Kicking the Habit (2). IMF and World Bank conditionality reform

’Policy reform has had a mixed track record...Adjustment has been a much slower, more difficult and more painful process than the Bank recognized atthe outset… What I am looking for…is a different way of doing business inthe future...’ Jim Wolfensohn, then President of the World Bank, letter to the Structural Adjustment Lending Civil Society Network, 9 April 1996 17

In 1999, in recognition of the failure of structural adjustment to deliver, and in the face of growing international criticism of their undue interference, the World Bank and the IMF announced a new way of delivering aid to developing countries; the Poverty Reduction Strategy (PRS) approach. The PRS approach was based around three key principles: poverty reduction, country ownership, and evidence-based policy-making. Poverty reduction was made a central objective of World Bank and IMF lending, in recognition that growth, though important, is in no way sufficient to ensure poverty reduction. Developing countries were to be put in the driving seat, with development policies to be domestically formulated and implemented, rather than devised by the IMF or the World Bank. Finally, it was made explicit that evidence-based policy-making was vital, in order to move away from ideologically driven policy design and take greater account of national economic, social, and political realities.

While never completely renouncing the use of economic policy conditionality under this new approach, the World Bank and IMF did both agree to use it far more sparingly and only when two important safeguards were met. Economic policy conditions had firstly to be country-owned, and secondly to be based on analysis of the impact of different policy choices on poor people prior to their application. In 2002, two years after the initial announcement of the supposed change of course, the IMF announced that it was going to streamline the number of conditions it attached to its lending, in recognition that there had been a proliferation of conditions during the 1990s.18 In 2004 the World Bank followed suit, initiating a new conditionality policy that stipulated that only those policies critical to ensuring programme success were to be set as conditions for lending, and these were to be drawn from policy and institutional frameworks agreed by the country.19

Importantly, the World Bank removed reference to specific economic policy reforms within its lending directive, in recognition that generalised policy prescriptions often fail, and that there is no single model of development.20 The directive also recognised the importance of assessing the social and poverty impacts of significant reforms prior to their being set as conditions.

Even more recently, the World Bank has issued new staff guidelines on conditionality to help ensure the implementation of this new approach. However, despite these reforms, seven years on from the announcement of a new poverty reduction approach, the World Bank and the IMF have failed sufficiently to change the way they do business. If ‘policy benchmarks’ are included, World Bank conditions have increased not decreased since 2000. Both institutions still have an unacceptable number of economic policy conditions attached to their aid. Ownership of conditions is inadequate. And too often analysis of the social impact of policies is too thin, or skipped altogether.

Failure to reduce the numbers of conditions
World Bank analysis shows a dramatic decline in policy conditions from 32 per loan on average in 1999 to 11 per loan in 2006.21 However, this figure is deceptive, as it does not account for a massive rise in the number of policy benchmarks attached to World Bank aid since 2000. According to World Bank data, policy benchmarks rose from eight per loan in 2001 to 27 per loan in 2006, an increase of over 300 per cent.22 The World Bank does not categorise policy benchmarks as full conditions. This is because if recipient countries fail to implement policy benchmarks aid cannot be stopped or delayed. The World Bank only counts ‘prior actions’, ‘tranche release’ and ‘triggers’ as full conditions, as these have the potential to stop aid if recipient countries do not implement them. But while policy benchmarks may not be as powerful as full conditions, they are very influential and as such constitute a form of conditionality. Before a loan is released, in addition to fulfilling prior actions and other measures, the Bank’s Executive Board must be assured that there has been satisfactory progress in implementing these benchmarks.23 Benchmarks are also perceived as conditions by
recipient governments. In a survey undertaken by the World Bank in 2005, 75 per cent thought that their country had to comply with prior action, trigger and policy benchmarks to access funding.24 If policy benchmarks are taken into account then World Bank policy conditions have risen since 2000 from 20 per loan to 38 per loan in 2006.25

In the case of the IMF, conditions were reduced substantially in the late 1990s, but new research in 20 countries by the European Network on Debt and Development (Eurodad) shows that they have started to slightly increase since 2002.26

There can be no doubt that progress has been made on economic policy conditionality, with a reduction in the number of prior actions and triggers that specify economic policy reforms attached to World Bank lending. However, around a quarter of all conditions still push specific economic policies. In the recent World Bank conditionality progress report, just under a third of the loans (32 per cent or six out of the 19 loans surveyed) contained prior actions or trigger conditions on privatisation, price liberalisation or trade reform. When benchmarks are added, ten out of the 19 loans sampled (52 per cent) have conditionality in one or other of those areas.27

The IMF does not do any better. A recent study by the Norwegian government reveals that privatisation and liberalisation still feature as important elements in IMF lending to poor countries. Of the 40 loans made to poor countries in 2006, 26 had privatisation and liberalisation conditions attached to them.28

Country-owned?
In addition, the World Bank and the IMF have failed to ensure that their policy conditions are truely country-owned. As a marker of ownership, both institutions look at whether policies are already contained within a national poverty reduction strategy. However, there are serious questions as to whether national poverty strategies are adequate proof of country ownership. Despite the fact that the development of national poverty strategies has undoubtedly opened up space for country policy-making, participation of civil society and parliamentarians is often extremely weak and sporadic. In addition, the strategies have been very broad making it easy for the Bank and the Fund to claim alignment. Moreover, national poverty strategies are hardly immune from World Bank and IMF influence, either directly or indirectly. The Bank and the Fund are important sources of advice to governments in the preparation of a Poverty Reduction Strategy Paper (PRSP), and they jointly assess the adequacy of the PRSP as the basis for their support.29 A World Bank conditionality survey, for example, showed that 50 per cent of governments surveyed felt that ’the Bank introduced elements that were not part of the country’s program’. 30 In addition, ‘37per cent of respondents said that negotiations with the World Bank significantly modified their original policy program’. 31 Even in the absence of direct influence, the strategies are open to a high degree of self-censorship, as they are essentially business plans for donor funding, meaning governments have an incentive to tell donors what they are likely to want to hear.

Finally, even against this dubious proxy for ownership the two institutions often fail. The Eurodad study found, for example, that four countries with specific privatisation conditions attached to their World Bank loans do not mention the reform in their national poverty strategies. By their own admission, in evaluations of their new PRS lending approach the World Bank and the IMF noted limited progress in ensuring their lending was aligned to national strategies.32

Missing poverty analysis
The World Bank and the IMF have also been weak in ensuring that the policy conditions they attach to their lending are sufficiently poverty road-tested. An IMF report notes that the PRSP approach ’has so far not contributed significantly to understanding the linkages between growth, poverty incidence and macroeconomic policies at the individual country level‘.33 The World Bank, more so than the IMF, has made a significant effort to ensure more of its analytical work looks at the impact of given reforms on poor people. The problem is that this analytical work rarely appears to feed in to policy conditionality design, something the World Bank itself acknowledges: ’in many cases a direct linkage is not made between such existing analytic work and the impacts of specific policy reforms‘.34 In addition most analytical work focuses on how to implement new economic policies rather than which economic policy would be best for fighting poverty. As a
recent Norwegian government study notes: ‘The International Financial Institutions exert considerable influence through providing policy advice, and have not generally elaborated alternative polices to those involving privatisation and liberalisation’.35

Policies with a high likelihood of a distributional impact are supposed to be subject to Poverty and Social Impact Analysis (PSIA). PSIA is meant to explore alternative policy options and be carried out in a participatory manner: recipient countries are supposed to both set the agenda and be involved in managing the analysis. Independent research on PSIA carried out by several nongovernmental organisations found that many such analyses were actually conducted after the policy was implemented, failed to look at alternative policies, and in the main were not sufficiently countryowned or transparent.36 In 2005, only ten of over 100 PSIAs funded by the World Bank were publicly available on their website.