Is the Provincial Capital Market Segmented in China?

• Published date: 01 August 2013
• Date: 01 August 2013
• DOI: http://doi.org/10.1111/rode.12041

Is the Provincial Capital Market Segmented

in China?

Kenneth Chan, Jennifer Lai, and Isabel Yan*

Abstract

This paper evaluates the degree of provincial capital mobility in China since 1978 by estimating the saving-

retention rate in the Feldstein and Horioka framework. It is found that the estimate of the saving retention

rate tends to be biased downward if we fail to decompose the investment and saving rates into the private

and government components. After the decomposition, we find that the private capital mobility was low

prior to the 1990s, but improved notably in the more recent period. We also find that the saving-retention

rate for the government sector remains negative throughout the sample, which can partially be explained by

the government’s reallocation of capital from the more productive regions to the less productive regions.

Comparing the results of China with those of Japan shows that Japan had a much higher degree of prefec-

tural capital mobility than China before the 1990s, but the gap gradually closed up afterwards.

1. Introduction

Since the seminal paper of Feldstein and Horioka (1980), numerous studies have used

the saving and investment correlation (also called the saving-retention rate) to evalu-

ate the degree of capital mobility across countries and within a country. Earlier

studies mostly apply the framework to study capital mobility across countries. For

example, Feldstein and Horioka (1980), Obstfeld (1986, 1993) and Tesar (1991) esti-

mate the saving-retention rates across Organisation for Economic Co-operation and

Development (OECD) countries. However, country level saving and investment rates

are subject to the current account solvency constraint that complicates the interpreta-

tion of the saving retention rate (Coakley et al., 1996). Some later studies like

Bayoumi and Rose (1993), Helliwell and Mckitrick (1999), Yamori (1995), Dekle

(1996) as well as Iwamoto and Van Wincoop (2000) adapt the framework to study

provincial capital mobility within a country that does not subject to the same con-

straint.1The first two studies focus on UK and Canada respectively, and the latter

three on Japan.2

A more recent study by Boyreau-Debray and Wei (2004) apply the framework to

study the provincial capital mobility in China. They obtain a saving-retention rate

estimate of around 0.5 using provincial saving and investment data spanning from

1952 to 2001. They conclude that there was no improvement in the provincial capital

mobility across China over the period. Li (2010) employs the panel cointegration

method to the provincial data of China and obtains similar estimates of the saving

* Lai: School of International Trade and Economics, Guangdong University of Foreign Studies, Guang-

zhou Higher Education Mega Center, Guangdong Province, P.R. China. Tel: 86-20-39328592; Fax: 86-20-

39328096; E-mail: econlai@gmail.com. Chan, Yan: Department of Economics and Finance, City University

of Hong Kong, Kowloon Tong, Hong Kong. We wish to thank Nelson Mark, Gregory Chow, Fred Y. K.

Kwan, Shangjin Wei, Chia Hui Lu, and the participants of the seminar held at Hong Kong Institute for

Monetary Research in March 2011 for their helpful comments. All remaining errors are ours. Lai would

like to thank the Research Center for International Economics in City University of Hong Kong for provid-

ing financial support for this project.

Review of Development Economics, 17(3), 430–446, 2013

DOI:10.1111/rode.12041

© 2013 John Wiley & Sons Ltd

retention rates (see Table 1 for a summary of the estimates). Moreover, he reaches

the same “imperfect capital mobility” conclusion as in Boyreau-Debray and Wei

(2004). Both studies found a moderate degree of provincial capital mobility within

China, which is at a similar level as that across the OECD countries but is at a signifi-

cantly lower level than that within each OECD country (see Tables 2 and 3 for a

summary of the inter-country and intra-country estimates of the saving–retention rate

for OECD countries, respectively). In other words, the barriers to provincial capital

mobility in China are still as high as those across the OECD countries. Chan et al.

(2012) applied fixed and random effects panel regressions to the provincial level data

of China and finds no significant improvement in capital mobility based on the esti-

mates of the total saving retention rate. However they observe that for private sector

there is a slight increase and then a decline in saving retention rate over the periods

1970–82 and 1995–2006 respectively. Chan et al. (2011) employs the bootstrap panel

cointegration method to regional level data of China and finds that there is a clear

divergence between saving retention rates of the total and private sector across differ-

ence regions in China. Their results show that the private sector saving retention rate

is much higher than the total saving retention rate.

This paper contributes to the Feldstein–Horioka literature in three aspects: first, we

decompose the provincial data of China from 1978 to 2006 into the government and

private components to gauge how economic reform affects the degree of private and

government capital mobility,3and the empirical results we obtain are robust.4Second,

we employ rolling window procedure to examine how the degree of provincial capital

mobility changes over time in China. To preview the result, we find a continuous

improvement in the provincial capital mobility for the private sector of China except

for a temporary period between 1993 and 1996 when double-digit inflation hit

the economy. Third, in view of the close proximity between China and Japan, we

compare our results of China with those of Dekle (1996) which studies the degree of

capital mobility across the Japanese prefectures.

Our paper differs from the previous works such as Chan et al. (2011, 2012) in the

following three aspects: first, we use time averaged cross sectional data while Chan

et al. (2011, 2012) use fixed (random) panel regression and bootstrap panel

cointegration respectively. Each method has its own merits, and it is useful to cross-

check if the results from different approaches affirm one another. Second, using cross

sectional regression allows us to compare our results with the previous literature,

which largely uses cross-sectional regressions to evaluate within country capital

mobility. Third, we use provincial data from 1978 to 2006 in 26 provinces, with

missing data excluded. Chan et al. (2012) uses a longer dataset ranging from 1970 to

2006 in which they include 29 provinces, and allow missing values in their unbalanced

panel.5However, the data shows that three provinces experienced abnormal gross

domestic product (GDP) growth rate of around 100% before 1978. The data after

1978 is more reasonable and there are no outliers. This suggests that the data before

1978 may have quality problem. In view of this, our paper only uses data after 1978.

In another study, Chan et al. (2011) forms a balanced panel with data from nine

regions between 1978 and 2006.6

2. Data

Data of the 26 provinces7in the years 1978–2006 is mainly collected from the China

Statistical Yearbooks, and is supplemented by the provincial statistical yearbooks. The

PROVINCIAL CAPITAL MARKET IN CHINA 431

© 2013 John Wiley & Sons Ltd