CGD NOTES

Multilateral Development Banks’ Crisis Response: What Instruments Do MDBs Offer and How Fast Do They Act?

Summary

In the face of sudden economic shocks and natural disasters that limit the ability of economies to rapidly raise revenue, MDBs have the potential to play a leading role in crisis response to minimize the damage. One key factor in their success in this role is the speed of their financing instruments—how quickly do they get funds to where they are needed after a crisis strikes? Using a calculated metric we term “relative disbursement rate,” our analysis finds that at both the World Bank and the Asian Development Bank (ADB), budget support operations are by far the fastest instrument, in some cases even outpacing crisis-specific instruments. We also find that most of the crisis-response specific instruments used by these institutions are contingency-based, meaning they are approved well before they would conceivably need to be disbursed. For these instruments, we find that speed is more accurately gauged by “response time” or the time between a crisis event and disbursement, and that at both institutions, this metric improved during the COVID-19 crisis. Together, these findings suggest that when it comes to crisis response, preparation is key—disbursement speed is much faster where instruments have advanced-design elements and policy actions that are already completed prior to approval.

Introduction: MDBs are important financiers for crisis response

MDBs have the potential to be at the forefront of crisis response for countries with limited domestic fiscal capacity. This is particularly the case when sudden shocks cripple the ability of economies to rapidly raise revenue and reallocate limited resources. Crises such as the COVID-19 pandemic, climate-induced hazards, and conflicts cause large-scale destruction of life, livelihoods and infrastructure. At present, MDBs provide a range of instruments that support countries in different ways and at different periods of crisis. Specifically, MDBs can: (1) provide financing to countries after crises strike; (2) pre-empt crises to be ready to step in quickly and minimize the time for funding to reach countries when crises strike; and (3) support countries in crisis prevention and mitigation, to minimize the chances and frequency of crises occuring, while maximizing cost-effectiveness and efficiency.[1] But how well do MDBs perform in using these instruments?

One leading performance measure of crisis response is its speed. As such, this analysis reviews crisis response instruments from the two largest MDBs – ADB and the World Bank – using speed as the primary aspect of interest. This analysis differs from previous CGD research on institution-level disbursement growth during Covid-19 in that our analysis focuses on relative speed across instruments.[2]We first take stock of each institution’s crisis-specific instruments, then analyze approval and disbursement speeds using two key measures (relative disbursement rate, RDR, and response time) to make suggestions for how crisis financing can be improved. Owing to the disparate nature of the World Bank and ADB datasets, we utilize slightly different methodologies when measuring performance. Because of this, results are not directly comparable across institutions. Instead, this analysis is intended to inform our understanding of each MDB’s respective crisis-specific instruments within their respective institutional contexts.[3] 

MDBs offer a variety of instruments to deal with crises

ADB provides multiple specialized crisis-specific instruments in addition to the non-crisis specific instruments offered under ordinary operations (see table 1).[4]Most are variations of the policy-based lending (PBL) instrument,[5]which, in normal circumstances, disburses general budget support to governments as pre-arranged policy conditions are met. One such variation which pre-empts crises and supports countries in crisis prevention and mitigation, is contingent disaster financing (CDF),[6]for which disbursement is made according to triggers concerning natural disasters or national emergencies. CDFs come with policy conditionality, but most are related to disaster preparedness requiring advance policy dialogue and implementation, and therefore rapid disbursement, when needed. Other variations take a different approach – instead of conditionalities, they simply require access criteria relating to the crisis and provide financing after crises strike. An example of this is the countercyclical support facility (CSF),[7]which was created in response to the Global Financial Crisis, and only requires fulfilment of general conditions relating to countercyclical investment. Similarly, in response to COVID-19, the ADB created the COVID-19 Pandemic Response Option (CPRO) under the general CSF,[8]for which the access criteria required countries to demonstrate that they are “proactively mitigating the current and potential future economic impact of COVID-19.” These budget-support instruments are complemented by the emergency assistance loan (EAL),[9]a traditional project loan with a medium-term disaster response framework which generally assists post-crisis response. Additional financing (see our section on “continued” projects) can also be drawn upon as extra project financing for disbursements during crises. Lastly, special assistance operations funded by financing partners (including both loans and grants) are available for smaller government spending needs.[10]

Table 1. ADB instruments

Instrument Acronym
Program Program
Policy-Based Lending PBL
Contingent Disaster Financing CDF
Countercyclical Support Facility CSF
Covid-19 Pandemic Response Option CPRO
Project Project
Emergency Assistance Loan EAL
Special Assistance Special Assistance
Sector Development Program SDP
Credit Line Credit Line
Activity Subgrant Activity Subgrant
Results-Based Lending RBL
Project Readiness Project Readiness
Sector Sector

Note: Red highlights crisis-specific instruments

The equivalent of ADB’s PBL instrument is known as development policy financing or lending (DPL) at the World Bank (see table 2).[11]DPL operations generally disburse as pre-specificied policy conditions are met. In situations of crisis or conflict, ordinary DPL speed can be further accelerated.[12]As far as we are aware, there are only two crisis-specific instruments currently on offer at the World Bank.[13]The first is a modified DPL, called a Deferred Drawdown Option (DPL DDO).[14]Unlike ordinary DPLs, once approved, disbursement only takes place when called upon by the borrower in the event of a negative economic shock. Disbursement can be deferred for up to 3 years, subject to one renewal (6 years in total).

Table 2. World Bank instruments

Instrument Acronym
Development Policy Lending DPL
Deferred Drawdown Option DPL DDO
Catastrophe Deferred Drawdown Option Cat-DDO
Investment Project Financing IPF
Program-for-Results Financing PFR

Note: Red highlights crisis-specific instruments

The second crisis-specific instrument, called a Catastrophe Deferred Drawdown Option (Cat-DDO) is based on the DPL DDO, but with two key differences: it is designed to provide liquidity to countries managing natural or health-related disasters rather than economic shocks, and its 3 year disbursement referral can be renewed up to 4 times, for a maximum of 15 years.[15]After creating the DPL DDO in 2001,[16]  the IBRD introduced the Cat-DDO to middle and lower-middle income countries in 2004, with IDA, the World Bank’s concessional lending arm, approving its first Cat-DDO in 2018.[17]As with CDFs from ADB, both DPL DDOs and Cat-DDOs disburse once triggered by a economic shock or natural disaster (respectively), but are approved well in advance of the potential crisis, minimizing any disbursement delays. Similarly, policy dialogue (usually centered around macroeconomic frameworks for DPL DDOs, or disaster preparedness for Cat-DDOs) takes place prior to approval, and implementation of conditions should take place well before disbursement. Like CDFs, both DPL DDOs and Cat-DDOs may be drawn down multiple times depending on the frequency of crises, so long as the country has not already utilized its full loan amount under the instrument. While there are only minimal restrictions for when a country can draw down a DPL DDO, for a Cat-DDO to be eligible for disbursement, a country must first satisfy a pre-specified drawdown trigger, typically declaration of a state of emergency, and then formally request disbursement from the World Bank. At that point, disbursement usually occurs in a matter of days, if not hours. Not every borrowing country had a DPL DDO or Cat-DDO in place prior to COVID-19,[18]but many that did chose to utilize the instrument during their response to the pandemic. Additionally, COVID-19 response was also channeled through the most traditional, non-crisis specific Development Policy Lending, Investment Project Financing (IPF), and Program-for-Results (PFR) instruments, despite the fact that none are specifically designed for crisis response.[19]As at ADB, there is also the option to use additional financing during times of crisis for speedy scale-up.[20]

Which instruments disburse the quickest?

To answer this, we calculate the relative disbursement rate (RDR) of different ADB sovereign lending instruments for projects approved between 2018 and 2021, by dividing the percentage disbursement of a given project by the amount of days between its approval date, and the latest available disbursement date (figure 1).[21]

Figure 1. Relative disbursement rate of ADB sovereign lending instruments, 2018-2020

An earlier CGD study analyzed how ADB’s CPROs have supported countries with unprecedented volumes and at rapid pace, when COVID-19 first hit in early 2020.[22]Indeed, the RDR of CPROs is speedier than the majority of other non-crisis specific modalities, but they underperform compared with traditional PBLs and their programmatic approaches (the fastest-disbursing non-crisis specific modalities on offer when measured by RDR). When analyzed by RDR, EALs and CDFs also appear to lag compared to other crisis-specific instruments. In the case of CDFs, this is a reflection of their contingency structure, with approvals taking place well before a crisis actually strikes. As such, response time – the time between a crisis event and disbursement – is the more informative metric for these modalities, as will be explored in the following section. As a variation of the traditional project loan (generally the slowest-disbursing modality) targeted at medium-term crisis response, EALs’ relatively slow performance is also unsurprising despite their classification as crisis-specific instruments.

It is unsurprising that PBLs (and its programmatic approach) are able to disburse relatively quicker than other instruments, since the conditions of disbursement (fulfillment of specific policy actions) are designated prior to approval and in some cases are already fulfilled upon approval. This means large amounts can be disbursed immediately once approved. We confirm this result by using a specific subset of ADB operations which had already been approved or started to be undertaken, which we collectively refer to as ‘continued’ operations. Continued operations include any projects which refer in their titles to: “additional financing,” (projects which added new funds to existing projects, with additional outputs and activities); “subprogram,” (projects consisting of at least two “parts” which support reform processes in stages, with basic designs of subsequent programs being pre-identified); and “tranche,” (any project which will disburse multiple tranches of funds over time through one approval). Given these are projects that are already designed or underway, they should be fast to reach any additional approval, and subsequently, disbursement (figure 2).

Figure 2. Crisis response process for continued operations

Indeed, when looking at the speed of response for continued vs. non-continued projects (figure 3), we find the “continued” subset ADB projects to be consistently faster disbursing than the non-continued sample across all years.

Figure 3. Relative disbursement rate for continued projects at ADB, 2018-2020

<p>We posit that one major factor delaying disbursement of committed funds of “Other” projects is the time required to undertake design during the initial response period – such that the initial steps outlined in figure 2 (design and approval) comprise a significant portion of response effort and time. For investment projects, this includes, at a minimum, country and subject diagnostics, procurement and financing plans and its associated administrative processes (including internal reviews). For additional finance projects, the design step is already taken care of, expediting the overall timeline as reflected in their faster RDRs.</p> <p>For the World Bank, we used data covering 2015-2021 and analyzed disbursement speed using the same RDR metric we used to compare ADB instruments (figure 4). Like the ADB’s PBLs, DPLs disburse as pre-designated prior actions are fulfilled. While DPLs are not a crisis-specific instrument, given their speed, they prove useful during crises. Previous CGD analysis has pointed out the shortcoming of DPL’s role in crisis response: many of the DPLs approved and disbursed during the COVID-19 crisis were burdened with excessive conditionality unrelated to the pandemic.<a data-cke-saved-href=" />[23]In addition, unlike during the global financial crisis, the World Bank did not significantly expand approvals of the instrument in the early days of the COVID-19 pandemic.  That said, even with such excessive conditionality and restrained involvement in crisis response, DPLs are considerably faster than all other non-crisis specific instruments, IPF and PFR.

Figure 4. Disbursement period and percentage disbursed of World Bank sovereign lending instruments, 2015-2021

There have been only 22 Cat-DDOs and 3 DPL DDOs approved since 2015[24]– but those graphed show their disbursement speeds lag behind DPL. One reason for this is because both instruments only disburse in the event of an economic shock or crisis, so that if such an event do not take place, the instrument will not disburse. As mentioned above, this is also true for ADB’s CDFs, which were designed to align with the World Bank’s Cat-DDOs.[25]

To get a better sense of the disbursement speed in the context of COVID, we tagged the subset of World Bank projects with “COVID” in their name and compared their average RDR to the average among the other projects in our dataset, maintaining our bucketing by modality. Results reveal that, except for PFR projects, COVID-19 -specific projects (classed as those with COVID-19 in their title) disbursed more quickly than non-COVID-19 projects.[26]DPLs, the quickest disbursing instrument, disbursed at an average rate of 4 percent per day, compared to only 2.3 percent per day for non-COVID-19 projects. For IPF projects, the margin was even greater – 0.11 percent per day for COVID-19 projects compared to 0.03 percent for non-COVID-19 projects.

Figure 5. Relative disbursement rate for COVID-19 and non-COVID-19 World Bank projects, 2015-2021

Up to now, the analysis has looked at RDRs across the suite of all available instruments to ADB and the World Bank. However, as mentioned previously, when looking at contingent crisis-specific instruments, initial approval time is not as important since such approvals take place well before the crisis event (see figure 2). Rather, the strength of these instruments is their ability to disburse quickly after the crisis hits because reform policy actions have already been completed at the time of approval. Hence, the important measure of speed in the case of contingent crisis-specific instruments is the time between the date when the crisis hits and disbursement. We term this the ‘response time’ and here, we specifically analyze CDFs and Cat-DDOs across COVID-19 and non-COVID-19 crises.

Crisis response time of contingent instruments: COVID-19 vs. previous crises

To determine a “response time” value for each project, we calculate the time between crisis date (date of crisis occurrence) and first disbursement. We merge our ADB dataset with an external international disaster database and determine whether the project was COVID-19 related (defined as whether it had COVID in the name, or was a CPRO),[27]or a response to another crisis.[28]For COVID-19 projects, we use the first reported COVID case date for each country, or, for countries where the first case occurred late in 2020, or not at all, the date where a State of Emergency or multi-country travel ban was implemented. For non-COVID crisis-response projects, the crisis date is as listed on the external international disaster database.

We observe that ADB’s COVID-19-related operations’ response time was slightly quicker than other crisis-response operations, by approximately 6 days. We attribute this pickup to three causes: (1) Instruments used to respond to the COVID crisis, namely CPROs, had fewer processes to go through before approval. CPRO operations had no procurement plans, which many other modalities would have had to include. (2) Given that countries were struck by the pandemic with varying speed and intensity, some had advanced warning of the crisis before the population was infected, allowing them and the ADB to prepare response packages before the “crisis date” used in our data set. (3) Given the widespread medical nature and urgency of the pandemic, barriers to approval that may have existed for other crises which respond to more politically sensitive crises were absent.

For the World Bank, the mechanics of Cat-DDOs require a slightly different data construction method and analysis than for ADB data on CDFs. As Cat-DDOs can disburse multiple times and requires declaration of a state of emergency and disbursement request by government, we manually examined the circumstances of each individual disbursement within each Cat-DDO (each tied to an individual crisis date).[29]We find that Cat-DDOs’ average response time hovers at approximately 29 days. Accounting for COVID-19 projects exclusively, this falls to around 21 days, reflecting the urgency of the pandemic. We did not extend our analysis of response times to DPL DDOs, which, given their much more general disbursement criteria, are more challenging to study from the crisis-to-response framework with which we analyze Cat-DDOs.

Conclusion

This study has analyzed the array of crisis response instruments and their speed of response, for two major MDBs, ADB and the World Bank. We find that both ADB and the World Bank make the quickest disbursements, whether crisis-specific or not, primarily in the form of budget-support (PBLs and DPLs and their variations). Our analysis of RDRs suggests preparation is key: disbursement speed is much faster where instruments have advanced-design elements and policy actions that are already completed prior to approval, such that large amounts can be disbursed in large chunks immediately upon approval. This result holds for ‘continued’ operations such as subprograms and additional financing, and some crisis-specific instruments (Cat-DDO, DPL DDO, CDFs).

While ADB introduced a new instrument, CPROs, for COVID-19 response, by looking at RDR it is unclear whether CPRO speed was markedly superior to pre-existing instruments such as PBLs and programs. However, COVID-19 response was rapid in the sense that projects were approved quickly relative to “crisis date” than other crisis response projects, whether using crisis-specific or non-crisis specific instruments. On the whole, the World Bank has a more limited list of crisis-specific instruments than the ADB, but one of its crisis-specific instruments, Cat-DDO, was particularly fast in responding to COVID-19.

To understand which instruments are better suited to respond to different types of crises, we need to further understand and disaggregate complete project timelines. Here we have analyzed the time frame between approval and disbursement, using RDRs, as well as the time frame between crisis and disbursement, using “response time.” Further disaggregation would need to assess the time between crisis and approval for non-contingent instruments, especially PBLs and DPLs. Once such dis-aggregations are done, it would be possible to assign instruments according to available lead-time/advanced warning for MDBs and borrowing countries. We also advocate the use of instruments which incentivize the application of advanced warning mechanisms and criteria to trigger approvals and disbursements sooner, or in advance, of crises. Similarly, any tweaks to instruments to accommodate crises that are not necessarily anticipated (much like COVID-19) would elevate their usefulness.

Specifically for instruments which require multiple disbursements, it is important to note that disbursement performance could be affected by a country’s experience in dealing with MDBs, and client relations. Countries that have more experience in understanding the range of available instruments and what they offer, may be better versed on how to draw upon resources faster. Accordingly, ADB - which has a higher proportion of wealthier countries with more developed capacities in its portfolio than the World Bank - may experience faster disbursement rates and our analysis here does not offer direct comparisons of the two MDBs for this reason. Rather, the analysis places the instruments on offer side by side as a starting point for further analysis. With more detailed data over time, it will be possible to make more direct inferences about relative performance and therefore evaluate the strengths and weaknesses of the different range of instruments and their characteristics on offer. Nonetheless, investing resources to increase technical capacity for procurement and financial management, among other technical support, should be beneficial for both governments and MDBs when considering how fast and useful crisis response instruments can be.

MDBs have the potential to support countries quickly when crisis hits. But prior actions and preparations significantly determine how fast resources reach those most in need. Further, given our concentration on analyzing speed, we do not explore here the quality of operations. While budget support and crisis-specific instruments clearly prioritize speed and urgency of response, achieving quality results and desired outcomes is challenging, and requires more attention.  To optimize crisis financing, MDBs should be clear on the available menu of instruments that look at development outcomes, continue to tweak and minimize administrative processes and support countries to implement projects which emphasize pre-design and approval, timed in advance of crises rather than responding once crises happen. Crises related to climate change and pandemics are inevitable; being under-prepared is not.

The authors are grateful to Jiro Tominaga and Clemence Landers for helpful comments on an earlier draft of this note.


[1]Idris, Iffat. “Cost-Effectiveness in Humanitarian Work: Preparedness, Pre-Financing and Early Action.” Governance and Social Development Resource Centre, May 13, 2020.

[2]Duggan, Julian, Scott Morris, Justin Sandefur, and George Yang. “Is the World Bank's Covid Crisis Lending Big Enough, Fast Enough? New Evidence on Loan Disbursements.” Center for Global Development, October 12, 2020.

[3]ADB data come from proprietary sources with limited information on disbursement. Because of this, we use closing date as a proxy for final disbursement date of projects which are 100% disbursed. However, since the final disbursement date generally precedes closing, this methodology likely underestimates the speed of ADB disbursement across all instruments. For the World Bank, we used the publicly available IBRD Statement of Loans and IDA Statement of Credits and Grants, which contains detailed information about disbursements at multiple points in time for any given project. As a result, our disbursement rate calculation for World Bank projects is more precise and less likely to underestimate the speed of disbursement across projects.

[4]What We Do: Lending and Grant Modalities.” Asian Development Bank.

[5]The programmatic approach is the most common mode of PBL, and finances a series of subprograms (over time, across levels of government, or inter-sector).

Strategy, Policy and Partnerships Department. “Operations Manual: Policy-Based Lending.” Asian Development Bank, January 19, 2021.

[7]Ibid.

[8]ADB’s Comprehensive Response to the COVID-19 Pandemic.” Asian Development Bank, April, 2020.

[9]Working Paper: Revised Emergency Assistance Loan Policy.” Asian Development Bank, June, 2021.

[10]What We Do: Funds and Resources.” Asian Development Bank.

[12]Results Office of the Vice President. “Bank Policy: Development Policy Financing.” World Bank Group, August 2, 2017.  

[13]Other instruments, such as Investment Policy Financing (IPF) or Program-for-Results (PFR) can be used during crisis but are not crisis-specific. See for a full list of such instruments used during the COVID-19 crisis. See

Saving Lives, Scaling-up Impact and Getting Back on Track World Bank Group COVID-19 Crisis Response Approach Paper.” World Bank Group, June 2020.

[14]Product Note - Deferred Drawdown Option: Major Terms & Conditions.” The World Bank Treasury, January 26, 2021.

[15]Product Note - Catastrophe Deferred Drawdown Option (Cat DDO).” The World Bank Treasury, January 26, 2021.

[16]Lending a Hand when Catasrophe Strikes.” Intergovernmental Panel on Climate Change.

[17]Product Note - Catastrophe Deferred Drawdown Option (Cat DDO).” The World Bank Treasury, January 26, 2021.

[18]Only three DPL DDOs were included in our sample, two of which disbursed during COVID-19.

[19]What We Do: Financing.” World Bank Group.

The World Bank previously had a pandemic emergency financing facility, but this was closed in 2021. See

Ritchie, Euan, and Mark Plant. “A Good Idea Executed Badly: Why the World Bank Should Not Renew the Pandemic Emergency Facility Insurance Window.” Center for Global Development, April 9, 2020.

[20]While 'Additional Financing’ operations made up a significant portion of the World Bank’s COVID response, most are not included in the IBRD Statement of Loans and IDA Statement of Credits and Grants datasets used to drive our analysis. Since these datasets are what provide us with information on disbursement timelines, we were unable to calculate RDRs of ‘Additional Financing’ operations at the World Bank in the way that we do for ADB.

[21]Data from the ADB projects and tenders page. Here, relative speed calculations do not include time between crisis occuring and approval date. We take this into account in a latter section.

Project & Tenders.” Asian Development Bank.

[22]Sato, Azusa, Rakan Aboneaaj, and Scott Morris. “How Effectively Is the Asian Development Bank Responding to COVID-19? An Early Assessment.” Center for Global Development, August 24, 2021.

[23]Landers, Clemence, Rakan Aboneaaj. “World Bank Budget Support in the Time of COVID: Crisis Finance… with Strings Attached.” Center for Global Development, July 8, 2021.

[24]Not including CAT-DDOs currently in the World Bank project pipeline, but not yet operational.

[26]Sample size of COVID-19-specific Program-for-Results Financing projects was very small (n=2), which may explain why the trend found across other World Bank instruments does not carry over. 

[27]EM-DAT: The International Disaster Database.” Centre for Research on the Epidemiology of Disasters (CRED).

As of Q1 2022, ADB will retrospectively tag COVID projects as: (i) all sovereign operations using the PBL modality, (ii) education projects with at least one social protection component, (ii) all health projects, and (iii) public sector management projects focusing on social protection. Note that this is a much broader categorization than the one utilized here.

[28]Crisis projects include CDFs, CPROs, Special Assistance projects, and anything with “Emergency” in the name. We use a different crisis-dating methodology for COVID-19-related crisis projects since the varied nature of the pandemic’s impact on different countries necessitates country-specific crisis dates. 

[29]The sample size for Cat-DDOs, is 24, at least 11 of which disbursed in response to COVID-19. If available, the date of the request was used as the crisis date. If unavailable, we used the date of the declaration of a state of emergency, and, in the absence of that, we used the crisis date itself (irrelevant for covid, since all countries at some point declared a crisis). Using the date of a declaration of emergency was sometimes a poor approximation of the actual request date that would have precipitated disbursement – for instances where this date resulted in a response time of greater than 150 days, we removed the observation. In addition, rather than a single response time measure being calculated for each project, we repeated the response time calculation for each individual disbursement within a project, and then averaged them.

Rights & Permissions

You may use and disseminate CGD’s publications under these conditions.