1
RIETI Discussion Paper Series 13-E-039
May 2013
Stochastic Macro-equilibrium and
Microfoundations for Keynesian Economics
∗
YOSHIKAWA Hiroshi
University of Tokyo
Research Institute of Economy, Trade and Industry
Abstract
In place of the standard search equilibrium, this paper presents an alternative concept of
stochastic macro-equilibrium based on the principle of statistical physics. This concept of
equilibrium is motivated by unspecifiable differences in economic agents and the presence of all
kinds of micro shocks in the macroeconomy. Our model mimics the empirically observed
distribution of labor productivity. The distribution of productivity resulting from the matching
of workers and firms depends crucially on aggregate demand. When aggregate demand rises,
more workers are employed by firms with higher productivity while, at the same time, the
unemployment rate declines. The model provides a micro-foundation for Keynes’ principle of
effective demand.
Keywords: Labor search theory, Microeconomic foundations, Stochastic macro-equilibrium,
Unemployment, Productivity, Keynes’ principle of effective demand
JRL: D39, E10, J64
∗
This work is supported by the Research Institute of Economy, Trade and Industry and the Program for Promoting
Methodological Innovation in Humanities and Social Sciences by Cross-Disciplinary Fusing of the Japan Society for
the Promotion of Science. The empirical works and simulation presented in the paper were carried out by Professor
Hiroshi Iyetomi and Mr. Yoshiyuki Arata. The author is grateful to them for their assistance. He is also grateful for
helpful comments and suggestions by Professor Masahiro Fujita and seminar participants at RIETI. He is indebted to
CRD Association for the Credit Risk Database used.
RIETI Discussion Papers Series aims at widely disseminating research results in the form of professional
papers, thereby stimulating lively discussion. The views expressed in the papers are solely those of the
author(s), and do not represent those of the Research Institute of Economy, Trade and Industry.
2
1. Introduction
The purpose of this paper is to present a new concept of stochastic
macro-equilibrium which provides a micro-foundation for the Keynesian theory of
effective demand. Our problem is cyclical changes in effective utilization of labor. The
concept of stochastic macro-equilibrium is meant to clarify the microeconomic picture
underneath the Keynesian problem of aggregate demand deficiency. It is similar to
standard search equilibrium in spirit, but it is based on the method of statistical physics.
The textbook interpretation of Keynesian economics originating with Modigliani
(1944), regards it as economics of inflexible prices/wages. If price and wages were
flexible enough, the economy would be led to the Pareto optimal neoclassical
equilibrium. However, whatever the reason, when prices and wages are inflexible, the
economy may fall into equilibrium with unemployment. Keynesian economics is meant
to analyze such economy. In this frame of thoughts, micro-foundations for Keynesian
economics would be to provide reasonable explanations for inflexible prices and wages.
Toward this goal, a number of researches were done summarized under the heading of
“New Keynesian economics” (Mankiw and Romer (1991)). The crux of these studies is
to consider optimizing behaviors of the representative household and firm which are
compatible with inflexible prices and wages. New Keynesian dynamic stochastic
general equilibrium (DSGE) models are built in the same spirit.
A different interpretation of Keynesian economics was advanced by Tobin (1993).
“The central Keynesian proposition is not nominal price rigidity but the
principle of effective demand (Keynes, 1936, Ch. 3). In the absence of
instantaneous and complete market clearing, output and employment are
frequently constrained by aggregate demand. In these excess-supply regimes,
agents’ demands are limited by their inability to sell as much as they would
like at prevailing prices. Any failure of price adjustments to keep markets
cleared opens the door for quantities to determine quantities, for example real
national income to determine consumption demand, as described in Keynes’
multiplier calculus. …
In Keynesian business cycle theory, the shocks generating fluctuations
are generally shifts in real aggregate demand for goods and services, notably
in capital investment.(Tobin, 1993)”
Tobin dubbed his own position an “Old Keynesian view”. Certainly, the main
message of Keynes (1936) is that real demand rather than factor endowment and
technology determines the level of aggregate production in the short-run simply because
the rate of utilization of production factors such as labor and capital endogenously
changes responding to changes in real demand. Keynes maintained that this proposition
holds true regardless of flexibility of prices and wages; he, in fact, argued that a fall of
prices and wages would aggravate, not alleviate the problems facing the economy in
3
deep recessions.
Following Tobin, let us call this proposition the Old Keynesian view. According to
the Old Keynesian view, changes in real aggregate demand for goods and services
generate fluctuations of output. The challenge is then to clarify the market mechanism
by which production factors are reallocated in such a way that total output follows
changes in real aggregate demand. A decrease of aggregate output is necessarily
accompanied by lower utilization of production factors, and vice versa. Since the days
of Keynes, economists have taken unemployment as a most important sign of possible
under-utilization of labor. However, unemployment is by definition job search, a kind of
economic activity of worker, and as such calls for explanation. Besides, unemployment
is only a partial indicator of under-utilization of labor in the macroeconomy. The
celebrated Okun’s law which relates the unemployment rate to the growth rate of real
GDP demonstrates the significance of under-utilization of employed labor other than
unemployment
1
. Without minimizing the importance of unemployment, in this paper,
we focus on productivity dispersion in the economy.
To consider Keynes’ principle of effective demand, we must obviously depart
from the Walrasian general equilibrium as represented by Arrow and Debreu (1954).
The most successful example of “non-Walraian economics” which analyzes labor
market in depth is equilibrium search theory surveyed by its pioneers Rogerson, Shimer,
and Wright (2005), Diamond (2011), Mortensen (2011), and Pissarides (2000, 2011).
Search theory stems from realizing limitations of the Walrasian equilibrium. The
standard general equilibrium abstracts itself altogether from the search and matching
costs which are always present in the actual markets. By explicitly exploring search
frictions, search theory has succeeded in shedding much light on the workings of labor
market.
While acknowledging the achievement of equilibrium search theory, we find
several fundamental problems with the standard theory. In particular, the theory fails to
provide a useful framework for explaining cyclical changes in effective utilization of
labor in the macroeconomy; Shimer (2005) demonstrates that the standard search theory
fails to account for stylized empirical facts on cyclical fluctuations of unemployment
and vacancy.
This paper presents an alternative concept of stochastic equilibrium of the
1
Okun (1963) found that a decline of the unemployment rate by one percent raises the growth rate
of real GDP by three percent. The Okun coefficient three is much larger than the elasticity of output
with respect to labor which is supposed to be equal to the labor share, and, therefore, roughly one
third. This finding demonstrates that there exists significant under-employment of labor other than
unemployment in the macroeconomy: See Okun(1973).
4
macroeconomy based on the basic method of statistical physics. Section 2 points out
limitations of standard search theory. After brief explanation of the concept of
equilibrium based on statistical physics in Section 3, Section 4 presents a model of
stochastic macro-equilibrium. Section 5 then explains that the stochastic
macro-equilibrium provides a micro-foundation for Keynes’ principle of effective
demand. It also presents suggestive evidence supporting the model. The final section
offers brief concluding remarks.
2. Limitations of Search Theory
The search theory starts with the presence of various frictions and accompanying
matching costs in market transactions. Once we recognize these problems, we are led to
heterogeneity of economic agents and multiple outcomes in equilibrium. In the simplest
retail market, for example, with search cost, it would be possible to obtain high and low
(more generally multiple) prices for the same good or service in equilibrium. This break
with the law of one price is certainly a big step toward reality. Frictions and matching
costs are particularly significant in labor market. And the analysis of labor market has
direst implications for macroeconomics. In what follows, we will discuss labor search
theory.
In search equilibrium, potentially similar workers and firms experience different
economic outcomes. For example, some workers are employed while others are
unemployed. In this way, search theory well recognizes, even emphasizes heterogeneity
of workers and firms. Despite this recognition, when it comes to model behavior of
economic agent such as worker and firm, it, in effect, presumes the representative agent
in the sense that stochastic economic environment is common to all the agents; Workers
and firms differ only in terms of the realizations of stochastic variable of interest whose
probability distribution is common. Specifically, it is routinely assumed that the job
arrival rate, the job separation rate, and the probability distribution of wages are
common to all the workers and firms. In some models such as Burdett and Mortensen
(1998), the arrival rate is assumed to depend on a worker’s current state, namely either
employed or unemployed. However, within each group, the job arrival rate is common
to all the workers.
The job separation includes layoffs as well as voluntary quits. It makes absolutely
no sense that all the firms and workers face the same job separation rate, particularly the
probability of layoffs. White collar and blue collar workers face different risks of layoff.
Everybody knows the difference between new expanding industry and old declining
industry. In any case, the probability of layoffs depends crucially on the state of demand
5
for the firm’s product, and as a result, among other things, on industry and region, and
ultimately the firm’s performance in the product market. The probability of layoffs
certainly differs significantly across firms.
We can always calculate the average job arrived rate and job separation rate, of
course. However, the average job arrival rate and the economy wide wage distribution
which by definition determine the average duration of an unemployment spell would be
relevant only to decisions made by say, the department of labor. When individual
private economic agent makes decisions, they are not relevant because the job arrival
rate, the job separation rate, and the wage distribution facing individual worker and firm
are all different. Beyond that, the job arrival rate and the wage distribution cannot be
independent in reality. Such complexity of actual labor market motivates our concept of
stochastic macro-equilibrium.
The unrealistic assumption that they are independent and common to all the
economic agents leads to an untenable consequence, that the reservation wage
R
is the
same for all the workers in the standard search model of unemployment.
On these assumptions, the standard analysis typically goes as follows. In the
equilibrium labor market, we must obtain
)1( usfu −
=
(1)
where
u
is the unemployment rate, and
s
and
f
are the separation rate and the job
finding rate, respectively. Equation (1) makes sure the balance between in- and
out-flows of the unemployment pool. If
λ
is the offer arrival rate and
)(wF
is the
cumulative distribution function of wage offers, the job finding rate
f
is
))(1( RFf −=
λ
. (2)
From equations (1) and (2), we obtain
))(
1( R
Fs
s
u
−+
=
λ
. (3)
Equation (2), the standard equation in the literature, presumes that the offer arrival
rate
λ
, the reservation wage
R
, and the cumulative distribution function of wage offers
F
are common to all the workers. However, as we argued above, it is obvious that in
reality,
λ
,
R
and
)(wF
differ significantly across workers. It is actually difficult to
imagine that workers of different educational attainments face the same probability
distribution of wage offers; it is plainly unrealistic to assume that a youngster working
at gas station faces the same probability of getting a job offer from bank as a graduate of
business school. And yet, in standard search models, the assumption that
)(wF
is
common to all the workers is routinely made, and the common
)(wF
is put into the
Bellman equations describing behaviors of firms and workers.
)(wF
for the i-th
6
worker must be
)(wF
i
.
Besides, although wages are one of the most important elements in any job offer,
workers care not only wages but other factors such as job quality, tenure, and location.
Preferences for these other factors which define a job offer certainly differ widely across
workers, and are constantly changing over time. Rogerson, Shimer and Wright (2005;
P.962) say that “although we refer to
w
as the wage, more generally it could capture
some measure of the desirability of the job, depending on benefits, location, prestige,
etc.” However, this is an illegitimate proposition. All the complexities they refer to
simply strengthens the case that we cannot assume that the probability distribution of
wage
)(wF
is common to all the workers. In fact, wage may be even a
lexicographically inferior variable to some workers. For example, pregnant female
worker might prefer a job closer to her home at the expense of lower wage. Workers and
firms all act in their own different universes.
The second problem of standard search model pertains to the behavior of firm. It
is assumed that the product market is perfectly competitive. The price is constant, and
individual demand curve facing the firm is flat. It is curious that the standard search
theory makes so much effort to consider the determination of wages within the firm
while at the same time it leaves the price unexplained under the assumption of perfect
competition. Most firms regard the determination of the prices of their own products as
important as (possibly more important than) the determination of wages.
Under the assumption of perfect competition in the product market, given firm’s
wage offer, whenever a worker comes, the firm is ready to hire him/her though the
worker could turn down the offer. The level of employment is determined only by the
number of successful matching. In Burdett and Mortensen (1998), for example, the flow
of revenue generated by employed worker,
p
is constant and a firm’s steady-state
profit given the wage offer
w
is simply
lwp )( −
where
l
is the number of workers
of this firm.
This analytical framework leads us to some awkward conclusion. Postel-Vinay
and Robin (2002), for example, interpret the empirically observed decreasing number of
workers at high productivity job sites as a consequence of less recruitment efforts made
by high productivity firms than by low productivity firms. However, there is no
reasonable reason why high productivity firms make less recruitment efforts. A much
more plausible reason is that firms are demand constrained in the product market, and
that demand for goods and services produced by high productivity firm is limited.
Shimer (2005) also demonstrates that in the standard search model, an increase in
the job separation rate raises both the unemployment and vacancy rates. This analytical
7
result is in stark contradiction to the well-known negative relationship between two
rates, the Beveridge curve. The awkward result is obtained because the standard model
assumes perfect competition in the product market, and considers strategic behavior of
firms under such assumption. In the economy constrained by real demand, an increase
in the job separation rate is likely to be generated by an increase in layoff which in turn,
is caused by a fall of aggregate demand. The unemployment rate rises while firms
laying off workers certainly post less vacancy signs.
From our perspective, the most serious problem with standard search theory is the
assumption that the product market is perfectly competitive simply because the firm’s
demand for labor depends crucially on the demand for the firm’s products. The
assumption is particularly ill-suited for studying cyclical changes in effective utilization
of labor. In this paper, following Negishi (1979), we assume that instead firms are
monopolistically competitive in the sense that they face downward sloping individual
demand curve in the product market.
Although the most important factor constraining the firm’s demand for labor is
demand constraint in the product market, it is absolutely impossible for us to know
individual demand curve facing each firm. The standard assumption in theoretical
models of imperfectly competitive markets is that the demand system is symmetric,
namely that firms face the same demand condition in equilibrium; See, for example,
Asplund and Nocke (2006). This assumption might be justified in some cases for the
analysis of an industry or a local market, but is absolutely untenable for the purpose of
studying the macroeconomy. Demand is far from symmetric across firms, and there is
no knowing how asymmetric it is in the economy as a whole. Our approach is based on
this fact.
In summary, standard search models are built on several unrealistic assumptions.
First, the job arrival rate, the job separation rate, and the probability distribution of
wages are common to all the workers and firms. Some models assume that workers and
firms are different in terms of their ability, preferences, and productivity (Burdett and
Mortensen (1998), Postel-Vinay and Robin (2002)). However, the probability
distributions of those characteristics are assumed to be utilized by all the economic
agents in common.
We can always find the distribution of any variable of interest for the economy as
a whole. The point is that such macro distribution is not relevant to the decisions made
by individual economic agent because the “universe” of each economic agent is all
different and keeps changing. In this sense, the macroeconomy is fundamentally
different from a local retail market. In the case of a local retail market, one may
8
reasonably assume that all the consumers in the small region share the same distribution
of prices. However, this assumption is untenable for the economy as a whole. Each
agent acts in his/her own universe. For example, a worker may suffer from illness. This
amounts to shocks to his utility function on one hand and to his resource constraint or
“ability” on the other. His preference for job including the reservation wage, the
location and working hours, necessarily changes. At the same time, the job arrival rate
and the wage distribution relevant for his decision making also changes; the
corresponding economy-wide information is not relevant. Note that these micro shocks
are not to cancel out each other in the nature of the case. In fact, it is frictions and
uncertainty emphasized by equilibrium search theory that makes the economy-wide
information such as the average job arrival rate and the wage distribution is irrelevant to
individual economic agent. Thus, among other things, the job arrival rate, the job
separation rate, and the wage distribution are different across workers and firms.
Secondly, the standard assumption that firms are perfectly competitive without any
demand constraint in the product market is particularly ill-suited for the analysis of
cyclical changes in under-utilization of labor. We must assume that instead firms face
downward-slowing demand curve. Because we analyze the macroeconomy, we cannot
assume that demand is symmetric across firms, and moreover, we never know the
details of demand constraints facing firms.
The point is not that we must explicitly introduce all the complexities
characterizing labor market into analytical model. It would simply make model
intractable. Rather, we must fully recognize that it is absolutely impossible to trace the
microeconomic behaviors, namely decision makings of workers and firms in detail. In
the labor market, microeconomic shocks are indeed unspecifiable.
Thus, for the purpose of the analysis of the macroeconomy, sophisticated
optimization exercises based on representative agent assumptions do not make much
sense (Aoki and Yoshikawa (2007)). This is actually partly recognized by search
theorists themselves. The recognition has led them to introduce the “matching function”
into the analysis. The matching function relates the rate of meetings of job seekers and
firms to the numbers of the unemployed and job vacancies. The idea behind it is
explained by Pissarides (2011) as follows.
“Although there were many attempts to derive an equilibrium wage
distribution for markets with search frictions, I took a different approach to
labor market equilibrium that could be better described by the term
“matching”. The idea is that the job search underlying unemployment in the
official definitions is not about looking for a good wage, but about looking
for a good job match. Moreover, it is not only the worker who is concerned
to find a good match, with the firm passively prepared to hire anyone who
9
accepts its wage offer, but the firm is also as concerned with locating a good
match before hiring someone.
The foundation for this idea is that each worker has many distinct
features, which make her suitable for different kinds of jobs. Job
requirements vary across firms too, and employers are not indifferent about
the type of worker that they hire, whatever the wage. The process of
matching workers to jobs takes time, irrespective of the wage offered by
each job. A process whereby both workers and firms search for each other
and jointly either accept or reject the match seemed to be closer to reality.
…… It allowed one to study equilibrium models that could incorporate
real-world features like differences across workers and jobs, and differences
in the institutional structure of labor markets.
The step from a theory of search based on the acceptance of a wage
offer to one based on a good match is small but has far-reaching
implications for the modeling of the labor market. The reason is that in the
case of searching for a good match we can bring in the matching function as
a description of the choices available to the worker. The matching function
captures many features of frictions in labor markets that are not made
explicit. It is a black box, as Barbara Petrongolo and I called it in our 2001
survey, in the same sense that the production function is a black box of
technology. (Pissarides, 2011; pp.1093-1094)”
What job seeker is looking for is not simply a good wage, but a good job offer which
cannot be uniquely defined but differs significantly across workers. It is simply
unspecifiable. Pissarides recognizes such “real-world features” as differences across
workers and jobs; the “universe” differs across workers and firms. Then, at the same
time, he recognizes that we need a macro black box. The matching function is certainly
a black box not explicitly derived from micro optimization exercises, and is, in fact, not
a function of any economic variable which directly affects the decisions of individual
workers and firms. Good in spirit, but the matching function is still only a half way in
our view.
The matching function is based on a kind of common sense in that the number of
job matchings would increase when there are a greater number of both job seekers and
vacancies. However, it still abstracts itself from an important aspect of reality. As Okun
(1973) emphasizes, the problem of unemployment cannot be reduced only to numbers.
“The evidence presented above confirms that a high-pressure economy
generates not only more jobs than does a slack economy, but also a different
pattern of employment. It suggests that, in a weak labor market, a poor job is
often the best job available, superior at least to the alternative of no job. A
high-pressure economy provides people with a chance to climb ladders to
better jobs.
The industry shifts are only one dimension of ladder climbing.
Increased upward movements within firms and within industries, and greater
geographical flows from lower-income to higher-income regions, are also
likely to be significant. (Okun, 1973; pp.234-235)”
Dynamics of unemployment cannot be separated from qualities of jobs, or more
10
specifically distribution of productivity on which we focus in the present paper.
Our analysis, in fact, demonstrates that the matching “function” is not a
structurally given function even in the short-run, but depends crucially on the level of
aggregate demand. To explicitly consider these problems, we face greater complexity
and, therefore, need a “greater macro black box” than the standard matching function.
This is the motive for stochastic macro-equilibrium we explain in the next section.
3. Stochastic Macro-equilibrium —— The Basic Idea
Our vision of the macroeconomy is basically the same as standard search theory.
Workers are always interested in better job opportunities, and occasionally change their
jobs. Job opportunities facing workers are stochastic depending on vacancy signs posted
by firms. Firms, on the other hand, make their efforts to recruit and retain the best
workers for their purposes. We assume that firms are demand-constrained in the product
market. Demand determines the level of production and, as a consequence, demand for
labor.
Unemployment, a great challenge to any economy, deserves special attention in
economic analysis, but actually a significant part of workers change their jobs without
experiencing any spell of unemployment. The importance of on the job search also
means that job turnover depends crucially on the distribution of qualities of jobs in the
economy. Postel-Vinay and Robin (2002), in fact, analyze on-the-job search explicitly
considering workers with different abilities on one hand and firms with different
productivities on the other. Their analysis, however, still depends on restrictive
assumptions that all the workers face the same job offer distribution independent of
their ability and employment status, and that the product market is perfectly
competitive.
While workers search for suitable jobs, firms also search for suitable workers.
Firm’s job offer is, of course, conditional on its economic performance. The present
analysis focuses on the firm’s labor productivity. The firm’s labor productivity increases
thanks to capital accumulation and technical progress or innovations. However, those
job sites with high productivity remain only potential unless firms face high enough
demand for their products; firms may not post job vacancy signs or even discharge the
existing workers when demand is low. As noted above, we assume that firms are all
monopolistically competitive in the sense that they face downward sloping individual
demand curves, and the levels of production are determined by demand rather than
increasing marginal costs.
Formally, a most elegant general equilibrium model of monopolistic competition
11
is given by Negishi (1960-61). Negishi (1979) persuasively argues that when the firm is
monopolistically competitive, the downward-sloping individual demand curve must be
kinked at the current level of output and price. The corresponding marginal revenue
becomes discontinuous as shown in Figure 1. The response of the firm’s sales to a
change in price is asymmetric because of different reactions on the part of vital firms.
With the perceived kinked individual demand curve, a shift of demand is basically
absorbed by a change in output leaving price unchanged. As Tobin (1993) says, the
model opens the door for “quantities determine quantities.
2
”
Negishi (1979; Chapter 6) establishes the existence of general equilibrium of
monopolistically competitive firms facing the kinked individual demand curve as shown
in Figure 1. Though the equilibrium exists, it is indeterminate because the initial levels
of the firms’ output and price (the kink point of the individual demand curve) must be
exogenously given. To give such initial condition amounts to determining the level of
individual demand curve facing each firm. The present analysis provides a rule for
allocating the aggregate demand to monopolistically competitive firms based on the
principle of statistical physics.
Negishi (1979)’s general equilibrium model of monopolistically competitive firms
abstracts itself from frictions and uncertainty in the labor market. Unlike capital, worker
either stays at a job, moves to another, or searches for one as the unemployed. Firms try
to recruit the best workers for their purposes. These are, of course, aspects of labor
market which search theory analyses. We follow its spirit, but not its analytical method.
In the present analysis, we simply assume that firms with higher productivity
make more attractive job offers to workers. This assumption means that whenever
possible, workers move to firms with higher productivity. However, for workers to
move to firms with higher productivity, it is necessary that those firms must decide to
fill the vacant job sites, and post enough number of vacancy signs. They post such
vacancy signs only when they face an increase of demand for their products, and decide
to raise the level of production. It also goes without saying that high productivity firms
keep their existing workers only when they face high enough demand.
The question we ask is what the distribution of employed workers is across firms
whose productivities differ. As we argued in the pervious section, because
microeconomic shocks to both workers and firms are so complex and unspecifiable,
optimization exercises based on representative agent assumptions do not help us much.
2
In Negishi (1979)’s model of monopolistically competitive firms, a change in wage does not affect
the level of production, and, therefore leaves the number of vacancy unchanged because marginal
revenue is discontinuous. This is in stark contrast to the standard search model.
12
In particular, we never know how the aggregate demand is distributed across
monopolistically competitive firms. To repeat, among other things, the job arrival rate,
the job separation rate, and the probability distribution of wages (or more generally
measure of the desirability of the job) differ across workers and firms. This recognition
is precisely the starting point of the fundamental method of statistical physics
3
. At first,
one might think that allowing too large a dispersion of individual characteristics leaves
so many degrees of freedom that almost anything can happen. However, it turns out that
the methods of statistical physics provide us not only with qualitative results but also
with quantitative predictions.
In the present model, the fundamental constraint on the economy as a whole is
aggregate demand
D
. Accordingly, to each firm facing the downward-sloping individual
demand curve, the level of demand for its product is the fundamental constraint. The
problem is how the aggregate demand
D
is allocated to these monopolistically
competitive firms. Our model provides a solution. The basic idea behind the analysis
can be explained with the help of the simplest case.
Suppose that
k
n
workers belong to firms whose productivity is
k
c
(
'kk
cc <
where
'kk <
). There are
K
levels of productivity in the economy
(
K
k
,,
2
,
1
=
). The total number of workers
N
is given.
Nn
K
k
k
=
∑
=
1
(4)
A vector
),,,(
21 K
nnnn =
shows a particular allocation of workers across firms
with different productivities. The combinatorial number
n
W
of obtaining this allocation,
n
, is equal to that of throwing
N
balls to
K
different boxes. It is
!
!
1
∏
=
=
K
k
k
n
n
N
W
(5)
Because the number of all the possible ways to allocate
N
different balls to
K
different
boxes is
N
K
, the probability that a particular allocation
),,
,(
21 K
nnn
n =
is obtained
is
!
!1
1
∏
=
==
K
k
k
NN
n
n
n
N
KK
W
P
. (6)
It is the fundamental postulate of statistical physics that the state or the allocation
),,,(
2
1 K
nnnn =
which maximizes the probability
n
P
or (6) under macro-constraints is
3
Foley (1994), in his seminal application of this approach to general equilibrium theory, advanced
the idea of “statistical equilibrium theory of markets”.
13
to be actually observed in the economy. The idea is similar to maximum likelihood in
statistics/econometrics.
Maximizing
n
P
is equivalent to maximizing
n
Pln
. Applying the Stirling formula
for large number
x
xxxx −≅ ln!ln
, (7)
we find that the maximization of
n
Pln
is equivalent to that of
S
.
k
K
k
k
p
p
S ln
1
∑
=
−=
)(
N
n
p
k
k
=
(8)
S
is the Shannon entropy, and captures the combinatorial aspect of the problem.
It can be easily understood with the help of a simple example. When we throw a pair of
dice, the possible sums range from two to twelve. Six is much more likely than twelve
simply because the combinatorial number of the former is five whereas that of the latter
is only one. Though the combinatorial consideration summarized in the entropy plays a
decisive role for the final outcome, that is not the whole story, of course. The
qualification “under macro-constraints” is crucial.
The first macro-constraint concerns the labor endowment, (4). The second
macro-constraint concerns the effective demand.
Dn
c
K
k
k
k
=
∑
=1
(9)
Here, aggregate demand
D
is assumed to be given. In our analyis, we explicitly analyze
the allocation of labor
),,,(
21 K
nnn
. Note, however, that the allocation of labor is
basically equivalent to the allocation of the aggregate demand to monopolistically
competitive firms.
To maximize entropy
S
under two macro-constraints (4) and (9), set up the
following Lagrangean form
L
:
−+
−+
−
=
∑∑∑
===
K
k
k
k
K
k
k
k
K
k
k
ncD
nN
N
n
N
n
L
111
ln
β
α
(10)
with two Lagrangean multipliers,
α
and
β
. Maximization of this Lagrangean form
with respect to
k
n
leads us to the first-order conditions:
k
k
NcN
N
n
βα
−−−=
1ln
(
Kk ,
,2,
1 =
) (11)
which are equivalent to
[ ]
k
k
NcN
N
n
βα
−−−= 1exp
. (12)
14
Because
N
n
k
sums up to one, we obtain
∑
=
−
−
=
K
k
Nc
Nc
k
k
k
e
e
N
n
1
β
β
(13)
Thus, the number of workers working at firms with productivity
k
c
is exponentially
distributed. It is known as the Boltzmann distribution in physics.
Here arises a crucial difference between economics and physics. In physics,
k
c
corresponds to the level of energy. Whenever possible, particles tend to move toward
the lowest energy level. To the contrary, in economics, workers always strive for better
jobs offered by firms with higher productivity
k
c
. In fact, if allowed, all the workers
would move up to the job sites with the highest productivity,
K
c
. This situation shown
in Figure 2 corresponds to the textbook Pareto optimal equilibrium with no frictions.
However, this state is actually impossible unless the level of aggregate demand
D
is so
high as equal to the maximum level
NcD
K
=
max
. Only when
D
is equal to
max
D
,
demand-constrained firms with the highest productivity face high enough individual
demand curves so that they have incentives to employ
N
workers. On the part of
workers, they would be able to find job sites with the highest productivity because firms
with
K
c
try hard to keep their existing workers, and at the same time post enough
vacancy signs to aggressively recruit. With frictions,
D
must be actually higher than
max
D
for the state shown in Figure 2 to be realized.
The maximum level of aggregate demand
max
D
is only imaginary, however.
When
D
is lower than
max
D
, the story is quite different. Some workers —— a
majority of workers, in fact, must work at job sites with productivity lower than
K
c
.
How are workers distributed over job sites with different productivity? Obviously, it
depends on the level of aggregate demand. When
D
reaches its lowest level,
min
D
,
workers are distributed evenly across all the sectors with different levels of productivity,
K
ccc ,,,
21
(Figure 3). Here,
min
D
is defined as
Kcc
cN
D
k
/)(
2
1min
+++
=
.
It is easy to see that the lower the level of
D
is, the greater the combinatorial
number of distribution
),,
,(
21 K
nnn
which satisfies aggregate demand constraint (9)
becomes. Note that this combinatorial number basically corresponds to the distribution
of demand for goods and services across job sites at monopolistically competitive firms
with different levels of productivity.
As explained above, the combinatorial number
n
W
of a particular allocation
),,,(
2
1 K
nnnn =
is basically equivalent to the Shannon entropy,
S
defined by (8).
S
increases when
D
decreases. In the extreme case where
D
is equal to (or more
realistically greater than) the maximum level
max
D
, all the workers work at job sites
15
with the highest productivity because whenever possible, workers move to better job
sites with higher productivity (Figure 2). In this case, the entropy
S
becomes zero, its
lowest level, because
1=Nn
K
and
)(0 KkNn
k
≠=
. In the other extreme where
KN
n
K
=
(Figure 3), the entropy
S
defined by (8) becomes
Kln
, its maximum
level. Thus, the relation between the entropy
S
and the level of aggregate demand
D
looks like the one shown in Figure 4.
At this stage, we can recall that the Lagrangean multiplier
β
in (10) for
aggregate demand constraint is equal to
D
S
D
L
∂
∂
=
∂
∂
=
β
. (14)
β
is the slope of the tangent of the curve as shown in Figure 4, and, therefore, is
negative
4
. Table 1 shows how negative
β
relates to the level of aggregate demand
D
.
With negative
β
, the exponential distribution (13) looks like the one shown in
Figure 5. Whenever possible, workers always move toward job sites with higher
productivity. As a result, the distribution is upward-sloping. However, unless the
aggregate demand is equal to (or greater than) the maximum level,
max
D
, workers’
efforts to reach job sites with the highest productivity
K
c
must be frustrated because
firms with the highest productivity do not need to employ a large number of workers
and are less aggressive in recruitment, and accordingly it gets harder for workers to find
such jobs. As a consequence, workers are distributed over all the job-sites with different
levels of productivity.
The maximization of entropy under the aggregate demand constraint (10), in fact,
balances two forces. On one hand, whenever possible, workers are assumed to move to
better jobs which are identified with job sites with higher productivity. It is the outcome
of successful job matching resulting from the worker’s search and the firm’s recruitment.
When the level of aggregate demand is high, this force dominates. However, when
D
is lower than
max
D
, there are in general a number of different allocations
),,
,(
21 K
nn
n
which are consistent with
D
.
As we explained in the previous section, micro shocks facing both workers and
firms are truly unspecifiable. We simply do not know, on one hand, which firms with
4
In physics,
β
is positive in most cases. This difference arises because workers strive for job sites
with higher productivity, not the other way round (Iyetomi (2012)). In physics,
β
is equal to the
inverse of temperature, or more precisely, temperature is defined as the inverse of
DS ∂∂
when
S
is the entropy and
D
energy. Thus, negative
β
means the negative temperature. It may sound
odd, but the notion of negative temperature is perfectly legitimate in such systems as the one in the
present analysis; see Ramsey (1965) and Appendix E of Kittel and Kroemer (1980).
16
what productivity face how much demand constraint and need to employ how many
workers with what qualifications, and on the other, which workers are seeking what
kind of jobs with how much productivity. Here comes the maximization of entropy. It
gives us the distribution
),,,(
21 K
nnn
which corresponds to the maximum
combinatorial number consistent with given
D
. The entropy maximization under
aggregate demand constraint plays, therefore, the role similar to the matching function
in the standard search theory. Note that unlike the standard matching function which
focuses only on the number of jobs, the matching of job quality characterized by
productivity plays a central role in the present analysis. The matching function
presumes that the number of job matching increases when the numbers of the
unemployed and vacancy increase. In the present analysis, the matching of high
productivity jobs is ultimately conditioned by the level of aggregate demand because
high aggregate demand loosens demand constraints facing monopolistically competitive
high productivity firms, and vice versa. Uncertainty and frictions emphasized by the
standard search theory are not exogenously given, but depend crucially on aggregate
demand; In a booming gold-rush town, one does not waste a minute to find a good job!
The distribution of workers over job-sites with different productivity levels which
results from this analysis is an upward-sloping exponential distribution as shown in
Figure 5. As one should expect, the higher the level of aggregate demand
D
is, the
steeper the distribution becomes meaning that more workers are mobilized to jobs with
higher productivity (Okun (1973)).
It is essential to understand that the present approach does not regard economic
agents’ behaviors as random. Certainly, firms and workers maximize their profits and
utilities. The present analysis, in fact, presumes that workers always strive for better
jobs characterized by higher productivity. However, firms are demand-constrained
facing downward-sloping demand curves. As a result of profit-maximization, the
optimal level of production is constrained by demand. Unless the level of aggregate
demand is extremely high and equal to (with frictions, greater than) its maximum level
NC
K
, micro demand constraints become effective. In this case, the entropy matters
because we can never know details of micro behaviors. Randomness underneath the
entropy maximization comes from the fact that both the objective functions of and
constraints facing a large number of economic agents are constantly subject to
unspecifiable micro shocks. We must recall that the number of households is of order
10
7
, and the number of firms, 10
6
. Therefore, there is nothing for outside observers,
namely economists analyzing the macroeconomy but to regard a particular allocation
under macro-constraints as equi-probable. Then, it is most likely that the allocation of
17
the aggregate demand and accordingly workers which maximizes the probability
n
P
or
(6) under macro-constraints is realized
5
.
4. The Model
The above analysis shows that the distribution of workers at firms with different
productivities depends crucially on the level of aggregate demand. Though the basic
model is good enough to explain the basic idea, it is too simple to explain the
empirically observed distribution of labor productivity.
Empirical Distribution of Productivity
Figures 6 (a), (b) and (c) show the distributions of workers at different
productivity levels for the Japanese economy ((a) total, (b) manufacturing and (c)
non-manufacturing industries). The data used are the Nikkei Economic Electric
Database (NEEDS, http://www.crd-office.net/CRD/english/index.html) and the Credit
Risk Database (CRD, http://www.crd-office.net/CRD/english/index.html) which cover
more than a million large and medium/small firms for 2007.
The “productivity” here is simply value added of the firm divided by the number
of employed workers, that is, the average labor productivity. Theoretically, we should be
interested in unobserved marginal productivity, not the average productivity. Besides,
proper “labor input” must be in terms of work hour, or for that matter even in terms of
work efficiency units rather than the number of workers. For these reasons, the average
labor productivity shown in the figure is a crude measure of theoretically meaningful
unobserved marginal productivity. However, Aoyama et. al. (2010; p.38-41)
demonstrates that when the average productivity and measurement errors are
independent, the distribution of true marginal productivity obeys the power law with the
5
This method has been time and again successful in natural sciences when we analyze object
comprising many micro elements. Economists might be still skeptical of the validity of the method
in economics saying that inorganic atoms and molecules comprising gas are essentially different
from optimizing economic agents. Every student of economics knows that behavior of dynamically
optimizing economic agent such as the Ramey consumer is described by the Euler equation for a
problem of calculus of variation. On the surface, such a sophisticated economic behavior must look
remote from “mechanical” movements of an inorganic particle which only satisfy the law of motion.
However, every student of physics knows that the Newtonian law of motion is actually nothing but
the Euler equation for a certain variational problem; particles minimizes the energy or the
Hamiltonian!. It is called the principle of least action: see Chapter 19 of Feynmann (1964)’s Lectures
on Physics, Vol. II. Therefore, behavior of dynamically optimizing economic agent and motions of
inorganic particle are on a par to the extent that they both satisfy the Euler equations for respective
variational problems. The method of statistical physics can be usefully applied not because motions
of micro units are “mechanical,” but because object under investigation comprises many micro units
individual movements of which we are unable to know. It leads us to a specific exponential
distribution (13).
18
same exponent as that for the measured average productivity. In other words,
distribution is robust with respect to measurement errors in the present case.
Incidentally, we also note that the Pareto efficiency of the economy pertains to marginal
products of production factors, not to total factor productivity (TFP) some economists
focus on.
Figure 6 drawn on the double logarithm plane broadly shows that (1) the
distribution of labor productivity is single-peaked, (2) in the low productivity (left)
region, it is upward-sloping exponential whereas (3) in the high productivity (right)
region, it obeys downward-sloping power-law (Aoyama et. al., (2010)). Ikeda and
Souma (2009) find a similar distribution of productivity for the U.S. while DelliGatti et.
al. (2008) find power-law tails of productivity distribution for France and Italy. In what
follows, we present an extended model based on the principle of statistical physics for
explaining the broad shape of this empirically observed distribution.
We explained in the previous section that the entropy maximization under macro
constraints leads us to exponential distribution. This distribution with negative
β
can
explain the broad pattern of the left-hand side of the distribution shown in Figures 6 (a)
(b) and (c), namely an upward-sloping exponential distribution (Iyetomi (2012)).
However, we cannot explain the right-hand side downward-sloping power distribution
for high productivity firms. To explain it, we need to make an additional assumption
that the number of potentially available high-productivity jobs is limited, and that it
decreases as the level of productivity rises.
Dynamics of Potential Job Creation/Destruction
Potential jobs are created by firms by accumulating capital and/or introducing new
technologies, particularly new products. On the other hand, they are destroyed by firms’
losing demand for their products permanently. Schumpeterian innovations by way of
creative destruction raise the levels of some potential jobs, but at the same time lower
the levels of others. In this way, the number of potential jobs with a particular level of
productivity keeps changing. They remain potential because firms do not attempt to fill
all the job sites with workers. To fill them, firms either keep the existing workers on the
job or post job vacancy signs, but it is an economic decision, and depends crucially on
the economic conditions facing the firms. The most important constraint on the firm’s
employment is the level of demand in the product market. The number of potential job
sites, therefore, is not exactly equal to, but rather imposes a ceiling on the sum of the
number of filled job sites, namely employment and the number of job openings or
19
vacancy signs.
The statistical theory we will discuss later explains how employment is
determined. In this sub-section, we first consider dynamics of potential job creation and
destruction. Causes of creation/destruction of potential job sites are micro-shocks, and
as explained in the previous section, unspecifiable. The best way to describe them is
Markov model. Good examples in economics are Champernowne (1950) on income
distribution, and Ijiri and Simon (1979) on size distribution of firms. Here, we adapt the
model of Marsili and Zhang (1998) to our own purpose. The goal of the analysis is to
derive a power-law distribution such as the one for the tail of the empirically observed
distribution of labor productivity.
Suppose that there are
j
f
”potential” jobs with productivity,
j
c
(
'
j
j
cc <
,
when
'
jj <
). In small time interval
dt
, the level of productivity
c
of a potential job
site increases with probability
dt
c
w
)
(
+
by a small amount which we can assume is
unity without loss of generality. We denote this probability as
)(c
w
+
because
+
w
depends on the level of
c
. Likewise, it decreases by a unit with productivity
dtcw )(
−
.
Thus,
)(
cw
+
and
)(cw
−
are transition rates for the processes
1+
→ cc
and
1−→ cc
, respectively. As noted above, the productivity of job site changes for many
reasons. It may reflect technical progress or innovations. Given significant costs of
adjusting labor, productivity also changes when demand for firm’s products change. For
example, when demand for firm’s products falls, labor productivity necessarily declines.
The decline of productivity in this case reflects labor hoarding (Fay and Medoff (1985)).
These changes are captured by transition rates,
)(cw
+
and
)(cw
−
in the model.
We also assume that a new job site is born with a unit of productivity with
probability
dtp
. On the other hand, a job site with productivity
1=c
will cease to
exist if
c
falls to zero. Thus the probability of exit is
dtcw
)1( =
−
. A set of the
transition rates and the entry probability specifies a jump Markov process.
Given this Markov model, the evolution of the average number of job sites of
productivity
c
at time
t
,
),( tcf
, obeys the following master equation:
),1()
1(),1
()1(
),(
tcfcw
tcfcw
t
t
cf
++
+−−=
∂
∂
−
+
1,
),(
)(),()(
c
ptcfcwtc
fcw
δ
+−−
−+
, (15)
Here,
1,c
δ
is 1 if
1=c
, and 0 if otherwise. This equation shows that change in
),( tcf
20
over time is nothing but the net inflow to the state
c
.
We consider the steady state or the stationary solution of equation (15) such that
0
/
)
,
(
=
∂
∂
t
t
c
f
. The solution
)(cf
can be readily obtained by using the boundary
condition that
pfw =
−
)1()1(
:
∏
−
=
−
+
+−
−
=
1
1
)1(
)(
)1()(
c
k
kcw
kcw
fcf
. (16)
Here, we make an important assumption on the transition rates,
+
w
and
−
w
.
Namely, we assume that the probabilities of an increase and a decrease of productivity
depend on the current level of productivity of job site. Specifically, the higher the
current level of productivity, the larger a chance of unit productivity change. This
assumption means that the transition rates can be written as
α
cacw
++
=)(
and
α
cacw
−−
=)(
, respectively. Here,
+
a
and
−
a
are positive constants, and
α
is greater
than 1. Under this assumption, the stationary solution (16) becomes
*
/
)(
)(
)/)1(1(
/)1(1
)1(
)(
cc
c
ec
c
Cf
Cf
f
cf
−−
≅
−
−
=
α
α
α
α
. (17)
where
1
)(
*
)/)1((
−
≡
α
Cfc
and
∑
∞
=
≡
1
)(
)(
c
cfcC
α
α
.
We use the relation
)(
/
)1(1
/
α
Cfa
a −=
−
+
. The approximation in equation (17)
follows from
1/)
1(
)(
<<
α
Cn
, and the exponential cut-off works as
c
approaches to
*
c
.
However, the value of
*
c
is practically large, and therefore, we observe the power law
distribution
α
−
∝
c
cn
)(
for a wide range of
c
in spite of the cut-off. Therefore, under
the reasonable assumption, we obtain power law distribution for job sites,
j
f
.
The above model can be understood easily with the help of an analogy with the
formation of cities. Imagine that
),
( tcf
is the number of cities with population
c
at
time
t
.
)(cw
+
corresponds to a birth in a city with population
c
, or an inflow into the
city from another city. Similarly,
)(cw
−
represents a death or an exit of a person
moving to another city. These rates are the instantaneous probabilities that the
population of a city with the current population
c
either increases or decreases by 1.
They are, therefore, the entry and exit rates of one person times the population
c
of the
city, respectively. In addition, a drifter forms his own one-person city with the
instantaneous probability
p
. In this model, the dynamics of
),( tcf
or the average
21
number of cities with population
c
, is given by equation (15). In the case of population
dynamics, one might assume that the entry (or birth) and exit (or death) rates of a person,
+
a
and
−
a
, are independent of the size of population of the city in which the person
lives. In that case
)(cw
+
and
)(cw
−
become linear functions of
c
, namely
ca
+
and
ca
−
. However, even in population dynamics, one might assume that the entry rate of a
person into a large city is higher than its counterpart in a smaller one because of better
job opportunities or the social attractiveness of such places, as encapsulated in the
words of the song, “bright lights, big city”. The same may hold true for exit and death
rates because of congestion or epidemics.
In turns out that in the dynamics of productivity of job site, both the “entry” and
“exit” rates of an existing “productivity job site” are increasing functions of
c
, namely
the level of productivity in which that particular job site happens to be located; to be
concrete,
ca
+
and
ca
−
. Thus,
)(cw
+
becomes
ca
+
times
c
which is equal to
2
ca
+
. Likewise, we obtain
2
)( cacw
−−
=
. This is the case of the so-called Zipf law (Ijiri
and Simon (1975)). Thus, under the reasonable assumption that the probability of a unit
change in productivity is an increasing function of its current level
c
, we obtain power
law distribution for job sites
j
f
.
Economists often presume that changes in productivity are caused by supply-side
factors such as technical progress and entry/exit of firm alone. However, an important
source of productivity change is actually a sectoral shift of demand. Indeed, Fay and
Medoff (1985) documented such changes in firm’s labor productivity by way of changes
in the rate of labor hoarding. Stochastic productivity changes as described in our
Markov model certainly include technical progress, particularly in the case of an
increase, but at the same time they embrace allocative demand disturbances. By their
careful study of industry-level productivity and worker flows, Davis et al. (1996) found
that while job creation is higher in industries with high total factor productivity growth,
job destruction for an industry is not systematically related to productivity growth; job
destruction is actually highest in the industries in the top productivity growth quintile
(Table 3.7, Davis et. al. (1996, P.52)). This finding suggests the importance of negative
demand shocks for job destruction.
Distribution of Productivity
Distribution of potential job sites with high productivity obeys downward-sloping
22
power law. However, the determination of employment by firms with various levels of
productivity is another matter. To fill potential job sites with workers is the firm’s
economic decision. The most important constraining factor is the level of demand
facing the firm in the product market. Whatever the level, to fill potential job sites, the
firm must either keep the existing workers on the job or post vacancy signs toward
successful job matching. Such actions of the firms and job search of workers are not
random but purposeful. However, micro shocks affecting firms and workers are just
unspecifiable. Then, how are workers actually employed at firms with various levels of
productivity? This is the problem we consider in what follows.
The number of workers working at the firms with productivity,
j
c
, namely
j
n
is
}
,,1,0{
jj
fn ∈
),2
,1
(
Kj =
. (18)
Here,
j
f
is the number of potential jobs with productivity
j
c
, and puts a ceiling on
j
n
6
. We assume that in the low productivity region,
j
f
is large enough meaning that
j
n
is virtually unconstrained by
j
f
. In contrast, in the high productivity region,
j
f
constrains
j
n
, and its distribution is power distribution as we have analyzed above.
Like many others, Postel-Vinay and Robin (2002) assume that there is no ceiling
for job-sites with high productivity. On this assumption, they regard a decreasing
number of workers with high productivity jobs as a consequence of less recruitment
efforts made by high productivity firms than by low productivity firms. This is an
awkward interpretation. There is no reasonable reason why high productivity firms
make less recruitment efforts than low productivity firms. A more plausible assumption
is that firms are monopolistically competitive facing the downward-sloping individual
demand curve rather than the price takers in the product market, and that jobs with high
productivity are limited in number by such demand constraints. Suppose, for example,
that automobile industry is a high productivity industry. It would be unreasonable to
argue that the level of employment in the industry is only the outcome of job matching,
and that a limited size of employment is due to a lack of the firms’ recruitment efforts.
The size of car producers is basically determined by the level of demand for cars and
their capacity.
6
When the number of potential jobs with high productivity is limited, behavior of economic agents
necessarily becomes correlated; If good jobs are taken by some workers, it becomes more difficult
for others to find such jobs. Garibaldi and Scalas (2010) suggest that we study the problem by
Markov model with such constraints.
23
The total number of employed workers is simply the sum of
j
n
:
∑
=
=
K
j
j
nN
1
. (19)
In the basic model explained in section 3, the total number of employed workers,
N
is
exogenously given (equation (4)). In the present model,
N
is assumed to be variable.
N
is smaller than the exogenously given total number of workers or labor
force,
L
(
L
N
<
). The difference between
L
and
N
is the number of the unemployed,
U
:
NLU
−=
. (20)
As in the basic model, firms are monopolistically competitive facing the
downward-sloping individual demand curve, and as a consequence the total output is
constrained by the aggregate demand,
D
(Equation (9)). In the basic model,
D
is
literally given. Accordingly, total output is also given by equation (9), and is constant. In
the present model, we more realistically assume that in accordance with fluctuations of
aggregate demand, total output
Y
also fluctuates. Specifically,
Y
defined by
∑
=
=
K
k
kk
ncY
1
(21)
is now stochastic, and its expected value
>< Y
is equal to constant
D
. That is, we
have
DY >=<
. (22)
Aggregate demand constrains total output in the sense of its expected value.
Under this assumption, the probability of total output
Y
turns out to be
exponential; The density function
)(Yg
is
∑
−
−
=
i
Y
Y
i
i
e
e
Yg
β
β
)(
(23)
This result is obtained by the method of Gibbs’ canonical ensemble. Gibbs
established the statistical mechanics by introducing the concept of “canonical ensemble”
which is a collection of macro states,
Y
in our present case. Suppose that there are
K
possible levels of
Y
denoted by
K
YY ,
1
. For the moment, we reinterpret
k
n
as the
number of cases where
Y
takes the value
k
Y
(
Kk ,,1 =
). The sum of
k
n
,
N
is
given. Then,
k
n
satisfies equation (4). We assume that the average of
Y
is equal to
constant
D
.
24
D
N
n
Y
K
k
k
k
=
∑
=1
(24)
Replacing
k
c
by
NY
k
/
, we observe that equation (24) is equivalent to equation (9).
Thus, we can apply the exactly same entropy maximization as we did in the basic model
in section 3. It leads us to equation (23). In (23),
β
is the Lagrange multiplier for
constraint (24) in the maximization of entropy (8).
Obviously,
Y
constrained by aggregate demand
D
affects the distribution of
workers,
k
n
(equation (21)). In the present model, the number of employed workers
N
is not constant, but changes causing changes in unemployment. Besides, the number
of potential job sites with high productivity,
j
f
constrains
j
n
. Under these
assumptions, we seek the state which maximizes the probability
n
P
or equation (6).
Before we proceed, it is useful to explain partition function because it is rarely
used in economics, but we will use it intensively in the subsequent analysis. When a
stochastic variable
Y
is exponentially distributed, that is, its density function
)(Yg
is
given by equation (23), partition function
Z
is defined as
∑
−
=
i
Y
i
eZ
β
. (25)
This function is extremely useful as moment generating function. For example, the first
moment or the average of
Y
can be simply found by differentiating
Zlog
with respect
to
β
.
∑
∑
∑
−
−
−
−
−=−=−
i
Y
i
Y
i
i
Y
i
i
i
e
eY
e
d
d
d
Zd
β
β
β
ββ
)(
)log(
log
)(
)(
)(
i
i
ii
i
Y
Y
i
i
Y
EY
gY
e
e
Y
i
i
=
==
∑
∑
∑
−
−
β
β
(26)
As in the basic model, we want to find the state which maximizes the probability,
n
P
or equation (6). We have two macro-constraints, equations (19) and (21). The total
number of workers employed
N
is, however, not constant but variable. The aggregate
output
Y
is also not constant but obeys the exponential distribution, namely equation
(23).
Because the level of total output depends on the total number of employed
workers
N
, we denote
i
Y
as
)
(NY
i
. Then, the canonical partition function
N
Z
can
25
be written as
∑
−
=
i
N
Y
N
i
eZ
)(
β
. (27)
Using equation (21), we can rewrite this partition function as follow:
{ }
)exp(
1
∑ ∑
=
−
=
i
n
K
i
i
iN
n
c
Z
β
. (28)
Unfortunately, it is generally difficult to carry out the summation with respect to
{ }
i
n
under constraint (19), namely
∑
=
i
nN
. Rather than taking
N
as given, we better
allow
N
to be variable as we do here, and consider the grand canonical partition
function
Φ
.
The grand canonical partition function is defined as
∑
∞
=
=Φ
0N
N
N
Zz
(29)
where
βµ
e
z =
. (30)
The parameter
µ
is called the chemical potential in physics, and measures the marginal
contribution in terms of energy of an additional particle to the system under
investigation. In the present model,
N
is the number of employed workers, and,
therefore,
µ
measures the marginal product of a worker who newly acquired job out
of the pool of unemployment. Thus,
µ
is equivalent to the reservation wage, or more
generally the “reservation job offer” of the unemployed. When
µ
is high, the
unemployed worker is choosy, and vice versa.
Substituting equation (28) into equation (29), the grand canonical partition
function
Φ
becomes as follows:
}exp{
0
∑
∑∑
−=Φ
∞
= nj j
j
j
N
N
cnz
β
where
βµ
ez =
(31)
Using the definitions of
z
, (30), and also
N
, (19), we have
])(
exp[}exp{
1
0
)(
1
jj
K
j
nj
n j
jj
N
nn
nccne
j
K
−=−=Φ
∏
∑∑ ∑∑
=
∞
=
++
µββ
βµ
. (32)
Because there is ceiling
j
f
for
j
n
(constraint (18)), (32) can be rewritten as follows:
26
∏∏
=
−
−+
−
−
=
−
−
=++=Φ
K
j
c
cf
cf
c
K
j
e
e
ee
j
j
1
)(
)()1(
)(
)(
1
]
1
1
[]1[
µβ
µβ
µβ
µβ
(33)
With this grand canonical partition function
Φ
, we can easily obtain the expected
value of the total number of employed workers
N
,
><
N
by differentiating
Φlog
with respect to
µ
which corresponds to the reservation wage of the unemployed worker.
This can be seen by differentiating (29) and noting the definition of
z
, (30).
>=<=
∂
∂
=Φ
∂
∂
∑
∑
∑
∞
=
∞
=
∞
=
N
Ze
ZNe
Ze
N
N
N
N
N
N
N
N
N
][
1
)]log([
1
]log[
1
0
0
0
βµ
βµ
βµ
β
βµβµβ
. (34)
In the present case,
Φ
is actually given by equation (33). Therefore, we can find
>< N
as follows.
]log[
1
Φ
∂
∂
=
µβ
N
)}1log()1{log(
1
)()()1(
1
jjj
ccf
K
j
ee
−−+
=
−−−
∂
∂
=
∑
µβµβ
µβ
∑
=
−
−
−+
−+
−
−
−
+
=
K
j
c
c
cf
cf
j
j
j
jj
jj
e
e
e
ef
1
)(
)(
)()1(
)()1(
]
11
)1(
[
µβ
µβ
µβ
µβ
(35)
The expected value of the number of workers employed on the job sites with
productivity
j
c
,
><
j
n
is simply the corresponding term in the summation of
>< N
or equation (35).
1
1
)1
(
)
(
)(
)()1(
)()1
(
−
−
−
+
=
−
−
−+
−
+
j
j
j
j
jj
c
c
cf
c
f
j
j
e
e
e
ef
n
µβ
µ
β
µβ
µβ
(36)
Equation (36) determines the distribution of workers across job-sites with different
levels of productivity in our stochastic macro-equilibrium.
Figure 7 shows how this model works. On one hand, there is dynamics of creation
and destruction of potential job sites with various levels of productivity (Figure 7 (a)).
We presented a Markov model which leads us to power-law tail of productivity
distribution in the steady state. At the same time, there is another dynamics of job
matching which corresponds to the standard search theory (Figure 7(b)). Convolution of
two dynamics determines the distribution of workers at job sites with various
productivities. The result is equation (36).
This distribution is fundamentally conditioned by aggregate demand. When the
27
level of aggregate demand is high, it is more likely that high productivity firms make
more job openings. They attract not only the unemployed but also workers currently on
the inferior jobs. As Okun (1973) vividly illustrates, “a high pressure economy provides
people with a chance to climb ladders to better jobs.” And people actually climb ladders
in such circumstances.
A Numerical Example
With the help of a simple numerical example, we can better understand how
equation (36) looks like, and also how the present model works. Figure 8 shows the
distribution of
j
n
given by equation (36). In the figure, the level of productivity
200,1
2001
== cc
are shown horizontally. The number of potential jobs or the ceiling
at each productivity level,
j
f
, is assumed to be constant at 10 for
501
, cc
, while it
declines for
)200,,50( =jc
j
as
j
c
increases. Specifically, for
)200,,50( =jc
j
,
j
f
obeys a power distribution:
2
1~
jj
cf
. This assumption means that low
productivity jobs are potentially abundant whereas high productivity jobs are limited. In
the figure, the number of potential jobs is shown by a dotted line.
In this example, we have two cases; Case A corresponds to high aggregate demand
whereas Case B to low aggregate demand. Specifically,
β
is assumed to be (A) -0.01,
and (B) -0.007. As explained in Table 1, Case (A)
01.0−=
β
corresponds to high
demand
D
whereas Case (B)
007
.0−=
β
to low demand. In both cases, the number of
workers or labor force
L
is assumed to be 680. The number of employed workers
N
is
endogenously determined. The chemical potential
µ
is assumed to be -1.
In Figure 8, we observe that
j
n
increases up to
50=j
, and then declines from
50=j
to 200 in both cases. Broadly, this is the observed pattern of productivity
dispersion among workers (Figure 6). What happens in this model is as follows.
Whenever possible, workers strive to get better jobs offered by firms with higher
productivity. That is why the number of workers
j
n
increases as the level of
productivity rises in the relatively low productivity region. The number of workers
j
n
becomes an increasing function of
j
c
because potential jobs with low productivity are
abundant. Note that the number of potential jobs or the ceiling is not an increasing
function of
j
c
but constant in this region.
The number of workers
j
n
turns to be a decreasing function of productivity
j
c
in the high productivity region simply because the number of potentially available
jobs
j
f
declines as
j
c
rises. Note, however, that
j
n
is not equal to
j
f
;
j
n
is strictly
smaller than
j
f
. The ratio of the number of employed workers to the potential jobs,
jj
fn /
is much higher in the high productivity region than in the low productivity region
28
(Figure 9). Again, this reflects the fact that workers always strive to get better jobs
offered by firms with higher productivity.
In this model, firms are assumed to be monopolistically competitive facing the
downward sloping individual demand curve for their own products. Job offers made by
such firms depend ultimately on the aggregate demand (Negishi (1979) and Solow
(1986)). In Figure 8, two distributions of
j
n
are shown: Case (A) and Case (B). They
depend on high and low levels of aggregate demand
D
, respectively. When aggregate
demand rises, the distribution of workers as a whole goes up. Figure 9 indeed shows
that more workers are employed by firms with higher productivity. Attractiveness of job
is not simply determined by wages. However, to the extent that wages offered by firm
are proportional to the firm’s productivity, the distribution shown in Figure 8 would
correspond to the wage offer distribution function in the standard search literature. It
depends crucially on the level of aggregate demand.
When aggregate demand
D
goes up, the number of employed workers
N
which
corresponds to the area below the distribution curve, increases. Specifically,
N
is 665
in case (A) while it is 613 in case (B). It means that given labor force
680=L
, the
unemployed rate
LNLLU /)(/ −=
declines when aggregate demand
D
rises. In this
example, the unemployment rate is 2.2% in case (A) while it is 9.9% in case (B).
Summary
Let us summarize economics of the present model. In the first place, the number
of potential job sites with various levels of productivity is assumed to be given. They
are determined not only by capital accumulation and technical progress but also by
allocative disturbances to demand. The existing stock of capital and technology only
slowly change, but allocative demand disturbances can swiftly change the economic
values of the job sites associated with those capital and technology. Creative
destructions due to Schumpeterian innovations raise the levels of productivity of some
job sites, but at the same time, lower the levels of productivity of others. We consider a
Markov model to describe dynamics of creations and destructions of potential job sites,
and derive the conditions for the stationary state. The number of job sites with high
productivity in the stationary state turns out to be power distribution. The important
point is that job sites with relatively low productivities are abundant whereas those with
high productivities are limited following power distribution
7
.
7
Houthakker (1955) shows that the aggregate production function becomes Cobb-Douglas on the
assumption that the distribution of productivity (labor and capital coefficients in his model) is the
power distribution: See also Sato (1974).
29
Now, let us visualize a potential job site as a “box” or a “desk” in the case of
white-collar workers. Each job site or box is associated with a particular level of
productivity. It is either empty or occupied by a worker. We can consider that a firm is
nothing but a cluster of many job sites or boxes. It is conceivable that productivity
differs across boxes in a single firm.
To fill a job site or a box with a worker is a firm’s economic decision. To do so,
the firm can either keep the existing workers on the job or post a vacancy sign for
workers searching better jobs. The most important factor constraining the firm’s labor
employment is the level of the product demand which depends ultimately on the level of
aggregate demand.
Because micro shocks and constraints facing firms and workers are so complex
and unspecifiable, we cannot usefully pursue the exact micro behaviors. The single
matching function for the economy as a whole will not do. Wages are obviously
important, but attractiveness of job is not fully determined by wages, but depends on
many factors; workers are interested in tenure, location, fringe benefits and other work
conditions. The relative importance of these factors differs across workers. Likewise,
the type of workers the firm is looking for cannot be simply defined, but depends on
many factors. Again, the relative importance of these factors differs across firms.
Besides, economic conditions facing workers and firms keep changing in unspecifiable
ways. Given such complexity, optimization exercises based on representative agent
assumptions do not help us find the outcome of random matching of workers and firms.
Here comes the method of statistical physics.
The basic assumption of the model is that firms with higher productivity can
afford to make more attractive job offers to workers. However, firms are
monopolistically competitive in that they face the downward-sloping individual demand
curve for their products. Then, their production and employment decisions depend
ultimately on the level of aggregate demand; see Negishi (1979) and Solow (1986). We
never know how aggregate demand is distributed across firms (or job sites) with
different levels of productivity, but when the level of aggregate demand is high, it is
more likely that high productivity firms face high enough demand, and as a
consequence, they keep more workers on the job and post more vacancy signs to attract
good workers. Workers are certainly aware of it; we know that quit rates go up in booms
and down in recessions. As a result of such actions of both firms and workers, the
distribution of productivity tilts to the direction of high productivity (Figure 8). As
Okun (1973) argues, when aggregate demand rises, workers on the job “climb ladders to
better jobs” without experiencing any spell of unemployment. At the same time, more
30
workers currently in the unemployment pool find acceptable jobs.
Employment
N
increases, and the unemployment rate declines.
5. The Principle of Effective Demand
Keynes (1936) argued that the aggregate demand determines the level of output in
the economy as a whole. Factor endowment and technology may set a ceiling on
aggregate output, but the actual level of output is effectively determined by aggregate
demand causing endogenous changes in utilization of production factors.
Modern macroeconomics is ready to regard factor endowment, technology, and
preferences as exogenous in the short-run, but rejects the idea that even a part of
demand is exogenous. Keynes did not regard all the demand exogenous, of course.
Plainly, a significant part of demand is endogenously created by production because
production makes more agents get greater purchasing power in the economy. This is
indeed the basic idea behind consumption function. Keynes’ principle of effective
demand regards a part of aggregate demand as exogenous, that is, almost impossible to
explain within the framework. The list of exogenous changes in aggregate demand
includes major technical change, changes in exports, financial crisis, and changes in
fiscal policy. Such exogenous changes in demand are a fundamental determinant of
changes of aggregate production; Keynes himself emphasized the volatility of
investment which is in turn caused by volatility of “long-term expectations” (Keynes
(1936, Chapter 12)).
Our economy really experiences occasional aggregate demand shocks. Figure 10
shows indices of exports and industrial production of Japan during the post-Lehman
“great recession”. It is most reasonable to regard a sudden fall of exports as exogenous
real demand shocks to the Japanese economy. Figure 10 demonstrates that the principle
of effective demand is alive and well! (See Iyetomi et. al. (2011))
When output changes responding to changes in aggregate demand, the level of
utilization of production factors must change correspondingly. An example is cyclical
changes of capacity utilization of capital. Unemployment of labor is another. In fact,
“involuntary” unemployment has been long taken as the symbol of the Keynesian
demand deficiency. However, unemployment is by definition job search, and, therefore,
the definition of “involuntary” unemployment is necessarily ambiguous. Our model
demonstrates that not only unemployment but also the distribution of labor productivity
across firms and job sites changes responding to changes in aggregate demand.
As we noted above, the post-Lehman “great recession” provides us with an
excellent example of the negative aggregate demand shock. The unemployment rate
31
certainly rose. The relatively low and cyclically insensitive Japan’s unemployment rate
was 3.6 percent as of July 2007, but after the global financial crisis, it rose to 5.5
percent for July 2009. Unemployment is not the whole story, however.
Figure 11 (a), (b), (c) compare the distributions of productivity before and after the
Lehman crisis, namely 2007 and 2009; (a) total, (b) manufacturing sector, and (c)
non-manufacturing sector. As our theory indicates, the distribution as a whole, in fact,
tilts toward lower productivity in sever recession. Figure 11 shows that the tilt of the
distribution toward low productivity is more conspicuous for the manufacturing
industry than for the non-manufacturing industry. It is due to the fact that in Japan, the
2009 recession after the bankruptcy of the Lehman Brothers was basically caused by a
fall of exports (Figure 10), and that exports consist mainly of manufactured products
such as cars. We can observe, however, that the distribution tilts toward low
productivity for the non-manufacturing industry as well, particularly in the high
productivity region. This is, of course, due to the fact that a fall of demand in the
manufacturing sector spills over to the non-manufacturing sector.
Our analysis clarifies the mechanism by which labor is reallocated responding to
changes in aggregate demand, and accordingly total output changes. It provides a
micro-foundation for the principle of effective demand.
6. Concluding Remarks
It is a cliché that the Keynesian problem of unemployment and under-utilization
of production factors arises because prices and wages are inflexible. Tobin (1993), in
fact, Keynes (1936) himself argued that the principle of effective demand holds true
regardless of flexibility of prices and wages.
The natural micro picture underneath the Keynesian economics is monopolistic
competition of firms facing the downward sloping individual demand curve, not perfect
competition in the product market. Negishi (1979)’s model of general equilibrium of
monopolistically competitive firms with the kinked individual demand curve provides a
nice microfoundation for what Tobin (1993) called the Old Keynesian view in which
“quantities determine quantities.” This model, however, abstracts itself from frictions
and uncertainty present in the labor market.
The standard equilibrium search theory has filled a gap by explicitly considering
frictions and matching costs in the labor market. While acknowledging the achievement
of standard search theory, we pointed out two fundamental problems with the theory.
First, the assumption that the job arrival rate, the job separation rate, and the probability
distribution of wages (more generally, some measure of the desirability of the jobs) are
32
common to all the workers and firms is simply untenable. There is always the
distribution of wages in the economy as a whole, of course. However, it is plainly not
relevant distribution facing each worker when he/she makes economic decisions. Each
worker faces different job arrival rate, job separation rate and probability distribution of
wages. He/she is even interested in different variables. Thus, he/she acts in his/her own
“universe”. The same is true for individual firm. It is, in fact, frictions and uncertainty
emphasized by the equilibrium search theory that makes the economy-wide distribution
or the average irrelevant to economic decisions made by individual economic agent. In
this respect, the labor market is fundamentally different from a local retail market.
Secondly, we maintain that the standard assumption that the product market is
perfectly competitive in the sense that the firm’s individual demand curve is flat is
unrealistic. It is particularly ill-suited for studying cyclical changes in effective
utilization rate of labor. To consider cyclical changes in unemployment without demand
constraint is to play Hamlet without a prince of Denmark.
Under the assumption of perfect competition, cyclical changes are identified with
changes in the average productivity (See, for example, Barlevy (2002), Shimer (2005)
and Hagedorn and Manovskii (2008)). This is essentially the real business cycle (RBC)
theory (Kydland and Prescott (1982)). Shimer (2005)’s comparative static analysis
however, in effect, demonstrates that a change in labor productivity a la RBC cannot
reasonably explain the empirically observed magnitude of fluctuations of the
unemployment and vacancy rates
8
. There is, in fact, a strong case that cyclical changes
are caused by demand shocks rather than productivity shocks (Mankiw (1989) and
Summers (1986)).
We assume that instead of perfect competition, firms face the down-ward sloping
individual demand curve. However, we never know how the aggregate demand is
distributed across firms or job sites with different levels of productivity. We are unable
to know micro behaviors of workers and firms, either. To eschew pursuing such
complex behavior of micro unit for understanding macro system is the basic principle of
statistical physics. This paper presented a model of stochastic macro-equilibrium based
on this principle. The concept of stochastic macro-equilibrium is motivated by the
presence of all kinds of unspecifiable micro shocks. At first, one might think that
allowing all kinds of unspecifiable micro shocks leaves so many degrees of freedom
that almost anything can happen. However, the methods of statistical physics ― the
8
The present paper shows that a change in the labor productivity and a change in aggregate demand
are different things. The Keynesian model with demand deficiency can reconcile the magnitudes of
fluctuations of labor productivity on one hand and unemployment and vacancy on the other.
33
maximization of entropy under macro-constraints ― actually provide us with the
quantitative prediction about the equilibrium distribution of productivity, namely
equation (36).
It is extremely important to recognize that the present approach does not regard
behaviors of workers and firms as random. They certainly maximize their objective
functions perhaps dynamically in their respective stochastic environments. The
maximization of entropy under the aggregate demand constraint (10), in fact, balances
two forces. On one hand, whenever possible, workers are assumed to move to better
jobs which are identified with job sites with higher productivity. It is the outcome of
successful job matching resulting from the worker’s search and the firm’s recruitment.
When the level of aggregate demand is high, this force dominates. However, as the
aggregate demand gets lower, the number of possible allocations consistent with the
level of aggregate demand increases. Randomness which plays a crucial role in our
analysis basically comes from the fact that the distribution of demand constraints in the
product market across firms with different productivity, and the optimizing behaviors of
workers and firms under such constraints are so complex and unspecifiable that those of
us who analyze the macroeconomy must take micro behaviors as random.
When the level of aggregate demand is high, it is most likely that high
productivity firms keep more workers on the job, and put more vacancy signs than in
the period of low demand. Workers are certainly aware of such a change. It is
demonstrated by the fact that quit rates are higher in high-demand periods. In this way,
whether or not better jobs are really offered and workers move to those jobs depends
ultimately on the level of aggregate demand. Our analysis demonstrates that the most
probable outcome of random matching of firms and workers is given by equation (36)
which depends on aggregate demand. It broadly coincides with the empirically observed
distribution of productivity. We emphasize that friction and uncertainty are not
exogenously given, but depend crucially on the aggregate demand. The entropy
maximization plays the role of matching function in standard search theory.
Keynes’ theory has been long debated in terms of unemployment or “involuntary”
unemployment. Though unemployment is one of the most important economic
problems in any society, to focus only on unemployment is inadequate for the purpose
of providing micro-foundations for the Keynesian economics. The real issue is whether
or not there is any room for mobilizing labor to high productivity jobs, firms, or sectors.
The famous Okun’s law demonstrates that there is always such a room in the economy
(Okun (1963)); See Syverson (2011) on more recent research on productivity dispersion.
Based on the method of statistical physics, the present paper quantitatively shows how
34
labor is mobilized when the aggregate demand rises. The level of aggregate demand is
the ultimate factor conditioning the outcome of random matching of monopolistically
competitive firms and workers. By so doing, it changes not only unemployment but also
the distribution of monopolistically competitive productivity, and as a consequence the
level of aggregate production.
35
Table 1: Negative
β
and Aggregate Demand
β
0-
……
-∞
Aggregate Demand D
Low
……
High
36
Figure 1: The Kinked Individual Demand Curve
Facing Monopolistically Competitive Firm: Negishi (1979)
p
q
０
A
B
C
D
E
F
M M’
P
p
q
０
A
B
C
D
E
F
M M’
P
37
Figure 2: All the Workers Work at Job sites with the Highest Level of
Productivity under Extremely High Aggregate Demand,
max
D
The Number of
Employed Workers
Productivity
ｃ
1
ｃ
2
・・・・・・ ｃ
K
０
ｎ
The Number of
Employed Workers
Productivity
ｃ
1
ｃ
2
・・・・・・ ｃ
K
０
ｎ
38
Figure 3: Workers are Distributed Evenly across All the Sectors with
Different Levels of Productivity under Extremely Low
Aggregate Demand,
min
D
The Number of
Employed Workers
Productivity
ｃ
1
ｃ
2
・・・・・・ ｃ
K
０
ｎ
N/K
The Number of
Employed Workers
Productivity
ｃ
1
ｃ
2
・・・・・・ ｃ
K
０
ｎ
The Number of
Employed Workers
Productivity
ｃ
1
ｃ
2
・・・・・・ ｃ
K
０
ｎ
N/K
39
Figure 4: Entropy S and Aggregate Demand D
40
Figure 5: Distribution of Labor across Sectors with Different Productivity
― : β＝ - 1/5000 --- : β＝ - 1/7500
41
Figure 6: Distribution of Labor Productivity in Japan (2007)
(a)
Mean= 1.29×10
4
, Standard Deviation=2.37×10
4
(b)
Mean= 1.30×10
4
, Standard Deviation=2.21×10
4
42
Figure 6: Distribution of Labor Productivity in Japan (2007)
(c)
Mean= 1.28×10
4
, Standard Deviation=2.48×10
4
43
Figure 7: Model of Stochastic Macro-equilibrium
(a) Dynamics of Creation and destruction of Potential Jobs
Note: Both productivity and the number of potential job sites are in the natural logarithm. The
straight line as drawn in the figure means that the distribution of productivity is
power-law.
(b) Stochastic Macro-equilibrium
The number of potential
job sites (in log)
Productivity (in log)
The number of potential
job sites (in log)
Productivity (in log)
The Number of
Employed Workers
The Level of Productivity
Aggregate Demand D or
β
Pool of Unemployment
Reservation
wages μ
The Number of
Employed Workers
The Level of Productivity
Aggregate Demand D or β
Pool of Unemployment
Reservation
wages μ
44
Figure 8: Distribution of labor Productivity
Note: (A) High Aggregate Demand, (B) Low Aggregate Demnd
See the main text for details.
The number of
employed workers
(A)
(B)
45
Figure 9: Percentage of Potential Job Sites Occupied by Employed Workers
Note: (A) High Aggregate Demand, (B) Low Aggregate Demnd
See the main text for details.
Ratio of employed workers
to potential job sites
(A)
(B)
46
Figure 10: Indices of exports and the Industrial Production,
normalized to 100 for 2005
Source: Cabinet Office, Ministry of Economy, Trade and Industry
60
70
80
90
100
110
120
130
2006 2007 2008
2009
(2005=100)
Exports
Industrial Production
47
Figure 11:
(a)
(b)
48
Figure 11:
(c)
49
References
Aoki, M. and H. Yoshikawa, (2007), Reconstructing Macroeconomics: A Perspective
from Statistical Physics and Combinatorial Stochastic Processes, Cambridge,
U.S.A., Cambridge University Press.
Aoyama, H., H. Yoshikawa, H. Iyetomi, and Y. Fujiwara, (2010), “ Productivity
Dispersion: Facts, Theory, and Implication”, Journal of Economic Interaction and
Coordination, Vol. 5, pp. 27-54.
Arrow, K. and G. Debreu (1954), “Existence of an Equilibrium for a Competitive
Economy,” Econometrica, Vol.22, No.3, pp.265-290.
Asplund M. and V. Nocke (2006), “Firm Turnover in Imperfectly Competitive
Markets,” Review of Economic Studies, Vol.73, pp295-327.
Barlevy (2002),
Burdett K. and D. T. Mortensen (1998), “Wage Differentials, Employer Size, and
Unemployment,” International Economic Review, Vol.39, No.2, pp257-273.
Champernowne, D. (1953), “A model of Income Distribution,” Economic Journal,
Vol.83, pp.318-351.
Davis, S. J., J. C. Haltiwanger and S. Schuh (1996) Job Creation and Destruction,
Cambridge, Mass.: MIT Press.
DelliGatti, D., E. Gaffero, M.Gallegatti, G. Giulioni and A. Palestini (2008), Emergent
Macroeconomics: An Agent-Based Approach to Business Fluctuations, Milano:
Springer.
Diamond, P. (1982), “Aggregate Demand Management in Search Equilibrium,” Journal
of Political Economy, Vol.90, pp.881-894.
——— (2011), “Unemployment, Vacancies, Wages,” American Economic Review
Vol.75, pp.63?-655.
Fay, J.A. and J.L. Medoff (1985), “Labor and Output over the Busines Cycles: Some
Direct Evidence,” American Economic Review Vol.101, No.4, pp.1045-1072.
Feynman, R. P., (1964), “The Principle of Least Action,” Chapter 19 of his Lectures on
Physics Vol.II, Reading MA: Addison Wesley.
Foley, D. (1994), “A Statistical Equilibrium Theory of Markets,” Journal of Economic
Theory, Vol.62, pp.321-345.
Garibaldi U. and E. Scalas (2010), Finitary Probabilitic Methods in Econophysics,
Cambridge: Cambridge University Press.
Hagedorn M. and I. Manovskii (2008) “The Cyclical Behavior of Equilibrium
Unemployment and Vacancies Revisited,” American Economic Review, Vol. 98,
50
No.4, pp. 1692-1706.
Hicks, J. R. (1936), “Mr. Keynes’s Theory of Employment.” Economic Journal 46
(182): pp.238–253.
Houthakker H., (1955/56), “The Pareto Distribution and the Cobb-Douglas Production
Function in Activity Analysis”, Review of Economic Studies, Vol.23, No.1,
pp.27-31.
Ijiri Y. and H.A. Simon, (1975) “Some distributions associated with Bose-Einstein
statistics”, Proceedings of the National Academy of Sciences of the United States
of America, 72(5): pp.1654-1657.
——— and ———, (1979), Skew Distribution and the Size of Busines Firms,
Amsterdam: North-Holland.
Ikeda Y. and W. Souma, (2009) “International Comparison of Labor Productivity
Distribution for Manufacturing and Non-manufacturing Firms”, Progress of
Theoretical Physics, Supplement 179, pp.93-102.
Iyetomi, H. (2012), “Labor Productivity Distribution with Negative Temperature”,
Progress of Theoretical Physics, Supplement, No.194, pp.135-143.
———, Y. Nakayama, H. Yoshikawa, H. Aoyama, Y. Fujiwara, Y. Ikeda and W. Souma
(2011), “What Causes Business Cycles? Analysis of the Japanese Iindustrial
Production Data,” Journal of The Japanese and International Economies, Vol.25,
pp.246-272.
Keynes, J. M. (1936), General Theory of Employment, Interest, and Money, London:
Macmillan.
Kirman, A.P. (1992) “Whom or What Does the Representative Individual Represent?,”
Journal of Economic Perspectives, Spring Vol.6, No.2, pp.117-136.
Kittel, C and H. Kroemer, (1980), Thermal Physics, Second ed., San Francisco: W H
Freeman.
Kydland, F. and E. Prescott, (1982), “Time to Build and Aggregate Fluctuation,”
Econometrica, November.
Marsil M. and Y. C. Zhang, (1998), “Interacting Individuals Leading to Zipf’s Law,”
Physical Review Letters, 80, No.12, pp.2741-2744.
Mankiw N. G. (1989), “Real Business Cycles: A New Keynesian Perspective,” Journal
of Perspectives, Vol.4, No.3, 79-90.
——— and D. Romer (1991), New Keynesian Economics, Vol.1 and 2, Cambridge: The
MIT Press.
Modigliani, F., (1944), “Liquidity Preference and the Theory of Interest and Money,”
Econometrica Vol.12, No.1, pp.45-88.
51
Mortensen, D.T., (2011), “Markets with Search Friction and the DMP Model,” The
American Economic Review Vol.101, No.4, pp.1073-1091.
Negishi, T. (1960-61), “Monopolistic Competition and General Equilibrium,” Review of
Economic Studies, Vol.28, pp.196-201.
——— (1979), Microeconomic Foundation of Keynesian Macroeconomics,
Amsterdam : North-Holland Publishing.
Okun, A (1963), “Potential GNP: Its Measurement and Significance,” Proceedings of
the Business and Economic Statistics Section, American Statistical Association,
98-104, reprinted in Okun’s The Political Economy of Prosperity, Washington
D.C.: The Brookings Institution, 1970.
——— (1973) “Upward Mobility in a High-pressure Economy,” Brooking Papers on
Economic Activity, Vol. 1, pp.207-261.
Pissarides, C.A., (2000), Equilibrium Unemployment Theory, second edition,
Cambridge: The MIT Press.
――― (2011) “Equilibrium in the Labor Market with Search Frictions,” The American
Economic Review Vol.101, No.4, pp.1092-1105.
Postel-Vinay, Fabien and Jean-Marc Robin, (2002), “Equilibrium Wage Dispersion with
Worker and Employer Heterogeneity,” Econometrica Vol.70, No.6, pp.2295-2350.
Ramsey, N. F., (1956), “Thermodynamics and Statistical Mechanics at Negative
Absolute Temperatures,” Physical Review, 103, No.1, pp.20-28.
Rogerson, Richard, Robert Shimer, and Randall Wright, (2005), “Search Theoretic
Models of the Labor Market: A Survey,” Journal of Economic Literature,
Vol.XLIII, December, pp.959-988.
Sato, Kazuo, (1974), Production Functions and Aggregation, Amsterdam:
North-Holland.
Shimer R. (2005) “The Cyclical Behavior of Equilibrium Unemployment and
Vacancies,” American Economic Review, Vol. 95, No.1, pp. 25-59.
Solow R. M. (1986) “Unemployment: Getting the Questions Right,” Economica, Vol.
53, Supplement, pp. S23-34.
Summers, L.H. (1986) “Some Skeptical Observation on Real Business Cycle Theory,”
Federal Reserve Bank of Minneapolis Quarterly Review, Vol. 10, 23-27.
Syverson, C. (2011) "What determines productivity?" Journal of Economic Litereture,
Vol.49, No.2, pp.326-365
Tobin, J. (1972), “Inflation and Unemployment,” American Economic Review. Vol. 62,
pp. 1-18.
――― (1993) “Price Flexibility and Output Stability: An Old Keynesian View,”
52
Journal of Economic Perspectives, Vol.7, pp.45-65.
Yoshikawa, H., (2003), “The Role of Demand in Macroeconomics,” Japanese
Economic Review, 54, (1), pp.1-27.
