Criticisms of Real Business Cycle Theory
In this section we will review some of the more important criticisms of real
business cycle theory. For critical surveys of the literature, the reader is
referred to Summers (1986), Hoover (1988), Sheffrin (1989), Mankiw (1989),
McCallum (1989), Phelps (1990), Eichenbaum (1991), Stadler (1994), and
Hartley et al. (1997, 1998).
The conventional neoclassical analysis of labour supply highlights two
opposing effects of an increase in the real wage. A higher real wage induces
an increase in labour supply through the substitution effect, but at the same
time a higher real wage also has an income effect that induces a worker to
consume more leisure. In real business cycle models the substitution effect
must be very powerful compared to the income effect if these models are to
plausibly explain large variations of employment induced by technology
shocks. But, as we have already noted, the available micro evidence relating
to the intertemporal elasticity of substitution in labour supply indicates a
weak response to transitory wage changes. If the wage elasticity of labour
supply is low, then technological shocks that shift the labour demand curve
(see Figure 6.3) will produce large variability of real wages and lower variability
of employment. However, the variations in employment observed over
the business cycle seem to be too large to be accounted for by intertemporal
substitution. In addition, Mankiw (1989) has argued that the real interest rate
is not a significant consideration in labour supply decisions. How, for example,
can agents be expected to accurately predict future interest rates and real
wages in order to engage in intertemporal substitution?
A second major criticism of real business cycle theory relates to the reliance
of these models on mainly unobservable technology shocks. Many
economists doubt whether the technology shocks required in order to generate
business cycle phenomena are either large enough or frequent enough. In
these models large movements in output require significant aggregate disturbances
to technology. Muellbauer (1997) argues that the kind of technological
volatility implied by REBCT is ‘quite implausible’ for three reasons, namely:
(i) technological diffusion tends to be slow; (ii) aggregation of diffusion
processes tends to produce a smooth outcome in aggregate; and (iii) the
technical regress required to produce recessions cannot be given plausible
microfoundations. In relation to this issue, Summers (1986) rejects Prescott’s
use of variations in the Solow residual as evidence of significant shocks to
technology. Large variations in the Solow residual can be explained as the
outcome of ‘off the production function behaviour’ in the form of labour
hoarding. Whereas real business cycle theorists interpret procyclical labour
productivity as evidence of shifts in the production function, the traditional
Keynesian explanation attributes this stylized fact to the quasi-fixity of the
labour input. The reason why productivity falls in recessions is that firms
retain more workers than they need, owing to short-run adjustment costs. In
such circumstances it will pay firms to smooth the labour input over the
cycle, which implies the hoarding of labour in a cyclical downturn. This
explains why the percentage reduction in output typically exceeds the percentage
reduction in the labour input during a recession. As the economy
recovers, firms utilize their labour more intensively, so that output increases
by a larger percentage than the labour input.
In general many economists explain the procyclical movement of the Solow
residual by highlighting the underutilization of both capital and labour during
periods of recession. Following Abel and Bernanke (2001), we can illustrate
this idea by rewriting the production function given by (6.13) and (6.14) as
(6.18):
Y = AF(μKK,μLL) = A(μKK)δ (μLL)1−δ (6.18)
where μK represents the underutilization rate of the capital input, and μL
represents the underutilization rate of labour input. Substituting (6.18) for Y
in (6.15) we obtain a new expression (6.19) for the Solow residual that
recognizes that the capital and labour inputs may be underutilized.
Solow residual = A(μKK)δ (μLL)1−δ /KδL1−δ = Aμ K δ μ L 1−δ (6.19)
Equation (6.19) shows that the Solow residual can vary even if technology
remains constant. If the utilization rates of capital and labour inputs are
procyclical, as the empirical evidence suggests is the case, then we will
observe a procyclical Solow residual that reflects this influence (for discussions
of this issue see Fay and Medoff, 1985; Rotemberg and Summers,
1990; Bernanke and Parkinson, 1991; Burnside et al., 1995; Braun and Evans,
1998; Millard et al., 1997).
A third line of criticism relates to the idea of recessions being periods of
technological regress. As Mankiw (1989, p. 85) notes, ‘recessions are important
events; they receive widespread attention from the policy-makers and the
media. There is, however, no discussion of declines in the available technology.
If society suffered some important adverse technological shock we
would be aware of it.’ In response to this line of criticism, Hansen and
Prescott (1993) have widened the interpretation of technological shocks so
that any ‘changes in the production functions, or, more generally, the production
possibility sets of the profit centres’ can be regarded as a potential source
of disturbance. In their analysis of the 1990–91 recession in the USA, they
suggest that changes to the legal and institutional framework can alter the
incentives to adopt certain technologies; for example, a barrage of government
regulations could act as a negative technology shock. However, as
Muellbauer (1997) points out, the severe recession in the UK in the early
1990s is easily explained as the consequence of a ‘massive’ rise in interest
rates in 1988–9, an associated collapse of property prices, and UK membership,
at an overvalued exchange rate, of the ERM after October 1990. Few of
these influences play a role in REBCT.
An important fourth criticism relates to the issue of unemployment. In real
business cycle models unemployment is either entirely absent or is voluntary.
Critics find this argument unconvincing and point to the experience of the
Great Depression, where ‘it defies credulity to account for movements on this
scale by pointing to intertemporal substitution and productivity shocks’ (Summers,
1986). Carlin and Soskice (1990) argue that a large proportion of the
European unemployment throughout the 1980s was involuntary and this represents
an important stylized fact which cannot be explained within a new
classical framework. Tobin (1980b) also questioned the general approach of
new classical economists to treat all unemployment as voluntary. The critics
point out that the pattern of labour market flows is inconsistent with equilibrium
theorizing. If we could explain unemployment as the result of voluntary
choice of economic agents, then we would not observe the well-established
procyclical movement of vacancy rates and voluntary quits. Recessions are
not periods where we observe an increase in rate of voluntary quits! In
Blinder’s view, the challenge posed by high unemployment during the 1980s
was not met by either policy makers or economists. In a comment obviously
directed at real business cycle theorists, Blinder (1988b) notes that ‘we will
not contribute much toward alleviating unemployment while we fiddle around
with theories of Pareto optimal recessions – an avocation that might be called
Nero-Classical Economics’. Although the intersectoral shifts model associated
with Lilien (1982) introduces unemployment into a model where
technology shocks motivate the need to reallocate resources across sectors,
the critics regard the neglect of unemployment in real business cycle theory
as a major weakness (see Hoover, 1988).
A fifth objection to real business cycle theory relates to the neutrality of
money and the irrelevance of monetary policy for real outcomes. It is a matter
of some irony that these models emerged in the early 1980s when in both the
USA and the UK the monetary disinflations initiated by Volcker and Thatcher
were followed by deep recessions in both countries. The 1990–92 economic
downturn in the UK also appears to have been the direct consequence of
another dose of monetary disinflation. In response to this line of criticism,
real business cycle theorists point out that the recessions experienced in the
early 1980s were preceded by a second major oil shock in 1979. However,
the majority of economists remain unconvinced that money is neutral in the
short run (see Romer and Romer, 1989, 1994a, 1994b; Blanchard, 1990a;
Ball, 1994; and Chapter 7).
A sixth line of criticism relates to the important finding by Nelson and
Plosser that it is hard to reject the view that real GNP is as persistent as a
random walk with drift. This finding appeared to lend support to the idea that
fluctuations are caused by supply-side shocks. The work of Nelson and
Plosser (1982) showed that aggregate output does not appear to be trendreverting.
If fluctuations were trend-reverting, then a temporary deviation of
output from its natural rate would not change a forecaster’s estimate of output
in ten years’ time. Campbell and Mankiw (1987, 1989), Stock and Watson
(1988) and Durlauf (1989) have confirmed the findings of Nelson and Plosser.
As a result, the persistence of shocks is now regarded as a ‘stylised fact’ (see
Durlauf, 1989, p. 71). However, Campbell, Mankiw and Durlauf do not accept
that the discovery of a near unit root in the GNP series is clear evidence
of real shocks, or that explanations of fluctuations based on demand disturbances
should be abandoned. Aggregate demand could have permanent effects
if technological innovation is affected by the business cycle or if hysteresis
effects are important (see Chapter 7). Durlauf has shown how, in the presence
of coordination failures, substantial persistence in real activity can result
from aggregate demand shocks. This implies that demand-side policies can
have long-lasting effects on output. Stadler (1990) has also shown how the
introduction of endogenous technological change fundamentally alters the
properties of both real and monetary theories of the business cycle. REBCT
does not provide any deep microeconomic foundations to explain technologi
cal change and innovative activity. But the plausible dependence of technological
progress on economic factors such as demand conditions, research
and development expenditures and ‘learning by doing’ effects (Arrow, 1962)
implies that changes on the supply side of the economy are not independent
of changes on the demand side. Hence an unanticipated increase in nominal
aggregate demand can induce technology changes on the supply side which
permanently increase output. In such a model the natural rate of unemployment
will depend on the history of aggregate demand as well as supply-side
factors. A purely monetary model of the business cycle where technology is
endogenous can also account for the Nelson and Plosser finding that output
appears to follow a random walk.
A seventh criticism relates to the pervasive use of the representative agent
construct in real business cycle theory. Real business cycle theorists sidestep
the aggregation problems inherent in macroeconomic analysis by using a
representative agent whose choices are assumed to coincide with the aggregate
choices of millions of heterogeneous individuals. Such models therefore
avoid the problems associated with asymmetric information, exchange and
coordination. To many economists the most important questions in macroeconomic
theory relate to problems associated with coordination and
heterogeneity. If the coordination question and the associated possibility of
exchange failures lie at the heart of economic fluctuations, then to by-pass the
problem by assuming that an economy is populated only by Robinson Crusoe
is an unacceptable research strategy for many economists (see Summers,
1986; Kirman, 1992; Leijonhufvud, 1992, 1998a; Akerlof, 2002; Snowdon,
2004a).
A final important criticism relates to the lack of robust empirical testing
(see Fair, 1992; Laidler, 1992a; Hartley et al., 1998). As far as the stylized
facts are concerned, both new Keynesian and real business cycle theories can
account for a broad pattern of time series co-movements (see Greenwald and
Stiglitz, 1988). In an assessment of the empirical plausibility of real business
cycle theory, Eichenbaum (1991) finds the evidence put forward by its proponents
as ‘too fragile to be believable’.
In this section we will review some of the more important criticisms of real
business cycle theory. For critical surveys of the literature, the reader is
referred to Summers (1986), Hoover (1988), Sheffrin (1989), Mankiw (1989),
McCallum (1989), Phelps (1990), Eichenbaum (1991), Stadler (1994), and
Hartley et al. (1997, 1998).
The conventional neoclassical analysis of labour supply highlights two
opposing effects of an increase in the real wage. A higher real wage induces
an increase in labour supply through the substitution effect, but at the same
time a higher real wage also has an income effect that induces a worker to
consume more leisure. In real business cycle models the substitution effect
must be very powerful compared to the income effect if these models are to
plausibly explain large variations of employment induced by technology
shocks. But, as we have already noted, the available micro evidence relating
to the intertemporal elasticity of substitution in labour supply indicates a
weak response to transitory wage changes. If the wage elasticity of labour
supply is low, then technological shocks that shift the labour demand curve
(see Figure 6.3) will produce large variability of real wages and lower variability
of employment. However, the variations in employment observed over
the business cycle seem to be too large to be accounted for by intertemporal
substitution. In addition, Mankiw (1989) has argued that the real interest rate
is not a significant consideration in labour supply decisions. How, for example,
can agents be expected to accurately predict future interest rates and real
wages in order to engage in intertemporal substitution?
A second major criticism of real business cycle theory relates to the reliance
of these models on mainly unobservable technology shocks. Many
economists doubt whether the technology shocks required in order to generate
business cycle phenomena are either large enough or frequent enough. In
these models large movements in output require significant aggregate disturbances
to technology. Muellbauer (1997) argues that the kind of technological
volatility implied by REBCT is ‘quite implausible’ for three reasons, namely:
(i) technological diffusion tends to be slow; (ii) aggregation of diffusion
processes tends to produce a smooth outcome in aggregate; and (iii) the
technical regress required to produce recessions cannot be given plausible
microfoundations. In relation to this issue, Summers (1986) rejects Prescott’s
use of variations in the Solow residual as evidence of significant shocks to
technology. Large variations in the Solow residual can be explained as the
outcome of ‘off the production function behaviour’ in the form of labour
hoarding. Whereas real business cycle theorists interpret procyclical labour
productivity as evidence of shifts in the production function, the traditional
Keynesian explanation attributes this stylized fact to the quasi-fixity of the
labour input. The reason why productivity falls in recessions is that firms
retain more workers than they need, owing to short-run adjustment costs. In
such circumstances it will pay firms to smooth the labour input over the
cycle, which implies the hoarding of labour in a cyclical downturn. This
explains why the percentage reduction in output typically exceeds the percentage
reduction in the labour input during a recession. As the economy
recovers, firms utilize their labour more intensively, so that output increases
by a larger percentage than the labour input.
In general many economists explain the procyclical movement of the Solow
residual by highlighting the underutilization of both capital and labour during
periods of recession. Following Abel and Bernanke (2001), we can illustrate
this idea by rewriting the production function given by (6.13) and (6.14) as
(6.18):
Y = AF(μKK,μLL) = A(μKK)δ (μLL)1−δ (6.18)
where μK represents the underutilization rate of the capital input, and μL
represents the underutilization rate of labour input. Substituting (6.18) for Y
in (6.15) we obtain a new expression (6.19) for the Solow residual that
recognizes that the capital and labour inputs may be underutilized.
Solow residual = A(μKK)δ (μLL)1−δ /KδL1−δ = Aμ K δ μ L 1−δ (6.19)
Equation (6.19) shows that the Solow residual can vary even if technology
remains constant. If the utilization rates of capital and labour inputs are
procyclical, as the empirical evidence suggests is the case, then we will
observe a procyclical Solow residual that reflects this influence (for discussions
of this issue see Fay and Medoff, 1985; Rotemberg and Summers,
1990; Bernanke and Parkinson, 1991; Burnside et al., 1995; Braun and Evans,
1998; Millard et al., 1997).
A third line of criticism relates to the idea of recessions being periods of
technological regress. As Mankiw (1989, p. 85) notes, ‘recessions are important
events; they receive widespread attention from the policy-makers and the
media. There is, however, no discussion of declines in the available technology.
If society suffered some important adverse technological shock we
would be aware of it.’ In response to this line of criticism, Hansen and
Prescott (1993) have widened the interpretation of technological shocks so
that any ‘changes in the production functions, or, more generally, the production
possibility sets of the profit centres’ can be regarded as a potential source
of disturbance. In their analysis of the 1990–91 recession in the USA, they
suggest that changes to the legal and institutional framework can alter the
incentives to adopt certain technologies; for example, a barrage of government
regulations could act as a negative technology shock. However, as
Muellbauer (1997) points out, the severe recession in the UK in the early
1990s is easily explained as the consequence of a ‘massive’ rise in interest
rates in 1988–9, an associated collapse of property prices, and UK membership,
at an overvalued exchange rate, of the ERM after October 1990. Few of
these influences play a role in REBCT.
An important fourth criticism relates to the issue of unemployment. In real
business cycle models unemployment is either entirely absent or is voluntary.
Critics find this argument unconvincing and point to the experience of the
Great Depression, where ‘it defies credulity to account for movements on this
scale by pointing to intertemporal substitution and productivity shocks’ (Summers,
1986). Carlin and Soskice (1990) argue that a large proportion of the
European unemployment throughout the 1980s was involuntary and this represents
an important stylized fact which cannot be explained within a new
classical framework. Tobin (1980b) also questioned the general approach of
new classical economists to treat all unemployment as voluntary. The critics
point out that the pattern of labour market flows is inconsistent with equilibrium
theorizing. If we could explain unemployment as the result of voluntary
choice of economic agents, then we would not observe the well-established
procyclical movement of vacancy rates and voluntary quits. Recessions are
not periods where we observe an increase in rate of voluntary quits! In
Blinder’s view, the challenge posed by high unemployment during the 1980s
was not met by either policy makers or economists. In a comment obviously
directed at real business cycle theorists, Blinder (1988b) notes that ‘we will
not contribute much toward alleviating unemployment while we fiddle around
with theories of Pareto optimal recessions – an avocation that might be called
Nero-Classical Economics’. Although the intersectoral shifts model associated
with Lilien (1982) introduces unemployment into a model where
technology shocks motivate the need to reallocate resources across sectors,
the critics regard the neglect of unemployment in real business cycle theory
as a major weakness (see Hoover, 1988).
A fifth objection to real business cycle theory relates to the neutrality of
money and the irrelevance of monetary policy for real outcomes. It is a matter
of some irony that these models emerged in the early 1980s when in both the
USA and the UK the monetary disinflations initiated by Volcker and Thatcher
were followed by deep recessions in both countries. The 1990–92 economic
downturn in the UK also appears to have been the direct consequence of
another dose of monetary disinflation. In response to this line of criticism,
real business cycle theorists point out that the recessions experienced in the
early 1980s were preceded by a second major oil shock in 1979. However,
the majority of economists remain unconvinced that money is neutral in the
short run (see Romer and Romer, 1989, 1994a, 1994b; Blanchard, 1990a;
Ball, 1994; and Chapter 7).
A sixth line of criticism relates to the important finding by Nelson and
Plosser that it is hard to reject the view that real GNP is as persistent as a
random walk with drift. This finding appeared to lend support to the idea that
fluctuations are caused by supply-side shocks. The work of Nelson and
Plosser (1982) showed that aggregate output does not appear to be trendreverting.
If fluctuations were trend-reverting, then a temporary deviation of
output from its natural rate would not change a forecaster’s estimate of output
in ten years’ time. Campbell and Mankiw (1987, 1989), Stock and Watson
(1988) and Durlauf (1989) have confirmed the findings of Nelson and Plosser.
As a result, the persistence of shocks is now regarded as a ‘stylised fact’ (see
Durlauf, 1989, p. 71). However, Campbell, Mankiw and Durlauf do not accept
that the discovery of a near unit root in the GNP series is clear evidence
of real shocks, or that explanations of fluctuations based on demand disturbances
should be abandoned. Aggregate demand could have permanent effects
if technological innovation is affected by the business cycle or if hysteresis
effects are important (see Chapter 7). Durlauf has shown how, in the presence
of coordination failures, substantial persistence in real activity can result
from aggregate demand shocks. This implies that demand-side policies can
have long-lasting effects on output. Stadler (1990) has also shown how the
introduction of endogenous technological change fundamentally alters the
properties of both real and monetary theories of the business cycle. REBCT
does not provide any deep microeconomic foundations to explain technologi
cal change and innovative activity. But the plausible dependence of technological
progress on economic factors such as demand conditions, research
and development expenditures and ‘learning by doing’ effects (Arrow, 1962)
implies that changes on the supply side of the economy are not independent
of changes on the demand side. Hence an unanticipated increase in nominal
aggregate demand can induce technology changes on the supply side which
permanently increase output. In such a model the natural rate of unemployment
will depend on the history of aggregate demand as well as supply-side
factors. A purely monetary model of the business cycle where technology is
endogenous can also account for the Nelson and Plosser finding that output
appears to follow a random walk.
A seventh criticism relates to the pervasive use of the representative agent
construct in real business cycle theory. Real business cycle theorists sidestep
the aggregation problems inherent in macroeconomic analysis by using a
representative agent whose choices are assumed to coincide with the aggregate
choices of millions of heterogeneous individuals. Such models therefore
avoid the problems associated with asymmetric information, exchange and
coordination. To many economists the most important questions in macroeconomic
theory relate to problems associated with coordination and
heterogeneity. If the coordination question and the associated possibility of
exchange failures lie at the heart of economic fluctuations, then to by-pass the
problem by assuming that an economy is populated only by Robinson Crusoe
is an unacceptable research strategy for many economists (see Summers,
1986; Kirman, 1992; Leijonhufvud, 1992, 1998a; Akerlof, 2002; Snowdon,
2004a).
A final important criticism relates to the lack of robust empirical testing
(see Fair, 1992; Laidler, 1992a; Hartley et al., 1998). As far as the stylized
facts are concerned, both new Keynesian and real business cycle theories can
account for a broad pattern of time series co-movements (see Greenwald and
Stiglitz, 1988). In an assessment of the empirical plausibility of real business
cycle theory, Eichenbaum (1991) finds the evidence put forward by its proponents
as ‘too fragile to be believable’.
No comments:
Post a Comment