What do we learn from Blanchard and Quah decompositions of output if aggregate demand may not be long-run neutral?

This paper structurally interprets empirical results obtained from Blanchard and Quah (1989) decompositions of output into permanent and transitory shocks. This is done based on assumptions about the qualitative responses of variables to structural shocks that are consistent with many different economic theories. Impulse responses of output to a permanent shock typically behave in a particular way before World War I that is unusual compared to post-World War II responses. Also, permanent shocks typically explain a larger share of output variance in that earlier period. We show these two empirical results support the hypothesis that a positive (negative) aggregate demand shock had a permanent positive (negative) effect on output in the pre-World War I period. Thus further support is obtained for that hypothesis based on more robust empirical findings and less restrictive structural assumptions. Another empirical result is that impulse responses from the data for post-World War II developed economies are typically qualitatively consistent with the effects of structural shocks from a standard textbook macro model. We show that Blanchard and Quah's statistical model will obtain impulse responses that appear consistent with simple textbook theories as the parameter measuring aggregate demand's long-run output effect is varied over a specific range. This analysis provides a range of non-neutralities for which results from Blanchard and Quah decompositions of output could be mistaken as structural.