The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
TY:
INDUSTRY RESTRUCTURING MEASURES AND PRODUCTIVI
EVIDENCE FROM THE 1980's
by
S. Brock Blomberg and Charles Steindel
Federal Reserve Bank of New York
Research Paper No. 9509
May 1995
only.
This paper is being circulated for purposes of discussion and comment
or quotation without
The contents should be regarded as preliminary and not for citation
r and do not necessarily
permission of the author. The views expressed are those of the autho
Reserve System.
reflect those. of the Federal Reserve Bank of New York or the .Federal
Single copies are available on request to:
Public Information Department
Federal Reser ve Bank of New York
New York, NY 10045
Industry Restructuring Measures and Productivity: Evidence ,From the 1980s
S. Brock Blomberg
and
Charles Steindel *
The Federal Reserve Bank of New York
April I0, 1995
* We-thank members•ofthe Federal Reserve Bank-of,Ncw York Seminar. The views expressed here arethose of the authors and do not necessarily reflect the views of the Federal Reserve Bank of New York or the
Federal Reserve System. We thank Marl,!aret Blair for sharing her data. All errors are our responsibility.
Absttact
Industry Restructuring Measures and Productivity:
Evidence from the 1980s
S. Brock Blombe,g & Charles Steiodel*
This paper -analyzes the empirical relationship between corporate restructuring and
productivity. We estimate neoclassical production functions and factor demand functions to
analyze the importance of restructuring in improving resource allocation and productivity. We
find, at most, restructuring may have spurred the substitution of capital for labor in some
industries, helping to set the stage for increased labor productivity. However, there is little
evidence that restructurings, themselves, aided in the improvement of true technological progress.
Keywords: Restructuring. Productivity .
.JEL Classification System: LI. ()4,
*Address Con-espondence to: S. Bnick Blomberg-or Charles Steindel, Domestic Research, The Federal Research
Bank of New York; 33 Libeny Street. New York. New York. 10045. Tel. (212) 72().;8100, FAX- (212) 720:,1379.
Restructuring and Productivity, Blomberg and Steindel
1. Introduction
The transformation of'many large American corporations·has been a prominent feature
of the economic landscape. The term '.'restructuring" takes in a great many phenomena, not only
balance sheet changes--in recent years, often involving the replacement of short-term debt by
long-term debt and equity, following the 1980s, when many firms sharply increased leverage--but
also the more diffuse changes in the basic organization of many large firms. These latter include
the divestiture of divisions, plant closings, elimination of layers of management, and outsourcing
many peripheral activities (for instance, getting rid of in-house provision of support services, such
as accounting, and reducing the degree of vertical integration in corporations by turning to
outside providers of materials· and ·intermediate products).
It has also been the case that a number of industries have seen significant improvements
in productivity growth for the period since the early 1980s, as compared to the sluggishness af
much of the 1970s. This paper attempts to examine the linkages between industry-wide indexes
of restructuring and productivity growth, using data from the 1980s.
ln··an likelihood, substantive gains in productivity by a firm or industry are accompanied
by significant changes in the structure of production, management organization, and finances.
The purpose of this paper is to see whether there are systematic relationships between several
broad measures ofchange. or restructuring. and productivity growth at the industry level. lfthere
are, this result would suggest that tax and regulatory policies which have the effect of
discouraging restructuring activities might have anti-productivity side effects. If no significant
positive linkage is found between restructuring indexes and productivity growth, the policy
·consequences are less clear.. Moreover, it can be argued that the evidence from the 1980s have
I
Restructuring and Productivity, Blomberg and Steindel
2
limited relevance for the 1990s..
, It wu,noted above that there are two-broad types ofrestructurings:.. financial--orbalance
:sheet--restructurings,. and .real-side-..restructurings. . :Restructuring announcements. are often
rewarded by increases in corporate stock prices.
From the point of view of corporate
shareholders, then, restructurings are quite often worthwhile. 1 However, it is not. true that
benefits to shareholders of an individual corporation are benefits to society as a whole. A
restructuring may be merely a devise to transfer income flows in a corporation from taxpayers,
workers, and suppliers to shareholders, and need not involve any efficiency gains in
production.
Financial:restructurings by·themselves obviously do not imply any improvements in
production since, by definition. they are merely paper transactions_.
It has, however, been
frequently •argued·'that financial restructurings will,spur efficiency gains. The•canonical,financial
restructuring,,.of the l 980s.involved, increasing the debt loads of corporations and concentrating
equity interests in small groups. often including senior management. Dr. Johnson said the
prospcl:t 0Pa man's being hanged i'concentrates his mind wonderfully"; the increased prospect
of bankruptcy connected with increased debt loads arguably spurred firms to look for new ways
to improve production techniques. 2 The carrot of increased equity interest on management's part
1Important issues are whether the stock price gains -associated with restructuring
announcements are sustained;-:and whether·the stock price gains are justified by future earnings
increases.
.2The genesis of the famed Johnson .quote may be relevant. Johnson had written the final
message of:a convicted ·clerygyman. A reader confronted Johnson with the uncharacteristic
Johnsonian force of the message. whereupon Johnson denied authorship. Johnson made the
Restructuring and Productivity, Blomberg and Steindel
. would add even more. incentive to boost productivity. Of course, though, the incentives given
the firm by: financial restructuring are actually .in the direction of redu,ced costs,(say by reducing
salaries· or: outsourcing)• or ,-increased revenues,. not necessarily toward jncreased .producti'lre
efficiency.
Previous research using firm-level or plant-level data has indeed .found.fairly significant
effects from restructuring to improving profitability. Amihud (1989) provides a nice survey of
the evidence for equity premiums associated with corporate restructuring. Others such as Cave
and Krepps (1993), Kaplan (1989), Smith (1990), Baker and Wruck (1989) have found positive
effects in terms of earnings and/or net cash flows,· or displacing nonproductive workers.
·· However, none
cif these.· studies
argue that -restructuring has led to greater productivity, only
••greater profitability. and are :therefore subject to our opening criticisms.
Lichtenberg-and Siegel-(1990) have ,addressed our opening criticisms by analyzing-the
··· direct impact restructuring has productivity. · Using the U.S. Census .Bureau's Longitudinal
research Database (LRD) and a list of LBO's provided by Morgan Stanley and Co., they link firm.
•·••. tevel productivity to restructuring ..• Their :analysis revealed that restructuring between 1983 to
1986 significantly improved productivity, but had no significant impact in the first years of their
time sample.
There are, however,.several drawbacks to their approach ... First, Lichtenberg and Siegel
measure restructuring as the number of LBO's rather than the dollar amount of the transaction
comment to Boswell when recounting the incident--the point being that the clerygyman's situation
.•• lenLcredence .. to Johnson's denial. not·:that there was an .actual transformation .in the condemned
man (see Boswell, p. 725).
3
Restructuring and Productivity, Blomberg and Steindel
4
itself. 3 ·It is not clear that the buyout .of RJR .Nabisco, for example, which led to the famQ\IS
book arid movie Barbarians at the Gate; should be treated in. the same way as every other LBO.
· Second/finding positive correlation between .restructuring ,and productiv~ty during the. boom ~f
··· I 983 to ·1986 might be misleading.in that it is exceedingly difficult to test whether restructuring
led to higher productivity or whether both restructuring and productivity increased in response
to the upswing of the business cycle. Finally, results at the firm-level or plant-level may not hold
when the unit of observation is an industry.
To addr~ss these criticisms, we analyze industry-level data over a wider range of years,
1977 to 1989, and employ actual dollar transactions of restructuring from the Brookings
Historical Merger Data file. By·examining the data-in this fashion we can better control for
business cycle effects and improve on the measure of restructuring.
The next section' deals with some ·conceptual issues, followed by an exploration,of the
empirical. evidence.
The conclusion is that, at most, restructuring .may .have spurred .the
substitution of capital for·labor, helping to set the, stage for a boost up in labor productivity, but
., "· there 'is nfrcevidence ··that restructurings aided true technological progress.
2. Possible Links Between Pmductivity and Restructuring
The term productivity is often misused.
A very common tendency is to associate
. productivity with the ratio of final .sales of a product to .employment by the final producer. ,:he
classic examples·are expressing .. productivity changes for. the auto. and steel industry by changes
in the ratio of unirautomobile shipments to.auto employment and changes in the ratio of tonnage
of steel shipments to steel employment. These are valid measures of productiyjty growth
3 As was also done more recently in McGuckin, Nguyen, and Remek (199.4).
if and
Restructuring and Productivity, Blomberg and Steindel
5
.••-only if two conditions•are satisfied: t, 1Ibere·are prQl!ortional chan11es in the Jabor·jnput•at all
levels
of production prion° the final shipment 2;
There
are no chani:es in the real value of the
physical product shipped.
Consider the first condition. In ·the auto and steel industries changes in the past decade
have often been said to involve the "outsourcing" of many functions; for instance, the spinningoff of divisions which supplied parts and raw materials.
Such moves would have reduced
employment in the final product industries, and work to increase the commonly reported
"productivity" measures. It is perfectly conceivable, however, that the spinning-off a division
may mean absolutely
no improvement in the actual physical production process: for instance,
an auto manufacturer Which spins off a division that produces_ batteries may still purchase the
same batteries made in the same factory by the same number of workers! Clearly, the spin-off
has not increased productivity in any meaningful way, even though the auto company might
· repon a ,handsome ·increase in the'· ratio of vehicle assemblies to employment. --Of course,- the
· spin-off of the battery,plant might spur innovation, as the managers now become owners start
' 'impleinenting changes,'etc: '.However, such reasoning--which is very common in many popular
discussions of restructuring--is speculative.
The second condition is rather simpler.
Obviously, an automaker which increases
assemblies per worker by switching to making a less-sophisticated car, or a steelmaker which
increases tonnage per-worker by switching from production of,plate.,to wire rods, may.not have
· increased productivity in any meaningful sense.
The cited studies of firm-level and plant-level manufacturing productivity and restructuring
-' ' - ,measure output by shipmentst· Although there are many benefits of using disaggregated data, ·the
Restructuring and Productivity, Blomberg and Steindel
6
,·,-,..:possible-<inaccuracy -ofthis ,output measure ·is one reason to turn .to industry ,measures. ,At,the
· •. two0 digit industry level,,government statisticians define1111 industry's output,by its constant-dollar
value-added--its constant-dollar sales less its constant-dollar purchases from outside sources. The
use--of :constant-dollar data helps prevenl the distortion· of productivity ,data-by switches in- the
specific mix of output.
The use of value-added prevents productivity indexes from being
distorted by changes in the legal relationship between different levels of the production process.
For instance, in the case of the auto company spinning-off the battery plant, measuring auto
output by value.-added means that the spin-off reduces both the output of the auto plant (batteries
are now included in purchases from outside sources, and are now an offset to sales in the
, calculation of value-added) and its employment.
Thinking about productivity in terms of real ·value-added per worker helps clarify possible
connections between -restructuring ·and productivity.· It is -immediately obvious ,that •purely
· - ,, financial,restructurings--changing lhe ~-pecifics of the balance sheet of a firm-shave no immediate
consequences for productivity, since they do not change any of the proximate determinants of
productivity. The sales of a firm are not: changed,by a change in its balance sheet, the purchased
input.~ are not changed--since capital costs like interest expense are not counted as a purchase-and employment is not changed! Only if the financial restructuring
provokes real-side changes
does it lead to productivity changes.
Surprisingly, the immediate,consequences-,of real-side,restructurings for. productivity-.are
murky.
In the course of a real-side restructuring a firm -may -drastically alter the level and
-· composition of its sales and expenses. It is plausible. of course,·that in-general, the changes will
!
;increase 'the,·,profitability of. the firm, undertaking them, as· measured, say, by return to equity
· Restructuring and Productivity, Blomberg and Steindel
,capitaL·:HoweverJbere is not necessarily
a simple or even
7
a predictably-signed relationship
between changes-in a:fum's profitabj)jty and changes in its productiyity.,measuring productivity
by real-value-added ·per worker. Consider the case ·of our automaker spinning off the .battery
' plant. The automaker may, use· the new. relationship with the battery maker to cut. the prices it
pays·for batteries (under the old captive relationship the battery plant may have been allowed.to
book above-market prices in the integrated firm's internal accounting).
The automaker's
profitability goes up, but there's no way to predict what happened to its real productivity.4
This disc::ussion has tried to get across the point that there is no necessary, definitional
connection between a "restructuring"--however defined--and enhanced productivity. Moreover,
higher profits as a result of a -restructuring, or higher share values following a restructuring
mover. are not~ signals of improved productivity, much heated rhetoric to the contrary. In
the next section we will look at statistical evidence to see the relationship of various measures
•Of restructuring activity to an industry's productivity.
4 lf the real productivit
y of the battery plant was higher than that of the assembly plant, the
spin-off will reduce the real value-added per worker of the auto company.Conceivably, the
demand of the automaker for lower battery prices could spur the battery plant to the achievement
of production improvements to reduce costs. In that sense. the real-side restructuring of the.auto
company encourages productivity gains which benefit the economy as a whole, although they
occur in a supplier rather than at the final producer. However, it is perfectly possible that the
demand for lower battery prices·could be met simply·by the battery maker's reducing wages and
· benefits at it~ plant. In· some long-term convoluted way this cost-cutting could result in some
efficiency gain for the economy (high-wage workers would ·exit battery makers for some other
industry where their productivity would justify there wages) but this would clearly not be a very
smooth or predictable process.
Restructuring and Productivity, Blomberg and Steindel
· 3. •Preliminary Statistical Evidence
•· Our aim .is to:-explore- the ·statistical ·Iinkages. ,between ,restructuring .proxies and
productivity growth.• The government•provides.value-addect productivity measures at both very
highly aggregated. and industry levels... ,At-times,,analysts.have cited improved growth in . the
aggregate productivity indexes--for nonfarm business as a whole, and, in .particular, .the. sharp
improvement in productivity growth in manufacturing--over the last decade as evidence
supporting a positive relationship between restructuring and efficiency gains. Whatever the merits
of the argument,,since the diverse indicators of restructuring are so closely linked in time, the
aggregate time series evidence can not be relied to discern ~ concept of restructuring is most
closely linked to productivity growth,ior to, distinguish the relative.importance of restructuring
proxies.
To deal with this problem we turn to.,the disaggregated data on productivity. and output
by 2-digit industry.
The .. presumed dispersion of .restructuring and .productivity across the
industries offers greater opponunities to pin down relationships.
• Our strategy will be· to take in to account a wide variety of factors that might influence
industry productivity, with restructuring proxies included as terms that might have either an
independent effect or will enhance the effect of other factors. Our basic framework is a standard
neoclassical production function, where; output, Q, is determined.by-·its inputs
· where J4 · is ·some· measure of technology• and Z is a vector of inputs- that include physical capital,
K, and labor; L
8
Restructuring and Productivity, Blomberg and Steindel
9
If production is homogenous of degree one, then productivity growth or growth per worker can
be expressed as a function of these inputs.
Assuming the production technology is Cobb-
Douglas, we express labor productivity, q, by the following identity 5
where growth rates for any variable X are given by AX and cr measures the returns to capital.
.. In this .framework, there are three potential ways restructuring proxies might affect
productivity growth:
1.c · There might. be an'independent restructuring. effect, independent -of all other variables--in the
jargon of ·growth studies, restructuring could be directly related to "technical progress''.. or the
"Solow residual" (A) ..... Such,.an effect could· come about if restructuring was linked to
• '.-'entrepreneurship" -in an industry:-for example. We depict this effect as a,parallel shift in a
production function from I to II. (See example on next page.)
-2.• ·Restructuring could alter the relationship of factors -of production to output. For instance,
restructuring could result in a reallocation of an industry's investment, resulting in increased
output. In this instance, restructuring leads to an increase in the productivity of capital, rather
than directly affecting·· technical progress.
In terms of the symbols above, the restructuring
increases ·o rather than AA ,--increasing the growth rate oflabor productivity rather than technical
progress.· We depict this effect as a shift in the slope of·the production function from I to
m.
, 5 11 is not necessary for production to be Cobb-Douglass, however, it greatly simplifies the
algebra.
·,
Restructuring and Productivity, Blomberg and Steindel
JO
\
\
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
~◄
\
\
\
\
\
\
.
\ ·.
\ ·.
\ ·.
\',
\',
\
\
',\
·. \
·. \
·. \
·. \
·.
·.
\
\
\
\
'\
.~. -~
..:, •uononpoJd
Restructuring and Productivity, Blomberg and Steindel
ll
3. Restructuring could alter the· demands of an industry for factors ofproduction, thus indirectly
affecting the level of output. For instance; an industry might change the overall level oLits
investment plans following a restructuring, thus ,changing its capital/labor ratio and altering . its
labor productivity level, though not necessarily its trend growth. We depict this as a movement
along a production function from I to IV.
There is a subtle distinction between possibilities 2. and 3. Possibility 2. implies that
restructuring is actually efficiency enhancing, implying that restructuring improves the marginal
products of one or more factors. Possibility 3. implies that restructuring changes the input mix,
leaving the underlying technological progress unchanged.
In short, Possibility 2. means
technology progressei;'!(though disembodied technology--the So\ow residual, or in our symbols,
A--is unchanged) whereas Possibility 3. means -at most that resources are allocated more
efficiently. Unfortunately. we-will not formally differentiate between possibilities 2. and .3.--in
a short sample period, a transition·~o a higher labor productivity level is hard to distinguish ,from
the emergence of a new higher growth path.
-·; Before we . begin the· formal statistical analysis that tests these possible links between
restructuring proxies and productivity. we must first check to see if there is an issue to be
explored--is there a large deviation in productivity trends across industries in the 1980s Bf1!:[ we
have taken into accounrchanges in standard variablesaffecting,productivity growth? If there.is,
then differences in restructurings across industries may help to.explain these differences.
As a first step, Tables lA-B presents the data on.average productivity growth by industry
for I 980-84 and I 985-89 (columns 2 and 3 l and the change in the average growth between the
:.two periods (column 4) (data for the years since '1989 are still somewhat preliminary). ·In
Restructuring and Productivity, Blomberg and Steindel
l_,:;;-'•\
12
. :':f.able IA. Industry Productivity,Growth 1980-84 versus 1984-89: The ·Winners
.
..
'~ ~,,;;...r-
•,
;,,
llul."til.r:l.
"ca~,a !6.0.-64. !
Oil & Gas Extract
Misc. Repair Serv.
Instruments*
Security Brokers
Leather
Machine & Equip.*
Legal Serv.
Coal Mining
Electronics*
Electric, Gas, etc.
Radio & T.V.*
·,Educational Serv.
Tobacco
Apparel
Social Serv.
Primary Metals
Personal Serv.*
Holding Co.
· Nonmetal Minerals
Chemicals
Real Estate
Health Serv.
Telephone & Tele.*
Water Transp.
Local Transit
Insurance
Transp. Serv.
1.15
-3.30
1.74
0.17
· 1.24
1.85
-5.22
8.40
2.81
-0.80
-3.75
-2.07
-4.71
3.07
0.64
0.75
-1.40
-0.28
3.04
4.51
-0.77
-1.97
6.18
0.51
-1.76
-0.24
0.01
"ca~,a 7.18
!6.2-6.2!
2.40
7.43
5.35
6.17
6.58
-0.83
12.54
6.80
2.76
-1.11
0.49
-2.38
3.94
1.47
1.54
-0.74
0.37
3.56
4.95
-0.42
-1.64
6.47
0.70
-1.58
-0.12
0.10
,~~-~21 - ,~a-~1
>
6.02
5.70
5.70
5.18
4.92
4.74
4.39
4.14
3.99
3.56
2.64
2.56
2.33
0.88
0.83
0.80
0.66
0.65
0.52
0.44
0.36
0.33
0.30
0.19
0.18
0.11
0.09
*Series break 86/87 due to SIC classification change.
principle, one can use these data and relate the changes in trend productivity growth by industry
to measures of restructuring. The conventional wisdom is that productivity growth may have
increased substantially during the latter period in quite a few industries due to the substantial
increase in restructuring. Such an observation. though, could easily fail to account for factors,
Restructuring and Productivity, Blomberg and Steindel
.•,,
--------------------------------Table 1B. Industry Productivity Growth 1980-84 versus 1984-89: The Losers
-~:,,,,•~,..,.
Industry
Grawth (80-841
Bus. Serv.*
Railroad Transp.
Paper & Allied
Printing & Pub.
Other Trans.*
Insurance Carriers
Lumber & Wood*
Textile Mill Prod.
Stone &Glass*
Trans. by Air
Amusement.& Rec*
Hotels
Retail Trade
Banking*
Auto Serv.
Construction
Fabricated Metals
Rubber*
Trucking & Ware.
Food
Furniture
Wholesale Tr.
Pipelines
Misc. Manu.
Petroleum & Coal
Motion Pictures*
Motor Vehicles
Metal Mining
Cred. Ag. no banks*
-0.06
9.82
3.41
-0.51
5.96
-1.27
2.73
4.74
3.60
3.62
2.71,
0.59
2.34
0.0 I
-0.97
0.85
3.67
5.34
1.39
4.51
3.0ll
5.ll6
3.K I
10.3K
13.40
3.91
7.67
29.43
42.28
-Growth (85-891
-0.32
9.29
2.77
-I.IS
5.20
-2.15
1.72
3.46
2.30
2.3 I
I.I I
-1.39
-0.02
-2.50
-3.69
-1.87
0.57
2.0 I
-2.08
0.95
-0.60
1.96
-1.20
3.24
4.39
-5.29
-3.25
15.42
16.58
C§5• 89 1• C8Q·§4
I
-0.26
-0.53
-0.64
-0.64
-0.76
-0.88
-1.01
-1.28
-1.30
-1.32
-1.60
-1.98
-2.36
-2.51
-2.72
-2.73
-3.10
-3.34
-3.47
-3.56
-3.68
-3.90
-5.02
-7.14
-9.01
-9.20
-10.92
-14.0 I
-25.70
*Series break 86/87 due to SIC classification change.
such as business cycle effects, that have significant impacts on productivity but are unrelated to
restructuring.
Looking at the tables, though, notice that there does not appear to be any
:systematic•-patterns ·in· productivity: performance--there are practically the identical number of
13
Restructuring and Productivity, Blomberg and Steindel
· ·-winners as•. there are losers. Even more telling, certain industries, which obviously .experienced
' major restructuring,-e.g,; telephone and-telegraph, exhibite~ virtually constant productivity growth
rates in the two periods.
Despite the fact that trend productivity does not seem to be higher in most ,industries. the
later period, we still have not provided evidence concerning the relationship between restructuring
and productivity.
depressed
For example, it could be that there exists some secular phenomenon that
productivity in the later period, independent of the increased incidence of
restructuring. For instance, the sharp cyclical increase in productivity in I983 and 1984 as the
economy emerged from the very deep I 981-82 recession could have inflated the first sample
relative ·to the· ·second, To properly evaluate this, possibility; we must carefully examine the
determinants of productivity. to see if there is a problem.to explain with the help of restructuring
indexes.
Therefore, as a benchmark:. Table 2 presents the results of a bare-bones model explaining
productivity, growth by 2-digit industry. The dependent variable is the growth of real gross
·, <output' per·full-tinie equivalent employee;; .The independent variables are, the ratio of start-of-year
real net fixed reproducible capital per employee, a dummy variable for each year (to account for
business cycle effects, such as scale economies), and a dummy variable for each industry (though
· not reported). The estimated,coefficients seem plausible; the-elasticity ofthe capital,labor,ratio
is 0.28 and the dummies . are negative and more significant in the recession years of the early
1980s.
··Now; we· begin by introducing hi-tech investment into the,regression because there is a
·-'r.;witlespread"fatuitive,notion•that '.'computerization" must ·have contributed greatly to productivity
14
Restructuring and Productivity, Blomberg and Steindel
15
4r-·····--------------------------------·:.Table 2: · Neoclassical Growth Model 1980-1989
"<:,',-;
-~·-.·~ ...
Erelqnatorv Ym: ·
CONSTANT
KIL
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
Adjusted R2
Ceeff,
-3.71
0.28
-0.41
-0.38
-0.31
-0.29
-0.23
-0.20
-0.16
-0.09
-0.05
-0.0 I
:ui.tlL
23.79
13.69
2.80
2.87
3.08
2.66
2.38
2.27
2.03
2.42
0.90
0.32
0.77
*The explanatory variables in the regressions include a constant, time (1980,.a,l 989) and industry
dummies. and the log of the capital labor ratio (K/L).
growth in the 1980s. It is interesting to analyze·what including "high-tech" investment .does to
the estimated coefficients.
For reasons that will not be discussed here, standard neoclassical
theory. combined with observations of the rapid declines in computer prices, strongly suggests·
· , that ,:'i,mputerizatiort' did not ·have :substantial impacts on· productivity (Steindel, 1992a, 1992b;
Oliner and Sichel, 1994). However, it is certainly worth seeing whether computer investments
helps explain industry productivity growth--perhaps industries which made heavy use of
computers in the I 980s saw unusually rapid growth in productivity. Accordingly, the ratio of real
net information processing equipment per full-time equivalent .employee was added to the
regression model of Table 2. Table 3 shows the results;•:the coefficient on the computer variable
at time t is indeed significantly positive, but the overall fit of the equation is little different from
Restructuring and Productivity, Blomberg and Steindel
Table 3: · Neoclassical Growth Model•with Hi-Tech Invesbnent 1980-1989
':i-;··¥·,{':i,,;/'s:.:1.,1;,
~lils/lUII.Q.~ !::at
CONSTANT
KIL
HI-TECH
HI-TECH(-!)
HI-TECH(-2)
HI-TECH(-3)
HI-TECH(-4)
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
Adjusted R2
,~, ,I
-3.25
0.22
0.09
-0.01
0.01
0.00
0.03
-0.41
-0.40
-0.40
-0.32
-0.26
-0.23
-0.20
-0.12
-0.08
-0.03
~
20.04
10.01
2.5 I
0.15
0.35
0.26
1.17
2.84
3.03
3.30
3.01
2.75
2.75
2.60
1.80
1.28
0.50
0.79
*The explanatory variables in the regressions include a constant.'time (1980, ...,1989) and industry
dummies. the log of the capital laborsratio (K/1.), and O to 4 lags of the ratio of real net information
processing c11uipmen1 per full-ume equivalent employee (HI-TECH).
that in Table 2. as the adjusted R2 just edges up from 0.77 to 0.79. 6 More importantly, there
is still a great deal of variation in productivity growth by industry left to explain. Indeed, a
listing of industries ranked by the swings in their residual productivity from the Table 2 equation
between 1980-84 and 1985-89 is not very different from a ~imilar ranking using the Table 3
6 Alan Krueger (1993) has uncovered evidence that people who work with computers tend
to earn higher salaries. < To the eittent that relative wages are associated with relative
productivities, and the real ,,stock of. computers is associated with the simple nose-count of
.. machines, the'se results are reminiscent of his.
16
Restructuring and Productivity, Blomberg and Steindel
Table 4A: Ranking or Industry by Tech. Progress (1985-89) - (1980-84): The Winners
~/-,•~•-i~:~i .
-~
"
Industry
Cred Ag. no banks
Metal Mining
Railroad
Textile
Apparel
Coal Mining
Rubber
Leather
Machinery
Hotels
Water Trans.
Stone & Glass
Lumber & Wood
Misc. Manu.
Primary Metals
Nonmetal. Min.
Rec. Serv.
Instruments
Furniture
Retail Trade
Fab. Metals
Trans. by Air
Electronics
Trans. Serv.
Other Trans.
Paper & Allied
Chemicals
Social Serv.
Trucking
Food
Personal Serv.
Local Transit
Telephone & Tele
Oil & Gas
~
z:,~ f:I.fliicw
1.26
0.92
0.60
0.48
0.40
0.35
0.34
0.29
0.27
0.26
0.25
0.25
.0.25
0.24
0.24
0.20
0.18
0.18
0.17
(l.15
0.15
0.14
0.14
0.14
O.IO
0.09
0.08
0.07
0.06
0.06
0.03
0.03
0.00
0.00
x.,,a ··f:Lfli ~c,rr.ii ~, u.,~z:-,,a
1.20
0.84
0.54
0.40
0.24
0.30
0.26
0.16
0.29
0.19
0.14
0.26
0.22
0.09
0.31
-0.13
0.25
0.22
0.04
0.21
0.02
0.17
0.22
-0.16
0.08
0.00
0.27
0.08
-0.02
0.11
0.09
0.07
0.33
0.02
*The explanatory variables in the regressions include a constant. time (1980,.... 1989) dummies. the log.of
the capital labor.ratio.(K/L). and/or O 10 4 lags or the ratio or real net information processing equipment
per lull-time equivalent employee CHI-TECH).
model. (See Table 4A-B.) Ho.wever. the average swing in the residual is somewhat smaller
17
Restructuring and Productivity, Blomberg and Steindel
Table-4B:
Ranking
by _
Tech.
Progress
(1980-84):
The
Losers
:f!.~;':.r'l\.·,:Y,'-"-"iiii_,
__
_ _ _of
_Industry
____
__
_ _ _(1985-89)
_ _ _ _•_
____
__
_ _ _ __
lndu'strv
Auto Repair
Wholesale Trade
Bus. Serv.
Petroleum & Coal
Motor Vech.
Misc. Repair
Printing
Construction
Radio & T.V.
Educational Serv.
Motion Pictures
Electric & Gas
Banking
Health Serv.
Insurance
Holding Co.
Insurance Agents
Security Brokers
Pipelines
Legal Serv.
Real Estate
Tobacco
z:,,a e.r:~~~,~
-0.03
-0.03
-0.06
-0.06
-0.10
-0.13
-0.19
-0.20
-0.21
-0.21
-0.25
-0.29
0 0.30
-0.31
-0.38
-0.45
-0.51
-0.58
-0.71
-0.85
-1.24
-1.28
. 'J:.WJ f.r:fi ~~~ ~£ tti- z:,,a
0.00
0.09
0.00
0.08
-0.19
-0.24
-0.28
-0.30
0.02
-0.28
-0.04
-0.25
-0.25
-0.19
-0.32
-0.28
-0.32
-0.44
-0.98·
-0.82
-1.04
-1.26
*The explanatory variables in the regressions include a constant, time (1980, ...,1989) and
industry dummies,-the log of:the capital, labor ratio {KIL), and 0 to 4 ;Jags of the ration of
·reah1et infonnation processing equipment per full-time equivalent employee (Hl~TECH).
when employing the Table 3 model.
As a second and more important step, we introduce actual measures of restructuring into
· the framework. We measure "financial" restructuring as net interest as a percent of capital and
· "real-side" restructuring as· the ·real value of ·merger and takeover transactions per full-time
·:.employee. : 'This·· "real-side". measure ·is calculated from both the perspective of the acquiring
18
Restructuring and Productivity, Blomberg and Steindel
industry and the targeted industry. 7
We use the "financial" measure to capture· the idea that firms inefficiently allocate debt
and equity prior to restructuring.
Conventional wisdom states that, in the course of a
restructuring, firms increase debt both to finance the restructuring itself and to resolve agency
issues within the firm that held back ·productive economies--bondholders may demand more
stringent performance goals in formal covenants in a high debt firm than do shareholders in a
firm with a diffuse ownership, and the switch to high debt levels has often been associated with
a concentration Qf ownership in the hands of management, who presumably then have a stronger
personal' interest in the best performance of the firm than when they were working on the behalf
of outside shareholders; Fonhese reasons, increases in net interest could be associated with more
efficient production. 8
We use both "real-side" measures,because there are cogent arguments for using either the
acquiring or targeted industry as the appropriate measure of restructuring. Suppose, for example,
7
The "real-side" measures are extracted from the Brookings Historical Merger Data file.
The Brookings Historical Merger Data file is a listing of corporate merger and takeover
transactions from 1955 to 19119. The data was assembled by Blair (1993) who merged data from
the Federal Trade Commission and the University of Texas College of Business Administration.
We merge the relevant variables from each data set for 43 industries over the period 1977 to
19119. and employ many of our measures of restructuring from this data set.
8 1t is also possible for debt to decrease following a restructuring because the firm had
previously been issuing too much debt for-efficient production--fears of bankruptcy, for
instance. may have hampered innovation. It is possible. then, that the measure of financial
•. restructuring that is associated with improved productivity is a large change of any .sign, not
just an-increase .. To.evaluate<this possibility. we included non-linear measures of "financial"
stfucturing·in our empirical specifications: None of the qualitative results were sensitive to
these changes.
19
Restructuring and Productivity, Blomberg and Steindel
· an automobile company targets a steel company for acquisition. The automobile company may
be targeting the steel -company because the steel -company's inefficiency, has continued to. raise
·· the·automobile company's production costs, thereby:providing an ,incentive for the-steel company
to be acquired. ·However; it could also be·the case that because of its own poor productivity, the
automobile company -is 'losing its competitive edge and so it acquires a more efficient steel
company to lower production costs. In the end, we find the results from the statistical analysis
are qualitatively the same regardless of the measure chosen.
We first consider the distribution of productivity and "real-side" restructuring. From 1977
to 1989, average real productivity increased by 1.3 percent per year whereas the average real
· value of restructurings increased by 19 percent annually over the same period.
Chart I then separates the -industries in our -sample into four di_stinct categories,based.on
the relationship between restructuring· and productivity. Industries that experienced high.levels
·of restructuring as well as positive productivity growth made·up 58 per cent of the entire
sample.9
Industries which experienced high levels of restructuring but lower productivity
·.,accounted-for the second largest portion of the sample, covering 32 percent. The pie chart
suggests that roughly 3/5 of the industries had a positive relationship between restructuring
whereas 2/5 did not. The distribution indicated in Chart I suggests there may be no systematic
relationship-between restructuring and productivity. This should not be surprising given that it
duplicates the impressionistic findings reponed earlier.
9The measure of restructuring chosen for the graphical analysis is taken from the "acquiring"
, iatid. not "target" industry; ,Howe1,1er. the. general qualitative results-hold regardless of definition.
20
:,a
Il
Chart 1: Industry Restructurlngsand Productivity
Lower Restructuring and Higher Productivity
5%
.,,l
<:l
~
Lower Restructuring and Lower Productivity
~
:.·
.s·
1:1:1
s;11
.
c,,
od
1
c,,
~
l
32%
Higher Restructuring and Lower Restructuring
Higher Restructuring and Lower Productivity
"'...
Restructuring and Productivity, Blomberg and Steindel
22
Charts 2-4 provide further impressionistic evidence that there is at best a weak relationship
between restructuring and productivity .. • These charts plot restructuring .activity versus
productivity, -where each square denotes a particular industry. The line plotted through these
points is the overalYfit between the variables. Chart 2 shows that the correlation between real side restructuring, as measured by the targeted industry, is practically non-existenL Charts 3 and
4 provide slightly more evidence of correlation, however, neither example is particularly
conclusive.
We now return to the neoclassical production function for more formalized analysis.
Recall that
We posited earlier that either L restructuring has· a direct.impact on production where ·
or 2.• 3. restructuring (R) influences the factors of production
where Z, is the vector of inputs.
We test 2.,3. by examining the relative importance of restructuring in the usage of capital
2
and labor. By employing simple x tests, we examine whether or not there are contemporaneous
or lagged effects of restructuring on each factor.
Restructuring and Productivity, Blomberg and Steindel
-
Productivity Growth
u)
l'I:)
......
......
0
0
·o
0
0
□
C.
C:
q
0
0
...
.
.....
...
..
..
Cf)
.i:,..
0
-i 0
...
...
...
..
....
.
....
.....
.....
..
C:
□
-
-I
D)
0
"""l
-i
co
-o
a.
<
~
C:
CD
CD
·······o· ··············f·············j'·····.-·······:··············~············· CD
...
.
C.
...
.....
..
.
□
(X)
S:: _o
CD
.....
(Cl
CD
(Q
0)
0
CD CD 0
0
Q.
~
CJ)
~
.....
<
<
CD
en
0
-o
DJ
---··
(")
........ : ............. ~ ............. ·I· ............. ;............ .
....
....
....
□:.
0
0
-a
"""l
0
□
I\)
0
0
0
□□
C d
3
0
01
.i:,..
0
0
0
0
0
□
C:
23
0
0
0
.....
········D ..
..
:
..
:
0
0
0
0
......
I\)
0
0
0
0
;..
:
<
-
D)
C:
··········+·············;··············[··············l············· CD
0
-s::
......
0
.
.
CD
"""l
.........: ............. (Q
:.
. . . . . . . . .. . . . . ......... -~ ............. :........ .... .........
.
.
..
..
... .
..
.:
.
.
CD
.
.
D
.
...
....
.....
.
..
..
...
..
.
...
...
.
"""l
(")
-::r
ll>
;:::::i.
I\:>
Restructuring and Productivity, Blomberg and Steindel
-.....
0
0
24
. Productivity Growth
.....
"'
0
0
0
0
0
.r:,.
c.:>
01
0
0
0
0
0
0
·o
q
q
:
O:
:
D
0
0
0
(J) 0
C: 0
3
-I
--
0
Ill
0)
.
..
.
.''
.
.r:,.
0
D
. . . .·r. ·. · . . T·. ··..... ··
"'0
0
.
T. ............
..
...
.
....
.
'
'
'
a.
-r .. · ··· ······
'
□
'
O
O
l
!
!
~
~
I
.
·1
!
!
0
-o
9. 0
'
i
)>
£.
ex,
0
D
mo0 ........... ..
Cl.
<
~
C:
CD
.....
0
CD
""'
l
:
.
···········
(
·1
.i
.i
_.
I
0
0
0
.....
.r:,.
0
0
0
0
.....
0)
0
0
0
0
·
.. -"""
CD
'
0
"'
C
.
·········
........ ·····:··············:··············:··············:····
..
.
'
'
0
<
en
"""
C
en
CD
a.
'
~o
CD
:
<"
-·
'<
)>
0
o?
-o
""'
<O
:
~
0
..0
(')
.Cl
C
,-+
'
. . .. . . . . . . . . . . .......... :............ .. !- ............ . :.. ·: .......... i.•...••....••
'
--a
'"""
0
'
...................... :· ·
!
~
[
'
.~
~.
! ..
<
Sl)
-CD
C
-s:
0
CD
.. '....... . "..... "· +' ...... '..... :'. "... "... "\" ........... ·t .......... .
.
.
.
..
...
...
..
...
....
...
...
.
"""
co
CD
"""
·o
:::r
ll>
;::::i
Cu
Restructuring and Productivity, Blomberg and Steindel
-
Productivity Growth
.....
0
0
.._,
25
.....
0
I\:)
0
0
0
0
c..:>
0
0
.i:,..
0
c,,
0
0
0
0
a.
D
----··
C:
q
0
<
:
:
0
(/)
C
c,,
3
z
0
~
0......
CD
-co
::,
CD
...a.
(/)
.
.
:
:
:.
.::
:.
:
<
CD
CJ)
"""'
C:
CJ)
zCD
-:::::,
CD
"""'
CD
CJ)
I ll
(/)
Ill
~
#0
......
CJ)
~
()
Ill
"O
CJ)
;::;:
Ill
'<
••.•········ ... ··········i··············i··············~··············!·············
.!
.~
.i.
.!.
:.
-I
'1J
0"""'
...a.
0,
-;:!:!.,
0
....
0
()
~
--·-
'"C
~
()
:::r
~
::I-
.j:l.
Restructuring and Productivity, Blomberg and Steindel
We· test 1. .by looking for ·significant increases ip productivity ·directly attributable to
-·• restructuring. 10 As· when •testing 2.3.,- we employ·x2 or. F-tests .to·examine whether or- not there
is a contemporaneous or lagged effect of restructuring on productivity.
It is, of course, ·•possible that the relationship between restructuring and productivity is
more attributable to "reverse-causation." The "reverse-causation" hypothesis implies that the
typical firm in more productive industries provide higher actual and potential profits and are
therefore more likely to be the recipient of takeover activity. Alternatively, though, firms in less
productive industries might in general offer greater scope for earnings growth. To test the
possibility of.''.reverse-causality". we also look at a system of jointly determined .equations to see
whether restructuring influences productivity or vice-versa.
4. Empirical Result~
-. Before discussing how we,estimate these relationships it is first necessary to describe how
we ensure that our estimates ·are robust. Since the data is a cross-sectional time-series data set,
in ·order to correct ,for industry "fixed effect~", we remove the industry mean from each variable.
We also include various lags of the variables and "year" dummies to correct for serial correlation
and business-cycle effects.
Also, when appropriate, we permit calculation of a consistent
covariance -matrix allowing for-heteroscedasticity. We begin by .analyzing the relationship
between restructuring (e.g. TARGET, ACQUIRE, NETINT/K) and capital, K, and labor, L:
· 1OAlternatively. we could estimate the affect on the" Solow" residual rather than productivity
itself. The results are qualitatively the same regardless of method of estimation.
26
Restructuring and Productivity, Blomberg and Steindel
27
where Z denotes .the log input for industry i at time t, and R denotes the log real value of .the
merger transaction per full-time employee (either target or acquiring industry) or the log of net
interest as a percentage of capital. If R affects Z, then we would expect to reject the following
hypothesis: 11
Table 5 reports the results from this exercise. The first column is the dependent variable,
Z, chosen in the estimation.--The second column reports the appropriate measure of restructuring.
The next two columns repon the appropriate test statistic accompanied by its p-value, and the last
column repons the calculated long-run elasticity of factor demand.
Table 5 shows that the acquired measure of restructuring and net interest as a percentage
. of capital seem to be important in :explaining.movements in capital, with p-values of 0.02- for·
ACQUIRE and 0'.09.'.for NETINT/K .. However, all of the restructuring variables seem to be
important in determining labor demand.
The · economic interpretation of these result~ is quite revealing.
In each case, the
restructuring elasticity of labor is negative, whereas only the "reaJsside" restructuring of elasticity
of capital is positive.
Taken literally, this means that "real-side" restructuring leads to• a
substitution of capital for labor. This also suggests that "financial" restructuring creates such a
,
11 Note •.that we are ·testing the value that all coefficients are zero; not merely the sum of
coefficients are zero.
Restructuring and Productivity, Blomberg and Steindel
. ,.,;~.
28
· Table 5: The Effects of Reslructuring on the Factors of Production*
.:.:.....;\':.•Ii~
Deueaclcat· Var ·
Other Var,
Test·SW
~-
ElmticiD'
CAPITAL
TARGET
ACQUIRE
NETINT/K
1.43
12.51
9.36
0.91
0.02
0.09
0.001
0.038
-0.113
LABOR
TARGET
ACQUIRE
NETINT/K
1l.63
26.02
33.97
0.02
0.00
0.00
-0.435
-0.349
-0.663
*The explanatory variables in the regressions include a constant, time trend, year dummies, and 1 to 5 lags
of CAPITAL, the log capital stock. and either 0 to 4 lags of ACQUIRE, the log real value of the acquired
industry per full-time employee. or Oto 4 lags of TARGET. the log real value of the targeted industry per
full-time employee. or 0 to 4 lags or NETINT/K. the log real value of net interest as a percentage of the
··· "' ·"GAPITAL.equa~ion. ,·,F""'•the; LABOR equation. the same variables were chosen except the lag length was
· reduced by· one and the log of the number of full-time employees was substituted as the dependent
variable.
-debt burden that firms must cut back on both· factors of production.. However, to put some
perspective on the magnitude of these effects, these results mean that if restructuring increases
by I percent. then labor usage falls •by .35 to .66 percent, while capital rises by, at most, a
modest .04 percent.
Reducing the demand for labor, cereris paribus, increases the capital labor ratio which
indirectly improves the level of labor productivity.
However, unless these changes are
productivity enhancing and not just taken to improve the firm's short-term earnings, productivity
growth may not improve over the long haul. Therefore, it is necessary to analyze the direct
impact of restructuring on productivity, controlling for these indirect affects on factor demand.
Restructuring and Productivity, Blomberg and Steindel
29
.· Table. 6: The 'Effect of Restructuring on Productivity*
·"' •:;;:·;,..;,~.£-·- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Restmct Yar
Test-SW
~
EWicitv
TARGET
ACQUIRE
NETINT/K
',.,.,,.
.1.55
1.69
4.72
0.46
0.43
0.09
-0.033
-0.041
-0.024
•Toe explanatory variables in the regressions include a constant. time trend, year dummies, and 1 lag of
CAPITAL, the log capital stock. 1 lag of LABOR, the log number of full-time employees, I lagged
dependent variable and either O to 1 lags of ACQUIRE, the log real value of the acquired industry per
full-time employee. or O 10 1 lags of TARGET. the log real value of the targeted industry per full-time
employee, or O 10 1 lags of NETINT/K, the log real value of net interest as a percentage of capital. Also
included were control variables such as the log real value of assets·of targeted or acquired industries per
full-lime employee.
_______________________
In this case, we estimate
where TIME denotes a linear time trend and q is the log of real output per full-time employee.
·Restructuring may take some time to affect productivity; hence, the lags. We test the hypothesis
that the coefficients associated with R are jointly significant.
·Table 6 ·reports,the result~ from this exercise.
The first column is the measure of
restructuring. the next two columns report the appropriate test statistic accompanied by its pvalue, while the fourth column reports the long run restructuring elasticity of production. Table
6 shows that neither "real-side" nor "financial-side" measures of·restructuring seems to be
extremely imponant in explaining movements in productivity. In fact, the elasticity of.each
measure of restructuring are negative, which implies . that restructuring may actually reduce
productivity. Furthermore, if we allow for even longer dynamic processes by increasing lag
,.•:•.lengths, the ·effect :becomes .even more. strongly negative.
Restructuring and Productivity, Blomberg and Steindel
A'possible criticism of these results is that,we have not considered the·possibility that•the
causation between restructuring and productivity runs in the opposite direction. In other words,
by estimating a production function- and not allowing the feedback ,fr.om production back .to
restructuring, our results may be biased. We formally tested this by employing Hausman ( 1978)
specification tests.
We ·tested the null that restructuring was independent of contemporaneous
and lagged productivity. We found that productivity significantly influenced restructuring as
measured by NETINT/K but had little influence on ACQUIRE or TARGET. Therefore, in order
to correct for this possible problem, we examine the joint relationship between corporate
restructuring and productivity.
By constructing the model in such a manner, we do not force any structure on the manner
in which restructuring and productivity are related. In this case, the estimated model is
R1c=Po+PiTIMB'.1t +~-o P2+jq.1t-J+~-o P,.,r'Bset1t-J+~ -1 Ps+JR.1c-1+B}t
where we examine the sign and significance of the coefficients associated with the endogenous
· ·variables. This system of equations is estimated using the Seemingly Unrelated Regressions
(SUR) estimation procedure.
Table 7 reports the key results from these systems. The fitst column is the dependent
variable of the relevant equation, the second column is the . measure of restructuring or
•. productivity included in the equation. The-Jast three columns report the appropriate test statistic
accompanied by its p-value and the sign of the sum of the coefficients of interest. Table 7 shows
30
Restructuring and Productivity, Blomberg and Steindel
.... ·., ..
31
Table 7: Restructuring and Productivity in the Jointly Determined System*
,·;.,!.,.:
'
.
•
,_
. "
.
'
'
•,
Deuendent Var,
Other End, Var.
Terxc~Stat
~
PRODUCTIVITY
TARGET
TARGET
PRODUCTIVITY
0.800
1.100
0.37
0.29
(Positive)
(Negative)
PRODUCTIVITY
ACQUIRE
ACQUIRE
PRODUCTIVITY
0.01 I
0.909
0.91
0.34
(Positive)
(Positive)
PRODUCTIVITY
NETINT/K
NETINT/K
PRODUCTIVITY
3.520
3.070
· 0.06
0.08
(Negative)
(Positive)
Siva ef Coeff.
*The explanatory variables in the regressions include a constant. time trend, year dummies, I to 2 lags of
CAPITAL. the log capitaLstock. 1 10 2 lags of the log of full-time employees and either O 10 I lags of
· Y''i\CQUtRE.'1he'log•,reuhvaluezof the acquired industry per full-time employee, or O 10 I lags of TARGET,
1he log real value of the targeted industry per full-time employee. or O to I lags of NETINT/K, the log
real value of net interest as a percentage of the capital stock. The ACQUIRE and TARGET equations are .
> identified by the log real value of the targeted industry .per full-time employee, or O to I lags of ASSET.
All other equations arc identified by including one more lagged dependent variable in that equation over
the other equation.
that no measure of restructuring is statistically different from zero at the .05 significance level
in any of· the productivity regressions.
NETINT/K is significant at the .IO level in the
productivity equation. however, the· sign of the sum of coefficients implies that financial
restructuring is prnductivity~rec/ucing not productivity-enhancing. However, the sign of the sum
of the coefficients are positive for the other measures. though not statistically different from zero.
It is also worthwhile to examine the restructuring equations. The sign of the coefficients
associated with productivity in these equations imply that -unproductive industries encourage
takeover activity; whereas productive industries are those industries more likely to acquire other
industries. However, we cannot reject the hypothesis that productivity has no significant impact
on-any "real-side" measure of restructuring. given the lack of significance associated with our
test-statistics.
·And. while there is some evidence that productivity leads to "financial"
Restructuring and Productivity, Blomberg and Steindel
restructuring, the null would be rejected at the -.05 and .0 l significance levels.
5. Alternative Hypothesis
Thus far, we have shown that restructuring-influences factors -of production but seems to
have little impact on productivity itself. It is conceivable that, restructuring plays a larger role
in making -labor markets more efficient rather than influencing the .goods market. As Jensen
(1989) argues, corporate restructuring achieves efficiencies by substituting "incentives and
compensation for direct monitoring by large bureaucratic staffs." If Jensen is correct, then as
industries restr,ucture themselves, compensation and layoffs should rise, creating an implicit
reward for good work and an implicit punishment for shoddy work.
•- Since ·we· already showed "that restructuring seemed to reduce employment, it is only
necessary to examine what restructuring does to wage growth. If restructuring were statistically
significant in changing wages. then··there may-be something to the· hypothesis that restructuring
improves the efficiency of the labor market. In the short run, restructurings presumably attempt
to align wages more with the value of marginal products--some workers gain, some lose. More
-fundamentally. though, a,more efficient firm -or industry might be presumed to see accelerated
growth in both productivity and wages (either because true underlying technological growth
improves or the entity is steadily gaining on state-of-the-art practice). However, if there is no
significant impact on wages, it suggests restructuring mightanot:hav.einfluenced efficiency.
To test the implication that restructuring influences wages, we estimate
E;.
Wu=Po +
0
P1+1R.1t-j+
E;.1 P,.1W.1t-j+e .!t
· ., where W is the growth rate of real compensation. If corporate .restructuring makes labor markets
more efficient, we would expect the coefficients associated with R to be positive and significantly
32
Restructuring and Productivity, Blomberg and Steindel
~~"-'::'1,::~_;:,,_ir~:-----------------------------------Table 8: The Effect of Restructuring on Real Compensation
Restruct Yqr
Tcst·Stat
~
Loar Run Eff,ct
TARGET
ACQUIRE
NETINT/K
3.17
2.58
4.51
0.67
0.76
0.47
0.007
0.002
-0.000
*The explanatory variables in the regressions include a constant, time trend, year dummies, and I to 4 lags
of WAGE. the growth in real compensation per employee, and either O to 4 lags of ACQUIRE, the log
real value of the acquired industry per full-time employee, or O to 4 lags of TARGET, the log real value
of the targeted industry per full-time employee. or O 10 4 lags of NETINT/K, the log real value of net
interest as a percentage of the capital stock.
different from zero. The results from this experiment are reported in Table 8.
Column one reports the appropriate measure of restructuring, the next two.columns report
the appropriate test statistic accompanied by its p-value and the last column reports the long run
. impact of restructuring on real wage growth; .'The long run effect from "financial" restructuring
·on compensation growth is negative. The coefficients .. associated with "real-side" restructuring
are positive but statistically insignificant.
6. Conclusions
Past research has found that restructuring has positive effects on firm profits and
shareholder wealth and some studies suggest a connection between certain types of restructurings
and plant or firm productivity. We find, however, that .the restructuring-productivity linkage is
virtually nonexistent at the jndustry level, at least for the indexes of productivity we used. Our
· result.~ need not be inconsistent with the plant-level studies; the productivity improvements found
in the earlier work could have been the fruit of firms shedding their costliest and least productive
, •! ?operatidns,and
concentrating ·technology. imprpvements at the best performing operations. If the
33
Restructuring and Productivity, Blomberg and Steindel
' least-productive operations remain at work under different ownership, the net result of the original
firm's restructuring on the industry's productivity in the short run could well be rero (in the long
fun, though, the emergence of more productive leaders in· an industry m;iy, of course, have. the
· effect of spurring growth throughouO. We do find, however, that restructuring has significant
impacts on the factors of production employed by firms. We find that restructuring led firms to
substitute capital for labor improving the labor productivity of all "dynamically" efficient firms.
As we noted at the outset, the implications of our results are unclear. Obviously, there
· , are many ways. to define and measure "restructuring" and it need not be the case that we made
· the· best choices. ·Furthermore, even if one would accept our results for the 1980s at face value,
it-may· well'be that more recent restructurings--often involving "downsizing" and reductions of
leverage--have common factors which will show-up more clearly in positive associations with
industry productivity.· Nevertheless. the evidence ·in this paper shows· that·the·restructuring
movement can not be easily associated with the revival of productivity growth in the .I 980s.
34
Restructuring and Productivity, Blomberg and Steindel
References
Amihud, Yakov ed:; 'Leveraged management buyouts: Causes and consequences Dow JonesIrwin,' Homewood: IL, 1989.
Baker, George P. and Karen H. Wruck, "Organizational Changes and Value Creation in
Leveraged Buyouts: The Case· of the O.M. Scott & Sons Co.", Journal of Financial
·
Economics, pp. 163-190, 1989.
Blair, Margaret, "Brookings Historical Merger Data File, 1955-89", The Brookings
lnstititution, 1993.
Boswell, James, The Life of Samuel Johnson, New York (Modern Library), no date.
· •, Caves, Richard, and Matthew Krepps, "Fat: The Displacement of Nonproduction Workers from
' · ,,... ,U;S::'Mmnrfacturing··lndustries," Brookings Papers: Microeconomics 2, pp. 227-288, 1993.
Hausman,.Jerry "Specification Tests in Econometrics," Econometrica, pp. 1251-1272, 1978.
Jensen, Michael, "Active Investors, LBO's and the Privatization of Bankruptcy,", Statement
before the House Ways and Means Committee, Journal of Applied Corporate Finance, p
35-44. 1989.
Kaplan, Steven. "The Effects of Management BuyouL~ on.,Operating Performance and Value,
Journal o/Financial El'lmomics, pp.217-54, 1989.
Krueger, Alan, "How Computers Have Changed the Wage Structure: Evidence from Microdata,
; 1984-1989", Quanerly Journal of Economics, pp. 33-60, 1993.
Lichtenberg Fnmk, and Donald Siegel. "The Effects of Leveraged Buyouts on Productivity and
Related Aspects of Firm Behavior." Journal of Financial Economics, pp. 165-94,
1990.
· McGucken, Robert, Nguyen, Sang, and-Arnold Reznek, "The Impact of Ownership Change on
Employment, Wages.and Labor Productivity in U;S. Manufacturing 1977-87," Center for
Economic Studies, U.S. Bureau of Census, mimeo ,I 994.
Oliner, Stephen and Dan Sichel, "Computers and Gutput Growth Revisited: How Big ls the
Puzzle." Brookings Papers: 2. pp. 273-334, I 994.
Smith, Abbie. "Corporate Ownership Structure and Performance: The Case of Management
1
1. Buyouts.' Journal ofFinanc,a/ Economics. pp. 143-64, 1990.
· Steindel, Charles, "Manufacturing Productivity and High-Tech Investment", Federal Reserve
35
Restructuring and Productivity, Blomberg and Steindel
Bank of New York Quanerly Review, pp39-47, 1992a.
Steindel, Charles, "Industry Productivity and High-Tech Investment:', Federal Reserve Bank of
New York Research Paper 9202, 1992b.
36
@FedFRASER