Is There a Capital Shortage in Europe? By
Michael C. Burda
Contents: I. I n t r o d u c t i o n . - II. T h e C a p i t a l - S h o r t a g e H y p o t h e s i s : T h e Facts. III, A M o d e l of t h e Capital Shortage. - IV. C o m p a r a t i v e D y n a m i c s E x p e r i m e n t s : Terms-of-Trade Shocks. - V. Empirical Evidence. - VI. C o n c l u s i o n .
I. Introduction
t is striking to most economists that European unemployment persists at high levels, despite the return of higher rates of per capita growth and capacity utilization. The reluctance of European policymakers to pursue aggressively expansive macroeconomic policies arguably corroborates the view that the relationship between growth and unemployment indicators has broken down, and that a large portion of employment reflects a "structural problem" rather than deficient demand. The substantial decline in product wages relative to productivity since 1980, however, casts considerable doubt on the simple view that labor costs alone can explain the current behavior of employment. A more robust variant of the structural account of European unemployment stresses shifts of - rather than movements along - the labor demand curve through the cumulative effects of investment shifts I. More specifically, deviations of the capital stock from its trend path can also explain reduced demand for labor and higher rates of unemployment. Trivially, in a price-taking economy with constant returns and classical (i.e., real wage-induced) unemployment, excessive product wages can always be offset by sufficiently augmenting the stock of physical capital. Conversely, a reduction in the capital stock can induce classical unemployment by reducing the demand for labor at any given product wage. It follows that adverse changes in the determinants of investment can exacerbate existing classical unemployment for any level of real wages. While the possibility of a classical capital or capacity shortage in Europe has been recognized in principle, there exist to date few formal models using
I
Remark: This paper is a revised version of Chapter 4 of my Ph.D. dissertation. It was written in part while visiting the Institut f/ir Weltwirtschaft, Kiel. I thank Olivier Blanchard, Henning Klodt, Warwick McKibbin, and Sabine Miltner for useful comments, the Institut fiir Weltwirtschaft for the generous use of its facilities, and the Deutscher Akademischer Austauschdienst for financial support.
See Giersch [1981; 1983], Malinvaud [1982], and Bruno and Sachs [1982; 1985].
Capital Shortage
39
modern theories of aggregate supply2. In this paper we investigate the capital shortage hypothesis in Western Europe. Firstly, we identify the existence of a capital gap in the European manufacturing sector. Secondly, we formalize these ideas in a model using conventional models of investment and the wage setting behavior of labor unions. Because scale effects predominate, the model predicts declining product wages and employment in the aftermath of a deterioration in the terms of trade. Finally, we examine some empirical evidence that supports the model's relevance for the West European economies.
II. The Capital-Shortage Hypothesis: The Facts That the data are at least mildly supportive of the capacity shortage hypothesis is indicated by the most recent Centre for European Policy Studies (CEPS) report, which focused on the role of capital formation in European unemployment (see Modigliani et al. [1986]). Firstly, gross capital stocks as computed by EUROSTAT and the OECD deviate significantly from pre-1973 trends, as depicted in the panels of Figure 1. Secondly, these deviations exhibit substantial variation across countries. The "capital gap", which we measure as the deviation of the real stock of fixed capital in manufacturing from a trend estimated through 1973, currently ranges from 53 percent in ER. Germany and 34 percent in Italy to less than 14 percent in Sweden, Austria, and the U.S. These "capital gaps" seem to be associated across countries with increases in the NAIRU; for the seven countries, the simple correlation of the capital gap in 1983 with changes in the average OECD standardized unemployment rate from 1970-1974 to 1980-1984 is 0.38. Clearly, the capital stock is only one facet of capacity, which is most naturally defined as the desired level of output of competitive firms at a given constellation of factor and output prices. In a constant returns world, capacity is conditioned on both the scale of the economy, summarized by the amount of fixed capital available for production, and relative factor and output prices. Yet conventional measures of capacity - estimated as the quotient of observed real output and aggregate capacity utilization rates - also exhibit deviations from earlier trend, ruling out a spurious slowdown in the growth of capital stock attributable to technical progress 3. Many researchers, especially in the U.S., have summarily dismissed the relevance of the capital constraint in Europe. In a recent paper, Blanchard and Summers [1986] argue that shifts in productive capacity should not be con2 Malinvaud [1982] argues that Keynesian unemployment can become classical unemployment through the channel of investment and Bruno and Sachs [1982; 1985] have formalized these effects in a dynamic model. a For the countries with percentage capacity utilization measures the corresponding "gaps" were in 1985: F.R. Germany, 44 percent; France, 51.3 percent; U.S., 21 percent. These numbers are corroborated by Modigliani et al. [1986, especially Figure 3, p. 15].
40
M i c h a e l C. B u r d a
Figure 1 [o 12.5
Log Real Gross Capital Stocks in Manufacturing (...) and 1960-1973 Trends (-)
-
log United Kingdom
122.9 %
12.2!
France
12.0
14.0
11.75-
13.5
11.5 ' ' ' I . . . . I . . . . 1/-~5- FR Germany
I ....
I ....
130
I
'
'
'
I
. . . .
I
'
''
'I
. . . .
I
'
''
'
I
...~
12.5
~0-
3~..6
13.0-
11.5 ' ' ' I ....
I ....
I''
' ' f ' ' ''1
n
>,6 ./.
15.0-
13.0-
'
'
'
r
'
'
r
'
I
'
t
'
'
[
. . . .
F
' '
'
S
'
I
}13.7"/o ~176
12.75 14.5-
1Z5
-
14.0-
12.25 ....
7.5-I 7.25t
o/,,
12.0-
13.5-
12.5
~ 28.6%
14.5
i
....
I ....
I
....
I
....
Uni t e d S t a t e s
o
~ } .. .
7 0 6 6 . 5 6 ~ i 1960
65
70
, , , , r , , , , l
I
1960
65
. . . .
70
i , , , , i
75
. . . .
80
l
85
,3.~'/o
""', , . . . . i 75
80
85
sidered differently from other supply shocks, which an economy with Hicks or Harrod neutral technical progress should be expected to outgrow. Moreover, they argue that despite "considerable disinvestment" during the Great Depres-
41
Capital Shortage
sion in the United States, the capacity constraint was nonbinding as the war-related recovery took hold. There are several difficulties with applying the U.S. experience in the 1930s to Europe in the 1980s. The first is, as Table 1 indicates, that European manufacturing is currently operating at historically high levels of capacity utilization, which is inconsistent with the "Depression" analogy, and indeed, the U.S. experience. Of all countries in Table 1, only U.S. and French manufacturing are currently operating at levels below their 1979 peaks. While it may be claimed that this capacity could be more extensively used, presumably through extra shifts [e.g. Franz, 1986], the "appropriate" rates of capacity utilization are probably best inferred from the historical behavior of these series 4. Secondly, the roles of trade unions and collective bargaining across the two regimes are hardly comparable. Membership in U.S. trade unions in 1930 was approximately 3.2 million or 9.3 percent of the total nonagricultural workforce [Bailey, 1983]; collective bargaining in Europe determines wages directly or indirectly for over 75 percent of all dependent status employees [OECD, 1979]. In the next section we present a suggestive account of a capital shortage induced by the interaction of wage setting and investment behavior as an alternative explanation of high levels of involuntary unemployment. It is a model of equilibrium from the viewpoint of the firm, or of employment at desired levels of capacity, and thus abstracts from nominal contracts or price stickiness. These phenomena, while important for business cycle fluctuations, are a priori less relevant for low frequency phenomena like the rise in the natural rate of unemployment. Table 1 - OECD Capacity Utilization Rates a
United Kingdom France ER. Germany Italy Sweden United States
1975-1979
1979
1985
1986
32.6 78.8 80.5 73.1 33.9 86.0
26.3 82.0 84.7 74.5 43.0 81.0
47.8 79.4 84.3 73.8 49.3 80.3
48.0 79.1 84.7 75.3 45.8 79.8
a Either percentage of firms operating at "full" capacity (U.K., I, S), or average capacity utilization rate (F, D, U.S.). Manufacturing industry only. Source: OECD [var. issues].
4 Recall that modern theories of monopolistic competition, which are often invoked in other contexts, predict excess capacity even when the individual firm is in equilibrium.
42
Michael C. B u r d a
Two features characterize our model. Firstly, price-taking, sales-unconstrained firms face installation costs of investment [Abel, 1982; Summers, 1981; Hayashi, 1982] which fixes the steady-state capital stock under constant returns in production. The second ingredient is a simple account of monopolistic union wage setting along the lines of Dunlop [1944] or de Menil [1971], in which unions maximize their surplus over the nonunion alternative wage, with recognition of the impact their wage policy may have on the latter. Individual labor suppliers may be constrained in the amount of labor that they supply, but the unions themselves are not. For their part, unions have no money illusion and are able to set the real wage in terms of consumer goods. This formulation also incorporates a sensitivity of the union to employment, in contrast to many models of its genre 5. III. A M o d e l of t h e C a p i t a l S h o r t a g e 1. The F i r m Our stylized economy consists of a single sector which sells output (value added) in a competitive product market. The representative firm employs capital K and labor L in a constant returns to scale production function: Vt
=
F(I~,Lt) = Ktf(Lt/Kt)
where f is the "intensive" form. The function F is assumed to possess the usual nice properties s. The representative firm, which borrows and lends at constant instantaneous real rate r, hires labor at (product) wage wt measured in terms of the output good. Firms invest at rate I t per instant with corresponding cash outflow of C(It/Kt)Kt, where c' > 0 and c" > 0. The convexity of the installation cost function provides the inducement to invest at a finite rate 7. The evolution of the capital stock is given by the transition equation I~ = I t - 6K t
(1)
We now characterize the optimal investment policy of the firm. The managers have subjective certainty over the path of product wages wt which they take as parametric. They choose investment and labor demand functions See McDonald and Solow [1981] or Oswald [1982a; 1982b]. Models in which membership matters for wage setting [Blanchard and Summers, 1986; Gottfries and Horn, 1986; Burda, 1987] generally include extrinsic uncertainty over the position of the labor demand curve, or allow for employment effects in a Nash bargaining framework. s Namely, FK, FL > 0; FKK,FLL< 0; FKL> 0, plus the Inada conditions: FL(K,0) = FK(0,L) = oofor positive K and L. 7 This feature could in principle be captured by two-sector models with decreasing returns to each, or in a representative agent model in which the investment process requires time [Kydland and Prescott, 1982].
43
Capital Shortage It
and Lt to maximize net discounted real cash flow at time t: 7 e-*-t)[F(K~, L~)-W~Ls-C(IJK~)K~]ds t
subject to Kt given and the transition equation (1). We solve the problem using Pontryagin's maximum principle. The Hamiltonian at instant s>t is H, = e-rCs-t)[F(K~,L,)-wsL,-c(Is/K,)K~ + qs (I~ - 5K,)] where we have redefined q~, the Lagrange multiplier function at each instant, to equal its undiscounted value. The necessary first-order conditions for a maximum are 8 OHJOI, = 0
=>
c' (IJK,) = q,
(2)
8H~/OL~ = 0
=>
f' (LJI~)= w~
(3)
8H~/SK~ = - d (e-rts-t)qs)/ds => FK - c + (IJI~) c ' = - dl, + (r+6) qs
(4)
and the transversality condition lim e- (r+~)t~-t)qt+~= 0
(5)
s~
As a differential equation in q, (4) may be integrated forward to any time s from some initial condition qt. The solution is qt+s = qt e(r+8) (s-t)_ e(r+~)(s-t)
7 e-(r+*) Iv-t) [FK _ c +
(I/K)vc']
dv
t
The transversality condition (5) chooses from this infinity of solutions the one consistent with profit maximization and nonnegativity of the capital stock 9. Multiplication of the above by e -(s-t)(r+~)and imposing (5) yields qt = 7 e -cr+~)(v-t) [FK(K~' Lv) _ c + (IJI(,) c'] dv
(6)
t
The conventional interpretation of q in (6) is the shadow value of an additional unit of capital in place. A familiar result of Hayashi [1982] is that under perfect competition in product markets and constant returns, q will equal the average shadow value of the capital stock or the value of the stock market. The optimal rate of investment and demand for labor are found by inverting (2) and (3), respectively: It = q/(qt)Kt
(7)
Lt = ~0(wt)K t
(8)
a See Kamien and Schwarz [1981] for an exposition of the maximum principle. 9 Profit maximization guarantees that the "cake be eaten" (here the capital stock); the Inada conditions on F bound the capital stock away from zero.
44
Michael C. Burda
so (6) may be rewritten as qt = ~ e-~r§ ~v-t~[FK (1, q0(wv)) - C + ~uC']dv
(9)
t
recalling that by homogeneity, F~(XK,XL)= FK(K,L) for X> 0. Note that current q depends on anticipated future q and wages. 2. Wage S e t t i n g The real product wage faced by the representative firm is set by a single union empowered to represent the interests of its membership, assumed constant at M. To abstract from insider/outsider effects we assume that membership equals the labor force, N. ~~ Members consume the produced good and an imported good, and have Cobb-Douglas preferences over both such that the following price index may be defined in terms of the produced good: p~ = y + ( 1 - y ) T t
where 0 < y < 1 and T is the price ratio of the imported good to the produced good; T may be considered roughly as the inverse of the terms of trade. The consumption wage, then, may be defined as W~t = w t / p ~ = Wt/[V 4r
(1-y)Tt]
(10)
At this juncture most union models posit a utility function for the representative member and a rule governing the allocation of employment among its members 11. It may be argued, however, that the union should provide unemployment insurance to those that do not have work, and the unions decisionmaking mechanism becomes a critical aspect of the model. It seems natural to employ the approach of Dunlop [1944] and de Menil [1971]: if rents may be distributed in arbitrary ways among the membership, the optimal union wage policy is separable from the allocation of the rents among members. We assume that the best alternative available to union members is either an unemployment benefit or work in the nonunion sector which we write as b(U), with b' < 0 and b" > 0, where U is the unemployment rate 1-L/N. Higher levels of unemployment are associated with lower opportunity wages, albeit at a decreasing rate ~2. We formulate the unions problem, then, as lo Clearly the model could be modified in a straightforward way to allow the dependence of M on N, or model the effects of deviations of M from N. i1 McDonald and Solow [1981], Oswald [1982a; 1982b], and Blanchard and Summers [1986] are examples of hiring hall union models in which all members have identical probability of employment; Blair and Crawford [1984], Grossman [1983], and Burda [1987] consider unions that allocate employment on the basis of ranking in a seniority queue. 2 Empirical studies indicate that there is substantial feedback from unemployment to wages in the OECD economies, probably reflecting the former's effect on the aggregate nonunion alternative wage or the unemployment benefit. See Newell and Symons [1985].
45
Capital Shortage
max [w~t-b(Ut)]L t or given rt max [wt(y+ (1-y)Tt)-b(Ut) ] Lt Wt
subject to the labor demand curve. The first-order conditions are: (y+(1-y)T t +
b'L'/Nt)
L t = - (wt[ u + (1-y)Tt] -
b)L'
or, rearranging, 1In - (1-Ut)b'/w~t = (w~t- b)/w~t
(11)
where n is the elasticity of labor demand with respect to the product wage. The standard "monopolist" interpretation applies here: the union wage optimal markup over b is inversely related to the elasticity of labor demand. In addition, the union recognizes its own effect on the level of the opportunity wage through the unemployment channel, and as long as b' < 0 will exploit it in wage setting. If n is roughly constant, (11) defines a "wage offer curve ''13 which traces out optimal w for combinations of K, N, and r: W t = O(Kt,Nt,Tt)
(12)
To determine the signs of OK, ON, and OT, the local derivatives of the implicit function (12), one can totally differentiate (11) and solve respectively for dw/dK, dw/dN, and dw/dT: -b'to/N + (- b"L/N + b') (8U/aK) >0 OK = (1 - n-') (V + (1 - y)'rt) + b'L'/N + (b"L/N-b') (SU/aw) b'L/N 2 + (- b"L/N + b') (8U/SN) ON = (1 -
n -1)
(y + (1 - y)rt) + b'L'/N + (b"L/N-b') (8U/Sw)
<0
(1 - n -1) (1-y)w
QT=
(1 - n -i) (V + (1 - y)Tt) + b'L'/N + (b"L/N-b') (au/aw)
>0
where the derivatives are evaluated at the optimum. They imply that increases in the capital stock and the price of imports increase, whereby increases in the labor force reduce, union wage demands.
~a The term "wage offer curve" is borrowed from Sachs [1986].
46
M i c h a e l C. B u r d a
3. E q u i l i b r i u m We now consider equilibrium capital stock and shadow price of installed capital q implied by the factor demands when wages are determined as above. Analysis of the economy's equilibrium is facilitated by a phase diagram in (q,K) space, which we display in Figure 2. The equations of motion for the system are: eli = -FK(1,cP(Q(I~, NtQ)) + c - c'u (qt) + (r+5)qt
(14)
and -- u (qt)Kt - 5I~
(15)
Figure 2 - The Phase Diagram
K~
K
The/1 = 0 locus is the first-order condition (4) with the wage offer curve O substituted in and represents the set of (q,K) combinations for which the shadow price of capital is constant. In most q-models with constant returns to scale, this locus is flat, i.e., the capital stock K has no impact on its marginal product FK as long as the other factor can be varied continuously in an optimal fashion 14. In the present model, however, the Cl = 0 locus for the aggregate economy is downward sloping. Ceteris paribus, aggregate increases in the capital stock reduce unemployment, raise the opportunity wage available in the economy and thereby the wage demands of unions. The subsequent negative reaction of labor demand to the higher wage leads to lower employment and marginal product of capital (F~>0). In order that Cl = 0 continue to hold, a lower level of q is required. Formally, differentiate (14), holding Cl = 0: ~' In Abel [1982], for example, profits of the firm are dependent the demand for labor is suppressed.
on
the capital stock only, and
Capital Shortage
47
FLKCp'QKdK+ qc"dq = (r+~)dq rearranging yields dq/dK = FLKq~'QK/(r+6--qc") which is negative as long as qc" < r + ~, which we assume 15. The second locus, I(=0, characterizes all (q,K) pairs for which the capital stock is constant. That it is fiat is easily established: at K = 0, qJ(q) = ~i which implies that equilibrium q is independent of K. The intersection of the two curves yields an equilibrium (q*,K*) depicted in Figure 2. It is saddle stable, a feature of models with forward looking agents or assets markets. This can be demonstrated heuristically by examining the equations of motion which govern the dynamics of the system. The arrows depict the direction of change for q and K implied by (14) and (15), and the unique saddle path trajectory to the equilibrium is given along the stable manifold FG. Any other path is "explosive" and ruled out by the conditions for profit maximization '6.
IV. Comparative Dynamics Experiments: Terms-of-Trade Shocks I. A n a l y s i s We can use the apparatus above to analyze the comparative dynamics of the model. In particular, we examine the response of the model to shocks in T, the terms of trade. It is well recognized that the relative price of manufacturing output in Western industrial economies has faced a secular decline in the face of increasing competition from the newly industrializing Third World. As Table 2 illustrates, the OPEC episodes of the last decade represent important punctuation marks in a long deterioration in the relative price of manufactured goods. The way in which the economy of the previous section responds to such shocks will now be studied in a series of comparative dynamics experiments. At the outset it should be stressed that the effects of terms of trade disturbances is an outcome of noncompetitive labor-market conditions. Under competitive wage setting, the labor market always clear, Ld = N = M, so w = qo-I(N/K). Since labor supply is assumed inelastic, changes in the "wedge" [y+(1-y)T] are fully shifted backwards onto workers. Under such a regime, only changes in labor supply, interest rates, and technology will move the steady state product wage w. ~5 if this condition does not hold the system is globally unstable. le This can be more rigorously shown by examining the eigenvalues ofa linearized version of the system around the steady state. In the two-dimension case, the finding of one positive and one negative eigenvalue would imply saddle point stability. The equation for the local saddle path around the steady state is given by the larger of the two associated eigenvectors.
48
M i c h a e l C. B u r d a
Table 2 - Terms of Trade in Manufacturing a
United Kingdom France F.R. G e r m a n y Italy Sweden Austria United States
1973
1975
1980
1985
1.000 1.000 1.000 1.000 1.000 1.000 1.000
0.896 0.613 0.849 0.795 0.959 0.937 0.765
1.059 0.584 0.738 0.705 0.822 0.890 0.568
1.099 0.561 0.698 0.700 0.779 0.952 0.666
a Defined as the ratio of the value added deflator in manufacturing divided by the import price index, 1973 = 1.00. Source: U.S. Bureau of Labor Statistics, Office of Productivity and Technology; OECD [b]; IMF
[var. issues].
Case 1: An unanticipated, permanent terms-of-trade deterioration: Consider first of all the effects of an unanticipated but permanent increase in T, or a deterioration in the terms of trade. The comparative dynamics are depicted in Figure 3. The economy begins at point A. The increase in T has no effect on the K = 0 locus, but does shift the Cl= 0 locus downward to the position (Cl= 0)', reflecting lower profitability at all capital levels. The capital stock is a state variable and cannot react immediately to the new information. Its shadow value q, in contrast, will jump once to incorporate the new information; the path of q to the new steady state cannot contain future "jumps" which would imply infinite arbitrage profits. The new equations of motion of the system apply immediately, and q falls to the new stable arm F'G' at point B. The path to the new steady state (point C) is characterized by rising q (stock market recovery), rising unemployment, and real product wage moderation. The period of protracted subnormal q induces lower investment and cumulatively a permanent reduction in capital per capita (K/M) and a rise in the unemployment rate. Because the I( = 0 locus is flat, only one q is consistent with equilibrium; ceteris paribus, this implies a unique capital-labor ratio, and by the factor price frontier a unique steady state product wage. As in the perfect competition case, the change in the wedge is completely shifted backwards, but now at the cost of capacity unemployment induced by cumulative net disinvestment. The time profiles of variables of interest are plotted in the lower half of Figure 3. Case 2: An unanticipated, temporary terms-of-trade deterioration: This experiment models an unexpected increase in T that is known to be temporary. It is depicted in Figure 4. As in the last case, the Cllocus shifts down: for
49
Capital Shortage
Figure 3
-
An Unanticipated, Permanent Increase in
T
k_-o
,,
1
,'x. I \
s ,
I I
I
I
K'
K
"K\ \ (~I--0)'
~=0 K
f
KI wI
f t1
t~rne
any given capital stock, q must decrease in order to equate the rate of return on capital in place to the required return r. The I( = 0 locus remains unchanged. However, the change is temporary so q does not fall to the new stable arm F'G', but rather to some point above it, say point B in the figure. The trajectory of the economy is now governed by the equations of motion of the new system, which will carry it along the path BC. When T returns to its previous value, the economy will be at point C. T h e / l = 0 locus and the associated saddle-stable path return to their initial positions. Since q cannot move discontinuously ("jump") again, the economy must then be on the stable manifold of the system, upon which it returns to its previous equilib-
Weltwirtschaftliches Archiv Bd. CXXIV.
4
50
M i c h a e l C. B u r d a
Figure 4
-
An Unanticipated, Temporary Increase in
~c
T
~e A
(,~
= o)'
(~ = o
KI
wI t1
t2
time
rium. As may be seen in the time paths plotted in Figure 4, labor input, the capital stock, and the unemployment rate return to their starting values, in contrast to the first example. 2. S u m m a r y and E x t e n s i o n s The above experiments may be extended in natural ways when changes are anticipated, just as in standard models of investment with adjustment costs [e.g., Abel, 1982]. In general, the forward-looking variable, q, will jump immediately when news arrives, even when the anticipated event occurs in the distant future. Until the regime change actually occurs, however, the old equations of motion will govern the economy; thereafter the economy find itself on the (saddle) stable path to the new steady state.
51
Capital Shortage
Other real interventions that are equivalent to altering the wedge can be analyzed in this framework. For example, employee tax reductions that are currently under discussion in Europe are isomorphic to a decrease in T. The model is also robust with respect to the addition of technical progress. The salient feature of the model is that the long-run unemployment rate appears state dependent, or exhibits hysteresis, in response to permanent changes in the terms of trade, without any change in union membership tastes. Because the path to the new level is monotone, past unemployment appears to cause current unemployment. Furthermore, the model is capable of generating periods of falling product and consumption wages with rising unemployment, although firms are continually on their notional demand schedules for labor. It is interesting to note that European product wages measured relative to productivity are currently at levels of the early 1970s.
V. Empirical Evidence One important feature of the model may be readily tested without specification or estimation of investment demand functions. The OPEC oil price increases of the 1970s are often cited as the most significant disturbance to the terms of trade over the past fifteen years. The model implies that growth in the capital stock should have changed around the dates of the oil crises. Our strategy is to allow the data, through the log likelihood function, to speak on the most likely date of a trend break in an ARIMA(1,1,1) specification of annual real gross capital stocks estimated by EUROSTAT and the U.S. Department of Commerce tT. The results are presented in Table 3. Remarkably, the most likely breaks occur in the 1970s and start in 1975, or the year after the oil price increase took effect. In West Germany the break occurs earlier, in 1972! Similarly, the second most likely breakpoint for these countries is 1981, similarly the year after the oil price increases of 1979 took effect. In contrast, the second most likely breakpoint in the U.S. given 1975 is 1976 rather than 1981. Moreover, the sign on this shift was positive, indicating that wage moderation allowed an improvement of profitability is. Significantly, none of the breaks in the capital stock growth is associated with the fiscal consolidation that has occurred since 1981 (with the possible exception of Italy). The breaks we document seem to be correlated with the onset of the so-called profitability squeeze [Bruno and Sachs, 1985, Chapter 8], the widely recognized decline in OECD manufacturing profitability following the OPEC ~7 Data kindly supplied by M. Emerson of the EC and J. Musgrave of the U.S. Commerce Department, Bureau of Economic Analysis. ~a In the late 1970s the real wage flexibility of the U.S. was widely documented. See Sachs [1979], Bruno and Sachs [1985]. 4*
52
M i c h a e l C. Burda
Table 3 - Maximum Likelihood Estimates o[ Three Most Likely Shifts in ARMA (1,1) Representations o[ Differenced Real Gross Manu[acturing Capital Stocks" Number of shifts
United Kingdom .. France ER. Germany Italy
1975 (5.95)*
1975, 1981 (7.02)*
1975, 1981, 1972 (6.24)*
1975 (12.15)*
1975, 1981 (6.59)*
1975, 1981, 1984 (1.50)
1972 (5.17)*
1972, 1981 (1.09)
1972, 1981, 1975 (0.45)
1975 (10.86)*
1975, 1983 (4.29)*
1975, 1981, 1984 (0.40)
United States . . . . .
1975 1975, 1976 1975, 1976, 1981 i (5.16)* (3.75) (5.84)* a Likehood ratios for the null that breakpoint is irrelevant are in parentheses. Sample period is 1962-1985. - Breaks restricted to 1970-1985. - An ARMA (2,1) specification was used for Italy. - Asterisks denote statistical significance at the 5 percent level.
shocks, especially in the European countries. Bruno [1986] has found strong statistical links between profitability and investment behavior. It is noteworthy that Blanchard and Summers [1984] document a European stock market boom since 1982, which is predicted by our model as a consequence of the oil price decline (decrease in r). Moreover, Modigliani et al. [1986] report a modest recovery in European investment since 1983. A final interesting implication of the model involves the response of equilibrium unemployment in the system described in the last section to a change in r. Ultimately, the sensitivity of the wage offer curve to changes in K determines how far unemployment must rise before product wages return to their initial conditions. From (13), if b' and b" are small in absolute value, a large increase in unemployment is needed to "beat down" the nonunion alternative wage, ceteris paribus. It follows that the optimal union wage will inherit this property. If, however, unemployment benefits or real wages in the low wage service sectors decline sharply in response to higher unemployment, it is likely that the increase in the NAIRU will be small. Some corroborating cross-country evidence for this implication may be found in Table 4. In the first column we display Coe's [1985] estimates of "short-run real wage rigidity", or the percentage increase in unemployment required to reduce wages by one percent given correct price expectations by workers. These estimates are a potential proxy for the responsiveness of union members' "best alternative" to labor market conditions. In the second
53
Capital Shortage
Table 4 - Real Wage Rigidity and the Re 9lacement Wage"
United Kingdom France F.R. Germany Austria Italy United States
Real wage rigidity b
Replacement wage c
1.94 1.52 1.76 0.83 1.48 0.67
41.60 38.22 65.52 13.50 24.00 15.00
a Correlation between these variables: 0.77. - b p e r c e n t a g e increase in unemployment required to reduce wages by one percent given correct price expectations by workers. c Number of weeks the unemployment insurance benefit applies times the replacement ratio. Source: Coe [1985]; Emerson [1986]; Minford [1985]; own calculations.
column we have transformed European unemployment insurance replacement ratios from Emerson's [1986] recent survey using the metric of percentage point-weeks, or the number of weeks the UI benefit applies times the replacement ratio. To the extent that unemployment benefits constitute a large component of the replacement wage, the correlation of 0.77 between the two columns supports the argument that increased generosity of the benefit in size and duration is associated with less flexible real wages w~. VI. Conclusion As was shown in the first section, the European concern over the availability of productive capacity to accommodate an aggressive fiscal expansion is not altogether unfounded. This essay has sought to explore how such a capital or capacity shortage might arise as a medium to long-run consequence of real wage rigidity when shocks impinge on the wedge between consumption and product wages. In the context of optimizing firms and unions, the model we develop endogenizes the response of the stock of productive capital to a terms-of-trade change and concomitant profitability reduction. An important finding is that the response of the opportunity wage or UI benefit to unemployment is crucial in determining the resting point of the system. In addition to the stylized facts, the time series evidence also provides support to a capital shortage interpretation of long-run increases in European unemployment over the past one and a half decades. Specifically, the most likely breaks in the behavior of European capital stocks are associated with ~0 This should not be confused with the argument that UI benefits alter household labor supply, which represents a voluntary phenomenon; rather we are stressing the potential effects of UI benefit inflexibility on union wage setting behavior.
54
Michael C. Burda
the oil price increases of the 1970s rather than the fiscal consolidation of the last five years. This evidence also supplies some direction for policy. While the model examines changes in the terms of trade, it is clearly germane to all elements of the so-called "wedge", including employee payroll taxes, social security and retirement contributions 2~ It follows that supply management policy accompanying a fiscal expansion should aim at "wedge relief' in the form of employer subsidies or tax reductions 21. Apparent state dependence in European unemployment rates has attracted considerable attention in recent years, and has encouraged the construction of models without unique stationary equilibria. In our model, state dependence in the unemployment rate is spurious, since the system contains no zero roots. Permanent changes in the terms of trade, however, can lead to permanent changes in total capacity and the (natural) unemployment rate through effects on profitability and investment for given union preferences. Furthermore, the model can explain the current contemporaneous decline in wages and rise in unemployment in a setting of classical unemployment along the adjustment path. An important shortcoming of this paper is the lack of a formal model of the best alternative wage available to union members, which represents some weighted average of unemployment benefits, income from self-employment, or employment in the nonunion economy. The responsiveness of this quantity to the unemployment rate has been shown to be central to determining the steady state of the economy. One option may be to follow McDonald and Solow [1985] and model a two-sector, Harris-Todaro economy in which workers queue for industrial jobs while either waiting and collecting unemployment benefits or working in the service economy. References Abel, Andrew B., "Dynamic Effects of Permanent and Temporary Tax Policies in a q-Model of Investment". Journal of Monetary Economics, Vol. 9, 1982, pp. 353-373. Bailey, Martin, "The Labor Market in the 1930s". In: James Tobin (Ed.), Maeroeconomics, Prices, and Quantities. Washington 1983, pp. 21-61. Blair, Douglas, David Crawford, "Trade Union Preferences and Collective Bargaining". The Quarterly Journal o[ Economics, Vol. 99, 1984, pp. 547-566. Blanchard, Olivier, Lawrence H. Summers, "Perspectives on Real World Real Interest Rates". Brookings Papers on Economic Activity, Vol. 2, 1984, pp. 273-324. -, -, "Hysteresis and the European Unemployment Problem". Macroeconomics Annual, 1, 1986, pp. 15-78. 20 For an empirical analysis of the role of the tax wedge in the evolution of the NAIRU in France, see Sachs and Wyplosz [1986]. 21 This constitutes an important component of the policy already suggested in the "two handed" approach of Blanchard et al. [1986].
Capital Shortage
55
Blanchard, Oiivier J., Jacques Dr~ze, Herbert Giersch, Richard Layard, Mario Monti, "Employment and Growth in Europe: A Two-Handed Approach". In: Olivier Blanchard, Rudiger Dornbusch, Peter R. Layard (Eds.), Restoring Europe's Prosperity. Cambridge, Mass., 1986, pp. 95-124. Bruno, Michael, "Aggregate Supply and Demand Factors in OECD Unemployment: An Update". Economica, Vol. 53, 1986, Supplement, pp. $35-52. -, Jeffrey D. Sachs, "Input Price Shocks and the Slowdown in Economic Growth: The Case of U.K. Manufacturing". Review of Economic Studies, Vol. 49, 1982, Special Issue, pp. 679-706. -, The Economics of Worldwide Stagflation. Cambridge, Mass., 1985.
Burda, Michael, "Trade Unions, Seniority, and European Wage-Employment Dynamics: An Investigation of the Membership Hysteresis Hypothesis". In: Michael Burda, Essays on the Rise of Unemployment in Europe. Harvard University, unpubl. Ph.D. Diss., May 1987, pp. 70-117. Coe, David T., "Nominal Wages, the NAIRU and Wage Flexibility". OECD Economic Studies, 5. Paris, 1985, pp. 87-126.
Dunlop, John T., Wage Determination under Trade Unions. New York 1944. Emerson, Michael, What Model for Europe? Brussels, June 1986, mimeo. Franz, Wolfgang, Hysteresis Effects and the NAIRU: A Theoretical and Empirical Analysis for the Federal Republic of Germany. 1986, mimeo. Giersch, Herbert, "Aspects of Growth, Structural Change, and Employment: A Schumpeterian Perspective". In: Herbert Giersch (Ed.), Macroeconomic Policies, Growth, and Stability. Tiibingen 1981, pp. 181-206. -, "Arbeit, Lohn und Produktivit~t". Weltwirtschaftliches Archly, Vol. 119, 1983, pp. 1-18. Gottfries, Nils, Hendrik Horn, Wage Formation and the Persistence of Unemployment. Institute for International Studies, Seminar Paper, 347. Stockholm 1986.
Grossman, Gene, "Union Wages, Seniority and Unemployment". The American Economic Review, Vol. 73, 1983, pp. 277-290. Hayashi, Fumio, "Tobin's Marginal q and Average q: A Neoclassical Interpretation". Econometrica, Vol. 50, 1982, pp. 213-224. International Monetary Fund (IMF), International Financial Statistics. Washington, var. issues. Kamien, Morton I., Nancy L. Schwartz, Dynamic Optimization: The Calculus of Variations and Optimal Control in Economics and Management. New York 1981. Kydland, Finn E., Edward C. Prescott, "Time to Build and Aggregate Fluctuations". Econometrica, Vol. 50, 1982, pp. 1345-1370. Malinvaud, Edmond, "Wages and Unemployment". The Economic Journal, Vol. 92, 1982, pp. 1-12.
McDonald, lan M., Robert M. Solow, "Wage Bargaining and Employment". The American Economic Review, Vol. 71, 1981, pp. 896-908. -, "Wages and Unemployment in a Segmented Labor Market". Quarterly Journal of Economics, Vol. 100, 1985, pp. 1115-1141.
56
Michael C. Burda
de Menil, George, Bargaining: Monopoly Power versus Union Porver. Cambridge, Mass., 1971.
Minford, Patrick, Unemployment: Cause and Cure. Second edition. Oxford 1985. Modigliani, France, Marie Monti, Jacques Dr~ze, Herbert Gierseh, Richard Layard, Reducing Unemployment in Europe: The Role of Capital Formation. Centre for European Policy Studies, Paper No. 28. Bruxelles 1986. Newell, Andrew, ]ames S.V. Symons, Wages and Employment in the OECD Countries. London School of Economics, Centre for Labor Economics. Discussion Paper, 219. London, May 1985. Organization for Economic Cooperation and Development (OECD), Collective Bargaining and Government Policies in Ten OECD Countries. Paris 1979. - [a], Main Economic Indicators. Paris, vat. issues. - [b], National Accounts. Paris, var. issues. Oswald, Andrew ]. [1982a], "The Microeconomic Theory of the Trade Union". The Economic Journal, Vol. 92, 1982, pp. 576-595. - [1982b], "Uncertainty and the Trade Union". Economic Letters, Vol. 9, 1982, pp. 105-111. Sachs, ]effreT D., "High Unemployment in Europe: Diagnosis and Policy Implications". NBER Working Papers, 1830, 1986. -, Charles A. Wyplesz, "The Economic Consequences of President Mitterand". Economic Policy, Vol. 2, 1986, pp. 261-305.
Summers, Lawrence H., "Taxation and Corporative Investment: A q-Theory Approach". Brookings Papers on Economic Activity, 1981, pp. 67-127.
Z u s a m m e n f a s s u n g : Gibt es eine Kapitalknappheit in Europa? - Die Hypothese yon der ,,Kapitalknappheit" erkl~'t den Anstieg der natiirlichen Arbeitslosenquote in Europa zum grol~en Teil mit Vertinderungen der Arbeitsnachfrage, die sich aus den langfristigen Wirkungen yon rigiden L6hnen auf die Kapitalakkumulation und auf den gesamten Umfang der Wirtschaft ergeben. In diesem Aufsatz werden zun~ichst einige Belege fOr eine Kapitalknappheit in Europa gegeben. Die q-Theorie der Investitionen wird dann mit einem einfachen Modell verbunden, in dem die L6hne von den Gewerkschaften bestimmt werden. In einem solchen Rahmen spielt der Keil zwischen den Reall6hnen beim Konsum und den Lohnkosten bei der Produktion eine entscheidende Rolle for die Bestimmung des Kapitalstocks und der Beschtiftigung im langfristigen Gleichgewicht. Die Anpassungslast f~illt in diesem Modell auf die Determinanten der ,,alternativen" LShne augerhalb des Gewerkschaftssektors, was auch im L~ndervergleich best~tigt wird. Das Modell kann nicht nut die starken negativen Abweichungen des Kapitalstocks yon seinem l~ingerfristigen Trend erkl~iren, die seit 1973 in der europtiischen Industrie beobachtet wurden, sondern auch den Anstieg der europ~iischen Aktienkurse in der jfingeren Vergangenheit.
R~sum~: Est-ce qu'il y a une p~nurie de capital en Europe? - L'hypoth~e de la ~p~nurie de capital~ attribue une grande partie de l'augmentation du taux de ch6mage naturel en Europe aux d~placements de la demande de main d'oeuvre qui apparaissent comme effets
Capital Shortage
57
long terme des salaires rigides a l'accumulation de capitaux et a l'6chelle g6n6rale de l'6conomie. Dans ce papier l'auteur pr6sente d'abord quelque 6vidence pour la ~p6nurie de capital,, en Europe. Ensuite, la q-th6orie d'investissement est int6gr6e dans un module simple de la d6termination du salaire syndical; en ce contexte le coin entre les salaires qui sont importants pour la consommation et ceux qui sont importants pour la production joue un r61e essentiel pour d6terminer le ~steady states du stock de capital et d'emploi. Dans ce module le faix d'ajustement p~se sur les d6terminants du salaire ~alternatif~ hors du secteur syndical, une implication support6e par l'6vidence trans-pays. Le module n'explique pas seulement les grandes d6viations en stock de capital de sa tendance a long terme observ6es depuis 1973 dans le secteur manufacturier en Europe, mais aussi l'augmentation des cours aux march6s des actions pendant les cinq ann6es avant la chute des cours.
R e s u m e n : / , Hay escasez de capital en Europa? - La hip6tesis de escasez de capital le atribuye el aumento de la tasa natural de desempleo en Europa en gran medida a cambios en la demanda de trabajo como consecuencia de los efectos de largo plazo de la rigidez del salario sobre la acumulaci6n de capital y el nivel general de la econom/a. En este trabajo se estudia esta hip6tesis. Se presenta evidencia de la escasez de capital en Europa. La teorfa ,,q,, de la inversi6n se integra a un modelo simple de determinaci6n del salario sindical. En tal situaci6n la diferencia entre el salario de consumo y el de producci6n juega un papel crucial en la determinaci6n del stock de capital y del empleo en ~steady state,,. En este modelo el peso del ajuste recae sobre las determinantes del salario ,,alternativo,, no sindical, una implicaci6n que es confirmada por la evidencia emptrica de una muestra de parses. El modelo no s61o explica las importantes desviaciones negativas de la tendencia observada despu~s de 1973 sufridas por el stock de capital manufacturero europeo, sino tambi6n el reciente aumento de los precios de las acciones europeas en los dltimos cinco aflos.