The trouble with econometric models is that they present correlations disguised as causal relations. The more obvious confusions between correlation and causation can often be avoided, but there are many subtle ways to confuse the two; in particular, the language of econometrics encourages this confusion.
The problem is so serious that econometric models are usually ineffective even for estimating supply and demand curves, despite efforts to use them for markets as diverse as money, gasoline and imports. While more experiments and better data analysis can sometimes be used to attack the problem, it is difficult to arrive at any general rules for solving the problem. It is doubtful, though, that traditional econometric methods will survive.