^{1}

^{1}

^{*}

^{1}

Sri Lanka is relying heavily on public debt to finance the budget deficit since its independence from British in 1948. Thus, it is much important to investigate the long run of public debt on economic growth of the country for the period 1977 to 2012 using time series data. Sri Lanka introduced fully liberalized economic policy in 1977. The study used domestic debt, external debt and educational expenditure as explanatory variables to determine their effect on GDP in the long run. Long run is estimated by employing Johansen test of cointegration analysis relies on Vector Error Correction Model (VECM). The coefficient of Error Correction Term (ECT) suggests disequilibrium that is corrected at the speed of 58 percent over the each year. Significant ECT is a proof of the existence of long run relationship.

Government needs to borrow for two broad categories; macroeconomic purposes such as higher investment, higher consumption (education and health) or to finance transitory balance of payments deficits [to lower nominal interest rates abroad, lack of domestic long-term credit, or to circumvent hard budget constraints] [

But it has been observed since 1900; most of the counties in both advanced and emerging category accumulate debt/borrowings mainly for the purpose of financing budget deficits. Situations where advanced economies also happened to depend on the borrowings because of large stock of defense expenditure spent on for world war period in early 20^{th} century and withstand for adverse effects of such as the oil prices crisis and debt crisis occurred in 1970-1980. However, for those less debt burdened nations, association of debt seemed to have higher growth rates where emerging countries and less developed countries kept accumulating debt for promoting economic growth due to their limited capacity to promote economic growth and to bridge budget deficits.

Governments prefer accumulation of debt in financing budget deficits because it is an anti-inflationary mechanism unlike printing money or imposing taxes. Although government can use taxes to finance its budget deficit, taxes tend to distort the structure of relative prices, and borrowing, if pushed beyond the carrying capacity of the economy, creates problems of intergenerational equity [

As the importance of the study, debt is improving welfare and enhancing growth at moderate level but there is damage from high level of debt [

Prior literature has shown a mixed impact of public debt on economic growth. Important point is according to the context those studies are carried out the results are varied thus there is no common agreement. A different political parties governing since the independence have been using debt finance to meet budget deficits.

Financing budget deficit through sources such as printing money, increasing taxes and cutting down government expenditure lead to increase price level of the country which may result in deteriorating living standards of the people [

This theory was developed by David Ricardo in the nineteenth century. He suggests that when a government tries to stimulate demand by increasing debt-financed government spending, demand remains unchanged. This is because the public will save its excess money in order to pay for future tax increases that will be initiated to pay off the debt.

According to the Ricardian equivalence theory Bernheim [

Later Harvard professor Robert Barro [

As stated by Buchchan Barro’s [

For more than eight decades, famed economist John Maynard Keynes has been the subject of much discussion and debate while the neoclassical and Ricardian schools of thoughts focus on the long run effects and Keynesian view emphasizes the short run effects [

As highlighted by Bernheim [

Keynes theories on economics ignored the basic laws of supply and demand. However, unlike economists before to Keynes, he argued that demand drives supply. In Keynes view, insufficient demand leads to excess supply of goods, which leads to cease production. Ultimately, weak demand leads to a downward trend in unemployment due to downsizing, and then poverty and even depression of economy. In order to remedy this weak/insufficient demand Keynes doctrine was to artificially stimulate demand by increasing government spending or cutting taxes, which then encourages the public to increase its spending as a result. The argument was to cause either increase government or the public spending to stimulate the under active economy (http://www.maynardkeynes.org/).

According to Keynes these objective of stimulating economy could be achieved through fiscal policy; expending public spending or tax cuts. Out of these two options Keynes concerned on cutting taxes rather than increasing expenditure to boost the underactive economy. Keynes favored government deficit spending through increasing expenditure only to handle economic depressions, not to uptrend low levels of unemployment created through weak demand. He also favored creating surplus budgets to eliminate government debt when the economy is overly active. Thus, Keynes was flexible for both situations concerning short term. (http://www.maynardkeynes.org/).

The primary concern about large government debt for financing deficit would lead inflation. Notably, Keynes advised that inflation could be treated with the help of budget surpluses and/or restrictive monetary policy. Though Keynes argued for budget deficits to stimulate demand, he also advocated for subsequent budget surpluses to eradicate debt [

Positive effects of public debt relate to the fact that in resource-starved economies debt financing if done properly leads to higher growth and adds to their capacity to service and repay external and internal debt [

In contrast, Shah and Pervin [

Barik [

As per debt guide (2013); debt liabilities owed by residents to residents of same economy are domestic debt. Reinhart, Rogoff, and Savastano [

Debt guide (2013); debt liabilities owed by residents to nonresidents are external debt. Reinhart, et al. [

In most literature evidence public debt is classified as sum of external debt and domestic debt [

Basically budget deficits are financed by; printing money, foreign borrowings, domestic borrowings and running down foreign exchange reserves [

When conducting time series studies prior literature has obliged to check univariate time series of variable by using a unit root test in each series before estimating any equation. It is considered as a problem if there is a unit root because particular series considered being non-stationary. Granger and Newbold [^{2} value and t statistics, but without any coherent economic meaning. Further they stated that in order to series to be stationary expected mean and variance of the variables must be stable over the period and cross-observation relations should not be remained. But the inherent feature of time series data used in such economic variables is to comprise of this non-stationary problem. Because there is a greater possibility of values in time series data are affected by temporary shocks occurred in previous time period, and such shocks would cause damages to remain the stationarity of entire time series appearing as permanent shocks. Variables which are non-stationary considered as unit root. Since, this study entirely a macroeconomic study and there is such a possibility of these variables containing non-stationary problem.

In order to perform the unit root test, this study has adopted the Augmented Dickey-Fuller (ADF) test of checking unit root because ADF unit root testing procedure is well established in the literature. Three different specifications of ADF are available. The first include both trend and intercept, second includes intercept without trend term, third exclude both trend and constant i.e. none. In this study first specification is adopted to test the null hypothesis; the series contain a unit root. Then the results are analyzed whether the variables are stationary at level or at 1^{st} difference I (1) at 5 per cent significance level. If all the variables contain unit root at level I (0) and stationary at 1^{st} difference considered as variables are integrated in same order. Therefore it is expected such integrated variables are cointegrated. Thus, unit root test satisfies the requirement to perform a cointegration analysis to examine whether there is a long run association from independent variables to the dependent variable.

It is also important to select appropriate lag length. Harris and Sollis (2003) stated [

After checking whether all the time series macro-economic variables deployed are integrated in the same order, subsequently the study test cointegration among the variables (Domestic Debt (DD), External Debt (ED), Educational expenditure (Edu) and Gross Domestic Production (GDP)).

The objective of adopting the cointegration technique is to determine the long run association among macroeconomic variables in the model. Previous authors who adopted Cunningham’s [

Two main cointegration techniques are generally used; Engle and Granger [

In order to determine the number of cointegrating vectors, this procedure contains two different likelihood ratio tests to decide the value of cointegrating vector, namely trace statistics test and Maximum Eigenvalue test statistics. It is expected to employ both these ratios to determine the number of cointegration equations. Trace test formulated as;

L R = T ∑ i = r + 1 n ln ( 1 − λ i ) (1)

Trace statistics test is combined test where null hypothesis is that the number of cointegrating vectors are less than or equal to r opposed to alternative hypothesis that there are more than r cointegrating vectors. And the Maximum Eigenvalue test statistics is;

L R = T ln ( 1 − λ r + 1 ) (2)

Here, null hypothesis is, number of cointegrating equations is r against the alternative hypothesis of r + 1 where Maximum Eigenvalue test conducts separate tests for each Eigenvalue. Thus, the null hypothesis was tested sequentially from low to high values of r. In this study, researcher expects to employ both these ratios to determine the number of cointegration equations and either one which provide better result is considered.

But the cointegrating vectors produced by Johansen test of cointegration still contain unit root and thus the results are spurious. Therefore as the next step of Johansen test of cointegration error terms of each cointegration vector should be corrected and to perform this estimation study adopts Vector Error Correction Model.

As stated by Johansen and Juselius [

Second implication is; Johansen and Juselius [

In VECM, variables are expressed in first difference (except for ECT) and thereby non-stationary problem is eliminated. Therefore model used in this study included in the following error correction Thus, vector error correction equation for the model deployed of the study will be;

ln GDP t = β 0 + ∑ i = 1 n β 1 ln GDP t − i + ∑ i = 1 n β 2 DD t − i + ∑ i = 1 n β 3 ED t − i + ∑ i = 1 n β 4 EDU t − i + ε t (3)

ln GDP t = β 0 + ∑ i = 1 n β 1 Δ ln GDP t − i + ∑ i = 1 n β 2 Δ DD t − i + ∑ i = 1 n β 3 Δ ED t − i + ∑ i = 1 n β 4 Δ EDU t − i + ∝ EC t − 1 + ε t (4)

where,

lnGDP_{t}: Per capita GDP in natural logarithmic form;

DD: Domestic Debt as a percentage of GDP;

ED: External Debt as a percentage of GDP;

EDU: (Proxy variable for human capital) Annual education expenditure;

ε_{t}: Random error term;

EC: Error correction term.

Researcher applied Equations ((3) and (4)) in the analysis which represents vector auto regression and Vector error correction models respectively.

Main source of the debt data are obtained from the publications of Public Debt Department of Central Bank of Sri Lanka (CBSL). The reliability of the publication “Public Debt Department in Sri Lanka, Performance in 2012 and Strategies for 2013 and beyond” for debt data is high because CBSL remains the responsibility of managing public debt in Sri Lanka under the Monetary Law Act No. 58 of 1949. Education Expenditure (proxy variable for Human Capital) is acquired from the publications of Census and Statistic Department of Sri Lanka. Researcher used annual data in the study.

Augmented Dickey Fuller (ADF) unit root analysis has been applied to test the stationary of data (

Variables | At level | At first difference |
---|---|---|

GDP | 0.3095 | 0.0082*** |

DD | 0.4492 | 0.0006*** |

ED | 0.2042 | 0.0000*** |

Edu. | 0.9999 | 0.0172** |

Note: *** and** indicate statistically significant at 1% level and 5% level respectively.

they are having trend.

According to the Schwarz information criterion (SC), and Hannanquinn information criterion (HQ), there is one lag length (

Trace statistics of Johansen test of cointegration propose three cointegration equations in to the VAR model at 5% level of significance (

Two Debt variables (DD, ED) and the education (EDU) have obtained negative coefficients. They say that in the long run there is a negative influence on

VAR lag order selection criteria Endogenous variables: GDP DD ED EDU | ||||||
---|---|---|---|---|---|---|

Lag | LogL | LR | FPE | AIC | SC | HQ |

0 | −375.0981 | NA | 111771.2 | 22.97564 | 23.15704 | 23.03668 |

1 | −164.3427 | 357.6456 | 0.843816 | 11.17228 | 12.07926* | 11.47745* |

2 | −152.9286 | 16.60231 | 1.168070 | 11.45022 | 13.08277 | 11.99952 |

3 | −128.1870 | 29.98981* | 0.778050* | 10.92042* | 13.27856 | 11.71386 |

Note: * indicates the number of lag length.

Hypothesized | Trace | 0.05 | ||
---|---|---|---|---|

No. of CE(s) | Eigenvalue | Statistic | Critical value | Prob.** |

None *** | 0.680013 | 72.19233 | 47.85613 | 0.0001 |

At most 1 *** | 0.454941 | 35.72918 | 29.79707 | 0.0092 |

At most 2 ** | 0.398703 | 16.30964 | 15.49471 | 0.0376 |

At most 3 | 0.001009 | 0.032316 | 3.841466 | 0.8573 |

Note: *** and** indicate statistically significant at 1% level and 5% level respectively.

economic growth. However this long run equilibrium is still inconclusive because significance of long run association of the variables has not been estimated. Therefore, in order to evaluate the long run causality/association, Vector Error Correction Model is employed in

Probability of error correction term that is 0.039 with negative beta of −0.58 and

Normalized cointegrating coefficients (standard error in parentheses) | |||
---|---|---|---|

GDP | DD | ED | EDU |

1.000000 | −0.560330 | −0.122317 | −7.45E−06 |

(0.03155) | (0.03292) | (5.5E−07) |

Dependent variable: D(GDP) | ||||
---|---|---|---|---|

Coefficient | Std. error | t-Statistic | Prob. | |

Error correction term** | −0.582624 | 0.259620 | −2.244145 | 0.0393 |

Gdp (−1)* | 0.319841 | 0.180720 | 1.769815 | 0.0958 |

Gdp (−2) | 0.034285 | 0.177651 | 0.192988 | 0.8494 |

Gdp (−3) | 0.254077 | 0.151937 | 1.672244 | 0.1139 |

DD (−1) | −0.096492 | 0.148709 | −0.648865 | 0.5256 |

DD (−2) | 0.147769 | 0.162885 | 0.907196 | 0.3778 |

DD (−3) | 0.061271 | 0.126894 | 0.482854 | 0.6357 |

ED (−1) | −0.040151 | 0.218625 | −0.183653 | 0.8566 |

ED (−2)** | −0.399007 | 0.182815 | −2.182565 | 0.0443 |

ED (−3) | 0.063040 | 0.209277 | 0.301226 | 0.7671 |

Edu (−1) | 3.61E−06 | 2.90E−06 | 1.247008 | 0.2303 |

Edu (−2) | −2.98E−06 | 1.94E−06 | −1.539036 | 0.1433 |

Edu (−3) | 1.16E−06 | 2.05E−06 | 0.566739 | 0.5788 |

Constant | 0.084719 | 0.129104 | 0.656207 | 0.5210 |

R-squared | 0.749128 | Mean dependent var | 0.137969 | |

Adjusted R-squared | 0.513935 | S.D. dependent var | 0.050610 | |

S.E. of regression | 0.035285 | Akaike info criterion | −3.543876 | |

Sum squared resid | 0.019920 | Schwarz criterion | −2.811008 | |

Log likelihood | 72.70201 | Hannan−Quinn criter. | −3.300951 | |

F-statistic | 3.185165 | Durbin−Watson stat | 2.311027 | |

Prob (F-statistic) | 0.013774 |

Note: ** and* indicate statistically significant at 5% level and 10% level respectively.

it is significant at 5% level (

Other coefficients represent the short run impact of lagged periods. In the short run two periods lagged of external debt is significant at 5% and there is a short run impact on economic growth. This indicates that 1% change of external debt has a negative effect on economic growth by 0.39%. One period lagged of Gdp has a positive impact on economic growth at 10%. Domestic debt and educational expenditure are not statistically significant for short period of times. Therefore, it can be said that the impact of changes in domestic debt and education are not transitory even though it affects economic growth in the long run as a permanent effect.

Probability of observed R-squared is 0.58 and the result is insignificant (

Correlogram Q statistics indicate that all the probabilities are insignificant and no serial correlation or auto correlation in the residuals (

Probability of Jarque-Bera test statistics is 0.805 and result is insignificant (

This study examines the impact of public debt on economic growth in Sri Lanka over the period of 1977 to 2012. Public debt is classified as public domestic debt and public external debt and their impacts towards economic growth in long run. For this purpose, the study adapted Johansen test of cointegration to estimate the long run impact followed by Vector Error Correction Model. Finally Error Correction Term was used to capture the speed of adjustment.

In addressing research objective, the study finds domestic debt, external debt and educational expenditure significantly affect economic growth in long run. As the policy implication, the results of the study are providing strategic information for the government in making policy decisions with regard to borrowing

Heteroskedasticity test: Breusch-Pagan-Godfrey | |||
---|---|---|---|

F-statistic | 0.743784 | Prob. F (16, 15) | 0.7185 |

Obs*R-squared | 14.15649 | Prob. Chi-Square (16) | 0.5871 |

Scaled explained SS | 3.074459 | Prob. Chi-Square (16) | 0.9998 |

Autocorrelation | Partial correlation | AC | PAC | Q-Stat | Prob | |
---|---|---|---|---|---|---|

. *| . | | . *| . | | 1 | −0.170 | −0.170 | 1.0127 | 0.314 |

. *| . | | .**| . | | 2 | −0.179 | −0.214 | 2.1733 | 0.337 |

. | . | | . *| . | | 3 | 0.011 | −0.067 | 2.1782 | 0.536 |

. |* . | | . | . | | 4 | 0.074 | 0.027 | 2.3931 | 0.664 |

. *| . | | . *| . | | 5 | −0.151 | −0.151 | 3.3169 | 0.651 |

. *| . | | . *| . | | 6 | −0.134 | −0.195 | 4.0735 | 0.667 |

. |* . | | . | . | | 7 | 0.188 | 0.071 | 5.6136 | 0.586 |

. | . | | . | . | | 8 | 0.045 | 0.032 | 5.7048 | 0.680 |

.**| . | | .**| . | | 9 | −0.340 | −0.320 | 11.184 | 0.263 |

. | . | | . *| . | | 10 | 0.031 | −0.123 | 11.231 | 0.340 |

. | . | | . *| . | | 11 | 0.042 | −0.167 | 11.324 | 0.417 |

. | . | | . | . | | 12 | 0.030 | −0.055 | 11.372 | 0.497 |

.**| . | | .**| . | | 13 | −0.210 | −0.270 | 13.896 | 0.381 |

. |* . | | . *| . | | 14 | 0.148 | −0.123 | 15.226 | 0.363 |

. | . | | .**| . | | 15 | 0.055 | −0.215 | 15.420 | 0.422 |

. |* . | | . |* . | | 16 | 0.098 | 0.085 | 16.070 | 0.448 |

debt. The government of Sri Lanka needs to pay attention to the financing budget deficit by properly maintaining the threshold for debt as a proportion of GDP. Foreign and domestic borrowings will influence on economic growth in the long run but threshold for debt is a factor to be evaluated in financial decisions. Public debt and economic growth gained must be paid attention over the time in budgetary decision making. This is vital to lead an appropriate “Debt-to-GDP” ratio; otherwise debt will drag economic growth negatively.

From the policy prospective, it is recommended to refrain from adverse effects of external debt and public domestic debt on economic growth. The negative effect on economic growth is more severe. As foreign exchange reserves are very important for paying back external debt; the revenue generated from export can help Sri Lanka to overcome the issue. Therefore, export diversification strategies and value added export must be strongly promoted. Also good relations with other countries to reduce trade barriers must be encouraged. At the same time, value of Sri Lanka Rupee in foreign exchange market should be protected because currency devaluation strategy to enhance exports has not been helpful against competitive devaluations. As external and domestic debt have negative impact on GDP, policy makers’ heavy reliance of debt to finance fiscal deficit must be discouraged because there is dire need to stimulate revenue.

Munasinghe, M.A.A., Attapattu, A.M.C.P. and Padmasiri, H.M.N. (2018) Long Run Association between Public Debt and Economic Growth in Sri Lanka. Modern Economy, 9, 775-789. https://doi.org/10.4236/me.2018.94051