Meta‐Analysis of Chinese Business Cycle Correlation
| Published date | 01 August 2018 |
| DOI | http://doi.org/10.1111/1468-0106.12173 |
| Date | 01 August 2018 |
Pacific Economic Review,••:•• (2016) pp. 1-29
doi: 10.1111/1468-0106.12135
bs_bs_banner
META-ANALYSIS OF CHINESE BUSINESS CYCLE
CORRELATION
JARKO FIDRMUC*Zeppelin University Friedrichshafen, and Henan University,
School of Economics, and Charles University Prague and Kaunas University of
Technology
IIKKA KORHONEN Bank of Finland Institute for Economies in Transition
Abstract. We review previous research on China’s business cycle correlation with other economies
applying meta-analysis. We survey 71 papers analysing the different periods of Chinese economic
development since the 1950s that were published in English or Chinese. We confirm that Pacific
Rim economies in particular have relatively high business cycle correlation with China. It appears
that many characteristics of the studies and authors influence the reported degree of business cycle
synchronization. For instance, Chinese-language papers report higher correlation coefficients.
Despite this, we do not detect robust evidence for publication bias in the papers. Moreover, we show
that the broad evidence does not confirm the popular decoupling hypothesis.
1. INTRODUCTION AND MOTIVATION
It is almosta truism that China’s economic significance has greatly increased during
the recent decades. This has, unsurprisingly, generated additional interest in busi-
ness cycle movements in China and in the synchronization of the cycles with other
economies. In this paper we use meta-analysis techniques to review the research on
international synchronization of the Chinese business cycle. This is important be-
cause it allows us to obtain a more comprehensive picture regarding the extent of
China’s integration with the rest of the world and especially with its neighbours.
Meta-analysis is used to summarize the findings of previous published litera-
ture on a selected topic in a systematic way (Stanley, 2001). Unlike the standard
literature reviews, meta-analysis applies statistical methods and is less subject to
subjective bias of the author. Meta-analysis is common in medicine and engi-
neering, but is also becoming more popular in economics. There are numerous
applications in different fields of economics, and these tools have become espe-
cially popular in international economics (e.g. Rose and Stanley (2005) investi-
gating the effect of currency unions on international trade). Any empirical
estimates can be summarized in this way: for instance, the link between financial
liberalization and growth (Bumann et al., 2012), the determinants of foreign
currency loans (Crespo Cuaresma et al., 2014), intertemporal elasticity of
substitution (Havránek et al., 2015), the misalignment of the renminbi
*Address for Correspondence: Zeppelin University Friedrichshafen, Am Seemooser Horn 20, 88045
Friedrichshafen, Germany. E-mail: jarko.fidrmuc@zu.de. We wish to thank Weijie Chen and Nanjun
Li for collecting and coding the data used in the analysis. We also thank Angela De Martiis, Nauro Cam-
pos and participants of seminars at the Hong Kong Institute for Monetary Research, the Henan Sympo-
sium of Development and Institutional Economics, the University of Jyväskylä and the Bank of Finland,
and an anonymous referee for useful comments. Iikka Korhonen wishes to thank the Hong Kong Insti-
tute for Monetary Research, where he worked on this project. Jarko Fidrmuc appreciates support
through a grant (No. MIP-016/2015) from the Research Council of Lithuania.
Pacific Economic Review,••:•• (2016)
doi: 10.1111/1468-0106.12173
© 2016 John Wiley & Sons Australia, Ltd
© 2016 John Wiley & Sons Australia, Ltd
Pacific Economic Review
, 23: 3 (2018) pp. 385–410
doi:10.1111/1468-0106.12173
© 2016 John Wiley & Sons Australia, Ltd
Pacific Economic Review
, 23: 3 (2018) pp. 385–410
doi:10.1111/1468-0106.12173
(Korhonen and Ritola, 2011) or the relationship between economic reforms and
growth (Babecky and Campos, 2011). However, so far there are only a few ap-
plications in business cycle analysis. De Grauwe and Costa Storti (2004) per-
form meta-analysis of the effects of monetary policies on output and prices.
Fidrmuc and Korhonen (2006) present a study on business cycle correlation
for the euro area enlargement. Rose (2008) reviews the literature on EMU, trade
and business cycle synchronization.
Meta-analysis can be used to summarize statistically and to review research
results on a given topic. It can be characterized as a systematic literature survey
in which all the papers published on a given topic are assigned at least some
weight in deriving a ‘consensus’or ‘aggregate’view on that topic. It is also a
means of assessing how authors’characteristics, variable specifications,
estimation techniques, data samples and other factors affect the reported results.
Finally, meta-analysis provides insights for understanding whether the
publication process may be biased in a particular direction.
A meta-regression analysis (Stanley, 2005, as well as Stanley, 2001) collects a
number of statistics from the surveyed literature (e.g. regression or correlation
coefficients) and regresses them on selected characteristics of the study design.
The significance of these study features indicates whether the results in previous
studies are systematically influenced by the data or methods used.
However, a possible drawback of meta-analysis is that it may compare apples
with oranges. This is especially important as there is no widely accepted approach
to measure the business cycle. The estimates of the business cycle may differ with
respect to the data (e.g. GDP vs industrial production, inflation or monetary ag-
gregates) or the methodology (simple correlations of e.g. growth rates or various
time series methods), thus introducing a significant degree of heterogeneity.
To deal with these difficulties, we proceed as follows. First, we look at the
synchronization between China and a selected group of economies measured
by correlation coefficients. Thus, the country-specific estimates of the business
cycle may be influenced by the authors’approach. However, this influence
should be eliminated once the bilateral degree of business cycle synchronization
is computed. Second, we control for the differences in the design of primary
studies, and we provide separate results for selected countries (e.g. the USA) ex-
amined in the published literature.
We survey 71 individual papers dealing with China’s business cycle
synchronization, which should represent all studies available on this topic, both in
the international and Chinese literature (including journals and unpublished work-
ing papers). All in all, these papers contain 1894 individual correlation coefficient es-
timates for China’s business cycle with other economies’business cycles, as all papers
report several correlation estimates. For many Asian economies (e.g. Japan, Korea,
Malaysia, Philippines, Singapore and Hong Kong) we have nearly 200 observations,
while the USA and Taiwan are represented by some 130 observations, respectively.
For the European countries we only have a small number of observations. While the
literature focuses on the most recent development, several authors use early data
(including 1950s and especially 1970s) on the Chinese business cycle as well.
J. FIDRMUC AND I. KORHONEN
© 2016 John Wiley & Sons Australia, Ltd
We contribute to the literature in several ways. First, we are able to systemat-
ically analyse the different economies’business cycle correlation with the
Chinese cycle. In particular, we find that correlation coefficients are relatively
high and statistically significant for many economies. This is especially true for
China’s immediate neighbours and ASEAN countries, but also for the USA
and New Zealand. The result confirms China’s eminent position within global
and regional production networks. Furthermore, it is possible that China’s
domestic demand is already so large that its imports are detrimental for the
economic development of the region.
Second, we show that some factors related to, for instance, the authors, ap-
plied methodologies and variables, affect the reported results. In fact, this does
not differ much from meta-analysis in other fields and countries. Based on our
preferred estimation specification, we observe, for example, that recent papers
are more likely to report higher correlation coefficients, and that papers that
do not have China as a specific focus are more likely to report lower correlation
coefficients. Moreover, we document several differences between Chinese and
English--language publications on business cycle synchronization, but no robust
publication bias is present in either stream of the literature.
While it is likely that China’s growing economic size and importance in global
supply chains will increase its business cycle synchronization over the coming
years, our results also warn against relying too much on any single estimate of
synchronization. In fact, we are able to show that the correlation coefficients re-
ported in any single study can be influenced by a large set of factors.
The paper is structured as follows. The next section describes our sample; that
is, papers published on the topic. The third section examines possible
publication bias in our sample and selected subsamples, in particular, the
differences between results published in English and Chinese, respectively. The
fourth section provides a statistical analysis of the literature on Chinese business
cycle correlation, and the last section concludes. The description of the
meta-search strategy (by search keywords) and the list of all publications used
for the meta-analysis are presented in Appendices A and B.
2. LITERATURE ON CHINA’S BUSINESS CYCLE SYNCHRONIZATION
The recent years have witnessed the publication of several papers related to
China’s business cycle synchronization with other economies. In this section
we start with the discussion of some broad trends in this strand of literature
and then present our data set; that is, the 71 related papers that we have
surveyed. The literature covers the whole period of China’s recent history, as
several authors include data since the 1950s and even more since the 1970s.
2.1. Meta-analysis design
Stanley (2001) identified different stages in carrying out meta-regressions. First,
all relevant studies are collected in a non-discriminatory manner to prevent any
CHINESE BUSINESS CYCLE
© 2016 John Wiley & Sons Australia, Ltd
© 2016 John Wiley & Sons Australia, Ltd
J. FIDRMUC AND I. KORHONEN
386
(Korhonen and Ritola, 2011) or the relationship between economic reforms and
growth (Babecky and Campos, 2011). However, so far there are only a few ap-
plications in business cycle analysis. De Grauwe and Costa Storti (2004) per-
form meta-analysis of the effects of monetary policies on output and prices.
Fidrmuc and Korhonen (2006) present a study on business cycle correlation
for the euro area enlargement. Rose (2008) reviews the literature on EMU, trade
and business cycle synchronization.
Meta-analysis can be used to summarize statistically and to review research
results on a given topic. It can be characterized as a systematic literature survey
in which all the papers published on a given topic are assigned at least some
weight in deriving a ‘consensus’or ‘aggregate’view on that topic. It is also a
means of assessing how authors’characteristics, variable specifications,
estimation techniques, data samples and other factors affect the reported results.
Finally, meta-analysis provides insights for understanding whether the
publication process may be biased in a particular direction.
A meta-regression analysis (Stanley, 2005, as well as Stanley, 2001) collects a
number of statistics from the surveyed literature (e.g. regression or correlation
coefficients) and regresses them on selected characteristics of the study design.
The significance of these study features indicates whether the results in previous
studies are systematically influenced by the data or methods used.
However, a possible drawback of meta-analysis is that it may compare apples
with oranges. This is especially important as there is no widely accepted approach
to measure the business cycle. The estimates of the business cycle may differ with
respect to the data (e.g. GDP vs industrial production, inflation or monetary ag-
gregates) or the methodology (simple correlations of e.g. growth rates or various
time series methods), thus introducing a significant degree of heterogeneity.
To deal with these difficulties, we proceed as follows. First, we look at the
synchronization between China and a selected group of economies measured
by correlation coefficients. Thus, the country-specific estimates of the business
cycle may be influenced by the authors’approach. However, this influence
should be eliminated once the bilateral degree of business cycle synchronization
is computed. Second, we control for the differences in the design of primary
studies, and we provide separate results for selected countries (e.g. the USA) ex-
amined in the published literature.
We survey 71 individual papers dealing with China’s business cycle
synchronization, which should represent all studies available on this topic, both in
the international and Chinese literature (including journals and unpublished work-
ing papers). All in all, these papers contain 1894 individual correlation coefficient es-
timates for China’s business cycle with other economies’business cycles, as all papers
report several correlation estimates. For many Asian economies (e.g. Japan, Korea,
Malaysia, Philippines, Singapore and Hong Kong) we have nearly 200 observations,
while the USA and Taiwan are represented by some 130 observations, respectively.
For the European countries we only have a small number of observations. While the
literature focuses on the most recent development, several authors use early data
(including 1950s and especially 1970s) on the Chinese business cycle as well.
J. FIDRMUC AND I. KORHONEN
© 2016 John Wiley & Sons Australia, Ltd
We contribute to the literature in several ways. First, we are able to systemat-
ically analyse the different economies’business cycle correlation with the
Chinese cycle. In particular, we find that correlation coefficients are relatively
high and statistically significant for many economies. This is especially true for
China’s immediate neighbours and ASEAN countries, but also for the USA
and New Zealand. The result confirms China’s eminent position within global
and regional production networks. Furthermore, it is possible that China’s
domestic demand is already so large that its imports are detrimental for the
economic development of the region.
Second, we show that some factors related to, for instance, the authors, ap-
plied methodologies and variables, affect the reported results. In fact, this does
not differ much from meta-analysis in other fields and countries. Based on our
preferred estimation specification, we observe, for example, that recent papers
are more likely to report higher correlation coefficients, and that papers that
do not have China as a specific focus are more likely to report lower correlation
coefficients. Moreover, we document several differences between Chinese and
English--language publications on business cycle synchronization, but no robust
publication bias is present in either stream of the literature.
While it is likely that China’s growing economic size and importance in global
supply chains will increase its business cycle synchronization over the coming
years, our results also warn against relying too much on any single estimate of
synchronization. In fact, we are able to show that the correlation coefficients re-
ported in any single study can be influenced by a large set of factors.
The paper is structured as follows. The next section describes our sample; that
is, papers published on the topic. The third section examines possible
publication bias in our sample and selected subsamples, in particular, the
differences between results published in English and Chinese, respectively. The
fourth section provides a statistical analysis of the literature on Chinese business
cycle correlation, and the last section concludes. The description of the
meta-search strategy (by search keywords) and the list of all publications used
for the meta-analysis are presented in Appendices A and B.
2. LITERATURE ON CHINA’S BUSINESS CYCLE SYNCHRONIZATION
The recent years have witnessed the publication of several papers related to
China’s business cycle synchronization with other economies. In this section
we start with the discussion of some broad trends in this strand of literature
and then present our data set; that is, the 71 related papers that we have
surveyed. The literature covers the whole period of China’s recent history, as
several authors include data since the 1950s and even more since the 1970s.
2.1. Meta-analysis design
Stanley (2001) identified different stages in carrying out meta-regressions. First,
all relevant studies are collected in a non-discriminatory manner to prevent any
CHINESE BUSINESS CYCLE
© 2016 John Wiley & Sons Australia, Ltd
© 2016 John Wiley & Sons Australia, Ltd
CHINESE BUSINESS CYCLE 387
© 2016 John Wiley & Sons Australia, Ltd
J. FIDRMUC AND I. KORHONEN
386
To continue reading
Request your trial