It ain't what you don't know that gets you into trouble. It's what you know for sure that just ain't so. Mark Twain.
At one level, it is easy to understand the popularity of the deficient demand hypothesis. First off, it's pretty much the first thing any undergrad learns in the way of macro theory. Second, they tend to learn it as a factual and self-evident explanation of the way the economy actually operates; not as an hypothesis or interpretation of the way an economy may work. Third, it is apparently easy to "see" evidence of deficient demand out there (much in the same way people can "see" the Phillips curve here?). They can "see," for example, that many firms cite a lack of product demand as a reason for holding back on making commitments to future capacity (including the addition of fulltime workers). The flip side of deficient demand is a "savings glut." People claim to see this as well; for example, in the form of low inflation and low Treasury yields.
Well, heck...I can see these things too. But the question, surely, is not what we record in our measurements. The question is how these measurements are to be interpreted. Interpretation (or explanation) necessarily entails a theory. (I define theory as a set of assumptions leading to a set of conclusions through the use of deductive logic.) And it is frequently the case that a given phenomenon has more than one plausible (or no less plausible) interpretation.
Before I go on, I want to make something clear. I do not disapprove of the practice of asking people what motivates their behavior. I would, in fact, like to see more in the way of this type of field work; see here. Having said this, we need to be careful in interpreting any given survey response as supporting one or some other theory. This is especially true in macroeconomics, where general equilibrium (system wide feedback effects) are likely to be important. According to Krugman, this is what makes macroeconomics hard.
And he is right. Unlike partial equilibrium analysis, it is conceptually difficult to identify independent "supply" and "demand" schedules in a dynamic general equilibrium system--everything is interelated, after all. Consider, for example, a shock that contracts the supply of some object (oil, credit, etc.). The ensuing price rise may lead oil-intensive sectors to curtail not only their demand for oil, but also their demand for a variety of complementary intermediate inputs. To the suppliers of these inputs, this will look very much like a "lack of demand" for their products. They are obviously not wrong for saying this. But upon hearing such reports, it would be rather hasty to conclude that these reports necessarily imply that "aggregate demand is too low."
With that out of the way, let me now discuss the deficient demand hypothesis. Some people may have been led to think that I don't believe that there is a demand problem. That's not quite true. It seems clear enough to me that the aggregate demand for investment (broadly defined to include investment in recruiting activities) is depressed. I'm just not very sure of the source of this depression. Understanding what these "fundamentals" are is necessary, I think, if we want to identify an appropriate policy response (more generally, the properties of an optimal policy rule).
Let me try to formalize what I mean here by way of a simple model that is, I think, sufficiently flexible to accommodate a range of views. I describe the model in some detail here (it is an OLG model). The fundamental economic friction is limited commitment, leading to an asset shortage; see here. The asset shortage gives rise to role for government debt (or money). The model highlights a portfolio choice problem: people must decide how to allocate a given amount of savings between two available asset classes, money and capital.
The key parameter in the model is the expected return to capital investment. A shock that depresses this expectation leads wealth-maximizing agents to substitute out of capital and into money. There is a collapse in aggregate investment spending (leading to a decline in future GDP); and there is a corresponding "flight" into government securities (money). For a fixed stock of money, the price-level drops (reflecting the increase in the market value of money); that is, the shock is deflationary. In short, the model generates something that resembles what we experienced in the recent recession.
Now, let's imagine that these expectations remain stubbornly depressed. What are the policy implications? Would it help if I told you that this is a model where increasing government spending (on investment), or lowering the interest rate (on government securities), or increasing the inflation rate, all serve to stimulate real economic activity? (This is, in fact, a property of the model.) It's tempting, isn't it? The economy is depressed and you have the tools to "fix" it. But hold on a minute. Before proceeding with your government stimulus plan, shouldn't you first ask why expectations appear to be so stubbornly depressed?
At the risk of oversimplifying, imagine that there are two possible answers to this question.  agents are rationally pessimistic; they forecast low returns on their investments because the environment (including the likely evolution of future government policies) dictate such a view.  agents are irrationally pessimistic; they forecast low returns on their investments for psychological reasons (e.g., Keynes' animal spirits).
Under interpretation , the decline in investment and flight to money is a rational response to an unfortunate event that has altered the economic landscape. One might imagine a rather muted enthusiasm for government stimulus under this interpretation. Under interpretation , the depression is caused by a collectively irrational flight to government bonds (Brad DeLong) or money (Nick Rowe); see their debate here. Under this interpretation--which is what I think most people have in mind when they speak of deficient demand--there is a strong case to be made for government intervention. There is obviously room here for reasonable people to disagree.
I want to conclude now with what I think is a shortcoming of the deficient demand hypothesis. It seems to me that the hypothesis leads us to think of a recession the way we might view a deflated balloon. The fundamental structure of the balloon remains intact, even if it is deflated. All that is needed to get back things to normal is a puff of fresh air. And if the private sector seems unwilling or unable to blow, then let the government do it instead. What could be simpler and more obvious?
My own observations over the years have led me to view recessionary events more like Humpty Dumpty after his great fall. Hands up all of you who think that the financial crisis had a severe impact on the economy's "structure." What do I have in mind here? Think about the disruptions that must have occurred in the form of terminated relationships (firm/worker, creditor/debtor, supplier/retailer, etc.). Think about the disruptions created out of a growing realization that resources have been misallocated (i.e., investments that looked good ex ante, now look like a bad idea ex post).
The main point is that a crisis destroys capital, broadly defined to include relationship capital--the glue that keeps the structure of economic relationships intact and productive. Sure, a breach in this structure may lead to deflated expectations and deficient-demand-like phenomena. But do not confuse symptoms with causes. The process of reallocating resources and rebuidling relationships after a traumatic event like the recent financial crisis is likely to take some time. This would be true even if all the king's men knew how to put Humpty Dumpty back together again.