Understanding Financial Frictions in the Chinese Capital Market

Transcription

1 Understanding Financial Frictions in the Chinese Capital Market Hui Chen MIT Sloan June 19, 2018 (MIT Sloan) Chinese Capital Markets June 19, / 51

2 experienced several phases. From 1990 to 1992, the initial experimental phase, only eight Chinese Capital Markets stocks were listed in Shanghai, the so-called old eight. Six were listed in Shenzhen in By the end of 1992, there were 53 stocks listed on the two exchanges combined. From 1993 What are the common perceptions of the Chinese capital markets? to 1997, the market experienced a robust growth. The number of firms listed on the two Already quite big. exchanges more than doubled in 1993, reaching 177 in by the end of the year. The total was Stock market capitalization: $8.5 Trillion (vs. USA: $34 Trillion) 311 by the2nd endlargest of 1995, in the 514world by 1996, and 720 by (a) Number of Listed Firms (Monthly) (b) Floating Market Capitalization (Monthly) Figure 1.1: Size of the Stock Market ( ) (MIT Sloan) Chinese Capital Markets June 19, / 51

3 Chinese Capital Markets What are the common perceptions of the Chinese capital markets? High percentage of (unsophisticated) individual investors. High volatility, high turnover. (a) Stock Market Volatility (Monthly) (b) Stock Market Turnover (Monthly) Figure 1.2: Volatility and Turnover of the Stock Market ( ) (MIT Sloan) Chinese Capital Markets June 19, / 51

4 Chinese Capital Markets What are the common perceptions of the Chinese capital markets? Complicated institutional details (a lot of restrictions). Daily price limit T + 1 ST (special treatment) Foreign access... Global impact of Chinese financial shocks. Example: sudden devaluation of the Yuan in 08/2015 (MIT Sloan) Chinese Capital Markets June 19, / 51

5 Two Examples Chen, Petukhov, Wang (2018): The Dark Side of Circuit Breakers Chen, Chen, He, Liu, Xie (2018): Pledgeability and Asset Prices: Evidence from the Chinese Corporate Bond Markets (MIT Sloan) Chinese Capital Markets June 19, / 51

6 Outline 1 Chen, Petukhov, Wang (2018): The Dark Side of Circuit Breakers 2 Chen, Chen, He, Liu, Xie (2018): Pledgeability and Asset Prices (MIT Sloan) Chinese Capital Markets June 19, / 51

7 Circuit Breakers What is it? Trading halt following extreme price movements. Market-wide CBs; CBs for individual stocks. First advocated by the Brady Commission following the Black Monday of Now widely adopted around the world. Why? To reduce excess volatility and improve price efficiency? To restore orderly trading in the market? To protect investors? What are the consequences? (MIT Sloan) Chinese Capital Markets June 19, / 51

8 Circuit Breakers: Chinese Experience First implemented on Jan 04, 2016, following the market crash in summer Abandoned after just 4 days. Jan 04, 2016 Jan 07, ,700 3,500 3,600 3,400 Level 1 Level 1 3,500 Level 2 3,300 Level 2 9:30 10:30 11:30 13:00 14:00 15:00 9:30 10:30 11:30 13:00 14:00 15:00 (MIT Sloan) Chinese Capital Markets June 19, / 51

9 Model Setup A continuous-time endowment economy over interval [0,T]. Aggregate stock: one unit, with terminal dividend D T. dd t = µd t dt + σd t dz t, D 0 = 1 Riskless bond: net supply, pays off 1 at time T. Two competitive agents: A and B Endowed with ω and 1 ω shares of the stock and bond. Log preferences over terminal wealth u i (W i T ) = ln(w i T ), i = {A,B} No intermediate consumption riskless bond as numeraire. (MIT Sloan) Chinese Capital Markets June 19, / 51

10 Heterogeneous Beliefs The two agents disagree about the growth rate of dividend. Agent A has objective beliefs: µ A = µ Agent B s belief: µ B t = µ + δ t Constant disagreement: Extrapolative disagreement: δ t δ dδ t = νdz t The two agents agree to disagree. Need trading. Heterogeneous risk aversion works similarly. (MIT Sloan) Chinese Capital Markets June 19, / 51

11 Circuit Breakers The stock market will be closed until T whenever the price of the stock S t falls below the level (1 α)s 0. α: circuit breaker limit, α [0,1] S 0 : initial stock price endogenous τ = inf{t 0 : S t = (1 α)s 0 } After stock market closure agents are not able to change their stock postions Bond market remains open throughout the interval [0,T]. (MIT Sloan) Chinese Capital Markets June 19, / 51

12 Equilibrium: No Circuit Breakers Markets are dynamically complete solve for equilibrium via the planner s problem. subject to resource constraint [ (Ŵ ) max E 0 λln A T + (1 λ)ηt ln ( Ŵ B )] ŴT A,Ŵ T B T Ŵ A T + Ŵ B T = D T + Stock price when 0: wealth-weighted average of the prices under two agents beliefs Ŝ t = ( ω A t Ŝ A t + ωb t Ŝ B t ) 1 Ŝt i : price in a single-agent economy with agent i ω i t : agent i s wealth share (MIT Sloan) Chinese Capital Markets June 19, / 51

13 Equilibrium: Circuit Breakers Two scenarios: CB is not triggered between 0 and T; CB is triggered at time τ < T. Markets are still dynamically complete over interval [0,τ T]. Solution strategy: 1 Pin down stopping rule τ consistent with a given stopping price S through equilibrium conditions upon market closure. 2 Given stopping time τ, solve for equilibrium allocation at τ T via planner s problem. 3 Compute price at t τ T for given τ and S, S t (τ,s). 4 Solve for S through the fixed point problem, S = (1 α)s 0 (S) (MIT Sloan) Chinese Capital Markets June 19, / 51

14 Equilibrium: Upon Market Closure Suppose agent i has wealth Wτ i at time τ T. Portfolio problem at time τ for competitive agents: V i (Wτ i,τ) = max s.t. θ i τ,ϕ i τ E i τ θ i τ S τ + φ i τ = W i τ W i T 0 [ln(θ i τ D T + φ i τ ) ] V i (Wτ i,τ): indirect utility function for agent i at time τ Market clearing conditions: θ A τ + θb τ = 1 φ A τ + φb τ = (stock market) (bond market) (MIT Sloan) Chinese Capital Markets June 19, / 51

15 Equilibrium ( 0 case): Upon Market Closure Market closure inability to rebalance between τ and T Illiquidity + log utility no short or levered position at τ Leverage constraint binds for the optimistic agent pessimistic agent becomes the marginal investor. Assumption 0 to be relaxed later. Proposition In the limiting case with 0, upon market closure at τ < T, both agents will hold all of their wealth in the stock with no bonds. The market clearing price is S τ = min{ŝ A τ,ŝb τ } Stopping rule τ is expressed in closed form as a function of state variables. (MIT Sloan) Chinese Capital Markets June 19, / 51

16 Characterizing the stopping time τ Lemma Take the stopping price S as given. Define a stopping time τ = inf{t 0 : D t = D(t,δ t,s)}. Then the circuit breaker is triggered at time τ when τ T. We have managed to characterize a stopping time that is based on the endogenous stock price S t as one that is based on the exogenous processes of D t and δ t. (MIT Sloan) Chinese Capital Markets June 19, / 51

17 Equilibrium: Before Market Closure Solve for optimal allocation at τ T through the planner problem, using the indirect utilities upon market closure: subject to [ max E 0 λv A (W A Wτ T A,W τ T B τ T,τ T) + (1 λ)η T V B (Wτ T B,τ T)] W A τ T + W B τ T = S τ T + The price of the stock at time t τ T: [ ] π A S t = E τ T t π A S τ T = ( ω A t E t[sτ T 1 ] + ωb t EB t [S 1 τ T ]) 1 t (MIT Sloan) Chinese Capital Markets June 19, / 51

18 Equilibrium: S Fixed point problem S = (1 α)s 0 (τ(s),s) ( ) Proposition There is a unique solution to ( ) for any α [0,1]. (MIT Sloan) Chinese Capital Markets June 19, / 51

19 Special Case: Constant Disagreement Calibration: T = 1 µ = 10%/250 σ = 3% α = 5% δ = 2% ω = 90% Agent B is pessimistic. Most wealth initially owned by rational agent. (MIT Sloan) Chinese Capital Markets June 19, / 51

20 Price and Agent A s Portfolio Holdings t = 0.25 t = St /Dt 0.99 θ A t Fundamental value: D t Fundamental value: D t Dotted line complete markets, solid circuit breakers, horizontal dashed S/D ratio in representative-agent economies (MIT Sloan) Chinese Capital Markets June 19, / 51

21 Conditional Return Volatility and Risk Premium t = 0.25 t = σ S,t (%) 5 4 µ A S,t (%) Fundamental value: D t Fundamental value: D t Dotted line complete markets, solid circuit breakers, horizontal dashed volatility ratio in a representative-agent economy (MIT Sloan) Chinese Capital Markets June 19, / 51

22 Stronger Effects Earlier during Trading Session t = 0.25 t = St /Dt σ S,t (%) t = 0.75 t = St /Dt σ S,t (%) Fundamental value: D t Fundamental value: D t (MIT Sloan) Chinese Capital Markets June 19, / 51

23 Circuit breaker vs. pre-scheduled trading halt 1 t = t = t = 0.25 St /Dt St /Dt t = Fundamental value: D t σ S,t (%) σ S,t (%) t = Fundamental value: D t θ A t θ A t t = Fundamental value: D t (MIT Sloan) Chinese Capital Markets June 19, / 51

24 Welfare Loss as a Function of Rational Agent Initial Share of Wealth Fractional C.E. loss (%) Agent A Agent B Total welfare ω ω initial wealth share of agent A Welfare loss is relative to the complete markets case. (MIT Sloan) Chinese Capital Markets June 19, / 51

25 Robustness and Extensions Non-zero bond supply Bounded shocks (discrete time) No need to completely delever/close short positions. Equilibrium can flip like with positive bond supply. Upside vs. downside CBs CBs based other variables: volatility, volume Multiple-tiered CBs (MIT Sloan) Chinese Capital Markets June 19, / 51

26 Conclusion A competitive benchmark to study the dynamic effects of CBs. CBs tend to have the following effects: Lower the price-dividend ratio (increase price distortion) Daily price range, conditional and realized volatility Magnet Effect: raise probability of the stock price to reach the threshold limit Main mechanism applies to other forms of disappearing liquidity: price limits, short-sale ban, trading frequency restrictions, sudden price jumps Policy implications: Reduce volatility : Which volatility? Lucas critique: Danger of using historical data to estimate the likelihood of CB trigger after implementation. (MIT Sloan) Chinese Capital Markets June 19, / 51

27 Outline 1 Chen, Petukhov, Wang (2018): The Dark Side of Circuit Breakers 2 Chen, Chen, He, Liu, Xie (2018): Pledgeability and Asset Prices (MIT Sloan) Chinese Capital Markets June 19, / 51

28 Introduction Equilibrium asset prices depend on both fundamental and liquidity. Duffie (2010) presidential address What exactly is liquidity? Asset pledgeability: the ability to be used as collateral to borrow (and help reduce financing costs). Closely related to repo specials (Duffie, 1996) and margin-based asset pricing (Garleanu & Pederson, 2010). Why are the Chinese bond markets suitable for this question? Same bonds are traded on two different markets, with different rules for repos. An unexpected policy change on one of the two markets: an arguably exogenous shock to pledgeability. (MIT Sloan) Chinese Capital Markets June 19, / 51

29 Chinese bond market: 3rd in the world behind U.S. and Japan (MIT Sloan) Chinese Capital Markets June 19, / 51

30 Two corporate bond markets Repo transactions are done differently in the two markets. Exchange (EX) market, centralized Repos are done with the exchange. Haircut schedules set unilaterally by the exchange; mainly based on bond ratings and liquidity. Interbank (IB) market, over-the-counter Repos are between two private parties. Haircuts set via negotiation; depend on counterparties as well as the bond. (MIT Sloan) Chinese Capital Markets June 19, / 51

31 Segmented markets (Limits to) arbitrage: Buying a bond in one market and selling it in the other requires the application for custody transfer. Lengthy process: EX IB: 3 days for enterprise bonds (1 day for Treasuries); IB EX: 5 days or more. (MIT Sloan) Chinese Capital Markets June 19, / 51

32 Exchange market premium Differences in pledgeability + limits to arbitrage same bond being traded at different prices in the two markets. Other factors could also lead to price differences: participants; trading protocols Exchange market premium: EX premium = s IB EX = spread IB spread EX (MIT Sloan) Chinese Capital Markets June 19, / 51

33 Rating-based haircuts Pledgeability in EX market: Borrowing capacity quoted as a fraction of bond face value (discount factor). discount factor book value haircut = 1 market value Haircuts largely depend on credit ratings (besides volatility, turnover). Rating explains 96.1% of the variation in haircuts (vs. 96.3% in a kitchen-sink regression). High rating bonds enjoy an EX premium. The ease of repo transactions in exchange market: the counterparty is always there + potentially more favorable haircuts Transparency in the ability to obtain collateralized borrowing (MIT Sloan) Chinese Capital Markets June 19, / 51

34 Event: Dec 8, 2014 Background (Chen, He, & Liu, 2017) 2009 stimulus funded by massive local government debt Rapid growth of Municipal Corporate Bonds (MCB) in MCB is a form of enterprise bonds, largely dual-listed In 2014, regulators started several rounds of coordinated effort to curb the local government debt problem E.g., swap program (swap between MCB and municipal bonds) After market closing on Dec 8, 2014, EX market suspended the repo eligibility of all enterprise bonds rated AA+ and below. Interesting policy shock Exchange market is not the main gate-keeper/regulator of MCB, but took an unexpected aggressive move. It caused significant market reactions. Blunt policy tool that depends on coarse (and often uninformative) ratings (MIT Sloan) Chinese Capital Markets June 19, / 51

35 Haircuts across ratings before and after the event (MIT Sloan) Chinese Capital Markets June 19, / 51

36 Haircut distributions (MIT Sloan) Chinese Capital Markets June 19, / 51

37 Haircut distributions (MIT Sloan) Chinese Capital Markets June 19, / 51

38 Yield spreads across ratings: EX (MIT Sloan) Chinese Capital Markets June 19, / 51

39 Yield spreads across ratings: IB (MIT Sloan) Chinese Capital Markets June 19, / 51

40 Exchange premia before and after the shock (MIT Sloan) Chinese Capital Markets June 19, / 51

41 EX premium during event Diff-in-diff: first cut Dual-listed and same-day trading fundamentals fully controlled for. Across ratings, treatment and control groups Control group consists of both higher- (AAA) and lower-rated (AA-) bonds. Ruling out lower-rating bonds are just more sensitive to macro shocks. (MIT Sloan) Chinese Capital Markets June 19, / 51

42 EX premium during event Diff-in-diff: with controls Rating-time dummy: For bond i on date t, define I j,τ = 1 when it has rating j during i,t period τ, where j {AAA,AA+,AA,AA } and τ is one of the sub periods around the event. Regression specification: s IB EX i,t = k j,τ I j,τ i,t + controls i,t + ɛ i,t j,τ Controls: Bond level characteristics: coupon, leverage, size, time to maturity, turnover ratio Macro factors: term spread, SHIBOR, CDBSpot, stock market index (MIT Sloan) Chinese Capital Markets June 19, / 51

43 EX premium during event (MIT Sloan) Chinese Capital Markets June 19, / 51

44 IV estimate of the effect of pledgeability on pricing Instrumental Variable: { 1 if rating {AA+,AA} & t is after 12/08/14 shock i,t = 0 otherwise Two-stage procedure: First stage: Regress haircuts on shocks. Second stage: Regress EX premium on fitted haircuts. Similar control variables as before Bond level characteristics: coupon, leverage, size, time to maturity, turnover ratio Macro factors: term spread, SHIBOR, CDBSpot, stock market index Sample: dual-listed and same-day trading (MIT Sloan) Chinese Capital Markets June 19, / 51

45 First Stage (MIT Sloan) Chinese Capital Markets June 19, / 51

46 Second Stage (MIT Sloan) Chinese Capital Markets June 19, / 51

47 IV estimate of the effect of pledgeability Our estimate based on the EX premium is likely to be downward biased. The IB prices of bonds hit by EX policy are affected as well due to potential arbitrage and over-controlling. A second estimate is based on matched AAA exchange bonds. AAA bonds with the same pledgeability and yield spread; but free of policy shock. Based on dual-listed sample (instead of dual-listed and same-day trading ) (MIT Sloan) Chinese Capital Markets June 19, / 51

48 Second Stage (Versus matched AAA) (MIT Sloan) Chinese Capital Markets June 19, / 51

49 IV estimate of the effect of pledgeability Economic magnitude 1% change in haircut translates to 0.6 bps change in yield, or about 3.7 cents change in price for an average dual-listed bond. Sensitivity table Price change given an X% change in haircut for Y year maturity Yield: 6.45%, coupon rate: 6.62%, face value: 100 (MIT Sloan) Chinese Capital Markets June 19, / 51

50 Sensitivity table (MIT Sloan) Chinese Capital Markets June 19, / 51

51 Conclusion Causal effect of pledgeability on asset prices Dual-list bonds, free of fundamental concerns Differential reactions based on ratings Quantify the effect of asset pledgeability First paper (to our knowledge) to estimate the effect of changes in pledgebality on asset prices 20% increase in haircut translates into 12 bps increase in yield change. For a 10-year bond, the price change is about 87 cents (per $100). Chinese bond markets are at least equally interesting compared to US bond market. (MIT Sloan) Chinese Capital Markets June 19, / 51