Speaker :Dr.Devlina Chatterjee
Indian Institute of Science Bangalore
Background of the Speaker:
Devlina Chatterjee got her BTech in Agricultural Engineering from IIT Kharagpur in 1989. She subsequently got a PGDM (Agribusiness) from IIM Ahmedabad, MS degrees from Penn State University (Agricultural Engineering and Agricultural Economics), and her PhD in Management Studies from IISc Bangalore in 2010. She also worked in industry for a total of eight years at organizations like IFCI, New Delhi; GE Capital, Bangalore; and Antrix Corporation, Department of Space, Bangalore. Her work experience in these organizations has included evaluation of project proposals, statistical analyses of credit card business data, and contractual negotiations in the space image marketing business.
The seminar:
Dr. Chatterjee began by stating the focus of her study, the liquidity of individual stocks traded in electronic market. She went on to quote Black and Kyle to elaborate on liquidity. According to Black, a market is said to be liquid if it is continuous and efficient. Kyle gave the transactional properties of such a liquid market as tightness, depth and resilience.
As Dr. Chatterjee told, it is difficult to quantify the multidimensional nature of liquidity and a solution to this, forms the focus of her paper. Her paper delves into this by finding the relation between the various proxies which can be used to quantitatively describe liquidity.
She utilised factor analysis and survival analysis to test two models, one which measures the various macro and micro proxies to find the factors which would help quantify liquidity and second, which finds how long it takes for given order to be executed.
The data for this analysis was taken from NSE in two time intervals and covered 124 and 127 stocks in the two intervals respectively.
In the first model, the multi-dimensional nature of liquidity was studied first through factor analysis of eleven commonly used liquidity proxies – trading volume, turnover, frequency of trading, turnover ratio, price range, liquidity ratio, relative bid ask spread, order depth, market depth, immediacy, price elasticity of buy/sell. The study revealed that across two different market conditions, five factors emerged consistently: depth, spread, volume, price elasticity and relative activity. Thus the 3 transactional properties as stated by Kyle are verified.
In the second model, execution probabilities of limit orders were studied using logistic models and survival analysis. The covariates used in the analysis were price premium, volatility, relative bid-ask spread, order imbalance, trading activity, depth, relative activity in script, last traded price of stock, short term changes in trading activity, time of day and day of week.
In related prior studies, the history of the entire limit order book had been reconstructed using high-frequency data and heavy computation. Here, using much less data, the idea of hypothetical orders, and interval censoring, most appropriate distribution of survival times were obtained. The covariates which were found to determine these were price premium, closing price, log of depth, volatility, relative activity and firm order size.
Finally Dr.Chatterjee closed the presentation by observing that, both the models developed in this study made good out-of-sample predictions as well.
The seminar was wrapped up with a free-wheeling discussion between Dr. Chaterjee and the esteemed faculty of IME Dept., which churned out some interesting perspectives and left the attendees with many issues to mull over.
Contributed by:
ArpitaPandey
PR & Media Cell,
MBA, IME Dept
IIT Kanpur
Indian Institute of Science Bangalore
Background of the Speaker:
Devlina Chatterjee got her BTech in Agricultural Engineering from IIT Kharagpur in 1989. She subsequently got a PGDM (Agribusiness) from IIM Ahmedabad, MS degrees from Penn State University (Agricultural Engineering and Agricultural Economics), and her PhD in Management Studies from IISc Bangalore in 2010. She also worked in industry for a total of eight years at organizations like IFCI, New Delhi; GE Capital, Bangalore; and Antrix Corporation, Department of Space, Bangalore. Her work experience in these organizations has included evaluation of project proposals, statistical analyses of credit card business data, and contractual negotiations in the space image marketing business.
The seminar:
Dr. Chatterjee began by stating the focus of her study, the liquidity of individual stocks traded in electronic market. She went on to quote Black and Kyle to elaborate on liquidity. According to Black, a market is said to be liquid if it is continuous and efficient. Kyle gave the transactional properties of such a liquid market as tightness, depth and resilience.
As Dr. Chatterjee told, it is difficult to quantify the multidimensional nature of liquidity and a solution to this, forms the focus of her paper. Her paper delves into this by finding the relation between the various proxies which can be used to quantitatively describe liquidity.
She utilised factor analysis and survival analysis to test two models, one which measures the various macro and micro proxies to find the factors which would help quantify liquidity and second, which finds how long it takes for given order to be executed.
The data for this analysis was taken from NSE in two time intervals and covered 124 and 127 stocks in the two intervals respectively.
In the first model, the multi-dimensional nature of liquidity was studied first through factor analysis of eleven commonly used liquidity proxies – trading volume, turnover, frequency of trading, turnover ratio, price range, liquidity ratio, relative bid ask spread, order depth, market depth, immediacy, price elasticity of buy/sell. The study revealed that across two different market conditions, five factors emerged consistently: depth, spread, volume, price elasticity and relative activity. Thus the 3 transactional properties as stated by Kyle are verified.
In the second model, execution probabilities of limit orders were studied using logistic models and survival analysis. The covariates used in the analysis were price premium, volatility, relative bid-ask spread, order imbalance, trading activity, depth, relative activity in script, last traded price of stock, short term changes in trading activity, time of day and day of week.
In related prior studies, the history of the entire limit order book had been reconstructed using high-frequency data and heavy computation. Here, using much less data, the idea of hypothetical orders, and interval censoring, most appropriate distribution of survival times were obtained. The covariates which were found to determine these were price premium, closing price, log of depth, volatility, relative activity and firm order size.
Finally Dr.Chatterjee closed the presentation by observing that, both the models developed in this study made good out-of-sample predictions as well.
The seminar was wrapped up with a free-wheeling discussion between Dr. Chaterjee and the esteemed faculty of IME Dept., which churned out some interesting perspectives and left the attendees with many issues to mull over.
Contributed by:
ArpitaPandey
PR & Media Cell,
MBA, IME Dept
IIT Kanpur
Comments
Post a Comment