There's an old joke about economics exams: over the years the questions are always the same, it's just the acceptable answers which change. Much more amusing than the joke though is the possibility that this might actually be true. That is, it's not just a comment upon what is the fashionably correct answer, it could be that the right answers really do change over time.
This is rather an addendum to Madsen's point that an economy never actually reaches equilibrium: I'd add to the issues he raises the point that technology is always developing. Thus any equilibira, even if they were ever reached, would be outdated as soon as they were by this new technology, new method of production, new change in the production mix and so on.
But what has this to do with the correct answers to questions economic changing? Let us take, just as an example, the New (or Neo-) Keynesian school of macroeconomics. This is a response to the New Classical school, itself a response to Keyesianism...and so forth back to the mists of time. The central point of the New K school is that of menu costs. The reason an economy once divergent from equilibrium doesn't immediately or quickly bounce back is because prices are sticky and thus government really should and really can do something about it all.
Now when the theory was first being pieced together in the 80s and early 90s sticky prices or menu costs looked like a reasonable assumption. However, we've seen much more recently that wages are a great deal less sticky downwards than had been assumed. And prices vastly less so:
It is time for everyone to realize that menu costs are a thing of the past. With current information technology, pricing is much more flexible than the production schedule (which gained tremendously in flexibility as well), and those New Keynesian models can safely be shelved now.
Which means that the New K explanations could have been true even if they are no longer so: and it's changes in technology which have made them wrong.
Now whether this applies to all macroeconomic theories is another thing: but I have a feeling that that is possible at least. For they're all trying to model past events and then using those models as predictions of the future. But if the underlying technology of the economy keeps changing, upsetting those modelled relationships, then we haven't really got a secure basis upon which to base our macroeconomics.
Not so much that the correct answers keep changing, but that by the time we've worked out what was correct for the past it is wrong for the present.