Tax hikes, according to IMF research praised by Paul Krugman

The surprising thing is this is an IMF study that usually gets cited to show that spending cuts don’t grow the economy—that “expansionary austerity” is a mere theorist’s dream.  But this same research also provides evidence that tax hikes cause more trouble than spending cuts in the short run.  
Quick summary of the method: The economists looked at 173 “fiscal consolidations” in rich countries, times when governments decided to reduce the long-run deficit.  They then checked to see whether consolidations based mostly on tax hikes turned out better or worse than ones based on spending cuts (Inside baseball: They followed a version of the Romer and Romer event study methodology, but applied it to exogenous-looking fiscal tightening instead of exogenous-looking monetary tightening).  
Here’s my favorite graph, one that I think smart people keeping drawing the wrong lesson from if they notice it at all.  It shows what happens to consumer spending and real output after the two different kinds of fiscal tightenings.  Time in years on the x-axis, percentage changes in spending on the y-axis:
Both GDP and consumer spending tell the same story: Spending cuts are the less painful path to fiscal rectitude.  When countries tried to get right with the bond markets, this IMF study found that  nations that mostly raised taxes suffered about twice as much as nations that mostly cut spending.  
The authors give a possible (I said possible) explanation of the results: Central banks play nice when governments cut spending, loosening up monetary policy.  They’re not as nice when governments raise taxes.  I’m sure somebody out there will say that the Federal Reserve and the ECB have run out of ammo so we can ignore Figure 9.  To those people I say QE3 and sovereign bond purchases.  Central banks still do stuff and they do more when things look bad: In 2012, you can just read the newspaper and you’ll see.  
Plus, possible.  
We might want to meditate on Figure 9 out here in the reality-based community, since both the U.S. and Europe will be spending some time this fall wrestling with how to get our fiscal houses in order. A benevolent social planner would like to take the least-cost path to solvency, a path probably based on spending cuts and loose money.  
Keynesian Coda: Notice that the graphs are saying that the tax multiplier is bigger than the spending multiplier, at least in these settings.  Quite the opposite of undergraduate Keynesianism.  This isn’t the final word on the matter, but if you’d like to see another study of multipliers that doesn’t fit neatly into the Keynesian box–and written by top New Keynesians–check out the abstract and conclusion of this paper by Blanchard and Perotti.