Policy-based lending (PBL) offered by ADB provides ADB developing member countries (DMCs) with fast-disbursing and untied financial resources. These loans are conditional on reforms agreed by the government and are designed to help meet budget financing needs, including the cost of the reforms.1
Initially known as program lending, the modality was first introduced by ADB in 1978, although its purpose and design has changed significantly since then.2
PBL initially provided liquidity (in a foreign currency) for countries in balance of payments distress, so its initial purpose was wholly crisis-related. The emphasis was on rapidly disbursing financial support to enable countries to purchase essential imports and to support macroeconomic stability, with little or no policy reform mentioned.3
However, conditions were soon added and, in 1983, PBL disbursement in ADB became explicitly linked to policy reforms, in line with the approach followed by other multilateral development banks (MDBs).4
Thus, in addition to crisis periods, PBL was increasingly used in non-crisis periods to respond to the financing needs of countries’ development programs with funds disbursed against policy reforms.5

• 1
  
  The term “financing” in this context encompasses grants and concessional loans in addition to non-concessional loans.
• 2
  
  ADB.1978.Program Lending. Manila.
• 3
  
  Funds were disbursed against a positive list of imports; in ADB’s case, against the importation of agricultural inputs. However, as money is fungible, this practice was replaced over time by disbursing financial support against an agreed list of policy reforms.
• 4
  
  ADB. 1983. A Review of Program Lending Policies. Manila. ADB reformulated its program lending to explicitly support policy reforms in this document. The reformulation included adding a policy matrix of required reform actions and a development policy letter that set out the country’s intended reform program.
• 5
  
  The link between the program size and the policy reform adjustment cost was, however, formally dropped in 2011, when it was decided to mainstream programmatic budget support “through determining the overall loan size based on the development financing needs of a country, with reference to specific elements of the development expenditure programs supported by the budget support (that may not include the costs of adjustment directly related to implementation of the envisaged reforms) and, where relevant, support provided by other development partners.” ADB. 2011. Review of Policy-Based Lending. Manila.

PBL has two objectives: to respond to country financing needs, generally indicated by a fiscal deficit, and to support policy reforms. In practice, however, the balance between these objectives is not automatic. In crisis periods, for instance, the balance tips toward financing because the short-term need for fast-disbursing liquidity (in a foreign currency) to support macroeconomic stability trumps the long-term need for policy reform, as demonstrated by the recent response to COVID-19, which so far has been largely budget support. 

This section argues that meeting country financing needs has been the primary objective of ADB PBL even in non-crisis periods, and that policy reform is the secondary objective. Policy reform is definitely an important objective, but for PBL to be effective in the policy arena it requires other inputs, including policy diagnostics and technical assistance, which go beyond budget support. These have not always been provided. Crisis periods have prompted ADB to introduce reform-free PBL variants.11
For instance, in response to the global financial crisis of 2007–2009, the food crisis of 2007–2008 and more recently the COVID-19 pandemic, ADB has used reform-free modalities that support an increase in public expenditure as a countercyclical response to an economic downturn. The switch to single-tranche PBL over the evaluation period also signaled that more predictable and reliable financing was in fact the primary objective of PBL and that there may be a trade-off between financing and reform. ADB also engages in policy reform without the use of PBL, e.g., through covenants attached to investment project loans or stand-alone technical assistance.

• 11
  
  In these PBL operations, reforms are substituted for countercyclical budget expenditures, which is also a policy response (a fiscal policy response).

ADB uses four PBL types: stand-alone, programmatic, countercyclical, and special.21
Within this group, there are also sector programs, which combine investment projects with PBL. Stand-alone PBL is a single loan usually configured into two or more tranches.22
Programmatic PBL is a series of single-tranche loans, each supporting policy reforms that are sequenced over time but linked by a common, longer-term development objective. While both these PBL types can support reforms over the longer term, in practice they are very different loans, and this is discussed in more detail below. In contrast, the Countercyclical Support Facility (CSF) and special PBL (SPBL) are crisis-response instruments. Following a crisis, CSF helps finance a fiscal stimulus that aims to protect the most vulnerable during an externally triggered economic downturn. The newly introduced COVID-19 pandemic response option (CPRO) also supports a fiscal stimulus to help countries tackle the economic and social impact of the COVID-19 pandemic.23
SPBL is used to support a country during a balance of payments crisis and until recently had never been used, despite being introduced nearly 3 decades ago.24
In addition to these different PBL types, ADB also uses PBL as a precautionary instrument in case an economic or disaster-related crisis may occur. In this case, policy reforms are directly related to reducing risks, e.g., building disaster resilience, which may include strengthening economic and fiscal resilience, providing social safety nets, and carrying out disaster preparedness. The use of PBL as a precautionary instrument that can respond rapidly in the aftermath of a disaster was formalized by ADB in August 2019.25
Precautionary PBL that anticipates the onset of a future economic crisis has not yet been widely used. 

Stand-alone and programmatic PBL have inherently different approaches to loan conditionality. Stand-alone PBL with several tranches is approved on the condition that the borrower undertakes policy reforms in the future, against which the loan is disbursed.26
It creates more uncertainty than single-tranche PBL since borrowers need to carry out the policy reforms attached to the loan after it has been approved, which can make timing and disbursement of subsequent tranches uncertain. The further into the future the reforms are, the greater the risk, which reduces the capacity of the PBL to provide timely financial support. Previous evaluations have noted that stand-alone operations have often resulted in significant delays to disbursements, or that waivers have been sought when tranche release conditions were not met. Evaluations of ADB PBL in 2001 (footnote 6) and 200727
described how overloading the second and subsequent tranches with complex and numerous conditions had led to severe disbursement delays, waivers, and cancellations, which was also experienced by other MDBs (Box 1.1).

To overcome the difficulties with stand-alone PBL with several tranches, around the turn of the century MDBs began making increasing use of single-tranche operations, either on their own or as part of a series of single-tranche loans in a programmatic series. ADB introduced programmatic PBL in 1999 and began to use it more frequently from the mid-2000s.28
 Policy actions in single-tranche PBL operations are completed before loan approval, eliminating the delays and cancellations associated with stand-alone operations, and increasing the likelihood of the PBL being successful. The share of stand-alone PBL in the ADB portfolio declined significantly over the evaluation period.29
 The shift to single-tranche PBL meant that loans could now disburse with certainty because policy reforms had been completed in advance. At the same time, the type of reforms supported also changed, and perhaps their ambition has also diminished. This is discussed in more detail below.

• 21
  
  New PBL types have also recently been established to respond to COVID-19.
• 22
  
  Stand-alone can also be designed as a single tranche.
• 23
  
  ADB. 2020. ADB’s Comprehensive Response to the COVID-19 Pandemic. Manila
• 24
  
  SPBL was used for the first time in 2019, as part of an international financial package to provide balance of payments support for Pakistan.
• 25
  
  Available https//www.adb.org/news/adb-introduces-contingent-disaster-financing-natural-disasters
• 26
  
  Although the first tranche in multitranche PBL is disbursed against policy actions already undertaken. Conditions in the second tranche are undertaken in the future
• 27
  
  OED. 2007. Policy-Based Lending: Emerging Practices in Supporting Reforms in Developing Member Countries. Manila: ADB.
• 28
  
  ADB. 1999.Review of ADB’s Program Lending Policies. Manila.
• 29
  
  In the post 2008 cohort of validated project completion reports, the number of multitranche PBL operations fell by two-thirds from 61 to 22, while the number of single-tranche operations increased by the same proportion from 9 to 27.

PBL has supported reforms across a broad range of sectors and policy areas in a variety of countries, from upper middle-income countries to fragile states emerging from conflict, countries impacted by natural disasters, and countries hit by the coronavirus pandemic. Over the last two decades, however, the focus of reform, in nearly all country categories, has shifted away from agriculture, natural resources, and rural development (ANR) to public sector management (PSM)—Figure 1.3. Before the late 1990s, on average 61% of the total number of approvals per year supported ANR, but lending for policy reform in agriculture has sharply declined since those days. In the period leading up to the Asian financial crisis, PBL support shifted to financial sector reforms30
 and, since the mid-1990s, ADB has used PBL mainly to support broad public sector management (PSM) reforms in such areas as public expenditure and fiscal management, public administration, economic affairs management, decentralization, state-owned enterprises, law and judiciary, and social protection. The number of PSM operations escalated in response to the global financial crisis, as it did at other MDBs. PSM accounted for nearly two-thirds of all PBL approved over the evaluation period, peaking in 2008 (78%) and 2009 (87%) and PSM continued to be the focus of over 60% of PBL in 2010–2020 (Figure 1.3). In the first half of 2020, PSM accounted for 73% of all PBL approvals.

ANR = agriculture, natural resources and rural development; FIN = finance; PSM = public sector management; IAT = industry and trade; INFRA = energy, transport, and water and other urban services; OTHERS = education, health, information and communication technology, and multisector.

Source: Asian Development Bank (Controller's Department) database.

Surprisingly few PBL operations are to be found in ADB’s key areas of investment—transport, energy, and water infrastructure—the areas in which ADB has most expertise.31
 PBL operations in energy and the financial sector increased only in the last 3 years of the evaluation period 2008-2017. Recent approvals in the financial sector have focused on capital market development and improving the environment for private sector investment, rather than on the areas that appeared before 2008, such as privatization of state-owned banks or the appropriate policy and institutional structure for developing rural financial services. However, it is the relevant and changing role of the state in public service management and the delivery of essential services that is the focus of attention in policy reform and hence the areas of investment may be less important than arriving at a clearer understanding of why particular reforms are selected for PBL support. For instance, ADB selected capital market development as the focus for PBL support in Bangladesh, not transport or energy development, despite the fact that reforms were needed for longer-term investment sustainability in these areas.32
 ADB also selected capital market development in Sri Lanka, despite the fact that energy sector pricing was well below the cost of production and the tax ratio to GDP was among the lowest in the world.33

In a limited number of cases, PBL is packaged alongside investment lending and TA into a sector development program. However, the use of this type of program, combining all three instruments in a single intervention, has declined since the early 2000s. Nevertheless, there are still examples of investment projects and PBL being used in parallel to help improve the policy context for investment projects. However, these are generally not packaged as an SDP. For instance, in Indonesia, ADB supports energy sector reform through a PBL operation without an explicit link to its energy sector projects. 

Nevertheless, as mentioned above, given the scale of ADB investment in the development of infrastructure in the energy, water, and transport sectors in Asia and the Pacific, there is a notable lack of PBL-supported reforms in these sectors, even though infrastructure gaps were identified as key constraints on growth and poverty reduction in ADB’s long-term strategic framework, Strategy 2020.34
 These are also areas where ADB has significant expertise and capacity to influence reform agendas and, given the scale of its infrastructure investments, one expect to see ADB capitalizing on its long-term engagement in these sectors. While ADB can address specific sectoral issues through covenants in investment loan agreements and through technical assistance, this may not be sufficient to remove binding constraints on growth in areas where political economy issues are deeply entrenched and go beyond the concerned sector. For example, the removal of consumer subsidies for energy, water, and transport requires broader engagement with central government agencies and other stakeholders.

• 30
  
  Financial sector reforms often overlap with PSM as the focus of reform is about developing appropriate financial market regulation, removing the state from bank ownership, or introducing appropriate policies for financial market development and competition.
• 31
  
  This is not to say that ADB does not support sector reforms but, rather, that these are often supported through project interventions and not necessarily through PBL.
• 32
  
  IED. Forthcoming. Country Assistance Program Evaluation: Bangladesh, 2010-2020. Manila: ADB.
• 33
  
  IED. 2016. Country Assistance Program Evaluation: Sri Lanka, 2006-2015. Manila: ADB.
• 34
  
  ADB. 2008. Strategy 2020. The Long-Term Strategic Framework of the Asian Development Bank, 2008–2020. Manila.

This section discusses the performance and results of ADB PBL, especially in those countries that have used PBL the most. Measures of PBL performance were drawn from project completion reports that have been validated by the Independent Evaluation Department (PVRs) while the broader focus on development results draws from evidence of achievements at the country level. The performance of PBL operations since 2008—as measured by the success ratings assigned by PVRs—has improved sharply compared with those approved before 2008.35
 Other MDBs have witnessed similar improvements. While several factors have contributed to the dramatic turnaround in ADB, two stand out: (i) the increasing use of single-tranche PBL since 2008, and (ii) the shift in the reform focus to public sector management, particularly public finance management.36
 Nevertheless, while the performance of PBL improved over the evaluation period, attributing country development outcomes to ADB-supported policy reforms is not straightforward, as is discussed in more detail below.

• 35
  
  The evaluation portfolio is all policy-based loans approved and evaluated since 2008. Policy-based loans evaluated post 2008 are included in the sample even if they were approved before the period. The validated project completion reports of 49 PBL operations approved since the beginning of 2008 found that 80% were rated successful, almost twice that for the 70 PBL operations approved before 2008.
• 36
  
  The evaluation referred to five factors: including the high proportion of regular PBL in the portfolio that responded to the 2008–2009 global financial crisis, 93% of which were rated successful; the reduction in the number of policy actions; and a reduction in the share of Pakistan operations in the portfolio, a higher number of which were rated unsuccessful. These factors were trends in ADB performance data rather than structural changes in PBL design and reform focus.

The notable increase in PBL performance since 2008, as judged by the high success rate in project completion reports validated by IED, should indicate that these loans are very well-designed, especially in relation to policy reform. Nevertheless, because PBL is presented to the Board in the same format as an investment loan and, in turn, is assessed at project completion as if it were an investment loan, this has raised several design issues. 

Macroeconomic assessment. First, like other MDBs, ADB relies too much on the IMF for its macroeconomic assessments, yet this assessment is rarely presented in the approval documentation because the investment loan template does not require it.58
 In Indonesia, for instance, none of the reports and recommendations of the president (RRPs)—the ADB approval documents—for PBL that were reviewed as part of the evaluation adequately assessed the macroeconomic framework or contained tables presenting key macroeconomic indicators or forecasts. In the Philippines, RRPs for PBL typically discussed the size of the budget deficit and how an operation would help close the financing gap, but this was rarely linked to a macroeconomic assessment. In Bangladesh, PBL designs seldom contained a judgment on whether the macroeconomic policy framework provided a satisfactory backdrop for PBL use. Staff guidance on what constitutes essential aspects of the macroeconomic framework is not clearly set out in the ADB Operations Manual. None of the minutes of the management review meetings for PBL approved in 2016 for example, raised any issues in connection with the macroeconomic framework. Nor was the adequacy of the macroeconomic framework considered at project completion. The appropriateness of injecting liquidity (in a foreign currency) into the macroeconomic framework of a DMC is taken as a given at approval, supported by an attached IMF assessment letter, and completion. The evaluation found only one case where ADB’s decision to proceed with a PBL was contrary to the IMF’s position. This implies that ADB itself should maintain a robust capacity to evaluate the views of the IMF on the macroeconomic conditions of the country as part of PBL design. The IMF assessment letter cannot be definitive, as the decision to proceed with the PBL lies with ADB. The risk implied by the IMF assessment, including the reputational and precedent-setting risks of acting contrary to the IMF views, must be fully assessed, and borne, by ADB.

Political economy analysis. While a macroeconomic assessment should underpin the use of budget support, the policy reform side of a PBL also requires good analytical work and technical assistance. Good political economy analysis is also important for sound loan design. The evaluation found that, while PBL designs drew on available political economy analysis, such work was rarely undertaken specifically in designing PBL operations. As a result, the political feasibility of, and risks associated with, specific PBL-supported measures tended not to receive much focus. RRPs for PBL operations in Bangladesh, the Philippines, and Viet Nam did not feature any significant political economy analysis, despite the substantial political risks involved. Similarly, in Indonesia, substantive political economy analysis was seldom part of the preparation for PBL, although interviews suggested that some analysis was done informally. Several programs in Pakistan would have benefited from political economy analysis. The sector analysis that was prepared for several ADB PBL-supported programs fell noticeably short of adequately assessing the economic and political dynamics that would be crucial determinants of ownership and commitment. Starting in the early 2000s, ADB built up significant PBL support to help shape local government systems and service delivery. The devolved social services programs (DSSPs) in Sindh,59
 Punjab,60
 and Balochistan,61
 for example, sought to develop formula-based, performance-oriented grants to district governments, despite providing little evidence to show that the provincial governments were genuinely interested in reducing their discretion over the financing of local governments (one provincial government later went so far as to claim that the DSSP was developed over its objections). Similar problems plagued the implementation of many other provincial PBL operations. In hindsight, the design of ADB support for the devolution process in Pakistan was based on an insufficient understanding of the political economy risks involved and this was a key reason why the PBL operations from that time were rated less than successful.

Analytical work. The evaluation found that, while the majority of PBL appeared to be informed by analytical work, it was difficult to find clear references or links to the work that had been undertaken. Generally, the key conclusions of such analysis were not described in the RRP and in most cases references to the analysis that provided the rationale for the PBL were unclear. One reason why this information was missing is that the RRP template introduced in 2010 was not tailored to the PBL modality. In Bangladesh, for example, two of the five PBL programs (on urban public health and regional trade facilitation) did not appear to be based on detailed diagnostic work. The sector assessments underlying several of the PBL operations were very general and often did not point to specific policy actions or reforms. Without clear analytical underpinnings, it was often unclear how policy actions were derived or why they were selected. In several cases, there was too little prior analysis to adequately inform PBL preparation. This was particularly the case in Pakistan. It was therefore often difficult to assess how specific policy actions would help to address policy issues and constraints.

Quality of the results framework. The ability to evaluate the policy reform side of PBL ultimately depends on the quality of the results framework in the design document. In ADB this is known as the design and monitoring framework (DMF), which is an appendix to the RRP, and includes the logic (or theory of change) linking program inputs with outputs and outcomes and the results indicators selected. In some MDBs, the policy matrix itself (i.e., the list of policy actions to be undertaken), has gradually also become the results matrix. In ADB, the policy matrix and the results matrix (the DMF) are separated, so linking policy actions, the first step on the results ladder, with results (outputs and outcomes in the DMF) is not as straightforward as it could be. 

The evaluation found several issues with the quality of the DMFs. Some of these related to basic attribution problems, which is a challenge for all MDBs. The use of very high-level outcomes in the DMF—e.g., GDP growth, levels of private sector investment, employment creation, and poverty reduction—which are influenced by many other factors exogenous to PBL reforms, limits any possibility of attributing significant changes in these areas to ADB-supported policy actions. Conversely, setting results that are too output-oriented, and which are identical or similar to policy actions, as some PBL operations do, is also a problem because it confuses inputs with intended changes generated by the policy actions, and may mislead the assessment of PBL results. Nor should the output and outcome statement be the same. 

Assessing the quality of the DMF for PBL is complex. Finding clear links between the policy actions supported and the outputs and outcomes stated in the DMF is often difficult, especially when there are numerous, often overly process-oriented policy actions, with results that are unclear or difficult to measure in practice. Such policy actions can include the approval of a document, the submission of a document to cabinet, the approval of an institution’s organization chart, and staff training. The evaluation found that over 50% of policy actions were overly process-oriented, with results that were not measurable or only distantly related to the outcome. Moreover, the number of policy actions was correlated with loan size, suggesting that the ADB Board of Directors expected to see a greater quantity of reform effort with larger loans. However, the size of the loan is determined by country financing needs, not by the cost of policy reforms. Moreover, this approach indicates shortcomings in PBL design. Other MDBs have set a limit to the number of policy actions, usually 10 or fewer. The average for ADB PBL operations over the evaluation period was 23. The evaluation also found that many policy actions had become outputs in the DMF or were listed as indicators for outputs. However, the problem with presenting policy actions as outputs in the DMF or as indicators of the output is that the result of the policy action is the policy action itself, which greatly reduces the credibility of achieving it. This can be seen in Table 1.1, where each indicator for the output is worded the same way as the policy action that was undertaken prior to loan approval. The result of all these actions, moreover, is not adequately captured in the output, which simply states that policies and actions are in place. The output and outcome statements are also similar. Even where the indicator for the outcome is achieved, it is difficult to attribute this entirely to the policy actions or to understand what the connection is. DMF outputs are achieved simultaneously with policy actions, making PBL successful on approval. A review of 49 PVRs of PBL approved since 2008 found that 20 contained policy actions that were used as equivalent to outputs in the DMF.

 Table 1.1: Example of a Results Chain and Performance Indicators for Policy-Based Lending

Source: ADB reports and recommendations of the President to the Board of Directors

Role of technical assistance. PBL design and implementation was widely, although not universally, supported by grant-financed technical assistance, which undertook analysis and supported the implementation of policy actions prior to loan approval. So, while the loan is absorbed into the countries’ budget, TA is provided to the implementing agency to help implement policy actions. As some interviews suggested, provision of TA can provide an important incentive for line ministries to undertake reforms. In the Philippines, for example, TA was delivered as an integral part of PBL during 2008–2011, although more recent PBL implementation has tended to rely on existing or stand-alone TA. In the Kyrgyz Republic, a system support grant attached to the Investment Climate Improvement Program 62
in 2008 helped set up an electronic single window for pre-customs clearance of imports and exports, clearly demonstrating how TA can help improve PBL design and implementation. 

Slow delivery and underuse of TA undermined operational effectiveness in some cases. In Bangladesh, for example, the PCR for the Public Expenditure Support Facility and the CSF Program suggested that the failure to provide TA had undermined the achievement of some of the desired outputs.63
 The Viet Nam countercyclical support loan also did not draw on TA, which may help to explain the stop–go character of the government’s stimulus program. In several of the Viet Nam programs, including the Third Financial Sector Program in 2007 and the Small and Medium- Sized Enterprise Program in 2010, TA came late or was ineffective. 

Many of the available PCRs noted that a significant portion of TA funds—as much as 40% in some cases—remained undisbursed. In Pakistan, TA was provided in some form for all the PBL approved during 2008–2016 but was often marked by low use of designated funds. Where TA was provided to assess the impact of crisis support on households, for example in Armenia (2009), Georgia (2009), and Kazakhstan (2009), it was cancelled, which was a missed opportunity for learning. A comprehensive report on the overall impact of the Economic Recovery Support Program on vulnerable groups in Cook Islands was not undertaken, largely due to a lack of household survey data. Similar TA provided to support the government with state-owned enterprise reforms was cancelled in the Maldives (2009). Other PBL operations had to add more TA to help reform objectives to be achieved. In the Marshall Islands, for instance, while the TA was fully utilized, its outputs were not fully achieved. Tax revenue increased but the approval of tax legislation (prepared with TA consultant support) remained pending in Parliament, so the reforms could not be fully implemented. Similarly, debt management and implementation of state-owned enterprise (SOE) reforms were also delayed pending approval of the SOE and fiscal responsibility legislation. The envisaged TA outcome of fiscal sustainability was not achieved because of continuing fiscal deficits. These reform objectives therefore had to be completed by additional TA even once the initial budget support had been completed.

The part played by TA in facilitating the achievement of policy actions is often not fully presented in project completion reports. A key issue is the extent to which the policy reforms that were undertaken could have been implemented with TA alone, i.e., without budget support. In Bangladesh, for instance, there were suggestions that TA was effective in supporting the Third Capital Market Development Program and reforms would likely have occurred without the need for, or incentive of, budget support. In nearly all cases, budget support is rarely undertaken without TA, further illustrating that finance and policy reform have separate purposes and require separate modalities. Moreover, policy reform is unlikely to occur with budget support alone; it also requires TA for good analytical work, diagnostics, and reform implementation.

• 58
  
  An IMF assessment letter no more than 6 months old attached as a linked document to the approval documentation (RRP) is generally considered to be adequate.
• 59
  
  ADB. 2003. Report and Recommendation of the President to the Board of Directors: Proposed Program and Technical Assistance Loans to the Islamic Republic of Pakistan for the Sindh Devolved Social Sector Program. Manila.
• 60
  
  ADB. 2004. Report and Recommendation of the President to the Board of Directors: Proposed Program Loans and Technical Assistance Grant to the Islamic Republic of Pakistan for the Punjab Devolved Social Sector Program. Manila.
• 61
  
  ADB. 2005. Report and Recommendation of the President to the Board of Directors: Proposed Program Loans, Technical Assistance Loan, and Asian Development Fund Grant to the Islamic Republic of Pakistan for the Balochistan Devolved Social Sector Program. Manila.
• 62
  
  ADB. 2008. Report and Recommendation of the President to the Board of Directors: Proposed Program Cluster, Grant for Subprogram 1, and Grant Assistance to the Kyrgyz Republic for the Investment Climate Improvement Program. Manila.
• 63
  
  ADB. 2012. Completion Report: Public Expenditure Support Facility Program and Countercyclical Support Facility Program in Bangladesh. Manila.

IED evaluated the use of PBL by ADB over the period 2008–2017. The design and reform focus of ADB PBL fundamentally changed over this period and success rates—as judged by project completion reports validated by IED—more than doubled, a trend also experienced in other MDBs. Improved performance appears to have coincided with the growing use of single-tranche PBL and, with this, the use of prior actions that are completed before loan approval. These changes have substantially reduced disbursement risks and increased the capacity of MDBs to provide more predictable and reliable budget support in response to country financing needs, the primary objective of the instrument. A key issue, however, is whether the need to respond efficiently to country financing needs has encouraged support for less critical reforms. Certainly, over time, PBL reform topics appear to have shifted from more politically sensitive reforms such as reform of state-owned banks, to more technical reforms connected with public financial management. PBL modalities also changed as the second tranche of a single loan, often containing more difficult policy actions, had generally required waivers or had been cancelled. These were no longer part of PBL design. So there appears to have been a trade-off between efficient, rapidly disbursing modalities to meet country financing needs and policy reform, which suggests that the two objectives of finance and policy reform are not automatically compatible.

While PBL performance dramatically improved over the evaluation period, the evaluation identified several issues. For example, except for Pakistan, PBL tended to be used in the region’s more developed economies and was rarely focused on reform areas concerned with infrastructure development, ADB’s main comparative advantage. Moreover, it was difficult to reconcile the high success rates in project completion reports validated by IED with the evaluation’s finding that there were shortcomings in the quality of PBL design. A key reason for this is that, while PBL performance significantly improved over the evaluation period, assessing country-level results, such as improved service delivery, better governance, greater business confidence, and poverty reduction, and linking these with ADB PBL interventions was not as straightforward. If there is doubt about attribution (e.g., whether the PBL outcome was the direct result of the policy actions taken), the responsibility often falls on the evaluator to prove this (e.g., by constructing a counterfactual to show whether the result would have been achieved with or without the PBL). In practice, if the outcome indicators were achieved, the PBL will usually be rated successful, giving the role of PBL the benefit of the doubt. 

Where budget support is delivered concurrently by several development partners, unambiguous attribution of outcomes to a specific donor or PBL operation can be almost impossible to establish. An alternative is to assess the additionality of the operation or its contribution to generating outcomes. In other words, the focus of evaluation should be on the quality of the whole of ADB’s contribution, including TA, policy analytical work, and ongoing policy dialogue and other design matters. While attributing PBL’s contribution to broader development outcomes is difficult, it is clear that positive outcomes are more likely if there is strong PBL design, including good quality analytical work underpinning the PBL’s reform content, strong policy actions critical to intended outcomes, good quality technical assistance, and a clear monitoring and evaluation framework against which results can be assessed. 

PBL remains an efficient modality for supporting DMCs through crisis periods. This was demonstrated by ADB’s rapid response to the global economic and financial crisis in 2007–2009, and by the COVID-19 pandemic in 2020. ADB’s use of PBL has spiked during crisis years and broken through the 20% ceiling imposed on this type of lending in the sovereign loan portfolio. The increase in PBL use was only possible through the introduction of reform-free and rapidly disbursing budget support modalities to finance DMC countercyclical public expenditure programs to mitigate the extent of the crisis. PBL played several different roles over the evaluation period. It supported countries through difficult periods, including economic downturns, natural disasters, and pandemics, and it supported broad public sector management and macroeconomic stability through non-crisis years. Other budget support mechanisms are also emerging, including results-based lending, which is more relevant to improving service delivery than PBL. 

It remains difficult to attribute the economic growth and reductions in income poverty that has been observed in several countries in the Asia and Pacific region to budget support. This is partly due to a paucity of data, weak country statistical systems, problems establishing a causal chain, and difficulties establishing an appropriate timescale; the difficulty in attributing results to PBL does not mean budget support had no poverty impact or had less poverty impact than that of other forms of MDB support. More country-specific knowledge is needed on the links between budget support, policy reform (especially in public sector management), and growth and poverty reduction. ADB needs to strengthen its efforts to strengthen the quality of its design and monitoring frameworks and to better articulate the link between policy actions (the first rung on the results ladder) and development policy outcomes. A PBL operation needs to be founded on a strong results framework in which a small number of critical reforms are more clearly linked to desired outcomes to foster effective monitoring and evaluation. In the past, ADB may have been given the benefit of the doubt in success ratings where outcomes in DMFs are achieved, but the link to policy actions has not always been clear. 

The evaluation published in 2018 made several recommendations, some strategic and others related to PBL design. At the strategic level, it recommended that ADB should make greater use of PBL to support policy reforms in sectors where significant project investments were also undertaken. This would enable ADB to help countries achieve more integrated and sustainable solutions to public policy problems in these areas. Although this recommendation was accepted by Management, in practice it is unlikely to materialize mainly due to the COVID-19 pandemic, which in 2020 led to higher use of the non-reform-based countercyclical support facility. If this continues, the opportunity to use PBL to support infrastructure-related policy reform is likely to be limited in the immediate term. 

The evaluation’s recommendation that ADB should develop an operational plan on the scope, objectives, and articulation of public sector management interventions was not accepted formally, but in practice ADB has made moves in this direction. An operational priority plan for governance and institutional capacity has since been developed as part of ADB’s Strategy 2030. This plan provides corporate guidance on the conditions under which public financial management loans should be provided. 64

The evaluation recommendations that (i) concessional assistance-only countries (Group A) should have access to a countercyclical facility, and (ii) the use of contingent disaster financing be formalized were accepted. Since the outbreak of COVID-19, countercyclical support has been expanded to include Group A and non-OCR eligible countries (Group B) as part of ADB’s response to the COVID-19 pandemic.65
 The response includes using both Asian Development Fund (ADF) grant resources and ADB concessional loan resources as part of its COVID response. It is not clear whether this arrangement will continue after the current crisis period. Contingent disaster financing was formally approved by ADB soon after the evaluation was issued.66
 

ADB did not agree to the evaluation’s recommendation that, in rare cases where a regional department’s view on the macroeconomic situation of a country diverges from that of the IMF, the risks involved be assessed independently of the regional department. Nevertheless, ADB has since strengthened the capacity of the Strategy, Policy, and Partnerships Department (SPD) to oversee PBL design prior to Board approval. SPD has revised the PBL provisions of the Operations Manual and the relevant staff instructions, which now include a specific loan approval template and a design and monitoring framework better suited to PBL. ADB’s relationship with the IMF has been clarified and ADB’s capacity to produce a clear macroeconomic assessment has been strengthened, helping to support the overall quality assurance mechanism for PBL. ADB Management decided against the recommendation of a separate 3-year PBL operational review such as the one produced by the World Bank, which may have helped to ensure a greater focus on results.67
 

ADB has taken steps to strengthen PBL design. The evaluation recommended that ADB limit the use of process-oriented actions and articulate policy actions as substantive outputs. It recommended tailoring the DMF so that policy actions, outputs, and outcomes are more clearly linked, and that the analytical work underpinning PBL design and policy actions is clearly referenced. While these recommendations are part of the revised Operations Manual and staff instructions, the outbreak of COVID-19 and the need to respond quickly to DMC financing needs during the pandemic has meant that implementation of these changes has been deferred. 

In due course, ADB needs to strengthen its assessment of PBL design at program completion. This assessment should cover: the justification for the use of PBL, the relevance of the policy reforms supported, and their significance to the development outcome. There should also be a greater focus on the role and quality of TA, given its central role in the preparation of a lot of PBL and its implementation. Just as PBL requires its own template and DMF, new approaches for assessing PBL performance need to be introduced to ensure that the success rating given to completed PBL is based on a robust evidence-based assessment of the design, especially with regard to the relevance and criticality of policy actions to development outcomes. In single-tranche PBL, policy actions have already been carried out at the time of the approval of the loan by the ADB Board, but their relevance and criticality to the outcome should still be assessed at completion. 

This chapter regards ADB management’s response to IED’s recommendations as generally positive. Several initiatives to strengthen PBL design are underway. SPD, which oversees PBL quality and design, has strengthened its procedures. A separate PBL loan template, and a new results framework, linking policy actions with their intended results, have been developed. The COVID-19 outbreak in early 2020 led to a surge in countercyclical budget support in the first half of that year and the introduction of a new type of PBL to enable ADB to respond quickly to those countries most severely affected by the pandemic. A revised project completion template which more closely examines design issues is yet to be developed.

• 64
  
  ADB. 2019. Strategy 2030. Operational Plan for Priority 6. Strengthening Governance and Institutional Capacity 2019-2014. Manila.
• 65
  
  ADB. 2020. ADB’s Comprehensive Response to the COVID-19 Pandemic. Manila.
• 66
  
  ADB. 2019. Contingent Disaster Financing under Policy-Based Lending in Response to Natural Hazards. Manila.
• 67
  
  World Bank Group. 2015. Development Policy Financing Retrospective. Results and Sustainability. Washington DC.