This document summarizes research on the relationship between fiscal decentralization and regional disparities. The key findings are:
1) Tax decentralization and higher fiscal authority are associated with lower regional disparities, especially for lower income regions, by stimulating sub-national governments to implement more pro-growth policies.
2) Revenue decentralization and higher vertical fiscal imbalances are linked to higher regional disparities.
3) The impact of fiscal decentralization on disparities depends on factors like countries' levels of development and quality of government institutions. Decentralization seems to reduce disparities more in developed countries with higher-quality government.
Fiscal relations across levels of government and regional disparities
1. FISCAL RELATIONS ACROSS LEVELS OF
GOVERNMENT AND REGIONAL
DISPARITIES
David Bartolini,
OECD Fiscal Network
ZEW Public Finance Conference,
Mannheim 25-26 April 2016
2. Motivation 1: Regional disparities
• Why is it important?
– Impact on economic growth: overall growth depends
on the contribution of each region (dynamic and size)
– Impact on income inequality: geographical inequality
adds to overall inequality (national GINI)
• How to measure regional disparities?
– Unit of analysis: OECD TL2 regions (NUTS2 regions)
– Coefficient of variation:
.
– Using regional per capita GDP in constant PPP 2005
US$
2
4. On the right-hand side of the Kuznets curve (?)
• Low income countries
seems to display larger
regional disparities
• Inequality is picking up
again for high levels of
development – signal
the importance of high-
tech activities and
services that tend to be
concentrated in cities
4
5. • Fiscal decentralisation might increase the efficient
provision of local public goods, but there is fear that it
would also increase regional disparity
• Main goal: investigate the impact of several indicators of
fiscal decentralisation on regional disparities
Motivation 2: fiscal decentralisation
5
Fiscal decentralisation indicators (from OECD Fiscal Decentralisation database):
• Revenue
• Tax
• Expenditure
• Tax autonomy
• Tax Authority (RAI)
6. Increases
disparities
Less endowed regions will
suffer – no level playing
field for competition
“race to the bottom”
(Prud’homme, 1995) and
“self selection”(Tiebout,
1956)
Corruption at the local
level (Tanzi, 1996)
Decreases
disparities
Efficiency (Oates, 1972) ,
Public choice (Brennan &
Buchanan, 1980),
transparency/political
economy (Salmon, 1987)
Incentive for growth-
enhancing policies (Qian &
Weingast, 1997)
Larger potential of
endogenous growth in
poor regions (Baldwin &
Krugman, 2004; Barankay
& Lookwood, 2007;
Rodriguez-Posé & Ezcurra,
2010)
Fiscal decentralisation framework:
existing literature
6
7. Fiscal autonomy,
balanced fiscal
structure
Incentive to
increase tax base
Competition
More
inequality
Less
inequality
Better use of
existing
resources
Less
inequality
• Key channel: better use of existing resources – there is
more scope for improvement in lagging regions than in
top performers, which are closer to the productivity
frontier
Importance of tax decentralisation and
vertical fiscal “balance”
7
8. Tax decentralisation may change the spending
decision of SNG
• Period 1995-2011
• Countries with large SNG tax share experience larger SNG spending
on economic affairs
8
9. Empirical strategy
Dependent variable (cv) • Coefficient of variation =
.
Fiscal decentralisation (FD) • SNG Revenue share
• SNG Tax share
• SNG Expenditure share
• Vertical fiscal imbalance (1-Rev/Exp)
Control variables (X) • GDP per capita of country i at time t
• Human capital, gross capital formation
• Trade openness
• Population concentration, pop, urbanisation
• Public expenditures, public debt
Fixed effects • Country
• Year
Unbalanced panel of 20 OECD countries over the period 1995-2011
9
10. Empirical results: fiscal decentralisation
(1) (2)
COV of per capita GDP robust SE
IV (2/3year
lag)
Tax decentralisation -0.278* -1.904***
(0.145) (0.707)
Revenue decentralisation -0.364* -0.433***
(0.189) (0.144)
Expenditure decentralisation 0.179*** 0.206***
(0.055) (0.069)
Vertical imbalance 0.127** 0.284***
(0.059) (0.085)
Fiscal autonomy -0.012*** -0.022***
(0.004) (0.005)
Observations 274 252
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
10
11. Empirical results: control variables
(2)
COV of per capita GDP
GDP pc 0.00163***
(0.000432)
Square GDP pc -1.55e-08**
(5.84e-09)
Capital formation -0.457***
(0.130)
Trade openness 5.204
(4.666)
Pop concentration -2.343**
(1.063)
Government expenditure size 0.157***
(0.0387)
Observations 274
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 11
12. Robustness
• Reverse causality – instrumental variable estimation
• Different time periods: (1995-2007) (2008-2011)
• Exclusion of one country at a time from estimation
• Different measure of regional disparities:
• Weighted CV
• GINI index
• percentile ratios (75/25 and 90/10)
• GDP per worker (labour productivity)
All these robustness checks provided results similar to the baseline model
12
13. • Cross section analysis
– Shankar and Shah (2003) decentralisation may increase disparities in “unitary”
countries
– Rodriguez-Posé & Gill (2004) cross section analysis of OECD countries –
decentralisation increases regional disparities because it favours economies of
agglomeration
• Panel data models:
– Lessman (2006, 2009): panel OECD countries (1980-2001) = all indicators of
fiscal decentralisation reduce regional disparity
• Differences in the level of development
– Rodriguez-Posé and Ezcurra (2010): impact of decentralisation depends on the
level of development of a country = political and expenditure decentralisation reduces
disparity only in developed countries
– Lessman (2012): panel of 54 countries (1980-2009) interaction decentralisation
and GDP pc has a negative impact on disparity
• Differences in the quality of government
– Kyriacou et al (2013): the quality of government rather than the level of
development may affect the impact of decentralisation on regional disparity
Selected literature review
13
14. Highest impact on “low income” regions
(catching up)
• Take regional income corresponding to the top (bottom) 25th percentile
• Use it a s dependent variable in the regression:
∆ !"#$% &'(# ) *'+ &',- ) *'+ &./ 0 1 2
(1) (2) (3) (5) (6) (7)
VARIABLES bottom bottom bottom top top top
SNG tax share 0.137* 0.0769
(0.0751) (0.0778)
SNG exp share 0.00188 0.0274 0.0692 0.0857*
(0.0578) (0.0595) (0.0526) (0.0508)
SNG rev share -0.00926 -0.0196
(0.0978) (0.108)
Fiscal authority 0.0112*** 0.0058
(0.00392) (0.00430)
• SNG tax share and Fiscal authority significant impact only on the
bottom 25th percentile
14
15. 1. FD reduces regional disparities,
if it does not increase vertical fiscal imbalance
2. Tax decentralisation stimulates SCG to
implement pro-growth policies
3. Tax decentralisation favours catching-
up of lagging regions
4. FD stimulate regional mobility in the
distribution
Main conclusions
15
17. Most of TL2 regional disparity depends on labour
productivity
3-4
5"5
3-4
675
∙
675
9:4
∙
9:4
5"5
Productivity Employment rate
Activity rate
Note: coefficient of variation for the year 2010 17