HOW DO FIRMS REACT TO SURPRISING CHANGES TO DEMAND?: A VECTOR AUTOREGRESSIVE ANALYSIS USING BUSINESS SURVEY DATA
| Author | Robert A. Buckle,Chris S. Meads |
| Date | 01 November 1991 |
| DOI | http://doi.org/10.1111/j.1468-0084.1991.mp53004006.x |
| Published date | 01 November 1991 |
OXFORD BULLETIN OF ECONOMICS AND STATISTICS, 53,4(1991)
0305-9049 S3.00
HOW DO FIRMS REACT TO SURPRISING
CHANGES TO DEMAND?: A VECTOR
AUTOREGRESSIVE ANALYSIS USING
BUSINESS SURVEY DATA
Robert A. Buckle and Chris S. Meads*
I. INTRODUCTION AND BRIEF SURVEY
There are essentially two views in contemporary business cycle theories
about how firms react to unexpected changes in demand. In the new-
Keynesian approach, which is characterized by sluggish price adjustment,
surprising changes in demand result primarily in unexpected changes to the
quantity of output and employment in the short run. In the new-classical
approach, surprising changes to demand cause unexpected price changes
which in turn precipitate previously unplanned quantity changes, at least in
the short run.
Price expectation errors are the crucial link between surprising changes in
demand and output in the new-classical model. This link evolved from an
attempt by Lucas (1972, 1973) to provide an equilibrium explanation for co-
movements of output and prices generally. In the initial models developed by
Lucas, imperfect information is the deits ex machina to resolve the conflict
between the assumption of market clearing and the existence of business
cycles. This idea is typically represented by a variant of the model used by
Sargent and Wallace (1975), see for instance McCallum (1980):
y,=a(m,p,)+u, (1)
y,=ß(pE,_1p,)+v, (2)
(3)
Notationally, y,, m, and p, denote logarithms of real aggregate output (or
demand), the money supply, and the price level respectively. Equation (1) is
an elementary aggregate demand function. Equation (2) is an aggregate
supply function based on the idea that producers cannot accurately dis-
*Comments by the Editors of the BULLETIN are gratefully acknowledged. We also wish to
thank Lew Evans, Fraser Jackson and Kun-Hong Kim and participants at a Victoria University
of Wellington seminar for helpful comments offered during the preparation of this paper.
Remaining errors are our own.
451
452 BULLETIN
tinguish between general and relative price components of prices for their
own output. Equation (3) is a policy reaction function describing the
generation of the money stock. E1- 1x1 denotes the rational expectation of x,
computed using the equations of the model and the values of all relevant
variables realized in periods t - j (for j =1, ... and earlier). The symbols, u,,
y,, z denote random disturbance terms representing unsystematic forces
impinging on the economy. These disturbances are assumed to have zero
means and constant variances and to be stochastically independent of past
values of all variables and disturbances.
Table 1 shows that in this rudimentary new-classical model, price and
output expectation errors are both related to contemporaneous stochastic
disturbances in the aggregate demand, supply and policy functions. In
particular, output and prices will be underestimated by firms when there is
a surprising rise in demand, whereas if there is a supply shock the model
predicts an inverse relationship between output and price expectation errors.
The key assumptions used to derive these relationships are that agents form
rational expectations, prices are flexible and markets clear continuously. The
last two assumptions are dispensed with in the new-Keynesian approach.
New-Keynesian macroeconomics almost exclusively focuses on explaining
the 'sluggish' behaviour of prices and wages and the implications for the
business cycle (see Chapters 8 and 9 of Blanchard and Fischer, 1989 for a
review). The consequence of this type of nominal price behaviour is that
prices are prevented from moving equi-proportionately to nominal demand
movements so that real output becomes a residual rather than a choice
variable. In particular, if prices are predetermined there need be no
systematic relationship between price and output expectation errors -
although that will depend on the precise nature of the pricing rule.
This point is easily illustrated by introducing an alternative to the new-
classical pricing rule. An example is the pricing rule suggested by McCallum
(1982) which is similar to the idea of staggered wage contracts first used in
the context of macroeconomic models by Fischer (1977) and Taylor (1979).
Prices for period t are set at the end of period t - 1 at a level that is expected
to make the quantity demanded equal to a weighted average of y11 and the
'natural' rate of output, 9,. There are two basic ideas justifying this price set-
ting rule: firms find it optimal to meet all demand at the quoted price; and
firms experience adjustment costs whenever y, differs from y1 -1 but also suf-
fer opportunity costs whenever there is a discrepancy between y, and 9,. If
both of these cost functions are quadratic, producers will aim at some value
of output between y,1 and 9,, which can be denoted as lAy, -1 + (1 - A) 9,],
where O < 1, reflecting the relative costliness of output changes. Conse-
quently, the price level is set to satisfy equation (1) expectationally but with
y,=Ay,_1+(l A)E,_19,,i.e.
Ay,_1 +(1 A)E,_19,= a(E,1m,p,) (4)
To continue reading
Request your trial