Public Capital and Growth
  • 1 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund

Contributor Notes

This paper estimates the impact of public capital on economic growth for forty-eight OECD and non-OECD countries during 1960 - 2001. Using the production function and its extensions, it finds a positive - but concave - elasticity of output with respect to public capital, which is robust to changes in time intervals and varying depreciation rates. Furthermore, in non-OECD countries the growth impact of public capital is higher once longer time intervals are considered.

Abstract

This paper estimates the impact of public capital on economic growth for forty-eight OECD and non-OECD countries during 1960 - 2001. Using the production function and its extensions, it finds a positive - but concave - elasticity of output with respect to public capital, which is robust to changes in time intervals and varying depreciation rates. Furthermore, in non-OECD countries the growth impact of public capital is higher once longer time intervals are considered.

I. Introduction

An understanding of the impact of public investment on growth is relevant for at least three reasons. First, it has been argued that tight budgets constrain public investment more than current spending because it is easier to cut the former for political and other reasons (Roy, Heuty, and Letouze, 2006). Since the late 1990s, this has led to calls to correct the bias against public investment, most importantly in infrastructure, and create “fiscal space” for funding such investment (Heller, 2005).2 Underlying this premise is the belief that public investment is productive.

Second, in a somewhat similar vein, it has been argued that constraints on external borrowing have prevented governments with large “infrastructure gaps” from undertaking productive investments. Although a country’s borrowing capacity depends primarily on its macroeconomic policies, ability to collect taxes, and strength of its public financial and debt management systems, the contribution of debt-financed public investment to growth and exports also plays a role in external borrowing limits.

Finally, fiscal policy has a countercyclical role to play in supporting growth and recovery, which has been recognized during the recent financial crisis. In this context, fiscal stimulus packages in many countries have included a large share of public investment spending in the belief that such investment is productive and better for future growth.3

However, the empirical evidence on the impact of public investment on growth is mixed. Previous studies on the impact of public investment on growth have not produced clear-cut results (IMF 2004 and IMF 2005b). This has led some to conclude that public investment is not productive. Some have also argued that total factor productivity (TFP), rather than capital accumulation, matters in explaining growth differentials (Easterly and Levine (2001).

At the same time, a more recent study by the World Bank (2007) concluded that there are positive growth effects of public spending in general, and that of infrastructure, education, and health spending in particular. The report from the Commission on Growth and Development (2008) came to an even stronger conclusion by noting that a common element in fast-growing countries is high public investment, defined as 7 percent of GDP or more. Other studies have argued that fiscal multipliers for investment spending are higher than those for other types of public spending or tax cuts (Perotti, 2005; Zandi, 2008).

Why is the evidence on the relationship between public investment and growth mixed? There could be three reasons. First, many empirical studies use the public investment rate, as opposed to the rate of change in public capital in explaining differences in growth across countries. This is because public investment data are easier to obtain, even though public investment and public capital variables can differ substantially for a country. In particular, these variables can grow at different rates, depending on the initial level of capital stock.

Second, there is an endogenous link between public investment and growth that can complicate econometric identification. Public investment and growth are flow variables determined in equilibrium and observed over the same period. For example, public investment may fall in an economic downturn simply due to lack of resources, as it is often among the first expenditure items to be cut in a downturn. In contrast, public capital stock does not suffer from this drawback, given that it is measured at the beginning of the period. A year in which growth is low would not affect the (beginning of the period) capital stock.

Third, most studies do not take into account the government’s budget constraint and the costs of financing public investment, which could lead to diminishing returns to investment. The relationship between public investment and growth could even turn negative once public capital is above a certain threshold. For example, maintaining and/or expanding the existing capital stock may require high (and potentially distortionary) tax rates, which would reduce growth, all else being equal (Aschauer 1998; Barro 1990). Thus, the productivity of public investment can vary depending on the initial stock of public capital.

We therefore use measures of the capital stock, rather than investment rates in our study, notwithstanding difficulties in estimating the capital stock. We find that the importance of assumed initial capital stock diminishes significantly in long time series.4 We address the issue of choosing an appropriate depreciation rate by using different rates. Indeed, our sensitivity analysis shows that under varying depreciation rates, the results of the paper hold.

An important contribution of the paper lies in the construction of public capital stock series for 26 middle-income and low-income countries. Such estimates already exist for 22 OECD countries (Kamps 2006). Our estimates rely on the same methodology as that used for OECD countries.

Our results for a panel of 48 OECD and non-OECD countries show that there is a positive elasticity of output with respect to public capital. This supports the view that changes in public capital stock can explain growth differences across countries, even though the evidence on the impact of public investment is mixed. The literature review in the appendix table, drawn from Romp and de Haan (2005), further indicates that studies that used changes in public capital stock have typically obtained positive results, while the picture is less clear for those using public investment rates.

We also find that the elasticity of output with respect to public capital depends on the income level of countries (OECD versus non-OECD) and the initial level of public capital. The elasticity is somewhat stronger for OECD countries, possibly suggesting the importance of institutional factors. At the same time, we find that the positive impact of public capital on output varies with the level of public capital, and that it can be partially offset by high levels of public capital. Such inefficiencies may be engendered by taxes levied to maintain or expand the existing capital stock.

The paper is organized as follows. Section II presents the theoretical models underlying the empirical estimation. Section III explains the construction of the public capital stock dataset. Section IV discusses the empirical results and Section V presents sensitivity analysis with respect to depreciation rates. Section VI concludes and discusses policy implications.

II. Theory

We use the standard production function approach in the literature and its extension to estimate the impact of public capital on output (Aschauer 1989; Aschauer 1998).

A. Standard Model

In this model, public capital is explicitly included in the aggregate production function by redefining the production function Y= A* F(L,K) as

Y=A˜*F(L,K,G)

where Y is the level of output, A is the level of productivity, L is employment, K is private capital, G is public capital and à is the total factor productivity purged of the influence of public capital.5 A commonly used specification is the Cobb-Douglas production function:

Y=A˜ LaKbGc

Taking natural logarithms yields the equation:

InY=InA˜+alnL+blnK+clnG

Taking first differences yields the equation:

ΔIn Y=constant+a ΔlnL+b ΔlnK+c ΔlnG

The elasticity of output with respect to public capital, c, is the main variable of interest in this study. The other production elasticities, a and b, are of interest mainly to assess the shape of the production function. Assuming perfect competition in factor markets, private capital and labor must be paid their marginal products (i.e., a + b = 1). In this case, given that c is expected to be positive, the model would generate increasing returns to scale (a + b + c > 1). On the other hand, if one assumes constant returns to scale (a + b + c = 1), then labor and private capital must be paid more than their marginal products (i.e., a + b < 1). That would be the case when public capital generates indirect income for labor and private capital, even though it is not paid directly for its services. In this paper, the model is estimated without any restrictions on the coefficients a, b, and c, but they are subsequently tested.

B. Extended Model

The extended model is the same as the standard model, except that it allows for a diminishing elasticity of public capital as a function of the initial stock of public capital (in percent of GDP). The motivation for this control variable is the potential non-linearities in the productivity of public capital. For example, if public investment is financed by capital taxes, a higher public investment would reduce after-tax profits for investors and curtail their incentive to invest. At some point, the disincentive effect from higher taxes would exceed the productivity gains from higher public investment, and the net impact on growth would become negative (Barro 1990). Furthermore, the marginal productivity of public capital could decline due to inefficiencies in the capital spending process and due to difficulties in finding investment projects with high returns.

To capture this relationship, we extend the standard model by adding an interaction on the right hand side of the equation. In particular:

ΔInY=constant+aΔlnL+bΔlnK+cΔlnG+dlnG*(G/Y)

The interaction term implies that the production elasticity of public capital is no longer just c, but is equal to (c + d G/Y). This implies that the productivity of public capital can vary according to its initial level. The coefficient on the interaction term, d, is expected to be negative, while the coefficient c is expected to remain positive.

III. Data

A. Construction of the Dataset

The dataset includes public and private capital stock series for 48 countries from 1960 to 2001. Table 1 lists the countries in the sample. The data for 22 OECD countries is from Kamps (2006).6 We construct public and private capital stock data for the other 26 non-OECD countries7 using the same methodology for OECD countries. The estimation is based on internationally comparable total investment series from Penn World Tables (Heston, Summers, Aten 2006) and public and private investment series from the World Economic Outlook database (IMF, April 2009).8

Table 1.

List of Countries in the Sample

article image
Note: Middle income non-OECD countries have per capita income of more than $3,700 in 2000 prices. Lower income non-OECD countries have per capita income of less than $3,700 in 2000 prices.

The Penn World Table (PWT) is one of the most widely used sources for cross-country growth studies. It provides data on internationally comparable output and investment series for a broad group of countries. The data are based on national accounts and adjusted for purchasing power parity (PPP) to make them internationally comparable.9 PWT also assigns grades to countries from A to D based on the quality and consistency of their data over time. Countries with a grade of D are excluded from the sample.

One drawback of PWT is that it does not provide a breakdown of investment into its public and private components. For that, we turn to the World Economic Outlook (WEO) database (IMF 2009), which provides disaggregated data on public and private investment.10 We use the share of public and private investment in total investment to split the PWT investment series into public and private components.

Based on the disaggregated PWT investment series, the public and private capital stocks are estimated using the perpetual inventory method employed by Kamps (2006) for OECD countries. First, the initial capital stock is set to zero for all countries in 1860. Second, a hypothetical investment series is constructed between 1860 and 1960 based on a 4 percent growth rate for investment during 1860–1960 to calculate the capital stock in 1960. Lastly, the investment series for 1960–2000 are accumulated to construct the capital stock series for 1960–2001.

B. Depreciation Rates

The net capital stock accounts for the wear and tear of an asset (i.e., depreciation), thus it excludes assets that are no longer used in production. The choice of depreciation rates present perhaps the biggest challenge in the construction of the capital stock estimates because country specific estimates of depreciation rates are typically not available, with the U.S. being an important exception. According to the Bureau of Economic Analysis, estimated depreciation rates for public capital in the U.S. were about 2.5 percent in 1960 and increased to 4 percent in 2001 (and from 4.25 to 8.5 percent for private capital, respectively). Kamps (2006) uses these U.S. depreciation rates to construct the capital stock estimates for other OECD countries.

Different depreciation assumptions may be more appropriate for non-OECD countries, given differences in the types of assets they hold. The composition of underlying assets affects the average depreciation rate of the aggregated stock because different types of assets have different life spans. For instance, concrete structures are typically estimated to have longer lifetime (e.g., 80–100 years), while assets related to IT tend to have much shorter life span (e.g., a few years). As countries become richer, the share of assets with shorter life span rises, thereby raising the overall depreciation rate.

The rising depreciation rates in OECD countries reflect the growing importance of high technology assets in their public capital. In contrast, non-OECD countries with relatively lower share of technological assets and more “traditional” physical infrastructure in their public capital are more likely to have lower depreciation rates, as the average life span of those assets is higher. Arestoff and Hurlin (2006) find that this is in fact the case for a number of developing countries.11

Table 2 shows the depreciation rates for public and private capital assumed for constructing the capital stock for non-OECD countries. For middle-income countries, we use a time-varying profile for public and private capital with a flatter slope than the one used for OECD countries. For low-income countries, we hold the depreciation rate constant over time.

Table 2.

Depreciation Rates

(In percent)

article image

C. Caveats

Our capital stock dataset is novel in several ways: It includes both OECD and non-OECD countries, differentiates between public and private capital, and applies time- and income-level varying depreciation rates to capture the nature of the underlying public and private assets. Previous studies, such as by Nehru and Dhareshwar (1993) and Marquetti and Foley (2008), have estimated the capital stock of non-OECD countries but have not made a distinction between public and private capital. On the other hand, Collier, Hoeffler and Pattillo (2001) have estimated the public capital stock for 22 sub-Saharan African countries, but assuming constant depreciation rates. Finally, Arestoff and Hurlin (2006) have estimated the public capital stock for a group of non-OECD countries using infrastructure specific depreciation rates, but not the private capital stock. Our dataset includes both public and private capital stock data, which are necessary to estimate production functions, and covers countries with different income groups from all five continents.

However, some caveats are in order:

First, investments in public capital may not always be productive (Pritchett 1996, Canning 1999, Easterly and Serven 2004). Reasons range from administrative inefficiencies to pork-barrel politics to corruption. This unobservable factor could cause public capital in some countries to be overestimated. While this is clearly an important issue, the estimated elasticity of output with respect to public capital should reflect this spending inefficiency. All else being equal, a country with a lower spending efficiency and overstated capital stock would have a lower elasticity of output with respect to public capital.

Second, non-OECD countries tend to spend less on operations and maintenance, which could cause depreciation rates to be higher in non-OECD countries compared to OECD countries.

On the other hand, as mentioned above, non-OECD countries tend to have a lower share of high-technology assets, which would make depreciation rates lower in non-OECD countries. While the net effect is unknown, we address this issue by conducting a sensitivity analysis using different depreciation rates (Section V).

Third, there may be differences in the way countries classify their investments as public or private, given the presence of public-private partnerships and the treatment of quasi-public enterprises in the national accounts. We do not address this issue in the paper, but note that this problem is common to all studies on the subject, regardless of whether they use public investment or public capital as the explanatory variable.

Finally, by applying the same depreciation rate to a group of countries, the approach disregards natural disasters and other Force Majeure events that may impact the capital stock of any particular country.

IV. Results

A. Descriptive Statistics

Table 3 provides the average real GDP growth, public investment rate, and the public capital stock growth for each income group during 1960–2000.12 It shows that the average GDP growth for the OECD countries was 3.4 percent during this period. For the average non-OECD countries, the growth rates were higher at 4.4 percent, suggesting that some catching-up took place during this period.

Table 3.

GDP Growth, Public Investment and Public Capital Stock Growth, 1960–2000

article image
Source: Authors’ calculations. Standard deviations in paranthesis.

The difference in public investment rates between OECD and non-OECD countries was relatively small during 1960–2000 (3.6 versus 3.9 percent of GDP). This contrasts with the difference in public capital growth, which was much higher—almost twice as much—in non-OECD countries (5.7 versus 3.3 percent).

Figure 1 plots a scatter of real GDP growth, public investment, and public capital growth in sample countries during 1960–2000. It shows that public capital growth is better in explaining the variation in GDP growth than public investment rates. Public investment rates lie in a narrow range, between 2 and 6 percent of GDP for most countries, whereas variation in public capital growth is larger. The public investment rate explains only 12 percent of the cross-country variation in GDP growth, whereas public capital growth explains 51 percent of the variation.

Figure 1.
Figure 1.

Long-Term Real GDP Growth, Public Investment and Public Capital Growth Rates, 1960–2000

Citation: IMF Working Papers 2010, 175; 10.5089/9781455201860.001.A001

Sources: PWT (Version 6.2), Kamps (2006), Authors’ calculations.

Figure 2 plots the public investment rate and public capital stock (both as a percent of GDP) for the average OECD and non-OECD countries. Note that public investment and public capital follow significantly different paths for each group. For example, public investment in the average OECD country started declining in the 1970s, whereas public capital was still increasing until the mid-1980s. For the average non-OECD country, both public investment and public capital stock “peaked” in the mid-1980 and since then has been on a declining trend.

Figure 2.
Figure 2.

Public Investment, Public Capital Stock and Real GDP Growth for OECD and non-OECD Countries, 1960–2000

(percent of GDP and annual percentage change)

Citation: IMF Working Papers 2010, 175; 10.5089/9781455201860.001.A001

Source: PWT (Version 6.2), Kamps (2006), Authors’ calculations.The red dotted lines indicate the average real GDP growth before and after the year in which the public capital stock peaked for each income group. Public capital stock (as a percent of GDP) peaked in 1983 for the average OECD country and in 1985 for the average non-OECD country. The peak levels were 60 percent of GDP for the average OECD country and 61percent of GDP for non-OECD countries.

Figure 2 also shows that the peak level of public capital was found in non-OECD countries. This is explained by the higher public investment rates in these countries (4 percent of GDP), compared to OECD countries (around 3 1/2 percent of GDP), during this period. Finally, Figure 2 shows the relationship between public capital and GDP growth. During the “pre-peak” years, the average real GDP growth was consistently high for OECD and non-OECD countries. In fact, Table 4 shows that real GDP growth was about one percentage point higher for both OECD and non-OECD countries in the pre-peak period when public capital stock was still increasing as a percent of GDP.

Table 4.

Average Real GDP Growth Before and After Public Capital Stock

article image
Source: Authors’ calculations. Standard deviations in paranthesis.Note: Public capital stock (as a percent of GDP) peaked in 1983 for the average OECD country and in 1985 for the average non-OECD country. The peak level was 60 percent of GDP for the average OECD country, 62 percent of GDP for the average non-OECD country.

B. Regression Results

We first examine the question whether regressions should be run in levels or differences. Table 5 shows that the variables of interest for non-OECD countries are stationary when expressed in first differences, extending a previous finding for OECD countries (Kamps 2006). We do not find any relationship between the variables in levels that would point to a common cointegrating relationship in the panel, neither for the combined dataset nor for non-OECD and OECD countries separately.13 We thus proceed to estimate the equation in differences to obtain consistent estimates of the coefficients.

Table 5.

Panel Unit Toot Tests for Non-OECD Countries

article image
Source: Authors’ calculations. IPS and LLC columns report the Im, Pesaran and Shin (2003) and the Levin, Lin and Chu (2002) corrected t-value of the lagged variable, which is normally distributed under the null hypothesis of nonstationarity. One lagged difference included. Panel unit root tests for OECD countries are conducted by Kamps (2006).

Table 6, 7, and 8 provide the main results of the paper. The OLS panel regressions include fixed effects. These are better at capturing cross-country differences in technological growth, human capital accumulation, and any other factor affecting total factor productivity, which are reflected in the intercept term (lnA˜). This is confirmed by Hausman tests on alternative random effect specifications. Table 6 shows results for the combined OECD and non-OECD dataset; Tables 7 and 8 provide separate results. The results are presented for three models: naive, standard and extended. We also present three different time intervals in the analysis, namely one year, five years, and ten years.14

Table 6.

Regression Results for All Countries

article image
article image
Note: Standard errors in parenthesis, statistical significance at the 10, 5 and 1 percent levels denoted by *, ** and *** respectively. Hausman test indicates p-value of the null hypothesis that the difference in estimated coefficients between the fixed effect and a random effects specification (not reported) is not systematic. L: Labor; K: Private Capital Stock; IG/GDP: Public Investment to GDP ti G P bli C it l St k Standard model (interacted)

In extended model, coefficient of the second quartile

Table 7.

Regression Results for OECD Countries

article image
Note: Standard errors in parenthesis, statistical significance at the 10, 5 and 1 percent levels denoted by *, ** and *** respectively. Hausman test indicates p-value of the null hypothesis that the difference in estimated coefficients between the fixed effect and a random effects specification (not reported) is not systematic. L: Labor; K: Private Capital Stock; IG/GDP Public Investment to GDP ratio; G: Public Capital Stock.
Table 8.

Regression Results for Non-OECD Countries

article image
Note: Standard errors in parenthesis, statistical significance at the 10, 5 and 1 percent levels denoted by *, ** and ***respectively. Hausman test indicates p-value of the null hypothesis that the difference in estimated coefficients between the fixed effect and a random effects specification (not reported) is not systematic. L: Labor; K: Private Capital Stock; IG/GDP: Public Investment to GDP ratio; G: Public Capital Stock.

One reason for analyzing different time intervals is the existence of both long and short-term effects of public capital accumulation on growth that may not be captured well in annual data. For example, some public investments may take more than a year to complete and, in addition, the payoff may accrue over a longer time horizon. The longer time horizons, especially the five-year interval, may be better suited to capture the indivisibility of investment and lags in its effectiveness.

Naive model

The naive model uses the public investment rate (IG/GDP) instead of public capital growth (Δln G) in the standard model.15 As expected, we do not find any statistically significant relationship between the public investment rate and GDP growth for the combined dataset and for the non-OECD countries. For the OECD, the coefficient is significant but negative only in the five-year interval specification.

Non-zero depreciation rates imply that the same gross investment can lead to very different rates of net capital accumulation. For example, in our sample of non-OECD countries, countries with a low public capital stock (30 percent of GDP, the 25th percentile) on average lose public capital worth around one percent of GDP annually due to depreciation, whereas those with a high stock of public capital (70 percent of GDP, the 75th percentile) lose more than double that amount, close to two percent of GDP. Consequently, a gross investment rate of 7 percent of GDP (Growth Commission report, 2008) would translate into net investment rates between 5 and 6 percent of GDP, different paths of capital accumulation, and different growth effects. Another way of making the same point is as follows: a 7 percent of GDP gross investment rate implies a public capital accumulation rate of 20 percent if the initial stock is 30 percent of GDP, while it implies an accumulation rate of only 7 percent, if the initial stock is 70 percent of GDP. This shows that there could be large differences between the public investment rate and the rate of change in the public capital stock depending on the initial capital stock.

Standard model

In the combined OECD and non-OECD panel dataset (Table 6) the standard model produces plausible and statistically significant results in the one-year specification, with a public capital stock coefficient of 0.05. Coefficients for L and K are significant and positive in all specifications. When varying the coefficient for G with the distribution of public capital stock (by multiplying with quartile dummy variables), a concave pattern emerges with respect to the elasticity of public capital, indicating a maximum impact of additional public capital (growth elasticity of 0.2) when the public capital stock is between 15 and 60 percent of GDP (Figure 3). For very high levels of public capital, the coefficient loses statistical significance. We now proceed to estimate separate models for OECD and non-OECD countries.

Figure 3.
Figure 3.

Standard Model with Varying Coefficients of Public Capital

Citation: IMF Working Papers 2010, 175; 10.5089/9781455201860.001.A001

Source: Authors’ calculations.

For OECD countries alone, we also find a positive and statistically significant relationship between public capital and output for OECD in the one-year interval specification. The coefficients for L, K, and G have values that are consistent with those found in the literature. Public capital is significant with a coefficient of 0.13, and the coefficients attached to labor and private capital are also significant and in the plausible range. When moving to five and ten-year time intervals, the coefficient on private capital is the only one to remain significant, and the estimate increases to 0.5. At the same time, the fit of the model improves when moving to higher time intervals, a feature that cannot be related to the literature since earlier studies do not report measures of fit.

For non-OECD countries, the standard model has a poor fit in the one-year interval specification, but the fit improves significantly when moving to the five and ten-year interval. While public capital remains insignificant in most specifications, private capital and labor are highly significant and the estimated coefficients increase with time intervals. In the five-year interval the coefficient on labor is estimated to be 0.5, whereas that on private capital is 0.2, in line with the one-year interval estimate obtained for OECD countries.

Extended model

The extended model produces estimates that are statistically more significant than those produced by the standard model, while the fit of the model also improves. For non-OECD countries, the coefficient on public capital, c, becomes statistically significant, while the coefficient on the interaction term, d, is negative, as expected.

For OECD countries, the coefficient c is estimated to be 0.13 (using one-year interval), however, when moving to higher time intervals—just as in the standard model—the coefficients for public capital lose significance. For example, in the five-year interval the estimated coefficient for public capital in OECD countries is smaller (0.08) and not significant. The interaction variable d—that captures non-linearities—is not significant in the case of OECD countries.

For non-OECD countries, the coefficient c on public capital averages 0.2 and is significant in all specifications. Its value and significance increases with the length of time interval, from 0.12 in the one-year specification to 0.26 in the ten-year specification. The coefficient d attached to the interaction term is negative and significant in all specifications, and averages -0.5. This suggests that in some instances, a significant scaling up of investment may not yield positive returns, depending on the size of the existing public capital stock. Evaluated at the 25 percentile of public capital stock for non-OECD countries (public capital is around 30 percent of GDP), the marginal effect of additional public capital is positive (0.1), the effect eventually reduces to zero around the 50th percentile (public capital is 50 percent of GDP), before turning slightly negative, reaching -0.06 at the 75th percentile (public capital is 70 percent of GDP).

Finally, the fit of the model improves as one moves from one-year to five and ten-year intervals. This indicates that in non-OECD countries, capital spending impacts growth over a period of time in reflection of indivisibility of investments.

In summary, once we enter the interaction term in the standard model, we find the following: (i) the fit of the regressions improve; (ii) the impact of public capital on growth becomes positive also for non-OECD countries; (iii) a negative relationship between growth and the level of public capital is uncovered, in particular for non-OECD countries. These results are robust to a variety of alternative specifications and to inclusion and exclusion of outliers.16

The results suggest that some non-OECD countries may not absorb a large scaling up of capital investment, reflecting implementation weaknesses, and that weak absorptive capacity should be taken into account in setting borrowing limits for non-concessional loans. Furthermore, from a policy perspective, the results also suggest that public investment can be used to boost aggregate demand in OECD countries, while it can boost aggregate supply in non-OECD countries.

C. Post Estimation Tests

First, even after controlling for non-linearities, non-OECD countries have a slightly lower coefficient for public capital than OECD countries (0.123 versus 0.132) in the one-year specification. All else being equal, this could be attributed to lower spending efficiency in non-OECD countries. However, the difference is not statistically significant.

Second, private capital has a higher coefficient than public capital in all models for both OECD and non-OECD countries. This is consistent with other studies that find private investment to be more productive than public investment (Khan and Kumar 1997). However, the difference is not significant for OECD countries and is significant only in non-OECD countries when using the standard model.

Third, the estimation results with one year intervals for OECD countries (standard model) yield elasticities of output with respect to public capital of 0.13, private capital of 0.25, and labor input of 0.39. Therefore, one cannot reject the null hypothesis that the production function has constant returns to scale (a+b+c=1). The estimation results for non-OECD countries (extended model, one year interval) are also reasonable with elasticities of output with respect to public capital of 0.12, private capital of 0.14, and labor input of 0.20 (all coefficients are significant at the 5 percent level, except for labor). However, one can reject the null hypothesis that the production function in non-OECD countries displays constant returns to scale in the one year interval. Results for non-OECD countries differ slightly if five-year intervals are considered. In this case, in line with results for OECD countries, the null hypothesis of constant returns to scale cannot be rejected, and the estimated coefficients move somewhat closer to the findings for OECD countries.

Finally, we ran Hausman tests, finding that the random effect specification is rejected in most cases.

V. Alternative Depreciation Rates

We vary the depreciation rates used in the construction of the public capital stock. We run three scenarios. In each scenario, depreciation rates are the same for both OECD and non-OECD countries. In the first scenario, they are time-varying and equal to the ones used by Kamps (2006), which increase from 2.5 to 4.0 percent between 1960–2001. In the second scenario, they are constant and equal to the average rate of 3.25 percent over the same period. In the third scenario, they are constant and equal to the maximum rate of 4.0 percent.

The results are provided in Tables 9 and 10. Using alternative depreciation rates does not change the results significantly. For OECD countries, the estimated coefficient c varies between 0.129 and 0.110 in the standard model, and 0.22 and 0.25 in the extended model. For non-OECD countries, c varies between 0.21 and 0.036 in the standard model and 0.09 to 0.2 in the extended model. More importantly, in all cases, c and d remain statistically significant, despite the use of different depreciation rates.

Table 9.

Sensitivity Analysis: Alternative Depreciation Rates for OECD Countries

article image
Note: Standard errors in parenthesis, statistical significance at the 10, 5 and 1 percent levels denoted by *, ** and *** respectively. Hausman test indicates p-value of the null hypothesis that the difference in estimated coefficients between the fixed effect and a random effects specification (not reported) is not systematic. L: Labor; K: Private Capital Stock; IG/GDP: Public Investment to GDP ratio; G: Public Capital Stock. Scenaro 1: Time-varying Kamps depreciation rates for public capital from 2.5 to 4.0 percent; Scenaro 2: Constant depreciation rate at 3.25 percent; Scenaro 3: Constant depreciation rate at 4.0 percent.
Table 10.

Sensitivity Analysis: Alternative Depreciation Rates for Non-OECD Countries

article image
Note: Standard errors in parenthesis, statistical significance at the 10, 5 and 1 percent levels denoted by *, ** and *** respectively. Hausman test indicates p-value of the null hypothesis that the difference in estimated coefficients between the fixed effect and a random effects specification (not reported) is not systematic. L: Labor; K: Private Capital Stock; IG/GDP: Public Investment to GDP ratio; G: Public Capital Stock. Scenaro 1: Time-varying Kamps depreciation rates for public capital from 2.5 to 4.0 percent; Scenaro 2: Constant depreciation rate at 3.25 percent; Scenaro 3: Constant depreciation rate at 4.0 percent.

VI. Conclusions

This paper revisits the debate on the effect of public investment on growth by estimating a production function for forty-eight OECD and non-OECD countries, using capital stock as the explanatory variable. The results indicate that increases in public capital stock are positively correlated with growth, after controlling for the initial level of public capital. The effect is stronger for OECD countries in the short-term, while it is stronger for non-OECD countries in the long-term. This is in contrast with the mixed results obtained by studies that mainly use the investment rate as the explanatory variable.

In some countries, the positive impact of public capital on output partially or wholly offset if the initial capital stock is high in relation to GDP. However, these considerations do not seem to matter in countries with a relatively low public capital stock. An important by-product of this study is the construction of a public capital stock series for 26 middle- and low-income countries. Such estimates are already available for 22 OECD countries.

A number of policy implications could be drawn from these results. First, while debate on fiscal space has centered on creating room in the budget for higher public investment, the results show that certain types of constraints (financing or the ability to absorb) can limit growth benefits of higher capital stock. Second, developing countries can avail of non-concessional foreign borrowing to finance new investments, provided these resources are invested in projects that have been subjected to a rigorous cost-benefit analysis and hold strong prospects for enhancing future growth. However, unlike OECD countries, the benefits tend to accrue over time. This would necessitate extending the timeframe of debt sustainability frameworks so that they can take into account the positive effects of public investments. Finally, public investment projects included in fiscal stimulus packages for 2009–10 should increase growth, even with lags in their implementation, provided public capital levels are not too high to begin with and the resulting financing costs and high tax rates do not negate the positive benefits of new public investments.

References

  • Albala-Bertrand, J. M., (2004), “Can the Composition of Capital Constrain Potential Output? A Gap Approach,” Queen Mary University of London, Department of Economics, Working Paper No. 510.

    • Search Google Scholar
    • Export Citation
  • Albala-Bertrand, J. M., and Mamatzakis, E. C., (2004), “The Impact of Public Infrastructure on the Productivity of the Chilean Economy,” Review of Development Economics, (8:2), pp. 266278.

    • Search Google Scholar
    • Export Citation
  • Arestoff, Florence, and Christophe Hurlin, (2006), “Estimates of Government Net Capital Stocks for 26 Developing Countries, 1970–2002,” Policy Research Working Paper Series 3858, The World Bank.

    • Search Google Scholar
    • Export Citation
  • Aschauer, D. A., (1989), “Is Public Expenditure Productive?Journal of Monetary Economics, 23, 177200.

  • Aschauer, D. A., (1998), “How Big Should the Public Capital Stock Be?The Jerome Levy Economics Institute of Bard College Public Policy, No: 43.

    • Search Google Scholar
    • Export Citation
  • Barro, Robert, (1990), “Government Spending in a Simple Model of Exogenous Growth,” Journal of Political Economy, 98, S103S125.

  • Batina, R. G., (1998), “On the Long-run Effects of Oublic Capital and Disaggregated Public Capital on Aggregate Output,” International Tax and Public Finance, (5:3), pp. 263281.

    • Search Google Scholar
    • Export Citation
  • Bonaglia, F., La Ferrara, E., and Marcellino, M., (2001), “Public capital and economic performance: Evidence from Italy,” IGIER Working Paper No. 163.

    • Search Google Scholar
    • Export Citation
  • Boscá, J. E., Escriba, J., and Murgui, M. J., (2000), “The Effect of Public Infrastructure on the Private Productive Sector of Spanish Regions,” Journal of Regional Science, (42), pp. 301326.

    • Search Google Scholar
    • Export Citation
  • Bose, N., Haque, M. E., and Osborne, D., (2003), “Public Expenditure and Economic Growth. A disaggregated Analysis for developing Countries,” Centre for Growth and Business Cycle Research, Economic Studies, University of Manchester; Unpublished.

    • Search Google Scholar
    • Export Citation
  • Cadot, O., Röller, L. H., and Stephan, A., (1999), “A Political Economy Model of Infrastructure Allocation: An Empirical Assessment,” CEPR Discussion Paper No. 2336.

    • Search Google Scholar
    • Export Citation
  • Cadot, O., Röller, L. H., and Stephan, A., (2002), “Contribution to Productivity or Pork Barrel? The Two Faces of Infrastructure Investment,” WZB Discussion Paper No. 02–09.

    • Search Google Scholar
    • Export Citation
  • Calderón, C., Easterly, W., and L. Servén, (2004), “Latin America’s Infrastructure in the Era of Macroeconomic Crises,” in The Limits of Stabilization: Infrastructure, Public Deficits and Growth in Latin America, Easterly W. and Servén L. (eds), Stanford University Press, California.

    • Search Google Scholar
    • Export Citation
  • Calderón, C., and Servén, L., (2002), “The output cost of Latin America’s Infrastructure Gap,” Central Bank of Chile Working Paper No. 186.

    • Search Google Scholar
    • Export Citation
  • Canaleta, C. G., Arzoz, P. P., and Gárate, M. R., (1998), “Public Capital, Regional Productivity and Spatial Spillovers,” Universidad Pública de Navarra, Lan Gaiak Departamento de Economía, Working Paper No. 9811.

    • Search Google Scholar
    • Export Citation
  • Canning, David, 1999, “Infrastructure’s Contribution to Aggregate Output,” The World Bank Policy Research Working Papers Series #2246, (November 30, 1999).

    • Search Google Scholar
    • Export Citation
  • Canning, D., and Bennathan, E., (2000), “The Social Rate of Return on Infrastructure Investments,” World Bank Working Paper No. 2390.

    • Search Google Scholar
    • Export Citation
  • Canning, D., and Pedroni, P., (1999), “Infrastructure and Long-run Economic Growth,” Mimeo.

  • Charlot, S., and Schmitt, B., (1999), “Public Infrastructure and Economic Growth in France’s Regions,” Paper No. 129 for ERSA 39th Congress, Dublin, Ireland.

    • Search Google Scholar
    • Export Citation
  • Cohen, J.P. and Morrison Paul, C.J. (2004), “Public Infrastructure Investment, Interstate spatial Spillovers, and Manufacturing Costs,” The Review of Economics and Statistics, (86:2), pp. 551560.

    • Search Google Scholar
    • Export Citation
  • Collier, Hoeffler, and Catherine Pattillo, (2001), “Flight Capital as a Portfolio Choice,” The World Bank Economic Review, Vol. 15, No. 1 (2001), pp. 5580

    • Search Google Scholar
    • Export Citation
  • Commission on Growth and Development, 2008, The Growth Report: Strategies for Sustained Growth and Inclusive Development.

  • Crowder, W.J., and Himarios, D., (1997), “Balanced Growth and Public capital: An Empirical Analysis,” Applied Economics, (29:8), pp. 10451053.

    • Search Google Scholar
    • Export Citation
  • Demetriades, P. O., and Mamuneas, T. P., (2000), “Intertemporal Output and Employment Effects of Public Infrastructure Capital: Evidence from 12 OECD economies,” Economic Journal, (110), pp. 687712.

    • Search Google Scholar
    • Export Citation
  • Dessus, S., and Herrera, R., (2000), “Public Capital and growth Revisited: a Panel Data Assessment,” Economic Development and Cultural Change, 48, 407418.

    • Search Google Scholar
    • Export Citation
  • Devarajan S., Swaroop, V., and Zou, H. F., (1996), “The Composition of Public Expenditure and Economic Growth,” Journal of Monetary Economics, (37), pp. 313344.

    • Search Google Scholar
    • Export Citation
  • Duggal, V. G., Saltzman, C., and Klein, L. R. (1999), “Infrastructure and Productivity: a Nonlinear Approach,” Journal of Econometrics, (92), pp. 4774.

    • Search Google Scholar
    • Export Citation
  • Easterly, William, and R., Levine, (2001), “It’s Not Factor Accumulation: Stylized Facts and Growth Models,” World Bank Economic Review 15 (2001), pp. 177219.

    • Search Google Scholar
    • Export Citation
  • Easterly, William, and Servén Luis, 2004, “The Limits of Stabilization: Infrastructure, Public Deficits and Growth in Latin America,” Stanford University Press, California.

    • Search Google Scholar
    • Export Citation
  • Esfahani, H., and Ramíres, M. T., (2003), “Institutions, Infrastructure and Economic Growth,” Journal of Development Economics, (70), pp. 443477.

    • Search Google Scholar
    • Export Citation
  • Everaert, Greetje., (2003), “Balanced Growth and Public Capital: an Empirical Analysis with I(2) Trends in Capital Stock Data,” Economic Modelling, (20), pp. 741763.

    • Search Google Scholar
    • Export Citation
  • Everaert, Greetje., and Heylen, F., (2004),, “Public Capital and Long-Term Labour Market Performance in Belgium,” Journal of Policy Modelling, (26), pp. 95112.

    • Search Google Scholar
    • Export Citation
  • Fernald, J. (1999), “Assessing the Link Between Public Capital and Productivity,” American Economic Review, (89:3), pp. 619638.

  • Ferrara, E. L. and Marcellino, M., (2000), “TFP, Costs, and Public Infrastructure: An Equivocal Relationship,” IGIER Working Paper No. 176.

    • Search Google Scholar
    • Export Citation
  • Flores de Frutos, R., Garcia-Diez, M., and Perez-Amaral, T., (1998), “Public Capital and Economic Growth: An Analysis of the Spanish Economy,” Applied Economics, (30:8), pp. 985994.

    • Search Google Scholar
    • Export Citation
  • Ghali, K.H. (1998), “Public Investment and Private Capital Formation in a Vector Error-Correction Model of Growth,” Applied Economics, (30), pp. 837844.

    • Search Google Scholar
    • Export Citation
  • Gupta, Sanjeev, Benedict Clements, Emanuele Baldacci, and Carlos Mulas-Granados, (2005), Fiscal Policy, Expenditure Composition, and Growth in Low-Income Countries, Journal of International Money and Finance, 24, 441463.

    • Search Google Scholar
    • Export Citation
  • Gwartney, J., Holcombe, R. G., and Lawson, R., (2004), “Institutions and the Impact of Investment on Growth,” Paper Presented at the Conference of the Association Private Enterprise Education, APEE, (April), Bahamas.

    • Search Google Scholar
    • Export Citation
  • Haque, M. E., (2004), The Composition of Government Expenditure and Economic Growth in developing Countries, Global Journal of Finance and Economics, 1, 97117.

    • Search Google Scholar
    • Export Citation
  • Heller, Peter S., 2005, “Understanding Fiscal Space,” IMF Policy Discussion Papers (May 4, 2005).

  • Heston, Alan, Robert Summers, and Bettina Aten, (2006), Penn World Table Version 6.2, Center for International Comparisons of Production, Income and Prices at the University of Pennsylvania, September 2006.

    • Search Google Scholar
    • Export Citation
  • Holtz-Eakin, D., and Schwartz, A. E., (1995), “Infrastructure in a Structural Model of Economic GrowthRegional Science and Urban Economics, (25), pp. 131151.

    • Search Google Scholar
    • Export Citation
  • Horton Mark, Mohan Kumar, and Paolo Mauro, (2009): The State of Public Finances: A Cross-Country Fiscal Monitor,” IMF Staff Position Note 09/21 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Im, K. S., M. H. Pesaran, and Y. Shin, (2003) “Testing for Unit Roots in Heterogeneous Panels,” Journal of Econometrics (115), pp. 5374.

    • Search Google Scholar
    • Export Citation
  • IMF, 2004, Public Investment and Fiscal Policy, available at http://www.imf.org/external/np/fad/2004/pifp/eng/pifp.pdf.

  • IMF, 2005a, Understanding Fiscal Space, IMF Policy Discussion Paper, (March 2005)

  • IMF, 2005b, “Public Investment and Fiscal Policy—Lessons from the Pilot Country Studies,” International Monetary Fund, (April 2005).

    • Search Google Scholar
    • Export Citation
  • IMF, 2009, World Economic Outlook, (April 2009).

  • Kamps, C., (2006), “New Estimates of Government Net Capital stocks for 22 OECD Countries 1960–2001,” IMF Staff Papers (53:1), pp. 120150.

    • Search Google Scholar
    • Export Citation
  • Kamps, C., (2004b), “The Dynamic Effects of Public Capital: VAR Evidence for 22 OECD Countries,” Kiel Institute of World Economics Working Paper No. 1224.

    • Search Google Scholar
    • Export Citation
  • Kemmerling, A., and Stephan A. (2002), “The Contribution of Local Public Infrastructure to Private Productivity and Its Political Economy: Evidence from a Panel of Large German Cities,” Public Choice, (113), pp. 403422.

    • Search Google Scholar
    • Export Citation
  • Levin, A., C.-F. Lin, and C.-S. J. Chu. (2002), “Unit Root Tests in Panel Data: Asymptotic and Finite-Sample PropertiesJournal of Econometrics (108), pp. 124.

    • Search Google Scholar
    • Export Citation
  • Ligthart, J. E. (2002), “Public Capital and Output Growth in Portugal: An empirical Analysis,” European Review of Economics and Finance, (1:2), pp. 330.

    • Search Google Scholar
    • Export Citation
  • Mamatzakis, E.C. (1999a), “Public Infrastructure, Private Input Demand, and Economic Performance of the Greek Industry,” Queen Mary & Westfield College Working Paper No. 406.

    • Search Google Scholar
    • Export Citation
  • Mamatzakis, E.C., (1999b), “Testing for Long run Relationship Between Infrastructure and Private Capital Productivity: A Time Series Analysis for the Greek Industry,” Applied Economics Letters, (6:4), pp. 243246.

    • Search Google Scholar
    • Export Citation
  • Marquetti, Adalmir and Foley, Duncan, 2008, Manuscript.

  • Milbourne, R., Otto, G., and Voss, G. (2003), “Public Investment and Economic Growth,” Applied Economics, (35:5), pp. 527540.

  • Mittnik S., and T. Neumann, (2001), “Dynamic Effects of Public Investment: Vector Autoregressive Evidence From Six Industrialized Countries,” Empirical Economics, Springer, vol. 26(2), pages 429446

    • Search Google Scholar
    • Export Citation
  • Moreno, R., López-Bazo, E., and Artís, M., (2003), “On the Effectiveness of Private and Public Capital,” Applied Economics, (35), pp. 727740.

    • Search Google Scholar
    • Export Citation
  • Nehru, Vikram and Ashok Dhareshwar, 1993, “A New database on Physical Capital Stock: Sources, Methodology and Results,” Revista de Análisis Económico, 8(1), pp. 3759.

    • Search Google Scholar
    • Export Citation
  • Perotti, Roberto, 2005, “Estimating the Effects of Fiscal Policy in OECD Countries,” CEPR Discussion Paper No. 4842 (London: Centre for Economic Policy Research).

    • Search Google Scholar
    • Export Citation
  • Pereira, A. M., (2000), “Is All Public Capital Created Equal?Review of Economics and Statistics, (82:3), pp. 513518.

  • Pereira, A. M., (2001), “On The Effects of Public Investment on Private Investment: What Crowds in What?Public Finance Review, (29:1), pp. 325.

    • Search Google Scholar
    • Export Citation
  • Pereira, A. M., and Andraz, J. M., (2001), “On the Impact of Public Investment on the Performance of U.S. Industries,” Public Finance Review, (31:1), pp. 6690.

    • Search Google Scholar
    • Export Citation
  • Pereira, A. M., and Flores de Frutos, R., (1999), “Public Capital Accumulation and Private Sector Performance,” Journal of Urban Economics, (46:2), pp. 300322.

    • Search Google Scholar
    • Export Citation
  • Pereira, A. M., and Roca Sagales, O., (2003), “Spillover Effects of Public Capital Formation: Evidence from the Spanish Regions,” Journal of Urban Economics, (53:2), pp. 238256.

    • Search Google Scholar
    • Export Citation
  • Pritchett, Lant, 1996, “Mind Your P’s and Q’s: The Cost of Public Investment is Not the Value of Public Capital,” Policy Research Working Paper 1660 (Washington: World Bank).

    • Search Google Scholar
    • Export Citation
  • Romp, W., and J. de Haan, 2005, “Public Capital and Economic Growth: a Critical Survey,” EIB Papers 2/2005, European Investment Bank, Economic and Financial Studies.

    • Search Google Scholar
    • Export Citation
  • Roy, R., A. Heuty, and E. Letouze, 2006, “Fiscal Space for Public Investment: Towards a Human Development Approach,” Paper Prepared or the G-24 Technical Meeting, Singapore, September 13–14, 2006.

    • Search Google Scholar
    • Export Citation
  • Sanchez-Robles, B., (1998), “Infrastructure Investment and Growth: Some Empirical Evidence,” Contemporary Economic Policy, (16), pp. 98108.

    • Search Google Scholar
    • Export Citation
  • Seung, C. K., and Kraybill, D. S. (2001), “The Effects of Infrastructure Investment: A Two Sector Dynamic Computable General Equilibrium Analysis for Ohio,” International Regional Science Review, (24:2), pp. 261281.

    • Search Google Scholar
    • Export Citation
  • Shioji, E., (2001), “Public Capital and Economic Growth: A convergence Approach,” Journal of Economic Growth, (6), pp. 205227.

  • Stephan, A., (2000), “Regional Infrastructure Policy and Its Impact on Productivity: A Comparison of Germany and France,” Applied Economics Quarterly, (46), pp. 327356.

    • Search Google Scholar
    • Export Citation
  • Stephan, A., (2003), “Assessing the Contribution of Public Capital to Private Production: Evidence from the German Manufacturing Sector,” International Review of Applied Economics, (17), pp. 399418.

    • Search Google Scholar
    • Export Citation
  • Sturm, J. E., Jacobs, J., and Groote, P., (1999), “Output Effects of Infrastructure Investment in the Netherlands, 1853–1913,” Journal of Macroeconomics, (21:2), pp. 355380.

    • Search Google Scholar
    • Export Citation
  • Vijverberg, W.P.M., Vijverberg, C.P.C., and Gamble, J.L., (1997), “Public Capital and Private ProductionReview of Economics and Statistics, (79:2), pp. 267278.

    • Search Google Scholar
    • Export Citation
  • Voss, G.M., (2002), “Public and Private Investment in the United States and Canada,” Economic Modelling, (19), pp. 641664.

  • World Bank, 2007, Fiscal Policy for Growth and Development Further Analysis and Lessons from Country Studies, March 28, 2007.

  • Zandi, Mark, 2008, “Financial Shock,” Pearson Education, INC., New Jersey.

Appendix Table: Literature Survey

article image
article image
article image
Source: Romp and de Haan (2005), IMF (2004) and World Bank (2007)