Another disadvantage is that, even though earnings may provide a useful reflection of value over a relatively short horizon in most situa tions, that is not always true. For example, one could not reasonably use
earnings prediction to study the impact on firm value of changes in the
market value of U.S. banks' investment portfolios, because, so long as the banks intend to hold and do hold the investments to maturity, earnings will never reflect that change in value. (The very accounting choice at
issue here [whether to mark-to-market], itself, affects earnings and pre
cludes the use of earnings as an object of prediction!) More commonly, earnings may not yet reflect the impact of an event, even though stock
prices do. For example, at this point in history, it would be difficult to assess the reliability of alternative measures of environmental exposure
through reference to associations with subsequent earnings impacts, because those impacts have for the most part not yet occurred. However, because stock prices should now reflect estimates of those future impacts, one could assess the reliability of various measures through examination of associations with stock prices-assuming that the stock market's esti mates provide a sufficiently well-informed benchmark.v