Growth in Government
• Structural models of growth in government
• Wagner’s Law
• Baumol’s cost disease
• Displacement hypothesis
The “law of increasing expansion of public and
particular state, activities” becomes for the fiscal economy
the law of the increasing expansion of
fiscal requirements. Wagner (1883)
The driving force: 2. Increasing incomes
3. Social progress
Modern (textbook) version:
As per capita incomes rise in industrial countries, their
public sectors grow in relative importance, or the income
elasticity of “public services” is greater than 1.
General government expenditure, 1870-1996
(percent of GDP)
0 17 OECD
1870 1913 1937 1960 1980 1990 1996 countries
Source: Tanzi and Schuknecht (2000), Table I.1.
Structural Direct effect Effect on government
change complexity of econ. banking services
relations. Law & order
income demand for income education
social progress elastic goods
demand for redistribution
New industries Government enterprises
economics of scale
Notice no feedback effects are considered. non-tax revenues
Testing Wagner’s Lawspending
d ln g
d ln y
GDP per capita
Absolute version: 0 Normal good
Relative version: 1 Superior good
(share of government spending
in GDP increases).
ln g i 0 ln yi X i i cross country
ln g t 0 ln yt 1 ln g t 1 X t t
LR 1 2
Time series in one country
• The absolute version has considerable
support, i.e., public services are normal
• The relative version yields mixed results
– most studies find estimates of the income
elasticity less than 1.
– those which do find larger effects are
typically not robust to proper econometric
Ball game estimate: 0.75.
Health care prices
The growth in government expenditure as a fraction of
GDP is driven by low productivity growth in the public
sector relative to that in the private sector.
Why not cumulative productivity growth in public services?
1. Education and other services are very labour intensive.
2. W up => higher average costs
3. Less units supplied but higher total cost
(1) Differential productivity growth in productivity
g t Lgt yt Lyte rt
Public sector Private sector
(2) Wage equalization across sectors wt e rt
(3) Fixed supply of labour: 1 Lgt Lyt
(4) Demand for public services: pgt
d wt pgt w
0 (price elasticity) tp t
0 (incomeelasticity) yt
(5) Demand for private goods:
d gt wt pgt gt
t p p
The average cost of producing public and private goods
wt Lg ,t wt Ly ,t
Pgt Cgt e rt
Pyt C yt 1
e rt increases over time
The relative size of government:
Pg ,t gt Pgt ( Pgt ) wt e rt e r ( )t
S gt e r ( )t
Pg ,t g t Py ,t yt wt e rt
r ( ) Growth if
Income Evidence Price
(1) Compensated price elasticity of public services:
(2) Income elasticity:
Can we assume that these elasticities are constant?
Central Government Spending in the UK
Central government spending %of GDP
1820 1840 1860 1880 1900 1920 1940 1960 1980
Source: Aidt et al. (2004)
The Displacement Hypothesis
(Peacock & Wiseman, 1961)
Government spending evolves in a step-like pattern,
where each step coincides with social upheavals, notably wars.
• The government can always find profitable ways to expend available
funds (in terms of generating political support).
• Citizens, in general, are unwilling to accept higher taxes than they
have grown accustomed to in the past.
• Governments must be responsive to the wishes of their citizens.
The “tolerable burden of taxation” T *
gt t *
but g t t
Actual spending spending desired by government
Social upheaval causes a permanent shift upwards
in the tolerable burden of taxation:
gt 1 t1 t gt
In times of crisis, formerly unacceptable means of
taxation will be tolerated and, importantly, this
higher tolerance persists.
normal war normal
• What qualifies as a major social upheaval:
– WWI and WWII
– Great Depression?
– Oil Shocks?
g t a1 a2 DWWI a3 DWWII X t t
0 if year < 1914
1 if year = 1923-38 or >1950 a1 0
0 if year < 1914 or 1923-38 a2 0
1 if year >1950
• Upwards trend, series correlation,
• No modelling of the wars (catch up
• Data relatively sparse before WWI.
• Model the times series properties of g.
• Model the war(s).
• Evidence much weaker when these things
are taken into account.
• The displacement hypothesis and (textbook)
Wagner’s Law have little empirical support.
• Baumol’s hypothesis has more support for
particular services (health and education), but
what about constant net size?
• Problem that the various structural theories
are considered in isolation. What is really
needed is a united (nested) framework.
• Problem that political institutions are not taken
• Problem that interactions between politics and
economies are not considered.
What is next?
• The median voter model
• Extension of the voting franchise
– relaxing income and property requirements
– relaxing gender restrictions
The weak version of the displacement
g total spending
y civilian spending
normal war normal
Underlying trend growth: level shift but not trend
ARIMA model of spending in
time series model
gt b0 b1T b2 gt 1 b3 gt 2
b4 PWt b5Wt X t b t
Dummy = 1 after WWII Dummy for WWII
Result: b4 negative and significant!!!
Little systematic evidence of the