Nominal Rigidities
Both orthodox and new Keynesian approaches assume that prices adjust slowly
following a disturbance. But, unlike the Keynesian cross or IS–LM approaches,
which arbitrarily assume fixed nominal wages and prices, the new Keynesian
approach seeks to provide a microeconomic underpinning for the slow adjustment
of both wages and prices. In line with the choice-theoretical framework of
new classical analysis, the new Keynesian approach assumes that workers and
firms are rational utility and profit maximizers, respectively.
As we have seen, new classicists adopt the flexible price auction model and
apply this to the analysis of transactions conducted in all markets, including
the labour market. In contrast, new Keynesians argue that it is important to
utilize the Hicksian (1974) distinction between markets which are essentially
fix-price, predominantly the labour market and a large section of the goods
market, and markets which are flex-price, predominantly financial and commodity
markets. In fix-price markets price setting is the norm, with price and
wage inertia a reality. In order to generate monetary non-neutrality (real
effects) Keynesian models rely on the failure of nominal wages and prices to
adjust promptly to their new market-clearing levels following an aggregate
demand disturbance. Keynesians have traditionally concentrated their attention
on the labour market and nominal wage stickiness in order to explain the
tendency of market economies to depart from full employment equilibrium.
However, it is important to note that for any given path of nominal aggregate
demand it is price, not wage, stickiness which is necessary to generate fluctuations
in real output. Providing profits are sufficiently flexible, nominal
prices could adjust to exactly mimic changes in nominal aggregate demand,
leaving real output unaffected (see Gordon, 1990).
Nevertheless the first wave of new Keynesian reaction to the new classical
critique concentrated on nominal wage rigidity.
7.5.1 Nominal wage rigidity
In traditional Keynesian models the price level is prevented from falling to
restore equilibrium by the failure of money wages (costs) to adjust (see
Figure 2.6). In the new classical models developed by Lucas, Sargent, Wallace
and Barro during the 1970s, any anticipated monetary disturbance will cause
an immediate jump of nominal wages and prices to their new equilibrium
values, so preserving output and employment. In such a world, systematic
monetary policy is ineffective. Initially it was widely believed that this new
classical policy ineffective proposition was a direct implication of incorpoThe
rating the rational expectations hypothesis into macroeconomic models. Fischer
(1977) and Phelps and Taylor (1977) showed that nominal disturbances were
capable of producing real effects in models incorporating rational expectations,
providing the assumption of continuously clearing markets was dropped
(see also Buiter, 1980). Following these contributions it became clear to
everyone that the rational expectations hypothesis did not imply the end of
Keynesian economics. The crucial feature of new classical models was shown
to be the assumption of continuous market clearing, that is, perfect and
instantaneous wage and price flexibility. But, as Phelps (1985) reminds us, it
is often through the rejection of a theoretically interesting model that a
science progresses and ‘even if dead wrong, the new classical macroeconomics
is still important because it demands Keynesians to fortify their theoretical
structure or reconstruct it’.
The early Keynesian attempts to fortify their theoretical structure concentrated
on nominal wage rigidities and the models developed by Fischer (1977)
and Taylor (1980) introduced nominal inertia in the form of long-term wage
contracts. In developed economies wages are not determined in spot markets
but tend to be set for an agreed period in the form of an explicit (or implicit)
contract. The existence of these long-term contracts can generate sufficient
nominal wage rigidity for monetary policy to regain its effectiveness. It
should be noted, however, that neither Fischer nor Phelps and Taylor pretend
to have a rigorous microfoundation for their price- and wage-setting assumptions.
Instead they take it for granted that there is a ‘revealed preference’ for
long-term wage contracts reflecting the perceived disadvantages that accompany
too frequent adjustments to wages and prices (for an innovative attempt
to explain nominal wage inflexibility, see Laing, 1993).
Fischer’s analysis has the following main features and involves the construction
of a model similar to the Lucas–Sargent–Wallace policy ineffectiveness
models discussed in Chapter 5. The output supply equation is the standard
rational expectations Lucas ‘surprise’ function (7.1), where ˙Pt and ˙Pt
e are the
actual and expected rates of inflation respectively:
Y Y P P t N t t
e
t = + α( ˙ − ˙ ), α > 0 (7.1)
Fischer assumes that inflation expectations are formed rationally, P˙t E(P˙
e
= t |
Ωt–1), so we can write (7.1) as (7.2):
Yt YNt Pt E Pt t = + α[ ˙ − ( ˙ |Ω −1)] (7.2)
Fischer’s model abstracts from growth, so wage negotiators are assumed to
aim for constancy of the real wage by setting nominal wage increases equal
to expected inflation. This is given by (7.3):
W˙ E(P˙ | ) t = t Ωt−1 (7.3)
Substituting (7.3) into (7.2) yields equation (7.4), which shows that aggregate
supply is a decreasing function of the real wage (note this implies a
countercyclical real wage).
Yt YNt Pt Wt = + α[ ˙ − ˙ ], and α > 0 (7.4)
For the multi-period contract nominal wage increases are fixed at W˙t = W˙t* .
Fischer (1977) makes the ‘empirically reasonable’ assumption that economic
agents negotiate contracts in nominal terms for ‘periods longer than the time
it takes the monetary authority to react to changing economic circumstances’.
Because the monetary authorities can change the money supply (and hence
inflation) more frequently than overlapping labour contracts are renegotiated,
monetary policy can have real effects in the short run although it will remain
neutral in the long run.
The argument presented by Fischer can be understood with reference to
Figure 7.1. The economy is initially operating at point A. Suppose in the
current period an unexpected nominal demand shock occurs (such as a fall in
velocity) which shifts the aggregate demand curve from AD0 to AD1. If prices
Figure 7.1 Nominal wage contracts, rational expectations and monetary
policy
are flexible but nominal wages are temporarily rigid (and set = W0) as the
result of contracts negotiated in the previous period and which extend beyond
the current period, the economy will move to point B, with real output falling
from YN to Y1. With flexible wages and prices the short-run aggregate supply
curve would shift down to the right from SRAS (W0) to SRAS (W1), to reestablish
the natural rate level of output at point C. However, the existence of
long-term nominal wage contracts prevents this and provides the monetary
authorities with an opportunity to expand the money supply which, even if
anticipated, shifts the AD curve to the right and re-establishes equilibrium at
point A. Providing the authorities are free to react to exogenous shocks at
every time period, while workers are not, there is scope for demand management
to stabilize the economy even if agents have rational expectations. In
effect, if the monetary authorities can react to nominal demand shocks more
quickly than the private sector can renegotiate nominal wages, there is scope
for discretionary intervention. The fixed nominal wage gives the monetary
authorities a handle on the real wage rate and hence employment and output.
The non-neutrality of money in the Fischer model is not due to an unanticipated
monetary surprise. Anticipated monetary policy has real effects because
it is based on information that only becomes available after the contract has
been made.
Wage contracts are an important feature in all major industrial market
economies. However, there are significant differences between countries with
respect to both contract duration and the timing of contract renegotiations.
For example, in Japan nominal wage contracts typically last for one year and
expire simultaneously. The synchronized renegotiation of contracts (the shunto
system) in Japan is consistent with greater macroeconomic stability than is
the case in the US economy, which has a system of non-synchronized overlapping
(staggered) contracts, many of which last for three years (see Gordon,
1982b; Hall and Taylor, 1997). In the UK contracts are overlapping but are
typically shorter than in the USA, usually lasting for one year. When contracts
are staggered, nominal wages will exhibit more inertia in the face of
shocks than would be the case if existing contracts were renegotiated in a
synchronized way so as to accommodate new information. Taylor (1980)
demonstrated that if workers are concerned with their nominal wage relative
to others, then staggered contracting will allow the impact of monetary policy
on real variables to persist well beyond the length of the contracting period.
Taylor (1992b) has shown that the responsiveness of wages to supply and
demand conditions is much greater in Japan than in the USA, Canada and
other major European countries, and this accounts for the more stable macroeconomic
performance in Japan during the 1970s and early 1980s.
An immediate question arises from the above discussion. Why are longterm
wage agreements formed if they increase macroeconomic instability?
According to Phelps (1985, 1990) there are private advantages to both firms
and workers from entering into long-term wage contracts:
1. Wage negotiations are costly in time for both workers and firms. Research
must be carried out with respect to the structure of wage relativities
both within and outside the negotiating organization. In addition, forecasts
are required with respect to the likely future paths of key variables
such as productivity, inflation, demand, profits and prices. The longer the
period of the contract, the less frequently are such transaction costs
incurred and in any case management will always tend to prefer a pre-set
schedule for dealing with the complex issues associated with pay negotiations.
2. There always exists the potential for such negotiations to break down,
with workers feeling that they may need to resort to strike action in order
to strengthen their bargaining position. Such disruption is costly to both
firms and workers.
3. It will not be an optimal strategy for a firm to ‘jump’ its wage rates to the
new ‘ultimate’ equilibrium following a negative demand stock because if
other firms do not do likewise the firm will have reduced its relative
wage, which would be likely to increase labour turnover, which is costly
to the firm.
Thus the responsiveness of wage rates during a recession does not follow the
new classical ‘precision drill process’; rather we observe a ‘ragged, disorderly
retreat’ as new information becomes available (Phelps, 1985, p. 564).
Another important question raised by this discussion relates to the absence
of indexing. Why are labour contracts not indexed to the rate of inflation?
Full cost of living agreements (COLAs) are simply too risky for firms (see
Gordon, 2003). The danger for firms is that not all shocks are nominal
demand shocks. If a firm agreed to index its wage rates to the rate of inflation,
then supply shocks, such as occurred in the 1970s, would drive up the price
level and with it a firm’s wage costs, so preventing the necessary fall in real
wages implied by the energy shock.
Finally, we should also note that the staggering of wage contracts does
have some microeconomic purpose even if it causes macroeconomic problems.
In a world where firms have imperfect knowledge of the current economic
situation, they can gain vital information by observing the prices and wages
set by other firms. According to Hall and Taylor (1997), staggered wage
setting provides useful information to both firms and workers about the
changing structure of wages and prices. In a decentralized system without
staggering, ‘tremendous variability’ would be introduced into the system.
Ball and Cecchetti (1988) show how imperfect information can make stag
gered price and wage setting socially optimal by helping firms set prices
closer to full information levels, leading to efficiency gains which outweigh
the costs of price level inertia. Thus staggered price adjustment can arise
from rational economic behaviour. In contrast, the case of wage setting in a
synchronized system would seem to require some degree of active participation
from the government.
7.5.2 Nominal price rigidity
Keynesian models based on nominal wage contracting soon came in for
considerable criticism (see Barro, 1977b). Critics pointed out that the existence
of such contracts is not explained from solid microeconomic principles.
A further problem relates to the countercyclical path of the real wage in
models with nominal wage contracts. In Fischer’s model, a monetary expansion
increases employment by lowering the real wage. Yet, as we have seen,
the stylized facts of the business cycle do not provide strong support for this
implication since real wages appear to be mildly procyclical (see Mankiw,
1990). Indeed, it was this issue that persuaded Mankiw (1991) that sticky
nominal wage models made little sense. A combination of price-taking firms,
neoclassical production technology and sticky nominal wages implies that
aggregate demand contractions will be associated with a rise in the real wage,
that is, real wages move countercyclically. As Mankiw notes, if this were the
case then recessions would be ‘quite popular’. While many people will be
laid off, most people who remain employed will enjoy a higher real wage! ‘If
high real wages accompanied low employment as the General Theory and my
Professors has taught me, then most households would welcome economic
downturns’. So ‘it was thinking about the real wage puzzle that originally got
me interested in thinking about imperfections in goods markets, and eventually,
about monopolistically competitive firms facing menu costs’ (Mankiw,
1991, pp. 129–30).
As a result of these and other criticisms, some economists sympathetic to
the Keynesian view that business cycles can be caused by fluctuations of
aggregate demand switched their attention to nominal rigidities in the goods
market, rather than continue with research into nominal wage inertia (Andersen,
1994). Indeed, the term ‘new Keynesian’ emerged in the mid-1980s as a
description of those new theories that attempted to provide more solid
microfoundations for the phenomenon of nominal price rigidity (see
Rotemberg, 1987). From this standpoint, the ‘fundamental new idea behind
new Keynesian models is that of imperfect competition’ (Ibid.). This is the
crucial innovation which differentiates new Keynesians from Keynes, orthodox
Keynesians, monetarists and new classicals.
If the process of changing prices were a costless exercise and if the failure to
adjust prices involved substantial changes in a firm’s profitability we would
certainly expect to observe a high degree of nominal price flexibility. A firm
operating under conditions of perfect competition is a price taker, and prices
change automatically to clear markets as demand and supply conditions change.
Since each firm can sell as much output as it likes at the going market price, a
perfectly competitive firm which attempted to charge a price above the marketclearing
level would have zero sales. There is also no profit incentive to reduce
price independently, given that the firm’s demand curve is perfectly elastic at
the prevailing market price. Thus in this world of perfect price flexibility it
makes no sense to talk of the individual firm having a pricing decision.
When firms operate in imperfectly competitive markets a firm’s profits will
vary differentially with changes in its own price because its sales will not fall
to zero if it marginally increases price. Price reductions by such a firm will
increase sales but also result in less revenue per unit sold. In such circumstances
any divergence of price from the optimum will only produce
‘second-order’ reductions of profits. Hence the presence of even small costs
to price adjustment can generate considerable aggregate nominal price rigidity.
This observation, due to Akerlof and Yellen (1985a), Mankiw (1985) and
Parkin (1986), is referred to by Rotemberg (1987) as the ‘PAYM insight’.
The PAYM insight makes a simple but powerful point. The private cost of
nominal rigidities to the individual firm is much smaller than the macroeconomic
consequences of such rigidities. A key ingredient of the PAYM insight
is the presence of frictions or barriers to price adjustment known as ‘menu
costs’. These menu costs include the physical costs of resetting prices, such
as the printing of new price lists and catalogues, as well as expensive management
time used up in the supervision and renegotiation of purchase and
sales contracts with suppliers and customers. To illustrate how small menu
costs can produce large macroeconomic fluctuations, we will review the
arguments made by Mankiw and by Akerlof and Yellen.
In imperfectly competitive markets a firm’s demand will depend on (i) its
relative price and (ii) aggregate demand. Suppose following a decline in
aggregate demand the demand curve facing an imperfectly competitive firm
shifts to the left. A shift of the demand curve to the left can significantly
reduce a firm’s profits. However, faced with this new demand curve, the firm
may gain little by changing its price. The firm would prefer that the demand
curve had not shifted but, given the new situation, it can only choose some
point on the new demand curve. This decline in demand is illustrated in
Figure 7.2 by the shift of demand from D0 to D1. Before the decline in
demand the profit-maximizing price and output are P0 and Q0, since marginal
revenue (MR0) is equal to marginal cost (MC0) at point X. For convenience
we assume that marginal cost does not vary with output over the range
shown. Following the decline in demand, the firm suffers a significant reduction
in its profits. Before the reduction in demand, profits are indicated in
Figure 7.2 Price adjustment under monopolistic competition
Figure 7.2 by the area SP0YX. If the firm does not initially reduce its price
following the decline in demand, profits fall to the area indicated by SP0JT.
Because this firm is a ‘price maker’ it must decide whether or not to reduce
price to the new profit-maximizing point indicated by W on the new demand
curve D1. The new profit-maximizing level of output is determined where
MR1 = MC0. With a level of output of Q1, the firm would make profits of SP1
WV. If there were no adjustment costs associated with changing price, a
profit-maximizing firm would reduce its price from P0 to P1. However, if a
firm faces non-trivial ‘menu costs’ of z, the firm may decide to leave price at
P0; that is, the firm moves from point Y to point J in Figure 7.2.
Figure 7.3 indicates the consequences of the firm’s decision. By reducing
price from P0 to P1 the firm would increase its profits by B – A. There is no
incentive for a profit-maximizing firm to reduce price if z > B – A. The loss to
society of producing an output of Q* rather than Q1 is indicated by B + C,
which represents the loss of total surplus. If following a reduction of demand
B + C > z > B – A, then the firm will not cut its price even though doing so
would be socially optimal. The flatter the MC schedule, the smaller are the
menu costs necessary to validate a firm’s decision to leave the price unchanged.
Readers should confirm for themselves that the incentive to lower
prices is therefore greater the more marginal cost falls when output declines
(see Gordon, 1990; D. Romer, 2001)
Figure 7.3 Menu costs v. price adjustment
In the Akerlof and Yellen (1985a, 1985b) model, inertial wage-price behaviour
by firms ‘may be near rational’. Firms that behave sub-optimally in
their price-setting behaviour may suffer losses but they are likely to be
second order (small). The idea of near rationality is illustrated in Figure 7.4.
As before, the profit-maximizing price following a decline in demand is
indicated by P1. The reduction in profits (π1 – π*) that results from failure to
reduce price from P0 to P1 is small (second order) even without taking into
account menu costs (that is, in Figure 7.3, B – A is small). Akerlof and Yellen
(1985a) also demonstrate that, when imperfect competition in the product
market is combined with efficiency wages in the labour market, aggregate
demand disturbances will lead to cyclical fluctuations (see Akerlof, 2002).
Although the firm may optimally choose to maintain price at P0, the impact
of their decision, if repeated by all firms, can have significant macroeconomic
effects. Blanchard and Kiyotaki (1987), in their interpretation of the PAYM
insight, show that the macroeconomic effects of nominal price rigidity differ
from the private costs because price rigidity generates an aggregate demand
externality. Society would be considerably better off if all firms cut their
prices, but the private incentives to do so are absent. As before, assume that a
firm’s demand curve has shifted left as a result of a decline in aggregate
demand. If firms did not face menu costs, then profit-maximizing behaviour
would dictate that all firms lowered their prices; that is, in terms of Figures
7.2 and 7.3, each firm would move from Y to W. Because all firms are
lowering their prices, each firm will find the cost of its inputs are falling,
including money wages. Hence each firm will find that its marginal cost
curve begins to shift down. This allows firms to reduce prices further. In
Figure 7.3, as MC0 shifts down, output will expand. Since all firms are
engaged in further price reductions, input prices will fall again, producing
another reduction of MC. Since this process of price deflation will increase
real money balances, thereby lowering interest rates, aggregate demand will
increase. This will shift the demand curves facing each firm to the right, so
that output will return to Q0.
If the presence of menu costs and/or near rational behaviour causes nominal
price rigidity, shocks to nominal aggregate demand will cause large
fluctuations in output and welfare. Since such fluctuations are inefficient, this
indicates that stabilization policy is desirable. Obviously if money wages are
rigid (because of contracts) the marginal cost curve will be sticky, thus
reinforcing the impact of menu costs in producing price rigidities.
We noted earlier that there are several private advantages to be gained by
both firms and workers from entering into long-term wage contracts. Many of
these advantages also apply to long-term agreements between firms with
respect to product prices. Pre-set prices not only reduce uncertainty but also
economize on the use of scarce resources. Gordon (1981) argues that ‘persua
sive heterogeneity’ in the types and quality of products available in a market
economy would create ‘overwhelming transaction costs’ if it were decided
that every price was to be decided in an auction. Auction markets are efficient
where buyers and sellers do not need to come into physical contact (as with
financial assets) or the product is homogeneous (as with wheat). The essential
feature of an auction market is that buyers and sellers need to be present
simultaneously. Because time and space are scarce resources it would not
make any sense for the vast majority of goods to be sold in this way. Instead
numerous items are made available at suitable locations where consumers
can choose to conduct transactions at their own convenience. The use of
‘price tags’ (goods available on fixed terms) is a rational response to the
problem of heterogeneity. Typically when prices are pre-set the procedure
used is a ‘mark-up pricing’ approach (see Okun, 1981).
As is evident from the above discussion, the theory of imperfect competition
forms one of the main building-blocks in new Keynesian economics.
Therefore, before moving on to consider real rigidities, it is interesting to
note one of the great puzzles in the history of economic thought. Why did
Keynes show so little interest in the imperfect competition revolution taking
place on his own doorstep in Cambridge in the early 1930s? Richard Kahn,
author of the famous 1931 multiplier article and colleague of Keynes, was
fully conversant with the theory of imperfect competition well before Joan
Robinson’s famous book was published on the subject in 1933. Given that
Keynes, Kahn and Robinson shared the same Cambridge academic environment
during the period when the General Theory was being written, it is
remarkable that Keynes adopted the classical/neoclassical assumption of a
perfectly competitive product market which Kahn (1929) had already argued
was unsound for short-period analysis (see Marris, 1991)! As Dixon (1997)
notes, ‘had Kahn and Keynes been able to work together, or Keynes and
Robinson, the General Theory might have been very different’. In contrast to
the orthodox Keynesian school, and inspired by the work of Michal Kalecki,
Post Keynesians have always stressed the importance of price-fixing firms in
their models (Arestis, 1997).
Both orthodox and new Keynesian approaches assume that prices adjust slowly
following a disturbance. But, unlike the Keynesian cross or IS–LM approaches,
which arbitrarily assume fixed nominal wages and prices, the new Keynesian
approach seeks to provide a microeconomic underpinning for the slow adjustment
of both wages and prices. In line with the choice-theoretical framework of
new classical analysis, the new Keynesian approach assumes that workers and
firms are rational utility and profit maximizers, respectively.
As we have seen, new classicists adopt the flexible price auction model and
apply this to the analysis of transactions conducted in all markets, including
the labour market. In contrast, new Keynesians argue that it is important to
utilize the Hicksian (1974) distinction between markets which are essentially
fix-price, predominantly the labour market and a large section of the goods
market, and markets which are flex-price, predominantly financial and commodity
markets. In fix-price markets price setting is the norm, with price and
wage inertia a reality. In order to generate monetary non-neutrality (real
effects) Keynesian models rely on the failure of nominal wages and prices to
adjust promptly to their new market-clearing levels following an aggregate
demand disturbance. Keynesians have traditionally concentrated their attention
on the labour market and nominal wage stickiness in order to explain the
tendency of market economies to depart from full employment equilibrium.
However, it is important to note that for any given path of nominal aggregate
demand it is price, not wage, stickiness which is necessary to generate fluctuations
in real output. Providing profits are sufficiently flexible, nominal
prices could adjust to exactly mimic changes in nominal aggregate demand,
leaving real output unaffected (see Gordon, 1990).
Nevertheless the first wave of new Keynesian reaction to the new classical
critique concentrated on nominal wage rigidity.
7.5.1 Nominal wage rigidity
In traditional Keynesian models the price level is prevented from falling to
restore equilibrium by the failure of money wages (costs) to adjust (see
Figure 2.6). In the new classical models developed by Lucas, Sargent, Wallace
and Barro during the 1970s, any anticipated monetary disturbance will cause
an immediate jump of nominal wages and prices to their new equilibrium
values, so preserving output and employment. In such a world, systematic
monetary policy is ineffective. Initially it was widely believed that this new
classical policy ineffective proposition was a direct implication of incorpoThe
rating the rational expectations hypothesis into macroeconomic models. Fischer
(1977) and Phelps and Taylor (1977) showed that nominal disturbances were
capable of producing real effects in models incorporating rational expectations,
providing the assumption of continuously clearing markets was dropped
(see also Buiter, 1980). Following these contributions it became clear to
everyone that the rational expectations hypothesis did not imply the end of
Keynesian economics. The crucial feature of new classical models was shown
to be the assumption of continuous market clearing, that is, perfect and
instantaneous wage and price flexibility. But, as Phelps (1985) reminds us, it
is often through the rejection of a theoretically interesting model that a
science progresses and ‘even if dead wrong, the new classical macroeconomics
is still important because it demands Keynesians to fortify their theoretical
structure or reconstruct it’.
The early Keynesian attempts to fortify their theoretical structure concentrated
on nominal wage rigidities and the models developed by Fischer (1977)
and Taylor (1980) introduced nominal inertia in the form of long-term wage
contracts. In developed economies wages are not determined in spot markets
but tend to be set for an agreed period in the form of an explicit (or implicit)
contract. The existence of these long-term contracts can generate sufficient
nominal wage rigidity for monetary policy to regain its effectiveness. It
should be noted, however, that neither Fischer nor Phelps and Taylor pretend
to have a rigorous microfoundation for their price- and wage-setting assumptions.
Instead they take it for granted that there is a ‘revealed preference’ for
long-term wage contracts reflecting the perceived disadvantages that accompany
too frequent adjustments to wages and prices (for an innovative attempt
to explain nominal wage inflexibility, see Laing, 1993).
Fischer’s analysis has the following main features and involves the construction
of a model similar to the Lucas–Sargent–Wallace policy ineffectiveness
models discussed in Chapter 5. The output supply equation is the standard
rational expectations Lucas ‘surprise’ function (7.1), where ˙Pt and ˙Pt
e are the
actual and expected rates of inflation respectively:
Y Y P P t N t t
e
t = + α( ˙ − ˙ ), α > 0 (7.1)
Fischer assumes that inflation expectations are formed rationally, P˙t E(P˙
e
= t |
Ωt–1), so we can write (7.1) as (7.2):
Yt YNt Pt E Pt t = + α[ ˙ − ( ˙ |Ω −1)] (7.2)
Fischer’s model abstracts from growth, so wage negotiators are assumed to
aim for constancy of the real wage by setting nominal wage increases equal
to expected inflation. This is given by (7.3):
W˙ E(P˙ | ) t = t Ωt−1 (7.3)
Substituting (7.3) into (7.2) yields equation (7.4), which shows that aggregate
supply is a decreasing function of the real wage (note this implies a
countercyclical real wage).
Yt YNt Pt Wt = + α[ ˙ − ˙ ], and α > 0 (7.4)
For the multi-period contract nominal wage increases are fixed at W˙t = W˙t* .
Fischer (1977) makes the ‘empirically reasonable’ assumption that economic
agents negotiate contracts in nominal terms for ‘periods longer than the time
it takes the monetary authority to react to changing economic circumstances’.
Because the monetary authorities can change the money supply (and hence
inflation) more frequently than overlapping labour contracts are renegotiated,
monetary policy can have real effects in the short run although it will remain
neutral in the long run.
The argument presented by Fischer can be understood with reference to
Figure 7.1. The economy is initially operating at point A. Suppose in the
current period an unexpected nominal demand shock occurs (such as a fall in
velocity) which shifts the aggregate demand curve from AD0 to AD1. If prices
Figure 7.1 Nominal wage contracts, rational expectations and monetary
policy
are flexible but nominal wages are temporarily rigid (and set = W0) as the
result of contracts negotiated in the previous period and which extend beyond
the current period, the economy will move to point B, with real output falling
from YN to Y1. With flexible wages and prices the short-run aggregate supply
curve would shift down to the right from SRAS (W0) to SRAS (W1), to reestablish
the natural rate level of output at point C. However, the existence of
long-term nominal wage contracts prevents this and provides the monetary
authorities with an opportunity to expand the money supply which, even if
anticipated, shifts the AD curve to the right and re-establishes equilibrium at
point A. Providing the authorities are free to react to exogenous shocks at
every time period, while workers are not, there is scope for demand management
to stabilize the economy even if agents have rational expectations. In
effect, if the monetary authorities can react to nominal demand shocks more
quickly than the private sector can renegotiate nominal wages, there is scope
for discretionary intervention. The fixed nominal wage gives the monetary
authorities a handle on the real wage rate and hence employment and output.
The non-neutrality of money in the Fischer model is not due to an unanticipated
monetary surprise. Anticipated monetary policy has real effects because
it is based on information that only becomes available after the contract has
been made.
Wage contracts are an important feature in all major industrial market
economies. However, there are significant differences between countries with
respect to both contract duration and the timing of contract renegotiations.
For example, in Japan nominal wage contracts typically last for one year and
expire simultaneously. The synchronized renegotiation of contracts (the shunto
system) in Japan is consistent with greater macroeconomic stability than is
the case in the US economy, which has a system of non-synchronized overlapping
(staggered) contracts, many of which last for three years (see Gordon,
1982b; Hall and Taylor, 1997). In the UK contracts are overlapping but are
typically shorter than in the USA, usually lasting for one year. When contracts
are staggered, nominal wages will exhibit more inertia in the face of
shocks than would be the case if existing contracts were renegotiated in a
synchronized way so as to accommodate new information. Taylor (1980)
demonstrated that if workers are concerned with their nominal wage relative
to others, then staggered contracting will allow the impact of monetary policy
on real variables to persist well beyond the length of the contracting period.
Taylor (1992b) has shown that the responsiveness of wages to supply and
demand conditions is much greater in Japan than in the USA, Canada and
other major European countries, and this accounts for the more stable macroeconomic
performance in Japan during the 1970s and early 1980s.
An immediate question arises from the above discussion. Why are longterm
wage agreements formed if they increase macroeconomic instability?
According to Phelps (1985, 1990) there are private advantages to both firms
and workers from entering into long-term wage contracts:
1. Wage negotiations are costly in time for both workers and firms. Research
must be carried out with respect to the structure of wage relativities
both within and outside the negotiating organization. In addition, forecasts
are required with respect to the likely future paths of key variables
such as productivity, inflation, demand, profits and prices. The longer the
period of the contract, the less frequently are such transaction costs
incurred and in any case management will always tend to prefer a pre-set
schedule for dealing with the complex issues associated with pay negotiations.
2. There always exists the potential for such negotiations to break down,
with workers feeling that they may need to resort to strike action in order
to strengthen their bargaining position. Such disruption is costly to both
firms and workers.
3. It will not be an optimal strategy for a firm to ‘jump’ its wage rates to the
new ‘ultimate’ equilibrium following a negative demand stock because if
other firms do not do likewise the firm will have reduced its relative
wage, which would be likely to increase labour turnover, which is costly
to the firm.
Thus the responsiveness of wage rates during a recession does not follow the
new classical ‘precision drill process’; rather we observe a ‘ragged, disorderly
retreat’ as new information becomes available (Phelps, 1985, p. 564).
Another important question raised by this discussion relates to the absence
of indexing. Why are labour contracts not indexed to the rate of inflation?
Full cost of living agreements (COLAs) are simply too risky for firms (see
Gordon, 2003). The danger for firms is that not all shocks are nominal
demand shocks. If a firm agreed to index its wage rates to the rate of inflation,
then supply shocks, such as occurred in the 1970s, would drive up the price
level and with it a firm’s wage costs, so preventing the necessary fall in real
wages implied by the energy shock.
Finally, we should also note that the staggering of wage contracts does
have some microeconomic purpose even if it causes macroeconomic problems.
In a world where firms have imperfect knowledge of the current economic
situation, they can gain vital information by observing the prices and wages
set by other firms. According to Hall and Taylor (1997), staggered wage
setting provides useful information to both firms and workers about the
changing structure of wages and prices. In a decentralized system without
staggering, ‘tremendous variability’ would be introduced into the system.
Ball and Cecchetti (1988) show how imperfect information can make stag
gered price and wage setting socially optimal by helping firms set prices
closer to full information levels, leading to efficiency gains which outweigh
the costs of price level inertia. Thus staggered price adjustment can arise
from rational economic behaviour. In contrast, the case of wage setting in a
synchronized system would seem to require some degree of active participation
from the government.
7.5.2 Nominal price rigidity
Keynesian models based on nominal wage contracting soon came in for
considerable criticism (see Barro, 1977b). Critics pointed out that the existence
of such contracts is not explained from solid microeconomic principles.
A further problem relates to the countercyclical path of the real wage in
models with nominal wage contracts. In Fischer’s model, a monetary expansion
increases employment by lowering the real wage. Yet, as we have seen,
the stylized facts of the business cycle do not provide strong support for this
implication since real wages appear to be mildly procyclical (see Mankiw,
1990). Indeed, it was this issue that persuaded Mankiw (1991) that sticky
nominal wage models made little sense. A combination of price-taking firms,
neoclassical production technology and sticky nominal wages implies that
aggregate demand contractions will be associated with a rise in the real wage,
that is, real wages move countercyclically. As Mankiw notes, if this were the
case then recessions would be ‘quite popular’. While many people will be
laid off, most people who remain employed will enjoy a higher real wage! ‘If
high real wages accompanied low employment as the General Theory and my
Professors has taught me, then most households would welcome economic
downturns’. So ‘it was thinking about the real wage puzzle that originally got
me interested in thinking about imperfections in goods markets, and eventually,
about monopolistically competitive firms facing menu costs’ (Mankiw,
1991, pp. 129–30).
As a result of these and other criticisms, some economists sympathetic to
the Keynesian view that business cycles can be caused by fluctuations of
aggregate demand switched their attention to nominal rigidities in the goods
market, rather than continue with research into nominal wage inertia (Andersen,
1994). Indeed, the term ‘new Keynesian’ emerged in the mid-1980s as a
description of those new theories that attempted to provide more solid
microfoundations for the phenomenon of nominal price rigidity (see
Rotemberg, 1987). From this standpoint, the ‘fundamental new idea behind
new Keynesian models is that of imperfect competition’ (Ibid.). This is the
crucial innovation which differentiates new Keynesians from Keynes, orthodox
Keynesians, monetarists and new classicals.
If the process of changing prices were a costless exercise and if the failure to
adjust prices involved substantial changes in a firm’s profitability we would
certainly expect to observe a high degree of nominal price flexibility. A firm
operating under conditions of perfect competition is a price taker, and prices
change automatically to clear markets as demand and supply conditions change.
Since each firm can sell as much output as it likes at the going market price, a
perfectly competitive firm which attempted to charge a price above the marketclearing
level would have zero sales. There is also no profit incentive to reduce
price independently, given that the firm’s demand curve is perfectly elastic at
the prevailing market price. Thus in this world of perfect price flexibility it
makes no sense to talk of the individual firm having a pricing decision.
When firms operate in imperfectly competitive markets a firm’s profits will
vary differentially with changes in its own price because its sales will not fall
to zero if it marginally increases price. Price reductions by such a firm will
increase sales but also result in less revenue per unit sold. In such circumstances
any divergence of price from the optimum will only produce
‘second-order’ reductions of profits. Hence the presence of even small costs
to price adjustment can generate considerable aggregate nominal price rigidity.
This observation, due to Akerlof and Yellen (1985a), Mankiw (1985) and
Parkin (1986), is referred to by Rotemberg (1987) as the ‘PAYM insight’.
The PAYM insight makes a simple but powerful point. The private cost of
nominal rigidities to the individual firm is much smaller than the macroeconomic
consequences of such rigidities. A key ingredient of the PAYM insight
is the presence of frictions or barriers to price adjustment known as ‘menu
costs’. These menu costs include the physical costs of resetting prices, such
as the printing of new price lists and catalogues, as well as expensive management
time used up in the supervision and renegotiation of purchase and
sales contracts with suppliers and customers. To illustrate how small menu
costs can produce large macroeconomic fluctuations, we will review the
arguments made by Mankiw and by Akerlof and Yellen.
In imperfectly competitive markets a firm’s demand will depend on (i) its
relative price and (ii) aggregate demand. Suppose following a decline in
aggregate demand the demand curve facing an imperfectly competitive firm
shifts to the left. A shift of the demand curve to the left can significantly
reduce a firm’s profits. However, faced with this new demand curve, the firm
may gain little by changing its price. The firm would prefer that the demand
curve had not shifted but, given the new situation, it can only choose some
point on the new demand curve. This decline in demand is illustrated in
Figure 7.2 by the shift of demand from D0 to D1. Before the decline in
demand the profit-maximizing price and output are P0 and Q0, since marginal
revenue (MR0) is equal to marginal cost (MC0) at point X. For convenience
we assume that marginal cost does not vary with output over the range
shown. Following the decline in demand, the firm suffers a significant reduction
in its profits. Before the reduction in demand, profits are indicated in
Figure 7.2 Price adjustment under monopolistic competition
Figure 7.2 by the area SP0YX. If the firm does not initially reduce its price
following the decline in demand, profits fall to the area indicated by SP0JT.
Because this firm is a ‘price maker’ it must decide whether or not to reduce
price to the new profit-maximizing point indicated by W on the new demand
curve D1. The new profit-maximizing level of output is determined where
MR1 = MC0. With a level of output of Q1, the firm would make profits of SP1
WV. If there were no adjustment costs associated with changing price, a
profit-maximizing firm would reduce its price from P0 to P1. However, if a
firm faces non-trivial ‘menu costs’ of z, the firm may decide to leave price at
P0; that is, the firm moves from point Y to point J in Figure 7.2.
Figure 7.3 indicates the consequences of the firm’s decision. By reducing
price from P0 to P1 the firm would increase its profits by B – A. There is no
incentive for a profit-maximizing firm to reduce price if z > B – A. The loss to
society of producing an output of Q* rather than Q1 is indicated by B + C,
which represents the loss of total surplus. If following a reduction of demand
B + C > z > B – A, then the firm will not cut its price even though doing so
would be socially optimal. The flatter the MC schedule, the smaller are the
menu costs necessary to validate a firm’s decision to leave the price unchanged.
Readers should confirm for themselves that the incentive to lower
prices is therefore greater the more marginal cost falls when output declines
(see Gordon, 1990; D. Romer, 2001)
Figure 7.3 Menu costs v. price adjustment
In the Akerlof and Yellen (1985a, 1985b) model, inertial wage-price behaviour
by firms ‘may be near rational’. Firms that behave sub-optimally in
their price-setting behaviour may suffer losses but they are likely to be
second order (small). The idea of near rationality is illustrated in Figure 7.4.
As before, the profit-maximizing price following a decline in demand is
indicated by P1. The reduction in profits (π1 – π*) that results from failure to
reduce price from P0 to P1 is small (second order) even without taking into
account menu costs (that is, in Figure 7.3, B – A is small). Akerlof and Yellen
(1985a) also demonstrate that, when imperfect competition in the product
market is combined with efficiency wages in the labour market, aggregate
demand disturbances will lead to cyclical fluctuations (see Akerlof, 2002).
Although the firm may optimally choose to maintain price at P0, the impact
of their decision, if repeated by all firms, can have significant macroeconomic
effects. Blanchard and Kiyotaki (1987), in their interpretation of the PAYM
insight, show that the macroeconomic effects of nominal price rigidity differ
from the private costs because price rigidity generates an aggregate demand
externality. Society would be considerably better off if all firms cut their
prices, but the private incentives to do so are absent. As before, assume that a
firm’s demand curve has shifted left as a result of a decline in aggregate
demand. If firms did not face menu costs, then profit-maximizing behaviour
would dictate that all firms lowered their prices; that is, in terms of Figures
7.2 and 7.3, each firm would move from Y to W. Because all firms are
lowering their prices, each firm will find the cost of its inputs are falling,
including money wages. Hence each firm will find that its marginal cost
curve begins to shift down. This allows firms to reduce prices further. In
Figure 7.3, as MC0 shifts down, output will expand. Since all firms are
engaged in further price reductions, input prices will fall again, producing
another reduction of MC. Since this process of price deflation will increase
real money balances, thereby lowering interest rates, aggregate demand will
increase. This will shift the demand curves facing each firm to the right, so
that output will return to Q0.
If the presence of menu costs and/or near rational behaviour causes nominal
price rigidity, shocks to nominal aggregate demand will cause large
fluctuations in output and welfare. Since such fluctuations are inefficient, this
indicates that stabilization policy is desirable. Obviously if money wages are
rigid (because of contracts) the marginal cost curve will be sticky, thus
reinforcing the impact of menu costs in producing price rigidities.
We noted earlier that there are several private advantages to be gained by
both firms and workers from entering into long-term wage contracts. Many of
these advantages also apply to long-term agreements between firms with
respect to product prices. Pre-set prices not only reduce uncertainty but also
economize on the use of scarce resources. Gordon (1981) argues that ‘persua
sive heterogeneity’ in the types and quality of products available in a market
economy would create ‘overwhelming transaction costs’ if it were decided
that every price was to be decided in an auction. Auction markets are efficient
where buyers and sellers do not need to come into physical contact (as with
financial assets) or the product is homogeneous (as with wheat). The essential
feature of an auction market is that buyers and sellers need to be present
simultaneously. Because time and space are scarce resources it would not
make any sense for the vast majority of goods to be sold in this way. Instead
numerous items are made available at suitable locations where consumers
can choose to conduct transactions at their own convenience. The use of
‘price tags’ (goods available on fixed terms) is a rational response to the
problem of heterogeneity. Typically when prices are pre-set the procedure
used is a ‘mark-up pricing’ approach (see Okun, 1981).
As is evident from the above discussion, the theory of imperfect competition
forms one of the main building-blocks in new Keynesian economics.
Therefore, before moving on to consider real rigidities, it is interesting to
note one of the great puzzles in the history of economic thought. Why did
Keynes show so little interest in the imperfect competition revolution taking
place on his own doorstep in Cambridge in the early 1930s? Richard Kahn,
author of the famous 1931 multiplier article and colleague of Keynes, was
fully conversant with the theory of imperfect competition well before Joan
Robinson’s famous book was published on the subject in 1933. Given that
Keynes, Kahn and Robinson shared the same Cambridge academic environment
during the period when the General Theory was being written, it is
remarkable that Keynes adopted the classical/neoclassical assumption of a
perfectly competitive product market which Kahn (1929) had already argued
was unsound for short-period analysis (see Marris, 1991)! As Dixon (1997)
notes, ‘had Kahn and Keynes been able to work together, or Keynes and
Robinson, the General Theory might have been very different’. In contrast to
the orthodox Keynesian school, and inspired by the work of Michal Kalecki,
Post Keynesians have always stressed the importance of price-fixing firms in
their models (Arestis, 1997).
No comments:
Post a Comment