1
ISS Discussion Paper Series F-162
Regionalization and Regionalism
in East Asia
Hiroyuki Hoshiro1
1 Institute of Social Science, The University of Tokyo (hoshiro@iss.u-tokyo.ac.jp)
2
Introduction
Since the end of the 1990s, an “East Asian Community” has attracted considerable
attention. A number of policy makers, academic scholars, and business executives have
discussed the possibility of the creation of a regional framework in East Asia 2
Regionalization is defined as an increase in the cross-border flow of capital,
goods, and people within a specific geographical area. It develops from the bottom up
through societally driven processes coming from markets, private trade, and investment
flows, none of which is strictly controlled by governments. The core players are
non-governmental actors—firms or individuals. Regionalization can be called a
spontaneous, bottom-up process. In contrast, regionalism is defined as a political will
(hence ism is attached as a suffix) to create a formal arrangement among states on a
geographically restricted basis. Since its main participants are governments, it can be
expressed as an artificial, top-down process. The “development of regionalization”
means an increase in the number of regional economic transactions such as money,
trade, and foreign direct investment (FDI). “Regionalism in progress” refers to the
agreement of regionally close governments to establish kinds of formal institutions such
as the Asia Pacific Economic Cooperation (APEC), the East Asia Summit, or bilateral
preferential trade agreements (PTAs) in order to cooperate with each other on various
issues. Few scholars oppose the idea that both regionalization and regionalism in East
Asia have been developing since the Asian financial crisis of 1997
.
Concerning this topic, there is mutual agreement among international relations scholars
that “regionalization” and “regionalism” are different phenomena, but there are
unresolved arguments as to whether there is a causal relationship between the two.
3
However, there is a huge disagreement over the relationships between the two
phenomena, in particular, the causal relationship between regionalization and
regionalism. One dominant view emphasizes that regionalization automatically
promotes regionalism. Contrary to this widely shared view, however, some scholars
.
2 “East Asia” here consists of the ASEAN 10 countries and Australia, the People's Republic of
China (hereafter China), India, Japan, New Zealand and the Republic of Korea (hereafter South
Korea).
3 However, as I will mention below, Ravenhill (2010) doubts this understanding.
3
argue that regionalization is not necessarily a catalyst for regionalism. Moreover,
because of the complexities of the relationship between the two, some scholars have
given up attempting to untangle this problem.
The purpose of this paper therefore is to shed light on the relationship between
regionalization and regionalism by use of statistical analyses for East Asian countries.
Since a number of political scientists and economists have verified that regional
arrangements (the outcome of regionalism) lead to regionalization, this paper focuses
on the reversed vector, that is, it investigates whether regionalization promotes
regionalism4
.
1 Existing Research on Regionalization and Regionalism
Since the mid-1990s, a certain number of scholars in international relations have
emphasized the necessity of distinguishing between regionalization and regionalism as
analytical concepts and of clarifying the relationship between the two (Fishlow and
Haggard 1992; Haggard 1993; Frankel 1993; Wyatt-Walter 1995; Pempel 2005). From this
time on, many scholars have wrestled with this issue.
First, there is a widely shared view that regionalism leads to a growth in trade, that
is, it is one demonstration of economic regionalization (Aitken 1973; Frankel 1993;
Winters and Wang 1994). Clearly, this argument is not surprising. As a result of a decrease
in tariffs and non-tariff barriers between two or more countries concluding trade
arrangements, a so-called “trade creating effect” would cause an increase of trade among
members. According to research that uses a CGE(Computable General Equilibrium)
model, if East Asian-wide PTAs (ASEAN+3 or ASEAN+6) were established, for example,
the economic growth of each member would increase (Kawai and Wignaraja 2009, 19-20).
Moreover, some argue that not only trade but also FDI would increase if trade agreements
were concluded. Since international institutions, such as PTAs, enable governments to
make more credible commitments to a liberal economic policy and prevent even
4 As only institutionalized regional agreements in the region are PTAs, I refer to regional
arrangements and PTAs interchangeably. This definition is consistent with that of Mansfield and
Milner (1999, 592).
4
developing countries from making arbitrary interventions such as regulation, taxation, or
tariff increases, foreign investors can safely make investments in countries joining regional
trade arrangements (Büthe and Milner 2008).
However, the economic effect of some individual PTAs in East Asia has not been
so clear. Ando (2007), for example, concludes that the Japan-Singapore EPA (Economic
Partnership Agreement) has had little impact on trade. Dent (2010, 218) also does not rate
the Singapore-Peru PTA, inferring that the expected annual trade and investment
liberalization “gains” are possibly less than the actual administrative cost of negotiating and
implementing the agreement itself. Thus, there are cases where participation in a regional
trade arrangement does not necessarily lead to an increase in economic interaction in East
Asia5
I will now examine the reversed vector, that is, whether increased regional
economic interdependence (regionalization) encourages governments in the region to
conclude or join economic arrangements such as PTAs (regionalism). The number of PTAs
has been increasing worldwide since the early 1990s and this tendency is particularly
remarkable in East Asia after 2000 (Figure 1). Can this be explained by enlarged
economic interdependence in the region?
.
At present, there are two competing views on this issue. The dominant view asserts that
regionalization is an inevitable driving force for regionalism. For example, Kawai (2005, 30)
argues that “[the] most fundamental rationale behind the emergence of recent economic
regionalism is the deepening of regional economic interdependence in East Asia.”
Munakata (2006, 29) shares Kawai’s stance by emphasizing that the intensity of economic
interaction contributes substance and depth and thereby a basis for institutionalized
intergovernmental cooperation, including preferential trade agreements (see also Lim
5 Although not the subject of this paper, the question of why states try to conclude PTAs even
though their expected economic effects are low is interesting and merits further research and
investigation.
5
1990, 21). Furthermore, Lincoln (2004, chap. 3) takes a negative view as to the necessity
and economic effect of East Asian regional trade arrangements because he argues that
intra-regional trade has been decreasing rather than increasing and identifies Japan’s
economic decline as the main factor in this. This logic derives its meaning from the
assumption that growth in regional economic interaction drives regionalism. All of these
views seem to be based on a conventional international relations thesis—institutions lower
transaction costs, reduce uncertainty, monitor compliance, and enhance opportunities for
more cooperation (Keohane 1984).
Contrary to this thesis, some scholars are skeptical about this simple, linear
relationship. Ravenhill (2010, 182) insists that regionalization in East Asia has indeed been
decreasing if we look at row shares in trade, which show the region’s changing share of
global commerce via the trade intensity index6
This existing research does have flaws, however. It does not examine the
relationship between the two phenomena systematically. The former is no more than policy
recommendation without empirical evidence (and does not even make an effort to probe its
own arguments). The latter suffers from a similar problem. Ravenhill (2010, 185) points out
the odd relationship (from the perspective of existing studies) between China and other
countries—those countries that have experienced trade increase with China but have
actually eschewed PTAs with it—as an illustration of his arguments. However, this
illustration is drawn from a small number of cases, as well as from outcomes, suggesting
that selection bias have occurred. Moreover, Ravenhill’s main argument is to reject
increased economic interdependence (regionalization) in East Asia, not the relationship
between regionalization and regionalism itself
. Therefore, Ravenhill argues that this is not
consistent with the widely shared view that both regionalism and regionalization in East
Asia have been increasing since the late 1980s. Moreover, Haggard (1997, 45) suggests
the possibility that regionalization may prevent regionalization, by stating that
“despite—and arguably because of—the extremely rapid growth of trade and investment,
there has not been strong demand within Asia for greater policy coordination.”
7
6 Ravenhill shares this understanding with Lincoln (2004).
.
7 In this point, Ravenhill (2010, 185) seems to accept the argument that PTAs are negotiated in
6
The influence of regionalization over regionalism has thus not been fully
examined. It is natural for some studies to avoid attempting to make clear which is cause
and which is consequence and to simply describe these relations as a “mutual
reinforcement process” (Dent 2008, 8). The aim of this paper, therefore, is to settle the
dispute by examining systematically the relations between regionalization and regionalism
by means of statistical analyses of East Asian trade data from 1948 to 2010.
2 The Method and the Model
The statistical method employed here is called “event history analysis” or “survival
analysis.” Since this paper aims to estimate the influence of regionalization over
regionalism (PTAs), the dependent variable is a binary outcome denoting whether a PTA
exists or not—it equals 1 if a PTA exists between state i and state j, 0 otherwise. The most
important independent variable is the amount of trade in the region from 1948 to 20108
response to the policy challenges posed by increasing interdependence. He points out that East
Asian countries have concluded or are currently negotiating with states outside the region that
have been experiencing growth in economic interaction with the East Asian countries.
. A
normal maximum likelihood method, such as logistic regression, raises the problem of
endogeneity, that is, if two countries concluded a PTA in 2005 for instance, the amount of
trade after 2006 would be affected by the agreement, which is the dependent variable.
Since this paper analyses the influence of economic factors over PTAs and not vice versa,
the analysis should end once partner countries enforce a PTA. Therefore, conclusion of a
PTA is read as the occurrence of the event (this is also called “failure”) and probability of it
is estimated. Survival analysis is the most appropriate statistical method for the purposes
of this paper. The two country pairs that have not signed PTAs, such as US-Japan, are
called “right censored” data. Since this analysis includes time-varying covariates (TVCs),
such as the amount of trade or GDP, I have used a discrete-time model. Data are shown in
Table 1 below.
8 As mentioned above, regionalization refers to a general term including increased trade, FDI,
and emigration within a specific region. Therefore, the amount of FDI and emigration should be
incorporated into independent variables. However, due to difficulties of data accessibility
(especially data for Southeast Asian nations), I use only trade data here. I will try to include FDI
and emigration data in future research.
7
The unit of analysis is a dyad product year and the dyad is cross-country data. East Asia
is defined here as ASEAN+6—the original members of the East Asian Summit. Since
ASEAN has unusually had PTAs with individual countries as if they are a single unit,
ASEAN is regarded as a country. Therefore, the total number of countries is 179. However,
given the fact that many of the current PTAs that East Asian governments have been
concluding are cross-regional trade agreements (Katada and Solís 2008), I have
expanded the data to APEC countries (except Taiwan)10
in a different model.
Dependent Variable
PTA data as a dependent variable are drawn from the WTO website11
9 These are ASEAN, Australia, Brunei Darussalam, Cambodia, China, India, Indonesia, Japan,
Republic of Korea, Laos, Malaysia, Myanmar, New Zealand, the Philippines, Singapore,
Thailand, and Vietnam.
(see also Figure 1).
If a country has PTAs with the same country more than twice through bilateral or
multilateral negotiations (for example, China has PTAs with ASEAN, signed in 2005, as
well as with Singapore, signed in 2009), I have chosen the bilateral one because bilateral
agreements are expected to have a more direct economic impact than multilateral ones.
Moreover, if countries concluded more than two multilateral PTAs separately (Indonesia
and Malaysia, for example, joined the GSTP [Global System of Trade Preferences among
Developing Countries] in 1989 and the AFTA [ASEAN Free Trade Area] in 1992), I have
chosen the earlier one. In the case of ASEAN, if a country has PTAs with ASEAN but not
with individual ASEAN countries, only the PTA with ASEAN is counted. For example,
since New Zealand concluded a PTA with ASEAN in 2010 but not with the Philippines, I
have counted only the ASEAN-New Zealand PTA, not the New Zealand-Philippines PTA.
10 Taiwan is excluded because no data are found in the IMF Direction of Trade Statistics
Dataset. The added countries are Canada, Chile, Hong Kong, Mexico, Papua New Guinea,
Peru, Russian Federation, and the United States.
11 http://www.wto.org/english/tratop_e/region_e/region_e.htm (accessed February 6, 2012).
8
Independent Variables
First, the most important variable is an enlarged economic interdependence
(regionalization). I measured this as the volume of trade between two countries. Trade data
are taken from IMF’s database whose timeframe is from 1948 to 2010 12 . As PTA
negotiations normally take two to five years from initiation to completion, I have taken three
years’ lag on the amount of trade (import plus export, expressed in US dollars) as an
independent variable13. Moreover, not only nominal value but also relative value (the
amount of trade between country j and country i divided by the sum of the total amount of
j’s worldwide trade and i’s worldwide trade)14
Furthermore, since regionalization implies an ongoing process, continuity of
growth in economic transactions may be more important than just one term’s trade.
Therefore, INCREASE(NO) and INCREASE(RE) are operationalized as dummy variables
that take a value of 1 if the amount of trade (whether nominal or relative) had been
increasing for five consecutive years in year t-3, and 0 otherwise. If the argument that
regionalization leads to regionalism is correct, the following hypothesis will be confirmed.
is taken as an independent variable. The
natural logarithm of the nominal trade is named LN_TRADE(NO) and the relative one
TRADE(RE) in year t-3.
H1: the greater the amount of trade (or if this is continuously increased), the greater the
likelihood that a pair of countries conclude PTAs.
Coefficients of LN_TRADE(NO), TRADE(RE), INCREASE(NO), and
INCREASE(RE) are expected to be positive.
Contrary to this linear relationship, some scholars have pointed out that bilateral PTAs in
East Asia are not between pairs of countries that have expectation of a high trade creation
12 IMF Direction of Trade Statistics Dataset, various years. If two countries’ data are not
corresponded, I used the developed countries ones.
13 Both four and five years’ lags are also estimated as Model1-2 and Model1-3 respectively.
14 Formally, Tradei, j (t-3)/(Tradei,world(t-3) + Tradej,world(t-3)).
9
effect nor those that have a large amount of trade interactions (Endo 2006; Ravenhill
2010; Dent 2010; Solís 2011). A large amount of trade means each country is also large
economically and this leads to an increase in the number of social economic actors—veto
players—which have a vested interest in foreign trade and tend to make policy change
difficult (Tsebelis 2002). They may prevent governments from promoting PTAs.
Furthermore, there seems little incentive for a government to invest time and money in
concluding PTAs with those with whom they have few trade relations. Therefore, the
relationship between the amount of trade and PTAs is assumed to be an inverted
U-shaped curve. So new variables, LN_TRADE(NO)_SQ, TRADE(RE)_SQ, DIFGDP, and
DIFGDP_SQ are introduced. LN_TRADE(NO)_SQ and TRADE(RE)_SQ are the square of
LN_TRADE(NO) and TRADE(RE) respectively. DIFGDP is measured as the absolute
value of difference of the natural logarithm of nominal GDPs (Gross Domestic Products)
between pair countries15. DIFGDP_SQ is the square of DIFGDP. GDP data are from the
World Bank’s Data Catalog16
. If previous studies are right, the following hypothesis will be
correct as well.
H2: The larger (and smaller) the amount of trade, the less likely a pair of countries have
PTAs.
LN_TRADE(NO)_SQ, TRADE(RE)_SQ, and DIFGDP_SQ are expected to have
negative coefficient.
Counter to these hypotheses, I expect those variables are not significant at all. As
regional economic ties are already strong in East Asia17
15 |ln GDPi - ln GDPj|
, the incentives for policy makers
and business elites to create formal arrangements should be weak. The purpose of trade
arrangements is to increase trade. Why do governments need to conclude new
agreements when economic transactions are already increasing sufficiently? This de-fact
integration might have prevented an EU-type regionalism in the region. That is:
16
http://databank.worldbank.org/ddp/editReport?REQUEST_SOURCE=search&CNO=2&topic=3
(accessed February 26, 2012)
17 This is often called “de-fact integration.”
10
H3: Growth in trade between pair countries has no effects on PTAs.
Other Independent Variables
Although it is not the objective of this project to seek determining factors for increased
PTAs in East Asia, other independent variables are introduced in order to control the main
variable mentioned above. These other variables are the number of worldwide PTAs in the
previous year, a dummy variable which indicates whether two countries are