Securitization in a Model of Regional Liquidity Shocks and Priv

Transcription

1 Securitization in a Model of Regional Liquidity Shocks and Private Information George Washington University June 2017

2 Questions If there is an incentive to misrepresent the quality of the long-term risky asset, how does it affect real output, credit allocation and liquidity insurance? Does it matter if credit risk from loan origination and securitization stays in the banking sector or if it is shifted to outside investors?

3 Approach: Examine the impact of ex post trading on ex ante optimality in the securitization process. Financial intermediation (FI) and financial markets (FM) may be competing mechanisms for credit allocation and liquidity provision - Jacklin 1987, Diamond JPE 1997, Allen and Gale [2004] Adverse selection in the securitization process Liquidity insurance through financial intermediation in a Diamond-Dybvig model

4 Diamond-Dybvig: 3 periods and a continuum of locations φ Φ [0, 1]. (Bhattacharya and Gale) Representative FI and a representative household at each location. At t = 0, all households have E units of the consumption good, which is allocated between risky long-term projects and risk-free short-term project. The risky project cannot be interrupted at t = 1.

5 At location φ t=0 t=1 t=2 Storage E = l + k η observed privately Dsn f(η) Securitize Fraction h m are early consumers l = E k Can t interrupt long term project θ i {θ 1,θ 2 } g i (η) θ i publicly observable θ i k α (1 h m ) fraction of late consumers Securitization is a process of insuring against random θ. The conditional mean is θ

6 θ 1 < θ 2 and η [η, η] with g 1 (η) decreasing. At t = 1, a household observes a liquidity shock υ {0, 1}. A household has utility U(c 1, c 2, υ) = (1 υ)u(c 1 ) + υβu(c 2 ) If υ = 0, the household is impatient and, if υ = 1, the household is patient. A fraction h m of the households is impatient - in the paper this is random so there are regional liquidity demand shocks.

7 Preferences - expected utility for a type η region h m U(c 1 (η)) + (1 h m )β i g i (η)u(c 2 (η, θ i )) (1) where 0 < β < 1. Individual incentive compatibility constraint (IICC) U(c 2 ) U(c 1 )

8 The first-best allocation: U (c 1 (η)) = βr f U (c 2 (η)) This is liquidity insurance. If βr 1, then the allocation satisfies the incentive compatibility constraint (υ private information and η public) Optimal long-term investment max [E k + θ(η)k α ] k r Solution k (η, r) c 2 (η) invariant wrt θ i and r f = U (c 1 (η)) βu (c 2 (η))

9 Highlight impact of trading process assuming full information: Financial Intermediation Financial Markets

10 Financial intermediation - maximize objective function subject to E l + k 0 q(η, θ i )z(η, θ i ) i l h m c 1 (η) θ i (k (η)) α + z(η, θ i ) (1 h m )c 2 (η) and IICC. Achieves first-best allocation Risky securities held by FI - equivalent to loan syndication

11 Financial Markets At t = 0 household holds safe assets E k and trades contingent claims E l + k 0 i q(η, θ i )z(η, θ i )

12 If an impatient agent - borrow against t = 2 income If a patient agent c 1 (η) = E k + θα k + z(η, θ i) r c 2 (η) = r(e k) + θ i k α + z(η, θ i ) Household can borrow and lend at rate r where value of securities is already incorporated into consumption.

13 Results - η is public information: r f > r m Interest rate higher with financial intermediation k m > k f Higher investment in long-term risky asset in financial markets How trade occurs affects real output and interest rates.

14 η is private information How to interpret adverse selection: Originate loan but don t screen. No due diligence - don t know quality Originate loan and know quality but choose to misrepresent Can a FI signal quality by portfolio choice (l, k)? No

15 Observe in general. q(η, θ i ) q(ˆη, θ) If nonexclusivity in contracts and if the individual contingent claims embedded in a contract can be traded, then elimination of arbitrage profits results in q(η, θ i ) = q(θ i ) Contracts can be written on the observables, including k, θ i.

16 FI maximizes the objective function with respect to k, c 1, c 2 The lifetime budget set is E k + i q(θ i )θ i k α = h m c 1 (η) + i q(θ i )(1 h m )c 2 (η, θ i ). The FOC for k is (2) 1 = i q(θ i )θ i αk α 1 Notice η doesn t enter in choice of k.

17 It follows a bank in region η will choose (c 1 (η), c 2 (η, θ i )) to satisfy q(θ 1 ) = βg 1(η)U (c 2 (η, θ 1 )) U, (c 1 (η)) q(θ 2 ) = βg 2(η)U (c 2 (η, θ 2 )) U, (c 1 (η)) Second period consumption varies with θ i. High return areas (high θ(η)) underinvest in k while low return areas over-invest. Affects provision of liquidity insurance through real output and the interest rate.

18 Stark Result All FI (or households if financial markets) pick the same portfolio (E k, k) regardless of risk.

19 When households directly hold securities and borrow and lend at rate r, still have similar result -all long term risky projects are represented as same quality but a different set of prices and interest rates.

20 Conclusion- when there is private information and private trading: Securitization leads to identical portfolios on the asset side of the balance sheet. Under or over investment in risky asset depending on region s type. Who holds the securities and how trading takes place has an impact on liquidity insurance and has real output effects. The distribution of investment in the risky asset is suboptimal, relative to first best.