The Effects of Monetary Policy in a Multi-Sector New Keynesian Model: An Analysis Using Recent Micro Evidence on Price Changes
Master Thesis by Aysegül Sanli
August 9, 2005
Abstract:
Economic modelling generally assumes
one-sector to represent the whole economy, whereas real economies consist of
many sectors with varying price durations. We try to model this variety by
setting up a multi-sector New Keynesian model based on the model outlined in
Gali (2002). The Calvo parameter estimates of our model are calculated using the
data set provided by Bils and Klenow (2004). We group similar price durations as
sectors and form ten sectors to represent the whole economy. Our first aim is to
study the responses of the multi-sector economy to policy shocks. The outcomes
show that aggregate price, interest and output display interesting dynamics
under such a setting. Then we try to match the responses of the multi-sector
economy with one-sector models. We find that for our baseline model, the
aggregate responses of the multi-sector economy can be well mimicked by a
one-sector model with a price duration of around 2.9 months. This model is able
to duplicate the aggregate responses, whereas the variety in the dynamics of
responses can only be captured with at least two sector-models. We test our
results under different parametrization. The proposed price duration for
one-sector model is still able to match the aggregate impulse responses well but
we see that it is not very precise in matching the volatility of the
multi-sector model. Under similar settings, the two-sector performs better. Next,
we change the labor assumptions of the baseline model and assume specific factor
markets. The results show that prices and inflation response by little to policy
shocks now. The new impulse responses are well matched with one- and two-sector
models with the proposed price durations. Whereas in terms of matching the
volatility, onesector estimate performs poorly and the new price duration
increases to around 4 months. However, this value lacks the precision of the
previously proposed price duration in matching the aggregate response levels.