Abstract
This paper investigates the determinants of private investment in Pakistan with special emphasis on the impact of government investment. Using annual data for the period 1973/74-1991/92, it is estimated that private investment was positively correlated to GDP growth, to credit extended to the private sector, and to government investment. When government investment is disaggregated into Its infrastructural and noninfrastructural components, the latter is found to be negatively correlated with private investment.
I. Introduction
Since the late eighties, Pakistan has been Implementing a comprehensive structural reform program with emphasis on private investment as an engine of growth. The program also stresses the need for increased public investment in infrastructure–such as roads, power projects and telecommunications–to improve the environment for private investment. In this context, the present paper attempts to analyze the main determinants of private sector investment in Pakistan, including the impact of public investment. It is organized as follows. Section II reviews the theoretical and empirical literature on investment determination, especially in developing countries. Section III examines the case of Pakistan and presents the results of a quantitative analysis of private investment behavior in the period 1974/75-1991/92. 1/ Section IV summarizes the findings of the paper.
II. The Determinants of Investment: A Selected Overview
1. Traditional theories of investment
Past analyses of the determinants of investment have tended to start essentially from four basic models: the accelerator, profit, neoclassical, and Tobin’s q models. In the fixed accelerator approach, once an increase in output is anticipated, the capital stock has to be increased consistent with the new level of output. Therefore, in this model, investment is determined as the sum of the difference between the existing and the desired capital stock, and the replacement needed to substitute the depreciation of the existing stock (Clark, 1917). According to the flexible version of this model, the desired increase in capital does not occur instantaneously. The speed of adjustment depends on various factors such as the level of capacity utilization. This is integrated in the original model by adding distributed lags to account for the delay in adjustment (Chenery, 1952, and Koyck, 1954). While the accelerator approach emphasizes the role of demand, the profit approach explicitly emphasize the role of profit in determining investment. According to this approach, profits are the major incentive for investment and they also play an important part in contributing to and facilitating internal and external financing.
According to the neoclassical approach, investment should take into account both anticipated earnings and the cost of capital. Therefore, optimal accumulation of capital (i.e., Investment) is achieved when the integral of the discounted expected net revenues is maximized (Jorgenson, 1963). The fourth approach is Tobin’s q model which suggests that firms invest as long as the increase in the value of their shares is higher than the increase in the replacement cost of their physical assets (Tobin, 1969). 1/ The q model has been regarded as both a modified version of the neoclassical model (Hayashi, 1982) and as a profit model because of its emphasis on the role of profitability. Indeed, this highlights a more general point–i.e., all of the above theories are not necessarily mutually exclusive. For example, in constructing investment models for developed countries, these models are usually integrated in what has been described as neoclassical-flexible accelerator models (Catinat et al, 1987).
2. Determinants of investment in developing countries
The above mentioned approaches have provided a basis for econometric analyses of investment determination in industrial countries, with the neoclassical-flexible accelerator models being most widely applied and empirically supported. However, they have appeared less successful in the case of many developing countries where assumptions regarding capital markets, and the way in which the economy as a whole operates differ significantly from the case of industrial countries. 2/ Therefore, studies on developing countries have tended to incorporate specific investment determinant factors, including most often, financing availability, and the role of government investment. Other factors–such as external Inflows, the size of external debt, market structures, the level of protection, and the degree of price distortions–have also been explored. As recent empirical studies have shown, once modified to accommodate these considerations, the classical models of investment are applicable to developing countries (e.g., Sundararajan and Thakur, 1980; Tun Wai and Wong, 1982; and Blejer and Khan, 1984). In these studies, however, the degree of required modification have varied from minimal to fundamental. The following discussion aims to provide a general review of several of these studies, focusing briefly on the major factors affecting investment in developing countries, many of which can be integrated within a quantitative model of investment behavior. 1/
a. Financing availability
Compared to industrial countries, the financing constraint is, in general, far more binding in developing countries. Self financing is less available due to two main reasons: first, many firms in developing countries are inchoate and, therefore, have a limited ability to accumulate adequate financing. Second, low per capita income is often accompanied by low savings leading to fewer resources being available to finance investment. Not surprisingly, some empirical studies have found a positive relation between investment and higher per capita income (e.g., Greene and Villanueva, 1991). The financing constraint is exacerbated due to the underdevelopment of capital markets. Capital markets are viewed not only as helping in mobilizing and channelling savings into productive investment, but also as enhancing efficiency and competitiveness (WIDER, 1990). 2/
In light of the above considerations, McKinnon (1973) and Shaw (1973) suggest that accumulated real money balances are an important determinant of investment in developing countries–implying a positive relationship between real interest rates and investment. However, some other studies (e.g., Bernanke (1983) and Greene and Villanueva (1991)) have found that the overall impact of high real interest rates is likely to be negative because of its significant impact on unit cost. However, since financial markets have tended to be repressed in several developing countries–albeit to a diminishing extent recently–investment may be expected to be less sensitive, ceteris paribus, to interest rates as compared to the availability of bank credit. This is further accentuated by the large import component of investment expenditure in developing countries which makes bank credit essential to substitute for advance import deposits (Tun Wai and Wong, 1982).
b. Foreign capital inflows
Whether they be in the form of debt flows or equities, foreign inflows can have an important impact on investment. Not only can they ease the domestic financing constraint, but they can also generate crowding in effects by creating linkages and externalities. Questions have arisen, however, as to the extent to which foreign direct investment may involve crowding out of domestic activities because of the comparative advantage which large multinationals tend to have over infant national industries. This issue was part of a larger debate in the sixties and early seventies concerning the “appropriateness” of foreign direct investment for developing countries. However, such concerns have diminished in recent years with greater recognition of the positive externalities that foreign direct investment has entailed, including the introduction of new technologies, upgrading of the capital stock, and enhancement of competitiveness and dynamism in the economy.
c. External debt
As foreign inflows accumulate, the issue of external debt arises in two respects. First, higher debt servicing involves a larger contractual claim on available resources. This means, ceteris paribus, that fewer resources will be available for investment. Moreover, as concern rises about declining creditworthiness, the cost of new external credit would tend to increase (Mirakhor and Montiel, 1987 and El-Erian, 1992). The second respect, leading to the “debt overhang effect,” relates to Investors’ perceptions of future returns on investment. Specifically, as debt grows significantly, investors’ risk aversion would tend to rise, reflecting concerns about the country’s ability to meet future contractual debt servicing without raising effective taxation (Corden, 1988 and Dooley et al, 1990).
d. Profitability and market structure
The role of profit was highlighted earlier both as an incentive to investment and as a source of financing. Needless to say, everything that affects profitability will affect investment in the same direction. The impact of taxes has been given special attention in the literature. This issue is important for designing an appropriate tax policy which aims at balancing the investment disincentive impact of taxation against its welfare benefits (Phelps, 1986). Nevertheless, the impact of profit taxes, per se, on investment in developing countries is believed to be less significant than in the case of industrial countries. The reasons are that taxation systems in developing countries depend mainly on indirect rather than income taxes, and, more importantly, that the impact of other investment constraints are believed to be more important.
Another important factor that affects profits is market structure. The more the market diverges from its competitive paradigm, the higher the profits are likely to be. Therefore, investment is expected to be inversely related to the degree of competition in the market. However, one should not underestimate the negative impact of monopolistic structures where output is produced at suboptimal levels and where there are other sources of inefficiency with a long term adverse impact on investment (Shafik, 1989). Moreover, it is also possible in monopolistic market structures that profits may not be abnormal if “contestability” applies due to the threat of potential entrants.
e. Uncertainty
Since expectations play a crucial role in investment decisions, economic and political instability and uncertainty can have a harmful effect. Indeed, it can be argued that the latter factors as being partially responsible for the incompatibility of the neoclassical-accelerator model in developing countries. The implications for investment behavior are straightforward. Efforts aimed at encouraging private investment need to be supported by sound macroeconomic policies and adequate regulatory and supervisory structures which maintain clarity and reasonable stability.
f. Government expenditure
The final factor that is considered here is that of government expenditure. 1/ A priori, government expenditure can have either a positive or a negative impact on private investment. Government expenditure can positively influence private investment by raising effective demand, which can lead to a rise in profitability and investment. Also, if government expenditure is geared toward investment in required infrastructure, it can help relax existing constraints, reducing transaction costs for the private sector, thereby raising expected profitability and encouraging private investment. Furthermore, government investment can crowd in private investment when it targets activities which have strong linkages with the rest of the economy (Tun Wai and Wong, 1982).
On the other hand, government expenditure can crowd out private investment both directly and indirectly. Direct crowding out takes place when public investment competes in activities within the scope of private investment and/or places claims on scarce financial resources. The latter is especially relevant when public investment is favored in allocating credit in repressed financial markets. Indirect crowding out occurs when excessive public expenditure leads to rising interest rates and increased inflationary pressures due to pressure on resources and/or inflationary finance (Friedman, 1978). Expected higher taxes to finance public expenditure and loss of confidence in public policy may act as additional factors to discourage investment. The impact of inflation may be further harmful as it raises uncertainty leading to a more volatile environment and to possible distortions in the information content of relative prices. Furthermore, financial constraints become even tighter under inflation, especially for long-term investments, as the average maturity of commercial lending is likely to be reduced (Greene and Villanueva, 1991). It is apparent that an important factor in determining the overall impact of government expenditure is its structure; the more it targets investment in infrastructure and in activities complementary to private investment, the stronger the crowding-in effect is likely to be (Aschauer, 1989). 1/
Several studies are available seeking to quantitatively investigate the impact of government investment on private investment. In a modified flexible accelerator model, Sundararajan and Thakur (1980) hypothesized that government investment affects private investment in two ways. Firstly, the stock of government investment provides infrastructure and other sources of externalities which may reduce the cost of production for private investment. Secondly, assuming that government investment competes with private investment for financing on a one to one basis, the amount of financing resources available for private investment will be total savings less government investment.
Starting from these assumptions, Sundararajan and Thakur constructed a behavioral model for investment, savings, and output for both the private and the public sectors. The explanatory variables in their private investment equation were: expected user cost of capital, the level of private output, both private and government capital stock, and finance availability for private investment, defined as total savings less government investment. Regressions were run for India (1960-76) and Korea (1958-76) with slight modifications to incorporate country specific features. The estimated coefficients of finance availability for private investment were positive and statistically significant in both cases. This result was taken by the authors as a validation of their hypothesis that government investment crowds out private investment by reducing the funds available for it. The results, however, for the impact of government capital stock on private investment were ambiguous and statistically insignificant both in the case of India and of Korea.
Tun Wai and Wong (1982) modified the above model emphasizing the role of bank credit and capital inflows. Their investment function contained three explanatory variables: government investment, finance available for private investment, and private capital stock. The equation was estimated for five countries (Greece (1960-76), Korea (1960-75), Malaysia (1961-71), Mexico (1965-75), and Thailand (1961-75)). The results generally indicated a positive impact of all the above determinants, including government investment, though statistical significance was not strong. In their discussion of the results, the authors suggested that the estimated coefficients of government investment overestimated the positive impact of government investment because they did not capture the financing crowding out effect. To address this issue, they constructed a three-equation recursive model based on the assumption of Sundararajan and Thakur that government and private investment compete one-to-one for finance with each other. In the reduced form specification, the coefficients of government investment became smaller than in the original model, consistent with a priori expectations. Their statistical significance, however, was lower.
Blejer and Khan (1984) made an important contribution by separating infrastructure and noninfrastructure government investment in analyzing determinants of private investment. The two proxies they suggested to represent infra- and noninfrastructure government investment were based on the expected or the trend values of government investment. Their final specification of the investment equation contained expected output, a proxy for cyclical factors, change in bank credit for private investment, expected government expenditure as a proxy for infrastructure government investment, unexpected government expenditure as a proxy for noninfrastructure government investment and, finally, a partial adjustment component. A regression for pooled data (1971-79) of 24 developing countries resulted in more consistent estimates than those obtained by an alternative regression based on a specification in which government investment was not separated into its two components. Furthermore, the differentiating equation had a better fit. The coefficients of expected and unexpected government investment were statistically significant and took the positive and negative signs respectively. To determine the relative importance of the explanatory variables, the beta coefficients were estimated and it was found that government investment in infrastructure was the second most important determinant of private investment (the first being change in real income). When government investment was included as an aggregate figure, its coefficient became statistically insignificant. Blejer and Khan concluded that this was because of the offsetting impact of crowding in and crowding out effects.
Greene and Villanueva (1991) used more recent data (1975-87) for 23 developing countries. In their specification, they placed explicit emphasis on the role of inflation and the rate of interest. Consequently, the ratio of private investment to GDP was specified as a function of the real interest rate, change in GDP per capita, change in the consumer price index (CPI), the level of per capita income, and external debt service to exports and external debt to GDP ratios. A country dummy variable vector was incorporated to allow for country specific factors. Their results showed that the ratio of government investment to GDP had a positive impact and both the real interest rate and inflation had a negative impact. All estimates were statistically significant. The estimation was also carried out for two subperiods, 1975-81 and 1982-87, to allow for changes due to the debt crisis. The F-test indicated a significant difference between the two periods. However, the results remained similar apart from the fact that the coefficient of real interest rate became statistically insignificant. In contrast with Blejer and Khan, the beta coefficient estimates indicated that, in importance, the impact of government investment (presented by the ratio of government investment to GDP) came last.
It is clear from the above cited studies that the relationship between private and government investment is still open to debate, as empirical findings have produced mixed results. However, it is possible to deduce the sources of such variation. It seems that both model specification and the time period and the countries considered in each study play an important role in shaping the results. This implies that the specific characteristics of the economy under consideration should be carefully taken into account in anticipating the impact of government investment. In addition, when constructing a model, It is important to differentiate between different types of government investment. As shown earlier, this improves the statistical significance, as such models better accommodate the characteristics of developing countries, as well as the theoretical points discussed earlier regarding the mixed impact of government investment.
III. The Case of Pakistan
Since the mid-1970s, the Pakistan economy has maintained relatively high growth rates with an annual average of some 6 percent in real terms (Chart 1). Investment in Pakistan has not kept up with GDP growth (Chart 2). Thus, while the average growth rate of real GDP in Pakistan in the eighties exceeded that of low income economies as a group (6.3 percent compared to 6.1 percent), its average growth rate of real gross domestic investment during the same period was considerably less than the group’s average (5.7 percent compared to 7.4 percent).
Pakistan Real GDP
(Percentage changes)
Citation: IMF Working Papers 1993, 030; 10.5089/9781451844863.001.A001
Pakistan Real Gross Fixed Capital Formation
(In billions of Pakistan rupees at 1980/81 prices)
Citation: IMF Working Papers 1993, 030; 10.5089/9781451844863.001.A001
With the fiscal Imbalance affecting the expansion in government investment, increased reliance has been placed on the private sector to provide a major part of the required increase in overall investment. The share of private investment fell in the 1970s In the context of the nationalization policy, reaching its lowest level (about 38 percent) In the middle of the decade. It then surged again in the eighties, with the gradual liberalization, to reach an estimated 52 percent in 1991/92 (Chart 3), The potential for a further expansion is encouraging, given the recent response of private sector investment activities to the authorities’ implementation of wide-ranging structural reform policies.
Pakistan Private Investment as Percent of Total Investment
(In percent)
Citation: IMF Working Papers 1993, 030; 10.5089/9781451844863.001.A001
Few attempts have been made to empirically investigate the determinants of private investment in Pakistan. Moreover, in macroeconomic studies that have been undertaken, only partial attention has been given to private investment. Notable efforts in this regard are those of Haque and Montiel (1991), and Haque, Husain, and Kontiel (1991).
In the context of formulating a macroeconomic model for Pakistan, Haque and Montiel (1991) estimated a cointegration equation for the period from 1972/73 to 1987/88 taking the private capital stock as the dependent variable, and the rental cost of capital and the government capital stock as the explanatory variables. Their results indicated an inverse relation between private investment and the rental cost of capital and a positive relation between it and the government capital stock. Both estimated coefficients were statistically significant.
Haque, Husain, and Montiel (1991) specified private investment as a function of both private and government capital stocks, the level of private output, and the change in credit to the private sector. However, since they were interested in the intersectoral allocation of investment, they estimated a separate equation for each of the tradable and nontradable sectors. Consequently, to allow for the intersectoral relative price incentives, they added to the explanatory variables the real exchange rate, taken as the ratio between the price of nontradables to the price of tradables. One of their most significant findings is a positive correlation between private investment and the government capital stock, and between it and available credit. The positive impact of credit was especially significant in the case of private investment in tradables.
Specification and results
In light of the above discussion and the findings of earlier studies, the following equations were specified and estimated for Pakistan using annual data for the period 1974/75 to 1991/92,
(1) IP - b0 + b1 ΔGDP + b2 ΔCRD + b3 IUG
and
(2) IP - b0 + bl ΔGDP + b2 ΔCRD + b4 GINF(-2) + b5 GNON
where
| IP | Is private Investment, |
| GDP | is gross domestic product, |
| CRD | is credit to private sector, |
| IG | is aggregate government investment, |
| GINF(-2) | is infrastructure government investment lagged by two years and, |
| GNON | is noninfrastructure government investment. |
| IP | Is private Investment, |
| GDP | is gross domestic product, |
| CRD | is credit to private sector, |
| IG | is aggregate government investment, |
| GINF(-2) | is infrastructure government investment lagged by two years and, |
| GNON | is noninfrastructure government investment. |
All variables are in real terms and are expressed In natural logarithms.
The first explanatory variable in equation (1), i.e., change in GDP, reflects the accelerator component in determining investment, with actual change in GDP used as a proxy for the expected change. 1/ The second explanatory variable represents the financing constraint. Since for most of the period covered, the financial markets in Pakistan were repressed through the use of administered interest rates set well below the market clearing level, credit provided to the private sector is specified as the effective financing constraint (Chart 4). 2/ The third explanatory variable covers government investment. An alternative specification may include lagged private investment as an additional explanatory variable to account for partial adjustment. However, when estimated, the lagged variable was insignificant. In the disaggregated formulation of government investment (equation 2), proxies for infrastructure and noninfrastructure investments are derived on the basis of adjusted disaggregated national accounts data for government investment expenditures in the following sectors 3/: agriculture, mining and quarrying, manufacturing, construction, electricity and gas, transport and communication, wholesale and retail trade, financial institutions, and other services; as well as investment undertaken by federal and local authorities and bodies. 1/ Government investment in infrastructure was lagged by two years to allow for the gestation period which is usually needed before such investment produces Its externalities. 2/
Pakistan Credit to Different Sectors
(In billions of Pakistan ruppes)
Citation: IMF Working Papers 1993, 030; 10.5089/9781451844863.001.A001
A priori, expectations were that bl, b2, and b4 would have positive signs and b5 a negative sign, while the sign of b3 would depend on the strength of the crowding-out and the crowding-in effects of government investment. The poor state of infrastructure in Pakistan was expected, on balance, to be associated with a positive b3.
Before running the regressions, unit root tests were carried out to check for stationarity. From the results reported in Table 1, it is obvious that most of the variables included were stationary (at the 5 percent level). Only growth of credit and government infrastructure investment using the less strict classification were nonstationary (even at the 10 percent level). Sims, Stock, and Watson (1990) showed that having some nonstationary variables in a regression does not affect the quality of the estimated coefficients of the stationary variables in that regression. Accordingly, apart from those concerning the coefficients of the two nonstationary variables, the results obtained are otherwise robust. To confirm the findings regarding growth of credit (the series of which is nonstationary), the basic investment regression (equation 1) was rerun using the first difference of that variable (regression 3). This need not be done for the other nonstationary variable, infrastructural government investment using the less strict definition, because the estimated coefficient of infrastructural government investment using the strict definition, which is stationary, can give the indication required. The results of the regressions are reported in Table 2. The beta coefficients were also calculated to get an indication as to the relative importance of each variable in determining private investment. They are reported in Table 3. All estimated coefficients carried the expected signs, including that of aggregate government investment (IG). The coefficient of noninfrastructure government investment was negative, as expected, but statistically insignificant. The results of rerunning the basic equation using the first difference of the growth of credit confirmed those which were previously obtained using the growth of credit.
Results of Augmented Dickey-Fuller Unit Root Test for Stationarity
- MacKinnon critical values:
1 percent = -4.5743
5 percent = -3.6920
10 percent = -3.2856
Results of Augmented Dickey-Fuller Unit Root Test for Stationarity
| Variable | A-D-F Stat. | Number of Lags | Stationarity |
|---|---|---|---|
| IP | -6.8851 | 1 | at 1 percent |
| ΔGDP | -3.9000 | 0 | at 5 percent |
| IG | -4.3691 | 3 | at 5 percent |
| ΔCRD | -2.8292 | 0 | nonstationary |
| GINFs | -3.3907 | 1 | at 10 percent |
| GINFls | -2.9999 | 0 | nonstationary |
| GNONs | -4.1623 | 2 | at 5 percent |
| GNONls | -3.2648 | 3 | almost 10 percent |
| Δ2CRD | -4.9426 | 0 | at 1 percent |
- MacKinnon critical values:
1 percent = -4.5743
5 percent = -3.6920
10 percent = -3.2856
Results of Augmented Dickey-Fuller Unit Root Test for Stationarity
| Variable | A-D-F Stat. | Number of Lags | Stationarity |
|---|---|---|---|
| IP | -6.8851 | 1 | at 1 percent |
| ΔGDP | -3.9000 | 0 | at 5 percent |
| IG | -4.3691 | 3 | at 5 percent |
| ΔCRD | -2.8292 | 0 | nonstationary |
| GINFs | -3.3907 | 1 | at 10 percent |
| GINFls | -2.9999 | 0 | nonstationary |
| GNONs | -4.1623 | 2 | at 5 percent |
| GNONls | -3.2648 | 3 | almost 10 percent |
| Δ2CRD | -4.9426 | 0 | at 1 percent |
- MacKinnon critical values:
1 percent = -4.5743
5 percent = -3.6920
10 percent = -3.2856
Regressions Results
Regressions Results
| C | ΔGDP | IG | ΔCRD | GINF(t-2) | GNON | Δ2CRD | R2 | |
|---|---|---|---|---|---|---|---|---|
| (1) | 5.660 | 1.620 | 0.341 | 0.348 | .982 | |||
| (4.399) | (2.708) | (2.701) | (2.633) | |||||
| (2.1) | 5.688 | 2.250 | 0.647 | 0.364 | -0.330 | .977 | ||
| (3.813) | (3.519) | (4.704) | (2.453) | (-1.138) | ||||
| (2.2) | 10.433 | 1.419 | 0.673 | 0.350 | -0.119 | .967 | ||
| (1.697) | (1.293) | (4.130) | (2.601) | (-1.014) | ||||
| (3) | 3.793 | 1.125 | 0.471 | 0.164 | .990 | |||
| (1.539) | (2.367) | (4.024) | (2.819) |
Regressions Results
| C | ΔGDP | IG | ΔCRD | GINF(t-2) | GNON | Δ2CRD | R2 | |
|---|---|---|---|---|---|---|---|---|
| (1) | 5.660 | 1.620 | 0.341 | 0.348 | .982 | |||
| (4.399) | (2.708) | (2.701) | (2.633) | |||||
| (2.1) | 5.688 | 2.250 | 0.647 | 0.364 | -0.330 | .977 | ||
| (3.813) | (3.519) | (4.704) | (2.453) | (-1.138) | ||||
| (2.2) | 10.433 | 1.419 | 0.673 | 0.350 | -0.119 | .967 | ||
| (1.697) | (1.293) | (4.130) | (2.601) | (-1.014) | ||||
| (3) | 3.793 | 1.125 | 0.471 | 0.164 | .990 | |||
| (1.539) | (2.367) | (4.024) | (2.819) |
Beta Coefficients
Beta Coefficients
| ΔGDP | IG | ΔCRD | GINF(t-2) | GNON | Δ2CRD | |
|---|---|---|---|---|---|---|
| (1) | 0.07009 | 0.09098 | 0.19596 | |||
| (2.1) | 0.09734 | 0.16915 | 0.29078 | 0.05646 | ||
| (2.2) | 0.06139 | 0.17595 | 0.28424 | -0.05627 | ||
| (3) | 0.04867 | 0.27066 | 0.04856 |
Beta Coefficients
| ΔGDP | IG | ΔCRD | GINF(t-2) | GNON | Δ2CRD | |
|---|---|---|---|---|---|---|
| (1) | 0.07009 | 0.09098 | 0.19596 | |||
| (2.1) | 0.09734 | 0.16915 | 0.29078 | 0.05646 | ||
| (2.2) | 0.06139 | 0.17595 | 0.28424 | -0.05627 | ||
| (3) | 0.04867 | 0.27066 | 0.04856 |
Although growth of GDP had a positive and statistically significant impact on private Investment, it was, in relative terms, the least important explanatory variable according to the beta coefficients’ estimates. This supports the argument that, in contrast to developed countries, the accelerator component does not play the major role in investment determination in developing countries. The estimates also confirm the relative importance of the role of growth In the credit given to private sector. 1/ This has important policy implications especially for investment in the tradable sectors which are suggested to be the most sensitive to change in credit according to the findings of Haque, Husain, and Montiel (1991) discussed earlier. One interesting finding is that aggregate government investment has a significant positive impact on private investment. As indicated by the beta coefficients, government investment is a relatively important variable in determining private investment in all estimated equations (the second in importance in regression 1 and the first in the other regressions). This implies that the crowding out effect was not strong enough to counterbalance the crowding in effect in the period covered and is consistent with the view, inter alia, as to the importance of upgrading the country’s road and transportation networks, and improving the drainage system. 1/
Pakistan Share of Private Sector in Total Domestic Credit
(In percent)
Citation: IMF Working Papers 1993, 030; 10.5089/9781451844863.001.A001
When government investment was disaggregated (in equation 2) to estimate the impact of its noninfrastructure component, the coefficients of the latter (in regressions 2.1 and 2.2) turned out to be negative, as expected, indicating an adverse impact on private investment. However, the estimated coefficients were statistically insignificant. A possible explanation is the relative weakness of both real and financial crowding-out effects in Pakistan due to the offsetting impact, inter alia, of external financing for government expenditures, as well as the tendency for the private and public sectors to target different activities and to operate through firms of different sizes. 2/
Two important policy implications would be that when applying fiscal and monetary discipline in Pakistan care should be taken so as not to suppress credit for productive private investment, especially in the tradable sector. Moreover, also within the confines of a prudent fiscal stance, government spending priorities would need to give due weight to investment in infrastructure. Indeed, this should not be taken as an argument for unconditional increase in government expenditure, even in infrastructure. As recognized by the Pakistan authorities, public investment should be undertaken after a careful assessment of the infrastructural needs of the country and financing possibilities. Arguments in favor of stimulating private investment through greater public investment need to give due weight to other economic considerations concerning the efficiency of the allocation of scarce resources and the importance of pursuing a sound macroeconomic policy to stimulate sustained investment in the medium and long run. Moreover, as the structure of both the private and government sectors and other factors concerning available finance and interest rate change–as has occurred under the authorities’ ongoing structural reform program–it is reasonable to expect the response to government investment and the appropriate specification of the private investment function as a whole to change. For example, with the liberalization and broadening of financial markets, it may be expected that an interest rate variable would eventually substitute for the availability of credit in the investment determination function. Moreover, as external sources available for government investment and the structure of both private and public sectors evolve in the context of internal (e.g. privatization and deregulation) and external factors, the crowding-out effect may become more pronounced, reducing the complementary impact of aggregate government investment. Nevertheless, given the current circumstances, appropriate government investment in infrastructure may be expected to continue to exert a significant crowding-in effect.
IV. Concluding Remarks
Following a selected review of the theoretical and empirical approaches to investment determinants, this paper has attempted to investigate the determinants of private investment in Pakistan with special emphasis on the role of government investment. The review indicated that investment functions in developing countries have tended to differ from those specified for developed countries in that credit availability and government investment, especially in infrastructure, play a more important role in the case of the former. Within this, it was argued that government investment has a mixed impact, with an ambiguous overall contribution a priori. Indeed, given the various sources of crowding-in/out effects that were discussed, the final impact was shown to depend primarily on both the structure of government expenditures and the investment characteristics of the economy under consideration. Generally, the more government investment is geared towards infrastructural investment, the more likely that it will have a positive impact on private investment–particularly in the case of developing countries with unexploited linkage opportunities and segmented sources of domestic and external financing.
Building on previous work, a private investment function for Pakistan was specified. Its estimation using annual data for the period 1973/74-1991/92 suggested that private investment in Pakistan was positively correlated to growth of GDP, to growth of credit to the private sector, and to government investment. To further Investigate the impact of government investment, it was disaggregated into its infrastructural and noninfrastructural components. The estimation results of the disaggregated specification were similar to the initial results apart from the finding that noninfrastructural government investment appeared to be negatively correlated with private investment. In terms of relative importance of explanatory variables, infrastructural government investment was estimated to be the most important positive determinant in the period under consideration, followed by credit extended to the private sector.
In assessing the policy implications of the quantitative analysis, the paper argued for the need to take account of the partial equilibrium nature of the analysis. Specifically, it was stressed that the positive relationship between private and public investment was itself influenced by the prevailing macroeconomic framework. Thus, important consideration must be given to maintaining macroeconomic stability which is essential to stimulate sustained private investment in productive activities. It was also stressed that, in designing infrastructural government investment, careful investigation of the country’s infrastructural needs should be carried out so as not to create artificial positive externalities that would lead to wasteful allocation of resources over the longer run. Similarly, the contribution of credit to private sector investment must not seriously jeopardize monetary stability.
Finally, it has to be noted that, as in other country cases undergoing fundamental structural changes, the appropriate specification of the investment function in Pakistan is most likely to evolve with the wide-ranging changes associated with the authorities’ comprehensive ongoing program of structural reforms. For example, with further liberalization of financial markets, it is expected that an interest rate variable will eventually take the place of the availability of credit in the investment function. Similarly, with the strengthening of infrastructure over the medium term and the anticipated evolution of the structure of both private and government investments (due, inter alia, to liberalization and privatization), it is anticipated that the impact of government investment on private investment will evolve, with a shift in the relative importance of crowding-out and crowding-in effects. It is unlikely, however, that the positive impact of infrastructural government Investment will significantly decline in the next few years provided the qualifications mentioned above regarding efficiency and macroeconomic stability are not seriously violated.
References
Aschauer, D., “Does Public Capital Crowd Out Private Capital?,” Journal of Monetary Economics (September 1989) Vol. 24, pp. 171–88.
Bernanke, B., “Investment, Savings, and Incentives. The Determinants of Investment: Another Look,” American Economic Review, Papers and Proceedings (May 1983) Vol. 73, pp. 71–5.
Blejer, M., and M. Khan, “Government Policy and Private Investment in Developing Countries,” IMF Staff Papers (June 1984) Vol. 31, pp.379–403.
Catinat, M. et al, “Determinants of Investment,” European Economy. Commission of the European Community, No.31 (March 1987), pp. 5–60.
Chenery, H. B., “Overcapacity and the Accelerator Principle,” Econometrica. No. 1 (January 1952) Vol. 20, pp.1–28.
Chhibber, A., M. Dailami, and N. Shafik, Reviving Private Investment in Developing Countries, Amsterdam, North Holland Press, 1992.
Clark, J. M., “Business Acceleration and the Law of Demand: A Technical Factor in Economic Cycles,” Journal of Political Economy, No. 1 (March 1917), Vol. 25, pp.217–35.
Corden, W., “Debt Relief and Adjustment Incentives,” IMF Staff Papers, No. 4 (December 1988), Vol. 35, pp. 628–43.
Dooley, M. et al, “Debt Reduction and Economic Activity,” IMF Occasional Paper, No. 68 (March 1990).
El-Erian, M. A., “Restoration of Access to Voluntary Capital Market Financing. The Recent Latin American Experience,” IMF Staff Papers, No. 1 (March 1992), Vol. 39, pp. 175–94.
Friedman, B., “Crowding Out or Crowding In? Economic Consequences of Financing Government Deficits,” Brookings Papers on Economic Activity, No. 3, 1978.
Greene, J. and D. Villanueva, “Private Investment In Developing Countries. An Empirical Analysis,” IMF Staff Papers, No. 1 (March 1991), Vol. 38, pp.33–58.
Haque, N., A. Husain, and P. Montiel, “An Empirical “Dependent Economy” Model for Pakistan,” IMF Working Paper, 1991.
Haque, N., and P. Montiel, “The Macroeconomics of Public Sector Deficits. The Case of Pakistan,” World Bank Country Economics Department Working Paper, 1991.
Hayashi, F., “Tobin’s Marginal q and Average q: A Neoclassical Interpretation,” Econometrica, 1982.
Jorgenson, D., “Capital Theory and Investment Behavior: A Survey,” Journal of Economic Literature, (December 1963).
Koyck, L. M., Distributed Lags and Investment Analysis, Amsterdam, North Holland Press, 1954.
McKinnon, R., Money and Capital in Economic Development, Washington, The Brookings Institution, 1973.
Mirakhor, A. and P. Montiel, “Import Intensity of Output Growth in Developing Countries, 1970-85,” IMF Staff Studies of the World Economic Outlook, August 1987.
Phelps, E., “Profits Theory and Profits Taxation,” IMF Staff Papers, No. 4, Vol. 33, 1986, pp. 674–96.
Shafik, N., “Private Investment and Public Policy in Egypt: 1960-1988,” D. Phil, thesis, Oxford University, 1989.
Shaw, E., Financial Deepening in Economic Development, New York, Oxford University Press, 1973.
Sims, C., J. Stock, and M. Watson, “Inference in Linear Time Series with Some Unit Roots,” Econometrica, No. 1 (January 1990), Vol. 58, pp. 113–144.
Singh, A., “The Stock Market and Economic Development: Should Developing Countries Encourage Stock Markets?,” unpublished, Faculty of Economics and Politics, University of Cambridge, October 1991.
Sundararajan, V., and S. Thakur, “Public Investment, Crowding Out, and Growth: A Dynamic Model Applied to India and Korea,” IMF Staff Papers, (December 1980), Vol.27, pp. 814–55.
Tobin, J., “A General Equilibrium Approach to Monetary Theory,” Journal of Money. Credit and Banking, (February 1969).
Turi Wat, U., and C-H., Wong, “Determination of Private Investment in Developing Countries,” Journal of Development Studies, (October 1982), Vol.19, pp.19–33.
WIDER, Foreign Portfolio Investment In Emerging Equity Markets, Study Group Series, No. 5, World Institute for Development Economics Research of the United Nations University, Helsinki, 1990.
This paper was written while the author was a summer intern in the Fund’s Middle Eastern Department. He wishes to thank Mohamed A. El-Erian, Antonio Furtado, Neven Mates, Syed Rizavi, Ghiath Shabsigh, Shamsuddin Tareq, Ulrich Thiessen, and Kal Wajid for their comments and insights. Any remaining errors are the responsibility of the author.
Fiscal years ending June 30.
This approach can be argued to be better suited than the neoclassical approach in accommodating some of the managerial theories of investment. In the model, the objective function of the managers is not merely to optimize profit, but also includes other objectives which may improve the strategic position of the firm. Accordingly, policies such as expansion and diversification, which are not always accommodated within the neoclassical framework, are included because they can lead to the increase in the market price of the shares.
The situation is further complicated by data limitations in most developing countries.
Although the above factors may also influence investment in industrial countries, they are less relevant there and mainly operate through the traditional determinants of investment. To take government investment as an example, its role in developed countries works mainly through aggregate demand and Is therefore included in the accelerator component of the investment function. In contrast, what is more relevant in the case of developing countries is the role of government investment in creating linkages and externalities. This is not represented in the accelerator component, and, therefore has been represented explicitly in the investment functions specified for these countries. It should be noted, however, that the investment incentives in developing countries do not differ from those in developed countries. Investors are mainly motivated by the expected rate of return which public Investment, in certain circumstances, may raise especially in the case of developing countries.
While there is some debate concerning the efficiency of capital markets in achieving these ends (Singh, 1991), their usefulness in mobilizing investment funds is well established.
In view of the emphasis given to government investment in the following sections, this discussion also reviews model specification and relevant empirical findings.
An interesting point to note is that when government investment is used as a counter-cyclical policy instrument, a negative correlation may appear between it and private investment. Such a correlation does not imply any causality, and this should not be interpreted as crowding out. However, this case is more relevant to the case of developed, rather than developing countries.
In Blejer and Khan (1984), expected change in GDP was estimated, rather than taking actual change as a proxy. The estimation can be made using a distributed-lags model or an adaptive expectation model. Reaching the correct specification Is important in order to avoid excessive approximations. To avoid this problem, the current study takes the realized change in GDP as a proxy for the expected change. There is no a priori reason to believe that any resulting bias would be significantly greater than that which would occur due to the approximation resulting from estimations using the above models. This issue, however, requires some further empirical investigation in order to ascertain which is the most efficient method for representing expected change in GDP.
An attempt was also made to estimate equation (1) using the real interest rate instead of ACRD. The resulting coefficient, however, had the wrong sign and was statistically insignificant.
Two definitions of noninfrastructure investment were used in this paper: a strict definition which included government Investment in manufacturing and wholesale and retail trade only, and a less strict definition where general and provincial government investments are included in noninfrastructure government investment.
This was preferred to the Blejer and Khan (1984) approach which used estimated expected (that which is following the trend) and unexpected (the residual) government investments as proxies for the infra and noninfrastructure components respectively. Using the latter methodology can lead to greater approximation. The reason is that many infrastructure projects are indivisible. When such projects are carried out, they violate the trend and would therefore be counted as noninfrastructure investment according to the latter methodology. On the other hand, a significant part of noninfrastructure investment can follow the trend. This is the case especially when the Government comes under political pressure to keep expanding different items of noninfrastructure or even unproductive investment. If this is the case, infrastructure investment, in contrast with the suggested methodology, may be the residual. These concerns about Blejer and Khan’s classification methodology are shared by Chhibber, Dailami, and Shafik (1992) who also point out that infrastructure investment is often financed by external aid– an important consideration in the case of Pakistan. This leads to less certainty and steadiness concerning such investment, and therefore, implies that it may not follow a steady trend as suggested by Blejer and Khan’s methodology.
Alternative lag structures were also attempted. However, the estimated coefficients of such lags were not significant. For non- infrastructure government investment (GNON), different lags were also experimented with and the results were insignificant suggesting that such investment manifests its main impact on private investment within the same year.
Regarding domestic sources, the availability of finance in Pakistan has been constrained by the low rates of domestic savings. Domestic bank credit has provided the major source of finance for private investment. Moreover, the Government resorted to direct credit as an instrument to promote private investment in certain sectors. As a result, the private credit share in total domestic credit increased from under 45 percent in 1981/82 to almost 60 percent in 1989/90 (Chart 5). It may also be noted that private investment in Pakistan has been less dependent on external borrowing which instead has been mainly geared toward financing government expenditure. An external inflow that may have had a direct impact on private investment is workers’ remittances, although it appears that a good part of this income was spent on improving the living standards of migrants’ families in Pakistan and on informal sector activities. As to portfolio and direct foreign investment, the private sector in Pakistan had only limited access to them in the period under analysis. More recently, the broadening stock market and participation of foreign investment are likely to alter this phenomenon.
According to the beta coefficients estimates, new credit to private investment had a relatively more important positive effect than total government investment (equation 1). This implies that the latter should not be financed at the expense of the former. However, it was also estimated that infrastructural government investment had the most positive impact (see the beta coefficients estimates of equations 2.1 and 2.2).
This analysis, however, requires qualification. As Chhibber, Dailami, and Shafik (1992) argue, the existence of public sector firms in a particular industry may deter private firms from entering even when the levels of both prices and demand are potentially favorable. Thus, private firms are usually cautious of governments’ tendency to favor public firms, thereby giving the latter a competitive advantage in the case of any potential competition in the future.
