ADB - Effectiveness of Counter Cyclical Fiscal Policy - Time-Series Evidence From Developing Asia

8/4/2019 ADB - Effectiveness of Counter Cyclical Fiscal Policy - Time-Series Evidence From Developing Asia

1/34

ADB EconomicsWorking Paper Series

Eectiveness o Countercyclical Fiscal Policy:Time-Series Evidence rom Developing Asia

Shikha Jha, Sushanta Mallick, Donghyun Park,

and Pilipinas Quising

No. 211 | August 2010


2/34


3/34

ADB Economics Working Paper Series No. 211

Eectiveness o Countercyclical Fiscal Policy:

Time-Series Evidence rom Developing Asia

Shikha Jha, Sushanta Mallick, Donghyun Park,

and Pilipinas QuisingAugust 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 Ofcer, Macroeconomics and Finance Research Division, Economics and

Research Department, ADB. This paper was presented at workshops in ADB. It has beneted 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.


4/34

Asian Development Bank

6 ADB Avenue, Mandaluyong City

1550 Metro Manila, Philippines

www.adb.org/economics

2010 by Asian Development BankAugust 2010

ISSN 1655-5252

Publication Stock No. WPS102320

The views expressed in this paper

are those of the author(s) and do not

necessarily reect the views or policies

of 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 Pacic 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 Asias development

and policy challenges; strengthen analytical rigor and quality of ADBs 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.


5/34

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


6/34


7/34

Abstract

As the global crisis hit developing Asia, several countries instituted scal 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 scal policy can still be used to

stimulate growth. To do so, we use a sign-restrictions-based structural vector

autoregression model. We nd that expansionary expenditure shocks have an

insignicant 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: decit-nanced tax cuts and decit spending. The experimentresults indicate that while decit-nanced tax cuts stimulate economic activity,

the impact of decit 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.


8/34


9/34

I. Introduction

The severe global crisis in 20082009 led to a deep contraction in developing Asias

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

inationary pressures subsided, governments in several Asian developing countries

implemented expansionary monetary policy. This helped stabilize the nancial sectorbut 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 scal policy by acting as the consumer of

last resort. This was facilitated by the relatively healthy state of public nances in Asia

vis--vis industrialized countries. Such proactive use of monetary and scal policy for

countercyclical purposes marks a signicant 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 signicant stress

of the crisis has dissipated, the question remains: Can scal 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 scal 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 governments 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 rms 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

scal policy enhances or retards long-run economic activity has long been debated in


10/34

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 measureshow 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 scal policy as a tool for ghting recessions. In comparison to the

very large empirical literature on scal 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 scal policy in the process of economic development can be

traced back to Easterly and Rebelo (1993). Since then a large number of studies haveconsidered the effects of scal 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 scal policy. See,

for example, Blanchard and Perrotti (2002); Burnside, Eichenbaum, and Fisher (2003);

Gal, Lpez-Salido, and Valls (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 scal 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 11.5 in large countries; 0.51 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

nd scal expansions based on government consumption to be more effective than those

1 Among others, two contrasting views come rom the basic Keynesian and Ricardian theories. In the

simple Keynesian world, where aggregate demand determines output given rigid prices, and whereconsumption responds to current income, scal expansion has a multiplier eect on growth. In

contrast, scal multiplier is zero under Ricardian equivalence between taxes and debt in a dynamic

ramework. In this case, Ricardian consumers are orward-looking and ully aware o the governments

intertemporal budget constraint. Since they know that a tax cut today will be nanced by higher taxes

in the uture, their consumption does not change because permanent income is unaected. Similarly,

the knowledge that an increase in government spending by borrowing today will be oset by uture

spending cuts leaves output unaected.

2 | ADB Economics Working Paper Series No. 211


11/34

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 thatgovernment spending has a bigger impact on output than tax cuts.2 Romer and Romer

(2009) nd 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, nd that decit-nanced tax

cuts have a bigger effect than decit-nanced spending or tax-nanced spending. In

a comprehensive analysis of 91 episodes of scal expansions in 21 OECD countriessince 1970, Alesina and Ardagna (2009) compared expansions that brought about output

growth with those that failed to do so. Strikingly, successful scal stimulus programs were

based almost entirely on tax cuts, whereas unsuccessful programs were based mostly

on additional government spending. Blanchard and Perotti (2002) nd that higher taxes

and higher government spending have a strong negative effect on private investment.

Similarly, M-U nd 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 nd 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 bylower private investment. Perotti (1999) also nds that government revenue shocks have

very different effects on private consumption in times of scal stress than in normal

times. In this context, for the US, Cogan et al. (2010) nd 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 scal policy in stimulating economic growth, note that given data

deciencies and institutional weaknesses, there is no clear conclusion on the sizeand sign of scal 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


12/34

spending, there is only limited evidence about the issue in the Asia and Pacic region.

To some extent, this simply mirrors the fact that industrialized countries have a longer

tradition of using scal 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 benets are an integral partof scal policy in the G3 but an underdeveloped novelty in much of the Asian region.

There is no study examining the issue specically in the context of developing Asia. In

this paper we attempt to ll 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 regions policymakers some guidance on how

to maximize the impact of scarce scal resources.

Based on econometric analysis using historical time-series data, this paper analyzes

what type of scal policytax cuts versus higher spendingworks 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 versusgovernment spending in boosting output in developing Asia and describes the data used.

Section IIIreports and discusses the main ndings emerging from the analysis. Section

IV reviews the central messages of the paper along with their implications for the regions

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 scal shocks in the data alongside

other shocks by imposing sign restrictions for the identication of each shock. Explicit

identifying sign-restrictions are imposed to isolate exogenous and unanticipated changes

in these variables by restricting the contemporaneous interaction of scal and nonscal

variables. The paper uses M-Us penalty function approach, which has the advantage of

only picking up large shocks and thereby reducing the variation in the identied 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-Us methodology, the VAR does not contain a constant or a time trend. But including



13/34

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 onthe 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 signicant, and which type of policy (decit-nanced tax cut or government spending)

has a dominant effect.

This methodology has some salient features. It distinguishes between unanticipated

scal shocks (news shocks) and automatic responses of scal variables to business

cycle conditions, e.g., automatic stabilizers such as unemployment benets and socialsafety nets that automatically kick in during downturns unlike discretionary scal policy.

The model also takes account of the announcement effect, namely the lag between

implementation and announcement of scal policy that may affect macro variables.

For instance, rms 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,

scal policy, and monetary policy variables. The basic intuition is that structural shocks

can be identied by checking whether the signs of the corresponding impulse responses

are in line with theoretical priors. Three structural shocks are identied by imposing sign

restrictions: (i) business cycle shocks; (ii) monetary shocks; and (iii) scal shocks (i.e.,

revenue shocks and government spending shocks). The sign restrictions help identify theeffects of unanticipated scal shocks and nonscal 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 deator

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 interestGDP,

consumption, and investmentto scal policy shocks. All variables are estimated with



14/34

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 unidentied. The horizon kof the imposed restrictions is four quarters. TheVAR model and its residuals can be represented as follows.

Table 1: Identiying Sign Restrictions

Nonscal and Fiscal

Policy Shocks

Real

GDP

Real

Government

Expenditure

Real

Government

Revenue

Policy

Rate

Real

Money

GDP

Deator

Real

Consumption

Real

Investment

Business cycle shock

(growth)+ ? + ? ? ? + +

Monetary policy shock

(tightening)? ? ? + ? ?

Government revenue

shock? ? + ? ? ? ? ?

Governmentexpenditure shock

? + ? ? ? ? ? ?

Note: The sign restrictions on the impulse responses or each identied shock are shown as + (positive response or our quarters

ollowing the shock); (negative response or our quarters ollowing the shock); and ? (no restriction).

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

Y A Y u

Y I A L u B L u

t t t

t t t

= +

= ( ) =

1 1

1

1( )

(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 Pis the unique lower triangularmatrix such that PP'=u. 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 satised. It will then be possible to derive

a new matrix C = PL, whose columns are the identied impulse vectors.

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

u C v C v C v C v C v t t

BC

t

MP

t

GR

t

GE

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 eachstructural shock, the associated impulse vector is Ci. As there is a lesser number of

shocks than the number of variables in the VAR, the unidentied shocks in vector v are

associated with the remaining columns of the matrix C . 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.



15/34

Following a business cycle shock (output growth/contraction) that should be identied

rst, a monetary shock should be identied next, wherein the monetary authority is

more likely to change its policy stance. Finally the scal authority introduces either

a spending rise or tax cut (a scal shock). Hence the impacts of these shocks are

checked sequentially to investigate whether scal policy shocks do help an economyto increase its output. From these effects, one can conclude in which countries scal

policy will be more effective. A scal shock is a surprise change in scal 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 scal shocks, we also

impose no restrictions on the variables of interest: GDP, consumption, investment, and

ination. 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 economiesthe Peoples 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. o 58 1995:12009:2

Hong Kong, China 68 1992:32009:2

India 53 1996:22009:2

Indonesia 66 1993:12009:2

Korea, Rep. o 74 1991:12009:2

Malaysia 74 1991:12009:2

Philippines 98 1985:12009:2

Singapore 86 1988:12009:2

Taipei,China 128 1977:32009:2

Thailand 66 1993:12009:2

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 deator is derived as (nominal GDP/real

GDP), which is used as price series for all countries.



16/34

(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 denition

of policy rate, however, differs as follows: the PRC: 1-year lending rate;

Hong Kong, China: discount rate; India: repo rate; Indonesia: Sertikat

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 xed investment are taken from

CEIC. Wherever it is available in nominal terms, we have deated the

series, using GDP deator 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 GDPof 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 deated by the GDP deator in order to be

expressed in real terms. We have converted annual scal 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 deated by the GDP deator to get realmoney 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

identication methodology allows for a very similar identication to be achieved across

countries despite these data problems. This is because the M-U identication strategy

identies shocks using mild restrictions on multiple time series in all economies

regardless of how they are measured and regardless of data idiosyncrasies, hence thedata denition used in an individual country for a particular variable is of secondary

importance.



17/34

III. Results rom the Empirical Analysis

The identication 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 theSVAR 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 identied 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 scal shock by imposing weaker identifying restrictions, such as

restricting government spending alone to be positive on impact, such identication will

hide the presence of other shocks at the same time, and may not uniquely identify the

impact of a scal shock.

Besides, the SVAR literature is primarily focused on assessing the effect of contractionary

monetary shocks, with little emphasis on scal 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 scal

contraction. The impacts of scal policy shocks would be inuenced by the presence of

nonscal 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. Identication 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 scal policies. The short-

run responses for business cycle shock in Table 3 satisfy the sign restrictions for

k= 1,...,Kquarters. In our analysis, the impulse responses of scal and nonscal

variables have been restricted for the rst four quarters following the shock.3 But the

signicant 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 prole o the eect o a shock (or impulse) on the (expected) uture values o

a variable. Short run is dened as our quarters over which the sign restrictions are imposed. Long-run response is

calculated as the sum o the coecients o the lagged variables in the VAR.



18/34

Table 3: Impact Efects o Diferent Shocks on GDP rom the Sign-VAR

Shocks in China,

Rep. o

Hong Kong,

China

India Indonesia Korea,

Rep. o

Malaysia

Business cycle +* +* +* +* +* +*

Monetary policy +* Government spending + + + +

Government revenue + * + *

Philippines Singapore Taipei,China Thailand

Business cycle +* +* +* +*

Monetary policy + +

Government spending +* +*

Government revenue * +* +

Note: * Indicates condence interval (16th and 84th percentiles) being signicantly dierent rom zero.

Source: Authors calculations.

A. Impulse Responses to Fiscal Policy Shocks

Applying the nonlinear method, which uses Monte-Carlo integration, we nd 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

decit-nanced expenditure stimulus is possible, the eventual costs are likely to be

much higher than the immediate benets, as the increased decit 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 scal 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 prole o the eect o a shock (or impulse) on the (expected) uture values o

a variable. Short run is dened as our quarters over which the sign restrictions are imposed. Long-run response is

calculated as the sum o the coecients o the lagged variables in the VAR.



19/34

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 signicant effect,

whereas a positive government expenditure shock can be more effective in terms of itsimpact on output. This is contrary to the result in M-U for the US.

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

PRC HKG IND INO

Immediate Impact

Long-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 = Peoples Republic o China; HKG = Hong Kong, China; IND = India; INO = Indonesia;

KOR = Republic o Korea; MAL = Malaysia; PHI = Philippines; SIN = Singapore;

TAP = Taipei,China; THA = Thailand.


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

Immediate Impact

Long-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



TAP = Taipei,China; THA = Thailand.




20/34

Higher taxes have a signicant 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 signicant in the long run, it is crowded out to some extent

by investment. Taipei,China shows the counterintuitive positive long-run growth effect ofhigher taxes, which is even enhanced through higher consumption and investment. The

apparently counterintuitive result for Singapore and Taipei,China may be depicting a case

ofexpansionary scal contraction. That is, a scal 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 scal 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 signicant

leakages of scal stimulus abroad through trade as well as remittances. The result is

insignicant 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 insignicant. Or, it increases ination, 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 insignicant. Some of the effects are negative,

though insignicant. They might imply that cuts in unproductive spending, especially

transfers, result in expansionary scal contractions. Given the insignicant coefcientof spending shocks on output in several countries, one could argue that scal policy has

had little destabilizing impact on real output and could reect a pro-cyclical nature of

scal policy from the spending side, whereas scal policy appears countercyclical from

the revenue side.

Government spending shocks increase ination in India; Indonesia; the Republic of Korea;

Singapore; and Taipei,China, whereas in the remaining ve countries, spending shocks

do not have any inationary impact. Any inationary 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 andpublic nance does not solely rely on distortionary taxation, whereas when government

expenditures are partially nanced by public debt, unit labor costs fall in response to a

scal expansion, such that ination tends to decline.

5 Eskesen (2009a and 2009b) conrm a case or countercyclical scal policy in the Republic o Korea and Singapore.6 Given data on real wages, one could explore this urther or developing Asian countries.



21/34

Table4:ImpulseR

esponsestoPositiveTaxReven

ueShock(%)

Shockson

Impactsin

Real

GDP

Government

Expenditure

Government

Revenue

Interest

Rate

GDP

Defator

Real

Money

Private

Consumption

Fixed

Investment

China,

Peoples

Rep.o

SR

0.0

171

0.0

399

0.0

386*

0.0

136

0.0

035

0.0

006

0.0

101

0.0

163

LR

0.0

030

0.0

634

0.0

372*

0.0

248

0.0

023

0.0

029

0.0

042

0.0

260

HongKong,

China

SR

0.0

023

0.0

028

0.0

723*

0.0

002

0.0

023

0.0

028

0.0

025

0.0

165

LR

0.0

178

0.0

616

0.2

195*

0.0

183*

0.0

708*

0.0

442

0.0

314

0.0

067

India

SR

0.0

017

0.0

170

0.0

698*

0.0

016

0.0

039

0.0

021

0.0

044

0.0

280

LR

0.0

185

0.0

086

0.1

015*

0.0

083

0.0

107

0.0

223

0.0

117

0.1

897

Indonesia

SR

0.0

064*

0.0

259

0.0

313*

0.0

030

0.0

136*

0.0

120

0.0

060

0.0

016

LR

0.0

038

0.1

312

0.0

474*

0.0

004

0.5

296*

0.0

371

0.0

269

0.0

355

Korea,

Rep.o

SR

0.0

006

0.0

416

0.0

643*

0.0

002

0.0

013

0.0

018

0.0

003

0.0

267

LR

0.0

076

0.0

619

0.0

456*

0.0

018

0.0

083

0.0

063

0.0

069

0.0

438

Malaysia

SR

0.0

070*

0.0

164

0.0

409*

0.0

013

0.0

061

0.0

048

0.0

039

0.0

014

LR

0.0

087

0.0

206

0.0

472

0.0

012

0.0

112

0.0

080

0.0

096

0.0

063

Philippines

SR

0.0

119*

0.0

109

0.0

345*

0.0

008

0.0

025

0.0

010

0.0

021

0.0

453*

LR

0.0

309*

0.0

243

0.1

081*

0.0

258*

0.0

859*

0.1

164*

0.0

088

0.0

682*

Singapore

SR

0.0

063*

0.0

021

0.0

502*

0.0

014*

0.0

052*

0.0

103*

0.0

066*

0.0

387

LR

0.0

529*

0.4

124

*

0.4

135*

0.0

050

0.0

353*

0.1

147*

0.0

139

0.0

487

Taipei,China

SR

0.0

021

0.0

464

*

0.0

517*

0.0

008

0.0

003

0.0

017

0.0

001

0.0

136*

LR

0.1

222*

0.3

508

*

0.2

236*

0.0

132*

0.1

389*

0.2

212*

0.2

182*

0.3

846*

Thailand

SR

0.0

035

0.0

022

0.0

150*

0.0

017

0.0

063

0.0

490

0.0

034

0.0

284

LR

0.0

076

0.0

127

0.0

444

0.0

024

0.0

098

0.0

821

0.0

097

0.0

587

*Indicatestheimpactbeingsignicantlydiferentromz

ero(bothupp

er[84thpercentile]andlower[16thpercentile]bandsaresignicantlydiferentromt

hez

eroline).

SR=shortrun;LR=long

run.

Source:Authorscalculat

ions.



22/34

Table5:ImpulseR

esponsestoPositivePublicSpendingShock

Shockson

Impactsin

Real

GDP

Government

Expenditure

Government

Revenue

Interest

Rate

GDP

Defator

Real

Money

Private

Consumption

Fixed

Investment

China,

Peoples

Rep.o

SR

0.0

065

0.0

322*

0.0

211

0.0

549

0.0

186*

0.0

110*

0.0

038

0.0

049

LR

0.0

100

0.0

245

0.0

011

0.4

607

0.0

592*

0.0

244

0.0

053

0.0

059

HongKong,

China

SR

0.0

015

0.0

793*

0.0

727

0.0

001

0.0

033

0.0

023

0.0

006

0.0

049

LR

0.0

174*

0.1

835*

0.2

284*

0.0

299*

0.1

041*

0.0

435*

0.0

344*

0.0

331

India

SR

0.0

027

0.0

365*

0.0

022

0.0

018*

0.0

108

0.0

052

0.0

013

0.2

229*

LR

0.0

526

0.0

724*

0.0

184

0.0

122*

0.0

072

0.0

952*

0.0

237

0.8

518*

Indonesia

SR

0.0

004

0.0

074*

0.0

248

0.0

021

0.0

163*

0.0

201*

0.0

054

0.0

059

LR

0.0

018

0.1

886*

0.0

068

0.0

043

0.6

758*

0.0

555*

0.0

266

0.0

235

Korea,

Rep.o

SR

0.0

086

0.0

798*

0.0

500*

0.0

010

0.0

015

0.0

016

0.0

004

0.0

185

LR

0.0

083

0.0

798*

0.0

320

0.0

012

0.0

058

0.0

231

0.0

169

0.0

673

Malaysia

SR

0.0

023

0.0

917*

0.0

125

0.0

027*

0.0

003

0.0

007

0.0

001

0.0

192*

LR

0.0

098

0.2

621*

0.0

223

0.0

138

0.0

539*

0.0

660

0.0

546*

0.0

257

Philippines

SR

0.0

053*

0.0

709*

0.0

110

0.0

003

0.0

046

0.0

072

0.0

002

0.0

274*

LR

0.0

113

0.1

104*

0.0

600

0.0

095

0.0

727

0.0

019

0.0

140*

0.0

743*

Singapore

SR

0.0

057*

0.1

539*

0.0

191

0.0

001

0.0

016

0.0

002

0.0

006

0.0

912

LR

0.0

230

0.2

883*

0.0

496

0.0

068

0.0

318

0.0

877*

0.0

268*

0.2

211

Taipei,China

SR

0.0

017

0.0

709*

0.0

449*

0.0

004

0.0

019*

0.0

037*

0.0

014

0.0

150*

LR

0.0

921*

0.3

527*

0.2

092*

0.0

134*

0.1

061*

0.1

580*

0.1

879*

0.3

369*

Thailand

SR

0.0

017

0.0

776*

0.0

029

0.0

029

0.0

039

0.0

361

0.0

042*

0.0

587*

LR

0.0

577*

0.0

114

0.1

169*

0.0

120*

0.0

212

0.9

564

0.0

470*

0.4

033*

*Indicatestheimpactbeingsignicantlydiferentromz

ero(bothupp

er[84thpercentile]andlower[16thpercentile]bandsaresignicantlydiferentromt

hez

eroline).

SR=shortrun;LR=long

run.

Source:Authorscalculations.



23/34

B. Cumulative Output Multipliers

Having observed the impact effects, we next derive cumulative output multipliers by

undertaking two key policy experiments: a decit-nanced tax cut policy scenario and

a decit-spending scal policy scenario. Calculation of the present value multiplieruses 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 o Present Value o the Multiplier

RIR Revenue

(percent o GDP)

Expenditure

(percent o GDP)

China, People's Rep. o 3.86 17.67 17.95

Hong Kong, China 2.76 16.05 15.26

India 2.43 10.72 15.65

Indonesia 1.24 17.61 18.59

Korea, Rep. o 2.21 23.11 22.04Malaysia 1.24 21.15 25.58

Philippines 1.83 15.49 18.57

Singapore 1.01 23.31 16.70

Taipei,China 2.06 19.15 21.63

Thailand 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 ollows: r=(i-pe)/(1+pe) where rdenotes the RIR, ithe nominal interest rate, and pe is the expected infation rate.

We use the current CPI infation as a proxy or pe and the policy rate as a measure o i.


Figures 3 and 4 display the present value GDP multipliers for both the experiments. Wecalculate 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 condence 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 signicantly

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 signicantly different from

zero. The multiplier represents the effect of a 1% cut in taxes or a 1% increase in public

spending in the rst quarter. This is calculated with the following formula:

Multiplier for GDP

rGDP response

rIni

j jj

k

j

=

+( )( )

+( )

=1

1

1

1

0

ttial fiscal shock

Average fiscal

jj

k

( )

=

0

var iiable share of GDPas( )(3)

where ris real interest rate or the discount rate, and k= 0,,4 periods.



24/34

Figure 3: Impulse Responses with Penalty Function Approach

Multiplier Government Revenue (percent)

6

4

2

02

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

02

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, Peoples Rep. of

Note: The present value multiplier has been calculated or the GDP multiplier eect due to decit nanced tax cuts. The error

bands are illustrated as the dotted lines above and below the middle response line (the thick line), which are composed o

the 16th (below), 84th (above) and median percentiles (middle) o the responses.




25/34

Figure 4: Impulse Responses with Penalty Function Approach

Multiplier Government Expenditure(percent)China, Peoples Rep. of

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

Quarters after the Shock

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


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

India

Indonesia

Korea, Rep. of

Philippines

Malaysia

Singapore

Taipei,China

Thailand

3

2

1

0

1

2

3

4

5

4

3

2

1

0

1

2

6

4

2

0

2

4

6

8

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 or the GDP multiplier eect due to decit nanced expenditure increase.

The error bands are illustrated as the dotted lines above and below the middle response line (the thick line), which are

composed o the 16th (below), 84th (above) and median percentiles (middle) o the responses.




26/34

We showed earlier that scal 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 scal instruments, we have calculated the

multipliers using the above formula. The cumulative multiplier is dened as the cumulative

change in output over the cumulative change in scal 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 ndings of

Blanchard and Perotti, M-U, and others.

The size of scal 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 themultipliers 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 decit spending scenario is insignicant,

whereas that for the decit-nanced tax cut is signicantly positive for ve countries,

namely, Malaysia; the Philippines; Singapore; Taipei,China; and Thailand. For theremaining ve countries, the impact is not signicant, although the tax cut has a positive

impact on output. The lack of signicance 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 tounderstand the impact o a decline in import prices on output via increase in import demand or

intermediate inputs, and thereby rise in private investment and aggregate output. In contrast, the

present paper specically calculates dynamic scal multipliers to understand the relative eectiveness

o dierent scal instruments in boosting economic activity or a broad group o 10 emerging Asian

countries. Our results or 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.



27/34

C. Variance Decomposition Analysis

This section presents the results of an analysis of variance decomposition by nding

the proportion of variance in GDP that is explained by different shocks across countries.

This decomposition (into discretionary and cyclical) is an ex postexercise, decomposingone-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 scal 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

reect 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 theyrely on highly volatile agricultural sector, the output variance in the Republic of Korea and

Singapore could be more demand-driven.

Figure 5: Decomposition o Variance (k=25) or GDP (percent)

100

75

50

25

0

Business

Cycle

Monetary

Policy

Government

Spending

Government

Revenue

IND KOR SIN INO PHI MAL HKG TAP PRC THA



TAP = Taipei,China; THA = Thailand.Note: We identied our shocks, when the number o variables in the VAR is eight. Even these our shocks still account or around

90% o the variation in GDP. I we identiy eight shocks (same as the number o variables), then it will add up to 100%. We

have not presented here the decomposition o other variables, but they are available upon request rom the authors.




28/34

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 signicant negative impact on

GDP. However, the share of variation in GDP is relatively less explained by scal shocksin India where government revenue as a percentage of GDP is lowest among the Asian

emerging countries. We also note that scal 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 scal shocks, with monetary shocks explaining the lowest

proportion of variation in GDP.

Our results conrm that business cycle shocks potentially explain the largest partof 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 xed regime such as in the PRC, scal shocks are critical

for any possible recovery. But in the PRC; India; Indonesia; the Republic of Korea; and

Taipei,China we nd that positive government expenditure shocks do have a signicant

effect on output, whereas a negative revenue shock (a tax cut) produces a desired

signicant impact on increasing output in the remaining ve countries.

The results consistently show that negative government revenue shocks (tax cuts) do

have a signicant positive multiplier effect on output, whereas positive spending shocks

have a positive effect in many countries (although insignicant). 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 1996Q22008Q2 and carry out a panel VAR to assess the effect of government expenditure

and revenue shocks (see Appendix). We nd 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.



29/34

It is worth mentioning here that government expenditure and revenue shocks are

dened as surprise increases in expenditure and revenues, respectively, which can

be termed as discretionary scal 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 variancedecomposition 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 decit-nanced tax cuts

can work as a (short-lived) stimulus to the economy.

IV. Summary and Concluding Observations

In response to the global nancial and economic crisis, governments throughout

developing Asia have decisively implemented scal stimulus packages. There are signs

that the region is rebounding. According to conventional wisdom, the sizable scal

stimulus packages put into effect by governments throughout the region helped to

kick-start the regions 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 regions governments represented an extraordinary policy response

to an extraordinary crisis, rather than a conventional countercyclical policy. Even though

Asias scal 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 scal 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 scal shockon output in developing Asia while accounting for business cycle and monetary policy

shocks. More specically, 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 scal policy as the region

shifts from crisis mode back to normal mode.



30/34

To answer our central question, i.e., which of the two main forms of scal 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 nonscal shocks on output. Second, on the basis of our estimated

responses from the rst stage, we perform and compare two policy experimentsdecit-nanced expenditure increase versus a decit-nanced tax cutby 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 difcult to reverse or adjust. By and large Asias scal

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

scal 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 scal measures is

through effective utilization of scarce scal resources. Minimizing waste and inefciency

in public programs will help reduce unproductive expenditures, whereas inefcient use of

public funds, combined with pilferage, diversion, and leakage may seriously impair the

effectiveness of public spending. More efcient use of public funds can help the region

better meet scal demands associated with population ageing, rebalancing, and otheremerging 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 scal 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.



31/34

Appendix

Appendix Figure 1: Panel VAR Responses on the Impact o Government Expenditure

Shock, 1996Q22008Q2 (percent)

1050 15 20

1050 15 20

1050 15 20

1050 15 20 1050 15 20

1050 15 20

1050 15 20

1050 15 200.004

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 Deator

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.




32/34

Appendix Figure 2: Panel VAR Responses on the Impact o Government Revenue Shock,

1996Q22008Q2 (percent)

Real GDP

10 15 200.0075

0.0050

0.0025

0.0000

0.0025

0.0050

0.0075

0.0100

0.0125

Government Expenditure

0.050

0.025

0.000

0.025

0.050

Government Revenue

0.020

0.015

0.010

0.005

0.000

0.005

0.010

0.015

0.020

Interest Rate

0.002

0.001

0.000

0.001

0.002

0.003

GDP Deator

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

0.0020.000

0.002

0.004

0.006

0.008

Private Consumption

0 5 10 15 200.006

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.




33/34

Reerences

Alesina, 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):132968.

Burnside, C., M. Eichenbaum, and J. Fisher. 2003. Fiscal Shocks and Their Consequences.

Journal of Economic Theory115(1):89117.

Burriel, P., F. de Castro, D. Garrote, E. Gordo, J. Paredes, and J. J. Prez. 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 Control34(3):28195.Easterly, W., and S. Rebelo. 1993. Fiscal Policy and Economic Growth. Journal of Monetary

Economics 32(3):41758.

Eskesen, L. B. 2009a. Countering the CycleThe 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.

Gal, J., J. D. Lpez-Salido, and J. Valls. 2007. Understanding the Effects of Government

Spending on Consumption. Journal of the European Economic Association5(1):22770.

Hemming, R., M. Kell, and S. Mahfouz. 2002. The Effectiveness of Fiscal Policy in StimulatingEconomic ActivityA 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 Economy53(1):2846.

Mountford, A., and H. Uhlig. 2009. What are the Effects of Fiscal Policy Shocks? Journal of

Applied Econometrics 24(6):96092.

Nijkamp, P., and J. Poot. 2004. Meta-analysis of the Effect of Fiscal Policies on Long-run Growth.

European Journal of Political Economy20(1):91124.

Pappa, E. 2009. The Effects of Fiscal Shocks on Employment and the Real Wage. International

Economic Review50:21744.

Perotti, R. 1999. Fiscal Policy in Good Times and Bad. Quarterly Journal of Economics

114(4):1399436.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.



34/34

About the Paper

Shikha Jha, Sushanta Mallick, Donghyun Park, and Pilipinas Quising use historical datarom 10 developing Asian countries and fnd evidence suggesting tax cuts are moreeective countercyclical policy instruments than government spending. However, Asiangovernments should be cautious about actively using tax cuts or countercyclical purposes,partly because a big part o the revenue shocks in the region are nondiscretionary ratherthan discretionary.

About the Asian Development Bank

ADBs vision is an Asia and Pacifc region ree o poverty. Its mission is to help its developingmember countries substantially reduce poverty and improve the quality o lie o theirpeople. Despite the regions many successes, it remains home to two-thirds o the worldspoor: 1.8 billion people who live on less than $2 a day, with 903 million struggling onless than $1.25 a day. ADB is committed to reducing poverty through inclusive economicgrowth, environmentally sustainable growth, and regional integration.

Based in Manila, ADB is owned by 67 members, including 48 rom the region. Its

main instruments or 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

< 0 1 0 2 3 2 0 6 >

Documents

ADB - Effectiveness of Counter Cyclical Fiscal Policy - Time-Series Evidence From Developing Asia