Formation Process And Early Warning Of Liquidity Black Hole On Stock Market

This paper studies the formation process and early warning of liquidity blackhole in a high frequency environment. There are two differences between thecondition in a high frequency environment and that in a normal environment. Thefirst is that the arrival rate of information becomes high, and the second is that thetraders might probably be changed-high frequency traders enter the market.Therefore, the features of liquidity and liquidity black holes in the high frequencyenvironment would also be different from that in the normal environment(somenon-linear characteristics emerge). Logically speaking, the key to study the problemsof liquidity and liquidity black hole in the high frequency environment is to sort outthe non-linear relationship among the net order flows, the order books and the price.Net order flows represent new traders entering into the market within a certain time,so they are the demand of liquidity. Order books represent the accumulatedquotations within a certain time and they are the supply of liquidity. When the twosides are in a dynamic equilibrium, the price is also in a relatively stable state. Oncethe dynamic balance between them is broken, the price would fall into a chaoticstate, which is pront to liquidity black hole event. Therefore, the study of dynamicrelationship between the net order flows, the order books and the price could helpus better understand the new difinition of liquidity and price dynamics in a highfrequency environment as well as hwo to prevent and monitor the occurrence ofemergencies(liquidity black hole events). In our country, the problem of how tocreate a liquidly stable market as well as how to create a market for long-terminvestors retained have become a top priority. So this paper has a certain theoreticaland practical significance.Firstly, this paper theoretically studies the formation mechaism of liquidity blackhole in a high frequency envirmonment in-depth. The modeling framework used inthe paper is mainly from the (Lux1995), and base on this we additionally consider the dynamic behaviors of high-frequency traders. Studies suggest that when theliqudity black holes occurs, the stock price in a short time will depart from thefundamental value because of the uninformed traders. And when different levels ofliquidity black holes occur, the degree of deviation of stock price from fundamentalvalue is diverse and non-monotonic. In addition, studies have shown that the impactof high-frequency traders are overall positive. In the vast majority cases whenliquidity black holes occur, high-frequency traders play a role in providing liquidity,whereas only in a few cases, their trading behaviors resonate with stock price, whichply a role in positively feed back trading.Secondly, this paper establishes the asymmetric simultaneous equationsmodel betwwen the net order flows and the price, and propose the asymmetricITH(Identification through Heteroskedasticity) method to solve the abovesimultaneous equations. The purpose is to study the dynamic relationship betweenthe net order flows and the price as well as the proportion of risk driven by liquiditydemand shocks. Studies show that the proportion of risk driven by liquidity demandshocks is about20%in CSI300Stock Index and the corresponding futures; Inadditioin, before the launch of stock index futures, the impact of liquidity demandshocks on price shocks does not exist asymmetric effects in CSI300Index. After thelanuch of stock index futures, the impact of liquidity demand shocks on price shocksshows obvious asymmetric effect in both markets. The effect of buying net orderflow shocks on price shocks is greater than that of selling net order flow shocks onprice shocks in the underlying market while the effect of buying net order flowshocks on price shocks is smaller than that of selling net order flow shocks on priceshocks in the corresponding futures market. Finally studies show that the entranceof high-frequency traders (HFT) play a positive role on improving market liquidity.Thirdly, this paper applies the mixed copula model to study the liquidity blackhole issues for the first time, the purpose of that is to study the non-lineardependence among the net order flows, the order books and the price. Logicallyspeaking, this paper is an upgrading work on the second part. The model is upgradedfrom the simultaneous equations to the more general dependent relationship and the market state is upgraded from the general state to the liquidity black hole state.Through establishing the indicator to measure the liquidity black hole state todistinguish the current state of the market, we studies the nonlinear tail dependencycharacteristics between net order flows and the price return when liquidity blackhole occurs. Studies have shown that there is a significant non-linear relationshipbetween the net order flows and the price return, and in extreme cases(whenliqudity black holes occurs) the relationship will change a lot. Especially when thelaunch of the CSI300stock index futures, due to high-frequency traders’intervention and the limited short-mechaism in the underlying market, the taildependence becames asymmetry. We add the order book information(the inverselimit order book slope) into the model, the nonlinear dependence between the netorder flow and the price significantly reduced whenever it is in a liquidity black holestate or in the normal state. This result indicates that the order book information canmostly interpret the non-linear relationship between the net order flow and theprice, which suggests that the invers limit order book slope might be an excellentliquidity proxy in a high-frequency environment.Finally, this paper divides the liquidity into three categories. The first category isthe supply level of liquidity(bid-ask spread, order book information, etc.); Thesecond category is the demand level of liquidity(net order flow, order imbalance);The third category is the integrated level of liquidity(ILLIQ, Integrated Measure ofLiquidity Indicators). We firstly use the quantile relationship method to study thewarning effectiveness of these three categories of liquidity on the price slump.Studies have shown that the supply level of liquidity forecasts the price slump best,the integrated level of liquidity forecasts the price slump secondly best, and thedemand level of liquidity forecasts the price slump worst. Later, theliquidity-black-hole based multivariate Logit model is proposed to predict the priceslump. The forecast consistency ratio of the model is about66.1%while the forecastinconsistency ratio is about32.4%. The model further confirm the result that thesupply level of liquidity is the most effective indicator in warning price slump. The data used in this paper is the high frequency data of CSI300IndexFutures(the data frequency is0.5second per quote, including5-bids and5-asksquotes), the high frequency data of CSI300Index(the data frequency is6secondsper quote), the high frequency data of SSE50Index(the data frequency is6secondsper quote), and the high frequency data of SME COMPOSITE Index(the datafrequency is6seconds per quote). The data source is from CSMAR Database.There are three main innovations in this paper: The first innovation is that thecontinuous-time model is firstly used to describe the occurrence and the process offormation mechanism of liquidity black hole, and the dynamics of trading behaviorsof high-frequency traders is firstly completely considered. This is a certain innovationin theory. The second innovation is that the asymmetric ITH method is proposed tosolve the asymmetrical simultaneous equations model of net order flows and theprice return. This is a certain innovation in methodology. The third innovation is thata mixed copula model is applied to study the non-linear dependency among the netorder flows, the order books, and the price return. This is a certain innovation in theresearch fields.After a multi-level in-depth research, we have a certain understanding onthe dynamics of liquidity and liquidity black hole in a high frequency environment.However, a lot of problems still need to be explored. Such as the ’fat finger’ event byEverbright Securities on August16,2013in China, which casued a huge impact onthe market. Until now, the market lakes sufficient knowledge of how high-frequencytradings perform, which leads to a slow response to certain events. Further, inaddition to the high-frequency trading, there are also other new types ofquantitative trading that might have certain impact on the market liquidity, such asarbitrage tradings, algorithmic tradings and so on. We will continue to explore theseproblems in future work.