De, Koustav

Research Topic: Investor Behavior

Reference Point Estimation

Using individual stock transaction data from Finland between 1995 to 2009 we try to investigate investor behavior. It’s well known that investors display strong disposition effect – tendency to sell winning stocks more readily than stocks at a loss. We study if this behavior is exactly around zero gains or if it is distinguishably non-zero. This is what we call the reference point. Furthermore, it is important to study if the reference point varies across individuals, across time and market conditions.

Participants:

Koustav De (University of Kentucky), Tyler Shumway (University of Michigan).

Students:

Quan Qi, Finance, Graduate, Added 10/19/2021

Investment and stock price

Stock price of the firm and/or competitors is often argued to influence investment decisions. Observing the portfolio being held, it might be possible to identify executives from their holding of stocks and options. We intend to study how executives and possibly high net worth individuals in the sample adjust their holding based on competitors and if these adjustments have any leading relation to investment decision of the firms.

Participants: Igor Cunha (University of Kentucky), Koustav De (University of Kentucky).

Software: SAS, R

Computation method:

The above projects require heavy computation due to the data used and the methods. The proprietary data comprises individual transactions by all Finnish stock investors daily between 1995-2009. The data also includes some transaction by non-individuals. For method, I primarily use proportional hazard models for a range of reference points which estimates models and compares likelihood across models in a finer grid. This intuitively presents the reference point that makes most sense for the behavior.

Projects