Luc

Research question#

  • To what extent does higher volume of trading and analysis lead to more accurate price discovery? Phrased differently, can trading and analysis be too high, leading to mispricing due to hype?

  • My phrasing:

    • Is there an optimum trading volume and analyst coverage from the perspective of efficient pricing?

Theory:

  • You have to explain how Zhang (2006) and Baker & Wurgler (2007) define mispricing, and how they attempt to measure it empirically
    • I see later that you have already done so
    • Argue for the relevance of mispricing
      • Answer the question: “Who cares?”
  • Hype is a very interesting concept
    • I have recently read a book by Robert Shiller, “Narrative Economics”.
      • z-lib.org
    • You might want to read a review (or articles in this vein), his ideas closely align to yours
    • He defines “narratives” (which are very similar to hypes) as “epidemics” that dissipate through the population and can cause bubbles in stock prices
  • You describe affect, but what is exactly its definition? And how does it differ from sentiment?
    • Overconfidence
    • Exhuberance
    • Overreaction

Methodology:

  • business press journalism as a measurement for hype

    This is a very nice and feasible idea that you could also incorporate!

    • You can analyze newspaper frequency of a company,
    • Maybe do sentiment analysis!
    • Natural Language Processing (in R, Python)
      • Access to newspaper archives (FD, FT)
      • Google Trends

  • Other ways to measure (literature)

  • Why would you want to categorize trading volume?

    • It is a continuous variable - why not keep it that way?

  • What is your unit of analysis? Is it a stock or a portfolio?

    • In my opinion, stock is more useful, because hypes usually exists around specific stocks or assets (e.g. Tesla, bitcoin)

  • I also do not see a clear definition of investor sentiment yet

Data:

  • Why do you opt for US Data?
  • If you decide to measure these concept in different ways (e.g. in ways I suggested), you could use different sources, e.g. newspaper archives or Google Trends

Conclusion:

  • Your research design is feasible. I think it has good potential, but some aspects (in particular, methodology) still has to be worked out more elaborately.