Sanjeev Gupta Deputy Director of Fiscal Affairs Department

International Monetary Fund

Tokyo Fiscal Forum June 10, 2015

Fiscal Policy and Long-Term Growth

Outline

Motivation

The Channels: How Can Fiscal Policy Affect Medium- to Long-Term Growth?

Empirical Evidence: Results of a Multi-Pronged Analysis

Other Key Lessons: Equity and Reform Design

Conclusions

2

Output across advanced and emerging market economies remains below expectations

707580859095

100105110115120

2000 2002 2004 2006 2008 2010 2012 2014

Fall 2007 Fall 2008 Fall 2014

Advanced Asia 1/ Index of Real GDP, 2007 = 100

Source: WEO. 1/ Hong Kong, Japan, Korea, Singapore, and Taiwan. 2/ China, India, Indonesia, Malaysia, Philippines, Sri Lanka, and Thailand.

50

70

90

110

130

150

170

2000 2002 2004 2006 2008 2010 2012 2014

Emerging Asia 2/ Index of Real GDP, 2007 = 100

3

How Can Fiscal Policy Affect Medium- to

Long-Term Growth?

4

At the macro level, fiscal stabilization reduces volatility and promotes growth…

0

1

2

3

4

5

6

7

8

-0.5 0.5 1.5

Out

put

vola

tility

(pe

rcen

t)

Fiscal stabilization coefficient

Significant coefficients Insignificant coefficients

Correlation = −0.40

0

1

2

3

4

5

6

7

8

-0.5 0.0 0.5 1.0 1.5

Real

GDP

gro

wth

rate

(per

cent

)

Fiscal stabilization coefficient

Fiscal Stabilization Coefficient vs Output Volatility

Fiscal Stabilization Coefficient vs RGDP Growth

5

..and allowing automatic stabilizers to operate in good times can avoid public debt buildup

30

40

50

60

70

80

90

t t+2 t+4 t+6 t+8 t+10 t+12 t+14 t+16 t+18 t+20

Publ

ic d

ebt

Symmetric stabilization

Revenue windfalls half spent

Asymmetric Stabilization: Unpleasant Public Debt Arithmetic (Percent of GDP)

6

Fiscal stabilization is much more common in advanced economies

0

2

4

6

8

10

12

Num

ber o

f cou

ntrie

s

Stabilization coefficient Insignificant Significant

Median coefficient (0.7)

0

5

10

15

20

25

30

35

Num

ber o

f cou

ntrie

s

Stabilization coefficient Insignificant Significant

-.06 -0.2 0.2 0.6 1.0 1.4 1.8 2.2

Median coefficient (0.3)

-0.2 0.2 0.6 1.0 1.4 1.8

Advanced Economies Emerging Market and Developing Economies

7

At the micro level, fiscal policy affects growth through four main channels

Labor supply Human capital

Physical capital Productivity/

Innovation

8

Fiscal Policies to Encourage Labor Supply

• Lowering the labor tax wedge increases after-tax earnings and the supply of labor (succeeded in Ireland, and the Netherlands)

• Use of in-work benefits can strengthen work incentives (used in Germany, the UK, Sweden)

• Targeted measures may be needed to increase LFP:

→ Women: closing the gender gap in education (e.g., in India); or providing better child care and flexible work options (e.g., in Japan)

→ Older workers: financial incentives (e.g., through tax rates); and increasing the retirement age;

→ Low-skilled workers: in-work tax credits; hiring subsidies; targeted reductions social contributions. 9

Fiscal Policies to Enhance Investment

• In AEs, taxing “excess returns” or rents can reduce distortions from CITs

• Infrastructure investment can boost growth directly and indirectly by raising the productivity of private capital: but efficiency is key:

→ The most efficient countries get twice the growth dividend from investment compared with the least efficient countries

• In developing economies, targeted and transparent incentives that reduce the cost of capital can promote investment

• Tax incentives can erode the revenue base without achieving any benefits from higher investment unless they are properly designed and limited 10

Open-ended and profit-based tax holidays should be avoided

0

20

40

60

80

100

Tax Holiday/TaxExemption

Reduced Tax Rate InvestmentAllowance/Tax

Credit

R&D Tax Incentive Super-Deduction

East Asia and Pacific Eastern Europe and Central AsiaLatin America and the Caribbean Middle East and North AfricaOECD South AsiaSub-Saharan Africa

Source: James (2013).

Regional Prevalence of Tax Incentives (Percent)

11

Fiscal Policies for Human Capital Development

• Improving access to education and health for disadvantaged groups is a priority, including by:

→ Increasing investment at lower levels of education and increasing cost-recovery in tertiary education (while protecting the poor)

→ Providing a basic health package; expanding services to remote areas; and reducing user charges for poor households

→ Conditioning cash transfers on school attendance and preventive health visits

• In AEs, allowing for the deductibility of education expenses can mitigate the adverse impact of progressive taxation

12

Fiscal Policies to Promote Productivity and Innovation

-2

0

2

4

6

8

10

12

-5 0 5 10 15 20

Aver

age

Real

GD

P Pe

r Cap

ita G

row

th

Average Real R&D Expenditure Per Capita Growth Source: WDI. Note: Excludes countries with fewer than five observations during the

R&D Expenditures and Growth, 2001-2012 (Percent)

13

Empirical Evidence: Results of our Multi-

Pronged Analysis

14

The growth dividend from fiscal reforms can be substantial

Source: IMF staff calculations; Supplement 1. 1/ 5-year averages for Germany and Poland. 2/ Chile (1) refers to the first reform episode (1974); Chile (2) to the second reform episode (1983); Australia (1) to the first reform episode (1985); and Australia (2) to the second reform episode (1998).

0

1

2

3

4

5

6Advanced EconomiesEmerging MarketsLow-Income Countries

Estimated Growth Gain (Percent, GDP per capita, 10-year average 1/ 2/)

Average, AE (excluding Ireland)

Average EME and LIC

15

Country Focus: Malaysia

• Malaysia’s reform period from 1986-90 chosen based on quantitative selection criteria;

• A large expenditure-based fiscal adjustment, reduction in the size of the public sector, and economic deregulation were key elements of the reforms;

• Growth picked up markedly in the period following fiscal reforms, increasing by 2 percentage points vis-à-vis the counterfactual;

• Fiscal policy appears to have contributed to boosting Malaysia’s growth by promoting private investment, job creation and gains in TFP.

16

Malaysia: Growth post-reform exceeded expectations

0

1

2

3

4

5

6

7

8

9

1976-1985 1986-1990 1991-2000 2001-2010Malaysia Synthetic Control Group 1/

Annual GDP Growth, 1986-1990 (Percent, 10-year average)

Reform period

1/ Indonesia, Korea, and Philippines. 17

Fiscal reforms increase the probability of growth accelerations

Source: IMF staff calculations. Note: Reported are the ratios of fiscal reforms followed by a growth accelerations within a 5-year period to the total number of fiscal reforms (in percent).

0

10

20

30

40

50Pe

rson

al in

com

e ta

xes

Hea

lth s

pend

ing

Soci

al s

ecur

ityco

ntrib

utio

ns

Corp

orat

e in

com

e ta

xes

Cons

umpt

ion

taxe

s

Educ

atio

n sp

endi

ng

Tran

spor

t and

com

mun

icat

ion

spen

ding

Prop

erty

taxe

s

Capi

tal s

pend

ing

Type of Reforms and Conditional Probability of Growth Accelerations (Percent)

18

Other Key Lessons

19

Social dialogue helps deepen and sustain reform efforts

-2

-1.5

-1

-0.5

0

0.5

t t+1 t+2 t+3 t+4 t+5 t+6 t+7 t+8 t+9 t+10Social Dialogue Little Social Dialogue

Cumulative Change in the Public Wage Bill (Percent of GDP)

Source: IMF staff calculations. 20

Equity-efficiency trade-offs can be avoided

Source: IMF staff calculations, SWIID 5.0. 1/ Refers to Gini coefficient after taxes and transfers.

Australia (1)

Australia (2) Ireland Poland

Chile (1)

Chile (2) Germany

Malaysia

Netherlands

-8

-4

0

4

8

12

16

-30 -25 -20 -15 -10 -5 0 5 10

Chan

ge in

Gin

i coe

ffici

ent

Maximum Reduction/Increase in Primary Spending (Percent of GDP)

Advanced Economies Emerging Markets

Net Inequality 1/ (10-year avg. post-reform minus 10-year avg. pre-reform)

Decrease in spending/ increase in inequality

Decrease in spending/ decrease in inequality

Increase in spending/ decrease in inequality

Increase in spending/ increase in inequality

21

Conclusions

22

Conclusions

• Fiscal policy can be an effective tool for supporting medium- to long-run growth.

• The mix of fiscal policy options should be tailored to country-specific conditions, administrative capacities and preferences.

• The growth dividends of fiscal reforms depend to a large degree on complementary structural reforms and supportive macroeconomic policies.

• Strategies—such as effective communication with stakeholders and compensatory measures for those made worse off— can help foster public support for fiscal reforms.

• Both growth and equity objectives can be achieved when fiscal reform packages are appropriately designed.

23

Thank you

Annex Slides

25

Three studies find a positive link

Methodology Growth impact

Synthetic control method ¾ pp for AEs and even higher for DCs

Endogenous growth simulations

½ pp from budget neutral tax reforms + ¼ pp for enhancing composition of spending

Statistical analysis Increased likelihood of growth following fiscal reform

Synthetic Control Method Overview

• Formal data-drive procedure to quantify the effect of fiscal policy on long-run growth.

• Removes discretion in selection of countries. • The effect of fiscal policy is difference between

growth in the country and its synthetic counterpart.

• Results should be treated with caution due to potential biases.

Synthetic Control Method Intuition

• Goal: Evaluate the impact of fiscal reforms on long-run growth in a country of interest

• Issue: Difficult to find counterfactual showing what the long-run growth

would be if the country did not implement the reform • How does synthetic control method address this issue?

o Use a panel data of countries (synthetic control) that did not implement fiscal reforms around the same time as the country of interest, but have similar observable characteristics (region, level of development, etc.)

o Assess the impact of fiscal reforms by taking the difference between the post-reform growth rate in the country of interest and weighted-average growth rate of synthetic control group

o Countries with more similar observable characteristics with the country of

interest in the pre-reform period carry higher weights

28

Synthetic Control Method Formal implementation

• Units: j = 0, 1, 2, … J countries, where j=0 is the treated or reforming country, and j = 1, 2, … J are control countries

• Time: t = 1, 2, ... T1 periods, where pre-reform period is t = 1, 2, … T0 ;

post-reform period is t = T0+1, … T1 • Variables of interest:

o YI0t : GDP growth in treated country 0 at time t assuming reforms

were implemented at T0 o YN

0t : GDP growth in treated country 0 at time t assuming reforms were not implemented at T0

• Effect of fiscal reform: YI

0t - YN0t = α0t (for t>T0), where YN

0t is not observable and needs to be estimated

29

Synthetic Control Method Formal implementation (continued)

Regression model:

where: Z: observed covariates of growth (GDP per capita, trade openness,

inflation rate, terms of trade index, human capital per person) δ: unobserved time effects µ: unobserved country effects λ, θ: time-varying coefficients (λ is constant in dif-in-dif regressions) • Counterfactual growth rate is , where vector of non-

negative weights w* is chosen to minimize the difference between observable characteristics of treated and control groups

jt t t jt t j jtY Zδ θ λ µ ε= + + +

*0

1j

JN Nt jt

jY w Y

=

=∑

30

Synthetic Control Method Advantages and Disadvantages

Main advantages Main limitations

Allows selection of control group based on a transparent and flexible statistical procedure, rather than ad-hoc reasoning

Can lead to over-fitting if initial sample of control group is not selected based on similarity to the reforming country

Allows for a study of the dynamic impact of reforms.

SCM suffers from reverse causation bias if reforms depend on expected future growth.

Robust to endogeneity bias due to time-varying omitted variables.

Tests of statistical significance are difficult with SCM

31

Synthetic Control Method Literature

• Abadie, A. and Gardeazabal, J. (2003) “The Economic Costs of Conflict: A Case Study of the Basque Country”, American Economic Review, 93(1): pp. 113-132 – impact of terrorism on output in the Basque Country

• Abadie, A., Diamond, A., and Hainmueller, J. (2010) “Synthetic Control Methods for

Comparative Case Studies: Estimating the Effect of California's Tobacco Control Program” Journal of the American Statistical Association, 105 (490): pp. 493-505 - effect of California’s 1988 tobacco control program on tobacco consumption

• Cavallo, E., Galiani, S., Noy, I., and Pantano, J. (2013) “Catastrophic Natural Disasters and

Economic Growth” Review of Economics and Statistics, 95(5): pp. 1549-1561 - effect of large natural disasters on economic growth

• Billmeier, A. and Nannicini, T. (2013) “Assessing Economic Liberalization Episodes: A

Synthetic Control Approach”, Review of Economics and Statistics, 95 (3): pp. 983-1001 – impact of economic liberalization on real GDP per capita growth

• Abadie, A., Diamond, A., and Hainmueller, J. (2014) “Comparative Politics and the Synthetic

Control Method”, American Journal of Political Science (forthcoming) – impact of Germany’s 1990 unification on real GDP per capita

32

Endogenous Growth Model Methodology

Key Features:

o Two sectors: final output and human capital

o Government investment in productive public capital

o Endogenous labor supply

o Constant returns to scale in public and private capital

o Accumulation of public and human capital offsets diminishing returns to physical capital accumulation

33

Endogenous Growth Model Results

Fiscal Reform Increase in LT Growth Relative to

Benchmark

Offsetting Measures

Δ Capital tax -5% Δ Labor tax -5%

0.4-0.5 pp ↑ consumption tax

Δ public investment +1% of GDP

0.15-0.2 pp ↓ unproductive spending by 1 pp of GDP

Budget-neutral experiments

34