Chinese Bond Market Challenges

1
Chinese Bond Market Challenges

• Sergey N. Smirnov
• State University Higher School of Economics Moscow

2
Contents
Chinese Bonds Market Overview
Zero-coupon yield curve used by CGSDTC
Applying EFFAS-EBC methodology to Chinese Bonds
Market
3
Market scale

• About 3000 bonds with total worth 34 704,6
  billion RMB issued since 1998
• Currently traded 1385 bonds with outstanding
  amount 16 600 billion RMB (rmb usd rate
• 1 Chinese yuan 0.15 U.S. Dollars)
• About 1000 transactions a day with average daily
  turnaround about 150 billion RMB
• Bonds are traded at Shanghai and Shenzhen
  exchanges, OTC and inter-bank markets

4
Chinese Bonds Market Structure
Source ChinaBond
5
Chinese Bonds Market Structure
Source ChinaBond
6
Market Infrastructure

• The primary market of bond issuance is largely
  completed through syndication.
• Governmental bonds were mostly underwritten by
  the four stated-owned banks while other
  commercial banks and securities companies play
  active role in forming syndicate to market
  financial and corporate bonds.
• Most of bonds are held by banks, insurance
  companies, securities firms, and corporations.
  Mutual funds hold a relatively small fraction of
  the total outstanding bonds.

7
Instruments Present

• Treasure Bonds (Ministry of Finance).
• All terms. Bullet redemption. Majority has
  coupons.
• Central Bank Bills (Peoples Bank of China).
• Very short and short term. Majority are
  discounted.
• Government owned Banks
• All terms. Majority has coupons. Some have
  options.
• Corporate and Commercial Banks Bonds
• Others

8
Outstanding Amount
9
Current position
Chinese Bonds Market Overview
Zero-coupon yield curve used by CGSDTC
Applying EFFAS-EBC methodology to Chinese Bonds
Market
10
History of Zero-Coupon Yield Curves Development
in China

• Research (1999-2001).
• With aid of Reuters was developed first yield
  curve for Treasure bonds
• Prototypes (2002-2005).
• Chinese development of 4 yield curves for
  Treasure bonds using information from different
  markets
• Exploitation and Modification (2006 - Present).
• New methods are developed and yield curves are
  constructed for different types of bonds.

11
CGSDTC Methodology

• Data Filtering
• Expert and Historical Estimates
• Hermite Polynomial Fitting

12
Data Sources

• Deals Prices from Interbank, OTC and Shanghai
  Exchange markets
• Bid-Ask Quotes from Interbank market
• Market Participants estimates

13
Data Filtering
Visual comparison with historical yield curves,
not historical market data
14
Expert and Historical Estimates
If too many entries are filtered out the data is
augmented with expert estimates or historical
values
15
Hermite Polynomial Fitting

• Fix bonds maturity dates as grid nodes
• Fit Hermite polynomial such that
• a) Bond pricing equation holds
• b) Yield expert estimates are recovered

16
Controversial results
17
Current position
Chinese Bonds Market Overview
Zero-coupon yield curve used by CGSDTC
Applying EFFAS-EBC methodology to Chinese Bonds
Market
18
Available data

• Chinabond has kindly supplied us with the
• following data
• Daily trading results.
• Daily OTC Bid/Ask quotes from 8 banks.

19
Quotes co-Movement
20
Bid-Ask Spread co-Movement
21
Quotes peculiarities

• Data contains errors Bid gt Ask
• Data is inconsistent BidsgtAskk for several days
  in a row
• Quoted YTM corresponds to smth slightly less than
  the Ask quote
• Bid-Ask spread is far too wide
• 4 banks quote all bonds, 4 banks quote only a
  subset.

22
Best inter-bank Bid-Ask spreads
23
Quotes Yield Curves
24
Quotes Forward Rates
25
Quotes (BidAsk)/2
26
Quotes Summary

• Different banks use different quoting schemes
  the way quotes move differs a lot
• Either quotes are non-committing or banks are
  isolated from each other systematic arbitrage is
  present
• Nobody uses spot forward rates
• Bonds are likely to be quoted in groups by time
  to maturity (duration is not used)

27
Prices peculiarities

• Highly illiquid market
• Prices may lie well outside Bid/Ask quotes
• Similar bonds are frequently priced unlike each
  other
• No filtering helps since unusual prices tend to
  repeat in time

28
Number of Deals
29
Turnover
30
Price vs. Bid-Ask
31
Prices Yield curves
32
More smoothing
33
High-Coupon Effect?

• Hypothesis Chinese traders like high coupons
  (bonds with high coupons are valued higher).
• Testing corr(coupon size, spread) 10

34
Visual Coupons
35
Prices Summary

• Data from different sources are likely to be
  mixed
• Yield curves should be constructed from quotes
• Extremely illiquid market, price information is
  unreliable