Conditionality in extreme circumstances: the IMF and emergency financing during the COVID-19 pandemic

IMF lending and conditionality have become two inseparable terms over the course of the past couple of decades. Any country that has run into balance of payment difficulties and needs external financial assistance knows better than to approach the IMF unless it was ready to embark on an extensive reform trajectory as part of the IMF's programme conditionality, also referred to as ex post conditionality by IMF staff. The IPE literature has therefore largely focussed on analysing the effects of conditionality on economic and social developments (Lee and Woo, 2020; Rickard and Caraway, 2018), taking the very existence of conditionality as a given.

Some scholars have usefully pointed to the scope and diversity of IMF conditionality (Dreher and Rupprecht, 2007; Stone, 2008; Woo, 2010), arguing that conditionality cannot be taken as a homogenous policy that offers the same unified criteria for each country case. Rickard and Caraway (2019) more specifically discover that adding a condition for public sector reform would reduce government spending on the country's public sector wages. Lee and Woo (2020) suggest a country's labour rights worsen when the country is subject to greater labour market conditionality. While there is ample sectoral research on IMF conditionality, the academic literature has been inconclusive about whether or not IMF conditionality has positive effects in terms of bringing a country onto the path of sustainable economic growth (Rickard and Caraway, 2019).

Moreover, the existing IPE literature tends to emphasise that the IMF lending and conditionality policy has remained largely unchanged despite attempts to encompass social policy in the post-Washington Consensus era (Broome, 2015). Conditionality has largely become synonymous with IMF lending policy as the IPE literature has been preoccupied with explaining the mechanism from constructivist or rationalist perspectives, or with assessing the effects and effectiveness of conditionality in bringing about inclusive and sustainable economic growth (Dimitrakopoulos, 2020; Rickard and Caraway, 2019).

The use of conditionality, and the analysis thereof in IPE and EU studies, has been growing in popularity since the IMF first exercised it in 1950s (Epstein and Sedelmeier, 2013). Judith G. Kelley provides a very useful definition of conditionality and distinction between normative pressure and conditionality: ‘Briefly, normative pressure occurs when an institution advises a government on the direction a policy should take, offering no reward other than the approbation of the institution. Conditionality, on the other hand, involves explicitly linking the change advocated to an incentive, a particular benefit provided by the institution’ (Kelley, 2006: 3). This distinction clearly indicates the causal link between the demands and rewards. In the case of economic relations, this correlation can be even stronger or can be legally binding.

Another valuable definition of conditionality that has been provided with regards to international lending programmes was put forward by Tony Killick, who finds that conditionality is ‘a mutual arrangement by which a government takes, or promises to take, certain policy actions, in support of which an international financial institution or other agency will provide specified amounts of financial assistance’ (Killick, 1998: 6). The lending policy and conditionality of the IMF have been widely criticised not only by Killick himself but also by John Williamson in a highly influential article outlining the ‘Washington Consensus’ (1989), as well as by Mark. D. Copelovitch (2010) and other authors (see Collier and Gunning, 1999; Kapur and Webb, 2000), who have concluded that the link between conditionality and structural reforms is failing: ‘In the general case, conditionality is not a credible commitment mechanism. Conditionality cannot substitute or circumvent domestic ownership of and commitment to reform’ (Santiso, 2001: 8).

However, as explained by Dani Rodrik, when addressing the lending and conditionality capacities of international public lenders like the IMF and the World Bank: ‘[…] as long as multilateral agencies [i.e. international financial institutions] retain some degree of autonomy from the governments that own them, their interaction with recipient countries, while official in nature, can remain less politicised than intergovernmental links. This in turn endows multilateral agencies with an advantage in exercise of conditionality, (that is, in lending that is conditional on changes in government policies)’ (Rodrik, 1995: 3). Rodrik's conclusion brings up two issues in relation to this article: Firstly, this explains the direct capacity of public economic institutions to impose legally binding conditions. Private economic institutions such as banks do not have such direct power. Secondly, unlike international public institutions such as the IMF, private banks will not immediately leave the negative image of a foreign agent imposing its rules.

Conditionality has also been tackled from the point of view of power relationships between various stakeholders: the IMF, its staff, its Executive Board in the form of member states and countries subject to IMF programmes. The external incentives model is a rational-choice institutionalist framework that is popularly applied to conditionality exerted by the EU upon enlargement; it focusses on self-interested agents (accession countries) that willingly fulfil external requirements laid upon them by the European Commission to win the rewards of EU membership (Schimmelfennig and Sedelmeier, 2004). The model has been used when determining the policy success of IMF conditionality, in particular when referring to IMF-EU conditions during the ‘euro crisis’ (Featherstone, 2015: 6).

Incentives for domestic reform may not be as strong when seeking IMF financing as they are when desiring access to the EU internal market and other EU membership rewards (Sippel and Neuhoff, 2008). Dimitrakopoulos (2020) illustrates that in the case of the Greek IMF programme and the reforming of the tax administration, the model was useful in explaining the direction of reform, but not the final result. Alternative sociological institutionalist or constructivist scholars would argue that conditionality is stronger where staff and the borrowing country do not share the same normative orientations (Chwieroth, 2012, 2015), perhaps due to a form of distrust between parties of different cultural and ideological backgrounds. Featherstone (2015) argues that reform failures in the case of Greece's IMF programme could be attributed to a conditionality strategy that ignored conflicting domestic interests and, most importantly, administrative traditions and cultural norms.

While cultural differences can explain some of the friction surrounding IMF programme design and success rates, scholars inspired by the ‘social learning’ model would emphasise the role of culture and norm transfer in the process of change once a country under an IMF programme becomes subject to cultural or normative transformations. The work of Heinrich (2020) illustrates that rather than seeing conditionality as an instrument for forced policy transfer, conditionality is a means for international organisations to ‘help transfer the intellectual matter that underpins policies’ as countries go through stages of the perception, adaptation and translation of knowledge (Heinrich, 2020: 2). While this is useful when seeking explanations for conditionality effectiveness or failure, this vast body of literature is challenged by a shift in structures—the possibility that the IMF's conditionality has changed, potentially transforming the motivations or balance of power of various stakeholders.

There is limited academic literature for explaining changes in the Fund's lending and conditionality policy. This strand of research, which focusses on policy change, covers a range of theoretical ground, including both rationalist and constructivist aspects (Vetterlein and Moschella, 2014: 143). Case in point, the work of Guven (2018), provides evidence of a change in the IMF's conditionality towards the developing world before and after the 2008 crisis. He argues that it reflects an adjusted policy practice of greater flexibility, as well as decision-making discretion being placed in the hands of IMF staff due to challenging crisis circumstances. Interestingly, Guven (2018) and Broome (2015) conclude that the global financial crisis narrowed the IMF's conditionality back to its traditional realms of expertise or ‘core business’—fiscal and financial policy—reducing its social and governance engagement within programme cases. One of the reasons for the narrowing of policy fields covered by conditionality appears to have been the operational constraints that arose amid a transforming development finance scene globally, with more alternatives to IMF financing (Guven, 2018: 3) becoming available. It has also been noted that there has previously been a weakening of conditionality in terms of its form: ‘Hard’ conditionality has been more often replaced with ‘soft’ conditionality, meaning structural performance criteria have over time been replaced with structural benchmarks (Guven 2017: 1157). However, Guven still points to the persistent presence of conditionality and criticises the IMF for not being adaptive enough in the light of the changing external environment and its own operational constraints as ‘there is the continued weight of conditionality’ (Guven, 2017: 1165).

Other research on understanding changes in IMF policy has pointed out that increased lending, including the doubling of concessional lending (resources for low-income developing countries) is ‘spectacularly good for business’ (Clegg, 2012: 61). Crisis episodes could be a window of opportunity for IMF staff to expand its balance sheets on the back of arguments such as the need to fight poverty or increasing global inequality. He argues that a crisis legitimises bloating the IMF balance sheets and lending activities against the traditionally more ‘minimalist’ preferences of the United States. Nevertheless, it is clear that the ‘developmentalist’ behaviour of the IMF explained by Clegg is not quite that of a typical development aid player—the distributed funds are not grants, but rather favourable loans with conditionality and ‘ring-fenced spending’ (Clegg, 2012: 61). While policies may have become more flexible and friendlier to borrowers during previous crisis episodes, the removal of conditionality in 2008 was out of question.

The IMF has not been lacking criticism for its conditionality policy in the 60 years since its origin, and it has several times discussed and implemented tweaks to the policy against growing criticisms by borrowing countries (Best, 2012). Best finds that ambiguities have long existed in the conditionality policy when compared to the formal guidelines, with key actors within the staff, the Executive Board and stakeholder countries being aware of the discrepancy, and, in fact, welcoming it (Best, 2012: 3). Such ‘mission creep’ of policy practice has theoretical explanations in traditional principle-agent theory as a form of power struggle: The staff desires to expand the organisation's budget or seeks greater autonomy and discretion (Pollack, 1997). Nevertheless, the IMF staff may weigh the benefits of increased power and autonomy against concerns about a decline in legitimacy due to compromised neutrality or even-handedness, should the level of discretion become too large (Best, 2012: 19).

While there is ample academic literature on conditionality, there is limited existing literature on explaining changes in IMF conditionality. This is regretful, given the sharp and transformative external circumstances the IMF found itself during the COVID-19 pandemic in 2020. There is a clear literature gap to be targeted on the newest developments in IMF conditionality, most notably, the weakening of conditionality through conditionality-free loopholes in the existing lending toolkit. By building on the existing body of work on structural changes in conditionality (Best, 2012; Clegg, 2012; Guven, 2017, 2018), this article looks into new patterns of change via the use of the emergency financing instrument in 2020. Bringing the conclusions of the existing literature into the context of the COVID-19 pandemic, three arguments are derived regarding a potential change in IMF conditionality: (1) more discretion tends to be granted to the IMF staff during times of crisis due to the time pressure and complexity associated with times of crisis; (2) the IMF uses crises episodes to expand its balance sheets and thus increase its power position in the global financial safety net; (3) conditionality is subject to normative interactions between the various stakeholders, with knowledge transfer being based on dominant ideas that do not rapidly change.

The IMF's support to its members during the first year of the COVID-19 pandemic was timely, but not particularly large in size compared to the amounts disbursed during and after the global financial crisis.

Credit outstanding increased by US$41 billion between 2020 and 2021. Credit outstanding increased by US$101 billion over the course of 2008–2011. Note that disbursements have been steeper in the first year of the COVID-19 crisis than the GFC, but they considerably slowed down in 2021.

The actual lending figures increased by slightly less than that the IMF originally anticipated, amounting to around US$70 billion over the course of 2020. When taking a closer look at the instruments through which financing was delivered, it becomes strikingly clear that the IMF's crisis response was largely delivered through a single channel—the emergency financing instruments, namely, the RFI and the variant for poorer states under the Poverty Reduction and Growth Trust (PRGT), the RCF (See Figure 1). Over 80 countries received financial assistance from one of the two emergency financing instruments in 2020. This calls for a closer look at what emergency financing instruments really are and why they were scarcely used before the COVID-19 crisis.

Figure 1.

The instruments were set up in their present name and form in 2011 by simplifying and merging the previous Emergency Natural Disasters Assistance and the Emergency Post-Conflict Assistance streams of financing. As the word ‘rapid’ in the title of the instruments suggests, they were designed to give fast-track relief to countries with balance of payment pressures due to a natural disaster or exogenous shock. Health- or pandemic-related emergencies were never explicitly mentioned in the terms of use; in practical terms, the instruments were reformed multiple times to adjust access limits to meet the needs of small island states that are particularly prone to natural disasters that cause a significant drop in their GDP (IMF 2020a). The scope was more broadly interpreted in 2020 to include temporary shocks such as health pandemics.

A key element to the instruments' great popularity in 2020 might have been the fact that emergency financing was made available to countries in which setting up full-fledged upper-credit tranche (UTC) programmes was neither necessary nor feasible. Therefore, it was achieved without involving the typical ex post conditionality. As such, it became one of the rare instruments in the IMF's lending arsenal that does not demand any reforms or periodical reviews (IMF 2020a). Although many country stakeholders in the Executive Board repeatedly called for sufficient safeguards on the use of Fund resources (IMF 2020a), emergency financing instruments only have two minor elements that act as safeguards: (1) a letter of intent that the borrowing country drafts ahead of receiving financing and (2) an arrangement with an auditing company to prevent corruption in the use of the funds. Perhaps the more noteworthy built-in ‘safeguard’ for IMF resources was the low access limit. In 2019, it was only 50% of a country's quota per year (and 100% of quota on a cumulative basis). For comparison purposes, the average Stand-by Arrangement (SBA) with ex post conditionality was nearly three times higher—145% of quota in the same 12-month period.

The yearly access limit for the RFI and RCF was subsequently doubled to 100% of quota in the spring of 2020 (with the cumulative access limit rising to 150%). While the SBA access limit was significantly increased, SBAs were not the most favoured instrument during the first year of the crisis. As illustrated in Figure 1, the largest increase in IMF credit outstanding was due to emergency financing instruments (RFI/RCF), with roughly US$30 billion disbursed in 2020.

Another noteworthy development in 2020 was related to the use of precautionary lending instruments—the Flexible Credit Facility (FCL) and the Precautionary and Liquidity Line (PLL). These instruments are also exempt from typical ex post conditionality, though they do contain the so-called ex ante conditionality, meaning a country needs to pre-qualify by showing strong economic fundamentals. The FCL and PLL have higher access limits than the other IMF lending instruments, with a complicated, case-dependent access level assessment procedure. The amounts set aside for Chile and Peru in 2020 were 1,000% and 600% of quota, respectively—potentially 10 times higher than a single disbursement under emergency financing instruments. It can be granted to a country to meet a potential or an actual balance of payment need, but before the COVID-19 crisis, it was merely used as a precaution, without drawing on the sums committed to a country, acting as a type of ‘insurance’ or backstop. It is therefore particularly noteworthy that in April 2020, Morocco set the precedent of actually drawing on its precautionary envelope of US$3 billion, or 240% of its quota. In other words, in 2020, the previous stigma around drawing from precautionary instruments had faded against the pressing need for external financing.

As illustrated earlier, the new IMF lending patterns that emerged over the course of 2020 point to a weakening in IMF conditionality. This appears both through the widespread use of emergency financing, including through the doubled access limits, and through requests for and the simultaneous drawing of precautionary lending instruments, as in Peru's case in 2020. If we divide IMF lending instruments into two groups—those (1) with ex post conditionality, being the SBA and the Extended Fund Facility, versus those (2) without ex post conditionality, being the Flexible Credit Line, the PLL, the RFI and the RCF—then the following stands true for 2020: 70% of the increase in credit outstanding was due to instruments without ex post conditionality.

Source: authors' calculations based on IMF weekly financial statistics. Note that figures for credit outstanding, not committed, are used. The share ‘without ex post conditionality’ would be even higher if figures of committed resources had been used as precautionary lending instruments take up large sums that are not drawn upon.

As the World Health Organization declared the COVID-19 outbreak a pandemic on March 11 and lockdowns were put in place across the globe, the IMF felt the urgency to be prepared to effectively serve its membership with its US$1 trillion lending capacity. Mirroring a certain level of urgency present among state leaders across the globe, the original communication of the IMF contained expectations for a serious economic crisis, with a forecasted 3% fall in the global GDP in 2020 (IMF, 2020b). In the IMF policy paper Enhancing the Emergency Financing Toolkit—Responding to the COVID-19 Pandemic published on 2 April 2020, staff envisaged a spike in demand for IMF resources, especially from emerging markets and developing countries:

‘… there could be an unprecedented surge in demand for Fund financing over the next 6–9 months. The Fund has already received requests or inquiries from 87 members, for emergency financing or augmentation of existing arrangements, involving financing of around US$27 billion, based on current access limits. This number is likely to increase as the crisis unfolds—and where necessary emergency support will pave the way for follow up Upper Credit Tranche (UCT)-quality programmes—with the potential for Fund financing to substantially exceed that provided during other global crises’

Interestingly, the demand for IMF resources plateaued at around the mentioned 87 emergency financing requests, neither increasing far beyond the initial interest nor shifting to UCT programmes that would demand resources with higher access limits. Nevertheless, it was natural for the IMF to react to the acute projections as they were lacking previous experience in dealing with global health pandemics. Of course, reacting immediately would prevent the need to give more financing later, after bankruptcies and other economic troubles increase. In fact, the IMF expected the sharp fall in global GDP to be followed by a quick recovery a year later—the so-called V-shaped recovery:

‘In a baseline scenario, which assumes that the pandemic fades in the second half of 2020 and containment efforts can be gradually unwound, the global economy is projected to grow by 5.8% in 2021 as economic activity normalizes, helped by policy support’

This original thinking helps explain the background against which the Fund was making decisions: It was believed that may countries would ask for financing to weather the storm, but the external nature of the shock would mean that macroeconomic adjustment, and therefore an IMF programme with conditionality, would not be needed.

On 9 April 2020, the IMF reacted by shortening procedures related to the approval of emergency financing (the RFI and the RCF). Before the pandemic, the time it took for the formal approval of emergency financing would be between 3 and 4 months. With this decision, it was cut to 4 working days (for review by the Board and the circulation of documents). This is longer than the already existing Emergency Financing Mechanism (EFM) approval period of 48–72 h that was used at the time of the GFC in 2008. A special streamlining mechanism was established for pandemics, be it the COVID-19 health pandemic or any future pandemic, keeping in mind that all work by Fund staff and the Executive Board needed to take place in a more challenging, freshly established virtual format. The response to time pressure was justified in the IMF in the policy paper as follows:

‘The risks of contagion, both within and across borders, mean that even the normal lags associated with the approval and disbursements under the RCF and RFI of “urgent” emergency financing could be costly’

If procedures for adopting emergency financing instruments needed streamlining, slow-burning negotiations between the Fund, Board and the recipient country on the appropriate conditionality seemed even more out of question.

The second proposed reason for the weakening of IMF conditionality, which is also documented in the academic literature in the previous section, is the opportunity to increase the IMF's power position in the global financial safety net through expanded lending and balance sheets. The possibility to forego conditionality would lead to an increase in the demand for IMF resources, as financing could be accessed freely without conditions for painful reforms.

There is no denying that the IMF expressed a readiness to react both in words and in deeds, with a long queue of interested countries forming at the start of the pandemic, even though the amounts disbursed through the emergency financing instruments were not large:

‘The IMF has moved with unprecedented speed—providing emergency financing to 72 countries in 4 months—and we will continue to support our member countries relentlessly’

The swift reaction was also associated with the desire to remain relevant in the network of international financial institutions:

‘The IMF will explore additional tools that could further help in this crisis that is like no other and play our role at the centre of the global financial safety net’

It also has to be noted that the concessional lending landscape was changing, making it harder than ever for the IMF to remain relevant:

‘There have… been changes in the credit landscape facing low income countries (LICs), with concessional financing becoming scarcer relative to countries' investment needs and with an increasing number of LICs beginning to access financing from international financial markets’

Various interests are always at play in decisions being made by the IMF. Despite the pressures for the IMF to remain relevant advocated by its staff, there were divergent views as regards the need for maintaining loan safeguards, given that no ex post conditionality existed in emergency financing instruments:

‘Many Directors emphasized the importance of implementing appropriate governance safeguards to mitigate the misuse of emergency financing, and welcomed staff's guidance encouraging commitments related to audits and procurement’

This protest by some Executive Directors is also a signal that there was indeed a change in risk aversion in the IMF. The IMF's communication illustrates that there was a readiness to expand balance sheets, particularly vis-à-vis emerging market economies, despite the cost of lower safeguards or control over the use of funds, which normally would be ensured thorough conditionality.

Thirdly, the IMF may have weakened conditionality due to an ongoing normative debate around the usefulness of conditionality. The empirical findings point to a more normative change in the IMF conditionality policy that, arguably, had already intensified in the 2 years before the COVID-19 pandemic.

During the Global Financial Crisis, the IMF accelerated its lending through IMF programmes with the granting of 18 new General Resources Account (GRA) programmes in the 20-month period between January 2008 and August 2009 (IMF, 2009). The decade after the crisis was a time of reflection for the IMF, given the widespread criticism of its stringent conditionality policy that came from receiving countries and the academic community. A periodic review of IMF conditionality took place already a few years after the crisis, in 2011. The conditionality policy was considered broadly appropriate with a few lessons learnt, notably in regard to the euro area, where conditions were deemed too stringent and the fiscal adjustment too steep (IMF, 2011). It was decided to improve the targeting of the conditions and make them more parsimonious, to give greater consideration to socio-economic circumstances and to give more ownership and transparency to the authorities of borrowing countries. However, more time was needed to be able to evaluate programme success rates and to clearly understand what challenges the conditionality policy was facing after the spike in lending during the GFC. The subsequent review in 2018 was more sobering. It was indeed ‘the first comprehensive stocktaking of Fund lending operations since the global financial crisis’ (IMF, 2019: 1). Looking at the GRA (regular IMF lending) and the PRGT (lending for poorer sates) separately, the following disappointing conclusions were drawn regarding success rates:

GRA: ‘One-third of GRA programmes are assessed as successful, one-quarter as unsuccessful, and the remainder as partially successful. Successful arrangements typically managed to both reduce vulnerabilities and eliminate or significantly reduce the BoP [balance of payments] need. Partially successful programmes accomplished one of these two objectives’.

A long list of issues was identified as reasons for the low success rate: GDP growth estimations were too optimistic, fiscal consolidation was more harmful than anticipated, debt sustainability often deteriorated (mostly in programmes that went off-track) and communication surrounding improved ownership on the part of authorities could still be enhanced (IMF, 2019). In the review, the IMF argued that the overly optimistic growth assumptions (which was the case in most programmes) was the dominant factor that correlated with the rate of failure. At the time, the IMF decided not to change its Guidelines on Conditionality, but rather to implement its recommendations through the Operational Guidance Note on Conditionality and to integrate them in other Fund workstreams. However, Board discussions around the review pointed at the need for a more fundamental shift in the IMF conditionality policy: ‘successful implementation of the recommendations would require a change in culture, and continued adaptation and learning’ (IMF, 2019: 6).

The IMF's Independent Evaluation Office published the report Adjustment and Growth in IMF-Supported Programs a year before the COVID-19 pandemic, in 2019. It confirmed the existence of a growth bias present in the 2018 Review of Program Design and Conditionality (IEO, 2019). Furthermore, the findings pointed at an overall fall in GDP growth in the sample countries of both GRA and PRGT countries during Fund-supported programmes and in the 5 years following them (IEO, 2019). This was surely identified as a problem—‘[it was] indicative of an excessive tightening bias and resulted in a perceived stigma, potentially discouraging use of IMF financing and challenging the Fund's reputation’ (IEO, 2019: 2).

The empirical analysis illustrates that there was a wind of change in the debate around IMF conditionality in 2018–2019. While the academic literature has been critical of austerity since the GFC, the IMF needed to carry out its own internal evaluations and review process based on accumulated data and lessons learnt before any reconsiderations of the decades-old policy could be made. The findings of the reports were brought to the attention of Fund staff and the Executive Board only a year or two before the outbreak of the pandemic, providing a confusing normative background against which to launch a new wave of crisis lending. Continuing with the old course of action in the light of the lacklustre success rates of IMF programmes may have created legitimacy concerns in an institution that prides itself on data-driven and impartial analysis.