Is the size of government optimal in Pakistan?
ul Husnain, Muhammad Iftikhar
INTRODUCTION
Historical evidence reveals that no society gained high level of
economic prosperity without the role of government. Societies without
proper governments faced different kinds of anarchy that halted their
growth over time. The emergence of governments in these societies
ensured the rule of law and protected property rights leading to high
level of economic development. Thus, the role of governments remains
important for the economic prosperity of the nations.
On the other hand, societies where all economic decisions were made
by the governments witnessed relatively low level of economic affluence
because large governments stifled the spirit of private agent, which
resulted in low level of economic development. Put differently, the
economic prosperity is limited both at zero and hundred percent level of
government. Thus, question arises; how large should be the size of
government?
Empirical as well as theoretical literature has focused on this
issue with conflicting results in recent decades. A group of economists
believes that large government size stimulates economic growth (Ram,
1986, 1989; Rubinson, 1977) among others. On the other hand some
economists deny this phenomenon and conclude that as the size of
government increases in relative term it reduces the growth of per
capita income (Landau 1983, 1986; Barro, 1991). Bairam (1990) and
Grossman (1990) conclude that government can affect growth positively as
well as negatively. However, within this diversity of explanations a
consensus emerges that up to a certain level, government activities are
pro growth but beyond this point the size of government may reduce
economic well-being. The debate concentrates the question that at what
point the public expenditures become counterproductive.
Only a handful of studies have attempted to determine the optimal
level of public expenditures for developed countries whereas, up to best
of my knowledge, no study can be found in case of developing countries
on the issue. This study attempts to determine the optimum level of
government expenditures/taxes in Pakistan using data for the period
1975-2008. This will help the policy makers to increase/decrease public
expenditures, as a percent of GDP, to bring the government size closer
to the optimal level. The rest of the study has been structured as
under: Section 2 shows public expenditures and growth scenarios in
Pakistan over time. Section 3 & 4 describe data and analytical
framework respectively. Section 5 presents estimation and results. The
last section, 6, concludes the paper with some policy implications.
GROWTH AND PUBLIC EXPENDITURES SCENARIOS IN PAKISTAN (1975-2008)
The growth path of Pakistan economy fluctuates over time. In some
decades it showed an impressive growth while in other decades this
momentum was lost. From mid 1970s to mid 1980s, Pakistan economy grew at
an impressive average percentage growth rate of 6.25. The decade from
1985 to 1995 witnessed an average growth rate of 5 percent. From 1996 to
2002, growth rate paints a dismal picture. Growth rate declined from 5%
in 1996 to 1.6% in 2002. This happened mainly due to the imposition of
economic sanctions by international community following the atomic
explosion on May 28, 1998. It affected growth adversely in Pakistan and
left the economy reeling at just more than 2 percent average growth rate
in this period. Since 2003 economy has regained its momentum and it is
growing at an impressive rate of more than 5 percent per annum.
[FIGURE 1 OMITTED]
In Pakistan, public expenditures to GDP ratio was 23.11 percent
between 1975 and 1985 fueled by several agendas. The major determinants
of this high level of public expenditures were nationalization of
institutions and government's efforts to increase level of
employment in the economy. Higher level of defense expenditures and
interest payments can also be accounted for this larger size of
government in Pakistan in this period. Public expenditures rose from
23.11percent to 25.78 percent of GDP from 1985 to 1995 showing greater
involvement of government in economic activity. The next ten years
(1996-2006) may be called the era of less government involvement as the
public sector squeezed significantly. Public expenditures shrunk from 26
percent of GDP in 1996 to 18 percent of GDP in 2005, a decrease of
substantial 30 percent. However, since 2006 public expenditures in
Pakistan are showing an upward trend.
Figure 1 clearly shows that both public expenditures and economic
growth have been volatile over the years in Pakistan. It can be noticed
that low level of government expenditures are associated with relatively
high growth rate and vice versa.
DATA
The consolidated central government expenditures as a share of GDP
are used as a proxy for the size of government. This variable comes from
Government Finance Statistics (GFS), a publication of the International
Monetary Fund (IMF). GDP at current prices is obtained from World
Development Indicators (WDI 2009), published by the World Bank (WB).
Time period of the study ranges from 1975 to 2008. Accordingly, 34
observations are available for the analysis.
ANALYTICAL FRAMEWORK
The analysis employs the methodology of Scully (1994 & 2006)
and Heerden et al (2008) to find the threshold level of government
expenditures in Pakistan. The mathematical formulation is based on a non
linear Cobb-Douglas production function with a government sector and a
private sector. The public sector provides goods financed with tax
revenues i.e. [sigma]/Y = [tau] and 1 - [tau] is the share retained by
private agents after taxation. Both public and private goods contribute
to output.
Y = [gamma][(G/Y).sup.[alpha]][(1 - [tau]).sup.[beta]] (1)
Where
Y = Gross Domestic Product
G = Government Expenditures
[tau] = Tax to GDP Ratio
[gamma] = Total Factor Productivity
[alpha] = Relative Share of Government Sector in Total Output
[beta] = Relative Share of Private Sector in Total Output
In logarithmic form equation (1) can be written as;
lnY = ln[gamma] + [alpha] ln(G/Y) + [beta] ln(1 - [tau]) (2)
The equation (2) is differentiated twice with respect to G to show
that an increase in government expenditures affect growth positively but
at a diminishing rate.
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]
The first derivative is positive while second derivative is
negative which verifies that public expenditures positively affect
growth but at diminishing rate.
By definition G/y = [tau], therefore, substitution into equation
(1) yields;
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (3)
To find growth maximizing tax rate, we differentiate equation (3)
with respect to [tau] and set it equal to zero.
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]
Rearranging terms;
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (4)
ESTIMATION AND RESULTS
We estimate equation (3) to find values of the parameters, [alpha]
& [beta]. Empirical evidence reveals that economic variables are
mostly non stationary. Thus the regression of non stationary variables
may lead to spurious results. Therefore, we start our empirical analysis
by testing the stationarity of the variables. Augmented Dickey Fuller
(ADF) unit root test has been employed for this purpose. The results are
reported in table 1.
Table 1 brings out that the variables are integrated of order one
i.e., non stationary. Therefore, variables are first differenced to make
them stationary before applying OLS to estimate the regression
coefficients. The OLS results are presented in table 2. It is clear from
table 2 that all the coefficients are highly significant.
Calculation of Optimal Level of Public Expenditures/Taxes
To measure the threshold government size, the values of regression
coefficients are substituted in equation 4.
[tau]* = 0.212/(0.212 + 0.775) x 100 = 21.48%
Our calculation shows that the optimal level of public
spending/taxes in Pakistan is 21.48% of GDP as against the 22.7% level
in 2008. Therefore, it is concluded that current size of government in
Pakistan is above the threshold size and there is a scope of 5.4 percent
to decrease public expenditures to attain the optimal size of
government.
The comparison of optimal size of governments in developing and
developed countries seems useful. In table 3, we present international
evidence on the optimal size of governments in developed countries.
Table 3 shows that the estimated optimal size of government is
larger in industrialized countries as compare to Pakistan. The high
optimal level of public expenditures/taxes in industrialized countries
is due to the efficiency of these governments that keep the functioning
of downward sloping phase of the Scully curve away for a longer time.
Contrary to this in developing countries, like Pakistan, due to many
factors like bad governance, corruption and political use of scarce
public resources efficiency of government remains low and the negative
effect of public expenditures start earlier than the developed
countries.
Our estimate of optimal size of government is in line with the
findings of Freedman (1997) who states that the optimal size of
governments ranges from 15 percent to 50 percent of gross domestic
product. Similarly our results match the findings of Scully et al (2008)
who find the level of tax rate that maximizes economic growth in US
about 19.3 percent of GDP. The findings are also supported by Vedder et
al (1998) and Mavrov (2007).
Reliability Analysis
Auto-correlation is a common phenomenon in time series analysis.
The Durban Watson statistics reported in table 2 reveals that residuals
are uncorrelated. Further, some more tests are applied to check whether
the results are reliable or not. The outcomes of these tests are
reported in table 4.
It is clear from table 4 that our model has no specification error
and residuals are not correlated. No Hetroskasdicity is found and on the
basis of Jarque-Bera test it is concluded that residuals are normally
distributed. In the presence of above tests results, it can be concluded
that our estimates of regression coefficients are reliable and need not
to be adjusted.
CONCLUSION AND POLICY IMPLICATIONS
We use time series data from 1975 to 2008 to estimate optimal size
of government in Pakistan under the assumption of balanced budget. The
analysis suggests that the optimal size of government is lower than the
current size of government (21.48 < 22.7) and there is a scope of
5.4% reduction in public expenditures. International comparison reveals
that both current and optimal sizes of public expenditures are high in
developed nations when compare to Pakistan. This shows that government
expenditures become counterproductive at earlier stage and the downward
phase of the Scully curve starts functioning at comparatively low level
of public spending in a country like Pakistan.
Some interesting policy implications emerge from the analysis.
Firstly, as the optimal size of government is lower than the current
size, hence there is a scope to decrease government expenditures in
Pakistan. Low tax burden will encourage the private sector that in turn
can stimulate economic growth in the country. Secondly, as the current
and optimal levels of public spending have merged recently in Pakistan,
the only issue is to enhance efficiency of the public sector as it is
already closer to the optimal level. Lastly, the current increasing
trend in public spending in the country needs to be checked as it may
hurt the economy in future. Fiscal discipline through cuts in public
expenditures can help achieve this objective.
Future research may focus to measure the gains that can be made by
reducing the size of government to the optimal level. In addition the
dead weight loss can be measured for the years in which government size
remained beyond the optimal level. The finding of optimal level of
different categories of public expenditures such as defense, education
and health can also be an interesting topic for the researchers.
REFERENCES
Akram, N., I. H. Padda & M. Khan (2008). The long term impact
of health on economic growth in Pakistan. The Pakistan Development
Review, 47(4), 487-500.
Bairam, E. (1990). Government size and economic growth: the African
experience, 1960-85. Applied Economics, 22, 1427-35.
Barro, R. (1991). Economic growth in a cross-section of countries.
Quarterly Journal of Economics, 106(2), 407-443.
Dickey, D & W. A. Fuller (1979). Distribution of the estimators
for time series regressions with a unit root. Journal of the American
Statistical Association, 74, 427-431.
Friedman, M, (1997). If only the U.S. were as free as Hong
Kong," Wall Street Journal, July 8, p. A14.
Grossman, P. J. (1990). Government and growth: cross- sectional
evidence. Public Choice, 65(3), 217-227.
Heerden, Y. & N. Scheoman (2008). University of Pretoria,
department of economic working paper series.
Husnain, M. I. (2010). Expenditure growth nexus: does the source of
finance matter? Empirical evidence from selected South Asian countries.
PSDE, 26th AGM and proceedings (28th-30th March), PIDE, Pakistan.
Husnain, M. I., M. Khan., I. H. Padda., N. Akram & A. Haider
(2011). Public Spending, foreign direct investment and economic growth:
A time series analysis for Pakistan (1975-2008). International Journal
of Finance and Economics, 61, 20-28
Landau, D. (1983). Government expenditure and economic growth: a
cross-country study. Southern Economic Journal, 49 (3), 783-92.
Landau, D. (1986). Government and economic growth in less developed
countries: an empirical study for 1960-80. Economic Development and
Cultural Change, 35-73.
Mavrov, H. (2007). The size of government expenditure and the rate
of economic growth in Bulgaria. Economic Alternatives, 1, 52-63.
Padda, I. H. & N. Akram (2009). The impact of tax policies on
economic growth: evidence from Asian economies. The Pakistan Development
Review, 48(4), 961-971.
Padda, I. H. & N. Akram (2010). Fiscal policy and growth nexus:
scenario of developing economies. The Empirical Economics Letters, 9,
405-412.
Pevcin, P. (2004). Does optimal size of government spending exist?
University of Ljubljana. http://soc.kulluven.be/io/egpa/fin/paper/slov.
Ram, R. (1986). Government size and economic growth: a new
framework and some evidence from cross-section and time-series data. The
American Economic Review, 76, 191-203.
Ram, R. (1989). Government size and economic growth: a new
framework and some evidence from cross-section and time-series data. a
reply: The American Economic Review, 79(1), 218-284.
Rubinson, R. (1977). Dependency, government revenue, and economic
growth, 1955-70. Studies in Comparative International Development, 12,
3-28.
Scully, G. (1994). What is the optimal size of government in the
US? National Center for Policy Analysis, Policy Report No. 188.
http://www.ncpa.org/studies/s159/s159.html
Scully, G. (2006). Taxes and economic growth. National Center for
Policy Analysis, Policy Report No. 292.
Scully, G. (2008). Optimal Taxation, Economic Growth and Income
Inequality in the United States. National Center for Policy Analysis,
Policy Report No. 316.
Vedder, R. & L. Gallaway (1998). Government size and economic
growth. Joint Economic Committee, Washington D.C., p. 5.
Muhammad Iftikhar ul Husnain, Federal Urdu University
Table 1
ADF Unit Root Test
Null Hypothesis: The series has a unit root
Level
Variables C C & T None
LnY -0.32 -3.15 7.64
Ln[tau] 0.07 -2.08 3.37
Ln[tau] -0.35 -3.03 4.84
First Difference
Variables C C & T None
LnY -4.70 * -4.56 * -1 92 ***
Ln[tau] -3.97 * -3.78 ** -2.09 **
Ln[tau] -5.15 * -4.98 * -1.88 ***
C and T denote Constant and Trend respectively.
*, **, *** mean significant at 1%, 5% and 10% respectively.
Table 2
Dependent variable is [DELTA] LnY
Variables Coefficients Standard Error T-Values
C 0.002 0.002 1.179
[DELTA]Ln[tau] 0.212 0.026 4.760
0.775 0.008 3.227
R-Square: 0.976 DW: 2.08
Note: [DELTA] is first difference operator. Ln denotes natural
log while Y, [tau]: and (1 - [tau]) represent GDP, government
share in output and share retained by private agent after tax.
Table 3
Optimal level of public expenditure in developed countries
Country Size of government Optimum size Percentage change
(% of GDP, 1996) (% of GDP) in spending as a
share of GDP
Italy 44.90 37.09 -17.39
France 54.73 42.90 -21.62
Finland 58.74 38.98 -33.64
Sweden 65.02 45.95 -29.31
Germany 48.72 38.45 -21.08
Ireland 39.60 42.28 +6.77
Netherlands 51.97 44.86 -13.68
Belgium 52.97 41.91 -20.88
Average percentage change -18.85
Reproduced From Pevcin (2004)
Table 4
Selected Diagnostic Test Results of the Model
Test-
Test Null hypothesis Statistics P-value
Ramsey RESET Model is stable with no 1.21 0.314
F Test specification error
Normality Test Residual are normally JB = 3.53 0.171
(Jarque- Bera) distributed
Breusch- No serial correlation 1.79 0.186
Godfrey LM F in the residuals up to
Statistics the 2nd order
ARCH F Test No auto regressive 0.160 0.692
conditional
hetroskedasticity
up to the 1st order
Test Conclusion
Ramsey RESET Can not reject null hypothesis and
F Test conclude that model is stable with no
specification error.
Normality Test Null hypothesis cannot be rejected and
(Jarque- Bera) conclude that residuals are normally
distributed.
Breusch- Can not reject null hypothesis and
Godfrey LM F conclude that residuals are not
Statistics correlated up to 2nd order.
ARCH F Test Null hypothesis cannot be rejected.
Hence there is no auto regressive
conditional hetroskedasticity up to the
1st order.
