I am going to try to stick to substance, and not do name-calling. But I think I am articulating a model that is in the spirit of Keynes, which probably puts me on the Brad DeLong and Paul Krugman side of the debate.
The basic idea is this: On the one hand, if I stop going to Chinese restaurants and start going to Mexican restaurants, we know that workers laid off by the Chinese restaurants can be hired by the Mexican restaurants. Not a problem. But on the other hand, if I stop going to Chinese restaurants because I want to save more, how does my saving get translated into specific demand for future goods? Perhaps nobody knows what I want to consume in the future, so that my saving does not create effective demand.
I do not want to take the Nick Rowe route, which is to say that I hold my savings as money, and since nobody produces money, that reduces demand. I have never liked that story. Similarly, I don't like Krugman's babysitting co-op model.
Instead, I am going to put together a story that has no financial assets or banks at all. It is my version of "what Keynes really meant." The rest below the fold.
There are four industries:
raw materials for construction
cheap construction (strip malls in New Jersey)
expensive construction (skyscrapers in Manhattan)
When workers want to save more, they buy fewer consumer goods and more raw materials. They invest these raw materials with entrepreneurs, who allocate them between cheap construction and expensive construction.
At the margin, it takes less labor to produce one unit of raw materials than one unit of consumer goods. Thus, if workers shift from consumption to saving, then this creates excess labor that can be employed in construction.
Labor in the cheap construction industry is fixed. All you need in order to produce more strip malls is to allocate more raw materials to producing them. Raw materials put into cheap construction this period will yield a risk-free one percent return, received next period. In this model, I do not care what happens next period. I am focused on this period.
Labor in the expensive construction industry is variable. The more skyscrapers we build this period, the more labor gets used. A skyscraper yields an uncertain return. It will either be + or - 10 percent. Everyone is risk neutral. The probability of success is entirely subjective. Obviously, if the probability of success is estimated at 0.5, you would rather build strip malls. If the probability of success is .6, you would build the skyscraper instead. The border line is a probability of success of .55.
The probability of success of a skyscraper declines with the number of skyscrapers we build this period. Other things equal, entrepreneurs view the first skyscraper as more likely to succeed than the second, and so on. Thus, there is a schedule of subjective probabilities of success.
The subjective probability of success also depends on animal spirits. Nobody knows the objective probability of success. If animal spirits are high, then entrepreneurs think that the tenth skyscraper will have a probability of success of .55. If animal spirits are low, then entrepreneurs think that only the third skyscraper will have a probability of success of .55.
The first point to note is that if animal spirits shift down, then employment will decrease. Raw materials will be shifted from skyscrapers to strip malls, and by assumption this reduces labor employed in skyscrapers without increasing labor employed in strip malls.*
The second point to note is that an increase in savings can reduce employment. More labor is now available for construction. However, the subjective probability of success for skyscrapers has not changed. Therefore, the labor is not employed.**
I could argue that these are PSST stories. There is a full-employment equilibrium embedded in the system somewhere, but it will require changes in prices, reallocation and retraining of labor, and so forth. In these stories, the shocks that require adjustment are changes in entrepreneurs' subjective probabilities or changes in workers' propensity to save. But there are many other phenomena that can require adjustments. We do not need to fit every period of macroeconomic history into a single model.
*In theory, wages could fall, this could reduce the cost of producing consumer goods, and demand for consumer goods could rise. But in the spirit of Keynes, we treat the share of income spent on consumer goods as constant.
**In theory, wages could fall, this could reduce the cost of building skyscrapers, and in theory this could increase the subjective estimate of the probability of success of the skyscraper. But these adjustments do not take place instantly. So along the way, you get lower employment, which means lower income and lower consumption, which causes multiplier effects, and so on.