Transcript Some instability puzzles in Kaleckian models of growth and

Some instability puzzles in
Kaleckian models of growth
and distribution
Eckhard Hein, Marc Lavoie and
Till van Treeck
The Kaleckian model
• Three essential equations
– A saving function
– A pricing function (income distribution)
– An investment function
• The actual rate of capacity utilization is
endogenous, determined by demand, even in
the long run, and so in general will not be equal
to the normal rate of capacity utilization
• Consequences: the paradox of thrift, and under
some specifications, the paradox of costs
Justification
• The Kaleckian model of growth has become a
workhorse of heterodox economics since the
early 1980s, proving to be highly flexible.
• Some authors claim however that one may be
Kaleckian or ‘Keynesian in the short run’ but
needs to be ‘Classical in the long run’ (Duménil
and Lévy 1999, Shaikh 2007).
• Others argue that ‘the current dominance of the
Kaleckian model (…) is unfortunate’ for postKeynesian and Structuralist macroeconomics
(Skott 2008).
• The purpose of this paper is to show that these
opinions are premature.
Consequences of the critique
• Somehow the actual rate of capacity
utilization must be brought back to its
normal rate
• Somehow Harrod’s warranted rate of
growth must constrain the economy, and
the classical equation must hold:
• gw = s/v =sprn
• The paradoxes of thrift and of costs don’t
hold anymore
Outline
• We distinguish Keynesian and Harrodian
instability
• We look at the various mechanisms that
have been suggested to bring back the
actual rate of utilization to its normal rate
• We question the need for such a
mechanism
• We offer alternative mechanisms that
retain Kaleckian features
Instability
• There have been two somewhat related kinds of
criticisms:
– Keynesian stability (the slope of the saving function is
steeper than that of the investment function), as
assumed in Kaleckian models, is doubtful.
– Harrodian instability (the investment function shifts
out if u > un)), as assumed away by Kaleckians, is
likely.
– Therefore, there must exist some other mechanism
that brings back the economy towards its normal rate
of capacity utilization (shifting back in the investment
function, or shifting the saving function)
Keynesian instability
Short period
u   (u  u ),
g
e
gi
K
e
gs
gi(ue)
g*
u
u* ue
uK
 0
   (u *  un ),
Harrodian instability
g
g3
γ3
g2
γ2
γ0=g0
 0
gs
C
gi
A
B
u
un
u1 u2
u3
Consequences
• Whether there is Keynesian or Harrodian
instability, the consequences are the
same: a boost in effective demand leads
to an ever-rising rate of accumulation and
rate of utilization.
• In practice, the two kinds of instability may
be difficult to disentangle.
The “classical” mechanisms designed to
bring back utilization to its normal rate
• Mechanisms acting on the saving function
– The Cambridge price mechanism, acting on
the profit margin
– The retention rate solution, acting on the
overall propensity to save
The Cambridge price mechanism
rn   (u *  un ),
,
 0
• (Robinson 1956, Kaldor 1956; Harcourt Kenyon 1977,
Eichner 1976, Wood 1975, Marglin 1984, Skott 1989)
• The profit margin rises as long as u > un, thus rotating
the saving function.
• The paradox of thrift is only retained if accumulation
depends on the profit rate
• The mechanism has been described either as an ultrashort run mechanism, or as a long-run mechanism.
• The profit margin, hence real wages fall, when
employment rates and growth are high. Doubtful?
The retention rate solution
s f   (u *  un ),
 0
• This is the Shaikh I solution (2007)
• “This classical synthesis allows us to preserve central
Keynesian arguments such as the dependence of
savings on investment and the regulation of investment
on expected profitability, without having to claim that
actual capacity utilization will persistently differ from the
rate desired by firms” (Shaikh)
• The retention ratio of firms rises as long as u > un, thus
rotating the saving function
• However, the paradoxes of saving and of costs
disappear, because: g** = γrrn
• Doubtful?
Shaikh I solution
g
gs
gi
g1
g2
g0
u
un
u1
The “classical” mechanisms designed to
bring back utilization to its normal rate
• Mechanisms acting on the investment
function
– Monetary authorities get scared of inflation
and raise real interest rates
– Capitalists get scared of full employment and
reduce the rate of growth of output
– Capitalists have perfect foresight and revise
their sales expectations
Monetary authorities get scared of inflation, and
raise real interest rates, thus lowering investment
    u  un ,
•
•
•
•
 0
The Duménil and Lévy (1999) mechanism
Similar to the New consensus mechanism
Also has resemblance with Robinson’s inflation frontier
Both the paradoxes of thrift and of costs get wiped out
The Duménil and Lévy mechanism: low saving
rates lead to high growth in the short run, but lower
growth in the long run
g
gs
gi
g1
g0
g2
u
us
u1
Drawbacks of the Duménil and
Lévy mechanism
• Higher rates of utilization may not mean higher or
accelerating inflation rates
• Higher rates of interest may not succeed in slowing
down demand (it may increase consumer demand
instead)
• When demand needs to be pumped up, it may be
impossible to lower real interest rates sufficiently (zero
lower-bound problem)
• Raising the interest rate is likely to lead to a lower
“normal” rate of utilization (a different NAIRU), as firms
raise their profit margins, and there is no guarantee that
the actual rate will converge to this evolving normal rate
(Hein 2006, 2008).
Capitalists get scared of full employment
and reduce the rate of growth of output
    u  un ,
 0
• This is the Skott mechanism (1989, 2007, 2008).
• In the ‘mature economy’, the rate of output growth is a
positive function of the profit share and a negative
function of the employment rate.
• If the employment rate rises above its steady state
value, capitalists reduce output growth, sales growth
declines, the actual rate of utilization falls, and the
constant in the investment function starts to shrink.
• The cause: firms have increasing problems to recruit
additional workers, workers and labour unions are
strengthened vis-à-vis management, workers’ militancy
increases, monitoring and surveillance costs rise, and
hence the overall business climate deteriorates.
Drawbacks of the Skott mechanism
• It is not clear why output growth should be a positive
function of the profit share: his excludes Kaleckian
effects by assumption.
• It is not clear why high employment rates, accompanied
by more powerful workers and labour unions, should
induce capitalists to reduce output growth in the first
place.
• One would rather think that high employment rates
generate rising nominal wage growth. This should cause
either rising inflation or a falling profit share, or both.
• But the latter would intensify Harrodian instability!
• What about labour supply growth being driven by labour
demand growth?
Capitalists have perfect foresight and revise their
sales expectations
g i  g y   u u  un ,
gy , u  0
• The Shaikh II (2007) mechanism
• Based on a special ‘Hicksian’ stock-flow
investment adjustment function.
• Investment depends on the rate of utilization and
the growth rate of sales that will be realized in
the current period
• If so, it can be shown that the actual rate of
capacity utilization necessarily converges to its
normal rate.
û
Figure 12
+
0
−
u
gs
g
gkal
gi = gy +
γu(u−un)
γ = gk0 = gy0
gk2
gy2
gk3 = gy3
un u2 u1 ukal
u
Drawbacks of the Shaikh II mechanism
• Firms must know the growth rate of their sales.
• But this rate depends on the investment
expenditures of all other firms.
• Thus, each firm needs to know what all other
firms simultaneously decide.
• The informational requirements are huge.
• The behaviour of managers is unlikely: following
a period of rising rates of utilization, firms need
to believe that sales will grow more slowly.
• If firms act in some adaptive way, Harrodian
instability reappears.
Questioning the necessity of any
adjustment of u towards un
• Provisional equilibrium; everything moves
anyway (Chick and Caserta 1997)
• Other stock-flow norms in growth models are not
realized, even when agents try to achieve them
– wealth/income targets (Godley)
• There is a large range of acceptable “desired” or
“normal” rates of capacity utilization (Dutt 1990).
• A firm may operate each running plant at optimal
capacity (cost-minimizing), while being unable to
to run all plants (idle capacity) (Caserta 1990).
Acceptable range
• ‘The stock adjustment principle, with its
particular desired level of stocks, is itself a
simplification. It would be more realistic to
suppose that there is a range or interval,
within which the level of stock is
“comfortable”, so that no special measures
seem called for to change it. Only if the
actual level goes outside that range will
there be a reaction.’ (Hicks 1974, p. 19)
Even Sraffians accept that firms
usually have idle capacity
• ‘It is virtually impossible for the investmentsaving mechanism … to result in an optimal
degree of capacity utilization…. It is, rather,
expected, that the economy will generally exhibit
smaller or larger margins of unutilized capacity
over and above the difference between full and
optimal capacity’. (Kurz 1994)
• ‘One must keep in mind that although each
entrepreneur might know the optimal degree of
capacity utilization, this is not enough to insure
that each of them will be able to realize this
optimal rate’. (Kurz 1993)
Still, we do not wish to sweep the
the problem under the carpet
• There are mechanisms that can bring together the actual
and the normal rates of capacity utilization, by making
the normal rate endogenous to the values taken by the
actual rate (Lavoie 1992, 1996, 2003; Cassetti 2006,
Commendatore 2006).
• So normal rates adjust to the actual rates.
• With some specifications, these hysteresis models
safeguard both the paradoxes of thrift and of costs.
• A critique of these has been: why would firms modify the
normal rate of utilization just because it has not been
achieved recently (Skott 2008)?
A possible answer: because firms
have multiple targets !
• This is the Dallery and van Treeck (2008) model, partly
based on Lavoie (2003).
• Firms may set themselves “target” or “normal” rates of
utilization.
• But they also may have other targets, such as target
rates of return (rsf), imposed or suggested by
shareholders.
• In addition, workers may have a target real wage,
equivalent to some target rate of return (rw), which stops
the firms from achieving their target rate of return.
• Thus, to follow Skott’s analogy, although I may always
be late arriving at work (u > un) , I may be unable to
leave any earlier to arrive on time (u = un), because of
other commitments (rw, rsf).
The Dallery and van Treeck (2008) mechanisms
rsf  1 ( r *  rsf )
s f    2 (rsf  r*)
• Firms may raise their target rate of return when
the actual rate is above the target.
– By doing this, running at an over-normal rate of
utilization, they may succeed in achieving the target
rate of return.
• Firms may also decide to decrease their
retention ratio sf (thus distributing more
dividends), as long as the actual profit rate is
below the target.
– This is similar to the Shaikh I mechanism. But the
utilization rate remains endogenous here.
Conclusion
Anything goes ?
Feyerabend