adb economics working paper series · the adb economics working paper series is a quick ......

ADB Economics Working Paper Series

Effectiveness of Countercyclical Fiscal Policy: Time-Series Evidence from Developing Asia

Shikha Jha, Sushanta Mallick, Donghyun Park, and Pilipinas QuisingNo. 211 | August 2010

ADB Economics Working Paper Series No. 211

Effectiveness of Countercyclical Fiscal Policy: Time-Series Evidence from Developing Asia

Shikha Jha, Sushanta Mallick, Donghyun Park, and Pilipinas Quising August 2010

Shikha Jha is Principal Economist, Macroeconomics and Finance Research Division, Economics and Research Department, Asian Development Bank (ADB); Sushanta Mallick is Reader, School of Business and Management, Queen Mary University of London; Donghyun Park is Principal Economist, Macroeconomics and Finance Research Division, Economics and Research Department, ADB; and Pilipinas Quising is Economics Officer, Macroeconomics and Finance Research Division, Economics and Research Department, ADB. This paper was presented at workshops in ADB. It has benefited from the comments of participants. The authors are also grateful for discussions and comments from Jong-Wha Lee, Joseph Ernest Zveglich, and Maria Socorro Bautista. However, they solely are responsible for any remaining errors.

Asian Development Bank6 ADB Avenue, Mandaluyong City1550 Metro Manila, Philippineswww.adb.org/economics

©2010 by Asian Development BankAugust 2010ISSN 1655-5252Publication Stock No. WPS102320

The views expressed in this paperare those of the author(s) and do notnecessarily reflect the views or policiesof the Asian Development Bank.

The ADB Economics Working Paper Series is a forum for stimulating discussion and eliciting feedback on ongoing and recently completed research and policy studies undertaken by the Asian Development Bank (ADB) staff, consultants, or resource persons. The series deals with key economic and development problems, particularly those facing the Asia and Pacific region; as well as conceptual, analytical, or methodological issues relating to project/program economic analysis, and statistical data and measurement. The series aims to enhance the knowledge on Asia’s development and policy challenges; strengthen analytical rigor and quality of ADB’s country partnership strategies, and its subregional and country operations; and improve the quality and availability of statistical data and development indicators for monitoring development effectiveness.

The ADB Economics Working Paper Series is a quick-disseminating, informal publication whose titles could subsequently be revised for publication as articles in professional journals or chapters in books. The series is maintained by the Economics and Research Department.

Contents

Abstract v

I. Introduction 1

II. Empirical Framework and Data Adjustments 4

A. Methodology 4 B. Data Sources and Adjustments 7

III. Results from the Empirical Analysis 9

A. Impulse Responses to Fiscal Policy Shocks 10 B. Cumulative Output Multipliers 15 C. Variance Decomposition Analysis 19

IV. Summary and Concluding Observations 21

Appendix 23

References 25

Abstract

As the global crisis hit developing Asia, several countries instituted fiscal stimulus measures to create domestic demand. With the region returning to normal times, in this paper we draw lessons using historical data from 10 developing Asian countries to examine if countercyclical fiscal policy can still be used to stimulate growth. To do so, we use a sign-restrictions-based structural vector autoregression model. We find that expansionary expenditure shocks have an insignificant effect on output but contractionary revenue shocks have a negative effect. On the basis of those estimated effects, we perform and compare two policy experiments: deficit-financed tax cuts and deficit spending. The experiment results indicate that while deficit-financed tax cuts stimulate economic activity, the impact of deficit spending is ambiguous. Our overall evidence thus suggests that tax cuts may be a more effective countercyclical policy instrument than government spending. However, a number of factors suggest that Asian governments should be cautious about actively using tax cuts for countercyclical purposes, in part because a big part of the revenue shocks in developing Asia are cyclical rather than discretionary.

I. Introduction

The severe global crisis in 2008–2009 led to a deep contraction in developing Asia’s growth. The pronounced negative impact on exports deprived the region of a traditionally vital source of demand. The consequent weakness of aggregate demand was further compounded by feeble private consumption and investment. As growth plummeted and inflationary pressures subsided, governments in several Asian developing countries implemented expansionary monetary policy. This helped stabilize the financial sector but did little to revitalize aggregate demand. In many countries interest rates were lowered so close to zero that it limited the scope for using monetary policy further. Asian governments then turned to activist fiscal policy by acting as the consumer of last resort. This was facilitated by the relatively healthy state of public finances in Asia vis-à-vis industrialized countries. Such proactive use of monetary and fiscal policy for countercyclical purposes marks a significant departure for a region that has traditionally used macroeconomic policy to promote macroeconomic stability rather than to smooth the business cycle. Although the region is now looking up again and the significant stress of the crisis has dissipated, the question remains: Can fiscal policy be used to stimulate growth as the region returns to normal times? What lessons can be drawn from the past and the present as developing Asia moves forward?

The effectiveness of countercyclical fiscal policy will depend not only on its size but also its composition, i.e., relative importance of tax cuts versus government spending. Intuitively, it is tempting to believe that government spending has a bigger impact on output since it has a more immediate and direct impact on aggregate demand. Public spending involves government’s direct purchase of goods and services produced by the economy. Spending on public works and infrastructure such as roads and power plants are examples of such direct demand. On the other hand, tax cuts have a less direct impact on aggregate demand since their effectiveness ultimately depends on the willingness of households and firms to spend the additional income resulting from the tax cuts. Economies in the real world are more complex, dynamic, open and uncertain, and do not always follow economic intuition. Therefore, the relative effectiveness of tax cuts versus government spending in boosting aggregate demand is ultimately an empirical issue that cannot be settled by economic intuition alone. The issue of whether fiscal policy enhances or retards long-run economic activity has long been debated in

the literature through the multiplier effects on output from increases in public spending or reductions in taxes.1 The evidence from such empirical studies is far from conclusive.

Fiscal multiplier, or the increase in output due to a one dollar increase in government spending or a one dollar reduction in taxes, is the underlying concept that measures how effectively tax cuts or government spending stimulates output. A large and growing number of studies have estimated the size of the multiplier. Those studies have produced a wide range of estimates, ranging from negative to more than one. This lack of clear-cut empirical evidence helps explain why economists are deeply divided about the usefulness of countercyclical fiscal policy as a tool for fighting recessions. In comparison to the very large empirical literature on fiscal multipliers, relatively few studies have explicitly compared the relative effectiveness of tax cuts versus government spending.

There is substantial recent literature on this topic, but much disagreement remains. The early literature on the role of fiscal policy in the process of economic development can be traced back to Easterly and Rebelo (1993). Since then a large number of studies have considered the effects of fiscal policy on growth in diverse countries and regions, applying a variety of methodologies and considering different types of data. In the empirical literature, most studies apply vector autoregressive methods aiming at identifying the usual reactions of the aggregate variables to the exogenous shocks in fiscal policy. See, for example, Blanchard and Perrotti (2002); Burnside, Eichenbaum, and Fisher (2003); Galí, López-Salido, and Vallés (2007); and more recently Burriel et al. (2009), and Mountford and Uhlig (2009). However, the evidence is mixed and most of the papers focus on countries of the Organisation for Economic Co-operation and Development, G7, and other industrial economies. Based on a meta analysis of a sample of 93 published studies, Nijkamp and Poot (2004) provide evidence that on balance, the positive effect of conventional fiscal policy on growth is rather weak. However, according to Spilimbergo, Symansky, and Schindler (2009), the overall evidence from the literature indicates that the multiplier for government spending is larger than that for tax cuts. Spilimbergo, Symansky, and Schindler (2009) put forth the following rule of thumb for government consumption: a multiplier of 1–1.5 in large countries; 0.5–1 in medium-size countries; and 0.5 or less in small open economies. The same rule of thumb postulates multipliers of only about half the above values for tax cuts and transfer payments and slightly larger multipliers for government investment. Baldacci, Gupta, and Mulas-Granados (2009) also find fiscal expansions based on government consumption to be more effective than those

1 Among others, two contrasting views come from the basic Keynesian and Ricardian theories. In the simple Keynesian world, where aggregate demand determines output given rigid prices, and where consumption responds to current income, fiscal expansion has a multiplier effect on growth. In contrast, fiscal multiplier is zero under Ricardian equivalence between taxes and debt in a dynamic framework. In this case, Ricardian consumers are forward-looking and fully aware of the government’s intertemporal budget constraint. Since they know that a tax cut today will be financed by higher taxes in the future, their consumption does not change because permanent income is unaffected. Similarly, the knowledge that an increase in government spending by borrowing today will be offset by future spending cuts leaves output unaffected.

2 | ADB Economics Working Paper Series No. 211

based on public investment or income tax cuts. Romer and Bernstein (2009) estimated the multipliers of government spending by the United States (US) and tax cuts to be 1.57 and 0.99, respectively.

Interestingly, a number of studies contradict the intuitively plausible notion that government spending has a bigger impact on output than tax cuts.2 Romer and Romer (2009) find that one dollar of tax cuts historically raised US gross domestic product (GDP) by about three dollars. A multiplier of 3 is far higher than most of the estimated multipliers for government spending. For example, the empirical results of Ramey (2009) imply a government spending multiplier of 1.4, and this is on the high end of estimates. The somewhat surprising implication is that tax cuts have a bigger effect on economic activity than government spending. However, Romer and Romer (2009) are not alone in uncovering a stronger impact of tax cuts. Applying the latest econometric methodology to US data, Mountford and Uhlig (2009), henceforth M-U, find that deficit-financed tax cuts have a bigger effect than deficit-financed spending or tax-financed spending. In a comprehensive analysis of 91 episodes of fiscal expansions in 21 OECD countries since 1970, Alesina and Ardagna (2009) compared expansions that brought about output growth with those that failed to do so. Strikingly, successful fiscal stimulus programs were based almost entirely on tax cuts, whereas unsuccessful programs were based mostly on additional government spending. Blanchard and Perotti (2002) find that higher taxes and higher government spending have a strong negative effect on private investment. Similarly, M-U find the cumulative multipliers for revenue (tax) shocks for the US to be typically greater than the spending shocks; and both increases in taxes and increases in government spending have a negative effect on consumption and investment spending. This suggests a possible explanation for why some studies find tax cuts to be more expansionary than government spending. Tax cuts may further boost output by stimulating investment while the positive effect of government spending may be largely offset by lower private investment. Perotti (1999) also finds that government revenue shocks have very different effects on private consumption in times of fiscal stress than in normal times. In this context, for the US, Cogan et al. (2010) find that the government spending multipliers from permanent increases in federal government purchases are much less in new Keynesian models than in old Keynesian models, with the multipliers being less than one as consumption and investment are crowded out, turning negative as the government purchases decline in the later years of their simulation.

Hemming et al. (2002), who summarize theoretical and empirical literature on the effectiveness of fiscal policy in stimulating economic growth, note that given data deficiencies and institutional weaknesses, there is no clear conclusion on the size and sign of fiscal multipliers in developing countries. Unlike a plethora of studies on industrialized countries debating whether tax cuts are more effective than government 2 A recent study (Caldara and Kamps 2008) that presents a vector autoregression-based comparative analysis, using

recursive approach, Blanchard-Perotti approach, sign-restrictions approach, and event-study approach, notes that government spending shocks yield very similar results both qualitatively and quantitatively, but tax shocks yield strongly diverging results.

Effectiveness of Countercyclical Fiscal Policy: Time-Series Evidence from Developing Asia | 3

spending, there is only limited evidence about the issue in the Asia and Pacific region. To some extent, this simply mirrors the fact that industrialized countries have a longer tradition of using fiscal policy for countercyclical purposes than developing Asian countries, which have accorded a higher priority to growth rather than output stability. For example, automatic stabilizers such as unemployment benefits are an integral part of fiscal policy in the G3 but an underdeveloped novelty in much of the Asian region. There is no study examining the issue specifically in the context of developing Asia. In this paper we attempt to fill this gap in the literature by empirically examining the relative effectiveness of tax cuts versus government spending in boosting aggregate demand in developing Asia and in so doing, give the region’s policymakers some guidance on how to maximize the impact of scarce fiscal resources.

Based on econometric analysis using historical time-series data, this paper analyzes what type of fiscal policy—tax cuts versus higher spending—works best during troughs in business cycles. The rest of this paper is organized as follows. Section II outlines the empirical methodology used for investigating the relative effectiveness of tax cuts versus government spending in boosting output in developing Asia and describes the data used. Section III reports and discusses the main findings emerging from the analysis. Section IV reviews the central messages of the paper along with their implications for the region’s policymakers.

II. Empirical Framework and Data Adjustments

A. Methodology

To analyze the dynamic effects of unexpected shocks in government spending and revenues on economic activity, which can be different across countries, this paper estimates an econometric model using historical time-series data from emerging Asian economies. It applies a new methodology due to M-U, based on a structural vector autoregression (VAR) framework that helps to identify fiscal shocks in the data alongside other shocks by imposing sign restrictions for the identification of each shock. Explicit identifying sign-restrictions are imposed to isolate exogenous and unanticipated changes in these variables by restricting the contemporaneous interaction of fiscal and nonfiscal variables. The paper uses M-U’s penalty function approach, which has the advantage of only picking up large shocks and thereby reducing the variation in the identified shocks, as we try to explain the variation in the macro economy using n or less number of shocks (where n is the dimension of the VAR). The methodology is implemented using the RATS econometrics software. All the endogenous variables in the model depend on each other through their lagged values. The optimal lag length is determined endogenously. In M-U’s methodology, the VAR does not contain a constant or a time trend. But including


a constant and a time trend does help smooth the impulse responses. Here we use a constant in the VAR, as there could be several shifts in the data series for emerging economies.

To see whether a particular theory (or channel) is accepted, restrictions are imposed on the contemporaneous effects of shocks, which makes it a structural model. The channel of transmission is either direct government spending or tax cut leading to higher private consumption or private investment and thereby output recovery. The responses can be sensitive depending on what variables and what sample periods are included in the model. The effectiveness depends on whether the cumulative multiplier of a policy shock is significant, and which type of policy (deficit-financed tax cut or government spending) has a dominant effect.

This methodology has some salient features. It distinguishes between unanticipated fiscal shocks (“news shocks”) and automatic responses of fiscal variables to business cycle conditions, e.g., automatic stabilizers such as unemployment benefits and social safety nets that automatically kick in during downturns unlike discretionary fiscal policy. The model also takes account of the “announcement effect”, namely the lag between implementation and announcement of fiscal policy that may affect macro variables. For instance, firms and individuals may adjust their choices before an announced tax increase takes effect. These features are modeled by imposing sign restrictions on the shape of the impulse responses to capture the relative movements of business cycle, fiscal policy, and monetary policy variables. The basic intuition is that structural shocks can be identified by checking whether the signs of the corresponding impulse responses are in line with theoretical priors. Three structural shocks are identified by imposing sign restrictions: (i) business cycle shocks; (ii) monetary shocks; and (iii) fiscal shocks (i.e., revenue shocks and government spending shocks). The sign restrictions help identify the effects of unanticipated fiscal shocks and nonfiscal shocks on the following endogenous variables:

GDP – real GDP EXP – real government expenditure REV – real government revenue INT – interest rate (benchmark policy rate) MON – real broad money DEF – GDP deflator CON – real consumption INV – real investment

All the variables are expressed in log levels except the interest rate. The set of sign restrictions imposed to identify different shocks is presented in Table 1. No restrictions are imposed on the signs of the responses of the key variables of interest—GDP, consumption, and investment—to fiscal policy shocks. All variables are estimated with


at least two lags. Although the VAR model includes eight macroeconomic variables, we identify only four structural shocks, which means the shocks and the associated impulse vectors are selected by discarding those vectors that do not satisfy the imposed sign restrictions. Since we estimate four structural shocks, we leave some structural disturbances unidentified. The horizon k of the imposed restrictions is four quarters. The VAR model and its residuals can be represented as follows.

Table 1: Identifying Sign Restrictions

Nonfiscal and Fiscal Policy Shocks

Real GDP

Real GovernmentExpenditure

Real Government

Revenue

Policy Rate

Real Money

GDP Deflator

Real Consumption

Real Investment

Business cycle shock(growth) + ? + ? ? ? + +

Monetary policy shock(tightening) ? ? ? + – – ? ?

Government revenue shock ? ? + ? ? ? ? ?

Government expenditure shock ? + ? ? ? ? ? ?

Note: The sign restrictions on the impulse responses for each identified shock are shown as + (positive response for four quarters following the shock); – (negative response for four quarters following the shock); and ‘? (no restriction).

Consider the reduced-form VAR that has a moving average representation:

Y AY u

Y I A L u B L u

t t t

t t t

= +

= −( ) =−

−1 1

11 ( )

(1)

The usual SVAR approach assumes that the error terms, ut, are related to structural macroeconomic shocks, vt, via a matrix P: ut = Pvt, where P is the unique lower triangular matrix such that PP'=Σu. The sign restriction-based orthogonalization can be recovered by post-multiplying P for a nonsingular matrix L exhibiting orthonormal columns such that sign restrictions on the impulse responses are satisfied. It will then be possible to derive a new matrix C = PL, whose columns are the identified impulse vectors.

Given the four identified shocks, the VAR residuals, ut, can be represented as:

u C v C v C v C v C vt tBC

tMP

tGR

tGE

t= + + + + ′1 2 3 4˘ ˘ (2)

where BC denotes the business cycle shock, MP the monetary policy shock, GR the government revenue shocks, and GE the government expenditure shock. With each structural shock, the associated impulse vector is Ci. As there is a lesser number of shocks than the number of variables in the VAR, the unidentified shocks in vector v̆ are associated with the remaining columns of the matrix C̆ . Penalty function approach is employed, which penalizes the opposite responses of the imposed sign restrictions for each shock under minimization of a criterion function over four quarters.


Following a business cycle shock (output growth/contraction) that should be identified first, a monetary shock should be identified next, wherein the monetary authority is more likely to change its policy stance. Finally the fiscal authority introduces either a spending rise or tax cut (a fiscal shock). Hence the impacts of these shocks are checked sequentially to investigate whether fiscal policy shocks do help an economy to increase its output. From these effects, one can conclude in which countries fiscal policy will be more effective. A fiscal shock is a surprise change in fiscal policy and has two dimensions, government revenue or expenditure, which should be separated from (orthogonal to) business cycle or monetary shocks. Thus, whenever taxes and output move in the same direction, this is attributed to a business cycle shock. This reasoning holds similarly for spending and output. While identifying the fiscal shocks, we also impose no restrictions on the variables of interest: GDP, consumption, investment, and inflation. We also carry out a variance decomposition exercise to show which shock (tax cut or spending increase) has a dominant effect on the economy.

B. Data Sources and Adjustments

We use the latest available historical data to assess how quickly and in which direction a particular economy responds to unexpected shocks. The series is based on quarterly data for 10 emerging Asian economies—the People’s Republic of China (PRC); Hong Kong, China; India, Indonesia; the Republic of Korea; Malaysia; the Philippines; Singapore; Taipei,China; and Thailand (Table 2).

Table 2: Sample Size

Economy Observations Sample Period

China, People's Rep. of 58 1995:1–2009:2Hong Kong, China 68 1992:3–2009:2India 53 1996:2–2009:2Indonesia 66 1993:1–2009:2Korea, Rep. of 74 1991:1–2009:2Malaysia 74 1991:1–2009:2Philippines 98 1985:1–2009:2Singapore 86 1988:1–2009:2Taipei,China 128 1977:3–2009:2Thailand 66 1993:1–2009:2

Data adjustments have been made as follows:

(i) Real GDP and nominal GDP are obtained from CEIC Data Company, Ltd. (in local currency unit) and GDP deflator is derived as (nominal GDP/real GDP), which is used as price series for all countries.


(ii) Short-term interest rate is obtained from CEIC, with the policy rate from each country used as a proxy for short-term interest rate. The definition of policy rate, however, differs as follows: the PRC: 1-year lending rate; Hong Kong, China: discount rate; India: repo rate; Indonesia: Sertifikat Bank Indonesia rate; the Republic of Korea: overnight call rate; Malaysia: overnight policy rate; the Philippines: repurchase rate; Singapore: benchmark Singapore Interbank Offered Rate 3-months rate; Taipei,China: rediscount rate; Thailand: Bank of Thailand policy rate.

(iii) Real private consumption and total fixed investment are taken from CEIC. Wherever it is available in nominal terms, we have deflated the series, using GDP deflator as calculated above. As the PRC does not release quarterly statistics for its GDP components, we have generated quarterly series from the annual data for private consumption and gross capital formation in real terms from national accounts by using a temporal disaggregation technique that follows the pattern in the quarterly real GDP of the PRC while maintaining the sum of the new quarterly series to the original annual series over each period.

(iv) Government total revenue and expenditure are compiled in current prices from CEIC; these are then deflated by the GDP deflator in order to be expressed in real terms. We have converted annual fiscal data to quarterly series for Indonesia before 2000 using the quarterly pattern in government consumption expenditure available from national accounts.

(v) Broad money supply is M2 for all countries and also come from CEIC. Nominal M2 values have been deflated by the GDP deflator to get real money balances.

To get a longer, consistent time series for Indonesia, the Republic of Korea, and Malaysia, we have also rebased all the earlier GDP data and its components (2000 base year) to be comparable with the recent data (2005 base year).

Given the different data availability across countries, the use of the M-U sign restriction identification methodology allows for a very similar identification to be achieved across countries despite these data problems. This is because the M-U identification strategy identifies shocks using mild restrictions on multiple time series in all economies regardless of how they are measured and regardless of data idiosyncrasies, hence the data definition used in an individual country for a particular variable is of secondary importance.


III. Results from the Empirical Analysis

The identification of the structural disturbances relies on the theoretical sign restrictions discussed earlier. The sign restrictions methodology is robust to the ordering of variables, which makes the results sensitive in the case of the traditional recursive ordering in the SVAR approach. Even the ordering of shocks is not relevant since the restrictions on the signs of impulse response functions are doing the job of identifying the shocks. Both revenue shocks and spending shocks are identified with separate sign restrictions. Hence it is immaterial whether they are either orthogonal only to the business cycle shock or orthogonal to both the business cycle and monetary policy shocks. Although one could check the effect of a fiscal shock by imposing weaker identifying restrictions, such as restricting government spending alone to be positive on impact, such identification will hide the presence of other shocks at the same time, and may not uniquely identify the impact of a fiscal shock.

Besides, the SVAR literature is primarily focused on assessing the effect of contractionary monetary shocks, with little emphasis on fiscal shocks, and mainly covering industrial countries. Although emerging markets do not share similar structural characteristics, it is worth investigating the macroeconomic effects of different policy shocks in individual countries in the Asian region, along with exploring the case of “expansionary fiscal contraction”. The impacts of fiscal policy shocks would be influenced by the presence of nonfiscal shocks in an economy. It is therefore important to compare the effect of revenue versus government spending shocks on the economy, while also accounting for the presence of business cycle and monetary policy shocks. Identification via sign restrictions is relevant in this context, as our objective is to investigate the effects of shocks due to surprise movements in business cycles, monetary policies, and fiscal policies. The short-run responses for business cycle shock in Table 3 satisfy the sign restrictions for k = 1,...,K quarters. In our analysis, the impulse responses of fiscal and nonfiscal variables have been restricted for the first four quarters following the shock.3 But the significant responses of a business cycle shock in Table 3 indicate that the sign effect of the business cycle shock on output does hold for all countries.

3 Impulse responses measure the time profile of the effect of a shock (or impulse) on the (expected) future values of a variable. Short run is defined as four quarters over which the sign restrictions are imposed. Long-run response is calculated as the sum of the coefficients of the lagged variables in the VAR.


Table 3: Impact Effects of Different Shocks on GDP from the Sign-VAR

Shocks in China, Rep. of

Hong Kong, China

India Indonesia Korea,Rep. of

Malaysia

Business cycle +* +* +* +* +* +*Monetary policy – – – – +* –Government spending + – + – + +Government revenue + – – –* + –*

Philippines Singapore Taipei,China Thailand

Business cycle +* +* +* +*Monetary policy – + + –Government spending +* +* – –Government revenue –* +* – +

Note: * Indicates confidence interval (16th and 84th percentiles) being significantly different from zero.Source: Authors‘ calculations.

A. Impulse Responses to Fiscal Policy Shocks

Applying the nonlinear method, which uses Monte-Carlo integration, we find that tax shocks have a negative effect on output in some countries, while the spending shocks have a positive effect but could get crowded out in the medium term. Although a deficit-financed expenditure stimulus is possible, the eventual costs are likely to be much higher than the immediate benefits, as the increased deficit needs to be repaid with a subsequent hike in taxes. As shown in Figure 1, the contractionary effect of tax increases (positive revenue shock) suggests that a tax cut will have a positive impact on output in Hong Kong, China; India; Indonesia; Malaysia; and the Philippines in both the short and long run.4 In the PRC, Singapore, and Thailand, the direction of impact is reversed and implies that output may not adjust favorably to tax changes. The short- and long-run effects are mixed in the remaining two economies (the Republic of Korea and Taipei,China). As shown in Figure 2, positive government expenditure shocks seem to have a positive impact on output only in the short term in the PRC, India, the Republic of Korea, and the Philippines, while in Indonesia and Taipei,China the effect is positive only in the longer term. Malaysia and Singapore display consistently positive impacts, whereas the opposite is the case with Hong Kong, China and Thailand. Although the magnitude of the responses is generally small, the directions of change are revealing.

Tables 4 and 5 present the above impacts of fiscal policy shocks on output working through other macro variables. Our results reveal that tax shocks have a negative effect on output in several countries, while the spending shocks have a positive effect but could get crowded out in the medium term. In general, tax cuts are expected to boost income, consumption, and investment by cutting payroll taxes or increasing investment tax credits.

4 Impulse responses measure the time profile of the effect of a shock (or impulse) on the (expected) future values of a variable. Short run is defined as four quarters over which the sign restrictions are imposed. Long-run response is calculated as the sum of the coefficients of the lagged variables in the VAR.


The effects of this channel may be limited in developing economies because of possibly lower tax rates, weak tax bases, and large informal sectors. In a high-tax country, a tax cut could thus be more effective in stimulating demand, whereas in a country where the tax rate is already very low, any further cut is less likely to have a significant effect, whereas a positive government expenditure shock can be more effective in terms of its impact on output. This is contrary to the result in M-U for the US.

Figure 1: Impacts of Positive Revenue Shocks on Real GDP (percent)

PRC HKG IND INO

Immediate ImpactLong-run Impact

KOR MAL PHI SIN TAP THA

0.14

0.12

0.10

0.08

0.06

0.04

0.02

0.00

−0.02

−0.04

PRC = People‘s Republic of China; HKG = Hong Kong, China; IND = India; INO = Indonesia; KOR = Republic of Korea; MAL = Malaysia; PHI = Philippines; SIN = Singapore; TAP = Taipei,China; THA = Thailand.Source: Authors’ calculations.

Figure 2: Impacts of Positive Expenditure Shocks on Real GDP (percent)

Immediate ImpactLong-run Impact

0.10

0.08

0.06

0.04

0.02

0.00

−0.02

−0.04

−0.06

−0.08

PRC HKG IND INO KOR MAL PHI SIN TAP THA

PRC = People‘s Republic of China; HKG = Hong Kong, China; IND = India; INO = Indonesia; KOR = Republic of Korea; MAL = Malaysia; PHI = Philippines; SIN = Singapore; TAP = Taipei,China; THA = Thailand.Source: Authors’ calculations.


Higher taxes have a significant effect on output only in Southeast Asian countries—negative in Indonesia, Malaysia, and the Philippines in the short run but surprisingly positive in Singapore in both the long and short run (Table 4). While the negative effect in the Philippines remains significant in the long run, it is crowded out to some extent by investment. Taipei,China shows the counterintuitive positive long-run growth effect of higher taxes, which is even enhanced through higher consumption and investment. The apparently counterintuitive result for Singapore and Taipei,China may be depicting a case of expansionary fiscal contraction. That is, a fiscal contraction by raising taxes (revenue) would reduce interest rates, thus increasing wealth, encouraging investment, and leading to higher output. A 2009 IMF study on Singapore notes that the countercyclical effect of discretionary fiscal policy is, at best, short-lived.5 The factors include high propensity to save among households and low consumption response; focus of countercyclical measures on easing the corporate cost burden through tax measures; and significant leakages of fiscal stimulus abroad through trade as well as remittances. The result is insignificant in other countries.

An increase in government spending seems to have a limited effect on GDP (Table 5). It is limited to a positive short-term impact in the Philippines (supported by higher investment) and Singapore, and to a negative long-term impact in Hong Kong, China and Thailand. Public spending shock discourages investment in Thailand in both the long and short run. In other countries, higher government expenditure discourages investment, making the growth impact insignificant. Or, it increases inflation, thereby discouraging consumption. For example, in India, higher public expenditure drives up short-run interest rates and crowds out private investment in both the short and long run, thereby making the growth impact insignificant. Some of the effects are negative, though insignificant. They might imply that cuts in unproductive spending, especially transfers, result in “expansionary fiscal contractions”. Given the insignificant coefficient of spending shocks on output in several countries, one could argue that fiscal policy has had little destabilizing impact on real output and could reflect a pro-cyclical nature of fiscal policy from the spending side, whereas fiscal policy appears countercyclical from the revenue side.

Government spending shocks increase inflation in India; Indonesia; the Republic of Korea; Singapore; and Taipei,China, whereas in the remaining five countries, spending shocks do not have any inflationary impact. Any inflationary impact due to a spending shock can impact real wages and the labor market (see, for example, Pappa 2009).6 Linnemann and Schabert (2006) show that government expenditures can lead to a rise in private consumption, real wages, and employment if the government share is not too large and public finance does not solely rely on distortionary taxation, whereas when government expenditures are partially financed by public debt, unit labor costs fall in response to a fiscal expansion, such that inflation tends to decline.

5 Eskesen (2009a and 2009b) confirm a case for countercyclical fiscal policy in the Republic of Korea and Singapore.6 Given data on real wages, one could explore this further for developing Asian countries.


Tabl

e 4:

Impu

lse

Resp

onse

s to

Pos

itiv

e Ta

x Re

venu

e Sh

ock

(%)

Shoc

ks o

nIm

pact

s in

Real

GD

PG

over

nmen

t Ex

pend

itur

eG

over

nmen

t Re

venu

eIn

tere

stRa

teG

DP

Defl

ator

Real

M

oney

Priv

ate

Cons

umpt

ion

Fixe

d In

vest

men

t

Chin

a, P

eopl

e‘s

Rep.

of

SR0.

0171

0.03

990.

0386

*−0

.013

60.

0035

−0.0

006

0.01

010.

0163

LR0.

0030

0.06

340.

0372

*−0

.024

80.

0023

−0.0

029

0.00

42−0

.026

0

Hon

g Ko

ng, C

hina

SR−0

.002

30.

0028

0.07

23*

−0.0

002

−0.0

023

0.00

28−0

.002

50.

0165

LR−0

.017

80.

0616

0.21

95*

−0.0

183*

−0.0

708*

−0.0

442

−0.0

314

−0.0

067

Indi

aSR

−0.0

017

0.01

700.

0698

*−0

.001

6−0

.003

9−0

.002

1−0

.004

4−0

.028

0

LR−0

.018

50.

0086

0.10

15*

−0.0

083

−0.0

107

−0.0

223

−0.0

117

−0.1

897

Indo

nesi

aSR

−0.0

064*

−0.0

259

0.03

13*

0.00

300.

0136

*−0

.012

0−0

.006

0−0

.001

6

LR−0

.003

80.

1312

0.04

74*

−0.0

004

0.52

96*

−0.0

371

−0.0

269

−0.0

355

Kore

a, R

ep. o

fSR

0.00

060.

0416

0.06

43*

0.00

020.

0013

−0.0

018

0.00

03−0

.026

7

LR−0

.007

60.

0619

0.04

56*

0.00

18−0

.008

30.

0063

−0.0

069

−0.0

438

Mal

aysi

aSR

−0.0

070*

0.01

640.

0409

*−0

.001

3−0

.006

10.

0048

0.00

39−0

.001

4

LR−0

.008

7−0

.020

60.

0472

0.00

12−0

.011

2−0

.008

00.

0096

−0.0

063

Phili

ppin

esSR

−0.0

119*

−0.0

109

0.03

45*

0.00

08−0

.002

5−0

.001

0−0

.002

1−0

.045

3*

LR−0

.030

9*−0

.024

30.

1081

*0.

0258

*−0

.085

9*0.

1164

*−0

.008

80.

0682

*

Sing

apor

eSR

0.00

63*

−0.0

021

0.05

02*

0.00

14*

0.00

52*

−0.0

103*

−0.0

066*

−0.0

387

LR0.

0529

*0.

4124

*0.

4135

*0.

0050

0.03

53*

0.11

47*

−0.0

139

−0.0

487

Taip

ei,C

hina

SR−0

.002

10.

0464

*0.

0517

*0.

0008

−0.0

003

0.00

170.

0001

−0.0

136*

LR0.

1222

*0.

3508

*0.

2236

*0.

0132

*0.

1389

*0.

2212

*0.

2182

*0.

3846

*

Thai

land

SR0.

0035

−0.0

022

0.01

50*

0.00

17−0

.006

30.

0490

−0.0

034

−0.0

284

LR0.

0076

−0.0

127

0.04

440.

0024

−0.0

098

−0.0

821

0.00

970.

0587

* In

dica

tes

the

impa

ct b

eing

sig

nific

antly

diff

eren

t fro

m z

ero

(bot

h up

per [

84th

per

cent

ile] a

nd lo

wer

[16t

h pe

rcen

tile]

ban

ds a

re s

igni

fican

tly d

iffer

ent f

rom

the

zero

line

).SR

= s

hort

run;

LR

= lo

ng ru

n.So

urce

: A

utho

rs’ c

alcu

latio

ns.


Tabl

e 5:

Impu

lse

Resp

onse

s to

Pos

itiv

e Pu

blic

Spe

ndin

g Sh

ock

Shoc

ks o

nIm

pact

s in

Real

GD

PG

over

nmen

t Ex

pend

itur

eG

over

nmen

t Re

venu

eIn

tere

stRa

teG

DP

Defl

ator

Real

Mon

eyPr

ivat

e Co

nsum

ptio

nFi

xed

Inve

stm

ent

Chin

a, P

eopl

e‘s

Rep.

of

SR0.

0065

0.03

22*

0.02

11−0

.054

9−0

.018

6*0.

0110

*0.

0038

0.00

49LR

−0.0

100

0.02

45−0

.001

1−0

.460

7−0

.059

2*0.

0244

−0.0

053

0.00

59H

ong

Kong

, Chi

naSR

−0.0

015

0.07

93*

0.07

27−0

.000

1−0

.003

30.

0023

−0.0

006

0.00

49

LR−0

.017

4*0.

1835

*0.

2284

*−0

.029

9*−0

.104

1*−0

.043

5*−0

.034

4*−0

.033

1In

dia

SR0.

0027

0.03

65*

−0.0

022

0.00

18*

0.01

08−0

.005

2−0

.001

3−0

.222

9*

LR−0

.052

60.

0724

*−0

.018

4−0

.012

2*−0

.007

2−0

.095

2*−0

.023

7−0

.851

8*In

done

sia

SR−0

.000

40.

0074

*0.

0248

−0.0

021

0.01

63*

−0.0

201*

−0.0

054

−0.0

059

LR0.

0018

0.18

86*

0.00

68−0

.004

30.

6758

*−0

.055

5*−0

.026

6−0

.023

5Ko

rea,

Rep

. of

SR0.

0086

0.07

98*

0.05

00*

0.00

100.

0015

−0.0

016

−0.0

004

−0.0

185

LR−0

.008

30.

0798

*0.

0320

0.00

12−0

.005

8−0

.023

1−0

.016

9−0

.067

3M

alay

sia

SR0.

0023

0.09

17*

−0.0

125

−0.0

027*

−0.0

003

0.00

070.

0001

0.01

92*

LR0.

0098

0.26

21*

−0.0

223

−0.0

138

−0.0

539*

−0.0

660

−0.0

546*

0.02

57Ph

ilipp

ines

SR0.

0053

*0.

0709

*−0

.011

0−0

.000

3−0

.004

60.

0072

−0.0

002

0.02

74*

LR−0

.011

30.

1104

*−0

.060

0−0

.009

5−0

.072

7−0

.001

9−0

.014

0*0.

0743

*Si

ngap

ore

SR0.

0057

*0.

1539

*0.

0191

−0.0

001

0.00

16−0

.000

2−0

.000

6−0

.091

2

LR0.

0230

0.28

83*

0.04

960.

0068

−0.0

318

0.08

77*

0.02

68*

−0.2

211

Taip

ei,C

hina

SR−0

.001

70.

0709

*0.

0449

*0.

0004

−0.0

019*

0.00

37*

0.00

14−0

.015

0 *

LR0.

0921

*0.

3527

*0.

2092

*0.

0134

*0.

1061

*0.

1580

*0.

1879

*0.

3369

*Th

aila

ndSR

−0.0

017

0.07

76*

0.00

29−0

.002

9−0

.003

90.

0361

0.00

42*

−0.0

587*

LR−0

.057

7*0.

0114

−0.1

169*

−0.0

120*

−0.0

212

0.95

64−0

.047

0*−0

.403

3*

* In

dica

tes

the

impa

ct b

eing

sig

nific

antly

diff

eren

t fro

m z

ero

(bot

h up

per [

84th

per

cent

ile] a

nd lo

wer

[16t

h pe

rcen

tile]

ban

ds a

re s

igni

fican

tly d

iffer

ent f

rom

the

zero

line

).SR

= s

hort

run;

LR

= lo

ng ru

n.So

urce

: A

utho

rs’ c

alcu

latio

ns.


B. Cumulative Output Multipliers

Having observed the impact effects, we next derive cumulative output multipliers by undertaking two key policy experiments: a deficit-financed tax cut policy scenario and a deficit-spending fiscal policy scenario. Calculation of the present value multiplier uses information on average real interest rate, average government expenditure as a percentage of GDP, and average government revenue as a percentage of GDP, which are shown in Table 6.

Table 6: Estimates Used in the Calculation of Present Value of the Multiplier

RIR Revenue (percent of GDP)

Expenditure (percent of GDP)

China, People's Rep. of 3.86 17.67 17.95Hong Kong, China 2.76 16.05 15.26India 2.43 10.72 15.65Indonesia 1.24 17.61 18.59Korea, Rep. of 2.21 23.11 22.04Malaysia 1.24 21.15 25.58Philippines 1.83 15.49 18.57Singapore 1.01 23.31 16.70Taipei,China 2.06 19.15 21.63Thailand 1.11 16.68 17.72

RIR = real interest rate, GDP = gross domestic product, CPI = consumer price index.Note: The average numbers are calculated with data over the sample period considered here. RIR is calculated using the Fisher

identity as follows: r=(i-pe)/(1+pe) where r denotes the RIR, i the nominal interest rate, and pe is the expected inflation rate. We use the current CPI inflation as a proxy for pe and the policy rate as a measure of i.

Source: Authors‘ calculations.

Figures 3 and 4 display the present value GDP multipliers for both the experiments. We calculate present value of these responses and then calculate the multiplier for several quarters ahead. The multipliers given are the median multipliers (the 50th percentile of the responses) and the confidence interval is graphed as the upper and lower bounds. If the bands lie on one side (both positive or both negative), then the response is significantly different from zero; if the upper band is in the positive territory and the lower band is in the negative territory; then we cannot say that the response is significantly different from zero. The multiplier represents the effect of a 1% cut in taxes or a 1% increase in public spending in the first quarter. This is calculated with the following formula:

Multiplier for GDP

rGDP response

rIni

j jj

k

j

=

+( )( )

+( )

=∑ 1

11

1

0

ttial fiscal shock

Average fiscal

jj

k

( )

=

∑0

var iiable share of GDP as ( ) (3)

where r is real interest rate or the discount rate, and k = 0,…,4 periods.


Figure 3: Impulse Responses with Penalty Function Approach Multiplier Government Revenue (percent)

−6

−4

−2

0

2

4

6

8

1 3 5 7 9 11 13 15 17 19 21 23 25 1 3 5 7 9 11 13 15 17 19 21 23 25

1 3 5 7 9 11 13 15 17 19 21 23 25

1 3 5 7 9 11 13 15 17 19 21 23 25

1 3 5 7 9 11 13 15 17 19 21 23 25

1 3 5 7 9 11 13 15 17 19 21 23 25

1 3 5 7 9 11 13 15 17 19 21 23 25

1 3 5 7 9 11 13 15 17 19 21 23 25

1 3 5 7 9 11 13 15 17 19 21 23 25

1 3 5 7 9 11 13 15 17 19 21 23 25

Hong Kong, China

−2

−1

0

1

2

3

4

India

−20

−15

−10

−5

0

5

10

15

20

Indonesia

−10

−8

−6

−4

−2

0

2

4

Korea, Rep. of

−5

−4

−3

−2

−1

0

1

2

3

Malaysia

−6

−4

−2

0

2

4

6

8

Philippines

−2

0

2

4

6

Singapore

−2

0

2

4

6

8

Taipei,China

−15

−10

−5

0

5

10

Thailand

−30

−20

−10

0

10

20

30

Quarters after the Shock Quarters after the Shock





China, People’s Rep. of

Note: The present value multiplier has been calculated for the GDP multiplier effect due to deficit financed tax cuts. The error bands are illustrated as the dotted lines above and below the middle response line (the thick line), which are composed of the 16th (below), 84th (above) and median percentiles (middle) of the responses.



Figure 4: Impulse Responses with Penalty Function Approach Multiplier Government Expenditure (percent)

China, People’s Rep. of

1 3 5 7 9 11 13 15 17 19 21 23 25Quarters after the Shock





3 5 7 9 11 13 15 17 19 21 23 25Quarters after the Shock





Hong Kong, China

India

Indonesia

Korea, Rep. of

Philippines

Malaysia

Singapore

Taipei,China

Thailand

−3−2

−1

01

23

4

5

−4

−3

−2

−1

0

1

2

−6−4−2

02468

10

−6

−4

−2

0

2

4

6

8

10

−4

−3

−2

−1

0

1

2

3

−2

−1

0

1

2

−3

−2

−1

0

1

−8

−6

−4

−2

0

2

4

−2

−1

0

1

2

3

4

−30

−20

−10

0

10

20

30

40

Note: The present value multiplier has been calculated for the GDP multiplier effect due to deficit financed expenditure increase. The error bands are illustrated as the dotted lines above and below the middle response line (the thick line), which are composed of the 16th (below), 84th (above) and median percentiles (middle) of the responses.



We showed earlier that fiscal policy is effective in stimulating output, but the effect varies across countries and instruments according to government spending and tax revenue. In terms of the size of the impact of the two fiscal instruments, we have calculated the multipliers using the above formula. The cumulative multiplier is defined as the cumulative change in output over the cumulative change in fiscal shocks at some time horizon k. In present value terms, tax cuts have a much greater effect on GDP than government spending, as tax cuts can produce a bigger boost in demand, not only by increasing consumption spending via a rise in disposable income but also by stimulating more investment spending. This result is broadly consistent with the empirical findings of Blanchard and Perotti, M-U, and others.

The size of fiscal multipliers depends on key country characteristics. As shown in Figures 3 and 4, the median tax-cut multipliers range up to a maximum of 2.0 across developing Asian economies in a 2-year horizon, whereas the median government expenditure multipliers range up to a maximum of around 1.0 in the same 2-year horizon. The median multipliers also turn out to be negative at different horizons over time. So the multipliers obtained here remain consistent with what has been found in the literature for other advanced countries, with some studies reporting multiplier values up to 4, although none of the studies calculated present value multipliers except M-U. The cumulative multiplier is the most appropriate measure, but it is typically larger than the impact or peak multipliers and hence it is rarely reported in empirical studies, with the exception of M-U. As cumulative multipliers can be bigger numbers, there is a need to calculate the present value of those cumulative responses as has been reported in this paper.

The present value of the GDP response to a deficit spending scenario is insignificant, whereas that for the deficit-financed tax cut is significantly positive for five countries, namely, Malaysia; the Philippines; Singapore; Taipei,China; and Thailand. For the remaining five countries, the impact is not significant, although the tax cut has a positive impact on output. The lack of significance in these countries could be due to data caveats, namely, very limited time dimension in some countries. This distinction becomes more obvious from the variance decomposition results of all the four shocks and what they account for in terms of proportional variation in GDP.7

7 Amjad and M. ud Din (2009) undertake a traditional “comparative-static” multiplier analysis to understand the impact of a decline in import prices on output via increase in import demand for intermediate inputs, and thereby rise in private investment and aggregate output. In contrast, the present paper specifically calculates “dynamic” fiscal multipliers to understand the relative effectiveness of different fiscal instruments in boosting economic activity for a broad group of 10 emerging Asian countries. Our results for the only common country, India, are consistent in the short run: government spending and tax cut have a positive impact on output, given that the Keynesian “comparative-static” multiplier approach provides only a short-run analysis.


C. Variance Decomposition Analysis

This section presents the results of an analysis of variance decomposition by finding the proportion of variance in GDP that is explained by different shocks across countries. This decomposition (into discretionary and cyclical) is an ex post exercise, decomposing one-step ahead prediction error into the part that is explained by its own policy shock and the (orthogonal) rest. This distinction cannot be made before identifying the shocks in the SVAR exercise. The variance decomposition for the VAR model is presented in Figure 5. In the variance decomposition analysis, nearly 50% of the variation in GDP is explained by business cycle shocks, and the remainder explained by monetary and fiscal shocks in several countries, except India, Indonesia, the Republic of Korea, the Philippines, and Singapore with higher proportions in GDP. The PRC; Hong Kong, China; Malaysia; Taipei,China; and Thailand have lower proportions of variation in GDP explained by business cycle shocks. The high output variance due to a business cycle shock could reflect higher output volatility due either to demand or supply shocks. While the high output variance could be supply-driven in India, Indonesia, and the Philippines as they rely on highly volatile agricultural sector, the output variance in the Republic of Korea and Singapore could be more demand-driven.

Figure 5: Decomposition of Variance (k=25) for GDP (percent)100

75

50

25

0

Business Cycle

Monetary Policy

GovernmentSpending

GovernmentRevenue

IND KOR SIN INO PHI MAL HKG TAP PRC THA

PRC = People‘s Republic of China; HKG = Hong Kong, China; IND = India; INO = Indonesia; KOR = Republic of Korea; MAL = Malaysia; PHI = Philippines; SIN = Singapore; TAP = Taipei,China; THA = Thailand.Note: We identified four shocks, when the number of variables in the VAR is eight. Even these four shocks still account for around

90% of the variation in GDP. If we identify eight shocks (same as the number of variables), then it will add up to 100%. We have not presented here the decomposition of other variables, but they are available upon request from the authors.



In most countries, notably the PRC; Indonesia; the Republic of Korea; the Philippines; Singapore; Taipei,China; and Thailand, government revenue shocks (due to a tax increase or tax cut) contribute more to output variance than spending shocks, as revenues rather than government expenditures have a significant negative impact on GDP. However, the share of variation in GDP is relatively less explained by fiscal shocks in India where government revenue as a percentage of GDP is lowest among the Asian emerging countries. We also note that fiscal shocks are more important in the PRC than India. In terms of contribution of the shocks to the variance of GDP, in Hong Kong, China, which is a small open economy, business cycle shocks account for 45.8% of the variation in GDP, followed by monetary policy shock explaining 10.8% of the variation in GDP, government expenditure for 16.6%, and revenue shock for 15.7% of the variation in GDP. Even for large economies, business cycle shocks account for a bigger proportion of variation in GDP followed by fiscal shocks, with monetary shocks explaining the lowest proportion of variation in GDP.

Our results confirm that business cycle shocks potentially explain the largest part of the variation in output, with revenue and spending shocks explaining less of the variation. As the business cycle shock accounts for a large variation in output in many countries, following a negative shock (for example, a downturn), output can contract. An expansionary monetary policy can help reverse the downturn, but when the monetary policy instrument is already at its lowest level, and when an economy has no independent monetary policy such as Hong Kong, China with its currency board type of exchange rate arrangement, or for a fixed regime such as in the PRC, fiscal shocks are critical for any possible recovery. But in the PRC; India; Indonesia; the Republic of Korea; and Taipei,China we find that positive government expenditure shocks do have a significant effect on output, whereas a negative revenue shock (a tax cut) produces a desired significant impact on increasing output in the remaining five countries.

The results consistently show that negative government revenue shocks (tax cuts) do have a significant positive multiplier effect on output, whereas positive spending shocks have a positive effect in many countries (although insignificant). The negative relationship between output (response) and positive revenue (shock) for many countries suggests that changes in tax rates may have created more revenue shocks, making output more volatile than spending shocks, as long-run responses show negative impact across six emerging Asian countries. That is probably the reason why the tax cut policy experiment produces a positive multiplier effect in many countries.

We also look at the panel of 10 Asian countries (Asia-10) over the time period 1996Q2–2008Q2 and carry out a panel VAR to assess the effect of government expenditure and revenue shocks (see Appendix). We find that government expenditure has a positive impact on output in the short run for Asia-10. Since a positive revenue shock has a negative impact on output in the long run even in the panel of Asia-10, a tax-cut hypothetical policy simulation can lead to increase in output as observed in individual country cases.


It is worth mentioning here that government expenditure and revenue shocks are defined as surprise increases in expenditure and revenues, respectively, which can be termed as discretionary fiscal policy changes. It is possible that the changes in government expenditures and revenues could be cyclical rather than purely driven by discretionary changes in government revenue and expenditure. For Asia-10, the variance decomposition suggests that a discretionary component of variation in government expenditure is around 55%, while the nondiscretionary component is around 45%. For revenues, the discretionary component is around 30%, and the nondiscretionary component is 70%. This decomposition of one-step ahead prediction error into the part that is explained by its own policy shock and the (orthogonal) rest suggests that a big part of the revenue shocks in developing Asia are cyclical rather than discretionary. Thus given the cyclical nature of government revenues, a tax cut policy may not be feasible to implement in developing Asian countries, although unanticipated deficit-financed tax cuts can work as a (short-lived) stimulus to the economy.

IV. Summary and Concluding Observations

In response to the global financial and economic crisis, governments throughout developing Asia have decisively implemented fiscal stimulus packages. There are signs that the region is rebounding. According to conventional wisdom, the sizable fiscal stimulus packages put into effect by governments throughout the region helped to kick-start the region’s struggling economies. The crisis, however, was not just another downturn in just another business cycle, but marked the deepest global recession and biggest contraction of global trade in the postwar era. As such, the stimulus measures put into effect by the region’s governments represented an extraordinary policy response to an extraordinary crisis, rather than a conventional countercyclical policy. Even though Asia’s fiscal stimulus packages have been tilted toward higher government spending, they have included tax cuts as well. In this context, as recovery takes hold in developing Asia, a timely and relevant issue is the relative effectiveness of tax cuts versus government spending in boosting aggregate demand and output.

Despite renewed interest in countercyclical fiscal policy, empirical literature on the issue is limited for developing economies, in particular, for Asia. Much of the literature is devoted to industrialized countries where there is a lively debate on this issue. The central objective of this paper is to empirically examine the impact of an unexpected fiscal shock on output in developing Asia while accounting for business cycle and monetary policy shocks. More specifically, we investigate the relative effectiveness of tax cuts versus government spending in boosting aggregate demand. We hope that our study can provide policymakers with some guidance about the optimal mix of fiscal policy as the region shifts from crisis mode back to normal mode.


To answer our central question, i.e., which of the two main forms of fiscal policy has been countercyclically more effective in Asia, we employ a two-stage empirical strategy. First, we examine the impacts of expansionary expenditure shocks, contractionary revenue shocks, and other nonfiscal shocks on output. Second, on the basis of our estimated responses from the first stage, we perform and compare two policy experiments—deficit-financed expenditure increase versus a deficit-financed tax cut—by calculating the present value of GDP multiplier effects. Our evidence suggests that across emerging Asian countries, tax cuts may be more effective for countercyclical purposes than higher spending. That is, unanticipated tax cuts stimulate economic activity more than public spending.

However, we should exercise a great deal of caution in interpreting our results as a general call for more active use of tax cuts. For one, once introduced, tax cuts may become entrenched and prove difficult to reverse or adjust. By and large Asia’s fiscal environment is characterized by a relatively small discretionary component of revenues, which implies that the scope for tax cuts may be limited in many countries. Furthermore, many countries in the region have weak tax bases, and the higher priority is to improve the tax revenue collection effort rather than to use tax cuts as a policy instrument for countercyclical purposes. In general, improvements in tax systems, enlargement of tax base, and rationalization of tax administration still remain relevant as ever. Such measures help to reduce the cost of compliance for taxpayers and generate larger fiscal space.

The relative ineffectiveness of government spending in increasing output suggests a clear need to improve the design of countercyclical spending to enhance its impact in the future. In particular, the composition of discretionary expenditure matters. In the short term, shifting expenditures to uses with larger multipliers and reducing misallocation, e.g., by targeting subsidies to the poor, will contribute to greater countercyclical effectiveness. In the long term, shifting expenditures toward growth-conducive public goods such as infrastructure and social protection will help create the physical and human capital required for growth.

In many countries, another way to enhance the impact of discretionary fiscal measures is through effective utilization of scarce fiscal resources. Minimizing waste and inefficiency in public programs will help reduce unproductive expenditures, whereas inefficient use of public funds, combined with pilferage, diversion, and leakage may seriously impair the effectiveness of public spending. More efficient use of public funds can help the region better meet fiscal demands associated with population ageing, rebalancing, and other emerging structural challenges.

Unlike discretionary public spending, automatic stabilizers have the advantages of speed, predictability, and reversibility. Once instituted, these measures are outside the political process and automatically kick in when the economy is on a downturn. Automatic stabilizers are especially valuable when fiscal institutions are weak, as in many developing economies. However, most developing countries in Asia have weak automatic stabilizers. Strengthening them will provide the region with an insurance mechanism against another economic crisis.


AppendixAppendix Figure 1: Panel VAR Responses on the Impact of Government Expenditure Shock, 1996Q2–2008Q2 (percent)

1050 15 20

1050 15 20

1050 15 20

1050 15 20 1050 15 20

1050 15 20

1050 15 20

1050 15 20−0.004

−0.002

0.000

0.002

0.004

0.006

0.008

0.010

0.012

−0.05

−0.03

−0.01

0.01

0.03

−0.04

−0.02

0.00

0.02

0.04

0.06

Real GDP

Government Expenditure

Government Revenue

Interest Rate

−0.005

−0.004

−0.003

−0.002

−0.001

0.000

0.001

0.002

0.003

GDP De�ator

−0.030

−0.020

−0.010

0.000

0.010

Real Money

−0.010

−0.005

0.000

0.005

0.010

0.015

0.020

Private Consumption

−0.0025

0.0000

0.0025

0.0050

0.0075

Private Investment

−0.04

−0.02

0.00

0.02

0.04

0.06

0.08

VAR = vector autoregression; GDP = gross domestic product.Source: Authors‘ calculations.


Appendix Figure 2: Panel VAR Responses on the Impact of Government Revenue Shock, 1996Q2–2008Q2 (percent)

Real GDP

10 15 20−0.0075

−0.0050−0.0025

0.00000.0025

0.00500.0075

0.01000.0125

Government Expenditure

−0.050

−0.025

0.000

0.025

0.050

Government Revenue

−0.020−0.015

−0.010−0.005

0.0000.005

0.0100.015

0.020

Interest Rate

−0.002

−0.001

0.000

0.001

0.002

0.003

GDP De�ator

0 5 10 15 20−0.020

−0.010

0.000

0.010

0.020

Real Money

0 5 10 15 20−0.006

−0.004

−0.002

0.000

0.002

0.004

0.006

0.008

Private Consumption

0 5 10 15 20−0.006

−0.004

−0.002

0.000

0.002

0.004

0.006

Private Investment

0 5 10 15 20−0.04

−0.02

0.00

0.02

0.04

0 5

10 15 200 5

10 15 200 5

10 15 200 5

VAR = vector autoregression; GDP = gross domestic product.Source: Authors‘ calculations.


ReferencesAlesina, A., and S. Ardagna. 2009. Large Changes in Fiscal Policy: Taxes Versus Spending. NBER

Working Paper 15438, National Bureau of Economic Research, Cambridge.Amjad, R., and M. ud Din. 2009. “Economic and Social Impact of Global Financial Crisis:

Implications for Macroeconomic and Development Policies in South Asia.” Pakistan Institute of Development Economics, Islamabad. Processed.

Baldacci, E., S. Gupta, and C. Mulas-Granados. 2009. How Effective is Fiscal Policy Response in Systemic Banking Crises? IMF Working Paper WP/09/160, International Monetary Fund, Washington, DC.

Blanchard, O., and R. Perotti. 2002. “An Empirical Characterization of the Dynamic Effects of Changes in Government Spending and Taxes on Output.” Quarterly Journal of Economics 117(4):1329–68.

Burnside, C., M. Eichenbaum, and J. Fisher. 2003. “Fiscal Shocks and Their Consequences.” Journal of Economic Theory 115(1):89–117.

Burriel, P., F. de Castro, D. Garrote, E. Gordo, J. Paredes, and J. J. Pérez. 2009. Fiscal Policy Shocks in the Euro Area and the US: An Empirical Assessment. Working Paper Series No. 1133, European Central Bank, Frankfurt.

Caldara, D., and C. Kamps. 2008. What are the Effects of Fiscal Shocks? A VAR-based Comparative Analysis. Working Paper Series No. 877, European Central Bank, Frankfurt.

Cogan, J. F., T. Cwik, J. B. Taylor, and V. Wieland. 2010. “New Keynesian versus Old Keynesian Government Spending Multipliers.” Journal of Economic Dynamics and Control 34(3):281–95.

Easterly, W., and S. Rebelo. 1993. “Fiscal Policy and Economic Growth.” Journal of Monetary Economics 32(3):417–58.

Eskesen, L. B. 2009a. Countering the Cycle—The Effectiveness of Fiscal Policy in Korea. IMF Working Paper No. 09/249, International Monetary Fund, Washington, DC.

———. 2009b. The Role for Counter-Cyclical Fiscal Policy in Singapore. IMF Working Paper No. 09/8, International Monetary Fund, Washington, DC.

Galí, J., J. D. López-Salido, and J. Vallés. 2007. “Understanding the Effects of Government Spending on Consumption.” Journal of the European Economic Association 5(1):227–70.

Hemming, R., M. Kell, and S. Mahfouz. 2002. The Effectiveness of Fiscal Policy in Stimulating Economic Activity—A Review of the Literature. IMF Working Paper No. WP/02/208, International Monetary Fund, Washington, DC.

Linnemann, L., and A. Schabert. 2006. “Productive Government Expenditure in Monetary Business Cycle Models.” Scottish Journal of Political Economy 53(1):28–46.

Mountford, A., and H. Uhlig. 2009. “What are the Effects of Fiscal Policy Shocks?” Journal of Applied Econometrics 24(6):960–92.

Nijkamp, P., and J. Poot. 2004. “Meta-analysis of the Effect of Fiscal Policies on Long-run Growth.” European Journal of Political Economy 20(1):91–124.

Pappa, E. 2009. “The Effects of Fiscal Shocks on Employment and the Real Wage.” International Economic Review 50:217–44.

Perotti, R. 1999. “Fiscal Policy in Good Times and Bad.” Quarterly Journal of Economics 114(4):1399–436.

Ramey, V. 2009. Identifying Government Spending Shocks: It's All in the Timing. NBER Working Paper 15464, National Bureau of Economic Research, Cambridge.

Romer, C., and J. Bernstein. 2009. “The Job Impact of the American Recovery and Reinvestment Plan.” Available: otrans.3cdn.net/45593e8ecbd339d074_l3m6bt1te.pdf.

Spilimbergo, A., S. Symansky, and M. Schindler. 2009. Fiscal Multipliers. Staff Position Note SPN/09/11, International Monetary Fund, Washington, DC.


About the PaperShikha Jha, Sushanta Mallick, Donghyun Park, and Pilipinas Quising use historical data from 10 developing Asian countries and find evidence suggesting tax cuts are more effective countercyclical policy instruments than government spending. However, Asian governments should be cautious about actively using tax cuts for countercyclical purposes, partly because a big part of the revenue shocks in the region are nondiscretionary rather than discretionary.

About the Asian Development BankADB’s vision is an Asia and Pacific region free of poverty. Its mission is to help its developing member countries substantially reduce poverty and improve the quality of life of their people. Despite the region’s many successes, it remains home to two-thirds of the world’s poor: 1.8 billion people who live on less than $2 a day, with 903 million struggling on less than $1.25 a day. ADB is committed to reducing poverty through inclusive economic growth, environmentally sustainable growth, and regional integration. Based in Manila, ADB is owned by 67 members, including 48 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 Bank6 ADB Avenue, Mandaluyong City1550 Metro Manila, Philippineswww.adb.org/economicsISSN: 1655-5252Publication Stock No. WPS102320 Printed in the Philippines

< 0 1 0 2 3 2 0 6 >

adb economics working paper series · the adb economics working paper series is a quick ......

Documents