Economic Growth, Foreign Direct Investment and Economic Openness An Empirical Study of the U.S.

Economic Growth, Foreign Direct Investment and Economic Openness: An Empirical Study of the U.S.

Jeffrey Breece

School of Finance, Renmin University of China,

Beijing, P. R. China

ABSTRACT: This paper analyzes the causal relationships between economic growth, foreign direct investment, and economic openness by relying on U.S. data from 1960 to 2010. These three factors have and continue to play a vital role in the development of the U.S, mandating an in depth analysis of their inter-workings. We establish a VAR framework to test for Granger Causality, in order to explore beyond mere correlation. We find a bilateral causal relationship exists between economic growth and economic openness as well as one from foreign direct investment to economic openness. We conclude with several policy implications that can be drawn from our findings.

1

I. Introduction The United States has the world’s largest economy, is the largest recipient of

foreign direct investment and one of the most economically open countries in the world, according to the Heritage Foundation’s Economic Freedom of the World Index. Many academic circles have noted that this concurrence can by no means be a coincidence and there must exist some underlying connections. These connections may hold the answer to how the U.S. became such a dominant power in such a short amount of time and how it can again enter into a period of prosperity.

Over the latter half of the 20th century and continuing until the present day, these three aspects have progressed in unison with one another, as can be seen in FIGURE 1. Especially evident in the data is the similar responses of economic growth (RGDP), foreign direct investment (FDI) and economic openness (measured as total trade to real gross domestic product) (EO) to recent economic shocks. All three suffered

synchronized periods of negative growth from the bust of the dot.com boom, as in the 3rd quarter of 2001 when Real GDP fell 0.3%, economic openness fell 0.033% and FDI fell 0.88%. In addition, the more recent recession caused by the sub-prime mortgage crisis is also visible through all three factors, particularly FDI which appears to be more sensitive than economic growth and economic openness.

FIGURE 1 543210-1-260657075EO8085IFDI9095000510RGDP

2

The tandem growth of these three factors suggests an obvious correlation over the past 50 years, but what about causation? How do the interdependent workings of

economic growth, foreign direct investment and economic openness all function together? And what role will this play in the future development of the U.S.?

This paper examines the previous questions through use of a Vector Auto-regression framework with hopes of understanding how these variables are related in the U.S. The remained of the paper is organized as follows. A review of relevant literature is put forth in Section 2. The data used in this study is presented in Section 3. Section 4 illustrates the models and discusses the empirical results. Section 5 concludes.

II. Literature Review

A large amount of research has been dedicated towards the respective fields of

economic growth, foreign direct investment and economic openness. However, the bulk of this work is concerned with only a pair-wise analysis of the factors, and even then, it is predominantly conducted on a series of developing countries. The specific case of the U.S. is normally not considered. Nevertheless, many recent findings are applicable to the U.S. over the past 50 years.

The relationship between the growth rate and inflow of foreign direct investment has been explored in a manner of different ways. Berthelemy and Demuger (2000), while investigating 24 Chinese provinces from 1985 to 1996, use endogenous simultaneous equations to reveal foreign direct investments’ large positive influence on economic growth. Similarly, Alfaro et al. (2000) arrive at a similar conclusion when they model a simple economy, finding that foreign direct investment plays an important role in economic growth, but is dependent on the maturity of financial markets. Other studies have been conducted that support a similar hypothesis; one finds that foreign direct investment can lead to higher growth rates, especially in countries that are more economically open (Nair-Reichert and Weinhold 2000). Another study using Granger Causality finds that foreign direct investment causes growth through knowledge transfers and the adoption of new technology (Hansen and Rand (2006). However, there have been papers which advocate the counter argument. Carkovic and Levine (2002), after

3

correcting for biases found in previous studies, conclude that foreign direct investment does not have a significant effect on growth.

Conversely, a paper put forth by Chowdhury and Mavrotas (2006) find that in the case of Chile, a developing country and in many ways comparable to the U.S. in the latter half of the 20th century, it is GDP that causes foreign direct investment. They also find evidence for bi-directional causality between the two variables for other countries.

While the relationship between economic growth and foreign direct investment has undergone much scrutiny recently, for decades academics have been investigating economic openness in a similar context. Harrison (1994), found a positive connection between growth and openness, using a varied assortment of openness measures, and finds that causality runs in both directions. This is conducive with the common economic theory that high growth provides an important momentum for the opening up of markets and falling growth rates facilitate a rise in protectionism. In conjunction with Harrison, Edwards (1997) and Sun et al. (2003) both report that open countries regularly have higher growth rates. However, they make no claim concerning causality for these factors.

The association between foreign direct investment and economic openness has received little attention until recently, but is becoming an increasingly important issue. This is especially true in the times following the recent financial crisis and the developing trade relations with China, where regulations and policies concerning economic openness are on the forefront of political agendas. Two studies, conducted by Quazi (2007) and Najarzadeh and Shahri (2008) both find that indices of economic freedom, most notably the Heritage Foundation’s Economic Freedom of the World Index, are a significant and robust determinant of foreign direct investment. Similarly, Bengoa and Sanchez-Robles (2003) determine that economic freedom has a positive causation on the level of inflowing foreign direct investment.

The goal of this paper is to build on previous literature by enlarging the

investigation of economic growth, foreign direct investment and economic freedom from bi-lateral to simultaneous. This is done in with the ambition of reinforcing or discounting past studies on the subject. To do this we will use the data and empirical methods mentioned below.

4

III. Data The data used in this study is quarterly data from the U.S. during 1960Q1 to

2010Q3 and is taken from the Federal Reserve Bank of St. Louis. We take economic growth as real gross domestic product (RGDP) in billions of chained 2005 dollars. Foreign direct investment (IFDIAV2) is a two period moving average of the net inflow of foreign direct investment and is also in billions of chained 2005 dollars. A moving average is used for foreign direct investment in order so smooth out the data. Economic openness (EO) is the sum of net imports and exports divided by the real gross domestic product to get a proportion of the percentage of trade to GDP. All data is seasonally adjusted and is summarized in TABLE 1.

TABLE 1

Variables

RGDP IFDIAV2 EO

Observations

202 202 202

Mean 744.969 87.502 0.156

S.D. 3241.007 132.090 0.0683

Minimum 2802.616 -70.132 0.071

Maximum 13363.49 729.494 0.292

We then conduct an Augmented-Dickey Fuller Test to test if RGDP, IFDI and EO

are all stationary series with the null hypothesis that the variables have a unit root and are stationary. The results are displayed in TABLE 2.

TABLE 2

Variables RGDP IFDIAV2 EO

ADF-statistic 0.883 -1.230 1.149

10% -2.574 -2.575 -2.574

5% -2.876 -2.876 -2.876

1% -3.463 -3.464 -3.46

P-value 0.995 0.6612 0.998

Judgment Accept Null Accept Null Accept Null

From TABLE 2 we can conclude that RGDP, IFDIAV2 and EO are all not

stationary. Therefore, we conduct the same Augmented-Dickey Fuller Test on their growth rates (first order conditions), which are RGDPGR, IFDIAV2GR and EOGR respectively. A graph of the first order conditions can be seen in FIGURE 2 and the outcome of the ADF Test are in TABLE 3. 5

FIGURE 2 1086420-2-4-6-860657075EOGR80859095000510FDIGRRGDPGR

TABLE 3

Variables RGDPGR IFDIAV2GR EOGR

ADF-statistic -6.797 -8.542 -20.301

10% -2.57 -2.575 -2.574

5% -2.876 -2.876 -2.876

1% -3.463 -3.464 -3.463

P-value 0.000 0.000 0.000 Judgment Reject Null Reject Null Reject Null

We can see from FIGURE 2 and TABLE 3 that RGDPGR, IFDIAV2GR and

EOGR all have p-values of 0.000, meaning that they are can reject the null hypothesis with absolute confidence. Now that the variables are stationary we can use them in our models.

IV. Empirical Methodology A.) Best Lag Order Selection

In order to generate a VAR model that will provide reliable results we need to choose the lag order that ensures that the error terms in the model will satisfy the vector white noise process. To determine the lag order we give a maximum lag order of 8 (2 years) and choose the lag order indicated by the majority of criterions. The results can be seen in TABLE 4.

6

TABLE 4

Lag LR FPE 0 NA 1.386131 1 95.45931 0.918257 2 33.90770 0.840099 3 31.86007 0.775004 4 14.78763 0.783925 5 7.745328 0.824148 6 22.24191 0.796767 7 11.43699 0.818940 8 33.02398* 0.739599* * indicates lag order selected by the criterion FPE: Final prediction error AIC: Akaike information criterion SIC: Schwarz information criterion HQ: Hannan-Quinn information criterion

AIC

8.840147 8.428336 8.339301 8.258466 8.269577 8.319082 8.284519 8.310901 8.207593*

SC

8.890863 8.631197* 8.694308 8.765620 8.928877 9.130528 9.248112 9.426639 9.475478

HQ

8.860685 8.510488 8.483067 8.463847* 8.536572 8.647692 8.674743 8.762739 8.721046

LR: sequential modified LR test statistic (each test at 5% level)

In TABLE 4 we can see that three out of the five criterions specify a lag order of

8, which is what we will use when constructing our VAR model.

B.) Johansen Co-integration Test

Now that we have ensured that the variables are stationary and determined the lag order, we are able to examine the existence of co-integration between them. We are interested in co-integration because it is an indicator of a common stochastic drift or long-term fluctuation between the variables. Since there are three variables we choose the Johansen Test as the Engle-Granger Test can only test for co-integration between two variables. The null hypothesis is that there is no con-integration between the three equations. The results are displayed in TABLE 5.

7

TABLE 5

Unrestricted Co-integration Rank Test (Trace)

Hypothesized No. of CE(s) Eigen value None * 0.211581 At most 1 * 0.117529 At most 2* 0.063304

**MacKinnon-Haug-Michelis (1999) p-values

Trace Statistic 45.64340 24.00569 12.55608

0.05 Critical Value 22.29962 15.89210 9.164546

Prob.** 0.0000 0.0021 0.0110

From TABLE 5 we discover that three co-integrating equations exist that are significant at the 5% level. This means that there are three long term relationships between the three variables. In order to determine the exact relationships we must conduct a Granger-Causality Test.

C.) VAR model and Granger Causality

Before we can perform a Granger-Causality Test we first need to establish our

VAR model that observes the relationships between GDPGR, IFDIAV2GR and EOGR. We use the VAR model framework because it allows for the simultaneous use of multiple equations in order to regress all the endogenous variables on their lags to estimate the relationship between all the endogenous variables. Below is the implicit representation of the VAR (8) model used.

RGDPGRc1a11a12a13RGDPGRa11a12a13RGDPGRe1ttt1t8IFDIAV2GRcaaaIFDIAV2GRaaaIFDIAV2GRett1t82122232t2212223c3EOGRta31a32a33EOGRt1a31a32a33EOGRt8e3t

8

We now conduct the Granger Causality Test to determine if the each variable

should be incorporated into the VAR model. The results are displayed in TABLE 6.

TABLE 6

Null Hypothesis

IFDIAV2GR does not Granger Cause EOGR EOGR does not Granger Cause IFDIAV2GR RGDPGR does not Granger Cause EOGR EOGR does not Granger Cause RGDPGR RGDPGR does not Granger Cause IFDIAV2GR

F-Statistic 2.57542 0.33740 4.80307 4.39082 0.38262

Probability 0.0111 0.9505 2.E-05 7.E-05 0.9288 0.6612

Adjustment Reject Accept Reject Reject Accept Accept

IFDIAV2GR does not Granger Cause RGDPGR 0.73413 Note: The conclusions are based on a 0.05 significance level.

From the Granger Causality Test we can conclude that EOGR and RGDPGR both

have Granger Causality on each other at the 1% significance level. This suggests that economic freedom and economic growth are tightly linked together in the U.S. and share a reciprocal relationship. In addition, IFDIAV2GR has Granger Causality on EOGR at the 5%, which advocates the use of foreign direct investment to predict the future trends of economic openness. These findings reinforce those from the Johansen Co-integration Test, as we now have discovered the specific long term relationships: economic freedom on economic growth, economic growth on economic freedom and foreign direct investment on economic freedom.

We can now use both real gross domestic product and foreign direct investment to predict economic openness in the U.S. This is an import finding as it has considerable policy implications for the U.S. government.

D.) Impulse Response Functions

Now that causality has been determined bilaterally between real gross domestic

product and economic openness, as well as foreign direct investment on economic openness, we examine the Impulse Response Functions for the specific relationships. Impulse Response Functions are used because they best ascertain the dynamic relationships in a VAR model, while a coefficient only represents a fraction of a single relationship. The blue line is the estimation while the red lines are boundaries set at two standard errors. The Impulse Response Functions of our VAR (8) model are displayed in FIGURES 3, 4 and 5 for 20 lags. 9

Response to Cholesky One S.D. InnovatiResponse of EOGR to EOGR.03.03Response of EOGR to IFDIAV2GRponse to Cholesky One S.D. Innovations ± 2 S.E.Response of EOGR to IFDIAV2GR.03.02FIGURE 3 .02.01.03Response of EOGR to RGDPGR.01.00.02.02.00-.01.01.01-.01-.02.00.002-.01-.0246810121416182024681012141618-.01-.022468101214161820504030-.022Response of IFDIAV2GR to EOGR468101214161820Response of IFDIAV2GR to IFDIAV2G50 50403020100-10-20-3024681012 40As seen in FIGURE 3, EOGR has a mostly positive response to a shock in Response of IFDIAV2GR to RGDPGR30Response of IFDIAV2GR to IFDIAV2GRRGDPGR, reaching a relative maximum in the second lag period, the equivalent of 6 205020104010months. It then fluctuates until slowly dissipating by lag period 12. This suggests that 00increases in Real GDP do cause an immediate and temporary increase in economic 3020100openness. 14161820-10-20-30-10-20-30468101214161820246810121416-102-20-30218FIGURE 4 468 RGDPGR to EOGR1012141618Response of20Response of RGDPGR to IFDIAV2GR1.01.0Response of RGDPGR to IFDIAV2GR1.00.80.60.40.20.00.80.60.40.20.41.00.80.6Response of RGDPGR to RGDPGR0.80.60.40.20.0-0.2-0.40.0-0.20.20.0-0.2-0.4-0.2-0.424681012141618202244668810101212141416161818202024681012141618-0.4 The inverse relationship presented in FIGURE 4 (which is on a different scale then FIGURE 3 and FIGURE 5) shows an immediate decrease in RGDPGR to a shock from EOGR, but then an increasing trend until peaking in period 8, then slowly diminishing. This result implies that a positive increase in economic openness, most likely through policy changes i.e. reduction of tariffs or quotas, decreases economic

10

growth in the short term as it takes time for firms to adjust to the new openness environment. However, once alterations are made to adapt to the new level of openness, taking a year or less (3 or 4 periods), economic growth will increase. This result is important as it allows government leaders in Washington D.C. to understand the direct effect their policies will have on U.S. Real GDP. The key point being that increases in economic freedom do not cause instantaneous beneficial results on Real GDP and short term losses should be anticipated in there undertaking. Response to Cholesky One S.D. Innovations ± 2 S.E.FIGURE 5 Response of EOGR to EOGR.03.03 Response of EOGR to IFDIAV2GR.03Response of EOGR to RGDPGR.02.02.02.01.01.01.00.00.00-.01-.01-.01-.022468101214161820-.022468101214161820-.02 FIGURE 5 displays the reaction of EOGR to a change in IFDIAV2GR over 20 Response of IFDIAV2GR to EOGRResponse of IFDIAV2GR to IFDIAV2GRResponse of IFDIAV2GR to RGDPGR50403020100-10-20-302468 2468101214161820lags or 4 1/2 years. We can first see that economic openness increases directly following 40405050a shock, remains positive for nearly 2 years and then decreases before returning back to a 30302020long-run equilibrium. The positive causation relationship from EOGR to IFDIAV2GR, in combination with the absence of the inverse relationship, suggests that it is economic 001010openness that attracts FDI and not FDI promoting the proliferation of economic openness. -10-10-20-20This is consistent with previous literature and reinforces the non-existence of FDI’s effect 10on economic openness. 1214161820-302468101214161820-3024681012141618201.00.80.60.40.20.0 Response of RGDPGR to EOGR 11 2468101214161820Response of RGDPGR to IFDIAV2GR1.00.80.60.40.20.0-0.2-0.424681012141618201.00.80.60.40.20.0-0.2-0.42Response of RGDPGR to RGDPGR-0.2-0.4468101214161820

.014.012.010.008.006.004.002.000FIGURE6 Accumulated Response of EOGR to CholeskyOne S.D. Innovations2468101214161820RGDPGRIFDIAV2GR

As both RGDPGR and IFDIAV2GR influence EOGR, we wish to examine the

relative sizes of these effects. FIGURE 6 shows the responses of EOGR to impulses from both RGDPGR and IFDIAV2GR simultaneously for comparison. It is evident that RGDPGR has a larger positive and longer lasting effect on EOGR. This is a critical discovery in the inter-workings of economic growth, foreign direct investment and economic openness. The implications being that it is more efficient to increase economic openness through increases in economic growth than increases in foreign direct investment.

It is also important to point out that in addition to the cross-relationships between RGDPGR, IFDIAV2GR and EOGR, each of the parameters has a significant effect on itself. While differing slightly from each other in the later lag periods, all three show an immediate increase followed by a decrease to the long run equilibrium. The magnitude of its effect on itself is also greater than the response shown in FIGURES 3, 4 and 5. This means that each factors lag is influenced more by itself then by the other two factors.

V. Conclusion

The VAR model used in the paper is designed to investigate the inter-workings

between economic growth, foreign direct investment and economic openness for the U.S. from 1960 to 2010. By using the VAR to conduct Granger Causality Tests and Impulse Response Functions we are able to determine the causal relationships between the three

12

variables, the bilateral relationship between economic growth and economic openness as well as from foreign direct investment to economic openness.

The conclusions drawn from our findings can be summarized into five specific points. First, economic growth causes an immediate and temporary increase in economic openness. Second, that economic openness decreases economic growth in the short-term, but increases it in the long-term. Third, economic openness attracts FDI, but FDI does not lead to increased economic openness. Forth, while both economic growth and foreign direct investment have an impact on economic openness, economic growth has a larger influence. Fifth, each parameter is primarily influenced by itself, with the other parameters having smaller effects. These points provide a good knowledge base when implementing new economic policy and should be considered as their effects are not negligible.

This paper has investigated the complexities surrounding the relative relationships between economic growth, foreign direct investment and economic openness as well as drawn several affirmative conclusions. However, more research needs to be conducted to truly understand and be able to manipulate the three factors. Our paper only provides an overview of the topic. Additional research should consider the recent changes in the domestic investment environment, the effects of global phenomena as additional endogenous variables, as well as considering alternative proxies to measure economic openness. It is only through these methods that a deeper understanding can be accomplished and to ensure the future development of the U.S. 13

References

Alfaro, Laura; Chanda, Areendam; Kalemli-Oxcan, Sebnem; Sayek, Selin (2000). “FDI and Economic Growth: The Role of Financial Markets.”

Bengoa, Marta; Sanchez-Robles, Blanca (2003). “Foreign direct investment, economic freedom and growth: new evidence from Latin America.” European Journal of Political Economy 19: 529-545

Berthelemy, Jean-Claude; Demurger, Sylvie (200). “Foreign Direct Investment and Economic Growth: Theory and Application to China.” Review of Development Economics 4.2: 140-155

Carkovic, Maria; Levine, Ross (2002). “Does Foreign Direct Investment Accelerate Economic Growth?” World Bank Conference (May 30-31, 2002), Financial Globalization: A Blessing or a Curse.

Chowdhury, Abdur; Mavrotas, George (2006). “FDI and Growth: What Causes What?” The World Economy 29.1: 9-19

Edwards, Sebastian (1997). “Openness, Productivity and Growth: What do We Really Know?” National Bureau of Economic Research: Working Paper 5978.

Hansen, Henrik; Rand, John (2006). “On the Causal Links Between FDI and Growth in Developing Countries.” The World Economy 29.1: 21-41

Harrison, Ann (1996). “Openness and growth: A time-series, cross-country analysis for developing countries.” Journal of Development Economics 48: 419-447

Najarzadeh, Reza; Shahri, Vaheed Shaghaghi (2008). “The Ranking of the OIC Members Countries Based on Factors Influencing Their Inward Foreign Direct Investments.” Iranian Economic Review 13.21: 107-122

14

Nair-Reichert, Usha; Weinhold, Diana (2000). “Causality Tests for Cross-Country Panels: New Look at FDI and Economic Growth in Developing Countries.”

Quazi, Rahim (2007). “Economic Freedom and Foreign Direct Investment in East Asia.” Journal of the Asia Pacific Economy 12.3: 329-344

Sun, Haishun; Hone, Philp; Doucoulago, Hristos (2003). “Economic openness and technical efficiency: A case study of Chinese manufacturing industries.” Economics of Transition 7.3: 615-636

15