Some countries have used PBL consistently and extensively since its introduction and have grown to rely on it as a source of budget financing even in non-crisis years. Consultations with country officials have suggested that demand for PBL is expanding, and in some countries it is the preferred lending modality. Indonesia, Pakistan, the Philippines, and Viet Nam have consistently accounted for nearly 59% of all PBL approved by value since 1978. These countries continue to be the largest, most frequent, and most consistent PBL users. Apart from Pakistan, where there is a strong poverty argument to justify the use of PBL, the evaluation found that PBL is mostly used by larger, more capable, and more developed countries. In terms of the number of PBL operations, Pakistan has been the largest PBL recipient, followed by Indonesia. In 2008–2017, PBL accounted for 71% of ADB’s sovereign lending portfolio in Indonesia and 57% in the Philippines. In terms of ADB regions, Southeast Asia has been the largest and most consistent user of PBL since 1978, accounting for nearly 40% of approvals by number each year and almost 50% by value. Central and West Asia accounted for 25% of total approvals, and South Asia for only 16% by value, despite its large reform agenda. The use of PBL in the Pacific region appears to be linked to crisis years: the Asian financial crisis (1997), the stock market crisis in the US triggered by the bursting of the dot-com bubble (2001), and the global financial crisis (2007–2009). Recently, ADB has used PBL to provide contingent financing operations in Cook Islands, Samoa, Tonga, and Tuvalu which have been used to build disaster resilience during non-crisis times and to release funds immediately following a natural disaster.16
 

The evaluation found that PBL is positively correlated with the size of fiscal deficits, and negatively correlated with GDP growth, suggesting that finance is the primary objective of countries requesting PBL, rather than policy reform,17
even if the key motivation of ADB is policy reform. Frequent PBL recipients, including Indonesia and the Philippines, base their requests for PBL on their budget financing needs (this was confirmed during consultations with ministry of finance officials in both countries). In the Philippines, for example, the Treasury divides financing of the budget deficit between domestic currency (80%) and foreign currency (20%) financing. Foreign currency financing includes financing by MDBs, so the larger the deficit the higher the demand for PBL. In general, ADB PBL approvals have increased after countries experienced large fiscal deficits. Policy reform is an important motivating factor as well (certainly for ADB) but the size of PBL operations is not necessarily determined by the size or cost of a government’s reform agenda and nor are such operations systematically triggered by the desire for reform. In the absence of a fiscal deficit, it would be difficult to justify PBL use, and policy reform would be better supported using other financing modalities.18
 

Despite growth in PBL use in crisis and non-crisis periods, the size of PBL in relation to gross domestic product (GDP) and as a share of budget financing has decreased since the 1980s. Asia has experienced rapid growth and poverty reduction over the last three decades and the significance of development assistance relative to GDP has fallen across the region, although less so in the Pacific. As a result, the average size of a PBL operation relative to a country’s GDP has declined more than threefold since the late 1980s, reducing the potential effectiveness of PBL as a countercyclical device during a crisis, although some of the value of PBL as a market confidence building device may have remained. The capacity of PBL in high-growth economies to act as an incentive for policy reform may therefore also have diminished. Moreover, finance (the provision of liquidity) alone cannot achieve policy reform.19
 Good policy making must be well-informed, supported by evidence-based analytical work, diagnostics, and continuous dialogue with stakeholders to build traction for reform. Hence, PBL should also come with technical assistance (TA) if policy reform is to be achieved, and ADB has learned to increasingly deploy TA alongside its PBL, on grant basis.20

• 16
  
  PBL has been used to clear debt arrears, for example, in the re-engagement of development partner support for Myanmar see. ADB. 2012. Proposed Policy-Based Loan Republic of the Union of Myanmar: Support for Myanmar Reforms for Inclusive Growth Program. Manila.
• 17
  
  Two panel fixed-effects models were estimated by the evaluation. Model 1 PBL_GDPit = ci + β1Res_Impit-1 + εit and Model 2 PBL_GDPit = ci + β1FiscalDeficitit + εit where i indicates country and t indicates year. The dependent variable, PBL_GDPit is policy-based lending (PBL) as a percentage of GDP.Res_Imp indicates total reserves in months of imports. ci is the country-specific effect. The sample period is 1990–2016. The regression analysis shows that PBL is negatively correlated with total reserves in months of imports and positively correlated with fiscal deficits. The results suggest that the use and the amount of PBL is determined by countries’ financing needs at the macroeconomic level.
• 18
  
  There is a view that higher demand for PBL in these countries also coincided with sharpened procurement and safeguard conditions attached to ADB investment loans, particularly in the first decade of the 2000s, and that some countries had negative experiences with the introduction and grade of these conditions, and therefore lowered their demand for infrastructure investment lending. Thus their demand for policy-based lending increased by default. However, the evaluation did not assess this view fully
• 19
  
  The Tinbergen Rule states that achieving multiple targets (or objectives) requires an equal number of instruments. Hence, the two objectives of providing finance and supporting policy reform requires two modalities: finance and technical assistance.
• 20
  
  For instance, from 2008 to 2017, 45% of all PBL had at least one TA project, 26% had two, 12% had between three and seven, but 17% had no TA.