This is a typical news source for the monthly unemployment report.
The sentence I would draw attention to is : "The report was discouraging in many regards. While the unemployment rate fell slightly to 7.4%, the drop was partly because 37,000 people dropped out of the labor force. "
Now, what happens in these labor numbers, which are quite noisy from month to month, is that there is a general trend of decreasing unemployment. However, labor force participation can move up or down from month to month by several 100,000s, due to sampling errors. So, what happens in practice is that over any 6 month period, the Unemployment Rate drops by a few tenths of a percent, and the months that the drop is noted will happen to be the months where the labor force noise is favorable to the Unemployment Rate.
To be fair, my impression is that this reporting is one-way, but my quick perusal of google actually showed a decent mix of reporters and analysts pointing out the issues with these numbers, including the demographic issues. News reports also sometimes include this caveat when the UER comes in high because of this noise. So, maybe it's just my bias that I especially notice this when LPF noise produces a high UER number, and my optimistic bias is causing me to misreport media evenhandedness as a bias. Please correct me in the comments if this is the case.
In any case, the point of this post, which may, admittedly simply be a reflection of my conceit, is that there appears to be a persistent excess return to holding periods that begin just after the beginning of a recession.
The 1 year holding period doesn't hold up during the last half of the sample period because business cycles have been longer during this time, so there are more bullish months that are unaffected by recessions. But, the 2 and 3 year holding periods are very strong for either the first half or second half of the sample period.
The holding period inceptions would be backward looking by 6 to 12 months in real time, so I don't think I have made the mistake of measuring predetermined returns. Results are similar if I move the holding period inception back another 6 months:
One reasonable explanation for this could be that risk premiums are higher coming out of a recession. But, the volatility of returns shown above is significantly lower for those periods.
I would expect market efficiency to be stronger than this. But, I wonder if socially enforced pessimism could be a strong enough force to keep prices too low during the early periods of recovery. At any given time, there should be an array of opinions about economic potential, and those opinions should be roughtly normally shaped, with prices reflecting the decision of the average investor.
But, imagining a news report, an analyst report for a broad audience, or even a private meeting with a wealth management client, it would be very difficult in the early recovery period of a recession to take an optimistic position. We have all seen comments on blogs along the lines of, "My brother in law has 15 years experience as an electrical contractor, and he's been busting his butt for 18 months trying to find work. Why don't you step outside your cushy office and see what it's like in the real world?" In thinking about that normal distribution of opinions that makes a market, there is an awful lot of public shaming aimed at the optimistic half of that curve during these economic periods.
Could that be enough to make markets inefficient, or am I just an optimist with sour grapes?