and the ratio of consumer credit to Net National Income. Lower values of the first
index and higher values of the second indicate a relaxation of credit constraints. In
all the countries of this sample, both indicators show clear evidence of relaxation of credit
constraints over time. In neo-keynesian models with liquidity constrained individuals, like
Galí, López-Salido and Vallés [2003], the positive effects of government spending on GDP
and private consumption fall as the share of constrained individuals falls, since for non-
ROT individuals private consumption is negatively correlated with government spending.
On the other hand, these models seem unsuited to explain the decline in the effects of
government spending on private investment: the latter is rather insensitive to government
spending to start with, and in any case it falls less as the share of ROT individuals falls.
(ii) The stronger real interest rate response
Can the larger response of the real interest rate in S2 help explain the decline of the
GDP response in the same period? Panel A of Table 20 displays the cumulative response of GDP to a spending shock in S1 and S2 at 1 and 3 years (the same as in Table 6);
Panel B displays the same response, but with the ex-post real interest rate response shut
off; that is, in each quarter the nominal interest rate response is forced to be equal to the
inflation rate response over the last 4 quarters.