Barrdear, John
(2013)
Incomplete information and the idiosyncratic foundations of aggregate volatility.
PhD thesis, London School of Economics and Political Science.
Abstract
This thesis considers two interrelated themes: the emergence of aggregate volatility from idiosyncratic shocks and optimisation under incomplete information when,
for reasons of strategic complementarity, agents are interested in both simple and
weighted averages of their competitors' actions.
I first develop a model of Bayesian social learning over a network. Unlike earlier
literature that abandons one of the assumptions that agents (a) act repeatedly; (b)
are rational; and (c) face strategic complementarities, I obtain tractability for arbitrarily large networks by also assuming that agents do not know the full structure of
the network, but do know its link distribution. An AR(1) process for the underlying
state induces an ARMA(1,1) process for the hierarchy of expectations, with current
and lagged weighted averages of agents' idiosyncratic shocks entering at an aggregate
level. For sufficiently irregular networks, these shocks do not wash out, thus causing
persistent aggregate effects.
I next apply this to firms' price-setting problem, demonstrating that even when
firms possess complete price
exibility, network learning induces considerable persistence in aggregate variables following monetary and real shocks and that network
shocks plausibly represent a source of aggregate economic volatility.
Finally, I explore price setting under monopolistic competition when facing TransLog preferences. I solve explicitly for a firm's best-response pricing rule under full
information and show that in partial equilibrium under incomplete information, larer firms will focus on their marginal costs while smaller firms will place more weight
on changes in consumer preferences and competitors' prices. In general equilibrium,
I estimate the effect of two distinct sources of real rigidity that emerge from TransLog preferences: the well-known curvature in demand and the dramatic increase
in complexity of firms' signal-extraction problems. With non-uniform preferences,
the model represents another channel through which idiosyncratic shocks can cause
aggregate volatility.
Actions (login required)
 |
Record administration - authorised staff only |
Download statistics
Download statistics