Have Free Trade Agreements Created Trade? Evidence from CEPA

Document

Cited in

Date	01 October 2019
DOI	http://doi.org/10.1111/1468-0106.12198
Author	Yonghong Zhou
Published date	01 October 2019

HAVE FREE TRADE AGREEMENTS CREATED TRADE?

EVIDENCE FROM CEPA

YONGHONG ZHOU*Jinan University

Abstract. This paper revisits the issue of trade creation effects in multi-stage free trade agreements

(FTA). In contrast to the empirical studies that use the gravity equation to study trade creation

effects, we apply a triple difference in differences approach to resolve their shortcomings, including

factors being omitted and endogeneity. After the identiﬁcation of treatment and control groups at

the eight-digit Harmonized System (HS) code product level, the regression analysis results show that

there are no signiﬁcant trade creation effects with the FTA, suggesting that the role of FTA may be a

signalling beyond trade.

1. INTRODUCTION

Trade creation is not a new topic in the study of international economics.

Originally, trade creation effects referred to the increase in trade brought about

through integration among membersof a trading community(Balassa, 1967). Over

the past 40 years, the development of free trade arrangements (FTA)1has pro-

moted regionalism around the world. Many published papers have examined the

trade creation brought about by FTAempirically; however, the results are mixed.

Since Balassa (1967), Aitken and Lowry (1972) and Aitken (1973), many re-

searchers have studied the trade creation brought about by FTA, including those

involving the European Economic Community(EEC)/the European Community

(EC), the European Free Trade Association (EFTA), the European Union (EU),

the Latin American Free Trade Association (LAFTA), the Central American

Common Market (CACM), the World Trade Organization(WTO) and its prede-

cessor the General Agreement on Tariffs and Trade (GATT). Some researchers

have found evidence for trade creation over their respective integration periods,

while others have not or have just found the trade creation effects to be fragile.

In the early studies, Balassa (1967), Aitken and Lowry (1972) and Aitken

(1973) found evidence of trade creation for LAFTA, CACM, EEC and EFTA.

Recently, Baier and Bergstrand (2007) analysed trade creation effects in 52 FTA

covering 96 countries and asserted that FTA approximately doubled two

*Address for correspondence: Department of Economics, Jinan University, and China Center for

Economic Development and Innovation Strategy of Jinan University, 601 Huangpudadao West,

Guangzhou, Guangdong 510632, China. E-mail: tzhouyhjnu@jnu.edu.cn. I am very grateful to Yi

Lu (National Singapore University) for his help. I thank the Editor, two anonymous referees, Anping

Chen (Jinan University), Bing Wang (Jinan University) and Peng Chen (Jinan University) for their

encouragement and constructive remarks. I gratefully acknowledge ﬁnancial support from the

National Natural Science Foundation of China (71103075, 71273116 and 71203077) and the

Fundamental Research Funds for the Central Universities of Jinan University (12JNYH002,

12JNKY001 and 15JNQM001).

1In reality, there are numerous types of international economic integration agreements, including

free trade agreements, regional trade agreements (RTA), customs unions, common markets and eco-

nomic unions. For ease of reference, we use the term ‘free trade agreements’generally.

Paciﬁc Economic Review,••:•• (2017)

doi: 10.1111/1468-0106.12198

members’bilateral trade over 10 years. A study by Anderson and Yotov (2011)

covered the period 1990–2002 for a total of 41 trading partners, which included

40 separate countries and the rest of the world, and showed evidence for trade

creation in FTA. Eicher et al. (2012) based their study on 164 importers and

177 exporters and spanned from 1960 to 2000, and the empirical results also

showed a strong trade creation effect. Estimates by Frankel and Wei (1993) sug-

gest that a country joining the EC in 1980 would have experienced an increase of

68% in trade with other members, but there was no such effect observed for the

EFTA. Based on the gravity model, the extreme bounds analysis carried out by

Ghosh and Yamarik (2004) tested the robustness of the trade creation hypothe-

sis and showed that the trade creation effect of most RTA was fragile. Rose’s

(2004) study covering 50 years and 175 countries also showed that there was

little evidence to support trade creation effects for the WTO/GATT.

Furthermore, the inﬂuenceof FTA may go beyond conventional trade impacts.

Fernandez (1997) examined the beneﬁts other than traditional gains produced

through regional trade arrangements, including credibility, signalling, bargaining

power, insurance and coordination. Antkiewicz and Whalley (2004) investigated

China’s recent cooperation in a wide range of FTA and implied that there was a

clear linkage between seeminglyconventional trade interests and China’s interests

in wider economic, diplomatic and strategic relationships. Banda and Whalley

(2005) evaluated the competition policy, investment, mutual recognition, move-

ment of persons and broader cooperation provisions in FTA. Hur and Park

(2012) found no signiﬁcant effect on growth performance but an upward trend

in the gap between the growth rates of per capita GDP of countries within an

FTA, which implied that uneven effects existed among countries within FTA.

Furthermore, theypointed out that technology transfer and trade diversionplayed

key roles in economic growth. FTA oftenconvey information about governments’

consideration of economicsupport for the regional economy, integration or future

policies. Such information can change market expectations and investors’

decisions. Therefore, the effects brought about by FTA would be quite great. In

fact, the purpose of FTA’s policy-makers would be possibly more than trade.

With the results of studies on trade creation being mixed, this paper will

conduct an analysis using both the nonparametric method and a quasi-natural

experiment with product-level data. We will apply triple difference in differences

(triple DID; also known as DDD (difference in differences in differences)) to un-

dertake our empirical study. In the empirical literature, most researchers esti-

mate the trade creation effects using the gravity equation or other similar

methods (see Aitken and Lowry, 1972; Frankel and Wei, 1993; Rose, 2004; An-

derson and Yotov, 2011; Eicher et al., 2012). In such regressions, the dependent

variable is bilateral trade ﬂow. The core dummy variable of FTA on the right-

hand side is used for identifying the average treatment effect (ATE). Other var-

iables include GDP, population, area, bilateral distance2and dummies for

2In a special case, Aitken and Lowry (1972) added the distance of the Latin American countries

from the United States to be control variables to measure the attraction and competitiveness of

the United States in international trade in these regions.

Y. ZHOU2

Paciﬁc Economic Review

, 24: 4 (2019) pp. 550–569

doi:10.1111/1468-0106.12198