IS THE TURKISH CURRENT ACCOUNT DEFICIT SUSTAINABLE? AN ECONOMETRİC ANALYSIS

The Prof. Dr. Muhammed Akdiş

Deparment of economics the University of Pamukkale

The assistant of professor Osman PEKER

Departmen of economics the University of Adnan Menderes

The assistant of professor Şakir Görmüş

Departmen of economics the University of Adnan Menderes

Abstract

The main objective of this study is to examine the sustainability of current acount deficit in Turkey over the 1992:1 - 2005:12 periods by utilizing co-integration and error correction methods. This study based on Husted’s (1992) model, which taking account some key variables such as exports of good and services plus current transfers, import of goods and services. In this study, we have tested long run relationship between several measurement of export and import utilizing Augmented Dickey Fuller (ADF) Unit Root test and standart Cointegration Regression Durbin Watson (CRDW) test. The results from ADF and CDRW tests showed that export and import are cointegrated and seems to be have a stable long-run relationship. In other words, Turkish current account deficit are sustainable. Also, we investigated short term deviations from long term trend by utilizing Error Correction Model. The results from Error Correction Model showed that short run changes in export have significant negative effects on imports and there is about 0.16 dısperancy between the actual and the long-run equilibrium. Therefore, deviation from short-term import can not be corrected.

Key Words: Current Account Deficit, Cointegration, Error Correction Model

I. Introduction

Literature explaines several causes of current account deficit such as: expansion of the fiscal deficit, decline in the private saving, decrease in productivity growth, overvalued exchange rate and trade deficit. Also, there are several criterias to measure the sustainability of current accout deficit such as: CA deficit to GDP ratio, import to GDP ratio, export to GDP ratio, change in reserves, change in capital flows and trade deficit to GDP ratio.

Both currency crises literature and 1994 and 2001 experiences teached us that the growing current account deficit was a mojor cause of currency crises. Also, data showed that trade deficit is the largest component of the current account deficit. A current account deficit of 4% (of GDP) in 1993 is accompanied by a trade deficit of 8% (of GDP) and a current account deficit of 5% (of GDP) in 2000 is accompanied by a trade deficit of 11% (of GDP), which end up with currency crises. In 2004, current account deficit and trade deficit reached 5% (of GDP) and 15% (of GDP), respectively, which is considered a warning level about sustainability of current account deficit. Therefore, current account deficit have become a major concern in Turkey during the last several years.

In this study, we will test long run relationship between several measurement of export and import. If there is a long run relationship between two, we can conclude that current account deficit is sustainable. Also, we will investigate short term deviations from long term trend by utilizing error correction model.

In the first part of this study, we will discuss definition, sustainability and criteria of current account. Second part, econometric model will be explained. Third part, the sustainability of current account deficit will be tested utilizing several econometric model. Final part, we explain the results.

II. Current Account And Criteria of Current the Account Deficit

Current account shows economic relationship between countries which inclued export-import, net factor’s payment and net transfers. If trade balance can not be compansated with net factor’s payment and net transfers then current account deficit occurs.

In the literature, current account can be defined in several way. First, current account can be defined as differences between saving and investment for overall economy. If investment is higher than saving then country faces current account deficit and to reduce it saving (investment) has to increase (decrease). Second, current account can be defined as difference between aggregate output and aggregate expenditure. If aggregate expenditure is higher than aggregate output then country faces current account deficit and to reduce it aggregate output (aggregate expenditure) has to increase (decrease). Third, current account can be defined as export plus net factor’s payment and transfers minus import. If import is higher than export plus net factor’s payment and transfers then country faces current account deficit and to reduce it export plus net factor’s payment and transfers (import) has to increase (decrease).

Therefore, we can say that decline in the private saving rate, decrease in growth rate, decrease (increase) in export (import) and increase in domestic interest rate are several causes of current account deficit .

There are several criterias to measure the sustainability of current accout deficit such as: budget deficit to GDP ratio, import to GDP ratio, export to GDP ratio, change in reserves, change in capital flows and trade deficit to GDP ratio. In general, if current account to GDP ratio is higher than %5, then sustainability of current acount deficit is questionble. Increase in export to GDP ratio, capital inflow, economic growth, reserves and saving will make current account deficit more attaniable. However, increase in import to GDP ratio, trade deficit to GDP ratio, investment and budget deficit to GDP ratio will make current account deficit less attaniable. Also, increase in political instability can make current account deficit less attaniable.

III. Model and Data

Husted (1992) presents a principal statistical analysis followed in this paper that implies a long-run relationship between export and import. The individual current-period budget constraint is:

C₀ = Y₀ + B₀ – I₀ – (1+r₀) B_-1 (1)

Where C₀ is current consumption; Y₀ is output; I₀ is the one period world interest rate;. B₀ is the size of international borrowing; and (1+r₀)B_-1 is the historically given initial debt of the representative agent, corresponding to the country’s external debt. In that case Husted (1992) makes several assumptions in order to derive a testable model which is given by the following regression:

X_t = a + b^* MM_t + e_t (2)

Where X is exports of goods and services, and MM is imports of goods and services plus net unilateral transfers. In order for the economy to satisfy its intertemporal budget constraint, b should be equal to 1 and e_t should be stationary. Thus if X ve MM are nonstationary, then under the null, they are cointegrated.

The data used in this study are monthly, adjusted flows of aggregate Turkey exports of goods and services plus current transfers, EX, and Tukey imports of goods and services IM, over the period 1992-2005. All variables dominated in terms of dollar. Dollar valuees of GNP were used to create export/GNP and import/GNP in (EXY and IMY). Data EX, IM and GNP were obtained from hhtp://www.tcmb.gov.tr. In addition, three crisis dummy variables are deployed in order to show the effect of crisis on the Turkish economy[1].

4 Methodology

Testing for the existence of cointegration among economic variables is an increasingly popular approach to study economic interrelationships. The consept of cointegration applies to a wide variety of economic models[2]. Enders (1996: 151), any equilibrium relationship among a set nonstationary variables implies that their stochastic trends must be linked. After all, the equilibrium relationship means that the variables cannot move independently of each other. This linkage among the stochastic trend necessitates that the variables have to be cointegrated. Since the trend of cointegrated variables are linked, the dynamic paths of such variables must bear some relation to the current deviation from the equilibrium relationship.

A principal feature of cointegrated variables is that their time paths are influenced by the extent of any deviation from long-run equilibrium. Thus, the sort-run dynamics must be influenced by the deviation from the long-run relationship. The dynamic model implied by this discussion is one of error correction. In a error-correction model, the short-term dynamics of the variables in the system are influenced by the deviation from equilibrium (Enders, 1995: 365-66).

Cointegration and error-correction modeling involves four steps. First, determine the orders of integration for each of the variables under consideration which difference of each series successively emerge to the stationary series. Second, estimate cointegration regressions with ordinary least square using variables with the same order of integration. Third, test for stationary residuals of the cointegration regressions. Finally, construct the error-correction models.

To test for cointegration between export and import measures, we follow the Engle-Granger (1987) methodology and use standart CRDW (Cointegration Regression Durbin-Watson) and ADF (Augmented Dickey-Fuller) tests. To explain the Engle-Granger testsing procedure, suppose that two variables say y_t and x_t are believed to be integrated of order one and you want to determine whether tere exists an equilibrium relationship between the two. Eangle and Granger (1987) proposed a straightforward test to detremine whether two I(1) variables are CI(1,1). By definition, cointegration necessitates that the variables have to be integrated of the same order. Thus, the first step in the analysis is to pretest each variable to determine its order of integration.

To infer the number of unit roots in each of the variables Dickey-Fuller test can be used. Dickey and Fuller (1979, 1981) devised a procedure to formally test for the presence of a unit root. If the variables are integrated of the same orders, it is possible to conclude they are cointegrated. The next step is to estimate the long-run equilibrium relationship in the form:

(3)

In order to determine if the variables are actually cointegrated, we can denote the residual sequence from (3) by ê_t. Thus, ê_t is the series of the estimated residuals of the long-run relationship. If these deviations from long-run equilibrium are found to be stationary, the y_t and x_t squences are cointegrated of order (3). It would be convenient if we could perform a Dickey-Fuller (DF) test on these residuals to determine their order of integration. Consider the autoregression of the residual:

(4)

If we cannot reject the null hypothesis a₁ =0, we can conclude that the residual series contains a unit root. Hence, we conclude that the y_t and x_t sequences are not cointegrated. İf the residuals of (.2) do not appear to be white-noise, an augmented Dickey-Fuller (ADF) test can be used instead of (4).

(5)

In the third step, if the variables are cointegrated, using the saved residuals from the estimation of the long-run equilibrium relationship, we can estimate the error correcting model as:

(6)

(7)

Where α_y and α_x are the speed of adjustment coefficients which they have important implications for the dynamics of the system; ε_yt and ε_xt are a white noise disturbances. Equations (6) and (7) constitute VAR in differences. In the finally step, there are several prodecures that can help determine whether the estimated error correction model is appropriate: if the variables are cointegrated the speed of coefficients’ adjusment must be significantly different from zero. After all, if both α_y and α_x are zero, there is no error correction.

4. Empirical Results

4.1.Unit Root Test

We have performed ADF unit root tests in levels and first differences for the series used in this study. These tests include a time trend and optimum lags of the variables and indicate that EXL and IML series are I(1), except EXYL and IMYL series. For the levels of the EXL and IML series, none rejects the null hypothesis of nonstationarity at the 5 percent. After first differencing EXL and IML series, reject the null hypothesis of nonstationarity at the 5 percent levels. According to these tests, EXL and IML variables are nonstationary in levels, EXYL and IMYL variables appears to be stationary in levels. Table 1 reports tests for ADF unit roots. Optimum lagged determined using AIC test.

Table 1: Unit Root Tests

(Level)
Variales	Test		ADF Statistic	MacKinnon Critical Value (%5)	Optimal Lagged
EXL	Trend+intercept)		-1.115824	-3.4379	3
IML	“		-2.498846	-3.4381	4
EXYL	“		-4.951893	-3.4374	1
IMYL	“		-4.181432	-3.4374	1
(First differences)
Variables	Test	ADF Statistic		MacKinnon Critical Value (%5)	Optimal Lagged
DEXL	none	-4.740855		-1.9417	3
DIML	“	-5.124993		-1.9415	3

4.2. Cointegration Tests

Our result for the Eangle-Granger cointegration tests are presented in Table 2. We report two results from the CRDW and ADF. For all proxies of the current account deficit, we can reject the null hypothesis of no cointegration. The other words, the rejection of the null hypothesis implies that the residual sequence is stationary. The Eangle-Granger 1%, 5%, 10% critical values of t statistic in the regression are 3.77, 3.17, and 3.03, respectively. Since in absolute terms the estimated t statistic value of 4.154 and 4.450 exceeds any of these critical values, the conclusion would be that the estimated e_t is stationary (i.e., it does not have a unit root), and, therefore, EXL and IML are being individually nonstationary and cointegrated.

CRDW test is an alternative and quicker method to find out whether EXL and IML are cointegrated. Critical values of CRDW test were first provided by Sargan and Bhargava (1983). In CRDW, we use the Durbin-Watson d value obtained from the cointegrating regression, such as d=0.374, d= 0.431 given in Table 2. But now the null hypothesis is that d=0 rather than the standart d=2. Based on 10.000 simulations formed from 100 observations each, the 1%, 5%, and 10% critical values to test the hypothesis that the true d=0 are 0.511, 0.386, and 0.322, respectively. Thus, if the computed d value is smaller than, say, 0.322, we reject the hypothesis of cointegration at the 10 level. In our example, the d value of 0.374 and 0.431 are above the critical level, which would suggest that EXL and IML are cointegrated. The result is consistent with one reached on the basis of ADF test.

Tablo 2: Cointegration Tests

Cointegration Regressions: 1992:01 to 2005:12
Coefficients of variable
	Constant	EXL	IML	D1	D2	D3	R^2	CRDW	ADF
EXL	-0.728	-	1.019	0.130	-0.300	0.044	0.88	0.374	-4.154
IML	2.874	0.869	-	-0.205	0.296	-0.034	0.88	0.431	-4.450

In summary, based on both the ADF and CRDW tests, our conclusion is that EXL and IML are cointegrated. Although they individually exhibit random walk, it seems to be there are a stable long- run relationship between the EXL and IML. Hence, Turkey current account deficits can sustainable.

4.3. Error Correction Models

We just showed that EXL and IML are cointegrated, that is, there is a long-term equilibrium relation between the two. Of course, in the short run there may be disequilibrium. Therefore, one can treat the error term in (6) and (7) as the equilibrium error. We can conclude that the series are cointegrated of order (1). Fortunately, both of the equilibrium regressions yield same conclusion. Therefore, we can apply the normalızation using any of the equations residuals. Estimating the error correction model which it is the first-order sistem is shown with t-statistics in parentheses.

In response to a positive discrepancy in e_yt-1 in equation (8), EXL tend to decrease while IML tend to increase. In response to a positive discrepancy in e_yt-1 in equation (9), both IML and EXL tend to decrease. The error-correction term, however, is significant only in equation (9). The error-orrection term, however, is significant only in (9). This result show that short-run changes in EXL have significant negative effects on IML and there is about 0.16 of disprepancy between the actual and the long-run equilibrium. Therefore, deviation from short-term import can not be corrected.

(8)

(9)

5. Conclusion

In thıs study, we examined the sustainability of current acount deficit in Turkey over the 1992:1- 2005:12 periods by utilizing co-integration and error correction methods. We have tested long run relationship between several measurement of export and import utilizing Augmented Dickey Fuller (ADF) Unit Root test and standart Cointegration Regression Durbin Watson (CRDW) test.

The results from ADF and CDRW tests showed that export and import are cointegrated and seems to be have a stable long-run relationship. In other words, Turkish current account deficit are sustainable. Also, we investigated short term deviations from long term trend by utilizing Error Correction Model.

The results from Error Correction Model showed that short run changes in export have significant negative effects on imports and there is about 0.16 disperancy between the actual and the long-run equilibrium. Therefore, deviation from short-term import can not be corrected.

Bibliography

ENDERS, W (1995), Applies Econometric Time Series, John Willey & Song, Inc.

ENDERS, W (1996), Rats Handbook for Econometric Time Series, John Willey & Song, Inc.

DICKEY, and W. A. FULLER (1979), “Distribution of the Estimates for Autoregressive Time Series with a Unit Root”, Journal of the American Statistical Association, 74, 427-431.

DICKEY, D. and W. A. FULLER (1981), “Likelihood Ratio Statistics for Autoregressive Time Series with a Unit Root”, Econometrica, 49, 1057-1072.

ENGLE, R. F. and B. S. YOO (1987), “Forecasting and Testing in Co-integrated Systems”, Journal of Econometrics, 35, 143-159.

ENGLE, R. F. and C. W. GRANGER (1987), “Co-integration and Error Correction: Representation, Estimation, and Testing”, Econometrica, 55, 251-276.

HUSTED, S. (1992), “The Emerging U.S. Current Account Deficit in the 1980s: A Cointegration Analysis”, The Review of Economics and Statics, 74, 159-166.

SARGAN J. D. and A. S. BHARGAVA (1983), “Testing Residuals from Least Squares Regression for Being Generated by the Gaussian Random Walk”, Econometrica, 51, 153-174.

FERRITTI, R. (1996a), “Sustainability of Persistent Current Account Deficits”, NBER Working Paper: 5467.

FERITTI, R. (1996b), “Current Account Sustainability: Selected East Asian and Latin American Experiences”, NBER Working Paper: 5791.

SEBASTİAN, E. (2005), “Is the U.S. Current Account Deficit Sustainable? And If Not, How Costly Is Adjustment Likely To Be?”, NBER Working Paper: 1154.

RUBINI, N. and P. WATCHEL (1998). Current Account Sustainability in Transition Economies, NBER Working Paper Series:6468.

---

§ Osman Peker, Adnan Menderes Üniversitesi Nazilli İİBF, Sümer Kampüsü, Nazilli/Aydın.

ottopeker@yahoo.com, 02563151972, 05055073864.

[1] D1, D2 and D3 are April 1994, November 2000 and February 2001 crises, respectively.

[2] See Engle and Yoo (1987) and Engle and Granger (1987) for further details.