Public-private investment and economic growth in Pakistan: an empirical analysis.
Bint-e-Ajaz, Maryam ; Ellahi, Nazima
1. INTRODUCTION
Investment is an important component of aggregate demand and a
leading source of economic growth. Change in investment not only affect
aggregate demand but also enhance the productive capacity of an economy.
A third important role highlighted in the literature refers to the
innovation and modernisation of the capital equipment via technological
progress. The investment plays an essential and vital role in expanding
the productive. Maryam capacity of the economy and promoting long term
economic growth [Jongwanich and Kohpaiboon (2008)]. Levine and Renelt
(1992) argued that investment in capital goods is the most robust and
vital determinant of economic growth. Gross domestic investment boosts
economic growth by increasing physical capital directly and indirectly
through technological spillovers [De Long and Summers (1995)].
1.1. The Role of Investment in Growth Process
There has been heated debate in policy-making and academic circles
regarding the roles of public and private investment in the process of
economic growth. In the 1950s and 1960s available economic models seemed
to offer only limited insight into the practical problems facing by
developing world. The dominant one-sector macro models of the day, from
Keynesian to Harrod-Domar [see Harrod (1939) and Domar (1957)] to Solow
1956, seemed to have relatively little relevance for developing
societies like Pakistan.
Available literature including recent extensions of the
neo-classical growth model as well as the theories of endogenous growth
has highlighted the role of investment in economic growth [see, for
example, Kormendi and Meguire (1985); Romer (1986); Lucas (1988); Grier
and Tullock (1989); Barro (1991); Levine and Renelt (1991); Rebelo
(1991); Mankiw, Romer, and Weil (1992); Barro and Lee (1993); Fischer
(1993) and Barro and Sala-i-Martin (1999)]. The effect of public
investment on economic growth depends on how the increased spending is
financed by the government [Bukhari (2006)]. If public and private
investments are perfect substitutes, then an increase in public
investment would have the same effect on growth as an increase in
private investment. Both contribute to the accumulation of physical
capital, which increases the productive capacity sustains a higher level
of output [Lachler and Aschauer (1964)]. Public investment in the
infrastructure has to boost up private investment indirectly that in
turn increases the marginal productivity of private capital and enhances
the growth of GDP [Looney, et al. (1997)]. It generates positive
spillovers by provision of health, education, basic scientific research
and physical infrastructure, and may also "crowd in" the
private investments. In contrast, the literature also suggests that
public investment negatively effects the private investment via the
well-known "crowding out" phenomenon via attracting the
domestic scare sources through bond floating [Erden and Holcomble
(2005)]. These contrasting views about the impact of public investment
on private investment are important, however yet unsettled.
So for as Pakistan is concerned, several studies have been carried
out, which concentrate on public and private investment and economic
growth. The most important are the studies inter-alia by Khan (1988),
Looney and Frederiken (1995), Loony, et al. (1997), Khan and Sasaki
(2001), Naqvi (2003), Ghani and Din (2006), Khan and Khan (2007), Ahmad
and Qayyum (2007) and Majeed and Khan (2008). In some studies the
relationship between growth and investment is investigated, while others
have attempted to examine the determinants of public and private
investment.
Given the vital importance of investment in the process of economic
growth, this study endeavours to develop an econometric model to examine
the relationship between public and private investment and growth. The
present study attempts to follow a comprehensive approach by examining
the overall effect of investment on growth, explaining the determinants
of public and private investment and evaluating the mutual relationship
of the both the components. Thus the rationale is obvious; instead of
following a piece meal strategy, it looks more efficient to place all
the components in one place and discuss the issue as a whole using
different models.
2. REVIEW OF EMPIRICAL LITERATURE
A number of empirical studies are available, which illustrate the
relationship between public investment, private investment and economic
growth with reference to Pakistan economy. This part presents a brief
review of the empirical literature relating to the issue concerned.
Looney and Frederiken (1995) estimated the relationship between
public and private investment and concluded that certain types of
government investment--especially in rural works 'crowded out'
private investment in non-manufacturing activities. Likewise, the public
infrastructure investment in energy projects provided the greatest
inducement to private investment. Side by side Loony, et al. (1997)
studied the impact of Government investment on private sector in
Pakistan over the period 1972 to 1995 and concluded private sector
investment depends on the lagged change in GDP, the change in private
sector credit, the lagged value of private investment, government
expenditure in the infrastructure and other projects.
Khan and Sasaki (2001) analysed the role of public capital in
Pakistan s' economy. The results showed that public labour ratio
and public capital had significantly positive effect on output. Public
capital productivity contributes largely at the aggregate and sectoral
level and so it played an effective role in the production process.
According to Naqvi (2003) public investment had a positive impact on
private investment, and that economic growth pushes forward both private
and public investment. Naqvi (2003) proved that long run estimates of
the elasticities of public and private investment are different under
different assumptions made about the evolution of technology. If
technology was considered exogenous, the elasticities of private and
public capital with respect to output and rate of return were similar to
each other.
Same relationship is examined by Ghani and Din (2006) and indicated
that public investment had a negative, though insignificant, impact on
output. In contrast, there was a positive relationship between private
investment and economic growth. Public investment had no favourable
impact on private investment; in other words, it 'crowded out'
private investment and this result raises some concern about the
efficiency of public investment.
Khan (1988) examined the impact of fiscal and monetary policies on
private investment in Pakistan. Private investment in aggregate as well
as investment in manufacturing and agriculture sector was estimated. The
study concluded that market conditions appear to have a strong influence
on private investment in general, while changes in output had minor
impact. Khan and Khan (2007) investigated the determinants of private
investment in Pakistan. The results showed that real GDP had positive
but insignificant impact on private investment while public investment
had negative but insignificant impact on private investment. According
Ahmed and Qayyum (2007) there was long run relationship between private
fixed investment, public consumption and development expenditure and
market activities. The relationship between public investment and
private investment was positive.
3. INVESTMENT AND GROWTH IN PAKISTAN
Pakistan economy has faced many crises since independence in 1947.
These crises have hampered the sustainable economic growth. During the
1950's decade, the Korean war boosted our exports and foreign
exchange earnings that helped maintaining high economic growth. In
1960's, the continuous inflow of foreign aid and assistance also
contributed to high and rapid growth. However, this momentum could not
continue during 1970's due civil war, oil price shock and
nationalisation policy. But above all, the political instability after
1970-71 has been the major cause of deterioration in Pakistan. High
level of defense spending since then is one of the critical factors,
which absorbs a significant fraction of scarce revenues and adversely
affects public savings otherwise meant for development purpose. The tax
revenues in Pakistan could not cope with faster growth in the
non-development spending.
Table 1 illustrate the rate of GDP growth and public/ private
investment and the total investment as percent of GDP.
Because of nationalisation policies during the period of
1970's, significant involvement of government in commercial
activity and increase in the share of public sector squeezed private
investment and adversely affected its growth. At that time, public
investment was twice in volume relative to private investment.
Domination of state owned/controlled institutions adversely affected the
financial sector development in Pakistan. In the decade of 1980's,
we notice some revival in private sector activity because of
encouragement and incentives provided by the government. However, due to
sever political instability during 1990's, the picture of the
economy remained gloomy. The growth rate fell from 6.2 percent in
1980's to 3.99 percent in 1990's. There was a slowing down of
public investment activity when compared to the trend level especially
in the latter part of the decade, while there was some acceleration in
the rate of private investment during 1990's relative to the
position in the 1980's decade. Political instability during the
1990's decade negatively affected the growth rate of the economy.
With the advent of 21st century, we observe some kind of revival in
growth and investment activities. Economic reforms programme such as
fiscal adjustment, privatisation of energy, telecommunication and
production, reforms in the banking and trade sectors launched in 2000,
played a vital role in the economic recovery of the Pakistan. Table 2
presents the year-wise percentage of public, private investment, total
investment and percentage of GDP from 2000-2001 to 2011-2012
respectively.
On the average GDP growth rate increased twice as compared to 1990s
decade and total investment also increased from 16.48 percent of GDP in
1990s to 18.08 percent of GDP in the early years of the current decade.
Private investment has increased overtime and public investment
relatively slowed down. Economy has grown by more than 6.5 percent per
year on the average since 2003-04. As a percentage of shares of GDP,
investment increased from 15.5 percent in 2001-02 to 20.4 percent in
2007-08, which is a healthy sign. After that it declined rapidly in
2008-09 and in 2009-10 its growth rate became negative both in public
and private investment. In 2011-12 its rose slightly due to some
increase in both type of investment due to recent election era.
The financial sectors reforms after 1990 have shown a positive
impact on the degree of interest rate liberalisation, moderate reduction
in credit subsidies and progress towards the market-based transactions.
However, because of high rate of inflation, the interest rate on
deposits became negative in real term and discouraged the financial
saving [Hassan (1997)]. To finance expenditures, the governments (both
democratic and authoritative) have to rely heavily on external and
internal borrowing and deficit financing. This practice has resulted
into high stakes of debt and high inflation, which has increased debt
servicing. The rising interest rate burden along with high defense
spending together absorb about two-third of gross revenues.
Consequently, nothing is left for the development budget and provision
of social services like health and education. The political conditions
deteriorated during 2007-08 and the new democratic government that took
over in March 2008, has to face a lot of challenges both on the internal
and external fronts. The rate of investment has surely slowed down
during 2008 and 2009 due to the terrorist activities and shortage of
electricity and gas for the industrial sector. The practice of out-wards
looking policies on part of the government continues as usual and the
prospects of growth and development depend heavily on the availability
of foreign aid and assistance. 2012 has passed on the dream of
self-sustaining growth and investment is yet far from turning into
reality.
4. DATA, MODEL AND METHODOLOGY
Most of the data is retrieved from the International Financial
Statistic (IFS) Yearbook published by International Monetary Fund (IMF).
The data on some variables is collected from various issues of Pakistan
Economic Survey compiled by the Federal Bureau of Statistics, Government
of Pakistan and from the Annual Reports of the State Bank of Pakistan.
All the data is expressed in million rupees except the Credit-to-GDP
ratio, inflation rate, exchange rate and lending rate.
4.1. The Model
The link between private, public investment and economic growth is
examined by the researchers like Ibrahim (2000). The relationship may be
expressed as under in somewhat modified form:
[Y.sub.t] = f([I.sub.pt], [I.sub.gt], [Cred.sub.t], [l.sub.rt]) (1)
Where Y = real GDP, Ig = public investment, Ip = private
investment, lr =lending rate, Cred = ratio of private sector credit to
GDP. Theoretically both types of investments are positively related to
the GDP but empirically it depends on the efficiency and productivity of
investment. Private sector credit and lending rate is also included in
the function as it affects the private investment directly and also the
growth rate of GDP indirectly since the availability of easy credit
provides incentives to private investors, which increases the growth
rate of GDP. Similarly, an increase in the real interest rate increases
the cost of borrowing and thus discourages new investment and growth of
GDP.
Public investment is mainly determined by foreign aid and
government revenue. It also depends on GDP. We expect positive
coefficients of these three variables. Exchange rate and inflation rate
also influence the public investment negatively. Following Rahman
(2008), we specify following public investment function as under:
[I.sub.gt] = f ([Y.sub.t], [Aid.sub.t], [er.sub.t], [Gr.sub.t],
[Inf.sub.t]) (2)
Where the symbols stand for: Aid = foreign aid, er = exchange rate,
Gr = Government Revenue, Inf= inflation rate.
GDP plays an important role in determining private investment. The
investment decisions are affected by domestic credit available to
private sector, lending rate and inflation, while public investment may
also include as explanatory variable to capture the "crowding
out" or "crowding in" effect on private investment.
Following Khan and Khan (2007), we specify the private investment
function as follows:
[I.sub.pt] = f ([Y.sub.t], [Cred.sub.t], [Ig.sub.t], [lr.sub.t],
[er.sub.t], [Inf.sub.t]) (3)
The above three functions can be written in a testable form as:
ln[Y.sub.t] = [[alpha].sub.0] + [[alpha].sub.1] ln[I.sub.pt] +
[[alpha.sub.2] ln[I.sub.gt] + [[alpha].sub.3]ln[Cred.sub.t] +
[[alpha].sub.4] [lr.sub.t] + [u.sub.t] (4)
ln[I.sub.gt] = [[beta].sub.0] + [[beta].sub.1] ln[y.sub.t] +
[[beta].sub.2] ln[Aid.sub.t] + [[beta].sub.3] ln[Gr.sub.t] +
[[beta].sub.4] ln[er.sub.t] + [[beta].sub.5] [Inf.sub.t] + [v.sub.t] (5)
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (6)
The terms u, v and w are the stochastic/error terms as usual.
4.2. Econometric Methodology
The above model will be estimated in three steps. First, using the
Augmented Dickey Fuller (ADF) unit root tests and assuming individual
time series as non-stationary, we examine the time series properties of
the data. Second, conditional to the results of the unit root test, we
check co-integration between the variables specified in each equation
using the method proposed by Johansen (1988) and Johansen and Juselius
(1990). Third, based on the results of the long-run co-integration
parameters, we will estimate the short-run error-correction models of
each equation.
4.2.1. Co integration Analysis
Let's we have an endogenous variable of kth order, which can
be written in a vector error correction model (VECM) as follows:
[Y.sub.t] = [[pi].sub.0] + [[pi].sub.1] [Y.sub.t-1] +
[[pi].sub.2][Y.sub.t-2] + ... ... ... + [[pi].sub.k][Y.sub.t-k] +
[v.sub.t] (7)
Where [Y.sub.t] is a (pxl) random time series vector (the variables
with order of integration of at most one are denoted by 1 (1), [PI]
represents the vector of constant term and [v.sub.t] is the vector of
error term which is I(0) and distributed with (0, [[sigma].sup.2]).
Defining [DELTA] = 1-L, where L is the lag operator, the dynamics of the
error correction model (ECM) is deduced as follows:
[DELTA] [Y.sub.t] = [[pi].sub.0] + [k-1.summation over (i=1)]
[[GAMMA].sub.i] [DELTA] [Y.sub.t-1] + [PI] [Y.sub.t-k] + [v.sub.t] (8)
[[GAMMA].sub.i] = - (I - [[PI].sub.l] - .... - [[PI].sub.i])
[v.sub.t] i = 1, 2, 3 (9)
where [PI][Y.sub.t-1] is a (px p) matrix of parameters, the rank of
which contains information about long-run relationships among the
variables in the model. If [PI][Y.sub.t-1] has full rank p, all elements
in [Y.sub.t] are stationary. If the rank of [PI] is zero, the model
reduces to VAR in the first-differences. When 0 < rank < p , there
exist co-integrating relationships equal to the rank. In this case there
exist (pxr) matrices [alpha] and [beta]. If the individual series is
I(1), then the first differences of the series are stationary. If there
is co-integration relationship between I(1) series, then the linear
combination of these variable is I(0), so that the [[PI].sub.t][Y.sub.t]
term is stationary.
To test whether there exists co-integration between the variable or
otherwise, two test statistics are used, which determine the rank of
co-integration space. One is the likelihood ratio test based on the
maximum Eigen value ([[lambda].sub.max]) of the stochastic matrix and
the second test is the value of the likelihood ratio test based on the
trace of the stochastic matrix ([[lambda].sub.trace]). The likelihood
ratio test statistics developed by Johansen are given below:
LR [[lambda].sub.trace] = - T [n.summation over (t=r+1)] ln (1 -
[[lambda].sup.[??].sub.1] (10)
Where [[lambda].sub.t+1], [[lambda].sub.t+2] ..., [[lambda].sub.n]
are the n-r smallest eigen-values and T stands for number of
observations.
LR [[lambda].sub.max] = - T ln (1 - [[lambda].sup.[??].sub.t+1])
(11)
The first statistics ([[lambda].sub.max]) tests the null hypothesis
that there are less than or equal to "r" co-integrating
vectors against the general alternative where "r" is the
number of co-integrating relations. The second statistics
([[lambda].sub.trace]) tests the hypothesis that there are "n"
numbers of co-integrating vectors against the alternative of r+1.
4.2.2. Short-run Analysis of the Variables
The short run dynamics are examined using the error correction
mechanism (ECM), the ECM is important for many reasons. It is a
convenient model, which is formulated in term of first differences. It
measures the correction from disequilibrium of the previous period. ECM
eliminates trend from the variables and resolves the problem of spurious
regression. This model follows the general to specific approach in
econometric modeling. By definition of co-integration disequilibrium,
the error term is stationary. Two variables are co-integrated implies
that there is some adjustment process which prevents the error into the
long-run relationship. Thus the concepts of co-integration and the error
correction mechanism (ECM) are closely related.
We formulate the error correction models for the real GDP, public
investment and private investment respectively as follows:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (12)
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (13)
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (14)
Where [DELTA] is the difference operator and [ECM.sub.t-1] is an
error correction term. The expected signs of the parameters [gamma],
[sigma] and [delta] should be negative, which will measure the speed of
adjustment towards long run equilibrium.
5. EMPIRICAL RESULTS
We examine the order of integration using Augmented Dickey Fuller
(ADF) unit root test. All variables, except the lending rate and
inflation are in log form. Table 3 reports the results.
It can be seen from above that all the variables are non-stationary
at their levels but stationary at their first differences.
5.1. The Long Run Growth Function
To examine the co integration between real GDP and its determinants
we use multivariate co-integration test. Two lags were selected on the
basis of Akaike information criterion (AIC). By applying the two stage
likelihood ratio tests the number of co-integrating vectors is
investigated. We follow the degree of freedom adjustment method (1) due
to Cheung and Lai (1993) for trace and max statistics. The results are
reported in Table 4 below.
The maximum Eigen-values test ([lambda]-max) indicates the
existence of two co-integrating vectors, while the trace statistics
([lamdba]-trace) indicates the existence of three co-integrating vectors
at the 5 percent level of significance. However, when we use the
adjusted max and the adjusted trace statistics, it is indicated that
there are one and three co-integrating vectors respectively included in
the model.
The long run output function (real GDP) is obtained by normalising
the first co-integrated vector on the growth rate. The results of long
run relationship are reported in Table 5 below.
It is evident from the table that in long run public investment
exerts negative impact on the growth rate of GDP. This is because
government is mainly investing in the sectors, which are unproductive
and inefficient. This result is line with Ghani and Din (2006). On the
other hand, private investment positively affects the GDP in long run
and enhances the growth rate. This result confirms the findings of Khan
and Sasaki (2002) and Ghani and Din (2006). The coefficient of private
sector credit relative to GDP is positive but insignificant. The lending
rate has negative and insignificant impact, which reflects that economic
growth is not much responsive to lending rate.
5.2. The Long Run Public Investment Function
The estimated results are quoted in Table 6. Two lags were selected
on the basis of Akaike information criterion (AIC). The likelihood ratio
statistics for ([lambda]-max) indicates the existence of five
co-integrating vectors where the ([lambda]-trace) indicates the
existence of six co-untegrating vectors at 5 percent level of
significance. By using the test with degree of freedom adjusted, the max
statistics indicates existence of four co-integration while the trace
statistics shows six co-integrating relationships in the model.
The long run public investment function is obtained by normalising
the first co-integration vector on public investment. The results are
reported in Table 7 below.
The above results indicate that the real GDP has positive and
significant impact on public investment. It confirms the theoretical
relationship of these two variables as implied by accelerator model. The
foreign aid is important but to a limited extent so far as public
investment in Pakistan is concerned. This result hardly supports the
findings of Rahman (2008) in the case of SAARC countries and the Blejer
and Khan (1984) that inflow of foreign capital positively affects the
investment rate. This is because the flow of foreign aid has been
irregular and too much fluctuating during the period of study. The
exchange rate shows negative but insignificant impact on public
investment. An increase in exchange rate makes imported goods relatively
expensive which is likely to compress investment. On the other hand, the
government revenue has (surprisingly) a negative and significant impact.
This could be explained by the fact that government revenue is merely
used to finance current expenditure of the government and seldom
available for development purposes [Rahman (2008)]. The inflation rate
exerts a negative and significant impact on public investment because an
increase in inflation leads to increase the nominal interest rate as
well as the cost of raw material and machinery/equipment.
5.3. The Long Run Private Investment Function
The co-integrating relationship between private investment and its
determinants based on Johansen co-integration test, is presented below
in Table 8. The model includes unrestricted intercept and no trend. Two
lags were selected on the basis of Akaike information criterion (AIC).
The likelihood ratio statistics ([lambda]-max) indicates the
existence of six co-integrating vectors while ([lambda]-trace) indicates
the existence of seven co-integrating vectors at 5 percent level of
significance. By using the degree of freedom adjusted test statistics,
the max-test indicates the existence of two co-integrating vectors and
the trace-statistics indicates that of five co-integrating vectors. Thus
the estimated results confirm the existence of long-run relationship
among the variables concerned. The long-run private investment function
is obtained by normalising the estimated co-integrated vector on the
private investment function. The results are reported in Table 9.
As revealed from the above, the effect of real GDP is positive but
statistically insignificant, showing weak accelerator. This finding is
consistent with Blejer and Khan (1984), Naqvi (2003), Ahmed and Qayyum
(2007) and Khan and Khan (2007). Surprisingly, the coefficient of
private sector Credit-to-GDP ratio has negative and significant impact
on private investment. This may be explained by the factual position
that credit was extended mainly to sick units who used the funds to
repay their outstanding loans to the banks [Khan and Khan (2007)]. The
negative and significant values of lending rate and inflation confirm
the theoretical relationship between these variables and private
investment. Likewise, an increase in the rate of inflation leads to
enhance the prices of raw material, machinery and equipment as well as
the wage bill, which discourage private investment. Same is the case
with exchange rate since depreciation of domestic currency definitely
increases the cost of imported goods. The public investment has negative
and significant impact on private investment, which implies the
"crowding out" effect. The results is consistent with findings
of Ghani and Din (2006), Khan and Sasaki (2001), Khan and Khan (2007)
and Majeed and Khan (2008).
5.3.1. The Short-run Growth Function
The results show that three regressors are important in
establishing the short run relationship with the growth rate of GDP and
the remaining two variables, being insignificant, are dropped front the
model following the general to specific methodology. The change in
private investment lagged by one year ([DELTA][LI.sub.pt-1]), current
public investment ([DELTA][LI.sub.gt]) and a dummy included for
uncertainty ([UN.sub.t]) (2) are significant variables while other
variables like the credit-to-GDP ratio and lending rate are proved to be
insignificant. The results are given below in Table 10.
The estimated error correction coefficient ([ECM.sub.t-1]) (3)
is--0.0058 has theoretically correct negative sign and significant at 5
percent level. In short run private investment positively and
significantly affects the growth rate of GDP, likewise public investment
is positive and significant, thereby indicating a strong impact on the
growth of GDP. The reason is that in short run it stimulates the demand
in some extant but in long run its effect dampen. The estimated
coefficient of uncertainty is negative which indicates that
macroeconomic instability and uncertainty has always depressed economic
growth in Pakistan. The estimated model passes different diagnostic
tests, such as ARCH test for serial correlation (F-statistics: 0.244,
probability: 0.784) and White test for Hetroscedasticity (F-Statistics:
2.21, probability: 0.669).
5.3.2. Short-run Public Investment Function
Estimated results for the short run relationship between the public
investment and its determinants like real GDP, foreign aid, exchange
rate, government revenue and inflation rate shows that all variables are
insignificant in the short run except changes in public investment
lagged by one year ([DELTA][LI.sub.gt-1]), current inflation rate
([DELTA]Linf) and lending rate ([DELTA]lr). These three variables show
significant short-run relationship with public investment. The results
are presented on Table 11.
The estimated coefficient of ECM shows that approximately 59
percent of disequilibrium in the public investment is instantly
corrected. The coefficient of lagged government investment is
significant and has positive sign, which indicates that changes in
previous period's public investment positively affect the short-run
changes in current public investment. The changes in inflation rate and
exchange rate exert significant and negative impacts on current public
investment. The estimated model passes different diagnostic tests, such
as ARCH test for serial correlation (F-statistics: 0.163, probability:
0.84) and White test for Hetroscedasticity (F-Statistics: 1.03,
probability: 0.44).
5.3.3. The Short-run Estimation of Private Investment Function
The results show that the variables significant in determining
changes in private investment include changes in public investment
lagged by one year ([DELTA][I.sub.gt-1]), changes in lending rate lagged
by one year ([DELTA][Ir.sub.t-1]) and current inflation rate
([DELTA]Linf). The remaining variables are insignificant in the
short-run. The results are presented below in Table 12.
The estimated error coefficient is -0.0117, with theoretically
correct sign and significant at 5 percent level. The coefficient of
lagged public investment is positive and significant, indicating a
positive effect on current private investment. The coefficients of
inflation rate and lending rate are significant. The negative signs
confirm the theoretical relationship that these variables negatively
affect private investment in the short-run. The estimated model passes
different diagnostic tests, such as ARCH test for serial correlation
(F-statistics: 0.355, probability: 0.703) and White test for
Hetroscedasticity (F-statistics: 2.13, probability: 0.069).
6. CONCLUSION
The study has attempted to evaluate the inter-relationship among
the three macrovariables, namely public and private investment and GDP
growth both in the long and short run with reference to Pakistan-
economy. We have tried to pinpoint the important determinants of each
variable, using the standard econometric techniques. As expected, the
GDP growth has a strong positive relationship with public and private
investment and there is a two-way causality between GDP and investment.
The public investment is affected by the level of GDP, inflation and
exchange rates. Likewise, private investment is affected by inflation
and exchange rates, the lending rate, besides the level of GDP. The
general negative theoretical relationship between public and private
investment is confirmed in the context of Pakistan economy, i.e. public
investment exerts a "crowdingout" effect on private investment
at large. This is because public investment has primarily been financed
in the past through internal and external borrowing. The government
revenues collected through taxation has little contribution in promoting
public investment.
Comments
Paper is a good contribution in the field of macroeconomics. It
shows the causal link among growth, private investment and public
investment. One of the most interesting result came out of the study is
that public investment crowds out private investment which is contrary
to the previous studies which shows crowding in hypothesis, i.e., public
investment crowds in private investment. After this paper government
should stay away from investing in all those sector where there is a
chance for private sector to invest. However to improve the paper
authors may need to revise it according to the following comments.
(1) As far as growth modelling is concerned. Why do we neglect
Harrod-Domer Model while studying the link between investment and
growth? Paper should include Harrod-Domer model and other theories
related to effect of private and public investment on growth.
(2) Paper used data till 2008 so they need to update it.
(3) They have written some blunt statement in the paper such as
"public and private investment would have same effect on GDP".
If this is true then how come there is crowding out, which shows that
public investment crowds out private investment. If author's
statement is true then crowding out does not have any impact on the GDP.
(4) As per literature, inflation variability and exchange rate
variability effects investment decision instead of just inflation and
exchange rate movements. Thus it would be good to use these two
variables as well in the model.
(5) Paper has used Johansen approach of cointegration. Model
explained in the paper does not represent the Johansen model since it
takes all the variables as endogenous variables. Otherwise there might
be problems of simultaneity and identification. Authors need to check
before sending the revised version of the paper since it is a big
mistake they are doing while estimating their parameters. Johansen
method applies on the system of equations where all the equations are
exactly identified.
M. Ali Kemal
Pakistan Institute of Development Economics, Islamabad.
REFERENCES
Ahmad, Imtiaz and Abdul Qayyum (2007) Do Public Expenditure and
Macroeconomic Uncertainity Matter to Private Investment? Evidence from
Pakistan. The Pakistan Development Review 46:2, 145-161.
Barro, R. J. (1991) Economic Growth in a Cross Section of
Countries. Quarterly Journal of Economics 106, 407-444.
Barro, R. J. and X. Sala-i-Martin (1999) Economic Growth.
Cambridge, MA: The MIT Press.
Blejer, M. and Mohsin S. Khan (1984) Government Policy and Private
Investment in Developing Countries. IMF Staff Papers 31:2, 379-403.
Blejer, M. I. and Mohsin S. Khan (19484) Government Policy and
Private Investment in Developing Countries. IMF Staff Papers 31:2.
Bukhari, S. A. A. S. Ali, L. and Sadaqat (2006) Public Investment
and Economic Growth in the Three Little Dragons. Evidence from
Heterogeneous Dynamic Panel Data. Karachi: Applied Economics Research
Centre, University of Karachi, Pakistan.
Cheung, Y. W. and K. Lai (1993) A Fractional Cointegration Analysis
of Purchasing Power Parity. Journal of Business and Economic Statistics
11, 103-112.
Erden, L. and H. R. (2005) The Effect of Public Investment on
Private Investment in Developing Economies. Public Finance Review 133:5,
575-602.
Fischer, S. (1993) The Role of Macroeconomic Factors in Growth.
National Bureau of Economic Research. (Working Paper No. 4565).
Ghani, E. and Musleh-ud Din (2006) The Impact of Public Investment
on Economic Growth in Pakistan. The Pakistan Development Review 45:1,
87-98.
Grier, K. and G. Tullock (1989) An Empirical Analysis of
Cross-National Economic Growth: 1951-1980. Journal of Monetary Economics
24:2, 259-276.
Hassan, P. (1997) Learning from the Past: A Fifty Year Perspective
of Pakistan's Development. The Pakistan Development Review 36:4,
355-402.
Ibrahim, M. H. (2000) Public and Private Capital Formation and
Economic Growth in Malaysia. HUM Journal of Economics and Management
8:1, 21-40.
Johansen, S. (1988) Statistical Analysis of Cointegration Vectors.
Journal of Economic Dynamics and Control 12, 231-254.
Johansen, S. and K. Juselius (1990) Maximum Likelihood Estimation
and Inference on Cointegration with Application of Demand for Money.
Oxford Bulletin of Economics and Statistics 12, 169-210.
Jongwanich, J. and A. Kohpaiboon (2008) Private Investment: Trends
and determinants in Thailand. World Development 36:10, 1709-1724.
Khan, Ashfaque H. (1988) Macroeconomic Policy and Private
Investment in Pakistan. The Pakistan Development Review 27:3, 277-291.
Khan, M. T. and Sasaki Komei (2001) Roles of Public Capital in
Pakistan s' Economy: Productivity, Investment and Growth Analysis.
Rural Urban Development Studies 13, 143-161.
Khan, Sajawal and Khan Arshad (2007) What Determines Private
Investment? The Case of Pakistan. (P1DE Working Papers 2007:36).
Levine, Ross and David Renelt (1992) A Sensitivity Analysis of
Cross-Country Growth Regressions. American Economic Review 82:4,
942-963.
Looney and Frederiken (1995) Privatization of Public Sector
Enterprises In Pakistan: Prospects For Reducing Regional Imbalances.
Review of Urban and Rural Development Studies 7, 143-152.
Looney, R. E. and Peter C. Frederic (1997) Government Investment
and Follow-on Private Sector Investment in Pakistan. Journal of Economic
Development 22:1, 91-100.
Lucas Jr., R. E. (1988) On the Mechanics of Economic Development.
Journal of Monetary Economics 22, 3-42.
Majeed, T. M. and S. Khan (2008) The Determinants of Private
Investment and the Relationship between Public and Private Investment in
Pakistan. NUST Journal of Business and Economics 1:1, 41-48.
Mankiw, G. N., D. Romer, and D. Weil (1992) A Contribution to the
Empirics of Economic Growth. Quarterly Journal of Economics 107:2,
407-437.
Naqvi (2003) Is Public Capital More Productive than Private
Capital? Macroeconomic Evidence from Pakistan, 1965-2000. School of
Finance and Business University of Durhan. (Working Paper in Economics
and Finance No. 03/03).
Pakistan, Government of (Various Issues) Economic Surveys.
Islamabad: Ministry of Finance.
Phillips, P. C. B. and Bruce Hansen (1990) Statistical Inference in
Instrumental Variables Regression with 1(1) Processes. Review of
Economic Studies 57, 99-125.
Rahman, M. M. (2008) Macro-economic and Trade Link Models of SAARC
Countries: An Investigation for Regional Trade Expansion. Journal of
Asia Pacific Economy 13:1, 50-62.
Rebelo, S. (1991) Long-Run Policy Analysis and Long-Run Growth.
Journal of Political Economy 94:5, 56-62.
Roberto, M.S. (2005) An Estimation of Residential Water Demand
Using Co-Integration and Error Correction Techniques. (MPRA Paper 615),
ppl-4.
Romer, Paul M. (1986) Increasing Returns and Long-Run Growth.
Journal of Political Economy 94:5 (October), 1002-1037.
Shahbaz, et al. (2008) Does Foreign Remittances Reduce Government
Spending? Longand Short Run Phenomenon. International Research Journal
of Finance and Economics ISSN 1450-2887 Issue 14.
(1) Cheung and Lai (1993) method is used to scale up the Johansen
Critical Value by the factor (T-K/T), where T indicates the number of
observations and K stands for the number of variables used in the study.
(2) A dummy for uncertainty is used in the short-run under the
assumption that investment decisions are likely to be affected by recent
uncertainty which is created by macro economic uncertainty.
(3) The term error correction (ECM) consists of residual obtained
from the long run output (real GDP), public investment and private
investment functions. The estimated error correction coefficient is
obtained by resetting the normalising coefficients obtained from long
run growth function.
Maryam Bint-e-Ajaz <maryabbasi@gmail.com> is Research Scholar
at University of AJ&K, Muzaffarabad. Nazima Ellahi is Assistant
Professor at Foundation University, Islamabad.
Table 1
Average GDP Growth Rate and Ratio of Public/
Private Investment to GDP Overtime
GDP Growth Public .Inv. Priv.Inv Total.Inv
Time Period (%) Ig/GDP Ip/GDP Ig+Ip/GDP
1971-80 4.78 9.44 5.32 14.76
1981-90 6.25 9.17 7.79 16.96
1991-2000 3.99 7.34 9.14 16.48
2001-2012 4.70 20.28 10.009 30.28
Source: Pakistan Economic Survey (various issues).
Table 2
Percentage of GDP Growth and Public/
Private Investment and Total Investment
GDP Growth Public .Inv. Priv.Inv Total.Inv
Time Period (%) Ig/GDP Ip/GDP Ig+Ip/GDP
2000-01 2.0 5.7 10.2 15.9
2001-02 3.1 4.2 11.3 15.5
2002-03 4.7 4.0 11.3 15.3
2003-04 7.5 4.0 10.9 14.9
2004-05 9.0 4.3 13.1 17.1
2005-06 5.8 4.7 15.7 20.4
2006-07 6.8 5.7 16.2 21.9
2007-08 3.7 5.4 15.0 20.4
2008-09 1.7 -0.34 1.4 1.06
2009-10 3.1 -1.74 -1.1 -2.84
2010-11 3.0 -0.133 0.3 0.167
2011-12 3.7 5.03 5.8 10.83
Source: Pakistan Economic Survey (various issues).
Table 3
Results for Augmented Dickey-Fuller Test of Unit Roots
ADF at First
Variables ADF at Level Difference I()(Decision)
In y 0.4973 -5.3540 I(1)
In [I.sub.g] -0.370 -6.2386 I(1)
In [I.sub.p -2.421 -4.8709 I(1)
In Cred -0.4618 -5.1788 I(1)
In Aid 0.9546 -8.7650 I(1)
In er -1.6619 -7.1341 I(1)
Lr -2.6125 -4.3394 I(1)
Inf -2.3405 -5.5965 I(1)
Note: ADF test is based on the Mackinnon (1991) critical values.
Table 4
GDP & Co-integrating Factors: Johansen Test
Maximum Eigen-values Test ([lambda]--max)
Null Alternative Test (T-K/T) Adjusted 5% Critical
Hypothesis Hypothesis Statistics Max Statistics Value
r=0 R=1 40.649 * 32.77 * 33.87
r=1 R=2 26.988 22.66 27.584
r=2 R=3 18.011 * 15.93 21.136
r=3 R=4 8.881 7.46 14.264
r=4 R=5 2.419 2.0 3.8416
Trace Test ([lambda]--trace)
r=0 R [greater 109.95 * 85.67 * 69.818
than or
equal to] 1
r=1 R [greater 67.301 * 54.9 * 47.856
than or
equal to] 2
r=2 R [greater 34.312 * 29.85 * 29.797
than or
equal to] 3
r=3 R [greater 17.300 12.492 15.494
than or
equal to] 4
r=4 R [greater 5.419 2.03 3.8414
than or
equal to] 5
Note: * Indicates significance at 5 percent level.
Table 5
Normalised Coefficients of Co-integrating Vector on Real GDP
Variables Coefficients Standard Error t-Value
In [I.sub.g] -0.785 * 0.153 5.78
In [I.sub.p 0.558 * 0.169 -4.181
In Cred 0.107 0.345 -0.227
Lr -0.0732 0.0437 0.980
Constant -9-44 -- --
Note: * Indicates significance at 5 percent level.
Table 6
Public Investment and Co-integrating Factors: Johansen Test
Maximum Eigen-values Test ([lambda]--max)
Null Alternative Test (T-K/T) Adjusted 5% critical
Hypothesis Hypothesis Statistics Trace Statistic Value
r=0 r=1 66.893 * 55.454 * 40.077
r=1 r=2 52.223 * 43.292 * 33.876
r=2 r=3 33.995 * 28.181 * 27.584
r=3 r=4 18.673 15.479 21.131
r=4 r=5 15.674 * 12.971 14.264
r=5 r=6 7.525 * 6.238 * 3.841
Trace Test ([lambda]--trace)
r=0 r [greater 194.96 * 161.62 * 95.75
than or
equal to] 1
r=1 r [greater 128.06 * 106.17 * 69.818
than or
equal to] 2
r=2 r [greater 75.842 * 62.87 * 47.856
than or
equal to] 3
r=3 r [greater 41.846 * 34.69 * 29.797
than or
equal to] 4
r=4 r [greater 23.173 * 19.21 * 15.494
than or
equal to] 5
r=5 r [greater 7.525 * 6.24 * 3.841
than or
equal to] 6
Note: * Indicates significance at 5 percent level.
Table 7
Normalised Coefficients of Co-integrating Vector on Public
Investment Function
Variables Coefficients Standard Error t-Value
In y 6.403 * 0.785 -5.585
In Aid 0.178 0.023 -0.981
In Gr -2.371 0.403 7.003
In er -0.276 0.172 1.256
Inf -0.109 * 0.013 9.654
Constant 25.134 -- --
Note: * Indicates significance at 5 percent level.
Table 8
Private Investment and Co-integrating Factors: Johansen Test
Maximum Eigen-values Test ([lambda]--max)
Null Alternative Test (T-K/T) Adjusted 5% Critical
Hypothesis Hypothesis Statistics Trace Statistics Value
r=0 r=1 97.994 * 57.995 * 46.231
r=1 r=2 87.880 * 75.995 * 40.077
r=2 r=3 43.805 * 30.391 33.876
r=3 r=4 35.395 * 20.708 27.584
r=4 r=5 21.151 * 15.630 21.131
r=5 r=6 13.992 11.225 14.264
r=6 r=7 11.078 * 7.022 3.841
Trace Test ([lambda]--trace)
r=0 r [greater 266.297 * 211.705 * 125.615
than or
equal to] 1
r=1 r [greater 186.302 * 148.74 * 95.753
than or
equal to] 2
r=2 r [greater 116.422 * 89.754 * 69.818
than or
equal to] 3
r=3 r [greater 77.617 * 66.362 * 47.856
than or
equal to] 4
r=4 r [greater 42.222 * 31.654 * 29.797
than or
equal to] 5
r=5 r [greater 18.070 * 15.193 15.494
than or
equal to] 6
r=6 r [greater 9.078 * 4.022 3.841
than or
equal to] 7
Note: * Indicates significance at 5 percent level.
Table 9
Normalised Coefficients of Co integrating
Vector on Private Investment Function
Variables Coefficients Standard Error t-Value
In y 0.375 0.437 -0.623
In Cred -0.567 * 0.038 5.946
In [I.sub.g] -1.973 * 0.387 5.271
In er -0.578 * 0.076 9.207
lr -3.262 * 0.766 5.255
Inf -0.243 * 0.026 5.946
Constant -4.761 -- --
Note: * Indicates significance at 5 percent level.
Table 10
Error Correction Model of Real (GDP)
Variables Coefficient Std. Error t-Statistic
[DELTA][ln.sub.pt-1] 0.0808 ** 0.0463 1.744
[DELTA][ln.sub.gt-1] 0.1195 * 0.0467 2.553
[UN.sub.t] -0.0035 ** 0.018 1.84
[ECM.sub.t-1] -0.0058 * 0.00085 6.861
R-squared = -0.20 Adjusted R-squared = -0.13
D.W Test = 2.32 F(4,33) = .466
Note: * Shows significance at 5 percent level
and ** shows significance at 10 percent level.
[ECM.sub.t-1] = (In [y.sub.t] + 0.785 * ln [I.sub.gt]
-0.585 * ln [I.sub.pt] -0.1017 * [lncred.sub.t] +0.0738 * lr)
Table 11
Error Correction Model of Public Investment Function
Variables Coefficients Standard Error t-Values
[DELTA][ln.sub.gt-1] 0.4102 * 0.148 2.769
[DELTA]ln inf -0.0906 * 0.035 -2.530
ln [er.sub.t] -0.842 * 0.254 -3.302
Constant -6.385 * 1.234 -5.17
[ECM.sub.t-1] -0.591 * 0.115 -5.14
R-squared = 0.54 Adjusted R-squared = 0.48
D.W stat = 2.59
Note: * Shows significant at 5 percent level
and ** shows significant at 10 percent level.
[ECM.sub.t-1] = (ln [I.sub.gt] + 2.371 * ln [G.sub.rt]
-0.195 * [lnaid.sub.t] -6.841 * [lny.sub.t] +0.10inf
* +0.276 [lner.sub.t])
Table 12
Error Correction Model of Private Investment Function
Variables Coefficients Standard Error t-Values
[DELTA][ln.sub.gt - 1] 0.00015 ** 7.78 E-05 1.936
[DELTA][lr.sub.t-1] -0.0438 * 0.0179 -2.438
[DELTA]ln inf -0.0892 * 0.0339 -2.630
[ECM.sub.t-1] -0.0117 * 0.0029 -3.952
R-squared = 0.37 Adjusted R-squared = 0.31
D-W Test = 2.421
Note: * Shows significant at 5 percent level and ** shows
significant at 10 percent level.
[ECM.sub.t-1] = [lnI.sub.pt] -0.375 * [lnY.sub.t-1] +5.567 *
[lncred.sub.t] +3.262 * lr +1.973 * [lnI.sub.gt] +0.243 * lninf
