A Tale of Two Markets: Interbank and Exchange Repo Markets in China
Abstract
Repurchase agreements, or repos, represent the largest, most important, and fastest growing money market in China. China repos trade in the interbank market as well as the stock exchanges. In light of the tremendous growth in China’s repo markets, emergence of shadow banking activities, aggressive interest rate liberalization, and increasing interest in China’s repo markets by foreign participants, this study investigates the behaviors and drivers of the interbank and exchange repo rates, as well as their spread. Using ten years of monthly data from October 2006 to October 2016, this paper documents continued segmentation between the two repo markets, and uncover the impact of shadow banking activities, monetary policies, stock IPO activities, bond issuance, and quarter-end effect on the rate dynamics in these two partially segmented repo markets.
JEL Classifications: G15, G21, G28
Keywords: Repo; Segmentation; Liquidity; Shadow Banking; Monetary Policy
A Tale of Two Markets: Interbank and Exchange Repo Markets in China
1. Introduction The repo market is a critical component of the modern global financial system. In a repurchase agreement, or repo, one counterparty (i.e., the borrower, collateral provider, repo seller) obtains financing by selling a financial security to the other counterparty (i.e., the lender, cash provider, repo buyer), and simultaneously agreeing to repurchase the security at the future date (usually in one day or a week) at a higher price. The difference between the selling price and repurchase price of the security reflects the implicit interest cost of the borrowing. In the case of a default by the borrower, the lender can quickly sell the security to recover from the loss. The security serves as a collateral to mitigate the lender’s credit risk exposure and reduce the borrower’s interest cost. In essence, repo is a short-term, highly liquid, and collateralized debt instrument.1 The repo market channels efficient allocation of short-term funds among banks, nonbank financial institutions, corporations, as well as retail investors, and also serves a major execution platform for central bank open market operations. While the history of the U.S. repo market can be dated back to 1920s, repo financing has been a new phenomenon in China since 1991, and has not fully taken off until 2006. China’s repos trade in the interbank market as well as the national stock exchanges, mirroring the interbank and exchange trading of bonds. The exchange repo market started in 1991 with the exchange providing the counterparty guarantee for all repo buyers and sellers, while the interbank repo market was initiated by the People’s Bank of China (PBOC) in 1997 to insulate deposit-based banks from funding nonbank institutions in equity and real estate investments through the exchange repo market.
1
See Baklanova, Copeland, and McCaughrin (2015) for an excellent reference guide to the U.S. repo market. 1
Porter and Xu (2009) examine the drivers of the 7-day interbank repo rate using daily data from October 2003 to August 2008 and show that the daily repo rate is highly persistent, and strongly driven by PBOC’s benchmark interest rates and calendar-related liquidity factors, but not PBOC’s monetary policy actions such as changes in the required reserve ratio or open market operations. The study is restricted to the interbank money market and focuses on the daily movement. Fan and Zhang (2007) study the segmentation and linkage between the interbank and exchange repo markets in China using weekly data from January 2000 to December 2005. They show that the exchange repo rates were significantly higher than the interbank repo rates, especially during the 2000-2002 period. Although these studies have investigated the drivers of interbank repo rate and the segmentation of the two repo markets, they were based on the earlier period when China’s repo market was still at its infancy. During the last 10 years from 2007 to 2016, annual trading volume of repos in China increased by 12 times to 564 Trillion RMB in the interbank repo market, and by 116 times to 211 Trillion RMB in the exchange repo market. In addition to the much larger market size, the dynamics of participation in the two repo markets have also changed considerably over last ten years. 2 In particular, nonbank financial institutions and enterprises have been permitted to participate in the interbank repo market, which was exclusively for banks at its initiation by the PBOC in 1997. On the other hand, the explosive growth of money market mutual funds, wealth management products, and shadow banking financing in China have led to far more supply and demand of short-term marketbased funding (see McLoughlin and Meredith (2017), Perry Weltewitz (2015), and Elliott, Kroeber and Qiao (2015)). The highly liquid and fully standardized exchange repo market has become an integral part of such market finance system.
2
See Shevlin and Chang (2015) for a thorough description of the differences in operations and trading mechanisms between China’s interbank and exchange repo markets. 2
During the summer of 2013, China experienced a severe liquidity crisis marked with extraordinary rate spikes in the repo market. Highest overnight repo rates in the interbank and exchange repo markets reached 30% and 32% in June 2013, respectively. Major financial outlets (such as Wall Street Journal, CNN Money, Reuters, and Bloomberg) called it as China’s “Funding Crisis” and “Liquidity Crisis.” The liquidity crunch of June 2013 was preceded by the rapid growth of China’s “Shadow Banking System,”3 in which funds were channeled from higher-yield “wealth management products” 4 through “shadow lenders” such as banks’ off-balance-sheet arms or nonbank financial institutions, to loans or investments in areas that were restricted or prohibited by traditional banking. China’s 2013 repo crisis shares some similarity with the “run on repos” during the U.S. financial crisis of 2007-2008,5 which was preceded by the explosive growth of “securitized shadow banking,” where financial institutions leveraged their investments in subprime loans and receivables through securitization6 and repos with MBS and ABS as collaterals. Krishnamurthy, Nagel and Orlov (2014) show that although the use of MBS and ABS as repo collaterals was relatively small in aggregate, the sharp decline in collateral value and the subsequent contraction in repo market have led to the failure of a number of repo dealers and caused systemic risk through the financial system. Unlike the 20072008 repo crisis in the U.S., China’s shadow banking system does not center on securitization, but mainly relies on the use of WMPs as “shadow deposits,” and the repo market as a source of liquidity
According to the Financial Stability Board (2015), “the shadow banking system can broadly be described as credit intermediation involving entities and activities outside of the regular banking system.” Gennaioli, Shleifer and Vishny (2013) describe “shadow banking” as “financial activities occurring outside the regulated banking sector.” 4 Wealth Management Products (WMPs) in China, often considered as “Shadow Deposits,” are sold by banks or nonbank financial institutions as higher-yield non-guaranteed alternatives to traditional bank deposits. According to http://www.chinawealth.com.cn/, China has 25.14 Trillion RMB of WMPs outstanding as of June 2016. 5 The term “run of repos” was used by Gorton and Metrick (2012) to describe the evaporation of liquidity in the market of repos collateralized by subprime mortgage-backed securities (MBS), which collapsed in valuation during the 2007-2008 financial crisis. 6 In the three years prior to the 2007-2008 U.S. financial crisis, more than 80% of the mortgage loans were financed through securitizations in MBS, and other illiquid risky assets (such as home equity loans, credit card receivables, and car loans) were also securitized through Asset-backed Securities (ABS). 3
3
to balance funding imbalance or maturity mismatching between “shadow deposits” and “shadow lending.” While the role of “securitized shadow banking” on the U.S. repo market during the 20072008 financial crisis has been extensively investigated by Gorton (2009), Brunnermeier and Pedersen (2009), Gorton and Metrick (2012), Comotto (2012), and Krishnamurthy, Nagel and Orlov (2014), the impact of shadow banking activities on China’s repo market has not yet been empirically examined. Another important development is the liberalization of interest rates in China. Although repo rates and bond yields are determined by the market, bank deposit and lending rates in China have traditionally been controlled by the PBOC through its setting of benchmark deposit and lending rates (see He and Wang (2011)). The PBOC’s setting of bank deposit and lending benchmark rates has become less binding as China gradually liberalizes interest rates since 2004. In particular, the PBOC removed the lending rate ceiling and deposit rate floor in 2004, lowered the lending rate floor in 2004 and 2012 before completely removing it in 2013, and gradually raised the deposit rate ceiling in 2012 and 2014 before full liberalization in 2015. Liberalization of deposit rates, in particular, allows banks to compete for deposit sources by setting market competitive rates. With tremendous growth in both interbank and exchange repo markets, rising prominence of exchange repo trading, greater pool of eligible nonbank institutional participants in both repo markets, aggressive interest rate liberalization reform, and emergence of the shadow banking system, are these two repo markets still segmented? What drives the repo rates in each of these two markets, and what drives their differences over time? The objective of this study is to thoroughly examine the behaviors and drivers of interbank and exchange repo rates, and to assess the degree of segmentation and integration of these two repo markets. We use monthly data in this study because data on many of the potential driving forces of repo rates, such as shadow banking activities, equity and bond issuance activities, hot money flow and home price growth, are only available on a monthly basis. Our sample
4
covers the ten-year period from October 2006 (when the money market rates of interbank lending, interbank repo and exchange repo are all available in Bloomberg) to October 2016, longest to date over which China repo markets have been studied. Because rate spikes and liquidity shocks are not fully captured in the closing rate alone at either the daily or monthly level, we examine the behaviors and drivers of the interbank and exchange repo rates and their differences using both the monthly closing rate as well as the monthly highest rate.7 We include the 1-day (i.e., overnight) repo rate in the analysis because it is the most actively traded term to maturity in both the interbank and exchange repo markets during the sample period. We also include the 7-day repo in our analysis because it is the second most active repo tenor, the most popular money market benchmark indicator, and the most studied China repo rate by earlier studies (such as Fan and Zhang (2007), and Porter and Xu (2009)). Using ten years of monthly data from October 2006 to October 2016, we find that the average monthly closing rate of 7-day interbank repos exceeds that of exchange repos by 0.683%, which contrasts the average negative spread of -0.298% from 2000 to 2005 from Fan and Zhang (2007). As the exchange repo market grows dramatically in volume and scope of participation over the years, the exchange’s counterparty guarantee and standardization of terms have led to a positive rate spread between the interbank and exchange repo markets during normal market conditions. On the other hand, we also find that the exchange repo market is much more volatile relative to the interbank repo market. The exchange repo rate has an average monthly high of 18.00% (7.32%) for the 1-day (7-day) maturity, which is 13.28% (2.23%) higher than its interbank repo counterpart. This striking difference reveals that the exchange repo market, especially in the 1-day maturity, is much more susceptible to liquidity
7
Rate spikes, often driven by demand shocks, were not sufficiently captured by the daily or monthly closing rates. Both interbank repo rates (from CFETS) and exchange repo rates (from SSE) data used in this study include the closing as well as the intraday highest rates during an interval, which allow us to uncover the rate spikes that occur during the interval. For example, during the liquidity crisis in June 2013, the interbank repo rate closed at 4.86% in June 2013, when the monthly intraday high was 30%. Even for the daily data, the closing rates were 4.31% and 8.89% at the peak of the crisis on June 20 and June 21, when the intraday highs were 30% and 25%, respectively. Research based on both closing and highest repo rates provide much more depth and insight into the behaviors and drivers of the repo rates in China. 5
dry-ups than the interbank repo market. Without large funding source from banks and monetary sterilization from the PBOC, the exchange repo market has much more frequent and sizable rate spikes than the interbank repo market. Although the size of the difference in monthly highest repo rates has declined substantially from first half to second half for the 7-day repos, but not for the overnight repos, indicating stronger and more persistent segmentation in overnight repo market due to limited capital mobility. We investigate the impact of the following factors to identify key drivers of the repo rates and their differences in the interbank and exchange markets: China’s monetary policy variables, such as PBOC’s policy and benchmark rates, change in required reserve ratio, funding injection from open market operations, China’s shadow banking activities, stock IPO activities, bond issuance, hot money flow, home price growth, and quarter-end calendar effect. The monthly interbank and exchange repo rates and their spread, based on monthly closing and highest rates, are used as the dependent variables. Our results show that the PBOC’s policy and benchmark interest rates have a positive and significant effect on the repo rates in both the interbank and exchange repo market, indicating an effective monetary policy transmission mechanism from the PBOC to the money market. On the other hand, the change in bank required reserve ratio has a positive effect on the interbank repo rates, but not the exchange repo rates. This is intuitive because higher reserve requirement drains liquidity from banks, leading to less supply of credit in the interbank repo market. We also examine the impact of liquidity injection from PBOC’s open market operations and find no significant effect on either the interbank or exchange repo market. This is consistent with the use of open market operations as a tool of liquidity sterilization by the PBOC (see Porter and Xu (2009)). Shadow banking activities in China have grown at an explosive 35% annual growth rate during the sample period, more than twice the 16% growth rate for the bank deposits. We find that the log
6
shadow banking activities has a positive and significant effect on both interbank and exchange repo rates, implying that an increase in shadow banking activities leads to more frequent liquidity shortage in the repo markets. This positive effect was more pronounced on the interbank repo rates than the exchange repo rates during the first half when banks’ shadow banking activities were largely unregulated. As the PBOC stepped up with regulatory oversight on banks’ wealth management products while liberalizing the deposit rate ceiling in the second half of the sample period, shadow banking activities have exhibited more influence on the rate spikes of exchange repo market than the interbank repo market. In addition, we find that the monthly high of exchange repo rates is positively and significantly driven by the amount of new equity issuance, consistent with Fan and Zhang (2007) who show that IPO activities drive up exchange repo rates. Subsample analysis shows that this positive effect was pronounced during the first half when new equity issuance was more sizable and exchange repo volume was smaller, but not evident in the second half when new equity issuance was thin and exchange repo market was much bigger. We also examine the impact of new bond issuance and find a significantly negative effect on the rate spikes in the exchange repo market. This is intuitive because new bond issuance by corporations and local governments represents long-term funding that reduces their need for short-term funding, which in turn leads to less liquidity shortage and lower rate spikes in the exchange repo market. Although hot money capital inflows and outflows have been shown to have caused a substantial impact on the boom and bust of equity and real estate markets in China (see Guo and Huang (2010), and Xu and Chen (2012)), we find no consistent evidence of the impact of hot money flow to China on the interbank repo or exchange repo market. In addition, we include China’s home price growth rate as a potential driver of the repo rates because real estate developers and real estate investment firms
7
represent a major group of borrowers in the repo market. We find no evidence of an upward pressure from home price growth rate on either the interbank or exchange repo rates. Finally, we include a quarter-end indicator to capture the possibility of tightening liquidity in March, June, September, and December. The positive effect of the quarter-end indicator is evident on the monthly closing rates of the interbank repos and monthly highest rates of the exchange repos. This is consistent with the conjecture that regulatory checkups and maturity of time deposits for banks lead to more predictable liquidity shortage at the quarter-end closing in the interbank repo market than the exchange repo market. On the other hand, there are more severe rate spikes during quarter end in the exchange repo market due to the maturity of nonbank institutions’ WMPs and absence of funding by banks or the PBOC. The rest of this paper is organized as follows. Section 2 presents the key developments of China’s repo markets, while Section 3 contrasts the interbank and exchange repo rates. Section 4 discusses monetary policy and interest rate liberalization in China. Section 5 characterizes the development of shadow banking activities. Section 6 formulates the potential driving forces of repo rates in the empirical framework, while Section 7 presents the empirical results. Section 8 concludes the paper.
2. Development of Repo Markets in China Financial market developments formally commenced in China with the official opening of two national stock exchanges: the Shanghai Stock Exchange (SSE) in December 1990 and the Shenzhen Stock Exchanges in July 1991. Along with the early development of the stock markets, nonstandardized repo trading entered the local exchange centers on a trial basis in 1991 and the SSE in December 1993 (see Xu (2007)). In 1995, China reformed the repo market by shutting down risky local repo centers and standardizing the national exchange repo market, signifying the official entry of exchange repos
8
into China’s financial market. Initial participants of the exchange repo market included commercial banks as well as nonbank financial institutions. At the center stage of China’s money market was the interbank money market. China’s interbank uncollateralized lending started in 1980s as a venue for banks to manage their excess reserves and credit imbalances, but was often exploited by banks to circumvent credit ceilings and channel credit to securities firms and real estate developers. The interbank lending before 1996 was mostly unregulated and highly fragmented. In January 1996, the PBOC embarked on the unification and modernization of the national interbank lending market with clear guidance on the member qualifications as well as the terms and limits of the lending, and required all trades be done electronically through the China Foreign Exchange Trading System & National Interbank Funding Center (CFETS). Since June 1996, banks were allowed to determine their own interbank offering rates through CFETS, symbolizing the start of the interest rate liberalization process in China.8 China’s interbank lending was initially developed without any securities serving as collateral, but the concerns over counterparty credit risk have prevented its further development. Interbank repos (with bonds as collateral) emerged in 1997 as the PBOC prohibited commercial banks from participating in the exchange repo market and launched the national interbank repo market exclusively for commercial banks. Over time, the eligible members of the interbank repo market have been expanded to include nonbank financial institutions such as insurance companies and securities firms since 2000, and further to include enterprises since 2002. Table 1 summarizes China’s interbank money market trading volume from 2001 to 2016 based on statistics from CFETS. Panel A shows that the interbank lending volume grew from 0.81 Trillion RMB in 2001 to 96 Trillion RMB in 2016, while Panel B shows that the interbank repo volume grew
8
See Iman (2004) for a review on the early stage development of China’s interbank money market. 9
from 4.01 Trillion RMB in 2001 to 568 Trillion RMB in 2016, representing an annualized growth rate of 39% over the last 15 years. It should also be noted that the interbank repo trading accounts for 86% of the interbank money market volume in 2016, clearly dominating the 14% in interbank lending. Finally, with various maturities ranging from overnight to 1 year, the overnight and 7-day tenors are most popular. As shown in Table 1, overnight and 7-day tenors contribute to 87.6% and 9.7% of the interbank lending volume (the last row in Panel A), and 85.5% and 10.9% of the interbank repo volume (the last row in Panel B), indicating the extremely short-term nature of interbank money market. Table 2 provides a snapshot of the trading activities by types of institutions in China’s interbank money market as of December 2016. Panel A shows that large commercial banks, joint stock commercial banks and urban commercial banks account for a total of 70% of the trading volume in interbank lending, while foreign capital institutions, rural financial institutions, and others account for the remaining 30%. Majority of the interbank repo trading takes the form of pledged repo (where bonds are “pledged” as collaterals to reduce counterparty risk), rather than the form of outright repo (where bonds are sold and repurchased).9 Panels B and C show that the number of trades, trading value, and outstanding balance of the interbank pledged repo are 4.9, 14.1, and 8.9 times those of the interbank outright repo, respectively. In addition, nonbank institutions account for 51% and 73% of the trading volume in interbank pledged repos and outright repos, respectively, indicating greater participation by nonbank institutions in interbank repos than interbank lending. According to the 2016 annual statistics from CFETS, 132, 1907, and 86 members participated in the interbank lending, pledged repos, and outright repos, respectively, suggesting broadest member participation in China’s interbank pledged repos.
9
Most of the exchange repos are also pledged repos, rather than outright repos. 10
Since the creation of interbank repos in 1997, China repos have traded in two distinctive market settings: the exchange repo market for nonbank financial institutions and the interbank repo market for banking institutions. The exchange repo market is more standardized and offers more credit protection since the exchange serves as the counterparty for all repo sellers and buyers. While all retail and institutional investors with an exchange account can participate, investments funds, securities firms, insurance companies and security brokers are the major participants in the exchange repo market. As retail investors seek high yield alternatives to bank deposits, money market mutual funds (MMMFs) have emerged as the largest and fastest growing segment of China’s mutual fund industry. According to the Asset Management Association of China (AMAC), China’s MMMF assets reached 4.28 Trillion RMB by the end of 2016. This represents half of the 8.53 Trillion RMB assets under management by all mutual funds in China, and the world’s second largest MMMF market. In the exchange repo market, MMMFs and retail investors normally use their cash holdings to lend cash to obtain higher yield, while bond funds, securities firms, trust companies, local government, and corporations typically use their securities holdings to borrow cash to balance fund flows or enhance leverage. Unlike exchange repos, interbank repos are traded with terms and conditions negotiated between the repo seller and buyer, exposing both sides to counterparty risk. Although both banks and nonbank institutions can participate in interbank repos, banks typically serve as lenders (cash providers, repo buyers) while nonbank financial institutions serve as borrowers (collateral providers, repo sellers). In addition, PBOC conducts weekly open market operations to inject liquidity into the interbank market through reverse repos or to take away liquidity from the interbank market through repos.10 The key advantage of the interbank market lies in its large trading volume and rich source of liquidity from banks and the PBOC.
10
A reverse repo is essentially the other side of a repo transaction. If the PBOC lends money in the repo market, it is said to conduct a reverse repo. If the PBOC borrows money in the repo market, it is said to conduct a repo. 11
Table 3 presents the annual trading volume of interbank repos and SSE exchange repos11 from 2007 to 2016, including the overnight, 7-day, 14-day, 1-month, and 3-month maturities. Relative to the sheer volume of interbank repo trading, the SSE exchange repo volume was much smaller during the earlier years. In 2007, the SSE repo volume was only 1.8 Trillion RMB when the interbank repo volume was 42 Trillion RMB. However, the SSE exchange volume has grown at a much rapid pace than its interbank repo counterpart. From 2007 to 2016, interbank trading volume increased by 12 times to 211 Trillion RMB, while SSE exchange repo volume grew 116 times to 564 Trillion RMB. The interbank to SSE repo volume ratio declined from 23.75 in 2007 to 2.68 in 2016, indicating greater prominence of exchange repo trading over the last ten years. It should be noted that China’s bond repo trading in the interbank market as well as the exchange market, mirroring the spot bond trading in these two parallel markets. According to statistics from Wind Info, bonds outstanding in China’s interbank market as of March 2017 reached 60 Trillion RMB, which is equivalent to 12 times the 5 Trillion RMB bonds outstanding in China’s exchange bond market. Considering the small market share of exchange bond market relative to the interbank bond trading, the statistics from Table 3 show that the exchange bond repo market is far more active than the exchange spot bond market. As for the distribution of trading activities across various maturities, the overnight and 7-day tenors attracted 92.9% and 6.3% of the SSE exchange repo trading volume in 2016, respectively. The rapid growth and increasing dominance of overnight SSE repo were clearly shown in Panel B, even more evident than the growth of overnight interbank repo as shown in Panel A. Chart 1 illustrates the evolution of China’s interbank lending, interbank repo and exchange repo trading volume in the last ten years. In terms of the market share of trading activities, the interbank repo market has maintained a leading position in the last 10 years, but the interbank lending market has
11
Most exchange repos trade on the Shanghai Stock Exchange (SSE), while only a small portion trades on the Shenzhen Stock Exchange. 12
been replaced by the exchange repo market as the second most active money market since 2003. Among these three markets, exchange repo has the highest growth, while interbank lending has the slowest growth.
3. Contrasting the Interbank and Exchange Repo Rates To promote marketization of interest rates and establish interest rate benchmarks for financial markets, the PBOC launched the Shanghai Interbank Offered Rates (SHIBOR) in October 2006, which was published daily by the National Interbank Funding Center based on the average of interbank unsecured lending rate quotes from a selected group of 16 commercial banks. 12 To examine the behaviors and fundamental drivers of interbank and exchange money market rates in China, we download the monthly closing SHIBOR, as well as the monthly closing, highest, and lowest rates of interbank repo and SSE exchange repo rates across the overnight, 7-day, 14-day, 1-month, and 3-month maturities, from Bloomberg. We use monthly frequency in this study because data on many of the possible driving forces of the repo rates are only available on a monthly basis. The sample period covers the ten-year period from October 2006, when data on SHIBOR, interbank and exchange repo rates, as well as shadow banking financing and other potential drivers, become available in Bloomberg, to October 2016. Table 4 presents the summary statistics of the overnight and 7-day SHIBOR, interbank repo rate, and SSE exchange repo rate for the full sample (October 2006 to October 2016), first half (October 2006 to September 2011), and second half (October 2011-October 2016). Panel A shows that the monthly closing rates of SHIBOR and interbank repo are 2.445% and 2.481% for the overnight
12
The SHIBOR as the RMB-based interbank money market benchmark is designed following the London Interbank Offered Rates (LIBOR), which is the globally-recognized USD-based interbank money market benchmark. There are currently 18 commercial banks in the group of SHIBOR quoting banks. 13
maturity, and 3.079% and 3.083% for the 7-day maturity. Theoretically speaking, for the same pair of counterparties, the interbank (secured) repo rate should be lower than the interbank (unsecured) lending rate to reflect the reduction in counterparty credit risk due to the use of bonds as collateral. However, the comparative analysis in Panel B shows that the differences in mean, median and standard deviation between SHIBOR and interbank repo rate of the same maturity are not statistically different from zero. This is intuitive as the SHIBOR reflects the interbank rate quotes from the largest commercial banks, while the interbank repo rate is based on transactions from a broader array of banks and non-bank financial institutions. Since there is no statistically significant difference between the SHIBOR and interbank repo rates, we move on to focus on examining the difference between interbank repo rate and SSE exchange repo rates. To capture the rate dynamics, we examine the monthly closing rate, monthly highest rate, as well as the monthly trading range (i.e., highest rate – lowest rate). Panel A shows that the mean monthly closing rate of SSE exchange repo is 2.29% (2.40%) for the 1-day (7-day) maturity, which is 0.19% (0.68%) lower than its interbank repo counterpart. Panel B shows that the difference in mean and difference in median between the 7-day interbank repo and exchange repo closing rates are positive and significant for the full, first half, and second half of the sample period. On the other hand, the standard deviation of the SSE monthly closing repo rate is 2.22% (1.84%) for the 1-day (7-day) maturity, which is 1.12% (0.54%) higher than its interbank counterpart. While the higher standard deviation in the exchange repo market is consistent with the earlier study by Fan and Zhang (2007), the positive mean (median) difference between the interbank repo rate and exchange repo rate in Table 4 contrasts the negative difference from Fan and Zhang (2007). As the exchange repo market grows dramatically in volume and scope of participation over the years, the exchange’s counterparty
14
guarantee and standardization of terms have led to a positive average rate difference between the interbank and exchange repo markets.13 The higher volatility in the exchange repo market, as shown by the higher standard deviation, is even more evident when we compare the monthly highest rate of the two repo markets. Panel A shows that the exchange repo rate has an average monthly high of 18.02% (7.32%) for the 1-day (7day) maturity, which is much greater than the average monthly high of 4.72% (5.09%) in its interbank repo counterpart. The monthly rate trading range (based on the difference between the highest and lowest rates of the month) in the exchange repo market is also much greater than that in the interbank repo market. Panel B shows that these differences are large and statistically significant for the full, first half, and second half of the sample period. This dramatic contrast reveals that the exchange repo market is far more susceptible to liquidity dry-ups and rate spikes than the interbank repo market, and this vulnerability is much more serious for the overnight repos than the 7-day repos. Although many nonbank institutions have been allowed to participate in both markets, banks and the PBOC only participate in interbank repo market while retail investors can only participate in the exchange repo market, and therefore the two markets are still partially segmented. When rate spikes in the interbank repo market, large banks with deep funding source, as well as the PBOC, can provide credit supply to combat the liquidity shortage. Without the participation of banks and the PBOC, rate spikes occur in the exchange repo market more often and at a larger magnitude than the case in the interbank repo market. Subsample analysis indicate that the size of difference in monthly highest repo rates between exchange and interbank markets declines substantially from the first half to the second half for the 7-
13
Freixas and Holthausen (2005) investigate cross-country interbank lending and found persistent interest rate differentials due to asymmetric information. They find that interbank repos with security collaterals, rather than unsecured interbank lending, help to reduce the cross-market rate spread and improve upon segmentation. However, the case of segmentation between China’s interbank and exchange repo markets is unique and differs substantially from the cross-county interbank segmentation case. 15
day repos, but not the overnight repos. Limited arbitrage activities by nonbank institutions may help to partially reduce the spike in the 7-day exchange repo, but the overnight interbank and exchange repo markets remain highly segmented due to limited capital mobility. In addition, the negative mean difference in rate spikes between the interbank and exchange repo markets is significant in both the first half and the second half of the sample period, suggesting continued presence of market segmentation.
1. Monetary Policy and Interest Rate Liberalization in China Interest rates in China have multiple facets: money market and bond market rates are determined by the market, with the 7-day repo rate being the most widely cited money market benchmark in China; on the other hand, bank deposit and lending rates were once strictly controlled by the PBOC, but they have been gradually liberalized from 2004 to 2015. Although interbank and exchange repo rates are market determined, it should be noted that the role of monetary policy by the PBOC could remain an important potential driver of the market rates. The PBOC has a variety of direct and indirect monetary policy tools, ranging from setting the policy rates, setting bank deposit and lending benchmark rates, setting the required reserve ratio for banks, to conducting open market operations. Table 5 presents descriptive statistics and unit root tests for various measures of China’s monetary policy variables, and their correlation with China’s 7-day interbank and SSE exchange repo rates, from the common sample period of all these variable from February 2008 to October 2016. Chart 2 presents a time series plot of PBOC’s policy rates, benchmark rates, and the required reserve ratio for the full sample period from October 2006 to October 2016. Similar to the Federal Reserve, the PBOC has full control over its own policy rates: the rediscount rate (the interest rate it charges to member banks when they borrow from the PBOC to get
16
liquidity), the interest rate it pays to banks on their required reserves, as well as the rate on excess reserves. However, these three PBOC policy rates traditionally play a minor role in China’s monetary policy execution. China’s financial system is dominated by a banking sector that has total RMB deposit almost twice the size of its GDP.14 To exercise interest rate control over the banking sector, the PBOC traditionally set benchmarks for commercial banks’ deposit rates and lending rates for various maturities, with the 1-year deposit rate and lending rate being most relevant to the money market. The PBOC’s setting of bank deposit and lending benchmark rates was highly binding before China embarked on the process of interest rate liberalization. Starting 2004, China removed the lending rate ceiling and lowered the lending rate floor to 0.9x the PBOC’s benchmark lending rate, which was further reduced to 0.7x in July 2012, and fully liberalized in July 2013. On the other hand, the deposit rate floor was removed in 2004, while the deposit rate ceiling was raised to 1.1x the PBOC’s benchmark deposit rate in 2012, 1.2x in 2014, 1.5x in 2015, and eventually fully liberalized in October 2015. The left axis of Chart 2 illustrates the PBOC’s rediscount rate (CDR), interest rate on required reserve (CRRI), interest rate on excess reserve (CERI), 1-year benchmark savings deposit rate (CSR), and 1-year benchmark lending rate (CLR). During the sample period from October 2006 to October 2016, the PBOC changed the CRRI and CERI only once in November 2008, but made more frequent adjustments to CDR, CSR and CLR. Although increase (decrease) in any of these five rates could exert upward (downward) pressure on the repo rates, including all of these rates in the regression will lead to high multicollinearity. In addition, as shown in Table 5, although CSR and CLR have high correlation with the repo rates, they are not suitable to serve as direct drivers of the repo rates due to nonstationarity as shown in their insignificant ADF (Augmented Dickey Fuller Unit Root Test)
According to annual statistics from the World Bank and Bloomberg, China’s total RMB deposits in commercial bank reached 150.59 Trillion RMB (equivalent to 20.918 Trillion USD) in 2015, which was 1.89 times its GDP in 2015. The size of the U.S. total bank deposits was at 11.303 Trillion USD in 2015, which was 62.67% of its GDP in the same year. 14
17
statistics. To address the issues of multicollinearity and unit root, we develop a China monetary policy composite rate indicator (CMP) to sum up all the above five rates in one measure. The CMP incorporates all the five policy and benchmark rates set by the PBOC and has a highly significant ADF, indicating that the measure is stationary. One unique feature of China’s monetary policy toolkit is that the PBOC increases (decreases) the banks’ required reserve ratio (CRR) to reduce (increase) available credit available by banks for lending. As shown in the right axis of Chart 2, CRR has been raised gradually from 8.5% in 2006 to a peak of 21.5% in September 2011, and then cut to 17% in 2016, in contrast to the 10% stable bank required reserve ratio set by the Federal Reserve System since 1992. Although CRR has an extremely high correlation with the repo rates, we use the change in the required reserve ratio (CRRD) as a potential driver for repo rates because the CRR has a unit root and therefore is nonstationary. Originated in 1998, China’s open market operations (COMO) mainly conduct reverse repos and repos in the interbank money market to inject or drain liquidity from the money market. The PBOC has also introduced short-term liquidity operations (SLO) in January 2013 to combat temporary liquidity shortage in the money market.15 The impact of COMO on repo rates is less straight forward as COMO could be used ex ante to guide the money market toward a new rate, or could be used ex post in its sterilization effort to reduce liquidity crunch. Since data on the PBOC’s net amount of money into the market via COMO starts from February 2008, and the correlation between COMO and the repo rates is much lower relative that of the CMP and CRR, we only use COMO as a potential driver in the second half of the sample period when COMO data is complete.
15
According to the PBOC, the short-term liquidity operation (SLO) introduced in January 2013 allows 12 banks to borrow from the PBOC using repos (less than 7 days) to manage their cash shortage. It was intended to supplement the PBOC-initiated open market operations conducted on Tuesday and Thursday each week. However, since the PBOC moved to daily OMO in 2016, the need for SLO has been reduced. 18
2. The Emergence of Shadow Banking in China Originating from a bank-dominant state economy, traditional banking in China has been highly regulated and subject to strict capital requirement, high reserve requirement, binding deposit rate ceilings, loan to deposit cap, and directions from the State on the quantity and directions of lending (through “window guidance” from the PBOC). To circumvent regulatory & policy constraints and improve profitability, banks have moved some of their financing activities from traditional banking to high yielding shadow financing activities such as entrusted loans, trust loans and undiscounted bankers’ acceptance.16 Nonbank financial institutions have also participated in shadow banking activities by engaging in similar credit and maturity intermediation outside of the formal banking system. Funke, Mihaylovski and Zhu (2015) develop a model to show that tighter interest rate regulation in the commercial banking sector leads to increasing shadow banking activities. Gennaioli, Shleifer and Vishny (2013) show that the shadow banking system is stable and welfare improving under rational expectations, but vulnerable to liquidity shocks when tail risks are ignored. While excessive growth in a shadow banking system may lead to greater systemic risk for the financial sector, it is considered as a competitive market-based financing system that provides the critical credit intermediation supplement when the traditional banking system is insufficient or incompetent in providing the needed credit to support economic growth. According to 2014 year-end statistics from the Financial Stability Board (2015), bank assets account for 73% and 25% of the total assets of all financial institutions in China and the U.S., showing that China’s financial system is still overwhelmingly bank-dominant. As shown in Panel A of Chart 3, shadow banking activities (based on economic functions tracked by the Financial Stability Board)
16
According to Elliott, Kroeber and Qiao (2015), Entrusted Loans are loans made on behalf of large corporations with banks or finance companies serving as intermediaries, while Trust Loans are loans made by Trust Companies. Allen, Qian, Tu, and Yu (2016) perform transactional level analysis on entrusted loans in China and show that the pricing of these loans incorporate fundamental and informational risks. 19
represent 26% of GDP in China in 2014, still far lower than its 82% counterpart in the U.S. in 2014. As shown in Panel B of Chart 3, China’s shadow banking activities have grown at a much faster pace than the total bank deposits over the past ten years. 17 According to statistics from Bloomberg Intelligence, China’s social banking financing activities grew from 1.2 Billion RMB in 2006 to 2,426 Billion RMB in 2016, with an explosive 35% annual growth rate, more than twice the 16% growth rate for the total bank deposits during the same period. As the repo market becomes a key source of liquidity for credit intermediation between shortterm WMPs and long-term risky shadow loans and investments, excessive funding imbalance and severe liquidity crunch hit the repo markets. Panel C of Chart 3 illustrates that China’s shadow banking as a % of the total bank deposits (on the right axis) grew from 3.39% in October 2006 to 20% in January 2014, before coming down to 15.4% in October 2016, consistent with the conjecture that shadow banking activities may have contributed to the pronounced interbank repo rate spike in June 2013 (on the left axis) when the PBOC refused to inject liquidity into the interbank market. Since 2013, the PBOC engaged in a number of steps to liberalize the interest rates, remove the 75% loan to deposit cap, regulate banks’ WMPs, and contain the excessive growth of the shadow banking system, which have led to the more stable and normal growth in shadow banking activities from 2014 to 2016.
3. Potential Drivers of the China Repo Market In this study, we examine the drivers of China interbank and exchange repo rates using the monthly closing rates as well as the monthly highest rates. As discussed before, this is necessary to capture the dynamics because rate spikes and liquidity shocks are not fully captured by the closing rates alone at either the daily or monthly level.
17
Hachem and Song (2016) provide evidence that tighter reserve requirement and other regulatory constraints in the late 2000s contributed to the rapid development of China’s shadow banking activities. 20
Before we conduct regression analysis, we examine the autocorrelation and stationarity of the repo rate series. As shown in Panel A of Table 4 for the full sample from October 2006 and October 2016, the first order autocorrelation function (ACF) for the 1-day (7-day) maturity is 0.679 (0.654) for the monthly closing interbank repo rates, and 0.312 (0.359) for the monthly highest interbank repo rates. When compared with the interbank repo rates, the SSE exchange repo rates show lower ACF in the monthly closing rates and higher ACF in the monthly highest rates. All the monthly interbank and SSE exchange repo rates are checked for unit root using the ADF test. Statistics in the last column in Panel A reject the presence of unit roots and confirm their stationarity in these monthly rate series. As a result, we will be using these rates (monthly closing, monthly highest) as the dependent variables directly in the modeling and analysis, and include the one-month autoregressive term (AR1) as an independent variable to control for the rate persistence. According to the analysis in Section 4 and statistics from Table 5, three variables related to PBOC’s monetary policy actions are used in this study as potential drivers. The first one is the China monetary policy composite rate indicator (CMP), which sums up the three PBOC policy rates as well as the 1-year bank deposit & lending benchmark rates controlled by the PBOC. We expect CMP to be a strong positive driver of repo rates in both the interbank and exchange repo markets as higher policy and benchmark rates set by the PBOC should lead to higher costs for lenders in the repo market, and indirectly drive up the repo rate. The second variable is the monthly change in required reserve ratio (CRRD), which is also expected to be positively driving the repo rates as higher reserve requirement tightens bank liquidity and leads to less supply of credit from banks in the repo market. It is not clear though, whether CRRD only affects interbank repo rates, or both interbank and exchange repo rates. As for China’s open market operations (COMO), because it could be used proactively to achieve policy goals or to sterilize the interbank market after liquidity dry-ups, its impact on monthly repo rates is not
21
as clear. Table 6 presents descriptive statistics of the potential drivers for the full sample, first half, and second half in Panels A, B and C, respectively. Table 6 shows that the standard deviation of CMP is 2.63% in the first half (Panel B) and 1.48% in the second half (Panel C), indicating much more stable policy and benchmark rates during the second half of the sample. On the other hand, the PBOC has relied on open market operations more extensively during the second half, as evident by the average monthly injection of 48.6 Billion RMB through COMO in second half, compared to the average 1.3 Billion RMB in COMO in the first half. To investigate the potential impact of shadow banking activities (CSB) on the repo markets, we used the log shadow banking activities (LCSB), instead of the raw CSB, in the analysis. The log transformation on CSB is used because the growth of CSB has been exponential over the last ten years and the raw CSB variable is highly nonstationary. As shown in Table 6, we also compute CSB as a percentage of total bank deposits (CSBTD), but CSBTD is highly nonstationary because the growth of CSB clearly outpaced the growth of total bank deposits during the sample period. As a result, we use the LCSB to capture the potential impact of shadow banking activities on the repo markets. This will be a new contribution to the literature as previous empirical studies on China repo markets have never incorporated shadow banking as a potential driver of repo rates. Fan and Zhang (2007) show that funds tied up during stock IPOs represent a key credit demand in the exchange repo market and therefore drive up the cross-market difference between exchange repo rate and interbank repo rate. Porter and Xu (2009) find that IPO activities affect the daily volatility of interbank repo rate, but not the level of repo rate. We also include the monthly total of China’s stock market A-share new issuance (CPOA) as a potential driver.18 Table 6 shows that the monthly average CPOA was 36.51 Billion RMB and 8.45 Billion RMB for the first half and second half of the sample
18
A shares are stocks issued in RMB while B shares are stocks issued in foreign currencies in China. Majority of the stocks issued in China are A shares. 22
period, respectively, suggesting much less credit demand to fund IPO activities during the second half. Considering that the exchange repo market has expanded dramatically in trading activities over time, we expect the exchange repo rates to become less sensitive to CPOA over time. In addition to stock issuance activities, we also include the monthly total of China’s bond issuance (CBI) as an independent variable. Because bonds are issued and traded in both the interbank and exchange markets, an increase in CBI could represent more supply of bond collateral (and therefore more demand for bond repos) in the repo markets, leading to higher repo rates. Lou, Yan, and Zhang (2013) examine the impact of regularly scheduled U.S. Treasury auctions and find that Treasury repo rates tend to increase after the Treasury auctions, consistent with this conjecture. On the other hand, more capital raised through long-term debt issuance (especially those by corporations and local governments) could lead to less demand for short-term debt financing in the repo market, leading to lower repo rates. Table 6 shows that the monthly average CBI was 680 Billion RMB and 678 Billion RMB for the first half and second half of the sample period, respectively. As 70% of debt issuance in China during the sample period was by local governments and corporations, who are key borrowers in the exchange repo market, we expect CBI to have a negative effect on the exchange repo rates. Since qualified foreign institutions may participate in China’s interbank and exchange repo markets, we also include the hot money flow to China (CHMF) as a potential driver of repo rates. “Hot Money Flow to China” typically refers to short-term speculative money flowing into China to seek higher investment returns.19 During the sample period, China’s 7-day repo rate averaged at 3.08%, much greater than the average 7-day repo rate of 0.92% in the U.S., showing a higher interest rate environment in China that could be attractive to foreign capital flows. Although China still has capital
19
According to Bloomberg, the monthly hot money flow to China statistics are calculated by taking the sum of FX purchases by the PBOC and change in FX deposits as total flows into the country, and netting out the monthly trade and direct investment balances. 23
control and its currency is still not fully convertible, hot money capital inflows and outflows have shown substantial impact on the boom and bust of equity and real estate markets in China (see Guo and Huang (2010), and Xu and Chen (2012)). According to Panels B and C of Table 6, China had an average monthly CHMF of positive 17 Billion RMB (inflow) during the first half and negative 27 Billion RMB (outflow) during the second half, showing a reversal from hot money inflow in the first half to outflow in the second half. Since capital inflow (outflow) may result in more (less) supply of liquidity to the repo market, our research will investigate if CHMF represents a significant negative driver to the repo rates in China. In addition, we include China’s monthly YOY home price growth rate (CHPGY) as a potential driver of the repo rate because real estate developers and real estate investment firms represent a major group of borrowers in China’s repo market. During the ten-year period from October 2006 to October 2016, China’s annual home price growth rate averaged at 3.62%, 5.49% and 1.79%,, for the full, first half and second half of the sample period, respectively. We expect that higher home price growth will drive more demand to fund investments and speculations in real estate, creating possible upward pressure on repo rates. Finally, we include a quarter-end indicator (QEND) to capture the possibility of tightening liquidity in March, June, September, and December. We expect a positive effect of QEND on repo rates as liquidity dries up due to quarter-end regulatory inspections, maturing time deposits, and maturing wealth management products.
4. Empirical Analysis Although repos trade with tenors ranging from one day to three months, and even possibly to one year, 86% (93%) and 11% (6%) of the interbank (exchange) repo trading volume are in the 1-day
24
and 7-day maturities, respectively. We include the 1-day repo rate in the analysis because it is the most actively traded term to maturity in both the interbank and exchange repo markets during the sample period. As for the 7-day repo, because it was the dominating term to maturity in the interbank market up to 2005, earlier studies on China repo markets (such as Fan and Zhang (2007) and Porter and Xu (2009)) focused on the 7-day repo rate in their analysis. Furthermore, the 7-day interbank repo rate is often considered as the best indicator of money market liquidity in China due to its long history and its frequent use by the PBOC in open market operations. In sum, we also include the 7-day repo in our analysis because it is the second most active repo tenor, the China money market benchmark indicator, and the most studied China repo rate in the literature. Table 7 presents the monthly regressions for China’s interbank repo rate on the list of possible drivers for the 7-day and 1-Day repos in Panel A and Panel B, respectively. In each of the panel, we summarize the regression estimates based on the monthly closing repo rate (monthly close) and monthly highest repo rate (monthly high) as two alternative dependent variables. The regressions were presented for the full sample period (October 2006 to October 2016), as well as the first half (October 2006 to September 2011), and second half (October 2011-October 2016). The main regressions include the following independent variables as possible drivers of repo rates: an autoregressive term (AR1), monetary policy composite rate indicator (CMP), monthly change in required reserve ratio (CRRD), log shadow banking activities (LCSB), new equity issuance (CPOA), home price growth rate (CHPRY), and quarter-end indicator (QEND). For the second half of the sample, we also include the open market operations (COMO) and new bond issuance (CBI) as two additional variables in a separate regression.20 Similar to the regression analysis for the interbank repo rates in Table 7, Table 8 presents regression analysis for the SSE exchange repo rates, while Table 9 reports the analysis for the interbank
20
The starting dates of the COMO and CBI variables are February 2008 and May 2007, respectively. Due to incomplete data in the first half of the sample period, we only include them in the second-half regression. 25
to SSE exchange repo rate spread. All the dependent and independent variables used in the regressions in this study have been tested to ensure stationarity. Appendix A presents the detailed definitions and data sources of all dependent variables and independent variables used in this study.
4.1 Drivers of China Interbank Repo Rates For the interbank repo monthly closing rate regressions in Table 7, Panel A (Panel B) shows that the lag one-month autoregressive team (AR1) has a positive and highly significant coefficient of 0.362 (0.462) on the 7-day (1-day) repo rate, showing persistence in the monthly repo rates. The coefficient of monetary policy composite rate indicator (CMP) is 0.190 (0.140) and highly significant, consistent with the expectation that an increase in PBOC’s policy and benchmark rates (an aggregate of the rediscount rate, interest rate on required reserve, interest rate on excess reserve, 1-year benchmark savings deposit rate, and 1-year benchmark lending rate) will drive up the interbank repo rates. As for the monthly change in the bank required reserve ratio (CRRD), the coefficients for both 7-day and 1-day regressions are positive and highly significant, consistent with the conjecture that higher (lower) bank reserve requirement leads to less (more) supply of credit in the interbank repo market and thus higher (lower) interbank repo rates. The open market operations (COMO) variable is only used in the second half regression due to data limitation, and fails to show significant impact on either the 7-day or 1-day repo rate. This is consistent with Porter and Xu (2009) who find no significant impact of COMO on the level and volatility of daily repo. As discussed before, the impact of COMO on repo rates is ambiguous as COMO could be used by the PBOC ex ante to guide the repo rate toward a new policy target, or ex post after liquidity shortage to adjust market liquidity of toward rate stability. The log shadow banking financing (LCSB) variable has shown a significantly positive coefficient in both Panel A (7-day repo) and Panel B (1-day repo) for the full sample. This is intuitive
26
because an increase in shadow banking activities leads to more frequent liquidity shortage in the interbank market and drives up repo rates. This positive effect is strong and significant for the first half of the sample period when shadow banking in China experienced the most dramatic growth, but insignificant during the second half of the sample, when the PBOC stepped up with interest rate liberalization and regulation of bank WMPs to contain the excessive growth of shadow banking activities. As for new equity issuance (CPOA) and bond new issuance (CBI), they do not show any significant impact on the monthly interbank repo rates during the sample period. While we do not find any significant impact of China hot money flow (CHMF) on the 7-day interbank repo rate, there is a marginally significant negative effect of CHMF on the 1-day interbank repo rate during the second half, showing weak evidence that the lower CHMF (as China experienced the outflow of hot money) during the second half put upward pressure on the overnight interbank repo rate. There is no significant effect of the home price growth index (CHPRY) for the full sample period, but some evidence of a significant (marginally significant) positive effect on the 7-day (1-day) interbank repo rate during the second half. Finally, the quarter-end indicator is highly significant with a positive coefficient of 0.424 (0.505) on the 7-day (1-day) interbank repo rate, indicating more liquidity shortage and upward rate pressure at quarter end. The positive quarter-end effect on the interbank closing repo rates is especially pronounced during the second half. In this study, we also explore drivers of the monthly highest repo rates as they can better capture rate spikes and liquidity shocks than the monthly closing repo rates. The results on the monthly high regressions are presented in Table 7 side by side with those from the monthly close repo rates in both Panel A and Panel B. The monthly high interbank repo rates are less persistent than their monthly close counterpart, as shown by the smaller AR1 coefficient. Similar to results from the monthly close
27
regressions, the effects of monetary policy composite rate (CMP) and log shadow banking financing (LCSB) on the interbank monthly high repo rates are positive and highly significant, and the effects of open market operations (COMO), bond new issuance (CBI), hot money flow (CHMF) and home price growth (CHPRY) are not significant. While there was no evidence of the impact of new equity issuance (CPOA) on the monthly close repo rate, we find evidence that CPOA has a significantly positive effect on the monthly high of 7-day interbank repo rate during the first half of the sample period when new equity issuance was most active. In addition, the change in bank required reserve ratio (CRRD) has a significantly positive effect on the monthly high of the 7-day repo rate, but not the 1-day repo rate. We also find no evidence on the significance of the quarter-end indicator variable on the monthly high interbank repo rates. Finally, none of the independent variables show any significant effect on the monthly high interbank repo rates during the second half of the sample period.
4.2 Drivers of China Exchange Repo Rates Table 8 presents the monthly regressions for China’s SSE exchange repo rate for the 7-day and 1-Day repos in Panel A and Panel B, respectively. As we compare the SSE exchange repo regression results in Table 8 with those from Table 7 on interbank repos, a few patterns emerge. For the monthly close regression, the AR1 coefficient is 0.006 (0.044) for the 7-day (1-day) exchange repo rate, which is small and insignificant, showing much less persistence than their interbank monthly closing repo counterpart. Unlike the monthly close, the monthly high of exchange repo rate has a highly significant AR1 coefficient of 0.255 (0.253) for the 7-day (1-day) maturity, indicating higher persistence for the monthly highs of the exchange repo rate. As for the China monetary policy composite rate (CMP), it has a positive and significant coefficient on the monthly close as well as the monthly high of exchange repo rates, consistent with the positive effect of PBOC policy and benchmark rates on repo market
28
rates. The change in bank required reserve ratio (CRRD) does not show any significant effect on either the monthly close or the monthly high of exchange repo rates. This contrasts the positive and significant impact of CRRD on the interbank repo rates, suggesting that CRRD is a driver of the interbank repo rates but not the exchange repo rates. This finding is intuitive because banks are the net lenders in the interbank repo market, but do not participate in the exchange repo market. The impact of log shadow banking activities (LCSB) was more pronounced on the interbank repo rates than the exchange repo rates during the first half when shadow banking activities by banks were largely unregulated. However, in the second half, LCSB showed a stronger effect on the exchange repo rates than their interbank repo counterparts. As PBOC stepped up with regulatory oversight on banks’ wealth management products while liberalizing the deposit rate ceiling in the second half of the sample period, shadow banking activities have grown at a more stable rate and exhibited more influence on the exchange repo market than the interbank repo market. This explains why LCSB has become a strong positive driver of the monthly high of both 7-day and 1-day exchange repo rates in the second half. In addition, we find that the monthly high of exchange repo rates is positively and significantly driven by the new equity issuance (CPOA), consistent with Fan and Zhang (2007) who show that IPO activities drive up exchange repo rates. Subsample analysis shows that the positive effect of CPOA was pronounced during the first half when new equity issuance was more sizable and exchange repo volume was smaller, but not evident in the second half when new equity issuance was thin and exchange repo market was much bigger.21
21
During the first half of the sample period, monthly new equity issuance averaged 36.51 Billon RMB, equivalent to 9.12% of the monthly SSE exchange repo trading volume (400 Billion RMB). In contrast, during the second half, monthly new equity issuance averaged 8.45 Billion RMB, equivalent to only 0.05% of the monthly SSE exchange repo trading volume (16.316 Trillion RMB). 29
In the second half regression, new bond issuance (CBI) has shown a significantly negative effect on the monthly high of both 7-day and 1-day exchange repo rates. As corporations and local governments are major participants in the exchange repo market, new bond issuance represents longterm funding that reduces their need for short-term funding, which in turn leads to less liquidity shortage and lower rate spikes in the exchange repo market. We find a negative and significant effect of hot money flow (CHMF) on the monthly high exchange repo rates during the first half, but results are mixed in the second half. The quarter-end indicator has a significant and positive coefficient of 1.610 (8.454) on the 7-day (1-day) monthly high exchange repo rate, consistent with more serious liquidity crunch at quarter end. The positive quarterend effect on the exchange repo monthly high rates is especially pronounced during the second half, and most sizable for the 1-day maturity. While nonbank institutions may participate in both interbank and exchange repo markets to arbitrage, the two different trading mechanisms lead to limited funding mobility for the overnight maturity. Finally, the monthly high regressions have R-squared of 43.9% (38.0%) for the 7-day (1-day) exchange repos in Table 8, much higher than the 26.9% (22.0%) R-squared for the 7-day (1-day) interbank repos in Table 7. In contrast, the monthly close regressions have R-squared of 13.4% (17.4%) for the 7-day (1-day) exchange repos in Table 8, much lower than the 57.9% (61.4%) R-squared for the 7-day (1-day) interbank repos in Table 7. These results suggest that our regression model has better explanatory power for the monthly highest rate in the exchange repo market and the monthly closing rate in the interbank repo market.
4.3 Drivers of the Interbank to Exchange Repo Rate Spread
30
In Table 9, we investigate the driving forces of the rate spread between the interbank repos and the exchange repos of the same maturity. Recall that the descriptive statistics from Table 4 show that the average rate spread based on monthly close is positive (reflecting the lower counterparty risk and standardized trading mechanism in the exchange repo market), while average rate spread based on monthly high is negative (reflecting deeper funding source and less vulnerability to liquidity crunch in the interbank repo market due to the participation of major banks and the PBOC). Panel A presents regression analysis for the rate spread for the 7-day maturity based on the monthly close repo rates between the interbank and exchange markets, as well as the spread between the monthly high rates of the two markets, for the full sample as well as the two subsamples. Panel B presents similar analysis for the 1-day repos. The potential driving forces in the regressive include all the factors used in Tables 7 and 8. Conditioning on all the factors included in the regression, the monthly close rate spreads for the 7-day and 1-day maturities do not show significant autoregressive term (AR1) coefficient, but there is highly positive AR1 coefficient for the monthly high rate spread for the 1-day maturity. This shows persistent segmentation between the interbank and exchange markets for the overnight repo, which can be explained by limited capital mobility between the two markets to exploit any arbitrage opportunity in the overnight rate spikes. The impact of monetary policy composite rate indicator (CMP) is not significant on the monthly close rate spread, but shows a significantly negative effect on the monthly high rate spread for the overnight repos in Panel B. This is consistent with the more dramatic positive reaction of overnight rate highs at the exchange repo market (relative to the interbank repo market) as PBOC policy & benchmark rates change. However, the change in required reserve ratio (CRRD) or the open market operations (COMO) show no significant effect on the rate spread between the two markets.
31
Results on the 7-day repo rate spread in Panel A shows that the log shadow banking financing (LCSB) has a positive effect on both the monthly close and high rate spreads in the first half, but a highly negative effect on the monthly high rate spread during the second half. Panel B also confirms the negative effect of LCSB on the monthly high rate spread for the overnight repos in the second half. These findings are consistent with the interpretation that more regulations on banks’ WMPs and deposit rate liberalization in the second half have led to greater reliance of shadow banks on the exchange repo market as their short-term wholesale funding source. New equity issuance (CPOA) has shown a negative and significant impact on the monthly high rate spread in both Panel A and Panel B, consistent with the liquidity drain from funds tied up to IPO subscriptions in the exchange repo market. However, the negative effect of CPOA on monthly high rate spread is no longer significant in the second half, when the average size of CPOA only represented only 0.05% of the exchange repo trading volume. There is weak evidence of the positive effect of new bond issuance (CBI) on the monthly high rate spread for overnight repos, consistent with less shortterm funding need in the exchange repo market from corporations and local governments that issue new bonds. In addition, the effect of the quarter-end indicator (QEND) on rate spread was positive and significant on the monthly close rate spread for 7-day repos, consistent with the conjecture that regulatory checkups and maturity of time deposits for banks lead to more predictable liquidity shortage at the quarter-end closing in the interbank repo market than the exchange repo market. On the other hand, QEND has a negative and significant effect on the monthly high rate spread for the 1-day repos, implying more severe rate spikes during quarter end in the overnight exchange repo market due to the maturity of nonbank financial institutions’ WMPs, quarterly corporate tax payment, and absence of funding by banks or PBOC.
32
Finally, the R-squared for the monthly high interbank to exchange repo rate spread regression is 38.7% (37.7%) for the 7-day (1-day) maturity, much higher than the 14.6% (5.7%) for the monthly close spread regression. This indicates high explanatory power of our regression framework in addressing the difference between rate spikes in the two repo markets.
5. Conclusion Repos in China channel efficient allocation of short-term funds among banks, nonbank financial institutions, corporations, as well as retail investors, and represent the most liquid and fastest growing financial market in China. China repos trade in the interbank market as well as the stock exchanges, with the interbank market being exclusively for banks when it was initiated by the PBOC in 1997. Currently, nonbank financial institutions and corporations can participate in both interbank and exchange repo markets. However, banks and the PBOC can only participate in the interbank repo market, while retail investors can only participate in the exchange repo market. In addition, exchange repos carry the counterparty guarantee from the exchange, while interbank repos expose participants to counterparty risk. Using ten years of monthly closing and highest rate data from October 2006 to October 2016, we examine the behaviors and key drivers of interbank and exchange repo rates, and their difference. As the exchange repo market grows dramatically in volume and scope of participation over the years, the exchange’s counterparty guarantee and standardization have led to a positive monthly closing rate spread between the interbank and exchange repo markets. On the other hand, comparisons of the monthly highest repo rates between the two markets indicate that the exchange repo market, especially the overnight tenor, is much more susceptible to liquidity dry-ups than the interbank market. Without large funding source from banks and monetary sterilization from the PBOC, the exchange repo market
33
has much more frequent and sizable rate spikes than the interbank repo market. Subsample analysis shows continued segmentation between both markets. In light of the tremendous growth in China’s repo market, aggressive interest rate liberalization, emergence of shadow banking activities, and increasing interest in China’s repo markets by foreign participants, we develop an empirical framework that thoroughly examines the impact of the following on the interbank and exchange repo rates, and their difference over the last ten years: China’s monetary policy variables, such as PBOC’s policy and benchmark rates, change in required reserve ratio, funding injection from open market operations; China’s shadow banking social financing activities; new equity issuance activities; new bond issuance activities; hot money flow; home price growth; quarter-end calendar effect. We find that the PBOC’s monetary policy rates serve as a significantly positive driver of repo rates in both the interbank and exchange repo market, but open market operations by the PBOC has little impact on the monthly repo rates in both markets. On the other hand, the change in bank required reserve ratio has a positive effect on interbank repo rates, but not exchange repo rates. Our empirical results show that an increase in shadow banking activities exerts upward pressure on repo rates, and this positive effect was more pronounced on the interbank market than the exchange market during the first half when banks’ shadow banking activities were largely unregulated. On the other hand, as PBOC stepped up with regulatory oversight on banks’ wealth management products while liberalizing the deposit rate ceiling in the second half of the sample period, shadow banking activities have exhibited more influence on the rate spikes of exchange repo market than the interbank repo market. Empirical results show that the monthly high of exchange repo rates is positively and significantly driven by the stock IPO activities, but that this positive effect diminishes in the second
34
half as new equity issuance represents much smaller funding pressure to the exchange repo market. Our results indicate that new bond issuance has a negative effect on the monthly highs of exchange repo rates, consistent with the long term debt substitution that results in less liquidity shortage and lower rate spikes in the exchange repo market. We also document a positive effect of the quarter-end indicator on monthly closing rates of interbank repos and monthly highs of exchange repos, consistent with quarter-end liquidity shortage due to regulatory checkups, corporate tax payments, and maturity of time deposits & wealth management products. This study contributes to the literature in two perspectives. First, extending earlier studies on China repo markets, this paper provides a comprehensive examination of the segmentation behaviors and fundamental drivers of China’s interbank and exchange repo rates and their differences. Second, to the best of our knowledge, this is the first empirical study to investigate the impact of shadow banking on repo markets in China, which complements the existing literature on the impact of “securitized shadow banking” on the “run on repo” during the global financial crisis. The continued segmentation between the interbank and exchange repo markets, partially due to the insulation of banks and the central bank from the exchange repo market, could represent a source of future systemic risk to China’s financial system. As China completes its decade-long interest rate liberation process and reforms its shadow banking system, macroprudential policy considerations should be given to mitigate the excessive volatility in the exchange repo market.
35
References
1. Allen, Franklin, Yiming Qian, Guoqian Tu, and Frank Yu. 2016. Entrusted Loans: A Close Look
at
China's
Shadow
Banking
System.
Available
at
SSRN:
https://ssrn.com/abstract=2621330 2. Baklanova, Viktoria, Adam Copeland, and Rebecca McCaughrin. 2015. Reference Guide to U.S. Repo and Securities Lending Markets. Federal Reserve Bank of New York Staff Report. 3. Brunnermeier, Markus K., and Lasse Heje Pedersen. 2009. Market Liquidity and Funding Liquidity. Review of Financial Studies 22: 2201–2238. 4. Comotto, Richard. 2012. Shadow Banking and Repo. ICMA European Repo Council. 5. Elliott, Douglas, Arthur Kroeber, and Yu Qiao. 2015. Shadow Banking in China: A Primer. Economic Studies at Brookings. 6. Fan, Longzhen, and Chu Zhang. 2007. Beyond Segmentation: The Case of China’s Repo Markets. Journal of Banking and Finance 31: 939-954. 7. Financial Stability Board. 2015. Global Shadow Banking Monitoring Report. 8. Freixas, Xavier, and Cornelia Holthausen. 2005. Interbank Market Integration under Asymmetric Information. Review of Financial Studies 18 (2): 459-490. 9. Funke, Michael, Petar Mihaylovski, and Haibin Zhu. 2015. Monetary Policy Transmission in China: A DSGE Model with Parallel Shadow Banking and Interest Rate Control. BOFIT Discussion Papers. 10. Gennaioli, Nicola, Andrei Shleifer, and Robert W. Vishny. 2013. A Model of Shadow Banking. Journal of Finance 68(4): 1331-1363 11. Gorton, Gary. 2009. Information, Liquidity, and the (Ongoing) Panic of 2007. American Economic Review: Papers and Proceedings 99(2): 567-572.
36
12. Gorton, Gary, and Andrew Metrick. 2012. Securitized Banking and the Run on Repo. Journal of Financial Economics 104: 425-451. 13. Guo, Feng, and Ying Sophie Huang. 2010. Does “Hot Money” Drive China's Real Estate and Stock Markets? International Review of Economics and Finance 19: 452–466. 14. Hachem, Kinda Cheryl, and Zheng Michael Song. 2016. Liquidity Regulation and Unintended Financial Transformation in China. NBER Working Paper. 15. He, Dong and Wang, Honglin, 2012. Dual-Track Interest Rates and the Conduct of Monetary Policy in China. China Economic Review 23(4): 928–947. 16. Imam, Michael. 2004. The Chinese Interbank Markets: Cornerstone of Financial Liberalization, China & World Economy 12(5): 17-33. 17. Krishnamurthy, Arvind, Stefan Nagel, and Dmitry Orlov. 2014. Sizing Up Repo. Journal of Finance 69(6): 2381-2417. 18. Lou, Dong, Hongjun Yan, and Jinfan Zhang. 2013. Anticipated and Repeated Shocks in Liquid Markets. Review of Financial Studies 26: 1890-1912. 19. McLoughlin, Kate, and Jessica Meredith. 2017. The Rise of Chinese Money Market Funds. Reserve Bank of Australia Bulletin March Quarter 2017: 75-83. 20. Perry, Emily, and Florian Weltewitz. 2015. Wealth Management Products in China. Reserve Bank of Australia Bulletin June Quarter 2015: 59-67. 21. Porter, Nathan, and TengTeng Xu. 2009. What Drives China’s Interbank Market? International Monetary Fund Working Paper. 22. Shevlin, Aidan, and Andy Chang. 2015. China’s Repo Markets. J. P. Morgan Asset Management. 23. Xu, Xiaoqing Eleanor, and Tao Chen. 2012. The Effect of Monetary Policy on Real Estate Price Growth in China. Pacific-Basin Finance Journal 20(1): 62-77. 24. Xu, Zheng. 2007. China’s Money Markets. China's Financial Markets: An Insider's Guide to How the Markets Work. Elsevier Academic Press. 37
Table 1. Annual Trading Volume of China Interbank Money Market from 2001 to 2016 Panel A. China Interbank Lending Volume (in 100 Million RMB) Year 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2,016 2016*
Overnight 1,039 2,015 6,419 2,833 2,230 6,352 80,305 106,514 161,666 244,862 273,200 402,814 289,636 294,983 539,953 839,762 87.6%
7 Days 5,607 8,523 14,563 10,414 8,963 12,904 21,780 35,005 21,348 24,269 42,401 41,934 44,024 61,061 76,974 92,764 9.7%
14 Days
1,468 2,736 4,744 5,978 5,061 9,986 12,068 11,579 11,767 15,305 12,771 1.3%
21 Days 934 1,003 566 307 604 381 502 1,107 1,022 650 2,283 2,370 1,828 899 1,372 2,210 0.2%
1 Month 353 292 441 189 299 191 342 1,135 2,048 1,613 2,705 4,476 5,070 4,665 4,243 4,463 0.5%
2 Months 3 Months 4 Months 6 Months 9 Months 94 47 9 108 48 118 101 102 28 92 58 26 75 141 15 120 52 14 279 318 133 31 24 445 666 185 292 213 538 710 62 97 13 466 1,340 198 185 30 1,120 1,674 351 601 39 1,626 1,170 81 379 29 1,034 1,748 67 119 2 1,237 1,670 60 100 22 1,006 2,445 120 146 17 2,130 3,477 264 510 260 0.2% 0.4% 0.0% 0.1% 0.0%
1 Year
16 185 23 10 54 97 83 163 553 523 0.1%
Total 8,082 12,107 22,220 13,920 12,328 21,484 106,466 150,492 193,505 278,684 334,412 467,044 355,190 376,626 642,135 959,134 100.0%
1 Month 917 1,335 1,729 2,332 2,311 3,197 4,932 7,350 4,200 8,735 13,804 13,155 24,745 22,896 18,661 23,673 0.42%
2 Months 3 Months 4 Months 6 Months 9 Months 490 251 19 40 25 798 440 216 108 23 925 633 89 143 0 904 696 145 132 36 788 535 55 211 87 561 458 9 42 115 1,513 907 143 138 21 1,086 1,046 133 155 77 1,015 1,360 284 181 31 2,852 1,913 392 550 84 5,133 4,513 1,028 1,052 36 8,120 4,421 612 804 89 8,264 7,068 613 1,045 234 6,722 9,854 1,214 1,464 123 5,372 10,193 768 849 60 7,801 9,345 678 742 83 0.14% 0.16% 0.01% 0.01% 0.00%
1 Year 6 55 0 34 149 239 239 82 160 58 225 182 384 311 73 740 0.01%
Total 40,133 101,885 97,418 72,055 84,037 263,021 440,672 563,830 677,007 846,533 966,650 1,366,174 1,519,757 2,124,191 4,324,109 5,682,686 100.00%
Panel B. China Interbank Repo Volume (in 100 Million RMB) Year 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2016*
Overnight
134,255 229,985 360,051 526,452 676,983 728,667 1,109,323 1,201,735 1,669,081 3,700,895 4,861,134 85.54%
7 Days 31,271 84,488 78,331 54,209 61,599 98,269 158,416 150,263 104,013 120,619 157,023 172,165 196,620 300,413 461,541 618,754 10.89%
14 Days 5,169 12,257 13,382 10,701 14,531 22,355 38,464 36,413 32,745 29,016 43,834 47,390 64,787 96,061 114,361 138,333 2.43%
21 Days 1,945 2,166 2,186 2,866 3,771 3,521 5,913 7,173 6,566 5,331 11,335 9,913 14,263 16,051 11,337 21,403 0.38%
* Market share of each maturity sector as a percentage of the total trading volume across all maturities. Source: China Foreign Exchange Trading System; The People's Bank of China
Table 2. China Interbank Lending and Repo Trading by Types of Institutions (December 2016) Type of Institution
Number of Deals
Trading Value (100M RMB)
Trading Value (in %)
Average Trade Size (100RMB)
Weighted Average Balance (100M Repo Rate (%) RMB)
Balance (in %)
1948 4129 2745 3177 2098 4259 18356
23166.17 62515.49 20441.36 6979.78 5577.86 32009.54 150690.20
15.37% 41.49% 13.57% 4.63% 3.70% 21.24% 100.00%
11.89 15.14 7.45 2.20 2.66 7.52 8.21
2.4037 2.3721 2.4294 2.3889 2.7522 2.5577 2.4390
2688.69 2984.90 2146.63 925.70 1364.99 3500.05 13610.95
19.75% 21.93% 15.77% 6.80% 10.03% 25.71% 100.00%
12794 10992 45356 17976 44566 108084 239768
88709.23 88615.96 175903.93 27660.75 100366.71 234804.30 716060.86
12.39% 12.38% 24.57% 3.86% 14.02% 32.79% 100.00%
6.93 8.06 3.88 1.54 2.25 2.17 2.99
2.4745 2.3909 2.4605 2.9106 2.5724 2.8255 2.6064
12789.89 5469.24 15296.71 4536.31 12620.06 42919.56 93631.77
13.66% 5.84% 16.34% 4.84% 13.48% 45.84% 100.00%
839 480 8140 839 7089 31407 48794
2403.84 544.94 10859.39 1034.57 8012.64 27974.46 50829.84
4.73% 1.07% 21.36% 2.04% 15.76% 55.04% 100.00%
2.87 1.14 1.33 1.23 1.13 0.89 1.04
2.5095 2.9156 2.7978 3.1349 3.0630 3.0998 3.0002
382.31 141.33 2377.02 309.44 2577.69 4688.24 10476.03
3.65% 1.35% 22.69% 2.95% 24.61% 44.75% 100.00%
Panel A. Interbank Lending Large Commercial Banks Joint Stock Commercial Bank Urban Commercial Bank Foreign Capital Institutions Rural Commercial Bank & Co-operative Others Total
Panel B. Interbank Pledged Repo Large Commercial Banks Joint Stock Commercial Banks Urban Commercial Banks Foreign Capital Institutions Rural Commercial Bank & Co-operative Others Total
Panel C. Interbank Outright Repo Large Commercial Banks Joint Stock Commercial Banks Urban Commercial Banks Foreign Capital Institutions Rural Commercial Bank & Co-operative Others Total
Note: The above summary is for the month of December 2016. Source: China Foreign Exchange Trading System; The People's Bank of China
Table 3. Interbank Repo vs. Exchange Repo Trading Volume (2007-2016) Date
Overnight
7 Days
14 Days
1 Month
3 Months
Total
904 1,045 1,355 1,913 4,520 4,421 7,042 9,594 9,459 9,346 0.17%
426,884 555,396 668,813 855,778 948,221 1,346,667 1,483,272 2,062,358 4,135,375 5,642,484 100.00%
46 11 4 1 21 23 95 184 116 84 0.00%
17,975 24,269 35,476 65,867 191,072 324,225 538,858 760,912 1,125,807 2,108,057 100.00%
Panel A. Trading Volume ( in 100 Million RMB) of Interbank Repos 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2016*
226,268 360,219 526,360 695,352 728,848 1,109,389 1,192,619 1,648,962 3,550,888 4,853,985 86.03%
157,527 150,370 104,090 120,719 157,161 172,237 194,493 297,281 452,014 617,318 10.94%
37,248 36,395 32,773 29,037 43,865 47,452 64,458 84,979 105,797 138,164 2.45%
4,937 7,366 4,234 8,757 13,827 13,167 24,660 21,542 17,217 23,670 0.42%
Panel B. Trading Volume ( in 100 Million RMB) of SSE Exchange Repos 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2016*
5,337 13,738 22,000 44,179 154,823 281,219 465,771 666,906 1,022,034 1,957,928 92.88%
11,441 10,339 13,455 21,619 33,613 38,791 62,025 79,826 91,161 132,052 6.26%
922 172 15 55 1,849 3,558 8,856 11,049 10,190 15,015 0.71%
228 9 3 13 766 633 2,111 2,948 2,306 2,979 0.14%
Panel C. Ratio of Interbank Repo Trading Volume to SSE Exchange Repo Trading Volume 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016
42.39 26.22 23.93 15.74 4.71 3.94 2.56 2.47 3.47 2.48
13.77 14.54 7.74 5.58 4.68 4.44 3.14 3.72 4.96 4.67
40.40 211.06 2256.78 526.15 23.73 13.34 7.28 7.69 10.38 9.20
21.63 818.38 1671.60 674.33 18.06 20.79 11.68 7.31 7.47 7.95
19.54 99.23 367.95 2046.43 210.24 190.34 74.08 52.19 81.65 111.77
23.75 22.89 18.85 12.99 4.96 4.15 2.75 2.71 3.67 2.68
* Market share of each maturity sector as a % of the total trading volume across all five maturities. Source: Bloomberg.
Table 4. Descriptive and Comparative Statistics of China Money Market Rates (October 2006-October 2016) Panel A. Descriptive Statistics of Monthly China Money Market Rates (in %) Name of China Money Market Rate Shanghai Interbank Offered Rate (SHIBOR)
Interbank Repo Rate
Shanghai Stock Exchange (SSE) Repo Rate
Tenor 1 Day 7 Days 1 Day 7 Days 1 Day 7 Days 1 Day 7 Days 1 Day 7 Days 1 Day 7 Days 1 Day 7 Days
a
Field Close Close Close Close High High High-Low High-Low Close Close High High High-Low High-Low
First Half Second Half Full Sample (10/2006-09/2011) (10/2011-10/2016) (10/2006-10/2016) c ACF Mean Median Stdev Mean Median Stdev Mean Median Stdev 2.119 1.926 1.016 2.765 2.608 0.915 2.445 2.289 1.016 0.673 2.646 2.500 1.414 3.504 3.394 1.046 3.079 2.874 1.309 0.736 2.135 1.925 1.015 2.820 2.700 0.928 2.481 2.330 1.028 0.679 2.628 2.550 1.355 3.531 3.450 1.071 3.083 2.850 1.297 0.654 4.005 3.520 2.263 5.430 5.000 3.804 4.723 4.500 3.204 0.312 4.445 3.850 2.673 5.709 5.500 3.429 5.082 4.600 3.130 0.359 2.562 1.850 2.064 3.672 2.900 3.610 3.122 2.590 2.986 0.249 2.846 2.112 2.466 3.487 3.070 3.103 3.169 2.600 2.812 0.294 1.845 1.483 1.259 2.721 2.100 2.806 2.287 1.615 2.216 0.176 2.261 1.685 2.246 2.538 2.805 1.332 2.400 2.050 1.840 0.124 15.945 8.450 21.648 20.024 14.060 14.561 18.002 11.950 18.456 0.400 7.486 4.500 8.512 7.149 6.000 3.538 7.316 5.480 6.474 0.484 15.121 7.743 21.651 19.538 13.730 14.528 17.347 10.745 18.464 0.397 6.383 2.998 8.555 5.924 4.950 3.699 6.152 4.190 6.548 0.425
Panel B. Comparison between Monthly Money Market Ratesb First Half (10/2006-09/2011) Pair for Comparison SHIBOR and China Interbank Repo Rate
Tenor
Fielda
1 Day
Close
7 Days 1 Day China Interbank Repo Rate 7 Days and SSE Exchange Repo 1 Day Rate 7 Days 1 Day 7 Days 7-Day and 1-Day SHIBOR 7 Day-1 Day 7 Day-1 Day 7-Day and 1-Day China Interbank Repo Rates 7 Day-1 Day 7-Day and 1-Day SSE Exchange Repo Rates a
Close Close Close High High High-Low High-Low Close Close High Close High
Diff. in Mean
Diff. in Diff. in Median Stdev
Second Half (10/2011-10/2016) Diff. in Mean
Diff. in Median
-0.016
0.001
0.000
-0.055
-0.092
0.018 0.290 0.367 -11.940 -3.042 -12.558 -3.537 0.527 0.493 0.440 0.416 -8.459
-0.050 0.443 0.865 -4.930 -0.650 -5.893 -0.886 0.574 0.625 0.330 0.203 -3.950
0.058 -0.243 -0.891 -19.385 -5.839 -19.587 -6.089 0.398 0.340 0.410 0.987 -13.137
-0.027 0.099 0.993 -14.594 -1.440 -15.866 -2.436 0.739 0.711 0.279 -0.183 -12.875
-0.056 0.600 0.645 -9.060 -0.500 -10.830 -1.880 0.786 0.750 0.500 0.705 -8.060
Full Sample (10/2006-10/2016)
Diff. in Stdev -0.013
Diff. in Mean -0.036
-0.026 -0.005 -1.878 0.194 -0.261 0.683 -10.758 -13.278 -0.109 -2.234 -10.918 -14.226 -0.596 -2.982 0.130 0.634 0.143 0.603 -0.375 0.359 -1.474 0.114 -11.023 -10.685
Diff. in Diff. in Median Stdev -0.041
-0.012
0.024 0.012 0.715 -1.188 0.800 -0.543 -7.450 -15.253 -0.880 -3.344 -8.155 -15.477 -1.590 -3.736 0.585 0.294 0.520 0.270 0.100 -0.074 0.435 -0.375 -6.470 -11.983
A field of "Close" refers to the monthly closing level of the interest rate, and a field of "High" and "Low" refers to the monthly highest and Lowest level of the interest rate. b The differences in Mean, Median, and Standard Deviation for each pair of money market rates are tested for significance. Bold - Significant at 5%. Bold and Italic - Significant at 10%. c ACF refers to the first order autocorrelation. d ADF Refers to the Augmented Dickey Fuller Unit Root Test. The ADF critical values for 5% and 10% significance are -2.89 and -2.58, respectively.
d
ADF -2.740 -4.223 -4.747 -3.633 -7.879 -7.458 -8.445 -8.020 -9.090 -3.195 -3.153 -2.666 -3.159 -2.794
Table 5. Measures of China's Monetary Policy and Their Correlations with Repo Rates (Common Sample: February 2008 to October 2016) * Symbol
Names of China Monetary Policy Variables
CDR
CSR
CLR
CRRI
1-year Savings 1-year Interest PBOC Deposit Lending Rate on Rediscount Benchmark Benchmark Required Rate Rate Rate Reserves
Pane A. Descriptive Statistics of the Monetary Policy Variables Mean 2.331 2.724 Median 2.250 3.000 Maximum 4.320 4.140 Minimum 1.800 1.500 Std. Dev. 0.645 0.725 ADF # -4.076 -1.776
5.738 6.000 7.470 4.350 0.814 -1.812
1.643 1.620 1.890 1.620 0.076 -3.218
CERI
CMP
CMPD
CRR
CRRD
COMO
COMOA
China Monetary Interest Policy Composite PBOC Open Market Rate on Monthly Required Monthly Operations Net Rate Indicator Excess (CMP=CDR+CSR+C change in Reserve change Injection to Markets LR+CRRI+CERI) Reserves CMP Ratio in CRR in 100 Billion RMB
0.743 0.720 0.990 0.720 0.076 -3.218
Absolute size of COMO
13.179 13.590 18.810 10.440 2.122 -2.954
-0.080 0.000 0.950 -4.050 0.481 -6.427
18.457 18.500 21.500 15.000 1.913 -1.674
0.019 0.000 1.000 -1.500 0.317 -4.707
0.288 0.160 6.900 -4.201 1.491 -11.412
0.955 0.590 6.900 0.000 1.176 -1.366
Panel B. Correlation between the Monetary Policy Variables and the 7-day Interbank Repo Rate Monthly Closing 0.133 0.467 0.420 -0.031 -0.031 Monthly High 0.051 0.331 0.293 -0.073 -0.073 Monthly High- Low -0.008 0.270 0.240 -0.105 -0.105 Change in Monthly Closing -0.026 -0.033 -0.030 -0.011 -0.011 Change in Monthly High -0.028 -0.020 -0.020 -0.027 -0.027
0.359 0.236 0.174 -0.031 -0.025
0.149 0.115 0.115 0.023 0.024
0.744 0.550 0.463 0.002 0.018
0.162 0.055 0.068 0.189 0.080
0.083 0.105 0.103 0.053 0.145
-0.158 -0.080 -0.070 0.010 0.084
Panel C. Correlation between the Monetary Policy Variables and the 7-day SSE Exchange Repo Rate Monthly Closing 0.076 0.245 0.209 -0.026 -0.026 Monthly High 0.091 0.267 0.212 -0.045 -0.045 Monthly High- Low 0.059 0.217 0.165 -0.062 -0.062 Change in Monthly Closing -0.053 -0.039 -0.041 -0.043 -0.043 Change in Monthly High -0.102 -0.047 -0.051 -0.105 -0.105
0.185 0.197 0.151 -0.048 -0.074
0.089 0.104 0.092 0.032 0.032
0.463 0.507 0.453 0.029 0.056
0.121 -0.031 -0.056 0.138 0.033
0.118 0.038 0.025 0.202 0.127
-0.088 -0.041 -0.020 0.123 0.108
Panel D. Correlation between the Monetary Policy Variables with the 7-day Interbank minus SSE Exchange Repo Spread Spread on Monthly Closing 0.055 0.215 0.203 -0.006 -0.006 0.168 0.059 Spread on Monthly High -0.053 0.026 0.051 -0.023 -0.023 0.011 -0.003 Spread on Monthly High-Low -0.027 0.033 0.063 0.016 0.016 0.029 0.013
0.277 -0.029 -0.078
0.043 0.092 0.083
-0.028 0.062 0.073
-0.068 -0.033 -0.001
Notes: All rates are in %. # ADF Refers to the Augmented Dickey Fuller Unit Root Test. The ADF critical values for 5% and 10% significance are -2.89 and -2.58, respectively. Bold - Significant at 5%. Bold and Italic - Significant at 10%. * The above statistics is based on the common sample period of all mometary policy variables from February 2008 to October 2016.
Table 6. Descriptive Statistics of Potential Drivers of Repo Rates Symbol
Names of Potential Drivers
Mean Median
Panel A. Full Sample (10/2006-10/2016) China monetary policy composite rate indicator # CMP CRRD Monthly change in China required reserve ratio COMO PBOC's net open market operations (in 100 bil. RMB) CSB China's shadow banking financing (in 100 bil. RMB) LCSB Log of CSB CSBTD CSB as a % of China Total Bank Deposits CSBM Monthly Growth Rate of CSB CPOA China's equity new issuance (in 100 bil. RMB) CBI China's bond new issuance (in 100 bil. RMB) CHMF Hot money flow to China (in billion RMB) CHPRY China's home price growth rate (YOY)
13.56 0.07 0.288 119.12 4.43 12.44 2.64 0.224 6.79 -5.36 3.62
Panel B. First Half (10/2006-09/2011) China monetary policy composite rate indicator # CMP CRRD Monthly change in China required reserve ratio COMO PBOC's net open market operations (in 100 bil. RMB) CSB China's shadow banking financing (in 100 bil. RMB) LCSB Log of CSB CSBTD CSB as a % of China Total Bank Deposits CSBM Monthly Growth Rate of CSB CPOA China's equity new issuance (in 100 bil. RMB) CBI China's bond new issuance (in 100 bil. RMB) CHMF Hot money flow to China (in billion RMB) CHPRY China's home price growth rate (YOY)
14.34 0.22 0.013 46.43 3.63 7.85 4.01 0.365 6.80 17.14 5.49
Panel C. Second Half (10/2011-10/2016) China monetary policy composite rate indicator # CMP CRRD Monthly change in China required reserve ratio COMO PBOC's net open market operations (in 100 bil. RMB) CSB China's shadow banking financing (in 100 bil. RMB) LCSB Log of CSB CSBTD CSB as a % of China Total Bank Deposits CSBM Monthly Growth Rate of CSB CPOA China's equity new issuance (in 100 bil. RMB) CBI China's bond new issuance (in 100 bil. RMB) CHMF Hot money flow to China (in billion RMB) CHPRY China's home price growth rate (YOY)
12.80 -0.07 0.486 190.61 5.22 16.94 1.31 0.085 6.78 -27.49 1.79
#
Max
Min
Stdev
Obs
13.59 18.81 10.44 0.00 1.00 -1.50 0.16 6.90 -4.20 105.74 236.60 11.17 4.66 5.47 2.41 13.41 20.00 3.38 1.94 20.98 -9.89 0.17 1.51 0.00 5.83 18.74 1.24 2.66 90.69 -170.01 4.35 11.48 -6.26
2.26 0.34 1.49 81.88 0.95 5.24 4.22 0.26 3.41 46.28 4.37
121 120 105 121 121 121 120 121 114 121 121
11.70 -1.50 -2.18 11.17 2.41 3.38 -9.89 0.00 1.61 -69.05 -0.68
2.63 0.39 1.06 29.79 0.67 3.07 5.40 0.29 2.84 23.32 3.21
60 59 44 60 60 60 59 60 53 60 60
13.59 14.65 10.44 0.00 0.00 -1.00 0.19 6.90 -4.20 215.49 236.60 104.66 5.37 5.47 4.65 17.12 20.00 13.22 0.98 6.86 -1.43 0.05 0.63 0.00 5.61 18.74 1.24 -27.07 90.69 -170.01 1.00 10.11 -6.26
1.48 0.20 1.72 45.36 0.27 2.02 1.87 0.11 3.85 52.39 4.60
61 61 61 61 61 61 61 61 61 61 61
14.15 18.81 0.00 1.00 0.06 2.41 33.99 107.81 3.53 4.68 7.16 13.70 3.84 20.98 0.31 1.51 6.13 13.52 19.11 61.20 5.34 11.48
China Monetary Policy Composite Rate Indicator (CMP) is the sum of PBOC Rediscount Rate, 1-year Benchmark Deposit Rate, 1-year Benchmark Lending Rate, Interest Rate for Required Reserves, and Interest Rate for Excess Reserves.
Table 7. Monthly Regressions for China's Interbank Repo Rates (October 2006-October 2016) Panel A: Dependent Variable - 7-Day China Internank Repo Rate
Explanatory Variables \ Dependent Variable Constant Lag 1 of the dependent variable (AR1) China's monetary policy composite rate indicator # Monthly change in China's bank required reserve ratio PBOC's net open market operations (in 100 bil. RMB) Log of China's shadow banking financing (in 100 bil. RMB) China's equity new issuance (in 100 bil. RMB) China's bond new issuance (in 100 bil. RMB) Hot money flow to China (in billion RMB) China's home price growth index Quarter-end Indicator R-squared Adjusted R-squared F-statistic
Full Sample (10/2006-10/2016) Monthly Close Monthly High -4.089 [-4.58] -9.604 [-2.97] 0.362 [3.67] 0.195 [2.05] 0.190 [3.52] 0.488 [2.54] 0.644 [2.28] 0.779 [1.84]
First Half (10/2006-09/2011) Monthly Close Monthly High -3.899 [-2.91] -6.091 [-2.41] 0.392 [2.31] 0.272 [1.70] 0.155 [2.77] 0.307 [2.48] 0.820 [2.91] 1.528 [2.92]
Monthly Close -7.983 [-1.75] 0.134 [1.47] 0.578 [3.60] 0.181 [0.27]
0.748 [5.90] -0.186 [-0.44]
1.436 [4.84] 1.515 [1.09]
0.862 [3.25] 0.020 [0.04]
1.228 [2.88] 3.404 [2.28]
0.615 [1.04] 0.630 [0.95]
0.003 [1.57] 0.008 [0.46] 0.424 [3.19] 57.9% 54.8% 19.057
0.002 [0.47] 0.018 [0.26] 0.795 [1.32] 26.9% 21.7% 5.118
0.004 [0.80] -0.031 [-0.79] 0.175 [0.91] 59.3% 52.8% 9.096
-0.005 [-0.59] -0.179 [-2.41] -0.073 [-0.17] 49.8% 41.8% 6.212
-0.004 [-1.24] 0.066 [2.74] 0.532 [3.11] 56.6% 49.9% 8.474
First Half (10/2006-09/2011) Monthly Close Monthly High -3.304 [-3.43] -7.181 [-2.84] 0.457 [4.21] 0.246 [1.64] 0.133 [3.74] 0.343 [3.34] 0.503 [2.78] 0.910 [1.50]
Monthly Close -2.636 [-0.70] 0.321 [2.37] 0.341 [2.87] 0.124 [0.31]
Second Half (10/2011-10/2016) Monthly High Monthly Close -19.430 [-1.27] -5.115 [-1.41] 0.061 [0.67] 0.097 [0.99] 1.263 [2.15] 0.467 [3.44] -0.332 [-0.51] 0.133 [0.19] -0.012 [-0.26] 1.461 [0.82] 0.427 [0.84] -0.515 [-0.20] 0.869 [1.16] -0.055 [-1.35] -0.009 [-0.76] -0.004 [-1.19] 0.156 [1.37] 0.075 [2.71] 1.517 [1.36] 0.533 [3.02] 23.8% 58.0% 12.1% 49.6% 2.030 6.910
Monthly High -1.542 [-0.12] 0.062 [0.85] 0.453 [1.06] -0.811 [-1.19] 0.138 [0.77] 0.424 [0.30] 1.096 [0.42] -0.362 [-1.58] -0.009 [-0.67] 0.198 [1.40] 1.621 [1.46] 30.6% 16.8% 2.209
Second Half (10/2011-10/2016) Monthly High Monthly Close -5.597 [-0.34] -1.057 [-0.32] 0.058 [0.66] 0.294 [2.07] 1.019 [1.64] 0.282 [2.68] -0.743 [-0.89] 0.088 [0.22] -0.002 [-0.04] -0.620 [-0.31] -0.142 [-0.30] -2.089 [-0.69] -0.679 [-1.11] -0.029 [-1.00] -0.009 [-0.69] -0.005 [-2.17] 0.182 [1.42] 0.052 [1.96] 1.380 [1.07] 0.620 [2.99] 20.8% 54.7% 8.6% 45.6% 1.704 6.036
Monthly High 13.777 [0.98] 0.063 [0.89] 0.117 [0.26] -1.335 [-1.46] 0.226 [1.08] -1.669 [-1.05] -0.264 [-0.09] -0.411 [-1.58] -0.008 [-0.57] 0.225 [1.45] 1.509 [1.19] 28.4% 14.1% 1.988
Panel B: Dependent Variable - 1-Day (Ovenight) China Internank Repo Rate
Explanatory Variables \ Dependent Variable Constant Lag 1 of the dependent variable (AR1) China's monetary policy composite rate indicator # Monthly change in China's bank required reserve ratio PBOC's net open market operations (in 100 bil. RMB) Log of China's shadow banking financing (in 100 bil. RMB) China's equity new issuance (in 100 bil. RMB) China's bond new issuance (in 100 bil. RMB) Hot money flow to China (in billion RMB) China's home price growth index Quarter-end Indicator R-squared Adjusted R-squared F-statistic
Full Sample (10/2006-10/2016) Monthly Close Monthly High -2.793 [-4.23] -9.061 [-2.91] 0.462 [5.24] 0.155 [1.73] 0.140 [3.81] 0.500 [2.95] 0.433 [2.23] 0.296 [0.63] 0.459 [4.46] -0.274 [-0.99]
1.333 [4.19] -0.348 [-0.27]
0.670 [3.59] 0.026 [0.10]
1.469 [3.21] 1.243 [0.97]
-0.033 [-0.06] -0.808 [-1.35]
0.001 [0.40] 0.014 [0.85] 0.505 [3.64] 61.4% 58.6% 22.041
0.003 [0.58] 0.060 [0.81] 0.693 [1.03] 22.0% 16.4% 3.914
0.002 [0.68] -0.024 [-1.04] 0.330 [1.91] 69.2% 64.3% 14.067
-0.008 [-0.99] -0.095 [-1.29] -0.070 [-0.17] 41.6% 32.3% 4.460
-0.005 [-2.30] 0.046 [1.85] 0.621 [3.04] 54.2% 47.1% 7.684
Notes: The monthly close (closing rates) and high (highest rates) of China Interbank Repo Rates are from Bloomberg. # China Monetary Policy Composite Rate Indicator is the sum of PBOC Rediscount Rate, 1-year Benchmark Savings Deposit Rate, 1-year Benchmark Lending Rate, Interest Rate for Required Reserves, and Interest Rate for Excess Reserves. The t-statistics in brackets are computed using the Newey-West (1987) robust standard errors. Bold -- Significant at 5%; Bold and Italic -- Significant at 10%
Table 8. Monthly Regressions for China's Shanghai Stock Exchange Repo Rates (October 2006-October 2016) Panel A: Dependent Variable - 7-Day China SSE Exchange Repo Rate
Explanatory Variables \ Dependent Variable Constant Lag 1 of the dependent variable (AR1) China's monetary policy composite rate indicator # Monthly change in China's bank required reserve ratio PBOC's net open market operations (in 100 bil. RMB) Log of China's shadow banking financing (in 100 bil. RMB) China's equity new issuance (in 100 bil. RMB) China's bond new issuance (in 100 bil. RMB) Hot money flow to China (in billion RMB) China's home price growth index Quarter-end Indicator R-squared Adjusted R-squared F-statistic
Full Sample (10/2006-10/2016) Monthly Close Monthly High -3.145 [-1.93] -13.413 [-1.78] 0.060 [1.16] 0.255 [2.09] 0.233 [3.16] 0.764 [1.85] 0.728 [1.65] 1.627 [0.97]
First Half (10/2006-09/2011) Monthly Close Monthly High -1.436 [-0.53] 0.576 [0.08] 0.050 [0.60] 0.097 [0.66] 0.175 [2.04] 0.548 [1.75] 0.766 [1.14] 3.352 [2.20]
Monthly Close -8.583 [-1.44] -0.031 [-0.29] 0.291 [1.60] 1.034 [1.62]
0.514 [1.75] 0.440 [0.46]
1.337 [1.60] 10.892 [1.80]
0.181 [0.32] 0.620 [0.49]
-1.746 [-1.28] 18.242 [3.21]
1.556 [1.79] -1.028 [-0.55]
-0.001 [-0.15] 0.006 [0.16] -0.651 [-1.37] 13.4% 7.2% 2.151
-0.010 [-0.91] -0.145 [-1.08] 1.610 [1.81] 43.9% 39.8% 10.854
-0.012 [-0.79] 0.054 [0.78] -0.225 [-0.24] 14.3% 0.6% 1.043
-0.007 [-0.34] -0.515 [-1.92] 1.100 [0.75] 61.9% 55.8% 10.150
0.005 [0.79] -0.031 [-0.59] -0.873 [-2.31] 27.7% 16.5% 2.485
First Half (10/2006-09/2011) Monthly Close Monthly High -3.534 [-1.48] -16.918 [-1.35] 0.017 [0.12] 0.161 [1.56] 0.158 [1.73] 1.477 [2.36] 0.370 [1.29] 9.951 [2.56]
Monthly Close 1.165 [0.12] -0.011 [-0.14] -0.082 [-0.25] 1.105 [1.41]
Second Half (10/2011-10/2016) Monthly High Monthly Close -70.121 [-2.52] -4.983 [-0.94] -0.029 [-0.20] -0.020 [-0.18] 2.018 [2.78] 0.108 [0.53] 0.329 [0.21] 0.862 [1.22] 0.108 [1.54] 9.846 [2.56] 1.409 [1.82] -0.394 [-0.11] -0.645 [-0.30] -0.090 [-1.90] 0.003 [0.38] 0.006 [0.86] -0.103 [-1.06] -0.024 [-0.49] 1.839 [2.32] -0.851 [-2.16] 42.7% 32.1% 33.9% 18.5% 4.840 2.364
Monthly High -50.407 [-2.25] -0.141 [-1.06] 0.986 [1.53] -0.261 [-0.19] 0.172 [0.94] 9.379 [3.05] 2.491 [0.74] -0.539 [-2.55] 0.005 [0.62] -0.057 [-0.76] 1.858 [2.95] 55.9% 47.1% 6.334
Second Half (10/2011-10/2016) Monthly High Monthly Close -187.755 [-2.20] 2.327 [0.28] -0.060 [-0.49] -0.027 [-0.31] 5.185 [2.24] -0.215 [-0.81] 2.715 [0.39] 0.770 [0.97] 0.291 [1.54] 27.198 [2.33] 0.862 [0.65] -8.406 [-0.60] -5.708 [-2.06] -0.096 [-0.91] 0.046 [1.40] 0.022 [1.69] -0.234 [-0.48] 0.022 [0.25] 10.016 [2.54] 0.960 [0.94] 30.6% 24.2% 19.9% 9.0% 2.865 1.597
Monthly High -116.562 [-1.24] -0.125 [-1.10] 1.694 [0.55] 0.471 [0.07] 0.663 [0.62] 24.312 [2.16] -0.255 [-0.02] -1.698 [-2.06] 0.052 [1.41] -0.064 [-0.13] 10.183 [2.79] 38.6% 26.3% 3.146
Panel B: Dependent Variable - 1-Day China SSE Exchange Repo Rate
Explanatory Variables \ Dependent Variable Constant Lag 1 of the dependent variable (AR1) China's monetary policy composite rate indicator # Monthly change in China's bank required reserve ratio PBOC's net open market operations (in 100 bil. RMB) Log of China's shadow banking financing (in 100 bil. RMB) China's equity new issuance (in 100 bil. RMB) China's bond new issuance (in 100 bil. RMB) Hot money flow to China (in billion RMB) China's home price growth index Quarter-end Indicator R-squared Adjusted R-squared F-statistic
Full Sample (10/2006-10/2016) Monthly Close Monthly High -3.614 [-1.99] -50.394 [-2.52] 0.044 [0.91] 0.253 [3.22] 0.152 [1.73] 2.348 [2.25] 0.399 [1.16] 5.320 [1.14] 0.838 [3.13] -1.407 [-1.54]
5.374 [2.52] 24.899 [1.77]
0.892 [1.98] 0.081 [0.16]
-0.124 [-0.07] 44.115 [3.30]
0.659 [0.49] -6.117 [-2.39]
0.011 [1.77] 0.049 [0.88] 0.538 [1.04] 17.4% 11.5% 2.931
-0.048 [-1.12] -0.202 [-0.52] 8.454 [2.69] 38.0% 33.6% 8.516
-0.003 [-0.33] -0.063 [-1.35] 0.336 [0.88] 19.3% 6.4% 1.497
-0.232 [-3.37] -1.187 [-1.95] 6.394 [1.80] 68.4% 63.3% 13.526
0.020 [1.54] 0.022 [0.28] 0.912 [0.89] 20.6% 8.4% 1.690
Notes: The monthly close (closing rates) and high (highest rates) of Shanghai Stock Exchange (SSE) Repo Rates are from Bloomberg. # China Monetary Policy Composite Rate Indicator is the sum of PBOC Rediscount Rate, 1-year Benchmark Savings Deposit Rate, 1-year Benchmark Lending Rate, Interest Rate for Required Reserves, and Interest Rate for Excess Reserves. The t-statistics in brackets are computed using the Newey-West (1987) robust standard errors. Bold -- Significant at 5%; Bold and Italic -- Significant at 10%
Table 9. Monthly Regressions for China's Interbank to SSE Exchange Repo Rate Spread (October 2006-October 2016) Panel A: Dependent Variable: 7-Day China Internank to SSE Exchange Repo Rate Spread Full Sample (10/2006-10/2016)
Explanatory Variables \ Dependent Variable Constant Lag 1 of the dependent variable (AR1) China's monetary policy composite rate indicator # Monthly change in China's bank required reserve ratio PBOC's net open market operations (in 100 bil. RMB) Log of China's shadow banking financing (in 100 bil. RMB) China's equity new issuance (in 100 bil. RMB) China's bond new issuance (in 100 bil. RMB) Hot money flow to China (in billion RMB) China's home price growth index Quarter-end Indicator R-squared Adjusted R-squared F-statistic
Monthly Close RP07-RPS07 -2.999 [-2.14] -0.044 [-0.60] 0.064 [1.10] -0.117 [-0.24]
Monthly High RP07H-RPS07H 5.118 [0.90] 0.169 [1.56] -0.376 [-1.38] -0.830 [-0.54]
0.621 [2.21] -0.840 [-0.86]
0.020 [0.03] -10.156 [-2.11]
0.007 [1.38] 0.003 [0.08] 0.920 [2.00] 14.6% 8.5% 2.381
0.013 [1.26] 0.169 [1.44] -0.654 [-0.77] 38.7% 34.3% 8.752
First Half (10/2006-09/2011) Second Half (10/2011-10/2016) Interbank to Exchange Repo Rate Spread based on Monthly Close Monthly High Monthly Close Monthly High Monthly Close RP07-RPS07 RP07H-RPS07H RP07-RPS07 RP07H-RPS07H RP07-RPS07 -5.123 [-1.89] -9.068 [-1.45] 0.621 [0.12] 50.273 [2.79] 0.462 [0.11] -0.084 [-0.92] -0.021 [-0.15] -0.177 [-1.44] -0.070 [-0.98] -0.156 [-1.21] 0.095 [1.55] -0.203 [-0.89] 0.392 [1.83] -0.653 [-1.04] 0.417 [1.99] 0.035 [0.06] -1.865 [-1.33] -0.782 [-0.85] -0.847 [-0.53] -0.690 [-0.74] -0.099 [-1.22] 1.251 [2.12] 3.657 [2.65] -1.045 [-1.46] -8.478 [-3.87] -1.105 [-1.67] -0.674 [-0.58] -15.028 [-3.54] 1.324 [0.58] -0.039 [-0.01] 1.256 [0.51] 0.024 [0.44] 0.019 [1.38] 0.006 [0.36] -0.009 [-1.24] -0.013 [-0.90] -0.010 [-1.30] -0.107 [-1.81] 0.288 [1.38] 0.108 [2.23] 0.285 [1.90] 0.109 [2.20] 0.280 [0.35] -1.017 [-0.86] 1.170 [2.90] -0.469 [-0.42] 1.183 [2.82] 19.6% 64.3% 35.6% 24.6% 36.8% 6.7% 58.5% 25.7% 13.0% 24.1% 1.521 11.237 3.597 2.124 2.910
Monthly High RP07H-RPS07H 43.255 [2.39] -0.094 [-1.31] -0.299 [-0.52] -0.690 [-0.44] -0.054 [-0.24] -8.194 [-3.73] -0.815 [-0.20] 0.161 [0.94] -0.014 [-0.87] 0.274 [1.63] -0.522 [-0.45] 25.8% 10.9% 1.735
Panel B: Dependent Variable: 1-Day China Internank to SSE Exchange Repo Rate Spread Full Sample (10/2006-10/2016)
Explanatory Variables \ Dependent Variable Constant Lag 1 of the dependent variable (AR1) China's monetary policy composite rate indicator # Monthly change in China's bank required reserve ratio PBOC's net open market operations (in 100 bil. RMB) Log of China's shadow banking financing (in 100 bil. RMB) China's equity new issuance (in 100 bil. RMB) China's bond new issuance (in 100 bil. RMB) Hot money flow to China (in billion RMB) China's home price growth index Quarter-end Indicator R-squared Adjusted R-squared F-statistic
First Half (10/2006-09/2011) Second Half (10/2011-10/2016) Interbank to Exchange Repo Rate Spread based on Monthly Close Monthly High Monthly Close Monthly High RP01-RPS01 RP01H-RPS01H RP01-RPS01 RP01H-RPS01H -2.081 [-1.65] 8.748 [0.75] -4.201 [-0.47] 178.319 [2.27] 0.001 [0.01] 0.174 [1.78] -0.042 [-0.55] -0.044 [-0.39] 0.092 [1.53] -1.058 [-1.87] 0.553 [1.52] -3.994 [-1.78] 0.097 [0.44] -9.127 [-2.38] -1.086 [-1.11] -3.618 [-0.52]
Monthly Close RP01-RPS01 -1.216 [-0.81] 0.029 [0.50] 0.107 [1.36] -0.057 [-0.13]
Monthly High RP01H-RPS01H 41.442 [2.27] 0.270 [3.64] -1.851 [-2.00] -5.071 [-1.10]
-0.039 [-0.19] 1.037 [1.26]
-4.125 [-2.05] -24.825 [-1.85]
0.227 [0.90] -0.046 [-0.13]
1.677 [0.93] -42.338 [-3.34]
-0.755 [-0.62] 5.387 [1.99]
-27.328 [-2.60] 6.663 [0.46]
-0.009 [-1.37] -0.028 [-0.65] -0.126 [-0.25] 5.7% -1.1% 0.833
0.049 [1.15] 0.257 [0.70] -7.797 [-2.54] 37.7% 33.2% 8.393
0.005 [0.51] 0.031 [0.80] -0.064 [-0.19] 8.0% -6.7% 0.544
0.224 [3.31] 1.084 [1.89] -6.587 [-1.90] 70.0% 65.3% 14.616
-0.025 [-1.87] 0.039 [0.56] -0.412 [-0.41] 17.0% 4.3% 1.335
-0.053 [-1.44] 0.429 [0.82] -8.836 [-2.24] 26.9% 15.7% 2.392
The monthly close (closing rates) and high (highest rates) of China interbank and SSE exchange repo rates are from Bloomberg. # China Monetary Policy Composite Rate Indicator is the sum of PBOC Rediscount Rate, 1-year Benchmark Savings Deposit Rate, 1-year Benchmark Lending Rate, Interest Rate for Required Reserves, and Interest Rate for Excess Reserves. The t-statistics in brackets are computed using the Newey-West (1987) robust standard errors. Bold -- Significant at 5%; Bold and Italic -- Significant at 10%
Second Half (10/2011-10/2016) Monthly Close RP01-RPS01 -2.397 [-0.29] -0.047 [-0.61] 0.559 [1.86] -0.815 [-0.82] -0.288 [-1.51] -1.142 [-0.88] 5.200 [1.76] 0.039 [0.36] -0.027 [-1.99] 0.048 [0.61] -0.446 [-0.43] 20.3% 4.4% 1.274
Monthly High RP01H-RPS01H 124.768 [1.32] -0.094 [-0.84] -1.354 [-0.44] -2.084 [-0.31] -0.435 [-0.39] -25.126 [-2.27] 0.725 [0.05] 1.256 [1.70] -0.058 [-1.36] 0.309 [0.53] -9.010 [-2.33] 31.6% 17.9% 2.309
Chart 1. Annual Trading Volume of China Interbank Lending, Interbank Repo, and Exchange Repo (in Bil. RMB)
600,000
Year 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Year 2007 2008
500,000
400,000
300,000
200,000
100,000
0 2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
Interbank Lending Volume Interbank Repo Volume SSE Exchange Repo Volume Source: Bloomberg. Chart 2. China Monetary Policy Interest Rates and Required Reserve Ratio
24 20 16 12
8
8
6 4 2 0 2006 2007
2008
2009
2010
2011
2012
2013
2014
2015
CDR (PBOC Rediscount Rate) CSR (12-month Savings Deposit Benchmark Rate) CLR (12-month Lending Benchmark Rate) CRRI (Interest Rate on Required Reseves) CERI (Interesr Rate on Excess Reserves) CRR (Required Reserve Ratio for Major Banks in China) Source: Bloomberg.
2016
Chart 3. Shadow Banking Financing and its Relative Size to GDP and Bank Deposits Panel A.
90 80 70 60 50 40 30 20 10 0 2010
2011
2012
2013
2014
China Shadow Banking as a % of GDP US Shadow Banking as a % of GDP
Panel B. 2,000
1,600
1,200
800
400
0 2006 2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
China Social Banking Financing (in 100 Billion RMB) China Total Bank Deposits (in 100 Billion RMB)
Panel C.
24 20 16 12 8 50 4 40
0
30 20 10 0 2006 2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
Monthly High of 7-Day China Interbank Repo Rate Monthly High of 7-Day SSE Repo Rate China Shadow Banking as a % of Total Bank Deposits
Data source for Panel A: 2000-2014 Shadow Banking Activitiy Measures are from the Global Shaddow Monitoring Report 2015 by the Financial Stability Board. Data source for Panel B and Panel C: Bloomberg Intelligence
Appendix A. Variable Definitions and Data Sources
Symbol
Variable Name
Data Source
Inception Date in the Sample
Bloomberg Bloomberg Bloomberg Bloomberg Bloomberg Bloomberg Bloomberg Bloomberg Calculation Calculation Calculation Calculation
Oct-2006 Oct-2006 Oct-2006 Oct-2006 Oct-2006 Oct-2006 Oct-2006 Oct-2006 Oct-2006 Oct-2006 Oct-2006 Oct-2006
Calculation
Oct-2006
Calculation People's Bank of China (PBOC) People's Bank of China (PBOC) People's Bank of China (PBOC) People's Bank of China (PBOC) People's Bank of China (PBOC) Calculation People's Bank of China (PBOC) People's Bank of China (PBOC) Calculation Bloomberg Intelligence; PBOC China Securities Regulatory Commission (CSRC) China Securities Regulatory Commission (CSRC) Bloomberg National Bureau of Statistics of China (NBSC) Calculation
Oct-2006 Oct-2006 Oct-2006 Oct-2006 Oct-2006 Oct-2006 Oct-2006 Oct-2006 Feb-2008 Oct-2006 Oct-2006 Oct-2006 May-2007 Oct-2006 Oct-2006 Oct-2006
Panel A. Dependent Variables RP07 RP01 RP07H RP01H RPS07 RPS01 RPS07H RPS01H RP07-RPS07 RP07H-RPS07H RP01-RPS01 RP01H-RPS01H
Monthly Closing of 7-Day Interbank Repo Rate Monthly Closing of 1-Day (Overnight) Interbank Repo Rate Monthly Highest of 7-Day Interbank Repo Rate Monthly Highest of 1-Day Interbank Repo Rate Monthly Closing of 7-Day Shanghai Stock Exchange (SSE) Repo Rate Monthly Closing of 1-Day Shanghai Stock Exchange Repo Rate Monthly Highest of 7-Day Shanghai Stock Exchange Repo Rate Monthly Highest of 1-Day Shanghai Stock Exchange Repo Rate 7-Day Interbank to SSE Exchange Repo Monthly Closing Rate Spread 7-Day Interbank to SSE Exchange Repo Monthly Highest Rate Spread 1-Day Interbank to SSE Exchange Repo Monthly Closing Rate Spread 1-Day Interbank to SSE Exchange Repo Monthly Highest Rate Spread
Panel B. Independent Variables AR1 Lag one month autoregresive term of the dependent variable CMP China's monetary policy composite rate indicator (=CDR+CSR+CLR+CRRI+CERI) CDR: PBOC Rediscount Rate CSR: 1-year Savings Deposit Benchmark Rate CLR: 1-year Lending Benchmark Rate CRRI: Interest Rate on Required Reserves CERI: Interest Rate on Excess Reserves Monthly change in CRR CRRD CRR: China's bank required reserve ratio COMO PBOC's net injection to market from open market operations (in 100 Billion RMB) Log of CSB LCSB CSB: China's shadow banking financing in 100 Billion RMB China's equity new issuance in A shares (in 100 Billion RMB) CPOA China's bond new issuance (in 100 Billion RMB) CBI CHMF Hot money flow to China (in Billion RMB) China's 70 cities new home price year-over-year growth rate CHPRY Quarter-end indicator (1 for Mar., June, Sep., and Dec., and 0 otherwise) QEND
