When formulating a financial decision, prediction, or guess, you have to start somewhere. The initial price or number you pick turns out to have enormous influence on your final conclusion. That is, you anchor to this initial price and make adjustments from there. These adjustments are frequently inadequate. Such biased results in forecasts underweight new information and can thus give rise to predictable, and yet surprising, forecast errors.
This anchoring bias impacts many financial decisions. For example, the awards from lawsuits are influenced by the plaintiff's initial demand—the plaintiff gets more if he or she requested more. In real estate, people are unconsciously influenced by arbitrary posted prices. In online auctions, the prices bid are anchored by the non-binding "buy-now price." And earnings forecasts by financial analysts are biased towards the previous months' data as an anchor. Issuers of initial public offerings anchor to the midpoint of the initially filed pricing range. This may cause them to become complacent about assessing the new information gathered in the ‘going public' process and thus may contribute to the underpricing of IPOs.
A particularly strong anchor is the purchase price of a stock. You might have heard (or said), "if the stock would just go up to what I paid for it, I'd sell it!" Not exactly a ringing endorsement of the investment. Our assessment of the stock performance and how we feel about it is determined by this buy-price anchor. This can lead to odd behaviors like selling your winners too soon and holding on to your losers too long. This is called the disposition effect (posting on this coming soon).
Daniel Kahneman and Amos Tversky, 1974, "Judgment under uncertainty: heuristics and Biases," Science 185 (4157), 1124-1131 and Sean Campbell and Steven Sharpe, 2008, "Anchoring bias in Consensus forecasts and its effect on market prices," forthcoming in Journal of Financial and Quantitative Analysis.