Downside Risk for the "New" GM

The new post-bailout, post-bankrupt GM has been generating some buzz on Wall Street this week. Yesterday, its stock jumped 2.8% in premarket trading after investment groups initiated coverage on the new stock. Credit Suisse gave the new GM an "Outperform" rating and a $43 per share price target (the stock was at 36.0 at the end of trading, see above). JP Morgan gave it an "Overweight" and a $44 price target while RBC Capital Markets set an "Outperform" rating. What should we think about this enthusiasm for "Government Motors." Is it just more "Irrational Exuberance"?
Let' s look at what history can tell us. In the time series above I've spliced together the old GM stock price history (MTLQQ.PK) with the new stock (GM) after the IPO in November. The red dotted line displays the one-step ahead predictions for the best fit model [1] to the stock trend. The best-fit model is a random walk, that is, today is like tomorrow except for random shocks! A "Random Walk Down Wall Street," indeed!
That's a surprise result for the world's largest multinational automaker, the engine of growth for the post-War U.S. economy [2]. The graph above shows a plot of the GM random walk model without the random shocks. Until mid-2005 (the beginning of the end?), the stock price did not stray far from it's initial value in the 1960s. What should we expect for the future?
"More of the same" would be a good guess. The graph above shows the random-walk forecast for 2011 with confidence intervals. What this shows is that any stock price between 10 and 50 is probable (within the 98% confidence interval for a random walk). The price targets from the investment groups seem a little more conservative. What's also interesting is that the investment houses don't talk about the downside risk.

My opinion: stocks that are random walks without even some observable drift are best left to the investment houses. Supposedly, the "smart money" can anticipate shocks and trigger events in ways that the average investor cannot. But, what do I know. I'm not even the 800-pound gorilla in the room!

[1] To find the best fit model, I used some statistical techniques to search among various candidate models ranging from business-as-usual models to models based on a broad index of secular and cyclical performance in the U.S. economy. None of these models fit any better than a random walk. The result doesn't mean that at some time in the future I won't find a model that outperforms the random walk.

[2] Some of the choppiness is due to stock splits. GM has had three stock splits since 1950, including a 2-for-1 split in October 1950, a 3-for-1 split in September 1955 and a 2-for-1 split in March 1989. The company has also adjusted its stock after spinning off subsidies such as Hughes and Delphi.