future uncertainty often associated with failing states, politicians currently holding office might be more disposed to undertaking such borrowing. The preceding finding also highlights the need to accurately forecast revenue estimates. Overly optimistic estimates of expected revenue will exacerbate the problem of public sectors not being able to meet their expenditure appropriations. Donors could assist in strengthening the ability of public sectors to undertake this exercise effectively. Further donors must be sympathetic to the financial requirements of public sectors which have experienced external shocks. Natural disasters and unexpected falls in the prices of key commodities can derail good public sector financial management. Donors should ensure their own aid flows are predictable and could supplement other unexpected revenue shortfalls to prevent higher levels of commercial borrowing. The finding that taxation and other recurrent revenues fall in response to aid inflows in failing states is a potentially serious matter for international aid donors. Trade liberalisation conditions, sometimes attached to the receipt of foreign aid could, at least partially, explain this finding. However, the finding is also likely to reflect recipients developing a long-run dependence on aid inflows as a revenue source. Failing states often have low revenue bases and sometimes a weak commitment or limited ability to collect taxes. To mitigate this impact of foreign aid it is important for international donors to work with recipient public sectors in strengthening revenue collection and management.