Liberalization, FDI, and growth in developing countries: a panel cointegration, approach.

Basu, Parantap ; Chakraborty, Chandana ; Reagle, Derrick 等

I. INTRODUCTION

Although there is an enormous body of literature dealing with the effect of foreign direct investment (FDI) on both long-run and short-run growth, there is no common consensus regarding the permanent as well as transitional effects of FDI on growth. Elias (1990) examines the Latin American experience and argues that removal of international credit and liquidity constraints has encouraged FDI in the 1980s in most of these countries. De Mello (1996) argues that the direction of causality depends on the recipient country's trade regime. In a recent article Nair-Reichert and Weinhold (2001) undertake panel causality tests and persuasively argue that the effect of FDI on growth is highly heterogenous across countries and this heterogeneity is more pronounced for more open economies.

Despite several empirical studies investigating the effect of FDI on growth, there is no serious effort to understand the two-way link between FDI and gross domestic product (GDP). The two-way link between FDI and GDP stems from the fact that increased FDI promotes growth in host countries, whereas brighter growth prospects in the host countries attract an increased flow of FDI. There are two disjoint branches now in the existing literature. The literature, including Nair-Reichert and Weinhold (2001) and other researchers, mainly focuses on the causality running from FDI to GDP. Very little attention has been given to examine the reverse causality. Taylor and Sarno (1999) look at the determinants of large portfolio flows from the United States to Latin America and Asian countries. However, they only look at the determinants of portfolio flows from the United States and do not focus on the two-way linkage between FDI and growth in a unified panel framework.

In this article we investigate the long-run and short-run two-way relationship between FDI and GDP using panel data. We investigate the two-way linkage between FDI and GDP by exploiting a cointegrated relationship between FDI and GDP. Cointegration implies that there is an error correction mechanism capturing the two-way linkage. In addition, we empirically examine whether the dynamics of the FDI and GDP relationship change when countries go through a process of liberalization. We take up the issue of heterogeneous relationships between FDI and growth as mentioned in Nair-Reichert and Weinhold (2001) further and investigate whether the short-run and long-run two-way linkages between FDI and GDP vary across countries that differ in terms of the degree of openness.

Using a panel of 23 countries and applying the latest panel cointegration technique of Pedroni (1995; 1997) we find that there is indeed a cointegrated relationship between FDI and GDP. To examine the short-run dynamics of FDI and GDP, we classify countries in closed and open economies using various criteria for openness. We find that for open economies the causality is bidirectional between FDI and GDP. The two-way linkage between FDI and GDP for open economies is evident from the dynamics of their relationship in both the short and the long run. On the other hand, for closed economies, the direction of causality is quite different. Although causality is bidirectional in the short run, causality runs mainly from growth to FDI in the long run. Permanent shifts in foreign capital are not realized in closed economies until after these economies exhibit considerable growth.

The following section explains the theoretical motivation. Section III explains the data. Section IV reports the empirical results. Section V concludes.

II. TWO-WAY LINKAGE BETWEEN FDI AND GROWTH: MOTIVATION

The two-way linkage between FDI and growth has long-run and short-run dimensions. The technology transfer brought by FDI could have permanent effects on the total factor productivity of the host country through a combined impact of human capital augmentation and technological change as discussed in Ramirez (2000), Barrell and Pain (1997), and Balasubramanyam et al. (1996). The reverse long run effect of GDP on FDI could arise through bilateral trade. A number of works, including Ajami and Barniv (1984), Ray (1989), and Grosse and Trevino (1996), argue that growth leads to increased bilateral trade. A larger bilateral trade would help the home country receive more information about the investment climate in the host countries. Through this channel, the home country creates permanent FDI opportunity in the host country as found in Liu et al. (1997). Integrating these two branches of literature dealing with technology and trade, we hypothesize a long-run steady-state relationship between FDI and the level of GDP for each country. This long-run relationship may vary across countries, but a panel cointegration can suitably capture this heterogeneity in the long-run relationship.

The bidirectional short-run causality between FDI and GDP could arise in the following way: An infusion of FDI, while bringing the economy to a higher long-run growth path, raises growth in the short run as well. The growth of the economy thus increases in the short run while the economy traverses along its transition path. (1) The reverse causality could arise from various "pull" factors associated with home country's opportunity and risk as in Taylor and Sarno (1999). In the short-run context, growth of national income in host countries could create a positive income effect. This could generate a higher aggregate demand in the host countries, thus attracting FDI in the short run, as in Grosse and Trevino (1996).

Causality between FDI and GDP is supported theoretically in many different forms: from FDI to GDP, GDP to FDI, permanent long-run movements, and transitory short-run changes. In the empirical sections that follow, we will examine the evidence for each of these effects using panel cointegration techniques.

III. DATA

Our panel includes 23 countries, and the study period spans from 1978 through 1996. A regional breakdown of these 23 countries is given in Table 1. As is apparent from the table, these countries provide a fair representation of all the major developing areas in the world, that is, Asia, Africa, Latin America, and Eastern Europe. Though a complete representation of the developing world would necessitate inclusion of many other countries, availability of data on relevant variables limits our choice to 23 countries listed in the table. The significance of these 23 countries, however, is evident from their share of FDI flows and the record of their moderate to strong growth performance over the study period. The available data set suggests that the country groupings listed in the table are major recipients of FDI flows in their respective regions. (2) Also, as reflected by country performance indicators, all these countries recorded real positive growth in GDP per capita over the period l990-99. (3) The choice of the time period for our sample data set is explained by the following two factors. First, before the 1970s, FDI flows to developing countries were dominated by flows of funds between multinational firms attempting to jump tariff barriers. Transportation and telecommunications breakthroughs coupled with increased efforts toward trade liberalization by developing countries have changed the strategies of multinational corporations and caused a noticeable acceleration of FDI flows since the late 1970s. Second, 1996 was the latest period for which data on relevant variables were available for all the countries included in the sample.

The annual series for net FDI flows and GDP for each of the selected countries and for each of the referred years have been collected directly from the combined sources of Global Development Finance and World Development Indicators published by the World Bank in 1998. Though the World Bank reports GDP figures in constant 1987 dollars, values for net FDI flows are reported in current U.S. dollars. We converted net FDI flows for each of the countries to real values by deflating them by the respective GDP deflator for each country using 1987 as the base year. The measure for degree of openness utilized in this study is an index of trade (exports plus imports) share in GDP. The annual average indices of openness were constructed with trade share data collected also from the World Development Indicators database.

IV. EMPIRICAL RESULTS

The empirical results are reported in three steps. First, we establish that both GDP and FDI have unit roots. In the second step, we look for cointegration using residual-based tests in the panel developed by Pedroni (1995; 1997) allowing for fixed and time effects. Finally, we look for short-run and long-run causality using Granger causality tests and weak exogeneity tests.

Unit Root

It is important to note that even though all 23 countries are being pooled in the unit root and cointegration tests, the direction and magnitude of the parameters are not constrained to be equal across countries, i (from i= 1 to 23). Furthermore, because a panel is being used we are able to control for time fixed effects common to countries for time t (t = 1978-1996). The flexibility of the specification is a major benefit of using panel time-series techniques, because they would not be possible without the increased degrees of freedom.

We test for the null hypothesis of nonstationarity versus the alternative that the variable is stationary. Using the unit root test of Im et al. (1995) the test is based on the augmented Dickey-Fuller (ADF) statistic for each country.

The top two rows in Table 2 report the results of the panel unit root statistic on the fdi and gdp series in levels. The results are reported with and without a trend, and with and without controlling for time fixed effects with a maximum of two lags for the ADF tests. (4) The results consistently fail to reject the null hypothesis of nonstationarity. The tests were also run allowing for up to five lags without a change in significance or magnitude, but as increased lags reduce the time horizon of the results, all reported results allow for a maximum of two lags of the annual data.

The third and fourth rows of Table 2 report the panel unit root statistics for the fdi and gdp series, respectively, in first differences. All specifications reject the null of nonstationarity at the 1% level, indicating that both series are I(1). Using these results, we proceed to test fdi and gdp for cointegration to determine if there is a long-run relationship to control for in the econometric specification.

Panel Cointegration

In the next step, we use a panel cointegration technique based on Pedroni (1995; 1997), which uses a residual-based ADF test. This technique allows for cointegrating vectors of differing magnitudes between countries, as well as country ([alpha]) and time ([delta]) fixed effects. The specific cointegrating relationship we estimate is:

(1) [fdi.sub.it] = [[alpha].sub.i] + [[beta].sub.i][gdp.sub.it] + [[epsilon].sub.it]

Equation (1) only estimates the long-run relationship. It does not make any assumptions about the direction of causality between fdi and gdp. The estimated residual, is the deviation from the modeled long-run relationship. If the series are cointegrated, [e.sub.it] will be a stationary (alternatively trend stationary) variable. As reported in the last row of Table 2, the panel unit root test rejects nonstationarity of [e.sub.it] under all specifications at the 1% significance level. We can therefore treat fdi and gdp as I(1) variables that share a long-run cointegrating relationship.

Casuality

As fdi and gdp are cointegrated, we estimate Granger causality using an error correction model to account for the long-run relationship using the two-step procedure from Engle and Granger (1987). The first step is the estimation of the long-run model from (1) to obtain the estimated residuals, [e.sub.it].

The second step is to estimate the Granger causality model with dynamic error correction:

(2) [DELTA][fdi.sub.it] = [a.sub.1j] + [[lambda].sub.1i][e.sub.it-1] + [summation over (k)][[phi].sub.11ik][DELTA][fdi.sub.i,t-k]

+ [summation over (k)][[phi].sub.12ik][DELTA][gdp.sub.i,t-k]+[u.sub.1it]

[DELTA][gdp.sub.it] = [a.sub.2j] + [[lambda].sub.2i][e.sub.it-1] + [summation over (k)[[phi].sub.21ik][DELTA][fdi.sub.i,t-k]

+ [summation over (k)][[phi].sub.22ik][DELTA][gdp.sub.i,t-k] + [u.sub.2it]

where k, the lag length, is chosen optimally for each country using a step-down procedure up to a maximum of two lags. (5)

There are several parameters of interest in the error correction model:

[[lambda].sub.1i]: Long-run effect of innovations in gdp on fdi

[[lambda].sub.2i]: Long-run effect of innovations in fdi on gdp

[[phi].sub.12i]: Short-run Granger causality from gdp to fdi

[[phi].sub.21i]: Short-run Granger causality from fdi to gdp

The long-run effect is in relation to the long-run relationship of the cointegrated process, so movements along this path can be considered permanent. The short-run effect is in terms of changes of the variables without reference to the long-run relationship and can be considered transitory. Because all variables enter the model in stationary form, a standard F test can be used to test the null hypothesis that none of the estimated country-specific parameters are significant as in Engle and Granger (1987) and Canning and Pedroni (1999).

Therefore, for short-run (Granger) causality, our hypothesis is formulated:

gdp to fdi [H.sub.0]: [[phi].sub.12ik] = 0 [for all] i, k

[H.sub.1]: [[phi].sub.12ik] [not equal to] 0 for at least 1, i, k

fdi to gdp [H.sub.0]: [[phi].sub.21ik] = 0 [for all] i, k

[H.sub.1]: [[phi].sub.21ik] [not equal to] 0 for at least 1 i, k

For long-run causality,

gdp to fdi [H.sub.0]: [[lambda].sub.1i] = 0 [for all] i

[H.sub.1]: [[lambda].sub.1i] [not equal to] 0 for at least 1 i

fdi to gdp [H.sub.0]: [[lambda].sub.2i] = 0 [for all] i

[H.sub.1]: [[lambda].sub.2i][not equal to] 0 for at least 1 i

We first test for the presence of the short-run Granger Causality. As reported in Table 3, the panel as a whole exhibits bidirectional causality, with gdp Granger causing fdi and fdi Granger causing gdp. Long-run causality also exhibits bi-directional causality as seen in Table 3. The null hypothesis of no long-run effect of gdp on fdi and the null hypothesis of no long-run effect of fdi on gdp can both be rejected at the 1% level of significance.

Effect of Liberalization

Although the causality tests for the whole sample period exhibit bidirectional causality, one gets a different picture when looking at subperiods before and after liberalization.

To this end, we reestimate (2) for two classes of economies labeled "closed" and "open." (6) The "open" subset is defined as being designated open by Sachs and Warner (1995) dates of trade liberalization. The complement of the "open" set is designated as "closed." Alternatively, we use the Bekaert and Harvey (2000) dates of financial liberalization to define "open." As can be seen in Table 4, there is a marked difference between the short- and long-run dynamics when comparing open and closed economies. For the closed economies, both the Sachs and Warner (1995) dates and the Bekaert and Harvey (2000) dates show long-run causality from gdp to fdi, but not in the other direction. Furthermore, only the Bekaert and Harvey (2000) classification shows short-run causality from fdi to gdp. Both classifications show short-run causality from gdp to fdi. (7)

Looking at the open subset yields a different picture. In both groups, the short-run and long-run causality tests show bidirectional causality in both groups. Although the full panel shows bidirectional causality, the only evidence of long-run causality from fdi to gdp is in open economies.

Thus far, we have shown a substantial difference in the dynamics of an economy from the time an economy is closed to when it is open. Another measure of openness, however, is the relative openness of a country compared to its counterparts. This measure adds to the analysis in two ways: (1) It is more objective than the dating of open and closed economies as it is based on one specific indicator of openness (thereby adding robustness), and (2) it recognizes that foreign funds may not view all open economies identically but move funds to open economies with fewer restrictions with more ease.

To construct the relative measure of openness, we calculate a world average of total trade as a proportion of GDP in developing countries, weighted by the volume of trade (measured by exports and imports) per country. We then divide the entire sample as above with "open" being defined as the subset with trade proportion higher than the world average. We use a world average rather than our sample mean to avoid selection bias in the subsets.

Table 5 reports the results of the panel error-correction Granger causality tests for the subsets. Again, the open and closed subsets show a difference in short- and long-run dynamics. For the closed economies, shortrun Granger causality is bidirectional, but the long-run movement is again only from gdp to fdi. For relatively open economies, long-run movement is again bidirectional as in the cases where liberalization dates were used, and short-run Granger causality is fromfdi to gdp at the 10% level of significance.

V. CONCLUSION

It is well known from the existing literature that FDI is a major engine of growth in developing countries. What is less understood is the two-way relationship between FDI and growth. Our empirical results suggest that there is a long-run steady-state relationship between FDI and GDP for a cross-section of countries after allowing for country-specific effects.

Regarding the short-run and long-run dynamics of FDI and growth, we find that the long-run causality mainly runs from growth to FDI in relatively closed economies. For relatively open economies, a richer dynamics emerges. There is evidence of bidirectional causality in both the long run and the short run, showing that FDI can lead to permanent changes in growth along the long-run path, not just transitory short-run fluctuations that have no reference to the cointegrating relationship. Our results suggest that permanent foreign capital does not reach closed economies until after the countries have exhibited growth, showing that trade and financial restrictions do indeed impede the inflow of foreign funds.

TABLE 1

List of Selected Developing Countries

Developing Region Country Groupings

East Europe and Central Asia Hungary
 Poland
 Turkey
East Asia China
 South Korea
 Singapore
Southeast Asia Indonesia
 Malaysia
 Philippines
 Thailand
South Asia India
 Pakistan
 Sri Lanka
Latin America Argentina
 Brazil
 Chile
 Colombia
 Mexico
 Peru
 Venezuela
Africa Kenya
 Nigeria
 Ghana

TABLE 2

Tests of Null of Nonstationarity

 [H.sub.1]: Trend Stationary [H.sub.1]: Mean
 Stationary
Series Time Fixed Effects No Time Effects Time Fixed Effects

fdi 4.06 4.92 6.08
gdp 8.95 3.11 20.11
[DELTA]fdi -13.43 * -14.73 * -9.45 *
[DELTA]gdp -4.93 * -7.10 * -2.55 *
Cointegration -3.51 * -3.42 * -2.73 *

 [H.sub.1]: Mean
 Stationary
Series No Time Effects

fdi 7.91
gdp 14.82
[DELTA]fdi -15.04 *
[DELTA]gdp -5.30 *
Cointegration -3.84 *

* Significant at the 1% level.

TABLE 3

Panel Causality Tests

 Short-Run
[H.sub.0]: No Causality [F.sub.25,272] p-Value

gdp to fdi ([[phi].sub.12i]) 3.9234 * <0.0001
fdi to gdp ([[phi].sub.21i]) 2.4196 * 0.0003

 Long-Run
[H.sub.0]: No Causality [H.sub.0]: No Causality

gdp to fdi ([[phi].sub.12i]) gdp to fdi ([[lambda].sub.1i])
fdi to gdp ([[phi].sub.21i]) fdi to gdp ([[lambda].sub.2i])

 Long-Run
[H.sub.0]: No Causality [F.sub.23,272] p-Value

gdp to fdi ([[phi].sub.12i]) 4.7058 * <0.0001
fdi to gdp ([[phi].sub.21i]) 2.1099 * 0.0020

Note: n = 368.

* Significant at the 1% level.

TABLE 4

Panel Causality Tests (Liberalization Dates)

[H.sub.0]: No Causality Closed Open

Sachs and Warner
Short run [F.sub.16,92] p-value [F.sub.21,106]
 gdp to fdi ([[phi].sub.12i]) 2.7885 * 0.0011 2.2385 *
 fdi to gdp ([[phi].sub.21i]) 1.4153 0.1520 9.0957 *
Long run [F.sub.14,92] p-value [F.sub.16,106]
 gdp to fdi ([[lambda].sub.1i]) 3.6729 * <0.0001 6.0437 *
 fdi to gdp ([[lambda].sub.2i]) 0.7698 0.7149 7.1296 *

Bekaert and Harvey
Short run [F.sub.17,84] p-value [F.sub.15,25]
 gdp to fdi ([[phi].sub.12i]) 4.1374 * <0.0001 3.8924 *
 fdi to gdp ([[phi].sub.21i]) 2.3368 * 0.0057 4.6806 *
Long run [F.sub.15,84] p-value [F.sub.17,84]
 gdp tofdi ([[lambda].sub.1i]) 3.6631 * <0.0001 6.6092 *
 fdi to gdp ([[lambda].sub.2i]) 1.5265 0.1052 4.5255 *

[H.sub.0]: No Causality

Sachs and Warner
Short run p-value
 gdp to fdi ([[phi].sub.12i]) 0.0039
 fdi to gdp ([[phi].sub.21i]) <0.0001
Long run p-value
 gdp to fdi ([[lambda].sub.1i]) <0.0001
 fdi to gdp ([[lambda].sub.2i]) <0.0001

Bekaert and Harvey
Short run p-value
 gdp to fdi ([[phi].sub.12i]) 0.0013
 fdi to gdp ([[phi].sub.21i]) 0.0003
Long run p-value
 gdp tofdi ([[lambda].sub.1i]) <0.0001
 fdi to gdp ([[lambda].sub.2i]) 0.0004

Note: For Sachs and Warner dates n = 152 for closed countries, n = 180
for open countries. For Bekaert and Harvey dates n = 148 for closed
countries, n = 77 for open countries.

* Significant at the 1% level.

TABLE 5

Panel Causality Tests (Trade Indices)

Trade Level
[H.sub.0]: No Causality Closed Open

Short run [F.sub.16,150] p-value [F.sub.11,82]
 gdp tofdi ([[phi].sub.12i]) 3.8438 * <0.0001 0.4997
 fdi to gdp ([[phi].sub.21i]) 2.1926 * 0.0075 1.8227 * *

Long run [F.sub.14,150] p-value [F.sub.9,82]
 gdp to fdi ([[lambda].sub.1i]) 3.2261 * <0.0001 10.6587 *
 fdi to gdp ([[lambda].sub.2i]) 1.2231 0.2566 4.5497 *

Trade Level
[H.sub.0]: No Causality

Short run p-value
 gdp tofdi ([[phi].sub.12i]) 0.8982
 fdi to gdp ([[phi].sub.21i]) 0.0631

Long run p-value
 gdp to fdi ([[lambda].sub.1i]) <0.0001
 fdi to gdp ([[lambda].sub.2i]) <0.0001

Note: n = 144 for closed countries, n = 211 for open countries.

** and * significant at the 10% and 1% level, respectively.

(1.) The transitional dynamics could arise in a standard Lucas (1988) type of endogenous growth model with two types of capital. In our context, the aggregate production function may involve complementarity between domestic and foreign inputs. Growth of output could result either due to use of an expanded variety of intermediate inputs in an FDI-related sector, as discussed in Feenstra and Markusen (1994), or due to human capital augmentation resulting form transfer of knowledge, referred to in De Mello and Sinclair (1995).

(2.) The data source of Global Development Finance (its 1999 CD-ROM data set) and World Development indicators (its 1998 CD-ROM data set) together indicate that the selected country groupings explain more than 80% of FDI flows in their respective regions.

(3.) Country performance indicators are listed in The Global Competitiveness Report (World Economic Forum, 2000).

(4.) The lag length for each country is allowed to differ in all ADF tests and is chosen endogenously using the standard step-down procedure.

(5.) We alternatively estimate all regressions allowing up to five lags with no substantive change in significance. A maximum of two lags is reported due to the longer time horizon.

(6.) The cointegrating vector for both closed and open economies is assumed to be obeying the same long-run relationship.

(7.) Chakraborty and Basu (2002) found similar results in the context of India.

REFERENCES

Ajami, R. A., and R. Barniv. "Utilizing Economic Indicators in Explaining Foreign Investment in the U.S." Management International Review, 24(4), 1984, 16-26.

Balasubramanyam, V. N., M. Salishu, and D. Sapsford. "Foreign Direct Investment and Growth: New Hypotheses and Evidence." Discussion Paper Ec7-96, Dept. of Economics, Lancaster University, 1996.

Barrell, R., and N. Pain, "Foreign Direct Investment, Technological Change, and Economic Growth within Europe." Economic Journal, 107(1), 1997, 1770-85.

Bekaert, G., and C. R. Harvey. "Foreign Speculators and Emerging Equity Markets." Journal of Finance, 55, 2000, 565-613.

Canning, D., and P. Pedroni. "Infrastructure and Long-Run Economic Growth." Mimeo, Indiana University, 1999.

Chakraborty, C., and P. Basu. "Foreign Direct Investment and Growth in India: A Cointegration Approach." Applied Economics, 34, 2002, 1061-73.

De Mello, L. R. "Foreign Direct Investment, International Knowledge Transfers and Endogenous Growth: Time Series Evidence." Mimeo, Department of Economics, University of Kent, UK, 1996.

De Mello, L. R., and M. T. Sinclair. "Foreign Direct Investment, Joint Ventures and Endogenous Growth." Mimeo, Department of Economics, University of Kent, UK, 1995.

Elias, V. J. Sources of Growth: A Study of Seven Latin American Countries. San Francisco: CA ICS Press, 1990.

Engle, R. F., and C. W. J. Granger. "Co-integration and Error Correction: Representation, Estimation, and Testing." Econometrica, 55, 1987, 251-76.

Feenstra, R. C., and J. R. Markusen. "Accounting for Growth with New Inputs." International Economic Review, 35, 1994, 429-47.

Grosse, R., and L. J. Trevino. "Foregin Direct Investment in the United States: An Analysis by Country of Origin." Journal of International Business Studies, 27(1), 1996, 139-55.

Im, K. S., M. H. Pesaran, and Y. Shin. "Testing for Unit Roots in Heterogeneous Panels." Working Paper 9526, Department of Applied Economics, Cambridge University, Cambridge, 1995.

Liu, Xiaming, H. Song, Y. Wei, and P. Romilly. "Country Characteristics and Foreign Direct Investment in China: A Panel Data Analysis." Weltwirtschaftliches Archive, 133(2), 1997, 313-29.

Lucas, R. E. "On the Mechanics of Economic Development." Journal of Monetary Economics, 22, 1988, 3-42.

Nair-Reichert, U., and D. Weinhold. "Causality Tests for Cross Country Panels: A New Look at FDI and Economic Growth in Developing Countries." Oxford Bulletin of Economic and Statistics, 63, 2001, 153-71.

Pedroni, P. "Panel Cointegration: Asymptotic and Finite Sample Properties of Pooled Time Series Tests, with an Application to the PPP Hypothesis." Indiana University Working Papers in Economics No. 95-013, 1995.

-----. "Panel Cointegration: Asymptotic and Finite Sample Properties of Pooled Time Series Tests, with an Application to the PPP Hypothesis: New Results." Mimeo, Indiana University, 1997.

Ramirez, M. D. "Foreign Direct Investment in Mexico: A Cointegration Analysis." Journal of Development Studies, 37(1), 2000, 132-62.

Ray, E. J. "The Determinants of Foreign Direct Investment in the United States: 1979-1985," in Trade Policies for International Competitiveness, edited by R. Feenstra. Chicago: University of Chicago Press, 1989.

Sachs, J., and A. Warner. "Economic Reform and the Process of Global Integration." Brookings Papers on Economic Activity, 1, 1995, 1-113.

Taylor, M. P., and L. Sarno. "Capital Flows to Developing Countries: Long and Short-Term Determinants." World Bank Economic Review, 11(3), 1999, 451-70.

World Economic Forum. The Global Competitiveness Report 2000. New York: Oxford University Press, 2000.

RELATED ARTICLE: ABBREVIATIONS

ADF: Augmented Dickey-Fuller

FDI: Foreign Direct Investment

GDP: Gross Domestic Product

PARANTAP BASU, CHANDANA CHAKRABORTY, and DERRICK REAGLE *

* We wish to thank Peter Pedroni and participants of the 2001 ASSA National Conference for valuable comments and discussion.

Basu: Associate Professor, Department of Economics, Fordham University, Bronx, NY 10458. Phone 1-718-817-4061, Fax 1-718-817-3518, E-mail basu@fordham.edu

Chakraborty: Associate Professor, Department of Economics and Finance, Montclair State University, Upper Montclair, NJ 07043. Phone 1-973-655-4125, Fax 1-973-655-7629, E-mail chakrabortyc@mail.montclair.edu

Reagle: Assistant Professor, Department of Economics, Fordham University, Bronx, NY 10458. Phone 1-718-817-4048, Fax 1-718-817-3518, E-mail reagle@fordham.edu