For a particular stock, the analyst usually looks at companies with similar growth rates or similar companies in different industries to find "comparables" which are then either tweaked higher or lower based on factors such as quality of management, size or stability of earnings. The problem is that this becomes the tail wagging the dog because everything is just viewed relative to everything else, not necessarily where they should be based on sound principals of finance. The big answer as to who really controls market valuation is that it is the retail investor, many of which do not know the first thing about stock market valuation, that really determines the market price. This is especially true today now that mutual funds have made it a practice to keep as little cash as possible on hand and will let inflows and outflows alone mostly control their net portfolio position. Stock market valuations are not the main factor driving the market, but it is the overall liquidity environment, a fact that was painfully obvious in the late 1990s when analysts betrayed their cluelessness on true market valuations by coming up with measures such as price to revenue or "price per click" to justify what was in reality just a liquidity bubble as emotional greed permeated the market.
The final point I would like to make regarding the problems with P/E analysis (also applying to other common measures such as enterprise value to EBITA or to cash flow), even for the purposes of comparing between companies, is that the P/E ratio is almost always artificially low given poor quality of earnings used for the analysis.
Given that valuations using P/E focuses on one year's number and not all subsequent cash flows in their entirety, analysts will often add back "non-recurring" charges to try to focus on the company's "true earnings power." The main problem with this it that companies that 10 or 15 years ago would rarely highlight nonrecurring charges, have come to make the reporting nonrecurring charges commonplace. If one takes into account these series of recurring nonrecurring charges, it actually makes a really big impact on the underlying discounting cash flow analysis, and arguable rips whatever theoretical validity of forward P/E analysis away. It also creates an unfair standard when comping to historical ratios where companies were more hesitant to report a charge as nonrecurring. In addition, it is also noted that the principal paid for acquisitions above and beyond the asset value (known as goodwill) is no longer required to be amortized. While investment bankers stress that it does not effect the ongoing cash flow from continuing operations and therefore is a "noneconomic" cost, it in fact does have a real, albeit nonrecurring, cash cost which can be quite substantial. For companies that make a habit of serial acquisitions, this again can have even more significant implications to a DCF analysis (not to mention that such acquisitions are almost always accompanied by large, nonrecurring charges). Essentially the only penalty for overpaying for an acquisition on the principal side now is simply the after tax cost of capital (if cash acquisition), which in our current interest rate environment is very minimal.
In conclusion, I assert that investors should be wary of the market valuation measures that most analysts throw at the public. The abuse of reporting of non-recurring expenses combined with an overly simplistic approach towards assigning a present value to a company's potential earnings stream seriously compromises the analysis. There is no reason a stock should ultimately trade within a certain P/E range over time. At the end of the day, it is really market psychology, the historical factors of fear and greed, combined with the liquidity environment largely provided by the Federal Reserve, that is going to determine where stocks trade. I suggest that the financial markets would be better served by the use of more sophisticated valuation models for determining where they are willing to buy or sell a stock.
