For persons who insist that aggregation at the high level at which it is typically done by textbook Keynesians (such as Krugman) is acceptable, I offer this Krugman piece  as Exhibit A for why such aggregation is poisonous to sound economic reasoning.
As Mario points out, wage adjustments are best thought of not as all wages falling (or all wages rising). There is no lump of L in the economy hired to do some all-purpose task aimed at producing, in combination with a lump of K, a lump of Q to be bought by a lump of spending (C + I + G + [X – M]).
Even if some measure of average wages is falling, some wages fall relative to other wages (meaning, some wages rise relative to others). And all wages that fall do so relative to the prices of capital goods — some of which are complementary to labor, others of which are substitutes for labor.
Such changes in the pattern of wages are necessary to allocate labor from less-productive to more-productive tasks.
Moreover, the demands for some forms of labor are elastic over the relevant range of wages — meaning, that as wages for such labor fall, the increased quantity of this labor demanded by employers results in these employers paying out more in total wages than before the wages fell. (It would be as if, say, McDonald's cuts the price of the Big Mac by ten percent and the resulting increase in the quantity of Big Macs demanded by buyers is twenty percent.) If the total spending power of workers as a whole is a function of the total amount of wages paid to workers as a whole, then a fall in wages when demand for labor is elastic over the relevant range of wages means that these wage cuts will contribute to higher consumption spending.
I have more to say about this silly Keynesian take on wages — more that, time permitting, I might offer in a follow-up post.