China s Model of Managing the Financial System by Markus Brunnermeier, Michael Sockin, and Wei Xiong

Transcription

1 China s Model of Managing he Financial Sysem by Markus Brunnermeier, Michael Sockin, and Wei Xiong Discussion by Neil D. Pearson Universiy of Illinois a Urbana Champaign May 9, 2017

2 Elemens of he Model Discree ime single asse model on an infinie horizon Asse is referred o as a sock, and discussion is of he Chinese conex, bu model can apply o any rading of risky asses (e.g., MBS, Greek sovereign bonds) and/or effors o impac prices by a cenral bank Normal (Gaussian), negaive exponenial noisy raional expecaions model All r.v. s are Normal, preferences are neg. exponenial, noise raders Implicaions: agens care only abou mean and variance, prices are no fully revealing, obain linear r.e. equilibrium Complicaed: 3 echnical appendices wih calculaions and proofs* Fundamenals: cash flow is an exogenous process : Dividends: D 1 D *Brunnermeier, Sockin and Xiong are smar guys, and we will assume ha he model is solved correcly D

3 Elemens of he Model Noise raders (inerpreed as reail invesors) N N N N1 N Noise raders don opimize; hey drive prices away from fundamenals and cause (non fundamenal) volailiy; gov rades in opposie direcion o limi deviaions of prices from fundamenals and/or limi volailiy Public informaion: All agens (invesors and governmen) observe hisory of dividends, asse prices, governmen noise* M F { Ds, Ps, Gs} s Model uses quaniies such as he expecaion of noise rading ˆ M M M N E[ N F ] *The noise induced by governmen rading, o be defined laer.

4 Elemens of he Model Governmen rades agains noise raders, and also inroduces is own noise G : X G ˆ M M M N, var[ N, F 1 ] N N gov' rades o noise is scaled up by offse is esimae magniude of gov' rading of rading by noise raders G noise gov'because perfec is no Governmen minimizes weighed sum of variance of prices and variance of deviaions beween price and fundamenal value: ˆ U G M M min var[ P F 1 ] var[ P /( R ) F 1 ] fundamenal value

5 Elemens of he Model Coninuum of invesors have shor horizons: each invesor lives for only one period, dies, and is replaced by anoher (Raional) invesors acquire a signal abou eiher nex period s fundamenal asses value +1 1/ 2 s ( i s or abou he nex period gov noise: Invesors maximize U i max i a {0,1} 1 s ) i 2 g G 1/ 1 g ), i ( g, i E max E[ exp( W i X i 1 ) F Then choose demand X i o maximize uiliy, based on invesor s informaion i ] F M 1 Invesors choose which signal o acquire, based on public informaion

6 Elemens of he model Markes clear (sum of invesor, governmen, and noise rader demands = 0) There is a linear raional expecaions equilibrium:

7 There are wo equilibria In one (he fundamenal cenric or good equilibrium), invesors ry o learn abou fundamenals In he oher (he gov cenric or bad equilibrium), invesors ry o learn abou he noise creaed by governmen rading The bad equilibrium only appears if he gov places a leas some weigh on minimizing price variance Provided he gov places a leas some weigh on price variance, hen Bad equilibrium appears if he weigh on price variance is large enough If weigh on price variance is non zero, bad equilibrium appears if variance of noise rading is large enough Gov cenric equilibrium is bad: Invesors do no gaher informaion abou fundamenals, so prices are less informaive, i.e. variance of difference beween prices and fundamenals is large Sepping ouside he model, real resources are spen on an aciviy (learning abou gov noise) ha presumably does no improve allocaion

8 Key resuls: Figure 4 hen we swich o an equilibrium in which invesors ry o learn abou governmen rading If governmen places a large weigh on price variance ( )

9 Key resuls: Figure 5 hen we swich o an equilibrium in which invesors ry o learn abou governmen rading If governmen places any weigh on price variance, and variance of noise rading ( ) is sufficienly high

10 Wha should one ake away from he paper? I is possible o consruc a sensible, inernally consisen model in which inervenion by a sophisicaed* gov (e.g., cenral bank), maximizing a reasonable objecive funcion, makes hings worse Here worse means ha he difference beween marke prices and fundamenal values is larger Mechanism is ha agens sop gahering informaion abou fundamenals and insead ry o learn abou he governmen s acions Noe ha he equilibrium is problemaic in anoher way Paper absracs away from he fac ha real resources are used up in informaion acquisiionin he model agens can learn abou eiher fundamenals or he regulaor s acions, wihou any resource cos In an acual economy, he bad equilibrium would involve ens or possibly hundreds of housands of very smar people no hinking abou fundamenals bu insead hinking abou wha he governmen will do The regulaor has a reasonable objecive funcion, opimizes given his/her informaion, uses Bayes Rule,. (hough is no perfec, i.e. here is gov noise )

11 Wha should one ake away from he paper? (coninued) You migh have assumed ha inervenion by a sophisicaed regulaor (e.g., he Fed) who has a reasonable objecive funcion, uses Bayes Rule, and opimizes given his/her (raional) beliefs would move prices closer o fundamenals If ha was your assumpion, you should rehink Even a good regulaor can make hings worse Ok so i is possible o consruc a model in which cenral bank/regulaory inervenion, even by a good regulaor, makes hings worse. Bu does his model describe he financial markes? Does i describe he Chinese marke? If, hypoheically, regulaors in Wesern economies followed similar policies, would he model describe a wesern financial marke?

12 Le s sar by rying o ake a scienific perspecive: Is he model consisen wih he daa? Wha are he implicaions of he model for he daa? This is a difficul quesion Model has a single asse no cross secional implicaions A saionary equilibrium no ime series implicaions Does he model have any esable implicaions? If so, he paper does no address hem. One can imagine rying o exploi some of he comparaive saics of he model. For example, he nex slide shows how he condiional price variance (lef panel) and he variance of he difference beween prices and fundamenal value (righ panel) vary wih noise rader risk N2.

13 Model implicaions

14 Tesable implicaions of he model? Bu i isn clear ha i makes sense o ry o ake hese comparaive saics o he daa Figure 3 was creaed by considering a sequence of differen models, indexed by N 2 You migh ry o convince yourself ha his provides guidance abou how he equilibrium will vary wih ime series variaion in N2, bu his is sepping ouside he model (in he model N2 is consan) And if you do convince yourself of his, i is no clear how you will measure 2 N I canno figure ou wha (if any) are he esable implicaions of he model. This is my quesion for he auhors (which I don expec hem o answer righ now): Wha are he esable implicaions of he model? Is i possible o confron he model wih he daa? Is i possible o falsify he model? If so, wha empirical finding would falsify he model?

15 Pending he answers o hese quesions, le s ry o hink abou he reasonableness of he model Main message is ha i is possible o consruc a sensible model in which inervenion by a sophisicaed gov, maximizing a reasonable objecive funcion, makes hings worse This does no happen if he regulaor s objecive funcion is o minimize he variance of he difference beween prices and fundamenals Agens only sop gahering informaion abou fundamenals if he regulaor s minimizaion problem places a leas some weigh on minimizing price volailiy (and some oher condiions are saisfied)

16 Le s hink abou China: Are Chinese regulaors likely o pu any weigh on minimizing price volailiy? Invesors ouside he 中国证券监督管理委员会 (China Securiies Regulaory Commission) reques a friendly dialogue wih regulaors following he Chinese sock marke collapse in summer 2015

17 Do Chinese regulaors care abou he difference beween prices and fundamenals and efficien allocaion of capial? Shor sales are resriced (only a limied lis of large cap. socks may be sold shor), and hisorically have been prohibied Allowing shor sales would be an easy way o move prices oward fundamenals Oher regulaory pracices sugges a lack of concern wih fundamenals IPOs are raioned, and here are limis on he P/E raio a which shares may be offered (P/E 23) This creaes various side effecs Only once has pary moved o reduce sock prices (he May 30, 2007 ripling of he samp ax) I hink of his as being driven by a concern abou volailiy, no he difference beween prices and fundamenals

18 Does he model in he paper describe he China marke? I sugges ha he Chinese regulaory objecive is mainaining social and poliical sabiliy, which will involve placing some weigh on minimizing price variance (and lile or no weigh on fundamenals) Provided he regulaor places some weigh on fundamenals, hen he bad equilibrium only occurs if noise rader volailiy is sufficienly high Is i high enough o pu us in he bad equilibrium? I will remind you of my previous commens abou how I wish I knew how o confron he model wih he daa Casual empiricism suggess ha Chinese invesors are very focused on wha he governmen (= pary) will do

19 If he Fed inervene in US (risky) asse markes, would he US fall ino he bad equilibrium? Well, of course he Fed is compeen, and will focus on fundamenals, and will never inroduce any noise. I imagine ha mos of you hink he US regulaors will no have he same focus on volailiy as Chinese regulaors focused on poliical sabiliy Bu I feel ha I should poin ou ha Chinese hink ha Chinese regulaors do a much beer job han US regulaors Are you really confiden ha US regulaors will be less subjec o social/poliical pressure a communis dicaorship?

20 Would he US fall ino he bad equilibrium? For one and half days I have been lisening o people a his conference, and I hear a lo of discussion abou when he Fed will reduce is balance shee when i will be unwound, wheher i will be fully unwound, and how aggressively i will be unwound Fed impacs fundamenals, so some of his concern is probably abou fundamenals Bu I suspec ha some would be concerned abou when he Fed would reduce is balance shee even if his did no impac fundamenals Tens, possibly hundreds, of housands of he mos highly educaed and (poenially) producive people in he US are no hinking abou fundamenals bu insead are hinking abou wha he Fed will do Perhaps we are already in he bad equilibrium If we are, i is an enormous wase of scarce and valuable resources, he ime and energy of very smar (and poenially producive)

21 Bu we are a a Fed conference You migh reply by poining ou ha we are a a Fed conferenceof course people a a Fed conference alk abou he Fed Bu perhaps he fac ha we are a a Fed conference is possibly a sympom of he problemwe are no a a conference abou produciviy and growh Even if we are in he bad equilibrium, I do no blame people for focusing on he Fed s balance shee If we are in he bad equilibrium hen i is opimal for people o ry o figure ou wha he Fed will do.

22 If, hypoheically, he Fed sared inervening in US equiy markes, would he US fall ino he bad equilibrium? Currenly (a leas some) equiy invesors hink abou fundamenals If, hypoheically, he Fed had a large posiion in US equiies, I hypohesize ha ens, possibly hundreds, of housands of he smares, mos highly educaed, and producive people in he US would sop hinking abou fundamenals and sar hinking abou wha he Fed will do I is difficul o see his as anyhing oher han a significan cos