Yet another chart

Courtesy of Antonio Fatas

I compare the change in real GDP in the 2008-2012 period with the change in real government consumption during the same years. This variable is an indication of the fiscal policy stance during those years. I include all OECD countries in the sample (removing some, including only the Euro countries makes no difference to the slope or fit of the regression). The correlation is very strong with a a coefficient that is not far from 1.

Ok. We have OECD nations subject to the GDP shock of 2008, and how did they react since that time under the following assumptions, They are agents with a G and an Private and Other components that meet the accounting identity for GDP. So a pure exogenous shock to any country, when at balance, would affect all components of GDP by the same percentage. For example, external bankers suddenly change interest rates, external oil producers suddenly switch the flow, something that affects all  components of GDP the same. That condition means countries should fall along a 45 degree angle on this chart, like Greece.  Greece was dominated by an external shock, and Antonio also claims, a specific shock to G, one of the components.  That is what he is showing, countries where G did not follow the fluid pattern, and what were the effects.

Note, absent any external shock, countries should be along the horizontal. On that line, GDP change is zero so any change is internal to the components and they offset. So two things to control for, how much was a macro shock to all components, and how much was a specific shock to a component.  Well Greece defaulted, and G got a gain from that, a multiplier greater than one. So, when GDP fell .25 units, and Private fell by the same ratio, and G only fell by .17 units, then we have a clear winner, default made gains over the period of the chart. But otherwise, Greece lies right along the 45, all sectors paying the same ratio for GDP.
Note the few countries that managed to escape the external shock, they lie along the horizontal, multiplier equal to one.  Most countries lie between the hoizontal and the 45.  The countries in the middle either had an over reacting Private and under reacting G; or visa versa.  That is the test, but the data points where to look, not much more.

How do we separate cause and effect here? Look at countries that had high GDP growth over the period.  If high GDP growth pushes you above the 45, then we get more growth with less G. As growth is less, the countries batch up below the 45, or along the horizontal. Here G is sticky but growth low.   Clear evidence that multipliers are a little less than one, over all. Let me crank my imaginary statisticalator,..., multipliers over all are .8.

Wait, do that again! OK. A country likely to have a high positive GDP change is more likely to have a smaller change in G. Successful countries rise above the 45 on the appearance of growth. Countries near zero on the GDP change will move toward the horizontal, they are mainly in a group called undetermined. That cluster has about a 5% variance in everything, too close to separate effects. But Antonio says they don't matter, so follow the linear model in the chart. The computed  line is below the 45 degree, so on average, countries are getting less GDP growth for more G, a multiplier less than one.   That is, in response to a positive growth shock, G expands more than the other components.  I have separated the group into those with large, positive GDP shocks and those with negative or no GDP shocks.  The later group are governed by shocks to G, the former shocks to GDP.