answers to P. Skott -- 2

[Jim Devine <JNDF%LMUACAD.BITNET@xxxxxxxxxxxxxxxxx>]

part II of II

PS: JD suggests that price is a constant mark-up on unit costs and
that (assuming fixed coefficients, excess capital capacity and no
raw material inputs) this yields a constant real wage rate. It seems
to me that this constant real wage curve is precisely the supply
side of PD's Marshallian scissors. Given the special assumptions
(fixed coefficients etc.) the system takes a particularly simple
form: supply is infinitely elastic at the mark-up-determined real
wage; the supply side determines real wages and aggregate demand (at
that real wage) gives the level of production.

JD: In general, this makes sense to me for a general description of
price determination in either a competitive market under conditions
of uncertainty and imperfect information or in a monopolistically
competitive market. It is true that in the Carlin/Soskice framework,
the constant mark-up on unit costs (assumed to be a flat curve due
to the existence of unused capacity rather than due to simple fixed
coefficients) and the price-determined real wage (PDRW) curve should
be seen as only one side of the determination of real wages. Given
money wages, the mark-up and the absense of diminishing returns to a
fixed factor determines a flat pseudo-supply curve.

The complication in Carlin/Soskice is that along with the PDRW curve
is the wage-determined real wage (WDRW) curve, which slopes upward
because workers are more able to raise money wages in line with
prices as employment rises (unemployment falls). To my mind, both of
these lines represent goals that the two sides of the wage bargain
struggle to achieve. The first curve is the goal of employers (given
their profit-rate goal), while the WDRW curve represents the goals
of workers. (C/S don't see the former, only the latter. But no
matter.) In trying to attain these goals, prices and nominal wages
rise or fall until the economy attains the WDRW=PDRW equilibrium. As
Carlin and Soskice note, however, this equilibrium may not be
unique: the WDRW may be a "fat line" (a range) rather than a
geometric line, so that there is a range of possible solutions
intersecting the PDRW line. In this case, the PDRW is more crucial
and there is (at least along some range) an infinitely-elastic
pseudo-supply curve. So in this range, the aggregate demand for
products determines the level of production (y) and the level of
hiring (y/APL).

(BTW, Victoria Chick argues that history, i.e., last period's money
wages, should be brought into the C/S story.  I agree completely and
don't see this as a hard task. C/S stress equilibrium conditions,
but history is crucial since the economy is usually not in WDRW=PDRW
equilibrium. I guess this is a hard task if one is wedded to
equilibrium economics.)

PS: Differences between JD's system and PD's general argument only
seem to arise if it's assumed that the actual real wage deviates
from the mark-up determined real wage [the PDRW]. This deviation
could result from sluggish price movements in combination with
changes in the money wage rate: the realised mark-up is too high
(the real wage too low) and firms reduce prices but a falling wage
level maintains the actual mark-up at high level. This story - which
invites questions about the causes of sluggish price adjustment and
the formation of wage expectations - implies short run
disequilibrium, and as a result the Marshallian scissors don't cut
quite as neatly as PD has it. Is this a reasonable interpretation of
your argument, JD, or have I got it completely wrong? ...

JD: If I understand it correctly, the difference between PD and
myself might be thought of for two cases. (A) In the case of total
and utterly atomistic competition, the difference is that PD assumes
that when aggregate demand is below the full-employment level, firms
hire according to w/p = f'. Now, I see that as *one* possible result
(for high w/p), either when the sales constraint is non-binding or
at the point that Keynes emphasized, what I'll term the "kink"
(where both the sales constraint and the production function are
binding). There are also the situations with low w/p < f', including
the case where w/p equals its FE equilibrium level.

In either the kink case or this latter case, the idea of atomistic
competition as involving universal price-taking becomes more
nonsensical than usual, since _firms can't sell all that they
produce at given prices_. The realized gap between f' and w/p is
"too high" relative to the f' = w/p kink. (I'd forgotten that the
firm's inability to sell all they can produce at the given price --
and the textbook image of perfect competition -- went out the window
at Keynes' kink. But it does.)

In this latter (w/p < f') case, are real wages "too low"? Yes and
no. They are too low relative to f' at the level of employment below
FE given by the sales constraint. But they may be "too high"
compared to the equilibrium real wage if employment were actually
at FE.  Or it might be below this FE employment real wage. They are
usually _too high_ relative to the marginal disutility of labor (the
labor supply curve), as Keynes emphasized. It is possible, though,
that real wages might be low enough to be actually on the
labor-supply curve, if the sales constraint intersects that curve.
(This seems unlikely, since the aggregate LS curve is inelastic.)

However, it is reasonable, given uncertainty and the normal absense
of complete information, that firms would follow a mark-up-style
"rule of thumb" in setting prices, even in atomistic markets, even
at FE. (Under atomistic competition, the mark-up on variable costs
would be high enough to cover both fixed costs and normal profits.)
At less than FE, in a situation where general equilibrium does not
apply, such pricing rules also apply. Given the absence of
diminishing returns to a fixed factor, i.e., excess capacity
coinciding with unemployed labor, and given a money wage, we might
model prices as p=(1+m)w/APL -- where m is the mark-up and APL is
the (constant) average product of labor. It seems quite likely that
atomistically competitve firms would keep the same mark-up as they
did at FE. After all, being atomistic, they don't know what's going
on. Then, a hypothesized fall in w automatically leads to a fall in
p, leaving w/p unchanged at APL/(1+m). Lowering m would raise w/p,
sliding up the vertical sales constraint toward the kink where the
production function binds. Raising m would lower w/p, moving in the
opposite direction. Unless the demand for products or the APL
changes, there will be no change in employment.

BTW, in the above, I assume that price changes do not change the
quantity of products demanded (no Pigou or Keynes effect, no debt
deflation, etc.)

(B) If we drop the atomistic competition model, one might model the
difference between PD and myself differently. For a firm with some
price-setting power, w/p is set equal to (1+m)f', where m =
1/(perceived price elasticity of demand). One might argue, in an
effort to reconcile PD and myself, that the notion of a sales
constraint can be summarized as saying that when the product market
is (more) quantity-constrained, the firm's perceived demand curve
becomes (more) inelastic, so that w/p falls _relative to_ f'. This
misses a lot: in the vertical "sales constraint" case, m goes to
infinity and the formula doesn't apply.

One might explain the inelasticity of the effective demand-for-labor
curve by the sales constraint and its failure to be _totally_
inelastic by long-term employment relations: employers expect to
have many of the same workers in the future when the labor market
will no longer be quantity-constrained. So they do respond a little
to wages in determining employment (as in early editions of
Dornbusch and Fischer's MACROECONOMICS).

I hope that this overly-long missive answers your questions.

sincerely,

Jim Devine   BITNET: jndf@lmuacad    INTERNET: jdevine@xxxxxxxxxxxxxxx
Econ. Dept., Loyola Marymount Univ., Los Angeles, CA 90045-2699 USA
310/338-2948 (off); 310/202-6546 (hm); FAX: 310/338-1950