The Difficulty of Measuring the Gains To Fundamental Research

Here's a paper by Bradford Cornell that I've had in my in box for a while. It's titled "Investment Research: How Much Is Enough?" Here's the abstract

Aside from the decision to enter the equity market, the most fundamental question an investor faces is whether to passively hold the market portfolio or to do investment research. This thesis of this paper is that there is no scientifically reliable procedure available which can be applied to estimate the marginal product of investment research. In light of this imprecision, investors become forced to rely on some combination of judgment, gut instinct, and marketing imperatives to determine both the research approaches they employ and the capital they allocate to each approach. However, decisions based on such nebulous criteria are fragile and subject to dramatic revision in the face of market movements. These revisions, in turn, can exacerbate movements in asset prices.

I raises some interesting issues about the difficulties in measuring gains to fundamental research. To name a few:

• The difficulty in measuring "abnormal" performance", given the stochastic (i.e. random) nature of stock returns
• The time-varying nature of any possible gains to analysis (funds and strategies change over time).
• Given the needs for sample size and duration necessary to get high levels of statistical significance, most findings are of pretty low confidence
• The ad hoc nature of many analysis strategies and the role that judgement plays
  

It's worth reading, and give some good points for discussion in a class module on efficient markets (and the related topic of "anomalies" like the size and value effects). You can read the working paper on SSRN here