It was gratifying to see that Professor Richard Thaler was featured in the film I saw at the weekend, “The Big Short”, because I had scheduled today to write about an important paper he co-wrote in 1987.
Richard Thaler is a brilliant behavioural economist – his acting could do with a brush-up though! The paper has a bearing on understanding why Company B might be underpriced (see last Friday’s post)
(By the way, the film is amazingly well written, directed, acted and edited; I recommend it despite not all the detail on the workings of CDOs being explained properly for a stickler like me (see last chapter of my book The Financial Times Guide to Financial Markets for more on causes of the crash and how CDOs CDS, etc. work). Musing: Maybe this is a Hollywood trend – aging Professors of Finance hit the silver screen with beautiful young actresses. My wife says “no” though, so that career avenue is blocked).
Anyway, back to this paper. It presents evidence supporting the view that Mr Market fails to allow sufficiently for the potential of mean reversion in earnings of companies.
As a result, those that have a recent string of falls of earnings find their share prices are pushed unreasonably low (everyone agrees they should go down, but do they go too far?), while those that have recently reported gains in earnings find their share prices pushed up to an unreasonable extent.
Or, as De Bondt and Thaler put it in this 1987 paper (Further Evidence on Investor Overreaction and Stock Market Seasonality):
“Experimental and survey evidence indicates that in probability revision problems people show a tendency to “overrreact”, i.e. they overweight recent information and underweight base rate data [i.e. the larger statistical data that is available if you take wider perspective].“
Their evidence comes from taking data on hundreds of US shares each year from 1966 to 1983 and putting the companies into five groups (each portfolio contains 1452 in each of the years).
Group 1 contains the 20% of shares with the lowest ratio of market capitalisation to balance sheet net assets at the selection date. These are the ‘value’ shares.
The share prices of these are low relative to assets (and other variables) because Mr Market has pushed them down.
One reason to push them down is that profits have fallen during the years before they were allocated to Group 1 – three years before the selection year the average earnings per share were 42.1% higher for this group.
This makes sense: it is rational that on average falling-profit companies should have falling shares prices.
But the question is: does Mr Market go too far, and fail to allow for a reversion to the mean in earnings? Does the extent of the fall in share price make sense?
On the other hand the companies placed in Group 5 each year are those which are in the top 20% in terms of the ratio of market capitalisation to balance sheet net assets.
These companies on average show a rise in earnings of (100 – 69.8)/69.8 = 43% over four years. These are the “glamour shares” – their prices have risen a lot over, hence the tendency to sell for much more than MCap.
The question is: have investors extrapolated the trend in earnings growth into the future when they should have allowed greater weight for reversion to the mean?
The second table helps us with some evidence on what actually happened to average earnings in each group in the four years AFTER the companies were allocated to portfolios based on ratio of MCap to net asset.
Earnings per share average for lowest and highest MCap/Net asset Groups before selection date – average set at 100 in selection year.
Earnings per share average for lowest and highest Mcap/Net asset Groups before selection date – average set at 100 in selection year…………..To read the rest of this article, and more like it, subscribe to my premium newsletter Deep Value Shares – click here http://newsletters.advfn.com/deepvalueshares/subscribe-1
Lowest Mcap/Net asset | Highest Mcap/Net asset | |
3 years before | 142.1 | 69.8 |
2 years before | 128 | 83.5 |
1 year before | 104.2 | 89.3 |
Selection date | 100 | 100 |