WTO accession, foreign bank entry, and the productivity of Chinese manufacturing firms

ARTICLE IN PRESS

JID: YJCEC

[m3Gsc;July 21, 2015;22:30]

Journal of Comparative Economics 000 (2015) 1–17

Contents lists available at ScienceDirect

Journal of Comparative Economics journal homepage: www.elsevier.com/locate/jce

WTO accession, foreign bank entry, and the productivity of Chinese manufacturing firmsR Tat-kei Lai a, Zhenjie Qian b,∗, Luhang Wang c a

Department of Economics, Copenhagen Business School, Denmark School of Banking and Finance, University of International Business and Economics, Huixin East 10, Beijing 100029 China c Wang Yanan Institute for Studies in Economics and Department of International Economics and Business, Xiamen University, China b

a r t i c l e Article history: Received 31 July 2014 Revised 19 June 2015 Available online xxx JEL Classfication: D24 O14 G21 Keywords: China Foreign bank entry WTO TFP Technical efficiency Reallocation

i n f o

a b s t r a c t Lai, Tat-kei, Qian, Zhenjie, and Wang, Luhang—WTO accession, foreign bank entry, and the productivity of Chinese manufacturing firms After China’s accession to the World Trade Organization (WTO) in December 2001, foreign banks are allowed to enter the Chinese banking market in phases. Using firm-level data from the National Bureau of Statistics of China which cover all state-owned and non state-owned manufacturing firms with sales over 5 million RMB, we examine the relationship between foreign bank entry and the industry-level productivity growth of China’s manufacturing sector. Our empirical results suggest that (a) on average, opening up a region for foreign bank entry has no impact on aggregate productivity growth, (b) however, industries more dependent on external finance grow faster after a region is opened up for foreign bank entry, and (c) these results are due to changes in technical efficiency rather than reallocation. Overall, this paper provides new evidence on the relationship between banking market structure and manufacturing productivity in a fast growing developing country. Journal of Comparative Economics 000 () (2015) 1–17. Department of Economics, Copenhagen Business School, Denmark; School of Banking and Finance, University of International Business and Economics, Huixin East 10, Beijing 100029 China; Wang Yanan Institute for Studies in Economics and Department of International Economics and Business, Xiamen University, China. © 2015 Association for Comparative Economic Studies. Published by Elsevier Inc. All rights reserved.

1. Introduction The Chinese banking sector has traditionally been dominated by the “Big 4” state-owned commercial banks which, in general, have worse performance than other banks (Lin et al., 2009).1 After China’s accession to the World Trade Organization (WTO) on R We thank two anonymous referees, Colin Xu and seminar and conference participants at the Copenhagen Business School, Chinese Economics Association (U.K./ Europe) Conference 2012, Comparative Analysis of Enterprise Data & COST Conference 2012, European Association for Research in Industrial Economics Annual Conference 2013, and the Second Annual Xiamen University International Workshop on Economic Analysis of Institutions (2014) for comments and suggestions. Zhenjie Qian acknowledges financial support of Beijing Planning Office of Philosophy and Social Science (grant no.: 14JGB063) and the Fundamental Research Funds for the Central Universities in UIBE (grant no: CXTD4-03). Luhang Wang acknowledges the financial support from the Scientific Research Foundation for the Returned Overseas Chinese Scholars, State Education Ministry (grant no.: SRF-1231/K1300003). All remaining errors are our own. ∗ Corresponding author. Fax: +861064493330. E-mail addresses: [email protected] (T.-k. Lai), [email protected], [email protected] (Z. Qian), [email protected] (L. Wang). 1 The “Big 4” banks include the Bank of China (BOC), the Agricultural Bank of China (ABC), the Construction Bank of China (CBC), and the Industrial and Commercial Bank of China (ICBC).

http://dx.doi.org/10.1016/j.jce.2015.06.003 0147-5967/© 2015 Association for Comparative Economic Studies. Published by Elsevier Inc. All rights reserved.

Please cite this article as: T.-k. Lai et al., WTO accession, foreign bank entry, and the productivity of Chinese manufacturing firms, Journal of Comparative Economics (2015), http://dx.doi.org/10.1016/j.jce.2015.06.003

JID: YJCEC 2

ARTICLE IN PRESS

[m3Gsc;July 21, 2015;22:30]

T.-k. Lai et al. / Journal of Comparative Economics 000 (2015) 1–17

December 11, 2001, foreign banks can enter the local currency market in phases; since the end of 2006, there have been no restrictions on foreign bank entry. Recent studies have documented that the entry of these foreign banks tends to be associated with a more competitive and efficient banking environment in China (e.g., Jiang et al., 2009; Lin et al., 2009; Xu, 2011).2 During the same period, the Chinese economy has grown rapidly with GDP growth rate increasing from roughly 8% in 2001 to above 14% in 2007.3 To the extent that the banking sector becomes more competitive and efficient after foreign bank entry is allowed in China, does opening up the banking sector for foreign bank entry contribute to the phenomenal growth of China’s manufacturing sector, the main engine of the Chinese economy? The answer to this question is not straightforward for at least two reasons. First, the existing literature suggests a non-trivial relationship between banks and manufacturing firms because of sorting. Second, domestic banks may not all compete against foreign banks in the same way. Therefore, the impact of foreign bank entry will depend on the sorting pattern between banks and manufacturing firms. Furthermore, the external dependence of the industries (Rajan et al., 1998) may also influence the impact of foreign bank entry and the associated change in banking market structure. In this paper, we try to shed light on these issues by relating different dimensions of the performance of Chinese manufacturing firms to the removal of foreign bank entry barriers. Theoretically, banking competition can improve allocation efficiency and promote economic growth (Pagano (1993)); with higher credit availability, firms can benefit, and those in more financially dependent industries can benefit even more. On the other hand, due to information asymmetry, banks with market power can internalize the benefits of helping financially constrained firms (Petersen et al., 1995). Banking competition thus discourages banks from forming lending relationships with firms so that firms should suffer. However, firms that are more financially dependent can suffer less (or can even benefit) because banks’ returns for the formation of lending relationship with these firms should be higher. Empirically, studies using industry-level data do show mixed results.4 This relationship is further complicated since foreign bank entry may have different effects on the access to credit by firms of different sizes. The extant literature finds that small- and medium-sized enterprises (SMEs) tend to borrow from smaller banks which have informational advantages over the larger banks (e.g., Berger et al., 2005; Cole et al., 2004).5 In the case of foreign bank entry into developing countries, Detragiache et al. (2008) argue that, relative to domestic banks, foreign banks have comparative disadvantage in monitoring “soft” information customers so that these customers may be hurt by foreign bank entry. Detragiache et al. (2008) also find evidence that developing countries with more foreign bank penetration have a shallower banking sector. Other existing studies also suggest that foreign bank entry tends to benefit larger firms and may even hurt SMEs (see, e.g., Gormley (2010); Mian (2006)).6 One main limitation of the existing empirical studies in the literature using industry-level data is that with these data one cannot separate productivity improvement due to within-firm technological progress from the efficiency gain following acrossfirm reallocation. Such differentiation is crucial in evaluating the impact of foreign bank entry when there is sorting between banks and manufacturing firms. In this paper, we use the firm-level manufacturing data obtained from the National Bureau of Statistics of China to shed light on the interaction between banking market structure and industry growth. This data set covers all manufacturing state-owned firms and non state-owned firms having sales over 5 million RMB between 1998 and 2007, representing about 90% of the gross total output in the manufacturing industries. Motivated by recent studies using microlevel data to study productivity of Chinese firms which highlight the importance of technical efficiency and allocation efficiency (e.g., Brandt et al., 2012a; Hsieh et al., 2009), we use these data to examine whether foreign bank entry into China is associated with the productivity improvements of the domestic manufacturing firms and whether such changes are related to technical or allocation efficiency.7 Interestingly, Brandt et al. (2012b) find that reallocation at the extensive margin contributes substantially

2 Other empirical studies also document that foreign banks or foreign ownership of local banks are more efficient than state-owned banks (e.g., Berger et al., 2009). In an international context, using a sample of bank observations from 80 countries between 1988 and 1995, Claessens et al. (2001) find that increased presence of foreign banks is related to a lower profitability and net interest margins for local banks. Besides, foreign banks are found to be more efficient than the local ones in developing countries (see, e.g., the survey by Clarke et al. (2003)) and their presence enhances the access to credits across firms in developing countries (Clarke et al., 2006). 3 See data from the World Bank: http://data.worldbank.org/indicator/NY.GDP.MKTP.CD. 4 For instance, using industry-level cross-country data (similar to the set of countries studied by Rajan et al. (1998)), Cetorelli et al. (2001) find that bank concentration has a negative impact on industry growth; but bank concentration promotes growth of those industries that are more dependent on external finance. Claessens et al. (2005) use industry-level data in 16 European countries and find that bank competition has a positive impact on growth of more financially dependent industries. Moreover, the state of economic development may also affect such interactions. For example, Demirgüç-Kunt et al. (2013) use data on 72 countries between 1980 and 2008 and find that the banking sector is less important relative to the securities market as the economy develops; Cull et al. (2013) examine the relationship between labor growth rate and the development of banking and financial sectors using data from the World Bank Enterprise Survey which covers 89 countries in various years between 2000 and 2009. They find that in poorer countries labor growth is positively related to ratio of private credit to GDP whereas in richer countries, labor growth is positively related to the level of stock market capitalization. 5 Within China, Shen et al. (2009) find that bank size alone does not matter much for SME lending; other factors such as hierarchical levels, local bank incentives, competition and law enforcement are also important. On the other hand, Chong et al. (2013) find that lower market concentration in general reduces financing constraints of SMEs, and the effect also depends on bank size and ownership structure. 6 One exception is Clarke et al. (2006), who use data from the World Business Environment Survey (WBES) covering 35 countries and find that foreign bank entry helps reduce the financing constraints for all firms (including SMEs). 7 Hsieh et al. (2009) examine the efficiency loss driven by the differential between actual and optimal marginal revenue caused by distortions in goods and factor market. Using an efficiency index, defined as the ratio of a hypothetical TFP without distortions to the actual TFP, to reflect the allocation efficiency in manufacturing industry, they find that allocation efficiency has improved since 1998 in China’s manufacturing industry. In contrast to the misallocation emphasis in Hsieh et al. (2009), Brandt et al. (2012b) calculate TFP with industry data and find that the creative entry has taken over half over the TFP growth at industry while the incumbents contribute the rest. On the whole, these previous works on productivity of Chinese manufacturing industry show that technical efficiency, allocation efficiency, and net entry of new firms are all relevant.

Please cite this article as: T.-k. Lai et al., WTO accession, foreign bank entry, and the productivity of Chinese manufacturing firms, Journal of Comparative Economics (2015), http://dx.doi.org/10.1016/j.jce.2015.06.003

JID: YJCEC

ARTICLE IN PRESS T.-k. Lai et al. / Journal of Comparative Economics 000 (2015) 1–17

[m3Gsc;July 21, 2015;22:30] 3

to the productivity growth during this period of time.8 On the other hand, studies by Buera et al. (2011) and Midrigan et al. (2014) reveal that distortion at the extensive margin accounts for the majority portion of TFP losses due to financial frictions. Against this backdrop, this paper will evaluate the impact of allowing foreign bank entry on the TFP growth along different margins in China’s manufacturing sector to shed some light on the effectiveness of this reform in promoting financial development in China. In the empirical analysis, we follow Petrin et al. (2012) to compute the aggregate productivity growth, defined as the change in output net of input for surviving firms, weighted by their corresponding output share. This term can be further decomposed into components reflecting the contribution of more productive firms keeping input constant (technical efficiency), and the output change resulted from the resource movement from firms with different marginal revenue (reallocation). Foreign bank entry is measured by a binary variable to indicate whether foreign banks are allowed to enter a certain region after China’s accession to WTO. To characterize the reliance on external finance of different industries, we use U.S. Compustat data to compute the external dependence measure (Rajan et al., 1998). The within-region and over-time variation of the foreign bank entry dummy together with the cross-industry variation in the degree of dependence on external finance allow us to identify the heterogeneous effects of foreign bank entry on the industry-level productivity growth. We find that on average foreign bank entry does not affect aggregate productivity growth. However, industries with greater dependence on external finance have higher aggregate productivity growth relative to those with less dependence on external finance. Examining the two growth components, namely technical efficiency and reallocation, we find that the effect of potential foreign bank entry on aggregate productivity growth is reflected mostly in changes in technical efficiency rather than allocation efficiency. Besides, foreign bank entry is also unrelated to net entry. We further examine the aggregate productivity impact of foreign bank entry on firms with different sizes and ages. We find that when foreign bank entry is allowed, only larger and older firms in industries with higher external dependence have significantly larger aggregate productivity growth. Our paper contributes to several strands of literature. Firstly, our paper is related to the studies on the relationship between banking market structure and industry growth, especially on the impact of increasing competition in the banking sector associated with the entry of foreign banks.9 One caveat of using cross-country variation to identify the impact of the market structure of the banking sector on the manufacturing sector (such as Cetorelli et al., 2001; Claessens et al., 2005) is that the estimated coefficient captures only the average of the effect across countries with different institutional backgrounds, which call for caution when using the estimate to draw country specific policy implications. Inklaar et al. (2012) use a sample of small- and medium-sized firms from Germany and find that bank competition has a positive effect on growth in industries with low external dependence but the relationship becomes negative for industries with high external dependence. In this paper, we use a more representative sample of manufacturing firms in China and find very different patterns, suggesting that firm size and country-specific factors may play a role in explaining the complex relationship between banking market structure and productivity growth. Secondly, our paper contributes to studies on the determining factors of China’s phenomenal productivity growth since the late 1990’s. Substantial growth in China has been documented in the literature and certain aspects of trade liberalization, such as tariff reduction, are found to be among the most important contributing factors.10 Not much has been done regarding the role of the liberalization of the financial sector though. Closely related to our study, Lin (2011) examines the effect of foreign bank entry after China’s WTO accession on domestic firms’ access to bank credit. She finds that after foreign bank entry, more profitable firms and non-state-owned firms use more long-term bank loans whereas firms with higher value of potential collateral do not use more bank loans. We contribute to this literature by focusing on the impact of one specific reform in the banking sector on firm productivity in a more representative sample. Finally, our paper is also related to the literature on the impact of financial frictions and the resulting capital misallocation on aggregate TFP growth. Hsieh et al. (2009) documents substantial TFP losses due to misallocation in China and India. More recent studies, for example Buera et al. (2011) and Midrigan et al. (2014), specifically investigate the role of financial frictions and find that financial frictions affect aggregate TFP mainly through distorting the entry behavior at the extensive margin. This paper focuses on one specific reform in the financial sector, foreign bank entry, and evaluates its impact on different margins of TFP growth of China’s manufacturing sector. We find no evidence that foreign bank entry helps relieve the financial constraints of small or young firms. Our results suggest that the entry of foreign banks in China reduces capital misallocation at the intensive margin rather than the extensive margin. One possible explanation is that banking competition after foreign bank entry only increases among larger banks which mainly serve the larger firms; as a result, distortions at the extensive margin may not have declined. Consistent with the findings in Buera et al. (2011) and Midrigan et al. (2014), without improvements at the extensive margin the impact on the aggregate TFP growth is small. This is not too surprising given the widespread existence of other forms of restrictions in the financial sector, such as designated interest rate of deposits. Overall, our results call for deeper domestic reforms in the financial sector in order to reduce capital misallocation in China. The remaining part of the paper is organized as follows. In Section 2, we provide an overview of the history of the reform in China’s banking sector. Section 3 describes the data we use in this study. Section 4 presents the main results of our empirical analysis and robustness checks. Section 5 concludes.

8 According to their study, the share of contribution of the extensive margin is 62% (versus 38% by continuing firms). The share of the extensive margin over the longer period of 1998 to 2007 is 72% . For the United States, the comparable share is 26% over the period 1977 to 1987. 9 See also a related study by Zhang et al. (2012) which focuses on the role of financial development in China’s economic growth. 10 For example, see Brandt et al. (2012b).

Please cite this article as: T.-k. Lai et al., WTO accession, foreign bank entry, and the productivity of Chinese manufacturing firms, Journal of Comparative Economics (2015), http://dx.doi.org/10.1016/j.jce.2015.06.003

JID: YJCEC 4

ARTICLE IN PRESS

[m3Gsc;July 21, 2015;22:30]

T.-k. Lai et al. / Journal of Comparative Economics 000 (2015) 1–17 Table 1 Foreign bank activities in China.

Operational bank entities∗ Assets (in billion RMB) Share of total banking assets

2004

2005

2006

2007

2008

188 582.3 1.84%

207 715.5 1.91%

224 927.9 2.11%

274 1252.5 2.38%

311 1344.8 2.16%

Note: Including head offices of locally incorporated banks, branches and subsidiaries of locally incorporated banks, and foreign bank branches. Source: Annual report of China Banking Regulatory Commission, 2008.

2. Chinese banking sector and foreign bank entry: an overview In this section we discuss some background information about the banking sector and foreign bank entry in China. For a recent survey of the development of the Chinese financial system, see also Allen et al. (2008). 2.1. The Chinese banking sector before WTO accession Prior to 1978, the People’s Bank of China (PBC) was the only bank in the country. It not only served as the central bank but also provided commercial banking service to the country. Since the economic reform of China began in 1978, the Chinese banking sector has undergone different changes. At the inception of the economic reform, a two-tier banking system was created with the establishment of four state-owned commercial banks (the “Big 4”). Under the two-tier system, PBC continued to serve as the central bank, and the other four banks specialized in commercial banking operations. Initially the Big 4 commercial banks served their own designated sectors and, in 1985, they were allowed to compete in different sectors of the economy. To foster greater competition in the banking sector, other new small and medium-sized commercial banks, many of which being joint-stock, were established between mid 1980s and early 1990s.11 In the 1990s, the commercial banks had gradually accumulated large non-performing loans because they made most of their loans to state-owned enterprises which were not as efficient in generating profits and did not have incentives to repay loans. In 1994, the government recapitalized the Big 4 commercial banks and established three policy banks to take over the policylending activities from the Big 4.12 Other measures were introduced, including formalizing the role of PBC as the central bank and reducing the influence by local governments, liberalizing the controls on interest rate and credit, strengthening the supervision of banks, and allowing foreign banks to do local-currency business. 2.2. Foreign bank entry While foreign banks can operate in China by different ways, including setting up branches, wholly-owned banks, joint ventures, and strategic investment, their activities in China were subject to strict regulations and entry requirements. Before 1993, foreign banks were only allowed to open branches in Special Economic Zones and conduct foreign-currency business with foreigners. Since 1993, the geographic restrictions were gradually relaxed. For example, foreign banks could take deposits and make loans in local currency to Chinese firms in Shanghai Pudong New Zone and Shenzhen Special Economic Zone in 1996. They were further allowed to conduct business in nearby regions in 1999. However, foreign banks were not allowed to provide consumer banking services to local residents in the 1990s. Upon China’s entry to the WTO on December 11, 2001, the restrictions on foreign bank entry were lifted in phases. In particular, according to the protocols of accession: “For foreign currency business, there will be no geographic restriction upon accession. For local currency business, the geographic restriction will be phased out as follows: upon accession, Shanghai, Shenzhen, Tianjin and Dalian; within one year after accession, Guangzhou, Zhuhai, Qingdao, Nanjing and Wuhan; within two years after accession, Jinan, Fuzhou, Chengdu and Chongqing; within three years after accession, Kunming, Beijing and Xiamen; within four years after accession, Shantou, Ningbo, Shenyang and Xi’an. Within five years after accession, all geographic restrictions will be removed.”13 Nevertheless, these foreign banks are still governed by different regulations, including minimum entry capital, minimum total assets, and previous presence in China. Table 1 summarizes the activities of foreign bank entities (including head offices of locally incorporated banks, branches and subsidiaries of locally incorporated banks, and foreign bank branches.) in China between 2004 and 2008. During this period, the number of foreign bank entities operating in China increased from 188 in 2004 to 311 in 2008. On the other hand, foreign banking assets have increased from about 582 billion RMB in 2004 to about 1,345 billion RMB in 2008 while the share of foreign banking 11 These banks include the Bank of Communications, the CITIC Industrial Bank, the Shenzhen Development Bank, the Guangdong Development Bank, the China Merchants Bank, the China Everbright Bank, and the Hua Xia Bank. 12 The three policy banks are the Agricultural Development Bank of China, the China Development Bank, and the Export-Import Bank of China. 13 See page 34 of the document WT/ACC/CHN/49/Add.2, which is available at http://www.wto.org/english/thewto_e/acc_e/completeacc_e.htm.

Please cite this article as: T.-k. Lai et al., WTO accession, foreign bank entry, and the productivity of Chinese manufacturing firms, Journal of Comparative Economics (2015), http://dx.doi.org/10.1016/j.jce.2015.06.003

ARTICLE IN PRESS

JID: YJCEC

T.-k. Lai et al. / Journal of Comparative Economics 000 (2015) 1–17

[m3Gsc;July 21, 2015;22:30] 5

assets remains small, roughly at 2%, over the same period. Their activities are concentrated in larger cities such as Shanghai, Shenzhen, Beijing, and Guangzhou (Wang, 2007). 3. Data and variable definitions 3.1. Data source In our main empirical analysis, we use the firm-level manufacturing data obtained from the National Bureau of Statistics of China. This data set covers all state-owned and non-state-owned manufacturing firms having sales over 5 million RMB. The sample period of the whole data set is between 1998 and 2007 with number of firm observations ranging from about 160,000 (in 1998) to about 330,000 (in 2007). This unbalanced panel of firms represents about 90% of the gross total output in the manufacturing industries. The data set provides firm-level information about their value-added, capital stock, employment, labor compensation, location, and ownership type. Brandt et al. (2012b) provides an excellent description of the data set. We refer interested readers to their paper for further details. Note that because of merger, privatization and other reasons, firms often change their identifiers during the sample period. Brandt et al. (2012a) argue that this issue would affect the accuracy of the productivity estimates. They fix the problem by identifying firms not only with the reported identifiers, but also their telephone numbers, registration addresses, and other qualitative information. For the same reason, we adjust firm identifiers following their approach. Besides, we notice that industry standard changed from 2002 to 2003 which would potentially affect our empirical results. We follow Brandt et al. (2012b) to align the industry code before 2002 (the “GBT/T4754-1994”) with that used since 2003 (the “GBT/T4754-2002” or “CIC 2002” hereafter) to come up with a consistent industry classification over time.14 3.2. Variable definitions Productivity measures. Our main outcome variables of interest are aggregate productivity growth and its components for surviving firms. Using our data, we follow Petrin et al. (2012) to construct aggregate productivity growth, APG, at the region and industry levels, and decompose it into changes in technical efficiency, TE and reallocation efficiency, RE. •

Aggregate productivity growth: Petrin et al. (2012) define aggregate productivity growth as the change in the aggregate final demand minus the change in aggregate expenditure on labor and capital inputs. In our context, aggregate productivity growth of surviving firms from year t − 1 to year t in region r and industry j, and year t is computed as:

APGr jt =



D¯ vir jt [ log VAir jt − s¯ir jt  log Lir jt − (1 − s¯ir jt ) log Kir jt ],

(1)

i∈C

•

where i is the index for the firms in each region-industry-year cell surviving from year t − 1 to year t, and C is the set of all surviving firms, and VAirjt , Lirjt , and Kirjt , respectively denote value-added, employment, and real capital stock of the firm, and D¯ vir jt is the average share of value-added of the firm i between years t − 1 and t (i.e., the Domar weight), and s¯ir jt is the average labor income share of firm i in industry j and region r between years t − 1 and t. In this paper, we measure nominal valueadded by income approach following Qian et al. (2011).15 With nominal value-added by income approach, we calculate labor income share as labor compensation over value-added net of production tax. Domar weight is also computed using nominal value-added, while VAirjt is deflated using price index proposed by Brandt et al. (2012a). We also follow Brandt et al. (2012b) to compute real capital stock, Kirjt . Technical efficiency: Petrin et al. (2012) define technical efficiency as the contribution to the aggregate productivity growth from more productive firms holding inputs fixed. Specifically, technical efficiency for surviving firms is computed as follows:

T Er jt =



D¯ vir jt ( log VAir jt − βL j  log Lir jt − βK j  log Kir jt ),

(2)

i∈C

where β Lj , and β Kj are the value-added elasticities of labor and capital, estimated from the following log-linearized production function for each industry (at 2-digit CIC 2002 level) with the method suggested by Ackerberg et al. (2006):

log VAi jt = β0 j + βL j log Li jt + βK j log Ki jt + εi jt ,

(3)

14

See the online appendix of Brandt et al. (2012b) for further details. The National Bureau of Statistics China (2007), National Bureau of Statistics China (2008) actually measures value-added using the average of value-added by the income approach and value-added by the production approach. While the latter is readily available in the data set, the former needs to be estimated following the approach stated by the National Bureau of Statistics China (2007), National Bureau of Statistics China (2008). However, not all variables necessary for valueadded by the income approach are available in the manufacturing data. We construct proxies for these variables using the most relevant information available in the data set as in Qian et al. (2011). We use value-added by the income approach in this paper as our baseline estimate because the National Bureau of Statistics has completely abandoned value-added by the production approach for the lack of accuracy in this measure since 2008 (National Bureau of Statistics China, 2010). Furthermore, our unreported results show that computing outcome variables with value-added by production or the average of the two value-added measures does not change our main results. 15

Please cite this article as: T.-k. Lai et al., WTO accession, foreign bank entry, and the productivity of Chinese manufacturing firms, Journal of Comparative Economics (2015), http://dx.doi.org/10.1016/j.jce.2015.06.003

ARTICLE IN PRESS

JID: YJCEC 6

[m3Gsc;July 21, 2015;22:30]

T.-k. Lai et al. / Journal of Comparative Economics 000 (2015) 1–17 •

Reallocation: we define reallocation as the difference between aggregate productivity growth and technical efficiency to account for the output changes arising from the resource movements from firms with different marginal revenues:

REr jt = APGr jt − T Er jt .

(4)

We want to emphasize that by reallocation efficiency we only consider resource movement among incumbents from year t − 1 to year t. As we do not include entry and exit effect in our baseline estimation, our reallocation efficiency has a different sample coverage from that of Hsieh et al. (2009). Nevertheless, we will consider the entry and exit effects as a robustness check. Foreign bank entry. According to the WTO accession protocol, 20 Chinese cities are selected for foreign bank entry between the end of 2001 and the end of 2005: • • • • •

Open up by the end of 2001: Shanghai, Shenzhen, Tianjin, and Dalian. Open up by the end of 2002: Guangzhou, Zhuhai, Qingdao, Nanjing, and Wuhan. Open up by the end of 2003: Jinan, Fuzhou, Chengdu, and Chongqing. Open up by the end of 2004: Kunming, Beijing, and Xiamen. Open up by the end of 2005: Shantou, Ningbo, Shenyang, and Xi’an.

After the end of 2006 there is no geographical restriction on foreign bank entry. We call the above 20 cities “regions opened up before the end of 2005” (or simply “opened regions”) and the other provinces (excluding the “opened regions”) “other regions”.16 We follow Lin (2011) and define a dummy variable (denoted by FB) if foreign banks are allowed to enter region r in year t and 0 otherwise.17 One may argue that this dummy variable is not informative about (a) whether foreign banks actually enter a particular region and (b) if they do, the extent of their lending, especially to manufacturing firms. Ideally, one would want to measure foreign bank activities in terms of their total assets or total loans in a given region and year. Besides, one would also want to measure the banks’ distributions of loans to different types of firms. There are two main empirical challenges of employing such measures of foreign bank activities. The first is about data limitation. From the annual reports of the banks, normally we do not observe the distribution of assets in different regions and the distribution of loans to different types of firms. On the other hand, from our manufacturing data, we could observe such financial ratios as the debt ratio (total liability/total assets) but we do not observe the source of funds (whether the funds come from a domestic or a foreign financial institution). Therefore, we are unable to find data to construct such measures of foreign bank activities at detailed levels. The second reason is that, even though such measures are available, they are likely to be endogeneous in the baseline regression. One may still want to use the same foreign bank entry dummy or other policy changes as an instrument to estimate the impact of foreign bank activities on the productivity of Chinese manufacturing firms. Given these empirical challenges, we use the simple foreign bank entry dummy, taking the view that when this dummy equals to 1, it indicates that the region is subject to potential foreign bank entry. Thus, this dummy is associated with the extent of competition in the domestic banking market. Some recent studies find that foreign banks are more efficient, and in the case of China, their entry makes the banking market more competitive (Xu, 2011). So even if foreign banks do not actually enter, their potential entry would impose competitive pressure to the domestic incumbents and should therefore have an impact on industry growth. Overall, we consider that the coefficient estimate of the foreign bank entry dummy is in itself a meaningful measure of the overall impact of this financial market opening policy, both directly through the increased banking activities of the actual new foreign entrants and indirectly through the increasing banking competition pressure due to the potential entry threat to the domestic banks. External dependence. We follow Rajan et al. (1998) and use U.S. Compustat data to construct the following measure of firms’ dependence on external finance:18

ED =

Capital expenditure − Cash flow from operations . Capital expenditure

This ratio measures the fraction of capital investment that cannot be financed by internal cash of the firm. Thus, a larger ratio indicates that the firm requires more external funds to finance its investment projects. We compute the sums of capital expenditure and cash flow from operations over the period 1990–1997 (i.e., we use data before the start of the sample period of the Chinese manufacturing data) to calculate the ratio for each firm and take the industry median as the external dependence measure for each industry. We calculate the external dependence measure at 3-digit SIC levels and then match with the CIC 2002 codes (at 2-digit levels) in our Chinese manufacturing data.19 Table 2 shows the list of industries and their external dependence measures. 16

We use the first 4 digits of region code in the raw data to match with the names of the “opened regions.” “Other regions” are aggregated to 2-digit levels. In the appendix we show some simple correlations between foreign bank presence and some banking market characteristics. 18 Since the U.S. economy is one of the most developed in the world, there should be less financial constraints in different industries. The financial dependence measure constructed using U.S. data can serve as a “benchmark” for other countries. On the other hand, endogeneity may exist when we use Chinese data to create the above measures. 19 Specifically, we use the concordance between CIC 2002 and ISIC rev3 and the concordance between ISIC rev3 and U.S. SIC 1997. 17

Please cite this article as: T.-k. Lai et al., WTO accession, foreign bank entry, and the productivity of Chinese manufacturing firms, Journal of Comparative Economics (2015), http://dx.doi.org/10.1016/j.jce.2015.06.003

ARTICLE IN PRESS

JID: YJCEC

[m3Gsc;July 21, 2015;22:30]

T.-k. Lai et al. / Journal of Comparative Economics 000 (2015) 1–17

7

Table 2 List of industries and their external dependence. Industry (2-digit)

External dependence

Smelting and pressing of ferrous metals Manufacture of rubber Manufacture of paper and paper products Manufacture of foods Processing of food from agricultural products Manufacture of metal products Manufacture of plastics Printing, reproduction of recording media Manufacture of furniture Manufacture of textile wearing apparel, footware and caps Manufacture of leather, fur, feather and related products Manufacture of textile Manufacture of special purpose machinery Manufacture of transport equipment Manufacture of electrical machinery and equipment Manufacture of general purpose machinery Manufacture of non-metallic mineral products Manufacture of measuring instruments and machinery for cultural activity and office work Manufacture of raw chemical materials and chemical products Manufacture of artwork and other manufacturing Manufacture of articles for culture, education and sport activities Processing of timber, manufacture of wood, bamboo, rattan, palm and straw products Smelting and pressing of non-ferrous metals Processing of petroleum, coking, processing of nuclear fuel Manufacture of communication equipment, computers and other electronic equipment Manufacture of chemical fibers Manufacture of beverages Manufacture of medicines

−0.052 0.078 0.087 0.108 0.148 0.267 0.295 0.302 0.305 0.313 0.314 0.333 0.352 0.370 0.407 0.428 0.457 0.551 0.560 0.582 0.585 0.688 0.712 0.723 0.764 1.076 1.194 1.385

Table 3 Descriptive statistics. Panel A: summary statistics Obs.

Mean

Panel B: correlation matrix

S.D.

Min.

1st Q.

Median

3rd Q.

Max.

APG 1.000 0.868 (0.000) 0.240 (0.000) 0.107 (0.000) −0.044 (0.000)

APG TE

9363 9363

0.060 0.076

0.199 0.201

−0.840 −0.847

−0.044 −0.026

0.063 −0.012

0.167 0.178

0.918 0.988

RE

9363

−0.016

0.103

−0.576

−0.051

0.021

0.022

0.501

FB

9363

0.272

0.445

0.000

0.000

0.000

1.000

1.000

ED

9363

0.432

0.405

−2.470

0.295

0.352

0.560

1.385

TE

RE

FB

ED

1.000 −0.274 (0.000) 0.124 (0.000) −0.028 (0.007)

1.000 −0.035 (0.001) −0.030 (0.004)

1.000 0.025 (0.018)

1.000

Note: Panel A shows the summary statistics of various variables. Panel B shows the pairwise correlation coefficients. The numbers in parentheses are p-values.

3.3. Descriptive statistics In the empirical analysis, we focus on manufacturing firms (those firms with 2-digit CIC 2002 industry codes between 13 and 42). Excluding missing values, we have 9363 region-industry-year observations in the regression sample. Table 3 reports some descriptive statistics for various variables: Panel A shows the summary statistics, whereas Panel B shows the pairwise correlation coefficients among different variables. In Fig. 1(a), we show the distributions of the aggregate productivity growth, technical efficiency, and reallocation; in Fig. 1(b), we show the annual means of the productivity measures over time. 4. Empirical analysis 4.1. Baseline results To examine whether opening up a region for foreign bank entry is related to changes in aggregate productivity growth and its components, and whether any such effects are related to the external dependence of the industries, we estimate the following linear regressions:

yr jt = αr + α j + αt + β1 F Brt + εr jt ,

(5)

Please cite this article as: T.-k. Lai et al., WTO accession, foreign bank entry, and the productivity of Chinese manufacturing firms, Journal of Comparative Economics (2015), http://dx.doi.org/10.1016/j.jce.2015.06.003

JID: YJCEC 8

ARTICLE IN PRESS

[m3Gsc;July 21, 2015;22:30]

T.-k. Lai et al. / Journal of Comparative Economics 000 (2015) 1–17

Fig. 1. Aggregate productivity growth, technical efficiency, and reallocation.

Please cite this article as: T.-k. Lai et al., WTO accession, foreign bank entry, and the productivity of Chinese manufacturing firms, Journal of Comparative Economics (2015), http://dx.doi.org/10.1016/j.jce.2015.06.003

ARTICLE IN PRESS

JID: YJCEC

[m3Gsc;July 21, 2015;22:30]

T.-k. Lai et al. / Journal of Comparative Economics 000 (2015) 1–17

9

Table 4 Main regression results. (1)

(2)

(3)

(4)

(5)

(6)

Panel A: region-year and industry-year specific shocks not controlled for Dependent variable: APG TE RE FB

0.001 (0.014)

FB × ED Region fixed-effects Industry fixed-effects Year fixed-effects Observations Adjusted R2

Yes Yes Yes 9363 0.070

−0.014 (0.013) 0.035∗∗∗ (0.008) Yes Yes Yes 9363 0.070

0.007 (0.015)

Yes Yes Yes 9363 0.085

−0.007 (0.015) 0.031∗∗∗ (0.011) Yes Yes Yes 9363 0.085

Yes Yes Yes 9363 0.031

Panel B: controlling for region-year and industry-year specific shocks Dependent variable: APG TE FB

0.000 (0.006)

FB × ED Region fixed-effects Industry fixed-effects Year fixed-effects Observations Adjusted R2

Yes Yes Yes 9363 0.106

−0.015∗∗∗ (0.004) 0.035∗∗∗ (0.008) Yes Yes Yes 9363 0.106

0.003 (0.008)

Yes Yes Yes 9363 0.120

−0.011∗ (0.007) 0.031∗∗∗ (0.011) Yes Yes Yes 9363 0.120

−0.007∗ (0.004) 0.004 (0.007) Yes Yes Yes 9363 0.031

−0.005 (0.004)

−0.004 (0.003)

Yes Yes Yes 9363 0.034

RE −0.006∗ (0.003) 0.004 (0.008) Yes Yes Yes 9363 0.034

Note: Standard errors are clustered at region and industry levels, and are reported in parentheses. ∗∗ significance at 5% level; ∗ significance at 10% level; ∗∗∗ significance at 1% level.

yr jt = αr + α j + αt + β1 F Brt + β2 (F Brt × ED j ) + εr jt ,

(6)

where r, j, and t are indexes for region, industry, and year, respectively; yrjt ∈ {APGrjt , TErjt , RErjt , NErjt } is an outcome variable of interest; FBrt is the foreign bank entry dummy; EDj is the external dependence measure; α r , α j , and α t are the regions, industry, and year fixed-effects, and ε rjt is the error term. In the first regression, the coefficient of interest is β 1 ; it tells us the average change of the outcome variable across industries when a region is opened up for foreign bank entry. Its identification comes from the within-region and over time variation of the foreign bank entry dummy. For illustration, consider the case of Shanghai. According to the WTO accession protocol, foreign banks can enter the market of Shanghai starting from the end of 2001. Thus, Shanghai is an “opened region” starting from 2002. To evaluate the treatment effect, we could have compared the outcome variables before and after 2002. However, other relevant factors (such as some national-wide policies implemented in 2002) might also affect the outcomes in Shanghai. We control for this possibility by including other regions which are not yet opened at the same time (e.g., Guangzhou, which is opened for foreign bank entry by the end of 2002). The difference in the outcome variables between these two regions before and after 2002 should give us an estimate of the average effect over industries of opening up Shanghai for foreign bank entry. The coefficient of interest in the second regression is β 2 which tells us the differential effects of foreign bank entry according to the external dependence of the industries. There is an additional source of identification, coming from the exogenous variation in the external dependence of different industries. For example, due to exogenous technological differences, machinery industries require more external funding to finance investment projects than food producing industries. Therefore, the machinery and food producing industries have different “exposures” to changes in banking market structure followed by foreign bank entry. This variation allows us to identify the effect of foreign bank entry on the outcome variables of interest for industries with different degrees of external dependence. The regression results for (5) and (6) are reported in Panel A of Table 4. The dependent variables are aggregate productivity growth (in columns (1) and (2)), technical efficiency (in columns (3) and (4)), reallocation (in columns (5) and (6)), respectively. In each set of regressions, we regress the dependent variable on the foreign bank dummy with region fixed-effects, industry fixed-effects, and year fixed-effects in the first specification; we also include the interaction between the foreign bank dummy and the external dependence measure in the second specification. In all these regressions, we cluster the standard errors by regions and industries. In column (1), the coefficient of the foreign bank dummy is statistically insignificant, suggesting that on average, opening up a region for foreign bank entry does not have any effect on the aggregate productivity growth of the manufacturing firms. In column (2), when we interact the foreign bank dummy with the external dependence measure, we find that the coefficient of the interaction term is positive and statistically significant. In other words, industries more dependent on external finance grow faster, relative to those less dependent on external finance, when a region is opened up for foreign bank entry. Please cite this article as: T.-k. Lai et al., WTO accession, foreign bank entry, and the productivity of Chinese manufacturing firms, Journal of Comparative Economics (2015), http://dx.doi.org/10.1016/j.jce.2015.06.003

JID: YJCEC 10

ARTICLE IN PRESS

[m3Gsc;July 21, 2015;22:30]

T.-k. Lai et al. / Journal of Comparative Economics 000 (2015) 1–17 Table 5 Main regression results: economic significance.

Percentile of ED 5th 10th 25th 50th 75th 90th 95th

Marginal effect of FB on APG −0.013∗∗∗ (0.004) −0.012∗∗∗ (0.004) −0.005 (0.004) 0.003 (0.004) 0.004 (0.005) 0.011∗ (0.006) 0.026∗∗∗ (0.009)

Note: Standard errors are clustered at region and industry levels, and are reported in parentheses. ∗∗ significance at 5% level; ∗ significance at 10% level; ∗∗∗ significance at 1% level.

We then examine the relationship between foreign bank entry, external dependence, and the two components of aggregate productivity growth. From columns (3) and (5), we find that opening up a region to foreign bank entry has no association with technical efficiency and reallocation of the firms. However, when we include the interaction of the foreign bank dummy and the external dependence measure (in columns (4) and (6), respectively), we find that the coefficient of the interaction term is positive and statistically significant in the technical efficiency regression but not reallocation regression. Therefore, the interacting effect of opening up a region for foreign bank entry and external dependence is reflected in technical efficiency. Put it another way, when a region is opened up for foreign bank entry, industries that are more dependent on external finance display larger increases in technical efficiency, relative to those industries that are less dependent on external finance. One potential problem for the specifications in (5) and (6) is that uncontrolled region-year or industry-year specific shocks may be correlated with both the foreign bank entry dummy and the outcome variables. For example, the schedule of foreign bank deregulation may not be randomly determined but may depend on the productivity growth prospect of firms in different regions or industries. In unreported regressions, we include region and industry-year fixed effects in the regressions. The regression results are similar to the baseline regression results, which suggests that omitted variables that vary at the industry-year level are less of a concern. However, since the foreign bank entry dummy only varies at the region-year level, it is not possible to include region-year fixed-effects. To address this concern, we follow the spirit of Bertrand et al. (2003) by including the mean of the outcome variable for each region-year (excluding the interested industry itself from the mean), denoted by y¯ rt (− j) , and that for each industry-year (excluding the interested region itself from the mean), denoted by y¯ jt (−r) , as extra covariates to control for region-year specific and industry-year specific shocks, respectively. In other words, we estimate the following regressions:

yr jt = αr + α j + αt + β1 F Brt + y¯ rt (− j) + y¯ jt (−r) + εr jt .

(7)

yr jt = αr + α j + αt + β1 F Brt + β2 (F Brt × ED j ) + y¯ rt (− j) + y¯ jt (−r) + εr jt .

(8)

In Panel B of Table 4, we report the regression results of (7) and (8) for the different outcome variables. We can see that, adding these extra region-year and industry-year specific controls gives us the same results. Based on the regression results reported in Panel B of Table 4, we consider industries at various percentiles of the external dependence measure (5th, 10th, 25th, 50th, 75th, 90th, and 95th) and compute the marginal effects of opening up a region for foreign bank entry. The results are reported in Table 5. We find that opening up a region for foreign bank entry has a negative and statistically significant effect on the aggregate productivity growth for industries at the 5th and 10th percentiles of the external dependence distribution; however, the effect becomes positive and statistically significant for industries at the 90th and 95th percentiles of the external dependence distribution. The results in Table 5 indicate that there is a positive relationship between the foreign bank dummy and aggregate productivity growth in industries in the top end of the external dependence distribution; the relationship becomes negative in industries in the bottom end of the external dependence distribution. These results can be rationalized in the context of the ambiguous relationship between banking market structure and industry growth. Recall that banking competition can be good for growth because it results in higher credit availability so that firms can benefit. The positive effect should be stronger for firms in high ED industries; those firms in low ED industries get smaller benefits from the extra credit in the financial market. On the other hand, banking competition can also be bad for growth because banking market power is needed to encourage banks to form lending relationships with firms. Therefore, firms should suffer under a higher degree of banking competition. However, firms in high ED industries can suffer less (or can even benefit) because the banks’ returns for the formation of lending relationship with Please cite this article as: T.-k. Lai et al., WTO accession, foreign bank entry, and the productivity of Chinese manufacturing firms, Journal of Comparative Economics (2015), http://dx.doi.org/10.1016/j.jce.2015.06.003

ARTICLE IN PRESS

JID: YJCEC

[m3Gsc;July 21, 2015;22:30]

T.-k. Lai et al. / Journal of Comparative Economics 000 (2015) 1–17

11

Table 6 Firm-level regression results: R&D activities. (1) All

(2) Large

(3) Small

(4) Old

(5) Young

Firm fixed-effects Industry fixed-effects Year fixed-effects

0.005 (0.016) 0.123∗∗∗ (0.041) −0.029∗∗∗ (0.007) −0.048∗∗∗ (0.016) Yes Yes Yes

0.010 (0.020) 0.180∗∗∗ (0.048) −0.001 (0.008) −0.062∗∗∗ (0.022) Yes Yes Yes

0.041 (0.036) −0.077 (0.094) −0.063∗∗∗ (0.020) −0.028 (0.031) Yes Yes Yes

0.006 (0.020) 0.146∗∗∗ (0.052) −0.024∗∗ (0.011) −0.027 (0.023) Yes Yes Yes

−0.003 (0.070) 0.039 (0.178) −0.046∗∗ (0.023) −0.044 (0.060) Yes Yes Yes

Observations Adjusted R2

958935 0.349

479744 0.378

479191 0.308

525508 0.428

433427 0.306

FB FB × ED Log assets Long-term debt/assets

Note: Dependent variable is R&D expenses divided by assets × 100. Column (1) includes all available observations. Column (2) includes firms with total assets greater than the median. Column (3) includes firms with total assets less than the median. Column (4) includes firms with age at least three years. Column (5) includes firms with age less than three years. Robust standard errors are reported in parentheses. ∗ significance at 10% level; ∗∗ significance at 5% level; ∗∗∗ significance at 1% level.

these firms should be higher. To the extent that opening up a region is associated with higher competition in the banking sector, ultimately the questions about “whether banking competition (associated with foreign bank entry) is good for industry growth” and “whether such a relationship depends on the external dependence of the industry” are empirical. One possibility is that on average the two opposite forces cancel out and we do not see an average effect. For firms in low ED industries, the effect from the “banking competition is bad” story dominates so that we have an overall negative effect; on the contrary, for firms in high ED industries, the effect from the “banking competition is good” story dominates so that we have an overall positive effect. This is exactly what we find in our empirical analysis. Why is technical efficiency but not reallocation affected? We provide one explanation as follows. Foreign bank entry seems to be related to higher banking competition and more abundant credit in the market. When credit is more abundant, firms (irrespective of their dependence on external finance) become less financially constrained. Relative to firms in a low external dependence industry, firms in a high external dependence industry should require more funds for capital intensive projects such as R&D activities. Such projects could not be implemented previously due to financial constraints. However, once this constraint is relaxed, these firms can make use of the extra funds to implement those projects. Therefore, we should expect that technical efficiency for the firms in the high external dependence industries should be higher. We provide further firm-level evidence about whether R&D activities increase for firms in financially dependent industries after foreign bank entry is allowed by estimating the following firm-level regression:

yir jt = αi + α j + αt + β1 F Brt + β2 (F Brt × ED j ) + Xir jt δ + εi jrt ,

(9)

where i is a firm and yirjt is R&D expenses (divided by assets × 100), Xirjt contains log of total assets and the debt ratio (long-term debt divided by total assets). In the regression, we control for firm, industry, and year fixed-effects.20 We estimate the above regression using different samples. In the baseline regression, we use all available observations. We also re-estimate the regression separately for large and small firms, and for old and young firms. Large (small) firms are defined as those with total assets larger (smaller) than the median. Old (young) firms are defined as those with age at least (below) three years. The regression results are reported in Table 6. We find that the coefficient estimates of the interactions between the foreign bank entry dummy and th0e external dependence measure are positive in all specifications and are statistically significant when the regression sample includes all firms (column (1)), large firms only (column (2)) and old firms only (column (4)). Regarding why reallocation is not affected, it is possible that higher banking competition also reduces the profit for banks investing in new relationships with firms. Banks therefore tend to lend money to existing clients who may or may not have projects with higher returns. A common situation in China is that a state-owned bank has a state-owned firm as a client. Both of them are expected to be less efficient than their respective non-state-owned counterparts. As a result, ex ante, bank-firm relationship can increase or decrease reallocation; and it is unclear whether opening up a region for foreign bank entry (together with external dependence) can affect reallocation.

20 Industry fixed-effects are included because certain firms change sectors during the sample period. In unreported regression results, we find that excluding sector-switchers do not affect the main results.

Please cite this article as: T.-k. Lai et al., WTO accession, foreign bank entry, and the productivity of Chinese manufacturing firms, Journal of Comparative Economics (2015), http://dx.doi.org/10.1016/j.jce.2015.06.003

ARTICLE IN PRESS

JID: YJCEC 12

[m3Gsc;July 21, 2015;22:30]

T.-k. Lai et al. / Journal of Comparative Economics 000 (2015) 1–17 Table 7 Characteristics of cities opened up before the end of 2005 and other cities, as of 1998. Obs.

Mean

S.D.

Min.

1st Q.

Median

3rd Q.

Max.

213 213 213 213

15.035 8.846 3.686 2.200

0.668 0.583 2.003 1.004

12.876 7.636 −2.120 0.230

14.676 8.476 2.482 1.587

15.081 8.745 3.825 2.326

15.501 9.158 5.013 2.880

17.236 11.631 8.199 5.390

20 20 20 20

15.517 9.756 6.604 2.002

0.860 0.647 1.117 0.912

13.451 8.564 3.541 0.646

15.422 9.407 6.092 1.037

15.644 9.616 6.827 2.044

15.915 10.099 7.324 2.589

17.236 11.631 8.199 3.639

193 193 193 193

14.985 8.752 3.383 2.221

0.627 0.489 1.825 1.013

12.876 7.636 −2.120 0.230

14.630 8.443 2.371 1.596

15.038 8.694 3.568 2.378

15.455 9.054 4.636 2.891

16.160 10.125 7.035 5.390

Panel A: all cities Log population Log GDP per capita Log utilized foreign capital (million USD) % emp. in banking and insurance sector

Panel B: cities opened up before the end of 2005 Log population Log GDP per capita Log utilized foreign capital (million USD) % emp. in banking and insurance sector Panel C: other cities Log population Log GDP per capita Log utilized foreign capital (million USD) % emp. in banking and insurance sector

4.2. Robustness checks Differences between “opened regions” and “other regions?”. As discussed earlier, after China’s accession to WTO in December 2001, foreign banks can enter 20 Chinese cities (the “opened regions”) in phases, and all geographic restrictions were removed by the end of 2006. Since the outcome variables are defined over the period between 1998 and 2006, the foreign bank entry dummy therefore takes a value of 0 for all the regions not opened up until the end of 2006. An implicit identifying assumption is that the timing of the opening for foreign bank entry is random, conditional on other observable characteristics of the cities. However, it is likely that the timing is endogenously determined and the “other regions” may not form a good control group for the “opened regions” in the regression analysis. When all the regions are included in the baseline regressions, the regression results may only pick up the productivity or other unobserved differences between the “opened regions” and “other regions” rather than the impact of allowing foreign banks to enter. To examine whether the timing of the opening up for foreign bank entry is related to the observable characteristics of different regions, we obtain some city-level variables from the China city statistics yearbook. These variables include: population, GDP per capita, utilized foreign capital (to proxy for foreign investment in the city), and the shares of employees in the banking and insurance sector. They are measured in 1998, i.e., before our sample period begins and before China’s WTO accession. Table 7 shows some statistics of these variables. The sample include 213 cities with non-missing values in these variables. These cities are at prefecture-level or above, where 20 of them are “opened regions” we considered earlier and the rest are located in “other regions.”21 We then estimate a simple probit model where the dependent variable takes a value of 1 for the “opened regions” and 0 for the other cities, and the explanatory variables are those in Table 7. The regression results, in terms of marginal effects, are reported in Table 8. These results show that, before China’s WTO accession, the “opened regions” are larger in terms of population, GDP per capita and foreign investment but have a smaller share of employees in the banking and insurance sector. The main takeaway from the above exercise is that we may not take the choice of the “opened regions” as completely exogenous. To address this problem, we conduct two robustness checks: In the first robustness check, we restrict our attention to the “opened regions” and re-estimate the regressions in (7) and (8). The results of these regressions are reported in Panel A of Table 9. These regression results are similar to those reported in Panel B of Table 4. In particular, the coefficients of the interaction terms between the foreign bank dummy and external dependence in the aggregate productivity growth regression and technical efficiency regression are positive and statistically significant. One may argue that the above exercise still does not adequately address the potentially endogenous selection of “opened regions,” because the results can still be driven by other unobserved factors not controlled for in the regressions. Therefore, in the second robustness check, we use propensity score matching to identify a control group of “other regions” (which have comparable pre-WTO accession characteristics as those “opened regions”) and re-estimate the main regressions using these regions and “opened regions” as the regression sample.22 Panel B of Table 9 reports the regression results. We also find that the

21

Strictly speaking, these other cities only form part of the “other regions.” In particular, this control group includes Dongguan, Handan, Hangzhou, Jieyang, Quanzhou, Shijiazhuang, Taiyuan, Taizhou, Tangshan, Weihai, Wuxi, Yantai, Zhenjiang, and Zibo. These cities have comparable observable characteristics as those of Chengdu, Chongqing, Dalian, Fuzhou, Jinan, Kunming, Nanjing, Ningbo, Qingdao, Shantou, Shenyang, Wuhan, Xi’an, Xiamen, and Zhuhai. In unreported regressions, we run the same probit model to see whether the observable characteristics are associated with the status of being an “opened region” within this 29 cities. We find that the coefficients of the observable characteristics have statistically insignificant coefficients. Note that we cannot find cities with comparable characteristics as those of Beijing, Guangzhou, Shanghai, Shenzhen, and Tianjin. These five “opened regions” are excluded in this robustness exercise. 22

Please cite this article as: T.-k. Lai et al., WTO accession, foreign bank entry, and the productivity of Chinese manufacturing firms, Journal of Comparative Economics (2015), http://dx.doi.org/10.1016/j.jce.2015.06.003

ARTICLE IN PRESS

JID: YJCEC

[m3Gsc;July 21, 2015;22:30]

T.-k. Lai et al. / Journal of Comparative Economics 000 (2015) 1–17

13

Table 8 Relationship between city characteristics and whether the city is opened before the end of 2005: probit regression marginal effects. (1)

(2)

(3)

(4)

(5)

0.097∗∗ (0.043)

Log population

0.178∗∗∗ (0.029)

Log GDP per capita

0.077∗∗∗ (0.009)

Log utilized foreign capital (million USD)

−0.019 (0.019)

% emp. in banking and insurance sector Observations Log likelihood

213 −60.821

213 −41.506

213 −34.226

213 −65.897

0.096∗∗∗ (0.025) 0.146∗∗∗ (0.039) 0.027∗∗ (0.013) −0.018∗ (0.010) 213 −24.365

Note: Dependent variable takes a value of 1 for regions opened up before the end of 2005. Covariates are measured in 1998. Robust standard errors are reported in parentheses. ∗ significance at 10% level; ∗∗ significance at 5% level; ∗∗∗ significance at 1% level.

Table 9 Robustness checks: controlling for differences between “opened regions” and “other regions”. (1)

Dependent variable: FB

(2)

Dependent variable: FB

−0.002 (0.002)

Yes Yes Yes 4100 0.083

(5)

−0.019∗∗∗ (0.005) 0.038∗∗ (0.017) Yes Yes Yes 4100 0.084

0.005 (0.007)

Yes Yes Yes 4100 0.101

−0.009 (0.008) 0.031∗∗ (0.015) Yes Yes Yes 4100 0.102

−0.003 (0.009)

Yes Yes Yes 5518 0.109

−0.021 (0.015) 0.041∗ (0.023) Yes Yes Yes 5518 0.109

−0.003 (0.010)

Yes Yes Yes 5518 0.129

(6)

RE −0.011∗∗ (0.005)

−0.014∗∗∗ (0.005) 0.006 (0.012) Yes Yes Yes 4100 0.036

Yes Yes Yes 4100 0.036

Panel B: propensity score matching APG TE

FB × ED Region fixed-effects Industry fixed-effects Year fixed-effects Observations Adjusted R2

(4)

Panel A: focusing on “opened regions” only APG TE

FB × ED Region fixed-effects Industry fixed-effects Year fixed-effects Observations Adjusted R2

(3)

RE ∗

−0.023 (0.013) 0.046∗∗ (0.019) Yes Yes Yes 5518 0.130

0.001 (0.004)

Yes Yes Yes 5518 0.050

0.003 (0.005) −0.005 (0.011) Yes Yes Yes 5518 0.050

Note: Region-year and industry-year specific shocks are also controlled for. Standard errors are clustered at region and industry levels, and are reported in parentheses. ∗ significance at 10% level; ∗∗ significance at 5% level; ∗∗∗ significance at 1% level.

coefficients of the interaction terms between the foreign bank dummy and external dependence in the aggregate productivity growth regression and technical efficiency regression are positive and statistically significant. Taking these two sets of results together, we consider that our baseline results are unlikely to be driven by differences between the “opened regions” and the “other regions.” Alternative production function estimation method. In the baseline regressions, we construct our outcome variables based on the production function estimated using Ackerberg et al. (2006). As a robustness check, we use another production function estimation method, proposed by Olley et al. (1996), to construct the outcome variables and re-estimate the regressions in (7) and (8). The regression results, presented Table 10, are also similar, suggesting that our results are not specific to a particular production function estimation method. Taking entry and exit into account. In the baseline results, we only use continuing firms to compute aggregate productivity growth and its components. In other words, new entrants and exiters were excluded in the analysis, which may bias our baseline result Please cite this article as: T.-k. Lai et al., WTO accession, foreign bank entry, and the productivity of Chinese manufacturing firms, Journal of Comparative Economics (2015), http://dx.doi.org/10.1016/j.jce.2015.06.003

ARTICLE IN PRESS

JID: YJCEC 14

[m3Gsc;July 21, 2015;22:30]

T.-k. Lai et al. / Journal of Comparative Economics 000 (2015) 1–17 Table 10 Robustness checks: based on Olley et al. (1996). (1) Dependent variable: FB

(3)

(4)

(5)

TE

(6) RE

0.001 (0.005)

−0.014∗∗∗ (0.004) 0.034∗∗∗ (0.008)

0.001 (0.008)

−0.011 (0.008) 0.027∗∗ (0.012)

0.001 (0.005)

−0.002 (0.006) 0.007 (0.006)

Yes Yes Yes 9370 0.107

Yes Yes Yes 9370 0.107

Yes Yes Yes 9370 0.118

Yes Yes Yes 9370 0.118

Yes Yes Yes 9370 0.019

Yes Yes Yes 9370 0.019

FB × ED Region fixed-effects Industry fixed-effects Year fixed-effects Observations Adjusted R2

(2) APG

Note: Region-year and industry-year specific shocks are also controlled for. Standard errors are clustered at region and industry levels, and are reported in parentheses. ∗ significance at 10% level; ∗∗ significance at 5% level; ∗∗∗ significance at 1% level.

Table 11 Robustness checks: taking firm entry and exit into account. (1)

(2) APG

Dependent variable: FB

RE

(4)

−0.001 (0.005)

−0.012 (0.008) 0.024∗ (0.013)

−0.009 (0.009)

−0.005 (0.011) −0.008 (0.017)

Yes Yes Yes 9030 0.068

Yes Yes Yes 9030 0.068

Yes Yes Yes 9030 0.040

Yes Yes Yes 9030 0.040

FB × ED Region fixed-effects Industry fixed-effects Year fixed-effects Observations Adjusted R2

(3)

Note: Region-year and industry-year specific shocks are also controlled for. Standard errors are clustered at region and industry levels, and are reported in parentheses. ∗∗ significance at 5% level; ∗∗∗ significance at 1% level. ∗ significance at 10% level;

because entry and exit are generally regarded as an extreme form of resource reallocation. According to Brandt et al. (2012b), this extreme form of resource reallocation, net entry of firms, has contributed over half of the growth in manufacturing sector during 1998 and 2007. Since the banking sector plays an important role in the allocation of resources, it is meaningful to ask whether to some degree the significant growth from net entry has been due to the restructuring in banking sector after foreign bank entry. To examine whether our main results change once we take into account the new entrants and exiters, we first compute similar aggregate productivity growth terms for entrants and exiters:

APGNrjt =

  i∈N

APGXrjt =









Dvir jt svir jt log Lir jt + (1 − svir jt ) log Kir jt Dvir jt svir jt log Lir jt + (1 − svir jt ) log Kir jt

,

(10)

,

(11)

i∈X

We then define the “net entry effect” as the difference between the above terms: NEr jt = APGN − APGXrjt−1 . Then we re-define r jt the aggregate productivity growth and reallocation to take into account the net entry effect, as follows:

APGr jt = APGr jt + NEr jt ,

(12)

REr jt = APGr jt − T Er jt .

(13)

Note that the technical efficiency term is unaffected by entry and exit. In Table 11, we show the regression results using the above two outcome variables. Essentially, after considering the net entry effect, in the aggregate productivity growth regression, we still observe a positive and statistically significant (at 10%) coefficient for the interaction between the foreign bank entry dummy and the external dependence measure; on the other hand, in the reallocation regression, the coefficient of the interaction term remains statistically insignificant. Please cite this article as: T.-k. Lai et al., WTO accession, foreign bank entry, and the productivity of Chinese manufacturing firms, Journal of Comparative Economics (2015), http://dx.doi.org/10.1016/j.jce.2015.06.003

ARTICLE IN PRESS

JID: YJCEC

T.-k. Lai et al. / Journal of Comparative Economics 000 (2015) 1–17

[m3Gsc;July 21, 2015;22:30] 15

Table 12 Aggregate productivity growth by size and by age. (1)

(2)

Panel A: by size Large FB

0.003 (0.007)

FB × ED Region fixed-effects Industry fixed-effects Year fixed-effects Observations Adjusted R2

Yes Yes Yes 9342 0.073

−0.010 (0.009) 0.029∗∗∗ (0.010) Yes Yes Yes 9342 0.074

Panel B: By age Old FB

0.001 (0.006)

FB × ED Region fixed-effects Industry fixed-effects Year fixed-effects Observations Adjusted R2

Yes Yes Yes 9358 0.077

−0.015∗∗ (0.006) 0.034∗∗∗ (0.011) Yes Yes Yes 9358 0.078

(3)

(4)

Small 0.003 (0.014)

Yes Yes Yes 9258 0.046

0.001 (0.017) 0.004 (0.017) Yes Yes Yes 9258 0.045

Young −0.002 (0.015)

Yes Yes Yes 8478 0.036

−0.018 (0.013) 0.036 (0.029) Yes Yes Yes 8478 0.036

Note: Dependent variable is aggregate productivity growth (APG). Large (small) firms are those with total assets greater (less) than the median. Old (young) firms are those with age at least (less than) three years. Region-year and industry-year specific shocks are also controlled for. Standard errors are clustered at region and industry levels, and are reported in parentheses. ∗ significance at 10% level; ∗∗ significance at 5% level; ∗∗∗ significance at 1% level.

4.3. Are firms with different sizes and ages affected differently? The baseline analysis assumes homogeneous impact on all firms. However, as mentioned in the Section 1, foreign bank entry may have different effects on the access to credit by firms of different sizes. Therefore, the impact of foreign bank entry in China may be different for larger (and older) firms and for smaller (and younger) firms. To examine whether the aggregate productivity growth of firms with different sizes and ages is affected differently, we reconstruct the aggregate productivity growth measures as follows. We first classify the firms as large (small) if their total assets are above (below) the median; we then follow the same procedures as in the baseline analysis to construct the aggregate productivity growth measures. Besides, we define old firms as those which have been in operation for three years or more and young firms as the remaining firms; again, we follow the same procedures as in the baseline analysis to construct the aggregate productivity growth measures. We re-estimate the baseline regression in (8) using these newly-constructed aggregate productivity measures. Panel A of Table 12 reports the regression results by firm size and Panel B of Table 12 reports the regression results by firm age. We find that only larger and older firms in industries with higher external dependence have significantly larger aggregate productivity growth when foreign bank entry is allowed. These results seem to suggest that allowing foreign bank entry only helps existing and larger firms that are financially constrained but may not solve the financing problems of newer and smaller firms. One potential reason why distortions at the extensive margin have not declined after foreign banks are allowed to enter is about the nature of banking sector competition in China.23 In addition to the differential impact of foreign bank entry on access to credit by firms of different sizes, we note that the literature about banking market competition in China focuses on competition among such large banks as state-owned banks and foreign banks. While competition among large banks has increased, these banks mainly serve larger firms in larger cities. Before China’s accession to the WTO, especially before 1998, there were many small banks (such as urban credit cooperatives and rural credit cooperatives) competing to provide services to smaller firms. These banks were later restructured and consolidated into larger city commercial banks (see the discussion in, e.g., Garcá-Herrero and Alicia, 2006). As a result, competition may have increased in larger cities for services to larger firms but decreased for smaller firms, so that distortions at the extensive margin may remain even after foreign bank entry.

23

We thank an anonymous referee for raising this point.

Please cite this article as: T.-k. Lai et al., WTO accession, foreign bank entry, and the productivity of Chinese manufacturing firms, Journal of Comparative Economics (2015), http://dx.doi.org/10.1016/j.jce.2015.06.003

ARTICLE IN PRESS

JID: YJCEC 16

[m3Gsc;July 21, 2015;22:30]

T.-k. Lai et al. / Journal of Comparative Economics 000 (2015) 1–17

Recent studies on financial frictions and capital misallocation by Buera et al. (2011) and Midrigan et al. (2014) find that the distortion at the extensive margin accounts for the major share of aggregate TFP loss due to financial frictions. Our results are consistent with these studies, because we find that the entry of foreign banks does not reduce frictions at the extensive margin and therefore the impact on aggregate TFP growth is small. The small aggregate impact of foreign bank entry may also be related to the fact that it is one of the many dimensions of the financial sector reforms. Particularly relevant in the case of China’s banking sector, the deposit interest rate is still officially chosen, which may limit the activities of entering foreign banks and therefore the overall impact of this reform. Without complementary reforms, this reform alone may only increase banking sector competition and reduce financial frictions marginally (especially at the extensive margin). Therefore, our results call for deeper domestic reforms in the financial sector to reduce capital misallocation in China.

5. Concluding remarks In this paper, we examine the relationship between foreign bank entry and the productivity growth of the Chinese manufacturing firms after China’s accession to the World Trade Organization in 2001. Using the firm-level manufacturing data obtained from the National Bureau of Statistics of China, we find that on average, opening up a region for foreign bank entry is not associated with aggregate productivity growth; but industries more dependent on external finance grow faster, relative to those less dependent on external finance, after a region is opened up for foreign bank entry. These results are due to changes in technical efficiency rather than reallocation or net entry. We also find that allowing foreign bank entry tends to help the growth of existing and larger firms that are in financially dependent industries, suggesting that this policy reform seems to have little effect in lowering financial frictions that affect the extensive margins the most. Further research is needed to understand how financial liberalization (and/or other reforms) may help reduce capital misallocation significantly in China.

Appendix A. The relationship between foreign bank presence and banking market characteristics In this appendix, we present some descriptive statistics about the relationship between foreign bank presence and banking market characteristics. Specifically, we use country-level data obtained from Bankscope between 1998 and 2007 to construct the following variables: Total bank loans, total bank assets, concentration ratios in terms of bank loans and bank assets, HerfindahlHirschman Index in terms of bank loans and bank assets, and the H-statistic proposed by Panzar et al. (1987). We relate these variables with the shares of opened cities in a given year, defined as follows:

Share of opened cities in year t =

Number of cities opened for foreign bank entry in year t , 213

where 213 is the total number of prefectural-level or above cities in Table 7. This variable is used to proxy for the extent of foreign bank presence in China. In Table A, we show the correlation matrix for these variables. We can see that the share of opened cities is significantly associated with more bank loans, more bank assets, and higher competition (measured by a lower concentration ratio, a lower Herfindhal-Hirschman Index, and a higher H-statistic).

Table A Correlations between shares of opened cities and other banking market characteristics. (1) (1) Shares of opened cities

(2)

(3)

(4)

(5)

(6)

(7)

1.000

(2) Total bank loans

0.719 (0.019)

1.000

(3) Total bank assets

0.774 (0.009)

0.994 (0.000)

1.000

(4) CR4 (bank loans)

−0.672 (0.033)

−0.992 (0.000)

-0.983 (0.000)

1.000

(5) CR4 (bank assets)

−0.744 (0.014)

−0.991 (0.000)

-0.986 (0.000)

0.991 (0.000)

1.000

(6) HHI (bank loans)

−0.627 (0.052)

−0.982 (0.000)

-0.966 (0.000)

0.996 (0.000)

0.985 (0.000)

1.000

(7) HHI (bank assets)

−0.683 (0.030)

−0.985 (0.000)

−0.970 (0.000)

0.991 (0.000)

0.995 (0.000)

0.992 (0.000)

1.000

0.803 (0.005)

0.645 (0.044)

0.681 (0.030)

−0.647 (0.043)

−0.688 (0.028)

−0.633 (0.049)

−0.659 (0.038)

(8) H-statistic

(8)

1.000

Note: p-values are in parentheses.

Please cite this article as: T.-k. Lai et al., WTO accession, foreign bank entry, and the productivity of Chinese manufacturing firms, Journal of Comparative Economics (2015), http://dx.doi.org/10.1016/j.jce.2015.06.003

JID: YJCEC

ARTICLE IN PRESS T.-k. Lai et al. / Journal of Comparative Economics 000 (2015) 1–17

[m3Gsc;July 21, 2015;22:30] 17

References Ackerberg, Daniel A., Caves, Kevin, Frazer, Garth, 2006. Structural Identification of Production Functions. Working Paper. Allen, Franklin, Qian, Jun, Qian, Meijun, 2008. China’s financial system: past, present, and future. In: Brandt, Loren, Rawski, Thomas G. (Eds.), In: China’s Great Economic Transformation. Cambridge University Press. Berger, Allen N., Hasan, Iftekhar, Zhou, Mingming, 2009. Bank ownership and efficiency in China: what will happen in the world’s largest nation? Journal of Banking and Finance 33, 113–130. Berger, Allen N., Miller, Nathan H., Petersen, Mitch A., Rajan, Raghuram G., Stein, Jeremy C., 2005. Does function follow organizational form? Evidence from the lending practices of large and small banks. Journal of Financial Economics 76, 237–269. Bertrand, Marianne, Mullainathan, Sendhil, 2003. Enjoying the quiet life? Corporate governance and managerial preferences. Journal of Political Economy 111, 1043–1075. Brandt, Loren, Van Biesebroeck, Johannes, Wang, Luhang, Zhang, Yifan, 2012a. WTO accession and firm-level productivity in Chinese manufacturing. CEPR Discussion Paper No. 9166. Brandt, Loren, Van Biesebroeck, Johannes, Zhang, Yifan, 2012b. Creative accounting or creative destruction? Firm-level productivity growth in Chinese manufacturing. Journal of Development Economics 97, 339–351. Buera, Francisco J., Kaboski, Joseph p., Shin, Yongseok, 2011. Finance and development: a tale of two sectors. American Economic Review 101, 1964–2002. Cetorelli, Nicola, Gambera, Michele, 2001. Banking market structure, financial dependence and growth: international evidence from industry data. Journal of Finance 56, 617–648. Chong, Terence Tai-Leung, Lu, Liping, Ongena, Steven, 2013. Does banking competition alleviate or worsen credit constraints faced by small- and medium-sized enterprises? evidence from China. Journal of Banking and Finace 37, 3412–3424. Claessens, Stijn, Demirgüç-Kunt, Aslı, Huizingac, Harry, 2001. How does foreign entry affect domestic banking markets? Journal of Banking and Finance 25, 891–911. Claessens, Stijn, Laeven, Luc, 2005. Financial dependence, banking sector competition, and economic growth. Journal of the European Economic Association 3, 179–207. Clarke, George, Cull, Robert, Martinez Peria, Maria Soledad, 2006. Foreign bank participation and access to credit across firms in developing countries. Journal of Comparative Economics 34, 774–795. Clarke, George, Cull, Robert, Martinez Peria, Maria Soledad, Sánchez, Susana M., 2003. Foreign bank entry: experience, implications for developing economies, and agenda for future research. World Bank Research Observer 18, 25–59. Cole, Rebel A., Goldberg, Lawrence G., White, Lawrence J., 2004. Cookie cutter vs. character: the micro structure of small business lending by large and small banks. Journal of Financial and Quantitative Analysis 39, 227–251. Cull, Robert, Xu, L. Colin, 2013. Job growth and finance: are some financial institutions better suited to the early stages of development than others? World Bank Economic Review 27, 542–572. Demirgüç-Kunt, Aslı, Feyen, Erik, Levine, Ross, 2013. The evolving importance of banks and securities markets. World Bank Economic Review 27, 476–490. Detragiache, Enrica, Tressel, Thierry, Gupta, Poonam, 2008. Foreign banks in poor countries: theory and evidence. Journal of Finance 63, 2123–2160. Garcá-Herrero, Alicia, Gavilá, Sergio, Santabárbara, Daniel, 2006. China’s banking reform: an assessment of its evolution and possible impact. CESifo Economic Studies 52, 304–363. Gormley, Todd A., 2010. The impact of foreign bank entry in emerging markets: evidence from India. Journal of Financial Intermediation 19, 26–51. Hsieh, Chang-Tai, Klenow, Peter J., 2009. Misallocation and manufacturing TFP in China and India. Quarterly Journal of Economics 124, 1403–1448. Inklaar, Robert, Koetter, Michael, Noth, Felix, 2012. Who’s afraid of big bad banks? Banking Competition, SME, and Industry Growth. Working Paper. Jiang, Chunxia, Yao, Shujie, Zhang, Zongyi, 2009. The effects of governance changes on bank efficiency in China: a stochastic distance function approach. China Economic Review 20, 717–731. Lin, Huidan, 2011. Foreign bank entry and firms’ access to bank credit: evidence from China. Journal of Banking and Finance 35, 1000–1010. Lin, Xiaochi, Zhang, Yi, 2009. Bank ownership reform and bank performance in China. Journal of Banking and Finance 33, 20–29. Mian, Atif, 2006. Distance constraints: the limits of foreign lending in poor countries. Journal of Finance 61, 1465–1505. Midrigan, Virgiliu, Xu, Daniel Yi, 2014. Finance and misallocation: evidence from plant-level data. American Economic Review 104, 422–458. National Bureau of Statistics China, 2007. The GDP accounting method for economic census year of China. (Zhong Guo Jing Ji Pu Cha Nian Du Guo Nei Sheng Chan Zong Zhi He Suan Fang Fa). China Statistics Press, Beijing (in Chinese). National Bureau of Statistics China, 2008. The GDP accounting method for non-census year of China. (Zhong Guo Fei Jing Ji Pu Cha Nian Du Guo Nei Sheng Chan Zong Zhi He Suan Fang Fa). China Statistics Press, Beijing (in Chinese). National Bureau of Statistics China, 2010. The GDP accounting method for non-census year of China (First Amendment). (Zhong Guo Fei Jing Ji Pu Cha Nian Du Guo Nei Sheng Chan Zong Zhi He Suan Fang Fa, Di Yi Ci Xiu Ding). Unpublished Manual of National Burean of Statistics, (in Chinese). Olley, G. Steven, Pakes, Ariel, 1996. The dynamics of productivity in the telecommunications equipment. Econometrica 64, 1263–1297. Pagano, Marco, 1993. Financial markets and growth: an overview. European Economic Review 37, 613–622. Panzar, John C., Rosse, Jammes N., 1987. Testing for “monopoly” equilibrium. Journal of Industrial Economics 35, 443–456. Petersen, Mitchell A., Rajan, Raghuram G., 1995. The effect of credit market competition on lending relationships. Quarterly Journal of Economics 110, 407–443. Petrin, Amil, Levinsohn, James, 2012. Measuring aggregate productivity growth using plant-level data. RAND Journal of Economics 43, 705–725. Qian, Zhenjie, Zhu, Xiaodong, 2011. Measuring factor income share using china annual survey of industry. Working Paper. Rajan, Raghuram, Zingales, Luigi, 1998. Financial dependence and growth. American Economic Review 88, 559–586. Shen, Yan, Shen, Minggao, Xu, Zhong, Bai, Ying, 2009. Bank size and small- and medium-sized enterprise (sme) lending: evidence from China. World Development 37, 800–811. Wang, Songqi, 2007. Annual Report on the Competitiveness of China’s Commercial Banks. (Zhong Guo Jing Rong Fa Zhan Bao Gao). Social Science Academic Press, Beijing (in Chinese). Xu, Ying, 2011. Towards a more accurate measure of foreign bank entry and its impact on domestic banking performance: the case of China. Journal of Banking and Finance 35, 886–901. Zhang, Jin, Wang, Lanfang, Wang, Susheng, 2012. Financial development and economic growth: recent evidence from China. Journal of Comparative Economics 40, 393–412.

Please cite this article as: T.-k. Lai et al., WTO accession, foreign bank entry, and the productivity of Chinese manufacturing firms, Journal of Comparative Economics (2015), http://dx.doi.org/10.1016/j.jce.2015.06.003