The graph shows, instead of a data trace, two types of behavior. The gray bands indicate times when government policies, whether by design or not, were consistent with a Keynesian economic approach. The turquoise bands indicate times when government policies were inconsistent with a Keynesian approach.
Let's review what this means. Here's Mike:
The basic Keynesian idea is this: economic downturns (and meltdowns) can occur and/or be prolonged and worsened when the private sector becomes worried and cuts back. In those circumstances, the government should step in and buy things, lots and lots of things, replacing the shrunken private sector demand. Once the economy picks up again, the government should cut back on its spending and start saving up money, first to pay for its recent spending bout and second to have cash in hand to cover its next necessary spending bout.
Put another way – a government thinking along Keynesian lines will tend to run a deficit when real private sector spending falls below some prior highwater mark. It will run a surplus in years real spending exceeds prior real private sector spending. There may, of course, be exceptions in any given year, but a Keynesian government will generally follow that sort of behavior. A government that runs a deficit when real private sector spending is rising, or runs a surplus when real private sector spending is falling, and behaves this way in general is most definitely not operating under Keynesian principles.
And here's his graph. (Click to enlarge.)
Remember - gray means Keynesian, turquoise means non-Keynesian. It's worth emphasizing that this chart says little to nothing about the intent of an administration to follow Keynes, or not. (Though not following appears to be considerably more deliberate than following - at least since the '70's.*) It only reflects the actual results, as described in Mike's text above. For example, I have no reason to believe that Ike was particularly Keynesian, or that Bush Sr. had a short-term economic epiphany in 1991. Nor, for that matter, that the B. Hoover Obama administration has decided to "step in and buy things, lots and lots of things, replacing the shrunken private sector demand."
But, irrespective of intent, it's results that count - and that is what is being indicated here.
Given all that, let's have another look at the graph, in the context of THE GREAT STAGNATION, which was Mike's point in presenting it. What we see is that from the beginning of FDR's administration (in fairness, from the previous year, 1931) until (an arbitrarily selected) 1975 Keynesian policies were followed 34 of 44 years, or 77% of the time. From 1976 through 2009 (end of the data set) non-Keynesian policies were followed 25 out of 33 years, or 76% of the time.
The symmetry is pretty remarkable, but also pretty misleading. Pick a different inflection point, and the ratios will skew. More importantly, one thing this approach does not address is the magnitude of the Keynesian policy. As the last couple of years have shown, half-hearted and underfunded stimulus might help a bit, but it's not going to get at the core of an aggregate demand shortfall problem. (And, sadly, plays into the hands of the Keynes denialists like Tabarrok.) Also, deficits resulting from reduced revenues - whether induced by tax cuts or depressed tax reciepts in a downturn - have little to no stimulative power. Further, as alluded to above, wandering across the border into Keynesian territory for a single year - as Poppy did in '91 - means nothing. He was running serial deficits anyway, and the recession incidentally provided the other requirement: that the private sectors cut back.
Mike's point is that "Tabarrok is wrong – wrong that Keynesian economics hasn’t been tried, and wrong that it hasn’t worked. And Cowen, it turns out, is wrong about exactly the same thing in his book."
My point is - hey look - here is another example of something that changed at the end of the post WW II golden age, and led into the Great Stagnation. The fact that the Keynesian period included the golden age, and the non-Keynesian period defines The Great Stagnation is, of course, a mere coincidence.
Afterthought: Tabarrak's post (linked above) is sloppy, incoherent, and obviously ideologically motivated. In all honesty, it strikes me as idiotic bullshit motivated by malicious intent. Posts by Scott Sumner in comments are - seriously - even worse. The Krugman derangement syndrome Sumner displays is really quite depressing. And these guys are highly respected economists.
* Afterthougth 2: OTOH, the beginning of the New Deal was less Keynesian than a scattershot approach to something - anything - that might help. Programs like the CCC and WPA were probably motivated more by social than economic considerations. (I'm speculating here - though I'm sure this answer is available, I don't have time at the moment to go find it.) At any rate, Keynes' General Theory wasn't published until 1936, and I'm pretty sure the FDR administration did not have an inside track to Keynes' unpublished thoughts in 1932.