Estimating structural budget balances in developing Asia

Jalles, João Tovar Working Paper Estimating structural budget balances in developing Asia ADB Economics Working Paper Series, No. 719 Provided in Cooperation with: Asian Development Bank (ADB), Manila Suggested Citation: Jalles, João Tovar (2024) : Estimating structural budget balances in developing Asia, ADB Economics Working Paper Series, No. 719, Asian Development Bank (ADB), Manila, https://doi.org/10.22617/WPS240087-2 This Version is available at: https://hdl.handle.net/10419/298165 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/3.0/igo/ ASIAN DEVELOPMENT BANK ASIAN DEVELOPMENT BANK 6 ADB Avenue, Mandaluyong City 1550 Metro Manila, Philippines www.adb.org ESTIMATING STRUCTURAL BUDGET BALANCES IN DEVELOPING ASIA João Tovar Jalles ADB ECONOMICS WORKING PAPER SERIES NO. 719 April 2024 Estimating Structural Budget Balances in Developing Asia This paper reviews the current discussions, methods and practices surrounding the estimation of reasonable proxies for the underlying fiscal position, a useful anchor for fiscal policy. An empirical application to developing Asian economies is carried out. There is no one-size fits all type of approach and the sensitivity and discernment regarding specific economies are important in various stages. About the Asian Development Bank ADB is committed to achieving a prosperous, inclusive, resilient, and sustainable Asia and the Pacific, while sustaining its efforts to eradicate extreme poverty. Established in 1966, it is owned by 68 members —49 from the region. Its main instruments for helping its developing member countries are policy dialogue, loans, equity investments, guarantees, grants, and technical assistance. ASIAN DEVELOPMENT BANK The ADB Economics Working Paper Series presents research in progress to elicit comments and encourage debate on development issues in Asia and the Pacific. The views expressed are those of the authors and do not necessarily reflect the views and policies of ADB or its Board of Governors or the governments they represent. ADB Economics Working Paper Series Estimating Structural Budget Balances in Developing Asia João Tovar Jalles No. 719 | April 2024 João Tovar Jalles ([email protected]) is a senior associate professor of economics at the Lisbon School of Economics and Management (ISEG), University of Lisbon. Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO) © 2024 Asian Development Bank 6 ADB Avenue, Mandaluyong City, 1550 Metro Manila, Philippines Tel +63 2 8632 4444; Fax +63 2 8636 2444 www.adb.org Some rights reserved. Published in 2024. ISSN 2313-6537 (print), 2313-6545 (electronic) Publication Stock No. WPS240087-2 DOI: http://dx.doi.org/10.22617/WPS240087-2 The views expressed in this publication are those of the authors and do not necessarily reflect the views and policies ofthe Asian Development Bank (ADB) or its Board of Governors or the governments they represent. ADB does not guarantee the accuracy of the data included in this publication and accepts no responsibility for any consequence of their use. The mention of specific companies or products of manufacturers does not imply that they are endorsed or recommended by ADB in preference to others of a similar nature that are not mentioned. By making any designation of or reference to a particular territory or geographic area, or by using the term “country” inthis publication, ADB does not intend to make any judgments as to the legal or other status of any territory or area. This publication is available under the Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO) https://creativecommons.org/licenses/by/3.0/igo/. By using the content of this publication, you agree to be bound bytheterms of this license. For attribution, translations, adaptations, and permissions, please read the provisions andterms of use at https://www.adb.org/terms-use#openaccess. This CC license does not apply to non-ADB copyright materials in this publication. If the material is attributed toanother source, please contact the copyright owner or publisher of that source for permission to reproduce it. ADB cannot be held liable for any claims that arise as a result of your use of the material. Please contact [email protected] if you have questions or comments with respect to content, or if you wish toobtain copyright permission for your intended use that does not fall within these terms, or for permission to use theADB logo. Corrigenda to ADB publications may be found at http://www.adb.org/publications/corrigenda. ABSTRACT This paper reviews the current discussions, methods, and practices surrounding the estimation of reasonable proxies for the underlying fiscal position, a useful anchor for fiscal policy. An empirical application to developing Asian economies is carried out. There is no one-size fits all type of approach and the sensitivity and discernment regarding specific economies are important in various stages. The choice of the filter to decompose trends from cycles matters. The way to adjust revenues and expenditures entering the cyclically adjusted balance also matters—choices regarding economy-specific vs. panel estimations or the use of static vs. time-varying approaches need to be made. To deal with one-off operations, a narrative-based approach can complement the suggested identification based on large changes in cyclically adjusted government capital transfers. A discussion of other important factors that can affect the estimates of structural balances such as asset and commodity prices is also provided. Keywords: budget elasticity, time-varying estimation, trend-cycle decomposition, cyclically adjusted balances, one-off fiscal operations JEL codes: C33, E62, H30, H60, O53 1. Introduction For fiscal surveillance and monitoring, it is paramount to assess the state of public finances and fiscal policies, that is, to understand the nature of budgetary developments. Actual budget balances are inadequate measures for these purposes as they are affected by multiple factors that are both temporary and outside the direct control of the government. Prime amongst these factors are fluctuations in economic activity. The understanding that economic fluctuations, which are transitory in nature, have on the interpretation and evaluation of fiscal developments supported the notion that nominal budgets could not be taken at face value since they combine temporary and permanent disturbances (Larch and Turrini 2009). Disentangling the two elements led to the creation of the concept of the cyclically adjusted budget balance (CAB), which came to live more than 30 years ago in a seminal contribution by Blanchard et al. (1990).1 The CAB seeks to correct fiscal outcomes for the influence of cyclical movements. In other words, they aim at determining what fiscal position would prevail if the economy operated at its full capacity (ECB 2014). In addition to the impact of the economic cycle, other aspects that have become progressively important are the one-off measures. These are temporary government measures such as one-off revenues (for instance, stemming from the sale of licenses for telecommunication) or one-off capital transfers (for instance, financial aid to the banking sector). 1 In fact, Brown (1956) pioneered the extraction of the underlying budget balance by computing the full employment surplus, the predecessor of the CAB. The conceptual advantage of potential GDP of potential output over full employment did not facilitate things in practical terms. The computation of potential output still remains a speculative issue despite many methods and approaches having been developed (Section 3.1). 2 Stripping the impact of short-term fluctuations and one-offs yield what some international organizations refer to as “structural budget balance” (SB) (International Monetary Fund [IMF]) or the “underlying fiscal position” (European Commission [EC]).2 The attraction of the SB sits with its goal to measure, at low cost, the “true” fiscal position corrected of transitory cyclical factors. However, estimates of the structural balance can be subjected to significant measurements errors as they are surrounded by considerable uncertainty. These errors are related to accurately estimating the output gap—which is an unobservable variable— and the nonlinear reactions of tax revenues to sharp changes in real gross domestic product (GDP) growth. The definition or identification of temporary factors can also cause difficulties. These problems complicate the applicability of this concept in economies that are regularly affected by severe shocks (e.g., weather events); those that have a production structure that is not sufficiently diversified and, hence, render the economy hostage of commodity cycles; or those whose growth outlook is characterized by a lack of some anchored predictability. This report aims to review the current discussions surrounding the estimation of reasonable proxies of discretionary fiscal stance by reviewing methods and practices. An empirical application to a selection of developing member economies of the Asian Development Bank (ADB) is carried out but full results are available in the Appendixes.3 We can conclude from the empirical exercises that there is no one-size fits all and desk sensibility and judgment is important in various degrees. The choice of the filter to decompose trends and cycles matters with a preference to avoid 2 For the remainder of the document, I will be referring to structural balance as the measure stripped from both cyclical variations and one-off measures. 3 The appendixes are available at https://dx.doi.org/10.22617/WPS240087-2. 3 the outdated and problematic HP filter in favor of the Hamilton (2018) approach. Moreover, the take on how to adjust revenues and expenditure entering the CB or SB is paramount. There is a plethora of choices to be made, from prior empirical estimation at the economy-specific level, to whether such analysis should be done in a static or timevarying fashion, to which subcomponents of revenues and expenditures are likely to fluctuate with the cycle or be orthogonal to it. Furthermore, if one thinks about one-off operations, then a more narrative-based approach can complement the ad hoc identification of large changes in cyclically adjusted government capital transfers. Finally, there is the discussion of other factors in addition to the cycle, such as asset and commodity prices, which includes changes in housing prices and terms of trade adjustments that can prove to be difficult to be reflected in a measure of the underlying fiscal position. Ultimately, SBs are a useful anchor for fiscal policy, but should not be taken too seriously for real-time policy making. The remainder of the report is organized as follows. Section 2 elaborates on the rationale behind needing a measure of cyclically adjusted budget balances; Section 3 presents the existing methodologies to do so contrasting pros and cons of each one, the so-called aggregated vs. disaggregated approaches; Section 4 deals with one-off fiscal operations and the issues concerning the passage from cyclically adjusted balances to a measure of underlying balance or structural balance; Section 5 performs several empirical applications on ADB member economies; and the last section concludes. 4 2. Why Do We Need a Measure of the Cyclically Adjusted Budget Balance or Structural Budget Balance? Despite the flaws and problems, the CAB (and/or the SB) continues to be one of the main indicators necessary for fiscal policy analysis. More specifically, it is an important concept because it allows: (i) splitting the influence of discretionary fiscal policy to a given change in the deficit from the effect of the economic environment while simultaneously allowing automatic stabilizers to operate freely; (ii) evaluating fiscal impulses; and (iii) assessing if fiscal policy in the long-run is sustainable. On (i), the discussion in the literature surrounds the role of fiscal policy in macroeconomic stabilization. Calculating the structural balance means splitting the fiscal stance into the cyclical and the structural components. More generally, fiscal policy can help the stabilization of the business cycle by means of three different channels (Silgoner, Reitschuler, and Crespo-Cuaresma 2003). The first refers to the role of the cyclical component, that is, the automatic stabilizers. There is, nonetheless, a second indirect channel related to public spending items that continue fixed, independent of the phase of the business cycle and by not reacting to it also have a stabilizing function. The third, is the use of discretionary fiscal policy (despite most literature pointing to a pro-cyclical bias of these policies and, hence, a destabilizing effect (van den Noord 2000). Changes in the structural balance need policy actions and translate discretionary fiscal policy changes.4 While the Global Financial Crisis (and, more recently, the coronavirus disease [COVID4 To Dos Reis et al. (2007) it is preferable to refer to automatic stabilizers as “a passive policy response to the cycle.” 11 Table 1 shows a sample of OECD economies’ aggregate budgetary elasticities parameters retrieved from Larch and Turrini (2009). Table 1: Budgetary Sensitivity Parameters Economy Revenue Expenditure Budget Balance Economy Revenue Expenditure Budget Balance Belgium 0.47 -0.07 0.54 Hungary 0.45 -0.01 0.46 Bulgaria 0.35 -0.01 0.36 Malta 0.35 -0.01 0.36 Czech Republic 0.36 -0.01 0.37 The Netherlands 0.39 -0.17 0.55 Denmark 0.50 -0.15 0.65 Austria 0.43 -0.04 0.47 Germany 0.40 -0.11 0.51 Poland 0.33 -0.06 0.40 Estonia 0.29 -0.01 0.30 Portugal 0.41 -0.04 0.45 Greece 0.42 -0.01 0.43 Romania 0.28 -0.02 0.30 Spain 0.38 -0.05 0.43 Slovenia 0.42 -0.05 0.47 France 0.44 -0.06 0.49 Slovak Republic 0.27 -0.02 0.29 Ireland 0.36 -0.05 0.40 Finland 0.41 -0.09 0.50 Italy 0.49 -0.02 0.50 Sweden 0.48 -0.10 0.58 Cyprus 0.39 -0.01 0.39 United Kingdom 0.40 -0.02 0.42 Latvia 0.26 -0.02 0.28 Lithuania 0.26 -0.01 0.27 Euro area 0.42 -0.06 0.48 Luxembourg 0.48 -0.01 0.49 EU27 0.39 -0.04 0.43 Source: Larch and Turrini (2009). The disaggregated approach finetunes the overall previous sequence of steps by further decomposing the two main budgetary items of revenues and expenditures and estimating individual revenue elasticities. When doing so, one needs to take a stance on revenue or expenditure items that may not require cyclical adjustment. This means that only a subset of revenue components is adjusted, those typically being personal income taxes, corporate income taxes, indirect taxes or value-added taxes and social security contributions. For expenditure, some rely on cyclical adjustment of only current primary expenditures, therefore, excluding interest payments and capital expenditure).9 Others simply focus on one item of current primary expenditures, the unemployment benefits. 9 Interest expenses may display cyclicality in tandem with fiscal deficits, resulting in increased (decreased) borrowing needs when there is a positive (negative) output gap. This, in turn, can lead to fluctuations in the interest costs incurred. Countercyclical shifts in interest rates have the potential to offset the cyclical nature of borrowing requirements, potentially resulting in a minimal net impact, if any, as suggested by Farrington et al. (2008). 12 However, they normally account for a small proportion of total or primary spending in most economies (if at all). Mathematically, this version of the CAB can be expressed as: 𝐶𝐴𝐵 = ∑𝑅    −𝐺  +𝑅 −𝐺 (8) where 𝑅  is the cyclically adjusted component i of revenues, 𝐺  is the cyclically adjusted unemployment-related expenditures, 𝑅 and 𝐺 denote the non-cyclically adjusted revenue and expenditure components respectively. For cyclically adjusted revenues, the new 𝜌, can be obtained as: 𝜌, =∑𝜌,     (9) where 𝜌, is the elasticity of each revenue component with respect to the output gap and   is the share of each revenue component i in total revenues. For cyclically adjusted expenditures, the new 𝜌, can be obtained as: 𝜌, = 𝜌,   (10) where 𝜌, is the elasticity of unemployment benefits with respect to the output gap and   is the share of unemployment benefits in current primary expenditures. 3.2 Potential Output In addition to being a measure of aggregate supply in the economy, potential output is equally an estimate of trend GDP. GDP’s long-term trend is normally upward as more resources—primarily labor and capital—become available and as technological change allows more-efficient use of these resources (Arnold 2009). Alternative measures of potential GDP were originally conceived to influence decisions about monetary and 13 fiscal policy. If an economy was found to be below potential then monetary or fiscal policy could be used to accelerate the growth of output without risking too much inflationary pressure. The concept of potential output is perceived as a tool to assist policymakers manage aggregate demand and hence keep a steady economic growth path. Different definitions are often used by alternative international organizations (Box 1). Box 1: Alternative Definitions of Potential Output European Commission. “Potential output constitutes the best composite indicator of the aggregate supply side capacity of an economy and of its scope for sustainable, non-inflationary, growth.... Potential growth constitutes a summary indicator of the economy’s capacity to generate sustainable, non-inflationary, growth.”a OECD. “Potential gross domestic product is defined in the OECD’s Economic Outlook publication as the level of output that an economy can produce at a constant inflation rate. Although an economy can temporarily produce more than its potential level of output, that comes at the cost of rising inflation. Potential output depends on the capital stock, the potential labour force (which depends on demographic factors and on participation rates), the non-accelerating inflation rate of unemployment (NAIRU), and the level of labour efficiency.”b ECB. “Potential output is a key economic concept as its evolution determines how fast an economy can grow in a sustainable way. It is typically thought of as the highest level of economic activity that can be sustained by means of the available technology and factors of production, in particular labour and capital, without creating inflationary pressure.”c ECB. “Potential output is generally understood to provide an indication of the medium-to longterm level of sustainable real output in the economy and its rate of growth. It is also referred to as the level of output which can be achieved using available production factors without creating inflationary pressures.”d IMF. “Potential output is the maximum amount of goods and services an economy can turn out when it is most efficient—that is, at full capacity. Often, potential output is referred to as the production capacity of the economy.”e ECB = European Central Bank, IMF = International Monetary Fund, OECD = Organisation for Economic Co-operation and Development. a European Commission. 2014. “The Production Function Methodology for Calculating Potential Growth Rates & Output Gaps.” Economic Papers 535. November. b OECD Stat. https://stats.oecd.org/ (accessed 2022). c Andersson, Malin, Bela Szörfi, Máté Tóth, and Nico Zorell. 2018. “Potential Output in the Post-Crisis Period.” ECB Economic Bulletin Issue 7/2018. e Jahan, Sarwat, and Ahmed Saber Mahmud. 2013. “What is the Output Gap?” Finance & Development 50 (3). IMF. Source: European Parliament. 2020. “Potential Output Estimates and Their Role in the EU Fiscal Policy Surveillance.” 14 There are multiple approaches to obtain estimates of trend GDP. In spite of substantial progress in the econometrics over the years, there is still not a generally accepted method in the economics profession to calculate potential output (Borio et al. 2013, 2014).10 To summarize, various approaches for estimating the output gap exist, with some based on statistical techniques and others on economic theory-informed models. Each method has its own set of strengths and weaknesses, as outlined in CBO (2004). However, there is a consensus that estimating the output gap in real-time poses significant challenges, as noted by Orphanides and Van Norden (2002). In the following discussion, we will undertake a historical analysis that looks backward in time, primarily focusing on univariate statistical methods. It’s important to note that statistical methods are not entirely devoid of theoretical considerations, as highlighted by Bassanetti et al. (2010) and Oksanen (2018). Univariate statistical methods, in particular, rely on the fundamental theoretical assumption that actual output tends to fluctuate around its potential level. In other words, these methods are designed to ensure that actual output, by construction, only temporarily deviates from potential output. Taking into account the criticisms directed at the widely used Hodrick-Prescott (HP) filter, such as its tendency to identify false cycles (Harvey and Jaeger 1993, Cogley and Nason 1995), a more recent filtering technique introduced by Hamilton (2018) is employed. Hamilton’s critique of the Hodrick and Prescott filter revolves around three issues, namely spurious cycles, end-of-sample bias, and arbitrary assumptions concerning the smoothing parameter. As an alternative, Hamilton proposed a regression 10 For a review of different methods see Gibbs (1995), Giorno et al. (1995), Ladiray et al. (2003), Horn et al. (2007), Bassanetti et al. (2010), Anderton et al. (2014), and Alichi et al. (2017). 15 filter. However, it is worth noting that Schüler (2018) demonstrates that Hamilton’s regression filter shares some of the shortcomings of the HP filter, indicating that it is not a universal solution. Schüler argues that the choice of the “correct” filter depends on the researcher’s specific objectives, that is, what aspects of the data they aim to emphasize. To date, only a limited number of studies in addition to Schüler (2018) have empirically evaluated Hamilton’s methodology.11 Hodrick (2020) subsequently examined if Hamilton’s (2018) alternative approach was actually better than the HP filter at extracting the cyclical component. Hodrick (2020) found that for time series in which there are distinct growth and cyclical components, the HP filter is better at isolating the cyclical component. 3.3 Budgetary Elasticities To make it simple, many rely on Girouard and Andre’s (2005) assumptions of zero expenditure elasticity and one revenue elasticity. In fact, for this group of economies elasticities computed for specific tax categories yield an aggregate revenue elasticity close to 1 (European Commission 2005). Regarding the zero-elasticity assumption on expenditure (meaning that the cyclically adjusted expenditure is equal to actual expenditure, 𝐺 = 𝑔), this is justified economically because government expenditure is often viewed as discretionary in its entirety, and thus independent from the business cycle. Fedelino et al. (2009), for instance, assumed no adjustment on the expenditure side. While this may be a reasonably good approximation, in practice, some expenditure 11 In the context of New Zealand, Callaghan et al. (2018) employed both the HP filter and Hamilton's (2018) method to calculate labor force participation gaps. Drehman and Yetman (2018) evaluated the effectiveness of a HP trend versus Hamilton's linear projection of the credit-to-GDP gap as an early warning indicator for financial crises. Phillips and Shi (2019) introduced an enhanced HP filter and conducted an assessment of its performance in comparison to Hamilton's approach. 16 items, such as unemployment benefits, exhibit a cyclical pattern. This method does not separate between the various revenue and expenditure items. However, the loss of accuracy may be acceptable in some cases according to these authors. As Sancak et al. (2010) demonstrate, in some economies, tax buoyancy tends to increase during expansions and go down during recessions. Consequently, in these economies the contribution form automatic stabilizers may be overstated if one/zero elasticities are used to estimate automatic stabilizers. Against this background, economy-specific elasticities for overall revenue and expenditure should be used whenever possible, either by relying on pre-existing studies or by estimating them in a regression setting. What Does This Entail?12 Concerning revenue, the elasticity of individual revenue categories can be broken down into two components. The first is the output elasticity of tax revenue (𝜌,), which results from multiplying the elasticity of tax revenues concerning the corresponding tax base (𝜌,),, by the elasticity of the tax base in relation to the output gap (𝜌,): 𝜌, = 𝜌,×𝜌, (11) The first step is to assume or derive the value of the tax elasticity with respect to its base. Derivation requires, in addition to statutory tax rates, knowledge of the income distribution. But in practice, one might rely on results of existing studies. The second step involves an econometric process where we estimate how sensitive the relevant tax bases are to changes in the output gap. This estimation 12 This section draws heavily from Bornhorst, Dobrescu, Fedelino, Gottschalk, and Nakata (2011). 17 necessitates the definition of macroeconomic indicators that act as proxies for these tax bases. For instance, wage bills are often used as a proxy for income taxes and social security contributions, corporate profits are employed as a measure of the tax base for corporate income taxes, and private consumption is the basis for indirect taxes. Once these two elasticities are determined, we can then compute the tax revenue elasticities in response to changes in the output gap. Girouard and Andre (2005) present a table (see Table 2 below) with typical tax elasticities. The elasticities of revenue components concerning changes in the output gap are usually greater than one for income taxes, which is indicative of the progressive nature of many income tax systems. They tend to hover around one for indirect taxes, reflecting the typically uniform indirect (VAT) tax rates. In the case of social security contributions, these elasticities are somewhat lower than one. Table 2: Common Tax Elasticities Tax Category Elasticity of Tax Revenue Relative to its Base Elasticity of Base Relative to Output Gap Elasticity of Tax Revenue Relative to Output Gap Personal income tax 1.5–2.0 0.6–0.9 1.0–1.7 Corporate income tax 1.0 1.2–1.8 1.2–1.8 Social security contributions 0.8–1.1 0.6–0.9 0.5–0.9 Indirect taxes 1 1 1 Source: Girouard and Andre (2005). Regarding expenditures, it’s possible to break down the elasticities of current expenditure categories into two components. Current spending, especially for items like unemployment benefits, is more likely to exhibit cyclical behavior due to the nature of the benefit system. In contrast, nominal spending on other items such as public sector wages 18 and consumption of goods and services is likely to be largely unaffected by the business cycle, requiring no adjustment (Bornhorst et al. 2011). The output elasticity of expenditures (𝜌,) is then determined by multiplying the elasticity of current expenditures (𝜌, ) concerning its base, which could be unemployment, by the elasticity of that base concerning changes in the output gap (𝜌,): 𝜌, = 𝜌, ×𝜌, (11) Similar to revenues, the elasticities of expenditure concerning their respective bases can either be assumed or derived. In equation (11), if we are considering a specific expenditure category, such as unemployment benefits alone, it may be reasonable to assume a proportional relationship with its base, which in this case is unemployment, resulting in an elasticity of 1. Consequently, the output elasticity of that particular expenditure category is determined by the elasticity of unemployment concerning changes in the output gap. In previous research, due to a lack of additional data, the Okun’s method has been employed for this purpose. The elasticity of the unemployment rate in relation to changes in output is essentially the reciprocal of the Okun coefficient. Giorno et al. (1995) and Bouthevillain et al. (2001) adjusted expenditure to account for cyclical effects using an average elasticity of -0.2. 19 Box 2: Estimating Budgetary Elasticities and the Okun’s Coefficient All these budgetary-output elasticities can be estimated in a regression setting. Elasticities represent the percentage shift in one variable, denoted as X, in response to a one-percentage-point alteration in another variable, Y. A commonly employed method for calculating the elasticity of a time series, X, in relation to a proxy for economic activity, denoted as Y, involves estimating the following equation: ∆log (𝑋) = 𝑎 +𝜌, ×∆log(𝑌)+𝜀 where ∆ is a first difference operator. More specifically, the left-hand-side dependent variable should be expressed in real terms (that is, revenues or expenditures deflated by the GDP deflator) and the right-hand-side key regressor should be real GDP growth (that is, the first difference of the log of real GDP). The Okun’s Law assumes a negative relationship between cyclical fluctuations in GDP and the unemployment rate. Shocks to the economy are assumed to lead GDP to fluctuate around its potential; this, in turn, causes firms to hire and fire workers, changing the unemployment rate in the opposite direction. The estimation of the Okun coefficient is needed in the disaggregated approach for the expenditure side elasticity. Ball et al. (2019) suggest estimating a “changes” version of the original “cyclical” Okun’s law, as a relationship between changes in the unemployment rate (u) and the growth rate of output: ∆𝑢= 𝛼 +𝛽∆𝑌+𝜀 Our coefficient of interest is 𝛽 which gives how much a one percentage point increase in output growth translates into a reduction in the unemployment rate. Furceri et al. (2020) estimate an average Okun coefficient in advanced economies of -0.4 while for developing economies it is equal to -0.2. For the list of Asian Development Bank members, Appendix 2 shows the estimated economy-specific elasticities for those pairs of growth and unemployment rate with at least 15 continuous observations. Coefficients are estimated by ordinary least squares (OLS). For those that came out statistically significant, they vary between -0.02 in Azerbaijan and -0.19 in Fiji with an average of -0.12. GDP = gross domestic product. Source: Jalles (2019) and Furceri et al. (2020). 20 4. The Role of Asset and Commodity Cycles Standard cyclical adjustments can be accompanied with an adjustment—when justified— for significant changes in asset or commodity prices (Bornhorst et al. 2011):  Commodity prices or terms of trade adjustment. In cases where government budgets depend on income generated by commodity exports or are significantly influenced by fluctuations in the terms of trade, it may be beneficial to make adjustments for abrupt shifts in these prices in order to reveal the true fiscal situation beneath.  Asset price adjustment. The fiscal situation at its core can also be influenced by asset prices, particularly those of real estate and stocks. As demonstrated by Farrington et al. (2008), a sustained 10% rise in asset and housing prices resulted in an annual increase in cyclically adjusted tax revenues ranging from 0.1% to 0.4% of GDP. How to address empirically the above issues? As with the cyclical adjustment, both aggregate and disaggregate approaches can be used. In what follows, the former will be discussed and employed. Basically, the approach consists in adding a separate term for the deviation of asset prices or terms of trade from their benchmark level, denoted as asset price gap (A∗/𝐴) or terms of trade gap (T∗/𝑇): 𝑅, = 𝑟󰇡∗ 󰇢, 󰇡∗ 󰇢, (12a) 𝑅,, = 𝑟󰇡∗ 󰇢, 󰇡∗ 󰇢, 󰇡∗ 󰇢, (12b) where 𝑅, and 𝑅,, stand for revenues adjusted for the output and asset price gaps and the output, asset prices, and terms of trade gaps, respectively. If the elasticities of revenues with respect to asset prices or terms of trade is zero𝜌, = 0 or 𝜌, = 0, the 27 play a crucial role in identifying one-offs. More technically, for each economy, there is a level or a trend of net capital transfers considered as “normal” such that large departures of net capital transfers from this benchmark provides a good proxy for one-offs. Summing up, from an operation point of view, calculating structural balances involves a set of interrelated steps (Figure 1).  Step 1: Identifying and removing one-offs.  Step 2: Evaluating the effect of the business cycle by relying on the aggregated or disaggregated approach.  Step 3: Estimating the effects of other cycles or factors, including those associated with asset prices or commodity prices. One can still obtain the CAB by omitting step 1. 28 Figure 1: Steps for Obtaining Structural Budget Balances Source: Bornhorst et al. (2011). 6. Empirical Application: ADB Membership 6.1 Potential Output To extract the cyclical and trend components for a nonspecific variable 𝑥 (denoted 𝑥 and 𝑥, respectively) where in our case 𝑥= {𝐺𝐷𝑃}, we begin by employing the frequently used Hodrick-Prescott filter. This filter minimizes the following function: 𝑚𝑖𝑛 {∑ (𝑥−𝑥)+𝜆∑[(𝑥−𝑥)−(𝑥   −𝑥)]   } (13) where 𝜆 is the smoothing parameter. The larger the value of 𝜆, the bigger is the penalty on variations of the trend’s growth rate. Hodrick and Prescott suggest 100 or 1,600 as a Cross-economy comparisons? 29 value for 𝜆 for annual or quarterly data, respectively. Ravn and Uhlig (2002) argue that 𝜆 should equal 6.25 (, ) for annual data and 129,600 (1,600 × 3) for monthly data. For Hamilton (2018), we estimate: 𝑥 = 𝛾+∑𝛾+𝑥 +𝑢   (14) where 𝑥= 𝑥+𝑥. The non-stationary part of the regression provides the cyclical component: 𝑥= 𝑢 (15) while the trend is given by 𝑥= 𝛾+∑𝛾+𝑥   (16) Hamilton (2018) claimed that h and k should be chosen such that the residuals obtained from equation (15) are stationary and suggested that, for broader processes, the fourth differences of a series are stationary. We chose h = 2 and k = 3. Hamilton’s (2018) better alternative is, in essence, to use the regression of a variable at date t on the four most recent values as of date t-h since this would achieve “… all the objectives sought by users of the HP filter with none of its drawbacks.” (p. 831). In particular, this alternative “… can isolate a stationary component from any I(4) series, preserves the underlying dynamic relations, and consistently estimates well-defined population characteristics for a broad class of possible data-generating processes.” (pp. 839–840). Equations (14), (15), and (16) are estimated using ordinary least squares (OLS). Figure 2 shows for a selected set of ADB member economies the cyclical decomposition of real GDP, that is, the output gap expressed—as common in the 30 literature—in percent of potential or trend GDP.15 Evidence seems to suggest that the HP is smoother than the Hamilton-resulting gaps, that is, booms and busts are less pronounced. In the case of the end-points, this is more relevant in the case of HP due to its known limitations and results would imply a more favorable output gap than that obtained using instead the Hamilton approach. Figure 2: Comparing Output Gap between HP and Hamilton in Selected Member Economies (% potential GDP) ADB = Asian Development Bank, GDP = gross domestic product, HP = Hodrick-Prescott. Note: HP filter with lambda=100 and Hamilton filter-based decompositions. Output gaps expressed in percent of respective potential GDP. Source: Author. Figure 3 plots the potential GDP in national currencies for the same set of selected economies for a shorter period (the last 10 years) for visually better inspect differences. 15 The reason behind the selection of these four economies relates with representativeness: two are from Southeast Asia and two are from South Asia. Also, size-wise they do not represent small island states as data quality is typically scarcer and poorer and climate-related vulnerabilities makes them unique. -15 -10 -5 0 5 10 1990 1994 1998 2002 2006 2010 2014 2018 Year HP100 Hamilton Philippines -20 -10 0 10 20 1990 1994 1998 2002 2006 2010 2014 2018 Year HP100 Hamilton Indonesia -5 0 5 10 1990 1994 1998 2002 2006 2010 2014 2018 Year HP100 Hamilton Pakistan -2 -1 0 1 2 3 1990 1994 1998 2002 2006 2010 2014 2018 Year HP100 Hamilton Bangladesh 31 At the end of the period, the Hamilton-based potential GDP is always below that coming from the HP decomposition. Figure 3: Comparing Potential between HP and Hamilton in Selected Member Economies (local currency) ADB = Asian Development Bank, GDP = gross domestic product, HP = Hodrick-Prescott. Note: HP filter with lambda=100 and Hamilton filter-based decompositions. Source: Author. 6.2 Budget Elasticities 6.2.1 Output The primary challenge in calculating measures like the Current Account Balance (CAB), Cyclically Adjusted Primary Balance (CAPB), or the Structural Balance (SB) lies in determining the appropriate fiscal elasticities. A straightforward and simplistic approach would be to adopt Girouard and André’s (2005) convention, which assigns an elasticity of one to revenues and an elasticity of zero to expenditures, as previously discussed for the reasons provided. 12000 14000 16000 18000 20000 2011 2013 2015 2017 2019 Year real GDP HP100 Hamilton Philippines 7000000 8000000 9000000 10000000 11000000 2011 2013 2015 2017 2019 Year real GDP HP100 Hamilton Indonesia 9000 10000 11000 12000 13000 2011 2013 2015 2017 2019 Year real GDP HP100 Hamilton Pakistan 7000 8000 9000 10000 11000 12000 2011 2013 2015 2017 2019 Year real GDP HP100 Hamilton Bangladesh 32 For the ADB sample of members—Appendix 1 has the complete list—it is likely that the strength of the expenditure side of the budget to react to the economic cycle is small. Specifically, by examining IMF Government Financial Statistics data pertaining to unemployment-related expenses as a percentage of GDP, we can discern that across 17 economies with data spanning the most recent 5-year period (2015-2019, intentionally excluding the years impacted by COVID-19), this figure ranges from 0% (in several economies) to 1.9% of GDP (in Thailand). On average, it stands at 0.36%, in stark contrast to the average current government expenditure, which amounts to 14.85% of GDP. Figure 4 shows the economy-specific values. Additionally, one can estimate the aggregate expenditure elasticity with respect to a proxy of economic activity as detailed in Box 2. Doing so for a sample of 38 ADB member economies with available data yields the coefficients displayed in Figure 5. Elasticities vary from 0.809 (in Myanmar16) to 2.422 (in Timor-Leste) with an average value of 1.23. The majority of the members have an elasticity of zero (27 out of 38 economies). Against this evidence, it seems very reasonable to assume then for this sample of economies, at least as a starting point for our exercises, a budgetary spending elasticity of zero. 16 Effective 1 February 2021, ADB placed a temporary hold on sovereign project disbursements and new contracts in Myanmar. 33 Figure 4: Unemployment-related Benefits, Member Economies with Available Data: Average, 2015–2019 (% GDP) ADB = Asian Development Bank, GDP = gross domestic product, IMF = International Monetary Fund. Note: ADB placed on hold its regular assistance in Afghanistan effective 15 August 2021. Effective 1 February 2021, ADB placed a temporary hold on sovereign project disbursements and new contracts in Myanmar. Source: IMF Government Financial Statistics. Figure 5. Total Expenditure to Output Elasticity Coefficients of Member Economies ADB = Asian Development Bank Notes: OLS estimation of the change in real total expenditure revenues against real GDP growth. Only economies with at least 15 observations for real total government revenues were considered. Economies with no bars denote insignificant coefficient estimates (replaced by zero). ADB placed on hold its regular assistance in Afghanistan effective 15 August 2021. Effective 1 February 2021, ADB placed a temporary hold on sovereign project disbursements and new contracts in Myanmar. Source: Author. 0 0.5 1 1.5 2 2.5 3 Bhutan Brunei Darussalam Cambodia India Indonesia Republic of Korea Malaysia Maldives Philippines Thailand Viet Nam Solomon Islands Fiji Kiribati Vanuatu Papua New Guinea Samoa Tonga Tuvalu Armenia Kazakhstan Kyrgyz Republic Tajikistan People’s Republic of China Turkmenistan Uzbekistan Myanmar Sri Lanka Georgia Azerbaijan Afghanistan Mongolia Bangladesh Pakistan Marshall Islands Nepal Timor-Leste country-specific elasticity average economy - specific elasticity 34 Now, regarding the revenue side, one can estimate the aggregate revenue elasticity with respect to a proxy of economic activity as detailed in Box 2. Doing so for a sample of 38 ADB member economies with available data yields the coefficients displayed in Figure 6. Elasticities vary from 0.55 (in Papua New Guinea) to 2.5 (in Kazakhstan) with an average value of 1.6. The majority of the economies have an elasticity relatively close to 1 from above. Figure 6: Total Revenues to Output Elasticity Coefficients of Member Economies ADB = Asian Development Bank. Notes: OLS estimation of the change in real total government revenues against real GDP growth. Only economies with at least 15 observations for real total government revenues were considered. Economies with no bars denote insignificant coefficient estimates (replaced by zero). ADB placed on hold its regular assistance in Afghanistan effective 15 August 2021. Effective 1 February 2021, ADB placed a temporary hold on sovereign project disbursements and new contracts in Myanmar. Source: Author. 0.00 0.50 1.00 1.50 2.00 2.50 3.00 Papua New Guinea Bangladesh Bhutan Maldives Armenia Malaysia Uzbekistan Philippines Turkmenistan Timor-Leste 1. Republic of Korea Azerbaijan Nepal Myanmar Pakistan Indonesia Kyrgyz Republic Marshall Islands Singapore Sri Lanka India Kiribati Vanuatu Fiji Samoa Tonga Afghanistan Tuvalu Thailand Solomon Islands Georgia Mongolia Cambodia Tajikistan People’s Republic of China Kazakhstan country-specific elasticity average economy - specific elasticity 35 6.2.2 Asset Prices and Terms of Trade Let us begin by plotting the time series for real housing prices (from Bank for International Settlements, BIS) and the real terms-of-trade-adjustment (from World Bank, World Development Indicators, WDI) for our sample of ADB members for which these series have at least 15 continuous observations. Figures 7 and 8.1 plot the log of each series together with the log of real GDP in the secondary right hand side (RHS) axis for comparison.17 Looking at Figure 7 for housing price dynamics, we observe that in some economies, both patterns seem to go relatively hand-in-hand (e.g., Malaysia and Singapore) while in others, they seem disjoint (e.g. Indonesia and Thailand). Figure 7: Real Housing Prices and Real GDP (logs) Over Time of Member Economies ADB = Asian Development Bank, GDP = gross domestic product, ln = natural logarithm. Source: Author. 17 According to the World Bank, the list of commodity exporters that are part of the ADB membership are: Brunei Darussalam, Myanmar, Indonesia, Timor-Leste, Solomon Islands, Fiji, Papua New Guinea, Armenia, Azerbaijan, Kazakhstan, Kyrgyz Republic, Tajikistan, Turkmenistan, Uzbekistan, Mongolia. Many of these are islands and others are economies for which there is no available data to conduct a serious analysis. 15.215.415.615.81616.2 5.5 6 6.5 7 7.5 10 10.5 11 11.5 1212.51313.51414.5 5.25.45.65.8 6 6.2 8 8.5 9 9.5 4.6 4.7 4.8 4.9 4 4.5 5 4.5 4.55 4.6 4.65 4.7 4.4 4.6 4.8 5 4.2 4.4 4.6 4.8 4.6 4.7 4.8 4.9 19801990200020102020 19801990200020102020 19801990200020102020 Indonesia Malaysia People’s Republic of China Republic of Korea Singapore Thailand lnrealhouseprice lnrgdp year ln real house price ln real GDP 36 Figure 8.1: Real Terms-of-Trade Adjustment Prices and Real GDP (logs) Over Time of Member Economies ADB = Asian Development Bank, GDP = gross domestic product, ln=natural logarithm, TOT = terms of trade. Source: Author. Now let us estimate jointly the elasticity of revenue accounting additionally for housing and terms-of-trade. Results in Table 3 seem to suggest that, while the coefficient on real GDP growth is positive and significant as one would expect from a revenue elasticity (even if different from those displayed in Figure 6 as the underlying specification is not the same), the growth of real housing prices does not seem to have an effect on revenues (all coefficients for the 6 economies for which we can estimate with some degree of strength are statistically insignificant). In Table 4, we expand the specification used in Table 3 to also include the real terms-of-trade-adjustment growth. Here, the common set dictates an even smaller sample of only four ADB member economies. Still, in general, the role of housing prices remains orthogonal to revenues in these economies and only in Indonesia it seems that terms-of-trade have a certain role to play. In particular, the coefficient is positive and significant, but in terms, the one from real GDP growth 6.5 7 7.5 8 7.588.599.5 2.533.544.5 99.51010.511 1010.51111.512 1414.51515.516 8 8.5 9 9.5 4.24.44.64.855.2 7.588.599.5 88.599.510 1212.51313.51414.5 5 5.5 6 6.5 1.41.61.822.22.4 7.588.599.5 7.588.599.5 23.52424.52525.5 10152025 5 10 15 24262830 22242628 30313233 24262830 2122232425 2425262728 20222426 29303132 18202224 18.51919.520 2022242628 2022242628 19801990200020102020 19801990200020102020 19801990200020102020 19801990200020102020 Armenia Bangladesh Bhutan Cambodia India Indonesia Kazakhstan Kyrgyz Republic Pakistan Philippines Republic of Korea Singapore Solomon Islands Sri Lanka Thailand lnrealtotadj lnrgdp year r eal TOT a djustment p rices ln real GDP 43 For completeness, the charts equivalent to Figure 9.1 for all individual ADB member economies—for which it is possible to obtain the CAPB—are available in Appendix 3 Figure A3.1. Underlying data is available upon request. 2 Turning to exercise (2), we now use the estimated revenue elasticities from Section 6.2.1 to obtain the CAPB. The result is shown in Figure 10. As only Indonesia and the Philippines have elasticities that are different (for Pakistan and Bangladesh, the assumption of unitary revenue elasticity is applied), only these two economies are shown. Compared to Figure 9.1 for the Philippines, the lines seem to be closer to one another, while in the case of Indonesia, this is true only for the post-GFC period. Figure 10: CAPB (% GDP) with Estimated Revenue Elasticities in Selected ADB Member Economies (Alternative Filters) ADB = Asian Development Bank, CAPB = cyclically adjusted primary balance, GDP = gross domestic product, HP = Hodrick-Prescott, OB = overall balance. Source: Author. 3 Regarding (3), instead of assuming a zero-expenditure elasticity, we estimated these economy-by-economy similarly to what we did for revenues, again recall Section 6.2.1. Results of these elasticities were shown in Figure 5 where missing bars denote -8 -6 -4 -2 0 2 1990 1994 1998 2002 2006 2010 2014 2018 Year OB HP Hamilton Philippines -6 -4 -2 0 2 4 1990 1994 1998 2002 2006 2010 2014 2018 Year OB HP Hamilton Indonesia 44 statistically insignificant coefficients. As previously noted, most economies are characterized by the zero-expenditure elasticity. That said, the new resulting CAPB for Bangladesh and Pakistan is shown in Figure 11 (note that now the cases of Indonesia and the Philippines are the reverse of (2); they have insignificant expenditure elasticities, which in effect would yield the result shown in Figure 10). Figure 11: CAPB (% GDP) with Both Revenue and Expenditure Estimated Elasticities in Selected ADB Member Economies (Alternative Filters) ADB = Asian Development Bank, CAPB = cyclically adjusted primary balance, GDP = gross domestic product, HP = Hodrick-Prescott, PB = primary balance. Source: Author. 4 Finally, for (4), we now return to a zero-expenditure elasticity but now revenue elasticities are estimated in a time-varying fashion for each economy. Results of these elasticities are shown in Figure 12 where we add the 90% confidence bands. For the Philippines, revenue elasticity has been mostly positive and significant over the 1990–2021 time -6 -4 -2 0 2 4 1990 1994 1998 2002 2006 2010 2014 2018 Year PB HP Hamilton Philippines -6 -4 -2 0 2 4 1990 1994 1998 2002 2006 2010 2014 2018 Year PB HP Hamilton Indonesia -10 -5 0 5 1990 1994 1998 2002 2006 2010 2014 2018 Year PB HP Hamilton Pakistan -10 -5 0 1990 1994 1998 2002 2006 2010 2014 2018 Year PB HP Hamilton Bangladesh 45 period. This pattern contrasts with the other three economies for which revenue elasticities have been positive but surrounded by larger uncertainty (with most of the years with confidence bands above and below the zero line, except Indonesia in the last decade). Hence, Indonesia, using a time-varying version of the elasticity coefficient would potentially underestimate the true historical cyclically adjusted revenue, which was positive and around 1 according to Figure 6. For Bangladesh and Pakistan static and time-varying evidence is consistent as Figure 6 also shows an insignificant coefficient estimate. For completeness, the charts equivalent to Figure 12 for all individual ADB member economies are available in Appendix 3 Figure A3.2. Underlying data is available upon request. Figure 12: Time-Varying Revenue Elasticities in Selected Member Economies (Alternative Filters) Philippines Indonesia Pakistan Bangladesh ADB = Asian Development Bank. Source: Author. 0 5 10 1990 2000 2010 2020 Philippines Coefficient of growth ub10r lb10r year -10 0 10 20 1990 2000 2010 2020 Coefficient of growth ub10r lb10r year -5 0 5 10 1990 2000 2010 2020 Coefficient of growth ub10r lb10r year -10 0 10 20 30 1990 2000 2010 2020 Coefficient of growth ub10r lb10r year Graphs by Country 46 Figure 13: CAPB (% GDP) with Time-varying Revenue Estimated Elasticities in Selected Member Economies (Alternative Filters) ADB = Asian Development Bank, CAPB = cyclically adjusted primary balance, GDP = gross domestic product, HP = Hodrick-Prescott, PB = primary balance. Source: Author. To conclude this section, we plot in Figure 14 the comparison of the Hamiltonbased CAPB against the PB for each of the four exercises and the four economies under scrutiny. We see that the CAPB versions that follow closest the primary balance are those from exercises (1) and (3). Exercise (4) with the time-varying revenue elasticities is the one giving the most different time profile and generally suggesting a worse fiscal position (the blue line tends to be below the other ones). Still within each economy there is dispersion among the four lines that is larger at the beginning (end) of the time period of Bangladesh (Pakistan and the Philippines). Hence, no one size fits all can be applied. -10 -5 0 5 1990 1994 1998 2002 2006 2010 2014 2018 Year PB HP Hamilton Philippines -10 -5 0 5 10 1990 1994 1998 2002 2006 2010 2014 2018 Year PB HP Hamilton Indonesia -30 -20 -10 0 10 1990 1994 1998 2002 2006 2010 2014 2018 Year PB HP Hamilton Pakistan -15 -10 -5 0 5 1990 1994 1998 2002 2006 2010 2014 2018 Year PB HP Hamilton Bangladesh 47 Figure 14: CAPB (% GDP) Comparison across the Four Exercises in Selected Member Economies (Hamilton Filter) Philippines Indonesia Pakistan Bangladesh ADB = Asian Development Bank, CAPB = cyclically adjusted primary balance, GDP = gross domestic product, PB = primary balance. Note: “PB” denotes the primary balance (% GDP). “CAPB_ham_i”, for i=1,…,4, denotes the resulting Hamilton-based CAPB (% GDP) for exercises 1–4 described earlier in this section. Source: Author. 6.4 Estimating One-Offs Decomposing net capital transfers into trend and cycle poses its practical problems. First, a net variable can take both positive and negative values that renders impossible the application of logarithm, which is a prior condition, as commonly used in the literature, to apply a statistical filter. That said, since most one-offs are operations that increase net transfers to improve the fiscal stance, we will focus on credits alone. This variable is transformed into real quantities using the GDP deflator, then the logarithm is computed 48 and applied to a given filter. Out of the previous selection of economies, we need to pick a new set as the transfers gap in percent of GDP for the previous group is very small (below 0.1%, which is the value used to consider one-offs). In what follows, we look at Armenia, Bhutan, the Kyrgyz Republic, and Timor-Leste for more heterogeneity.19 The capital transfers (credit) gap (expressed in % of GDP) is shown in Figure 14 for these four economies. For completeness, the charts equivalent to Figure 15 for all individual ADB member economies for which there is data, are available in Appendix 3, Figure A3.3. Underlying data is available upon request. Figure 15: Government Capital Transfers-Gap (% GDP) in Selected Member Economies, Alternative Filters, 2000–2021 19 Other eligible economies that have transfers gap in percent of GDP larger than 0.1 include mostly islands in the Pacific: Kiribati, Samoa, Solomon Islands, Tonga, Tuvalu, and Vanuatu. .2 .4 .6 .8 Transfers Gap (credit) (% GDP) 2000 2004 2008 2012 2016 2020 Year HP Hamilton Timor-Leste 0 .2 .4 .6 .8 1 Transfers Gap (credit) (% GDP) 2000 2004 2008 2012 2016 2020 Year HP Hamilton Armenia Continued on the next page 49 ADB = Asian Development Bank, GDP = gross domestic product, HP = Hodrick-Prescott. Source: Author. Table 6 identifies the years for which the change in the transfers (credit) gap is in the economy’s top 90th percentile and at the same time, that gap is larger than 0.1% of GDP. Whether or not these years correspond to “true” one-offs, this is something to be confirmed by desk economists and local authorities using a more narrative-based approach. Appendix 4 provides the full list of ADB member economies with identified oneoff years computed using either the HP or Hamilton approaches. Table 6: Years of Deficit Reducing One-Offs in Selected Member Economies Computed Using Cyclical Capital Transfers (Credit) Approach HP Hamilton Economy Potential O ne - off Years Potential O ne - off Years Armenia 2002, 2018 - Bhutan 2010, 2013 - Kyrgyz Republic 2002 2002, 2003 Timor - Leste 2017, 2018 2009, 2011, 2018 ADB = Asian Development Bank, GDP = gross domestic product, HP = Hodrick-Prescott. Note: Years at the top 90th percentile of the distribution of the change in capital transfers gap and with capital transfers gap larger than 0.1% of GDP. Source: Author. 0 .05 .1 .15 Transfers Gap (credit) (% GDP) 2000 2004 2008 2012 2016 2020 Year HP Hamilton Bhutan 0 .5 1 Transfers Gap (credit) (% GDP) 2000 2004 2008 2012 2016 2020 Year HP Hamilton Kyrgyz Republic 50 7. Conclusion This report aims to review the current discussions surrounding the estimation of reasonable proxies of discretionary fiscal stance by reviewing methods and practices. An empirical application to ADB’s members is carried out but full results are available in the Appendixes. The main takeaway is that there is no one way to adjust fiscal balances. We can conclude from the empirical exercises that there is no one-size fits all and desk sensibility and judgment is important in various degrees. The choice of the filter to decompose trends and cycles matters with a preference to avoid the outdated and problematic HP filter in favor of the Hamilton (2018) approach. Moreover, the take on how to adjust revenues and expenditures entering the CAB or SB is paramount. There is a plethora of choices to be made, from prior empirical estimation at the economy-specific level, to whether such analysis should be done in a static or time-varying fashion, to which subcomponents of revenues and expenditures are likely to fluctuate with the cycle or be orthogonal to it. Furthermore, if one thinks about one-off operations, then a more narrative-based approach can complement the ad hoc identification of large changes in cyclically adjusted government capital transfers. Finally, there is the discussion of other factors in addition to the cycle, such as asset and commodity prices, which includes changes in housing prices and terms of trade adjustments that can prove to be difficult to be reflected in a measure of the underlying fiscal position. Ultimately, I would say that for the selected set of members looked at more closely, asset prices, commodity prices, and one-offs do not seem to be important. Moreover, as the government size is relatively small and stable in most economies, assuming a zero 51 elasticity for this item seems appropriate. As for revenue elasticities, where some fine tuning with economy-specific elasticities can be made, the imposition of a unitary value for simplicity also seems appropriate in most contexts. But if prior economy-specific estimations of revenue elasticities are to be conducted, I would rather go with a static estimation in the aggregated or disaggregated approaches. 52 REFERENCES Abiad, Abdul, Davide Furceri, and Petia Topalova. 2016. “The Macro-economic Effects of Public Investment: Evidence from Advanced Economies.” Journal of Macroeconomics 50: 224–40. Adarov, A., B. Clements, and J. Jalles. Forthcoming. “Growth Effects of Public Investment and Public Consumption: The Role of the Business Cycle, Fiscal Space and Efficiency.” World Bank Working Paper. Afonso, Antonio, and João Tovar Jalles. 2014. “Assessing Fiscal Episodes.” Economic Modelling, 37: 255–70. Alichi, Ali, Olivier Bizimana, Silvia Domit, Emilio Fernandez Corugedo, Douglas Laxton, Kadir Tanyeri, Hou Wang, and Fan Zhang. 2015. “Multivariate Filter Estimation of Potential Output for the Euro Area and the United States.” IMF Working Paper No. 15/253. Washington, DC: International Monetary Fund (IMF). Alichi, Ali, Olivier Bizimana, Douglas Laxton, Kadir Tanyeri, Hou Wang, Jiaxiong Yao, and Fan Zhang. 2017. “Multivariate Filter Estimation of Potential Output for the United States.” IMF Working Paper No. 2017/106. Washington, DC: International Monetary Fund (IMF). Andersson, Malin, Bela Szörfi, Máté Tóth, and Nico Zorell. 2018. “Potential Output in the Post-Crisis Period.” ECB Economic Bulletin Issue 7/2018. Anderton, Robert, Ted Aranki, Alistair Dieppe, Catherine Elding, Stephan Haroutunian, Pascal Jacquinot, Valerie Jarvis, Vincent Labhard, Desislava Rusinova, and Béla Szörfi. 2014. “Potential Output from a Euro Area Perspective.” Occasional Paper Series No. 156. Frankfurt: European Central Bank (ECB). Arnold, Robert W. 2009. “The Challenges of Estimating Potential Output in Real Time,” Federal Reserve Bank of St. Louis Review. July/August 2009. Auerbach, Alan, and Daniel Feenberg. 2000. “The Significance of Federal Taxes as Automatic Stabilizers.” Journal of Economic Perspectives 14 (3): 37–56. Auerbach, Alan, and Yuriy Gorodnichenko. 2013a. “Fiscal Multipliers in Recession and Expansion.” In Fiscal Policy After the Financial Crisis, edited by Alberto Alesina, and Francesco Giavazzi, 63–98. Chicago: University of Chicago Press/National Bureau of Economic Research (NBER). _____. 2013b. “Measuring the Output Responses to Fiscal Policy.” American Economic Journal: Economic Policy 4 (2): 1–27. Ball, Laurence, Davide Furceri, Daniel Leigh, and Prakash Loungani. 2019. ”Does One Law Fit All? Cross-Country Evidence on Okun’s Law.” Open Economies Review 30: 841–74. ASIAN DEVELOPMENT BANK ASIAN DEVELOPMENT BANK 6 ADB Avenue, Mandaluyong City 1550 Metro Manila, Philippines www.adb.org ESTIMATING STRUCTURAL BUDGET BALANCES IN DEVELOPING ASIA João Tovar Jalles ADB ECONOMICS WORKING PAPER SERIES NO. 719 April 2024 Estimating Structural Budget Balances in Developing Asia This paper reviews the current discussions, methods and practices surrounding the estimation of reasonable proxies for the underlying fiscal position, a useful anchor for fiscal policy. An empirical application to developing Asian economies is carried out. There is no one-size fits all type of approach and the sensitivity and discernment regarding specific economies are important in various stages. About the Asian Development Bank ADB is committed to achieving a prosperous, inclusive, resilient, and sustainable Asia and the Pacific, while sustaining its efforts to eradicate extreme poverty. Established in 1966, it is owned by 68 members —49 from the region. Its main instruments for helping its developing member countries are policy dialogue, loans, equity investments, guarantees, grants, and technical assistance.