We set out to investigate the relationship between public debt and private investment using a panel of four countries in East Africa for the period 1992-2015. The results from the Autoregressive Distributed Lag Models show that Public Debt (PD) crowds out both Private Domestic Investment (PD I ) and Foreign Direct Investment (FDI) in the long run, although the magnitude of the impact is greater for the former category. We fail to find evidence of any short run significant relationship in either case. However, the importance of institutional quality in enhancing relationship in question is unquestionably confirmed in the data. The effect of PD on either PDI or FDI is observed to change when the corruption control improves. The immediate recommendation is the need to design fiscal policies to tame the growing debt that appears to discourage private investment in the region. A proper debt management system coupled with clear policies to improve the institutional quality would likely boost private investment in East Africa. The anti-corruption measures already in place should be enhanced to create a conducive investment climate for the private sector to thrive.
The East African countries have ambitious plans to transform their economies to middle income levels in the next decade. In an ardent effort to achieve this target, member countries have for the last decade undertaken public investment in infrastructure as well as investor-friendly strategies including but not limited to investment incentives and institutional enhancements. Intuitively, the strategies aim at creating a private-sector-driven economy. Given the small tax base and inadequate revenue characteristic of developing countries and East Africa in particular, the main source of finances in the development and expansion of infrastructure is public debt secured both domestically and externally. While public investment may be complementary to private investment particularly where the former increases capital productivity of the private sector, increases demand for inputs, and improves aggregate demand and savings, the link between public debt and private investment is still an empirical question. In principle, however, since the supply of money is fixed, domestic borrowing by the government may be at the cost of private investment since it will be withdrawn from the productive uses. Limited credit availability offered at high interest rates crowds out the private sector just as increased domestic borrowing may also result into high interest rates that cause an increase in cost of production, making tradable goods expensive and noncompetitive in foreign markets. Regarding these theoretical arguments, the key question still attracting debate is the extent to which public debt would quantitatively impact private investment and via which channels this would be possible. The current paper seeks to appreciate these concerns with reference to East Africa.
Certainly the empirical arena is not in scarcity of studies that examine the issue in question. However, a detailed scrutiny of the available studies informs us of a divergence in the findings. For example, while the likes of inter alia [
1The other three primary convergence criteria are: a ceiling on headline inflation of eight percent; reserve cover of 4.5 months of import; and, a ceiling on the overall deficit of three percent of GDP, including grants. All the four must be must be attained and maintained by each Partner State, for at least three years, before joining the Monetary Union.
Perhaps the reasons for the mixture of evidence regarding the debt-investment nexus are not self-explanatory but could have basis in type of debt considered, type of investment examined, sampling, methodology and data used. It appears too that overall; the results depend on the country or region under analysis. We note that there is scanty literature on studies that capture the linkage in question for the east African countries. Yet the region is continuously engaged in achieving greater regional integration and the Eat African monetary union (EAMU) in particular which, via the EAMU Protocol inter alia, requires a ceiling on gross public debt of 50 percent of GDP in net present value terms as one of the four primary convergence criteria1 According to [
With the aim of harnessing the private investment potential to promote economic growth and development in the region, the five partner states of the East African Community (EAC) agreed to cooperate in the areas of investment and industrial development [
A country specific comparison, from
From
On the other hand, the EAC Region recorded the highest share of foreign direct investment (FDI) across the continent, achieving 26.3% of total projects
with Kenya being among the region’s major beneficiaries [
In light of the above facts at hand, the popular question is the extent to which public debt affects domestic investment here categorized in form of private domestic investment and foreign direct investment. The current paper attempts to offer contribution to the debate. In theory, the purported relationship is certainly well-grounded. For example, whereas PD could be used for some public
investment that compliments private investment, prudent and sustainable debt management is imperative because a continual rise of domestic debt causes interest rates to soar and crowd-out private investors and annual interest increments on external debt could exceed all other spending [
We focus our study on three specific objectives. First is to investigate the effect of public debt on the level of private domestic investment. Second, we examine the impact of Public Debt on Foreign Direct Investment. Lastly, we analyze the impact of institutions on Private Investment. The latter objective is grounded in the argument that the perception of the institutional quality may, in addition to, inter alia the fear of prudent debt management and sustainability, and, debt overhang, impact on private investment. Specifically, institutional quality might reduce economic uncertainties, determine the ease of establishment and doing business and ensure efficient utilization of resources to avoid unsustainable debt levels in the public sector. By implication, sound and efficient institutions are likely to enhance good governance which is a cornerstone to efficient resource allocation and utilization. On the other hand, bad institutional quality such as corruption would likely lead to inefficiency in government spending, poor tax administration, misallocation of resources and embezzlement. With weak institutions in place, public borrowing may continue to increase to fund the ever increasing public expenditure on ineffective administration. The consequence might be an increase in the production costs (for example, poorly done infrastructure), which discourages private investment. According to the World Economic Forum [
In East Africa, member countries have put in place ethics and anti-corruption institutions in order to improve governance quality. According to World Governance Indicators of 2015, Rwanda is ranked highly on the corruption control measure, scoring 70 per cent, followed by Kenya at 53 per cent, Tanzania with 36 per cent and Uganda with 20 per cent. Given the aforementioned theoretical perspective to institutions vis-a-vis the ease of doing business, it is important to establish the extent to which institutional quality would influence the relationship between public debt and private domestic investment.
Overall, the results provide evidence of the detrimental role played by public debt in private investment in the long run. The crowding-out impact however is much significantly felt for Private Domestic Investment in relation to Foreign Direct Investment. We fail to find evidence of any short run significant relationship in either case. However, the importance of institutional quality in enhancing relationship in question is unquestionably confirmed in the data. The effect of PD on either PDI or FDI is observed to change when the corruption control improves. These findings have implications for policy.
The rest of the paper is organized as follows. In the next section, a detailed analysis of both theoretical and empirical literature is presented. This is followed by Section 3 where the model and data are discussed. We then present and discuss our results Section 4, and conclude in Section 5.
Investment models usually distinguish two separate elements in the investment process: the determination of a desired capital stock and the specification of an adjustment process by which the gap between existing and desired capital stock is filled [
One such model that hold relevancy to our study is the accelerator theory that can be traced back to [
Besides the accelerator theory, we acknowledge existence of other theories to explain investment. For example, Jorgenson’s [
On the other hand, Tobin’s q-theory extends the neoclassical theory by incorporating adjustment costs to account for losses in output. In addition to postulating that investment depends upon the ratio of the market value of a firm’s assets to their replacement cost, i.e. the q-ratio [
The current study adopts the Accelerator Model which appears to have more relevance to the developing country setting where the underdeveloped equity and bond markets are part and partial of the economies. Other models seem to lack this attribute. For example, the Tobin Q theory of investment has been criticized for oversimplifying rational expectations and efficient markets, and the possibility of generating different investment behavior from the specification of the firm’s alternative objective and production function [
2A detailed analysis of the theories of investment can be found in [
It has been pointed out earlier that the field under analysis is not in scarcity of empirical references. Interestingly however, are the divergent findings on effects of public debt on private investment. The different estimation techniques, variations in the datasets used, the sample space, as well as the varied indicators for debt and investment would perhaps account for the observed divergences in the findings, although we also admit that alternative explanations would be elsewhere. The various schools of thought nevertheless appear to rotate around four strands of literature: those in support of the crowding-out role; those for the crowding-in effect; those that are inconclusive; and, those arguing for a non-linear relationship with focus on the threshold effect.
To begin with, in demonstrating crowding out effect, using growth accounting [
Support for the crowding-out effect can further be located in a study by [
The other school of thought identifiable from the empirics includes protagonists for the crowding-in effect of debt on private investment. For example, [
The third strand of literature consists of categories of authors that provide inconclusive evidence on the relationship in question. For example, the second analysis of the aforementioned study by [
Beside the three strands presented above, we can also point out that other studies that argue for nonlinearity of the model that produces varying results depending on the threshold level of public debt. In this category, we identify among others, a study by [
We have earlier on argued that the role of institutions in the perceived link between public debt and private investment is theoretically plausible. The empirical arena however, is still characterized by scarcity in terms of related studies. A few available scholarly works focus on the direct effect of individual as well as aggregate institutions on private investment. The results re however still mixed up. For example, while the likes of [
In sum, empirical literature provides mixed results and calls for further debate regarding the issue in question. The importance of institutions in the debt-investment nexus is theoretically plausible but generally ignored in the existing literature. We focus on the East African Community as a region overly neglected in the existing analysis regarding the area under discussion. Yet, as alluded to earlier, in her efforts to integrate into a monetary union and to further consolidate her gains in terms of private sector investment with a dream to make it an engine of growth, the region urgently needs evidence-driven policies relevant for the member countries. Our main contribution pertains not only to adding to the pool of literature in the field but also to examine for the first time in the east African studies the quantitative role of corruption in channeling the effect of public debt on private investment.
As presented earlier, the current study adopts the Flexible Accelerator Model with a few extensions because it takes into account uncertainty common to developing economies. According the model, Capital is adjusted towards its desired level hence firm net investment is proportional to change in desired capital. Demand for capital increases when demand for output increases, hence the name accelerator model [
K p t ∗ = α Y t e (1)
where: K p t ∗ stands for desired stock of capital by private sector in period t; α Y t e is the expected level of output in period t.
Following [
K p t − K p t − 1 = β ( K p t ∗ − K p t − 1 ) , 0 ≤ β ≤ 1 (2)
Simplifying Equation (2) by making K p t the subject, we obtain:
K p t = β K p t ∗ + ( 1 − β ) K p ( t − 1 ) , 0 ≥ β ≤ 1 (3)
where:
K p t is the actual stock of private capital while β is the coefficient of adjustment. For estimation purposes the study specified a partial adjustment function by using the Gross Domestic Private Investment ( P D I t ) and ( F D I t ) rather than the private capital ( K p t ) due to lack of data on capital stock in most of the developing countries. Hence the gross Private Domestic Investment is given by:
P D I t = ( P D I t − P D I t − 1 ) + δ P D I t − 1 (4)
where: P D I t is the private domestic investment while δ is the depreciation rate of private capital stock (in this case, the private domestic investment). Equation (3) states that the gross private domestic investment is composed of the net and replacement components with the former component equal to the changes in capital stock while the latter is the capital stock available at the previous period. Suppose PDI depends on the desired and the past gross private investment, we re-write Equation (3) to obtain:
P D I t = β P D I t ∗ + ( 1 − β ) P D I t − 1 , 0 ≤ β ≤ 1 (5)
where = P D I t is already defined while P D I t ∗ represents the desired gross private domestic investment in the steady state. However, in the steady state, gross private domestic investment is given by:
P D I t = [ 1 − ( 1 − δ ) L ] P D I t ∗ (6)
where, P D I t ∗ means that the desired capital stock is related to the expected level of output such that P D I t ∗ = Y t e . L stands for the Lag operator defined as L Y t e = Y t − 1 e , while δ stands for the depreciation rate of capital stock. Hence after making all the necessary substitutions and then by combining Equations (5) and (6) we get:
P D I t = α β [ 1 − ( 1 − δ ) L ] Y t e + ( 1 − β ) P D I t − 1 (7)
Assuming that PD and other relevant variables denoted by vector X affect the speed of adjustment which determines the closure of the gap between the desired and actual gross Private Domestic Investment in each period. The coefficient β will hence vary with the factors that influence Private Domestic Investment and FDI. Hence if PD complements PDI, it speeds up the adjustment of the actual investment to the desired level of private investment and vice versa. Hence the speed of adjustment β is written in a linear form as:
β = b 0 + [ b 1 P D t + b 2 X t P D I t ∗ − P D I t − 1 ] (8)
where b 0 is the intercept, b are the coefficients to be estimated, and X t is a vector of other relevant variables that may affect PDI. Substituting Equation (7) into Equation (8) and then re-arranging we obtain:
P D I ( t ) = b 0 [ 1 − ( 1 − δ ) L ] Y t e + b 1 P D i t + b 2 X i t + ( 1 − b 0 ) P D I ( t − 1 ) (9)
However Equation (9) cannot be estimated because it contains some variables that are unobserved, Y t e , as well as depreciation rate, δ, whose data in SSA is lacking [
P D I i t = α 1 P D I i t − 1 + α 2 P D i t + β ′ X i t + v i + u i t (10)
where, P D I i t is the private domestic investment; P D I i t − 1 is the private domestic investment for the previous period; α is the coefficient for past gross private investment and PD; β ′ are the coefficients to be estimated; i = 1 , ⋯ , N and t = 1 ⋯ , T stands for countries and time periods respectively; v i are unobserved country effects and u i t is the error term. Year fixed effects are controlled for in Equation (10).
To simplify Equation (10) for estimation purposes, few variables are selected in place of X i t . The choice of variables of interest is informed by the economic theory which stipulates that as the economy grows, per capita incomes and domestic savings grow, therefore enabling credit creation for private investment. Additionally, improved institutional quality is a pre-requisite for PDI to thrive. Also, since the capital markets are underdeveloped in the least developed and developing economies, most enterprises access finance from the financial markets. This would hence imply that the cost of capital (real interest rate) determines the demand for capital and levels of investment. Finally, Foreign Direct Investment could also positively impact PDI through technology transfers and human capital development.
The variables chosen are hence PD, GDP growth, credit available to private sector (CP), Real Interest rate (RINT), Corruption Control (COR) and FDI to obtain the following equation:
P D I i t = α 1 P D I i t − 1 + α 2 P D i t + α 3 G D P i t + α 4 C P i t + α 5 R I N T i t + α 6 C O R i t + α 7 F D I + μ i t (11)
where: G D P i t is the Gross Domestic Product; C P i t is the credit to private sector; R I N T i t is the interest rate; and, C O R i t is the corruption control index.
In addition to PDI model specification, the study also estimated the impact of PD on FDI using the following equation:
F D I i t = κ 1 F D I i t - 1 + κ 2 P D i t + κ 3 P D I i t + κ 4 G D P i t + κ 5 H C i t + κ 6 T r a d e i t + κ 7 I N F i t + κ 8 C O R i t + μ i t (12)
where: F D I i t is the Foreign Direct Investment for period t; F D I i t − 1 is the Foreign Direct Investment for the previous period; G D P i t measures growth in Gross Domestic Product of the host country; T R A D E i t measures the trade openness of the host country; I N F i t is the inflation rate measured by the GDP deflator; C O R i t is the corruption control.
To evaluate the complementarily between PD and some policy variables in stimulating PDI and FDI, we interact PD with corruption index. By interacting PD with corruption index (PD × COR) the argument is that the impact of PD on PDI and FDI might be lower if there is good governance and hence low corruption levels.
P D I i t = α 1 P D I i t − 1 + α 2 P D i t + α 3 G P D i t + α 4 C P i t + α 5 R I N T i t + α 6 C O R i t + α 7 F D I i t + α 8 P D i t ∗ C O R i t + μ i t (13)
F D I i t = κ 1 F D I i t − 1 + κ 2 P D i t + κ 3 G D P i t + κ 4 H C i t + κ 5 P D I i t + κ 6 T r a d e i t + κ 7 I N F i t + κ 8 P D i t ∗ C O R i t + μ i t (14)
where: H C i t is the human capital Development; P D i t ∗ C O R i t is the interaction term showing the indirect effect of PD on PDI via corruption.
The marginal impact of PD on PDI in the presence of corruption would for example be captured by differentiating Equation (11) with respect to COR to get:
∂ P D I i t ∂ P D i t = α 2 + α 8 C O R i t (15)
From Equation (15), if α 8 < 0 , and the absolute value exceeds α 2 > 0 it implies that a one percentage point increase in PD yields a negative impact on PDI as corruption decreases. Conversely, if α 8 > 0 , PD increases PDI if corruption levels decrease with it. The interaction term between PD and corruption in Equations (13) and (14) test if the rise in PD is as a result of increased corruption levels in the countries under review and whether it lowers or increases PDI or FDI.
Additionally, according to the crowding out hypotheses, the accumulation of a large debt may stifle economic growth through lower investment. On this basis, we assume that public debt would be beneficial to investment up to a certain threshold. Once debt surpass this threshold, it will start to be a constraint to investment. Therefore in order to check for the Public Debt sustainability threshold or to determine the PD-GDP ratio turning point, we introduce a nonlinear relationship between public debt and domestic investment, as in [
P D I i t = θ 0 + θ 1 P D i t + θ 2 P D i t 2 (16)
where P D i t 2 is the Public Debt Squared. The inclusion of the squared variable affords us the opportunity to investigate the non-linearity effect of public debt on private investment, as well as analyzing the values of public debt thresholds. As in [
∂ P D I i t ∂ P D i t = ∂ ( θ 0 + θ 1 P D i t + θ 2 P D i t 2 ) ∂ P D i t
0 = θ 1 + 2 θ 2 P D i t
⇔ P D i t ( threshold ) = − θ 1 2 θ 2 (17)
Equation (18) is the debt-GDP ratio turning point or the threshold level of debt. Here, θ 1 is the coefficient of the linear term and θ 2 is the coefficient of the quadratic term.
Since it is likely that investment in previous periods could affect current investments, and that the explanatory variables could be endogenous, estimating our model with static techniques would be inappropriate and lead to inconsistent outcomes. Various techniques can be used in such circumstances, among which dynamic panel ARDL, and the system Generalized Method of Moments is (GMM) by [
Expressing our model in an ARDL framework, we get:
Δ P D I i t = α 0 + α 1 P D I i t − 1 + α 2 P D i t − 1 + α 3 G D P i t − 1 + α 4 C P i t − 1 + α 5 R I N T i t − 1 + α 6 C O R i t − 1 + α 7 F D I i t − 1 + α 8 P D * C O R i t − 1 + β 1 ∑ i = 0 m Δ P D I i t − 1 + β 2 ∑ i = 0 n Δ P D i t − 1 + β 3 ∑ i = 0 p Δ G D P i t − 1 + β 4 ∑ i = 0 q Δ C P i t − 1 + β 5 ∑ i = 0 r Δ R I N T i t − 1 + β 6 ∑ i = 0 s Δ C O R i t − 1 + β 7 ∑ i = 0 y Δ F D I i t − 1 + β 8 ∑ i = 0 z Δ P D * C O R i t − 1 + u i t (18)
Δ F D I i t = α 0 + α 1 P D i t − 1 + α 2 G D P i t − 1 + α 3 H C i t − 1 + α 4 T R A D E i t − 1 + α 5 C O R i t − 1 + α 6 I N F i t − 1 + α 7 P D * C O R i t − 1 + β 1 ∑ i = 0 e Δ P D i t − 1 + β 2 ∑ i = 0 f Δ G D P i t − 1 + β 3 ∑ i = 0 g Δ H C i t − 1 + β 4 ∑ i = 0 h Δ T R A D E i t − 1 + β 5 ∑ i = 0 j Δ C O R i t − 1 + β 6 ∑ i = 0 k Δ I N F i t − 1 + β 7 ∑ i = 0 l Δ P D ∗ C O R i t − 1 + u i t (19)
The ARDL panel model offers two estimators: The Mean Group Estimator (MG) and the Pooled Mean Group (PMG). The MG suggested by [
However, before the estimation of the co-integration relationship, the ARDL model requires that variables are stationary at level, at first difference or variables are stationary at both level and first difference. The technique may not work in cases where variables are stationarity at second difference. For purpose of checking for stationarity therefore, the study adopted the two unit root test: the Levin, Lin and Chu (LLC) Test which assumes that there is a common unit root process so that the autoregressive coefficients are identical across countries; and, the Im Pesaran and Shin (IPS) Test which allows for individual unit root processes and hence heterogeneous autoregressive coefficients across countries.
The study is based on the EAC involving Kenya, Uganda, Tanzania and Rwanda leaving out Burundi and Southern Sudan due to unavailability of data, for the period under study. To some extent thee EAC countries are homogenous in terms of general policies on investment for implementation in the common market. Panel secondary data used covers the period from 1992 to 2015. The period is selected based on the availability of data. Note also that during this period these countries appear to have developed strategies and policies to promote good governance as well as private and foreign investment. It is also observed that the corruption levels in the region were increasing during this period despite the anti-corruption measures in place. Moreover, during this period, the IMF and World Bank came up with concessional facilities for HIPCs to augment their development prospects.
The Private Domestic Investment ranges between 9.983 and 33.24, indicating that 5.098 of the values deviate from the mean. The mean of the Private Domestic Investment is closer to the maximum, hence negatively skewed. The PD as a percentage of GDP ranges between 19.19 with a deviation of 30.15 from the mean. FDI ranges between 0 and 6.48 with a deviation of 1.82. Similarly, the correlation matrix, in
Variable | Definition | Source |
---|---|---|
P D I i t | Private Domestic Investment, Measured by Gross fixed capital formation, private sector (% of GDP) | WDI |
PD | Public Debt as percentage of GDP | IMF |
P D I i t − 1 | Private investment as a ratio of real GDP (lagged P D I i t ) | WDI |
G D P i t | Gross domestic product growth (%) | WDI |
C P i t | Domestic credit to private sector(% GDP) | WDI |
R I N T i t | Real interest rate (%) is the lending interest rate adjusted for inflation as measured by the GDP deflator. | WDI |
C O R i t | Corruption control | WDI |
H C i t | Human capital Development, proxied by secondary school enrolment | WDI |
T R A D E i t | Trade as a percentage of GDP (proxy for liberalization) | WDI |
I N F i t | Inflation (GDP Deflator) | WDI |
Note: WDI is world development indicators sourced from World Bank.
Variable | Obs (N) | Mean | Std Dev | Min | Max |
---|---|---|---|---|---|
PDI | 96 | 20.39 | 5.10 | 9.983 | 33.24 |
PD | 96 | 59.56 | 30.15 | 19.19 | 138.3 |
COR | 80 | 42.52 | 16.39 | 20.10 | 82.78 |
GDP | 96 | 5.488 | 7.24 | −50.25 | 35.22 |
FDI | 96 | 2.150 | 1.82 | 0.000133 | 6.48 |
CP | 96 | 14.03 | 8.06 | 3.09 | 34.89 |
RINT | 93 | 9.247 | 7.33 | −9.74 | 28.66 |
Trade | 96 | 45.35 | 11.27 | 23.83 | 72.86 |
INF | 96 | 11.26 | 10.31 | −9.19 | 51.27 |
HC | 96 | 2.20 | 2.6 | 0 | 9.01 |
Source: Author’s computation.
GDP | PD | FDI | RINT | COR | CP | HC | Trade | INF | |
---|---|---|---|---|---|---|---|---|---|
GDP | 1 | ||||||||
PD | 0.058 | 1 | |||||||
FDI | 0.128 | −0.416 | 1 | ||||||
RINT | 0.052 | 0.214 | −0.027 | 1 | |||||
COR | 0.418 | −0.015 | −0.021 | −0.008 | 1 | ||||
CP | −0.334 | −0.354 | −0.426 | −0.034 | −0.274 | 1 | |||
HC | −0.218 | −0.345 | −0.247 | −0.059 | −0.143 | 0.538 | 1 | ||
Trade | −0.374 | −0.579 | −0.036 | −0.188 | −0.424 | 0.711 | 0.543 | 11 | 1 |
INF | −0.217 | −0.071 | 0.052 | −0.858 | −0.149 | −0.068 | −0.099 | 0.159 | 1 |
Source: Author’s computation based on World Bank WDI and WGI statistics.
In
LLC Test (t-statistic) | IPS Test (t-statistic) | LLC Test (t-statistic) | IPS Test (t-statistic) | Order | |
---|---|---|---|---|---|
level | 1ST Difference | ||||
PI | −3.084 (0.001) | −3.372 (0.0003) | −6.7158 (0.0000) | −5.7547 (0.0000) | I (0) |
PD | −2.013 (0.022) | −2.605 (0.0414) | −3.2129 (0.0007) | −3.5842 (0.0001) | I (1) |
FDI | −1.382 (0.083) | −2.970 (0.0025) | −4.6783 (0.0000) | −7.1446 (0.0000) | I (1) |
COR | −2.1544 (0.0156) | −2.7539 (0.008) | −2.9228 (0.0017) | −4.8725 (0.0000) | I (1) |
RINT | −4.666 (0.000) | −6.5737 (0.0000) | I (0) | ||
CP | −1.9334 (0.0266) | −2.1535 (0.0684) | −3.2602 (0.0006) | −5.0612 (0.0000) | I (1) |
HC | −0.0712 (0.4716) | −2.701 (0.0095) | −0.9655 (0.1672) | −6.3866 (0.0000) | I (1) |
Trade | −3.5859 (0.0002) | −4.339 (0.000) | −11.8764 (0.0000) | −8.1651 (0.0000) | I (0) |
INF | −4.8041 (0.0000) | −5.189 (0.000) | −7.3573 (0.0000) | −6.9804 (0.0000) | I (0) |
Note: LLC is Levin-Lin-Chiu test; IPS is Im-Pesaran-Shin test; RINT lacks strongly balanced data which is a prerequisite for LLC test; p-values in parenthesis.
Pooled Mean Group and Mean Group Results | ||||||
---|---|---|---|---|---|---|
PMG | Hausman test | MG | ||||
Coefficient | Std. Error | h-test | p-value | Coefficient | Std. Error | |
Long Run | ||||||
PD | −0.0687*** | 0.1164 | −0.129** | 0.0581 | ||
GDP | 0.1579* | 0.0928 | −0.0200 | 0.124 | ||
CP | 0.5020*** | 0.0948 | 0.798*** | 0.190 | ||
RINT | −0.1494*** | 0.0419 | −0.130 | 0.112 | ||
FDI | 0.2886 | 0.3671 | 0.439 | 0.349 | ||
COR | 0.1034*** | 0.0243 | 0.109** | 0.0693 | ||
Short Run | ||||||
ECT | −0.5582*** | 0.0869 | −0.7980*** | 0.0257 | ||
PD | 0.0018 | 0.0537 | −0.0066 | 0.0921 | ||
GDP | 0.1670 | 0.1545 | 0.2206 | 0.1857 | ||
RINT | 0.0503* | 0.0269 | 0.0702 | 0.0208 | ||
CP | 0.1577 | 0.4595 | −0.1481 | 0.4066 | ||
FDI | −0.2920 | 0.2057 | −0.4460 | 0.2597 | ||
COR | 0.0366 | 0.0552 | 0.00599 | 0.0376 | ||
Hausman test ( χ 2 ) | 2.34 | 0.8860 |
Note: PD (Public Debt), GDP (Gross Domestic Product), CP (Credit to Private Sector), RINT (Real Interest Rate), FDI (Foreign Direct Investment) and COR (Corruption Control). The hausman test with a p-value of 0.8860 is greater than 5% and informs the choice of PMG over the MG estimator. *, **, *** indicate significance at 10%, 5% and 1% respectively. Source: Author’s computation.
the null hypothesis implies that PMG is the preferred estimator, where the estimator restricts Long Run equilibrium between variables to be homogenous across countries or a subset of them. The Hausman test with an h-statistic of 2.34 and a p-value of 0.8860, which is greater than 5 percent level of significance implies that there is slope homogeneity and that PMG is the preferred estimation model. An important note here is that error correction term (ECT) coefficient of −0.5582 at one percent level of significance, is indicative of a disequilibrium in the previous period being corrected at a speed of 55 per cent to reach a steady state.
In the Short Run, PD is observed not significant whereas in the Long Run both variables attract an inversely relationship. Specifically, a unit increase in PD would decrease PDI by approximately 0.0687 units, which indicates that PD crowds out PDI. This confirms the priori expectation and is in line with the findings in [
Besides the PD-PDI relationship,
From the same table, only inflation, as a measure of the macroeconomic environment, proxied by GDP deflator is significant at 10 percent and has a positive impact on FDI in the Short Run. In the Long run, in line with the expectation, inflation is found to negatively influence FDI. Perhaps once viewed as a future cost of investment inflation enters the investors’ objective function as a policy variable that measures policy soundness that governments and central banks undertake and therefore the marginal effect of inflation on FDI is observed negative.
Corruption control (COR) is neither significant in the Short Run nor in the Long Run. This could imply that institutions of host countries do not matter for
PMG | MG | |||
---|---|---|---|---|
Coefficient | Std. Error | Coefficient | Std. Error | |
Long Run results | ||||
PD | −0.0409*** | 0.0108 | −0.0266 | 0.0307 |
COR | −0.0171 | 0.0202 | 0.0543 | 0.0635 |
GDP | 0.1499 | 0.1020 | 0.0183 | 0.1104 |
HC | 0.0471 | 0.0852 | 0.0530 | 0.2728 |
Trade | −0.0705 | 0.0434 | 0.0676 | 0.1906 |
INF | −0.0658** | 0.0313 | −0.0813** | 0.0409 |
Short Run Results | ||||
ECT | −0.6426*** | 0.2106 | −1.0002*** | 0.0363 |
PD | −0.0104 | 0.0109 | −0.0058 | 0.0197 |
COR | 0.0130 | 0.0212 | −0.0055 | 0.0482 |
GDP | −0.0538*** | 0.0114 | −0.0605 | 0.0484 |
HC | 0.0188 | 0.0864 | 0.0225 | 0.1904 |
Trade | 0.0072 | 0.0373 | −0.1082 | 0.0994 |
INF | 0.0245* | 0.0132 | 0.0665 | 0.0499 |
Note: PD (Public Debt), COR (Corruption Control), GDP (Gross Domestic Product), HC (Human Capital), INF (Inflation), ECT (Error Correction Term); *, **, *** indicate significance at 10%, 5% and 1% respectively. Source: Author’s computation.
FDIs. This is consistent with the findings in [
The Long Run total impact of PD on PDI in the presence of an improving institutional environment from PMG regression turns out to be is 42.52 (i.e. 0.155 + [(−0.00388)(42.52)]). It can be noted that the impact of PD on PDI, when
Pooled Mean Group and Mean Group Results | ||||||
---|---|---|---|---|---|---|
PMG | MG | |||||
Coefficient | Std. Error | h-test | p-value | Coefficient | Std. Error | |
Long Run Results | ||||||
PD | 0.155*** | 0.0342 | −0.117 | 0.387 | ||
GDP | 0.247*** | 0.0237 | −0.221 | 0.307 | ||
CP | 0.343*** | 0.0584 | 0.807** | 0.355 | ||
RINT | 0.830*** | 0.0775 | 0.111* | 0.064 | ||
FDI | −0.774*** | 0.218 | −0.989*** | 0.354 | ||
COR | −0.166*** | 0.0229 | 0.117 | 0.526 | ||
PD × COR | −0.004*** | 0.0006 | 0.002 | 0.009 | ||
Short Run Results | ||||||
ECT | −0.634** | 0.217 | −0.987* | 1.409 | ||
PD | −0.257** | 0.120 | 0.0691 | 0.651 | ||
GDP | 0.0762 | 0.190 | 0.291 | 0.274 | ||
RINT | 0.0544 | 0.0210 | −0.0834 | 0.150 | ||
CP | 0.0593 | 0.408 | −0.558 | 0.495 | ||
FDI | −0.502 | 0.390 | 1.352 | 1.490 | ||
COR | −0.304 | 0.272 | −1.117** | 0.472 | ||
PD × COR | 0.005 | 0.004 | 0.009*** | 0.003 | ||
Hausman test χ 2 | 2.34 | 0.8860 |
Note: *, **, *** indicate significance at 10%, 5% and 1% respectively. Source: Author’s computation.
factoring in corruption (−0.01) is less than −0.0687 (impact of debt without factoring in institutions as in
PMG | MG | |||
---|---|---|---|---|
Coefficient | Std. Error | Coefficient | Std. Error | |
Long Run Results | ||||
PD | −0.1731*** | 0.0447 | −0.4739 | 0.5821 |
COR | −0.1172*** | 0.0423 | −0.7133 | 0 .9541 |
GDP | 0.1559 | 0.1079 | −0.1896 | 0.3283 |
HC | 0.0353 | 0.08754 | −0.3268 | 0.5505 |
Trade | −0.1182*** | 0.0364 | 0.1652 | 0.1480 |
INF | −0.1228*** | 0.0384 | −0.0408 | 0.0408 |
PD × COR | 0.0033*** | 0.0011 | 0.0143 | 0.0175 |
Short Run Results | ||||
ECT | −1.09*** | 0.2240 | −1.3009*** | 0.8778 |
PD | −0.0982 | 0.0880 | −0.0308 | 0.3053 |
COR | −0.0959 | 0.1548 | 0.2205 | 0.364 |
GDP | −0.0664 | 0.0346 | 0.389* | 0.2122 |
HC | −0.0177 | 0.0188 | 1.2495 | 1.4532 |
Trade | −0.0181 | 0.0220 | −0.365736 | 0.3718 |
INF | 0.0440* | 0.0231 | 0.1231* | 0.0710 |
PD × COR | 0.0010 | 0.0023 | −0.0017 | 0.0055 |
Note: *, **, *** indicate significance at 10%, 5% and 1% respectively. Source: Author’s computation.
value exceeds coefficient of PD (−0.1731), it implies that a one percent increase in PD yields an improvement on FDI as institutional quality grows with it. Therefore, the impact of institutions proxied by corruption-control supports the hypothesis that good institutions matter for Foreign Direct Investment. Intuitively, corruption reduces Foreign Direct Investment, a finding consistent with [
PDI (Dependent Variable) | Coefficient | Std Error | p-value | Z-Statistic |
---|---|---|---|---|
PD | −0.4317 | 0.0833 | 0.000 | −5.18 |
PD^2 | 0.0023 | 0.0006 | 0.000 | 3.60 |
PD Turning point | 94.93 | |||
FDI (Dependent variable) | ||||
FD | −0.0192912 | .0140124 | 0.169 | −1.38 |
FD^2 | −0.0000669 | .0001058 | 0.527 | −0.63 |
Note: A constant is included in each model. The threshold or turning point is calculated using equation (18) in the text. Source: Author’s computation.
Implicitly, for public debt to have a positive impact on private investment, borrowed amounts must be invested productively and in a manner that is private-sector-enhancing or else the impact of debt on private investment may always be negative.
However, given the popular argument shared by [
In order to check the validity of our study, we conduct several robustness checks. First, we carried out a country-by-country analysis allowed by the ARDL technique. Since the results do not substantially alter the original findings, and due to space limitations, we do not present them here but they are available on request. Second, we adopted a common practice in several empirical works of using three-year averages of all the variables to eliminate short run fluctuations. Here we notice a few ignorable changes that do not seriously alter interpretation of the original findings. Similarly, one would perhaps argue that the findings could be driven by the existence of more or less developed economies in East Africa in relation to their counterparts. We therefore dropped Kenya from the sample. In turn, we also dropped Rwanda to remain with the original EAC countries. Later we also dropped the most corrupt country just as we in turn dropped the country with the largest public debt on average during the study period. Unsurprisingly, the results were never substantially affected in each of the “droppings”, although some control variables turned out more significant. Given the high similarity rate with the original findings, still we have spared space and not presented the robustness results here but they are available on request.
The study analyzed the relationship between Public Debt and Private Domestic Investment and also investigated the impact of PD on FDI in the EAC countries. The findings indicate a crowding out effect of Public Debt. It is further noted that the magnitude of the Public Debt’s impact is greater for PDI as compared to FDI. Additionally, it appears that interaction of public debt and corruption control improves Private Investment in the Long run. This has the implication that enhancing of institutional quality is vital in the promotion and development of Private Investment. Our findings also point to a nonlinear relationship between public debt and domestic investment. The results for a panel of the 4 East African countries over the period 1992-2015, indicate that public debt lower than 94.93 percent of GDP is positively associated with private domestic investment. Otherwise, once the debt exceeds this threshold, the relationship between public debt and investment becomes negative.
The results from the study have policy implications. The immediate recommendation is the need to design fiscal policies to tame the growing debt that discourages private investment. In addition to the urgent necessity to reduce reliance on non-concessional borrowing in refinancing the debt and lowering fiscal vulnerabilities, there is need for a proper debt management system to lower fiscal vulnerabilities, coupled with clear policies to improve the institutional quality in order to boost private investment in East Africa. Any existing fiscal adjustment efforts that are focused on both expenditures and revenues together with complimentary monetary policies deserve commendation. Also, the anti-corruption measures already in place and those in the pipeline should be strongly supported to create a conducive investment climate for the private sector to thrive.
Besides the policy implications, the study appears to have insinuated further debate in related areas. For example, the finding of an insignificant relationship between FDI and PDI may not be taken on face value. Perhaps, it would be interesting to find out the empirical rationale behind such an outcome in at a more detailed level regarding FDI spillover effects. A related area of interest but which was outside the scope of our study is decomposing private investment by category and taking them as separate dependent variables. Perhaps such an analysis would provide a more detailed picture of the debt-investment link. However, such a kind of analysis would be limited by data availability. A similar limitation is likely a hindrance to repeating the analysis by disaggregating public debt in order to determine which category of debt impacts greatly on investment. Otherwise once data gets available in future, such would be an interesting area to better understand the debt-investment nexus.
The authors declare no conflicts of interest regarding the publication of this paper.
Aswata, M.B., Nnyanzi, J.B. and Bbale, J.M. (2018) Debt, Corruption and Investment in East Africa: A Panel ARDL Analysis. Modern Economy, 9, 2012-2038. https://doi.org/10.4236/me.2018.912126