The current debate over the size of the "government spending mulitiplier" is a perfect measure of the sway that conventional economic theorizing continues to grip the minds of people who should know better.
At the center of the theorizing is the income-expenditure identity: Y = C + I + G. This identity is not a theory; it is something that is true by definition. The theory comes in by way of behavioral assumptions that are imposed on C and I. All that is left is to determine how an exogenous change in G manifests itself as a change in Y. The government spending multiplier is dY/dG. Now all economic historians have left to do is to try to estimate the size of dY/dG. They frequently "discover" that dY/dG > 1. That is, it appears that for every dollar the government spends, the national income appears to increase by more than a dollar. Conclusion: Obama's stimulus package is a good idea.
There you have it. The only puzzle remaining is why it takes 4 full years of training to receive a PhD in macroeconomic theory; and why it should take a further 6 years to become a tenured professional by publishing papers examining what we all know to be the self-evident truth embedded in this "really useful ad hoc model."
Unfortunately, I am apparently one of few who have trouble absorbing this simple theory. But perhaps there is still some hope for me. I just need a few questions answered.
 What does this theory predict concerning the optimal level of Y? Is more Y always to be preferred to less? When Y was expanding rapidly above trend during WW2, were people really made materially better off? Were people made happier by their long hours employed in military manufacture and European adventures? Did people really enjoy the rationing of foodstuffs and gasoline associated with the increase in G? Was the general destruction of capital (both physical and human) during WW2 really associated with increasing wealth levels?
 Do *measured* increases in Y associated with increases in G correspond in any meaningful way to *true* increases in Y? Do we not know that when the government pays for something, the "value" of this purchase is measured by cost (instead of market-value) by the National Income and Product accountants? Is this not a serious problem in assessing the "true" value of an increase in G? Are measured government spending multipliers simply a figment of this questionable accounting exercise?
 What are we to make of multiplier estimates that are above unity? Is this a linear approximation? Does it represent the value of the marginal dollar spent? (If it does not, then is the implication that the entire economy should simply be nationalized?).
 Do the "microeconomic" details concerning the added G not matter at all? Does it matter, for example, that almost $100 million in "stimulus" money is being directed to the Milwaukee Public School system to construct new schools (a district that has been closing schools as a result of declining attendance; see here)? Are people still willing to argue that "digging up holes and filling them in" is a wise use of economic resources?
 What does this simple theory identify as the "cause" of recession? An exogenous decline in private sector "sentiment?" What does this mean? Is this purely pyschological? Can we not expect private sector agents to make estimates of the future based on the best information available? Does the government have better information? If so, why does it not make it available? Can declining consumer and business sector "confidence" not be reasonably interpreted as a symptom, rather than a cause, of any economic downturn? How can we know, one way or the other? What evidence can be brought to bear on the idea that recessions are "inefficient?" Is the arrival of a harsh winter that freezes construction to a halt also inefficient? Should the government promote construction activity during a freezing winter?
 What is the government spending multiplier in Japan since 2000?
 Should we follow Christina Romer's advice and take employment as a metric of economic welfare? Has she not studied economic theory? (Actually, I know the answer to this last question--it is no). I recall reading an article from the TASS news agency, published in 1957 that "the unemployment rate in the Soviet Union, as in previous years, was equal to zero." Should we seek "full employment" along the old Soviet model? Is this how we are to measure success?
I have many more questions, but these will do for now. Of course, some of these questions are rhetorical in nature. But many are not; I genuinely do not know the answers to them. Evidently, the likes of Romer, DeLong and Krugman do know the answers to them. But perhaps this is because none of these esteemed academics are macro theorists. They likely view my ignorance on these matters as evidence supporting their proposition that too much macroeconomic theory distorts the mind. Thank goodness we have these clear-thinking folk around to set me straight!
Brief Bio: Born in Vancouver, British Columbia. Flowering career in construction sector (drywall taper) aborted by severe recession (1982). Received Ph.D. in Economics from the University of Western Ontario (1994). Taught as a university professor for over 20 years. I now work in the Research Division of the Federal Reserve Bank of St. Louis and I write this blog mainly in my spare time (it is not a part of my formal duties). I welcome comments and (constructive) criticisms. Feel free to email me if you would like to discuss issues in greater detail.
Believe those who are seeking the truth. Doubt those who find it. Andre Gide