Beruflich Dokumente
Kultur Dokumente
13-1 (2013)
COMPARING THE SPEED OF CONVERGENCE
IN AMERICAN INTEGRATION AREAS
SPERLICH, Yvonne
*
Abstract
Economic growth theory predicts more success for North-South than for South-South
agreements. We compare the speed of convergence of the North-South area NAFTA
with that of the South-South area SICA and the one of the MERCOSUR. We apply
GSL fixed effects and a random effects regression with the Mundlak device,
respectively, both with an AR(1) errors. The results indicate that Southern agreement
areas clearly converge faster, whereas the within-convergence is faster for NAFTA.
But when looking at the confidence intervals, it is shown that the absolute information
value of the area-specific speed of convergence is rather limited
1
.
Keywords: regional integration, speed of convergence, Latin America,
NAFTA, MERCOSUR, CAIS, free trade areas, economic development
JEL code: F15, F43, O43, O11, E13
1. Introduction
In the last three decades, one important development is the increasing of North-South
(N&S) and South-South (S&S) agreements worldwide. The key motivation to create a
N&S-RIA is technology spill-over and enlargement of domestic market (Smulders and
Van de Klundert, 1996) whereas the institutional and political integration is often of
minor interest. However, the S&S-RIAs are founded mostly in order to improve the
regional security, solve cross-national problems, and strengthen the bargaining power
to third parties and reducing the economic dependence to the western industrialized
states. Consequently, the political and institutional issues are considered as being quite
important here. In this article we study the question whether different kind of RIAs
have different speeds of convergence or not.
For America we have the chance to compare a typical North-South with two kinds of
South-South integration areas (RIAs). The main focus is put on the speeds of
convergence, comparing integrating versus non-integrating countries for the different
types of RIAs and also comparing these RIAs in-between. We concentrate on the
following American RIAs as NAFTA (North American Free Trade Agreement), SICA
(Central American Integration System), and MERCOSUR (Mercado Comn del Sur).
The NAFTA is a N&S-RIA and the other both S&S agreements. We select these three
American RIAs along political topicality relating to pressure for reforms within the
existing RIAs, mainly due to the idea of the Free Trade of Americas (FTAA) as well as
*
Yvonne SPERLICH, Department of economics, University Geneva, Switzerland. E-mail:
yvonne.sperlich@bluewin.ch
1 Acknowledgement: The author appreciated a lot helpful comments of and discussion with
Inmaculada Martinez-Zarzoso (Universitat Jaume I), Ricardo Mora (Universidad Carlos III de
Madrid), Marcelo Olarreaga and Stefan Sperlich (Universit de Genve), Carola Grn and
Walter Zucchini (Universitt Gttingen, Research center for poverty, equity and growth), and
Ana Maria Alvarez Herrera (UNCTAD), the participants of the WTO, UNCTAD, UNIGE and
the Centre for Trade and Economic Integration seminars.
Applied Econometrics and International Development Vol. 13-1 (2013)
78
the foundation of the Dominican Rep.-Central America Free Trade Agreement (DR-
CAFTA) and the challenges of globalization (see Fugazza and Robert-Nicoud, 2006).
To obtain findings that are robust against misspecification, we calculate the speed of
convergence based on different modifications of the Solow growth model. As a by-
product when constructing confidence intervals we will also see why this measure must
not be taken too literally.
The regional integration and trade theory assume that N&S RIAs benefit in per capita
income and convergence though technological transfer, promote economic reforms and
FDI inflow. Along growth theory, however, this is much less predicted for the S&S-
RIAs, see for example Venables (1999). Nonetheless, several empirical studies
indicated that also the membership in S&S agreements has a positive impact on growth
and convergence which is partly in accordance (c.f. te Velde, 2011) and partly in
contradiction to a significant part of the literature based on economic theory. In this
present study we continue the discussion but also comment on the question which kind
of information can be obtained from the speed of convergence.
The speed of convergence is typically debated in the context of growth theory or trade,
as it is hoped to find out more about the long-term adjustment paths of countries in
terms of their per capita income development. The convergence rate provides
information about the time period in which the gap between the recent per capita GDP
of one country and the income of the leading country within a group will be closed.
Speed of convergence is a tool to bring all measures to the same scale making them
comparable. According to different predictions based on the human capital augmented
Solow model (HCA), regions are generally expected to converge at a speed of about
two per cent per year; see e.g. Sala-i-Martin (1996), though Evans (1997) stated even
faster rates. This implies that the half-life distance to its steady state is about 35 years.
If we take this value as a benchmark, it would be interesting to see whether the speed
of convergence for the different types of RIAs is higher or lower than two per cent p.a.
According to the new integration- and growth theory, the speed of convergence in
RIAs should be higher than in geographical regions due to increasing economic
cooperation (Smulders and Van de Klundert, 1996) or simply to certain similarities, see
De Benedictis and Tajoli (2007). As far as we know, there does not exist any empirical
study that compares the speed of convergence between RIAs or inside versus outside
RIAs for different American regions. For a most recent study of these regions see for
example Escobari (2011).
The model framework and methods
Let y(t) be the per capita income at time point t, y* the steady state income, n the
exogenous growth rate of the labour force, g the technological progress, and o the
depreciation rate. The speed of convergence is proportional to the distance to steady
state, cf. Barro and Sala-i-Martin (1995),
*) / ln(
)) ( ln(
y y
y
y
dt
t y d
for a particular .
Sperlich, Y. Comparing the Speed of Convergence in American Integration Areas
79
Let us consider the panel data equation for measuring unconditional -convergence,
it i
T
i i i
t y e t t y t y c
t
) ( ln ) 1 ( ) ( ln ) ( ln
1 2 1 2
, = t
2
- t
1
(1)
where
it
summarizes all heterogeneity,
i
are unknown (random or fixed) effects. Often
t
t
/ 1
e is denoted by , the so-called 'beta' coefficient. A particular extension of
the above equation is the human capital augmented (HCA)-Solow-model for panel data
with saving rates for capital and human capital, s
K
and s
H
constant returns to scale:
it
T
i i i
Hi Ki i i
t t y e g t n e
t s e t s e t y t y
q o
t o
t o
t o
t
t o
o
t t
t t
2 1 2
2 2 1 2
) ( ln ) 1 ( ) ) ( ln(
1
) 1 (
)) ( ln(
1
) 1 ( )) ( ln(
1
) 1 ( ) ( ln ) ( ln
(2)
with and being the unknown elasticises. The so-called convergence rate is used
for the estimation of the speed at which countries, respectively areas, converge to its
steady state. According to the panel equation above, the convergence rate is
calculated via t | t / )
1 ln(
)
and ub(