AGGREGATE GROWTH: WHAT HAVE WE LEARNED FROM MICROECONOMIC EVIDENCE? ECONOMICS DEPARTMENT WORKING PAPERS NO. 267, J Haltiwanger

Tags: productivity growth, reallocation, establishments, creative destruction, labour productivity, establishment, measurement error, effects, MICROECONOMIC EVIDENCE, employment, Haltiwanger, innovations, economic growth, decomposition method, endogenous growth, decomposition, economic growth and international trade, new innovations, entry terms, United States, output growth, AGGREGATE GROWTH, John Haltiwanger, Economics Department Working Papers, empirical findings, steel industry, productivity gains, labour productivity growth, continuing establishments
Content: ECO/WKP(2000)40 Unclassified
Unclassified
ECO/WKP(2000)40
Organisation de Coopйration et de Dйveloppement Economiques
OLIS : 12-Oct-2000
Organisation for Economic Co-operation and Development
Dist. : 20-Oct-2000
__________________________________________________________________________________________
English text only
ECONOMICS DEPARTMENT
AGGREGATE GROWTH: WHAT HAVE WE LEARNED FROM MICROECONOMIC EVIDENCE? ECONOMICS DEPARTMENT WORKING PAPERS NO. 267 by John Haltiwanger
English text only
Most Economics Department Working Papers beginning with No. 144 are now available through OECD's Internet Web site at http://ww.oecd.org/eco/eco. 96615 Document complet disponible sur OLIS dans son format d'origine Complete document available on OLIS in its original format
ECO/WKP(2000)40 ABSTRACT/RЙSUMЙ This paper provides a synthesis of what is known about the determinants of output growth based on studying microeconomic data sets. It starts with a summary of the theoretical explanations which help reconcile heterogeneous performance observed across establishments in the same sector. The related theoretical literature on creative destruction models of growth is also discussed. This is followed by a review of the recent empirical literature on the relationship between microeconomic and macroeconomic productivity growth. The final section discusses the main empirical findings, the caveats of interpretation and the main issues of interpretation underlying the relationship between reallocation and growth. JEL classification: D24, O12 Keywords: Industry productivity, reallocation, establishment level data ***** Ce papier document une synthиse de l'йtat des connaissances des dйterminants de la croissance sur la base de l'йtude de donnйes microйconomiques. Il commence par un rйsumй des explications thйoriques aidant а expliquer l'existence des performances hйtйrogиnes observйes entre les йtablissements au sein d'un mкme secteur. La littйrature thйorique correspondante sur les modиles de destruction crйatrice est йgalement discutйe. Suit une revue de la littйrature empirique rйcente а propos des relations entre la croissance de la productivitй aux niveaux microйconomique et macroйconomique. La derniиre section discute des principaux constats empiriques, les problиmes d'interprйtation sous-jacents а la relation entre rйallocation et croissance. Classification JEL : D24, O12 Mots-clйs : Productivitй sectorielle, rйallocation, donnйes individuelles d'йtablissement Copyright: OECD 2000 Applications for permission to reproduce or translate all, or part of, this material should be made to: Head of Publications Service, OECD, 2 rue Andrй Pascal, 75775 PARIS CEDEX 16, France. 2
ECO/WKP(2000)40
TABLE OF CONTENTS
AGGREGATE GROWTH: WHAT HAVE WE LEARNED FROM MICROECONOMIC EVIDENCE? .. 4 1. Overview.............................................................................................................................................. 4 2. Theoretical underpinnings ................................................................................................................... 5 3. Review of existing empirical evidence ................................................................................................ 8 3.1 Brief review of key micro patterns ............................................................................................. 8 3.2 Reallocation and aggregate productivity growth ...................................................................... 10 3.2.1 Measurement methodology .................................................................................................. 10 3.2.2 The evidence......................................................................................................................... 12 4. Caveats, questions and issues ............................................................................................................ 15 5. Concluding remarks ........................................................................................................................... 16 BIBLIOGRAPHY......................................................................................................................................... 22
Tables
1a.
Gross reallocation of employment, output, equipment and structures
1b.
Gross reallocation of employment and output for automobile repair shops
2.
A comparison of decompositions of aggregate productivity growth
3.
Sensitivity of decomposition results to business cycle and sector
3
ECO/WKP(2000)40
AGGREGATE GROWTH: WHAT HAVE WE LEARNED FROM MICROECONOMIC EVIDENCE? John Haltiwanger1
1.
Overview
1.
Recent research using establishment and firm level data has raised a variety of conceptual and
measurement questions regarding our understanding of aggregate growth.2 Several key, related findings are
of interest. First, in a well functioning market economy, there is large-scale, ongoing reallocation of
outputs and inputs across individual producers. Second, the pace of this reallocation varies over time (both
secularly and cyclically) and across sectors. Third, much of this reallocation reflects within rather than
between-sector reallocation. Fourth, there are large differentials in the levels and the rates of growth of
productivity across establishments within the same sector. The rapid pace of output and input reallocation
along with differences in productivity levels and growth rates are the necessary ingredients for the pace of
reallocation to play an important role in aggregate productivity growth. Indeed, the core finding in the
recent empirical literature exploring micro datasets to look at questions of aggregate growth is that
reallocation does contribute significantly to aggregate productivity growth. For example, for the
US manufacturing sector, roughly half of total factor productivity growth over the course of a decade can
be accounted for by the reallocation of outputs and inputs away from less productive to more productive
businesses.
2.
That reallocation contributes positively to growth is precisely what one would hope and expect
from a healthy, dynamic market economy. Of course, the idea that productivity growth in a market
economy invariably involves restructuring and reallocation across producers is not new. For example,
Schumpeter (p. 83, 1942) coined the term, "creative destruction", which he described as follows:
"The fundamental impulse that keeps the capital engine in motion comes from the new consumers' goods, the new methods of production and transportation, the new markets..[The process] incessantly revolutionizes from within, incessantly destroying the old one, incessantly creating a new one. This process of Creative Destruction is the essential fact of capitalism."
1.
University of Maryland, Bureau of the Census, and NBER. The analyses and results presented in this paper
are attributable to the author and do not necessarily reflect concurrence by the U.S. Bureau of the Census.
2.
Empirical papers of relevance that focus on the connection between aggregate and micro productivity
growth include: i) for the United States: Baily et al. (1992); Baily et al. (1996, 1997); Bartelsman and
Dhrymes (1994); Foster et al. (2000); Dwyer (1995, 1997); Haltiwanger (1997); and Olley and Pakes
(1996); ii) for other countries: Tybout (1996); Aw et al. (1997); Liu and Tybout (1996); and Griliches and
Regev (1995). Much of this paper draws heavily from Haltiwanger (1997); Davis and Haltiwanger (1999);
and Foster et al. (2000).
4
ECO/WKP(2000)40
3.
However, what is new in the emerging empirical literature is the growing availability of
longitudinal establishment level data that permit characterisation and analysis of the reallocation across
individual producers within narrowly defined sectors and, in turn, the connection of this reallocation to
aggregate productivity growth. From these new datasets striking findings have emerged. An especially
striking aspect of these finding are the respective magnitudes - the magnitude of the reallocation within
narrowly defined sectors, the magnitude of the productivity differences that seemingly occur across
businesses in the same narrowly defined sector and the resulting magnitude of the contribution of
reallocation to growth.
4.
In spite of these striking findings, it would be wrong to conclude immediately that economies that
exhibit a greater pace of reallocation are inherently more efficient. It would also be wrong to suspect an
immediate prediction of a simple monotonic empirical relationship between the pace of reallocation and
economic growth. If nothing else, different economies (across time or country) may be experiencing
different aggregate and structural shocks. Moreover, even in response to the same shocks, economies may
differ fundamentally in the nature of their market structure and institutions. In this vein, there are many
theoretical reasons why one might suspect that either the magnitude or the timing or the nature of the
reallocation process might be inefficient (see, e.g. Caballero and Hammour, 2000) due to market
imperfections and/or market institutions. Moreover, while the evidence suggests that reallocation does
contribute positively to growth, there are various indicators of inefficiencies in the pace and timing of
reallocation.
5.
The paper proceeds as follows. In Section 2, a summary of theories that can account for the
observed heterogeneous fortunes across establishments in the same narrowly defined sector is provided. In
addition, the related theoretical literature on creative destruction models of growth is discussed. This brief
discussion of theoretical underpinnings is of considerable help in putting the results on the relationship
between micro and macro productivity growth into perspective. In Section 3, a review and synthesis of the
recent empirical literature is presented. Section 4 contains some discussion of caveats and questions and
results that raise questions about the connection between reallocation and growth. Section 5 provides
concluding remarks.
2.
Theoretical underpinnings
6.
This section draws together theories and evidence related to the reasons for cross-sectional
heterogeneity in establishment-level and firm-level outcomes. A pervasive empirical finding in the recent
literature is that within-sector differences dwarf between sector differences in behaviour. For example,
Haltiwanger (Table 1, 1997) shows that 4-digit industry effects account for less than 10 per cent of the
cross-sectional heterogeneity in output, employment, capital equipment, Capital Structures, and productivity
growth rates across establishments.
7.
The magnitude of within-sector heterogeneity implies that idiosyncratic factors dominate the
determination of which establishments create and destroy jobs and which establishments achieve rapid
productivity growth or suffer productivity declines. An examination of the literature suggests that the
following may account for establishment-level heterogeneity: uncertainty; establishment-level differences
in managerial ability, capital vintage, location and disturbances; and diffusion of knowledge. Starting with
the first of these, one likely reason for heterogeneity in establishment-level outcomes is the considerable
uncertainty that surrounds the development, adoption, distribution, marketing and regulation of new
products and production techniques. Uncertainty about the demand for new products or the
cost-effectiveness of alternative technologies encourages firms to experiment with different technologies,
goods and production facilities (Roberts and Weitzman, 1981). Experimentation, in turn, generates
differences in outcomes (Jovanovic, 1982; and Ericson and Pakes, 1989). Even when incentives for
5
ECO/WKP(2000)40
experimentation are absent, uncertainty about future cost or demand conditions encourages firms to differentiate their choice of current products and technology so as to optimally position themselves for possible future circumstances (Lambson, 1991).
8.
Another possible reason is that differences in entrepreneurial and managerial ability lead to
differences in job and productivity growth rates among firms and establishments. These differences include
the ability to identify and develop new products, to organise production activity, to motivate workers, and
to adapt to changing circumstances. There seems little doubt that these and other ability differences among
managers generate much of the observed heterogeneity in establishment-level outcomes. Business
magazines, newspapers and academic case studies (e.g. Dial and Murphy, 1995) regularly portray the
decisions and actions of particular management teams or individuals as crucial determinants of success or
failure. High levels of compensation, often heavily skewed toward various forms of incentive pay
(Murphy, 1997), also suggest that senior managers play key roles in business performance, including productivity and job growth outcomes.3 A related idea is that it takes time for new businesses to learn
about their abilities.
9.
Other factors that drive heterogeneity in establishment-level productivity, output and input
growth outcomes involve establishment - and firm-specific location and disturbances. For example, energy
costs and labour costs vary across locations, and so do the timing of changes in factor costs. Cost
differences induce different employment and investment decisions among otherwise similar establishments
and firms. These decisions, in addition, influence the size and type of labour force and capital stock that a
business carries into the future. Thus, current differences in cost and demand conditions induce
contemporaneous heterogeneity in establishment-level job and productivity growth, and they also cause
businesses to differentiate themselves in ways that lead to heterogeneous responses to common shocks in
the future. The role of establishment-specific shocks to technology, factor costs and product demand in
accounting for the pace of job reallocation has been explored in Hopenhayn (1992), Hopenhayn and
Rogerson (1993), and Campbell (1997).
10.
Slow diffusion of information about technology, Distribution channels, marketing strategies, and
consumer tastes is another important source of establishment-level heterogeneity in productivity and job
growth. Nasbeth and Ray (1974) and Rogers (1983) document multi-year lags in the diffusion of
knowledge about new technologies among firms producing related products. Mansfield et al. (1981) and
Pakes and Schankerman (1984) provide evidence of long imitation and product development lags.4
11.
Part of the differences across establishments may reflect the vintage of the installed capital.5
Suppose, for example, that New Technology can only be adopted by new establishments. Under this view,
entering technologically sophisticated establishments displace older, outmoded establishments and gross
output and input flows reflect a process of creative destruction. A related idea is that it may not be the
vintage of the capital but rather the vintage of the manager or the organisational structure that induces
establishment-level heterogeneity (see, e.g. Nelson and Winter, 1982).
3.
Many economic analyses attribute a key role to managerial ability in the organisation of firms and
production units. Lucas (1977), for example, provides an early and influential formal treatment.
4.
Knowledge diffusion plays a key role in many theories of firm-level dynamics, industrial evolution,
economic growth and international trade. See, for example, Grossman and Helpman (1991); Jovanovic and
Rob (1989); and Jovanovic and MacDonald (1994).
5.
See Aghion and Howitt (1992); Caballero and Hammour (1994, 1996); Campbell (1997); Stein (1997);
Cooley et al. (1996); and Chari and Hopenhayn (1991).
6
ECO/WKP(2000)40
12.
These models of establishment-level heterogeneity are closely related to the theoretical growth
models emphasising the role of creative destruction. Creative destruction models of economic growth
stress that the process of adopting new products and new processes inherently involves the destruction of
old products and processes. Creative destruction manifests itself in many forms. An important paper that
formalises these ideas is Aghion and Howitt (1992). They consider a model of endogenous growth where
endogenous innovations yield creative destruction. Specifically, the creator of a new innovation gets some
monopoly rents until the next innovation comes along at which point the knowledge underlying the rents
becomes obsolete. The incentives for investment in R&D and thus growth are impacted by this process of creative destruction.6
13.
An alternative but related type of creative destruction growth model mentioned above as a source
of establishment-level heterogeneity is the vintage capital model. One form of these models (Caballero and
Hammour, 1994; and Campbell, 1997) emphasises the potential role of entry and exit. If new technology
can only be adopted by new establishments, growth occurs only via entry and exit, and this requires output
and input reallocation. An alternative view is that new technology is embodied in new capital (e.g. Cooper,
et al., 1997), but that existing establishments can adopt new technology by retooling. Under this latter
view, both within-establishment and between-establishment job reallocation may be induced in the
retooling process. If, for example, there is skill biased technical change, the adoption of new technology
through retooling will yield a change in the desired mix of skilled workers at an establishment. In addition,
there may be an impact on the overall desired level of employment at the establishment.
14.
In all of these creative destruction models, the reallocation of outputs and inputs across producers
plays a critical role in economic growth. In these models, stifling reallocation stifles growth. It is important
to emphasise, however, that there are many forces that may cause growth and the pace of reallocation to
deviate from optimal outcomes. As mentioned above in the context of Aghion and Howitt (1992), a generic
problem is that agents (firms, innovators, workers) do not internalise the impact of their actions on others.
In an analogous manner, Caballero and Hammour (1996) emphasise that the sunkness of investment in
new capital implies potential ex post hold-up problems that yield several harmful side effects. They
explore the hold-up problem generated by worker-firm bargaining over wages after the firm's investment in specific capital.7 One of the interesting aspects of their analysis is that it yields predictions about the
patterns of reallocation that will emerge in efficient and inefficient environments. One prediction is quite
intuitive and relatively easy to evaluate: economies with efficient reallocation should exhibit simultaneous
creation and destruction - observation of a long lag between an increase in destruction and creation is a
sign typically that market imperfections and/or institutional factors are distorting the process.
6.
Growth may be more or less than optimal since there are effects that work in opposite directions. On the
one hand, appropriability and intertemporal spillover effects make growth slower than optimal. The
appropriability effect derives from the fact that, in their model, research on new innovations requires
skilled labour as does the production of the intermediate goods where new innovations are implemented. A
fixed supply of skilled labour implies that skilled labour earns part of the returns from new innovations.
The inability of the research firms to capture all of the value from the innovations reduces their incentives
to conduct research. The intertemporal spillover effect derives from the fact that current and future
innovators derive benefits (i.e. knowledge) from past innovations but do not compensate past innovators
for this benefit. The fact that private research firms do not internalise the destruction of rents generated by
their innovation works in the opposite direction. This business stealing effect can actually yield too high a
growth rate. They also find, however, that the business stealing effect also tends to make innovations too
small.
7.
Indeed, Blanchard and Kremer (1997) argue that for transition economies, such hold-up problems are
potentially severe enough that the restructuring process is better described as "disruptive destruction"
rather than creative destruction.
7
ECO/WKP(2000)40
15.
A related point is that, even though reallocation may be vital for growth, there are clearly losers
in the process. The losers include the owners of the outmoded businesses that fail as well as the displaced
workers. This observation is important in its own right but also potentially related to sources of distortions
in the pace of reallocation. One question is whether markets are incomplete in terms of insurance and
related contingent claims - the inability of losers in this process to insure against idiosyncratic risks can be
a source of distortion. Moreover, the incompleteness of markets generates pressures for institutional
intervention to overcome the market limitations. Put differently, barriers to the reallocation process can
emerge through a variety of interventions in product, labour, trade, and credit markets that are rationalised
in terms of protecting against the potential losses to those that would be adversely affected in the
reallocation process.
3.
Review of existing empirical evidence
16.
The theoretical literature on creative destruction as well as the underlying theories of
heterogeneity characterise technological change as a noisy, complex process with considerable
experimentation (in terms of entry and retooling) and failure (in terms of contraction and exit) playing
integral roles. In this section, we review the evidence from the recent empirical literature that has
developed in parallel with the theoretical literature. We conduct this review in two parts: first, we provide a
brief review of the micro patterns of output, input and productivity growth; second, we consider the
aggregate implications of these micro patterns. Our review of micro patterns is brief since we regard the
results discussed in this section as well-established and there are excellent recent survey articles by
Bartelsman and Doms (1997) and Caves (1997) that cover much of the same material in more detail.
Moreover, it is the aggregate consequences of these micro patterns that are more open to debate and there
are a number of measurement/conceptual issues that generate the variation that is found across studies on
this dimension.
3.1 Brief review of key micro patterns 17. We begin our review by briefly summarising a few key patterns that have become well-established in this literature. They are: 18. Large-scale reallocation of outputs and inputs within sectors: The rate of within-sector reallocation of output and inputs is of great magnitude. Davis and Haltiwanger (1999) summarise much of the recent literature on gross job flows; they note that in the United States, more that one in ten jobs is created in a given year and more than one in ten jobs is destroyed every year. Similar patterns hold for many other market economies. Much of this reallocation reflects reallocation within narrowly defined sectors. For example, Davis and Haltiwanger (1999) report that across a variety of studies only about 10 per cent of reallocation reflects shifts of employment opportunities across 4-digit industries. 19. It is important to emphasise that it is not only labour that is being reallocated. There is large-scale output and capital reallocation accompanying (or perhaps in some cases substituting for) the labour reallocation. Table 1 reports tabulations from Foster et al. (2000) with estimates of the gross expansion and contraction rates of employment, output and capital (structures and equipment) over the 1977-87 period for US manufacturing. The rates of output and input expansion (contraction) are measured as the weighted average of the growth rates of expanding (contracting) establishments including the contribution of entering (exiting) establishments using the methodology of Davis et al. (1996).8 The pace of gross output
8.
This methodology entails defining establishment-level growth rates as the change divided by the average of
the base and end year variable. The advantage of this growth rate measure is that it is symmetric for
8
ECO/WKP(2000)40
and input expansion and contraction is extremely large over the ten-year horizon. Expanding establishments yielded a gross rate of expansion of more than 40 per cent of outputs and inputs and contracting establishments yielded a gross rate of contraction in excess of 30 per cent of outputs and inputs. Net growth rate of output is higher than that of inputs (especially employment) reflecting the productivity growth over this period. A large fraction of the output and input gross creation from expanding establishments came from entry and a large fraction of the output and input gross destruction came from exit.
20.
Table 1 also includes the fraction of excess reallocation within 4-digit industries in each of these
industries. Excess reallocation is the sum of gross expansion and contraction rates less the absolute value
of net change for the sector. Thus, excess reallocation reflects the gross reallocation (expansion plus
contraction) that is in excess of that required to accommodate the net expansion of the sector. Following
Davis et al. (1996) (see pages 52 and 53 for a description of the methodology) excess reallocation rates at
the total manufacturing level can be decomposed into within and between-sector effects. The far right
column of Table 1 indicates that most of the excess reallocation at the total manufacturing level reflects
excess reallocation within 4-digit industries. Thus, the implied large shifts in the allocation of employment,
output and capital are primarily among producers in the same 4-digit industry.
21.
Entry and exit play a significant role in this process of reallocation. For annual changes, Davis
et al. (1996) report that about 20 per cent of job destruction and 15 per cent of job creation is accounted for
by entry and exit. For five-year changes, Baldwin et al. (1995) report that about 40 per cent of creation and
destruction are accounted for by entry and exit, respectively.9
22.
Persistent differences in levels of productivity. There are large and persistent differences in
productivity across establishments in the same industry (see Bartelsman and Doms, 1997) for an excellent
discussion). The differences themselves are large - for total factor productivity the ratio of the productivity level for the plant at the 75th percentile to the plant at the 5th percentile in the same industry is 2.4 (this is
the average across industries) - the equivalent ratio for labour productivity is 3.5. In analysing persistence,
many studies report transition matrices of establishments in the relative productivity distribution within
narrowly defined industries (see, e.g. Baily et al., 1992; and Bartelsman and Dhrymes, 1994). These
transition matrices exhibit large diagonal and near-diagonal elements indicating that establishments that are
high in the distribution in one period tend to stay high in the distribution in subsequent periods. In contrast,
establishment-level productivity growth rates exhibit an important transitory component. Baily et al.
(1992) and Dwyer (1995) present strong evidence of regression to the mean effects in productivity growth
regressions.
23. Low productivity helps predict exit: Many studies (e.g. Baily et al., 1992; Olley and Pakes, 1996; and Dwyer, 1995) find that the productivity level helps predict exit. Low productivity establishments are more likely to exit even after controlling for other factors such as establishment size and age. A related set of findings is that observable establishment characteristics are positively correlated with productivity including size, age, wages, adoption of advanced technologies, and exporting (see, e.g. Baily et al., 1992; Doms et al., 1996; Olley and Pakes, 1996; and Bernard and Jensen, 1995). It has been more difficult to find correlates of changes in productivity. For example, Doms et al. (1996) find that establishments that
positive and negative changes and allows for an integrated treatment of entering and exiting establishments.
9.
The calculations in Baldwin et al. (1995) are an updated version of earlier calculations by Dunne et al.
(1989). The five-year gross flows and the shares accounted for by entry and exit are somewhat lower in the
later work for equivalent periods reflecting the improvement in longitudinal linkages in the Census of
Manufacturers over time.
9
ECO/WKP(2000)40
have adopted advanced technologies are more likely to be high productivity establishments but that the change in productivity is only weakly related to the adoption of such advanced technologies.
3.2 Reallocation and aggregate productivity growth
24.
Empirical analysis of the implications of the pace of reallocation and restructuring for
productivity dynamics has been recently provided by Baily et al. (1992), Olley and Pakes (1996),
Bartelsman and Dhrymes (1994), Dwyer (1995, 1997), Haltiwanger (1997), and Foster et al. (2000) using
establishment-level manufacturing data from the United States; Aw et al. (1997) using firm-level data from
Taiwan; Tybout (1996) and Liu and Tybout (1996) using data from Columbia, Chile, and Morocco; and Griliches and Regev (1995) using data from Israel.10 Virtually all of the studies consider some form of
decomposition of an index of industry-level productivity:
3 V S LW
H, HW HW
[1]
here Pit is the index of industry productivity, set is the share of establishment e in industry i (e.g. output share), and pet is an index of establishment-level productivity.
3.2.1 Measurement methodology
25.
The results that one obtains from decomposing [1] do turn out to be sensitive on some
dimensions to measurement methodology. To illustrate the sensitivity to measurement methodology, two
alternative decomposition methodologies are considered. The first decomposition method (denoted
method 1 in what follows) considered is a modified version of that used by Baily et al. (1992) and is given
by:11
3LW H& VHW SHW H& SHW 3LW VHW H& SHW VHW [2] H1 VHW SHW 3LW H; VHW SHW 3LW
where C denotes continuing establishments, N denotes entering establishments, and X denotes exiting establishments. The first term in this decomposition represents a within-establishment component based on establishment-level changes, weighted by initial shares in the industry. The second term represents a between-establishment component that reflects changing shares, weighted by the deviation of initial establishment productivity from the initial industry index. The third term represents a cross
10.
Baldwin (1995) presents some related analysis of the contribution of establishment turnover to productivity
growth for Canada but his methodology differs sufficiently from the rest of the literature that it is not easy
to integrate his work into this discussion.
11.
The first term in this decomposition (the "within component") is identical to that in Baily et al. (1992).
They essentially combined the second two terms by calculating a term based upon the sum of changes in
shares of activity weighted by ending period productivity. In addition, they did not deviate the terms in the
between and net entry terms from initial levels. As Haltiwanger (1997) points out, this implies that even if
all establishments have the same productivity in both beginning and end periods, the between component
and the net entry component in the Baily et al. (1992) decomposition will, in general, be non-zero. See
Haltiwanger (1997) for further discussion.
10
ECO/WKP(2000)40
(i.e. covariance-type) term. The last two terms represent the contribution of entering and exiting establishments, respectively.
26.
In this decomposition, the between-establishment term and the entry and exit terms involve
deviations of establishment-level productivity from the initial industry index. For a continuing
establishment, this implies that an increase in its share contributes positively to the between-establishment
component only if the establishment has higher productivity than average initial productivity for the
industry. Similarly, an exiting establishment contributes positively only if the establishment exhibits
productivity lower than the initial average, and an entering establishment contributes positively only if the
establishment has higher productivity than the initial average.
27.
This decomposition differs somewhat from others that have appeared in the literature in some
subtle but important ways. Key distinguishing features of the decomposition used here are: i) an integrated
treatment of entry/exit and continuing establishments; ii) separating out within and between effects from
cross/covariance effects. Some of the decompositions that appear in the literature are more difficult to
interpret because they do not separate out cross/covariance effects. For example, some measure the within
effect as the change in productivity weighted by average shares (in t and t-k - see method 2 below). While
the latter method yields a seemingly cleaner decomposition, it also allows the within effect to partially
reflect reallocation effects since it incorporates the share in period t. Another problem is in the treatment
of net entry. Some of the decompositions in the literature that consider net entry measure the contribution
of net entry via the simple difference between the weighted average of entrants and exiting establishments
productivity. Even if there are no differences in productivity between entering and exiting establishments,
this method yields the inference that net entry contributes positively to an increase (decrease) in
productivity growth if the share of entrants is greater (less than) the share of exiting establishments. There
are related (and offsetting) problems in the treatment of the contribution of continuing establishments.
28.
While this first method is our preferred decomposition, it is worthwhile to consider an alternative
decomposition closely related to that used by Griliches and Regev (1995). Consider, in particular, the
following alternative decomposition (denoted method 2 in the remainder of this paper):
3LW H& VH SHW H& SH3L VHW [3] H1 VHW SHW 3L H; VHW SHW 3L
where a bar over a variable indicates the average of the variable over the base and end year. In this decomposition, the first term is interpretable as a within effect that is measured as the weighted sum of productivity with the weights equal to the average (across time) shares. The second is interpretable as a between effect where the changes in the shares are indexed by the deviations of the average establishment level productivity from the overall industry average. In a like manner, the net entry terms are such that entry contributes positively as long as entering establishments are higher than the overall average and exiting establishments are lower than the overall average.
29. This second decomposition method is a modification of the standard within/between decomposition that is often used for balanced panels. The disadvantage of this method is that the measured within effect will now reflect in part cross/covariance effects (as will the measured between effect). However, this second method is apt to be less sensitive to measurement error in outputs or inputs relative to the first method as shown in equation [2]. Suppose, for example, we are considering labour productivity (e.g. output per man-hour) and that there is random measurement error in measured man-hours. Measurement error of this type will imply that establishments in a given period with spuriously high measured man-hours will have spuriously low measured productivity. Such measurement error will yield a
11
ECO/WKP(2000)40
negative covariance between changes in productivity and changes in shares (measured in terms of man-hours) and a spuriously high within-establishment effect under method 1. In a similar manner, consider the decomposition of multi-factor productivity using output weights. Random measurement error in output will yield a positive covariance between productivity changes and changes in shares and a spuriously low within-establishment effect under method 1. In contrast, the measured within effect from method 2 will be less sensitive to random measurement error in output or inputs since the averaging across time of the shares will mitigate the influence of measurement error.12
30.
Which method should be preferred? Absent measurement error, method 1 is clearly the preferred
method since it offers more detail and all of the terms are readily interpretable (there is not a confounding
of reallocation and within effects in the "within" term). Nevertheless, concerns about measurement error
are non-trivial. The latter concerns are likely to be especially of concern for analyses using high frequency
data where output and employment growth rates are apt to be especially noisy.
31.
In reviewing evidence in the next section, we restrict ourselves to comparing the components of
the evidence that use sufficiently similar methodology to permit a reasonable comparison. As will become
clear, this implies that for many studies we do not have and cannot compare all of the terms in the above
decompositions.
3.2.2 The evidence
32.
Using establishment-level data, the industry index and its components can be constructed for
measures of labour and multi-factor productivity. Many studies have used something analogous to one of
the methodologies above to decompose the time series changes in aggregate (i.e. industry-level)
productivity into components that reflect a within component (holding shares fixed in some manner) and
other effects that reflect the reallocation of the shares across establishments including the impact of entry
and exit. Table 2 presents a summary of results from a variety of studies using different countries, time
periods, frequency of measured changes, productivity concepts (i.e. multi-factor vs. labour) and
measurement methodologies (see the discussion in the prior subsection).13 The differences along these
many dimensions make fine comparisons difficult so the objective in considering the alternative studies is
to consider broad patterns. In the next section, some of these methodological issues are considered in more
detail. For now, an attempt is made to compare studies on dimensions that are relatively easy to compare.
33. One core aspect that is roughly comparable across studies is the contribution of the within establishment contribution to aggregate productivity growth. Even for this measure, there are differences in the methodology along a number of dimensions. These include whether the measure of productivity is multi-factor or labour, whether the share is based on output or employment weights, and whether the share is based on the initial share at the base period or the average share (averaged over base and end period). The latter difference often reflects a study using method 1 vs. method 2 in the analysis.
34. The fraction of within-establishment contribution to multi-factor productivity growth ranges from 0.23 to 1.00 across studies, while the fraction of the within-establishment contribution to labour productivity growth ranges from 0.79 to 1.20 across studies. At first glance, it appears to be difficult to draw conclusions even in broad terms about whether the within-establishment contribution is large or
12.
This discussion focuses on simple classical measurement error. There may be other forms of non-random
measurement error that are important in this context.
13.
In the case of Taiwan, a simple average (or simple median) of the industry-level results reported in the Aw
et al. (1997) paper is presented.
12
ECO/WKP(2000)40
small. However, careful examination of the individual studies indicates that this variation is due in part to there being systematic sensitivity to time period, frequency, and differences between the consequences of output vs. labour reallocation.
35.
Another dimension that many of the studies are comparable is the contribution of net entry to
productivity growth. The results here range from 0.25 to 0.26 for multi-factor productivity and from 0.02
to 1.00 for studies of labour productivity. Again, it would first appear that it would be difficult to draw
strong conclusions. Moreover, since the within plus the net entry components for some cases sum to
greater than one, this implies the reallocation among the continuing establishments is in those cases
working in the opposite direction. As we shall see, like for the within-establishment effect, the net entry
effect differences across studies reflect some systematic patterns.
36.
We begin by considering the contribution of establishment entry and exit to these aggregate
productivity dynamics. From Table 2, we see that one important factor is the horizon over which the
productivity growth is measured. By construction, the share of activity accounted for by exits in the base
year and entrants in the end year are increasing in the horizon over which the base and end year are
measured. At an annual frequency, we observe that the share of employment accounted for by exits in the
United States in the year t-1 is only 0.02 and by entrants in year t is only 0.01. In contrast, at a ten-year
horizon, the share of employment accounted for by establishments in the United States in year t-10 that
ultimately exit over the ten years is 0.28 while the share of employment accounted for by establishments in
year t that entered over the ten years is 0.26. These results imply that the contribution of any differences in
productivity between entering and exiting establishments will be greater for changes measured over a
longer horizon.
37.
The influence of the horizon also is likely to impact the observed productivity differences
between exiting establishments in the base year and entering establishments in the end year via selection
and learning effects. That is, one-year-old establishments are likely to have on average a lower productivity
than ten-year-old establishments because of selection and learning effects. Many studies (e.g. Olley and
Pakes. 1996; Liu and Tybout, 1996; Aw et al., 1997; Foster et al., 2000) present results suggesting that
selection and learning effects play an important role. The results in Table 2 reflect this in that the relative
productivity of entering establishments in the end-year to exiting establishments in the base year is increasing for changes measured over a longer horizon.14
38. Putting these results on entry and exit together helps account for the finding that studies that focus on high frequency variation (e.g. Baily et al., 1997; and Griliches and Regev, 1995) tend to find a small contribution of net entry to aggregate productivity growth while studies over a longer horizon find a large role for net entry (e.g. Baily et al., 1996; Haltiwanger, 1997; Foster et al., 2000; and Aw et al., 1997). There are two main factors that contribute to this sensitivity to frequency: i) entering and exiting businesses at high frequencies account for a small share of activity while entering and exiting businesses at lower frequencies account for a non-trivial share of activity; ii) entering businesses enter at higher productivity than exiting businesses but selection/learning effects imply that the productivity gap at the point of simultaneous exit/entry is small but grows over time.
39. Now consider the sensitivity to business cycles. Table 3 presents a few selected results from different time periods from the Baily et al., (1992) and Foster et al. (2000) studies. For the 1977-82 and 1987-92 periods, the within-establishment contribution for manufacturing in general is negative for both studies reflecting the fact that, while there is modest overall productivity growth over this period, its source
14.
Although the earlier vintage arguments suggest that it may be that younger establishments should have
higher productivity. While such vintage effects may be present, the evidence clearly suggests that the
impact of selection and learning effects dominate.
13
ECO/WKP(2000)40
is not the within-establishment component. In contrast, for the 1982-87 period the within-establishment contribution is large and positive during a period of robust productivity growth. This apparent sensitivity to the business cycle (1982 was during a severe slump in US manufacturing) is interesting in its own right. These results suggest that overall productivity is less procyclical than within-establishment productivity. The inference is that reallocation effects tend to generate a counter-cyclical "bias" and thus recessions are times that the share of activity accounted for by less productive establishments decreases either through contraction or exit. Baily et al. (1997) provide a more extensive analysis of the role of reallocation for the cyclical behaviour of productivity and present considerable evidence that the within-establishment component of aggregate productivity is more procyclical than the aggregate itself.
40.
Now consider the differences between the results decomposing labour productivity and
multi-factor productivity. For the contribution of net entry, the decompositions of multi-factor and labour
productivity have very similar findings. For the contribution of within establishment, the decompositions
are very different. The reason for these differences in the respective contributions to the total depends in
part on the role of reallocation among continuing establishments. That is, output reallocation across
continuing establishments is positively correlated with multi-factor productivity growth but labour
reallocation across continuing establishments is weakly or even negatively correlated with labour
productivity growth. Foster et al. (2000) show that for continuing establishments there are strong positive
correlations between output growth and employment growth, between multi-factor and labour productivity
growth, and between multi-factor productivity growth and output growth. However, there is a zero or
sometimes even a strong negative correlation for continuing establishments between labour productivity
growth and employment growth.
41.
These results show that it is inappropriate to infer that all or even most job reallocation reflects
the movement of employment from less productive to more productive businesses. Instead, employment
downsizing often accompanies large productivity gains. For example, as described in Davis and
Haltiwanger (1999), the US steel industry underwent tremendous restructuring during the 1970s and 1980s.
Much of this restructuring involved a shift from large, integrated mills to more specialised mini mills.
Entry and exit played a major role, but the restructuring of the industry also involved the retooling of many
continuing establishments. The employment-weighted mean number of workers at US steel mills fell from
7000 in 1980 to 4000 in 1985. Baily et al. (1996) find that continuing establishments in the steel industry
experienced substantial productivity gains while downsizing. Moreover, the downsizing episode in the
early 1980s was followed by dramatic productivity gains in the steel industry in later years (see, Davis
et al., 1996).
42. It is useful to compare and contrast the results on the reallocation involving net entry with those involving reallocation across continuing establishments. With entry and exit, labour and output (and capital) reallocation go hand-in-hand. Exit is clearly a sign of poor performance and both low labour productivity and low multi-factor productivity help predict exit. However, the same is not the case for output and labour reallocation across continuing establishments. Indeed, Foster et al. (2000) find that the correlation between labour productivity growth and increases in capital intensity to be quite large for continuing establishments. Thus, observing that a business is downsizing in terms of workers does not inherently imply a sign of distress but may reflect a restructuring involving increasing capital intensity (and perhaps skill intensity as suggested in the analysis of Dunne et al., 1997) with associated increases in labour productivity.
43. To summarise the main points from the evidence on the contribution of reallocation to aggregate productivity growth and fluctuations:
14
ECO/WKP(2000)40
- The contribution of net entry to aggregate productivity growth is disproportionate and increases over the horizon for which the change is being measured. The latter effect reflects both increasing shares of entry/exit over longer horizons and learning/selection effects. - Output reallocation among continuing establishments contributes positively to aggregate multi-factor and labour productivity growth. Labour reallocation among continuing establishments, while correlated with output reallocation, does not contribute positively to labour productivity growth as downsizing of a continuing business is often associated with rising labour productivity. Labour reallocation from net entry does contribute positively to labour productivity growth. - Reallocation yields a counter-cyclical "bias" to measured fluctuations in aggregate productivity growth at high frequencies. Put differently, the average continuing establishment exhibits even greater procyclicality in productivity than would be suspected from the aggregate data.
4.
Caveats, questions and issues
44.
The general theme thus far is that growth is a noisy process with much trial and error. Economic
growth inherently involves a continuous pace of restructuring and reallocation as businesses try out new
techniques, new products, and new locations and adapt to a continuously changing environment. The
resulting churning of businesses and outputs and inputs is large - moreover this churning contributes
significantly to growth. The policy implications at first glance appear clear - create a market environment
where successful innovators can indeed succeed and unsuccessful businesses can fail.
45.
The problem is that market imperfections (and in turn market institutions) may yield
inefficiencies in this process of reallocation. As discussed in section II and emphasised by Caballero and
Hammour (2000), the continuous process of reallocation has many potential pitfalls. The largest problem is
that reallocation involves businesses between themselves and with workers engaging in a variety of
relationships with a high degree of specificity. By specificity, we mean that some of the joint value of a
relationship is specific to that particular relationship and would be lost if the parties terminated their
relationship. The specific investments that firms and workers make in creating jobs and making matches
are some of the factors that yield such specificity. The problem with such relationships is that there is an
ex post hold-up problem and the question is whether the market structure and institutions yield an efficient
solution to the hold-up problem.
46. Caballero and Hammour (2000) argue that various countries and parts of the world have experienced problems in the institutional and market structure along these lines with significant adverse aggregate consequences. For example, many post-communist countries have seen their potential for growth and restructuring stifled by an underdeveloped legal and institutional environment. In Western Europe, labour-market regulation stifles this reallocation process. The crises in emerging markets can be attributed to economic systems with a lack of transparency and lax corporate governance standards. In all of these cases, the limitations of the institutional/market structure stifle and distort the ongoing reallocation process (and thus the ability to tap new technological developments and to adapt to a changing environment).
47. Analysis and evidence of the efficiency aspects of the reallocation process are still in their infancy. Davis and Haltiwanger (1999), Hopenhyn and Rogerson (1993), and Mortensen (1994) provide calibrations/simulations of the distortions in the magnitude of reallocation that can occur from various labour-market interventions. As noted in Section II, one indicator of efficiency that Caballero and Hammour (1996, 2000) emphasise is the presence of synchronised creation and destruction. The evidence
15
ECO/WKP(2000)40
for the United States suggests high rates of creation and destruction in the same period regardless of industry/time period with most workers involved in the reallocation process experiencing direct job-to-job movements - both of the features are suggestive of an efficient environment. However, over the course of a business cycle, the United States exhibits sharp spikes in job destruction in recessions followed by long and slow recoveries involving an increase in job creation. The staggered nature of creation/destruction at business cycle frequencies is a potential sign of inefficiency even for the United States. For other western economies, Davis and Haltiwanger (1999) survey the evidence on job creation/destruction. In their Figure 3.5, they show patterns of annual creation and destruction rates for eight countries. All of the countries exhibit creation and destruction being inversely correlated and a surge of destruction leading more gradual increases in creation over the course of several years.
48. Haltiwanger and Vodopivec (2000) provide some evidence on synchronisation/staggering of creation/destruction for transition economies. They compare and contrast the job and worker flows for a rapid reformer, Estonia, and a gradual reformer, Slovenia. Estonia made dramatic changes to open up product, labour, credit and trade markets quickly. For example, in terms of the labour market, they eliminated firing restrictions, established a very modest UI system, and allowed for decentralised wage determination with a very low minimum wage. In contrast, Slovenia retained restrictions of firing, established a quite generous UI system, and virtually all wages are determined via collective bargaining with a high minimum wage. Estonia experienced a surge in job flows from around zero in 1990 to almost 10 per cent creation and destruction rates by 1995 (similar to U.S. rates). Moreover, while destruction led creation and 1992 and 1993 were very difficult years in terms of a sharp downturn in aggregate activity, the increase in flows was relatively synchronised as evidenced by a dramatic surge in job to job movements for workers experiencing job destruction. In contrast, in Slovenia the magnitude of the creation/destruction flows has remained substantially less in Slovenia and destruction has led creation by a substantial period. Interestingly, both of these economies have done reasonably well in terms of recovering from the downturn in aggregate activity associated with the transition. Also, interestingly, Estonia has experienced a sharper downturn and a much sharper upturn than Slovenia.
49.
Overall, we are still in the early stages of the development of the theory, the appropriate
empirical specifications and the requisite data for the micro restructuring/reallocation approach towards
understanding economic growth. Limited data development is a major obstacle for understanding these
issues because the requisite data are measures of output and input reallocation (including the contribution
of entry/exit) linked to unemployment/worker flows and in turn linked to measures of business
performance. That is, to investigate these issues we need measures of the churning and to be able to relate
these to worker and firm outcomes. Very few countries have developed the data that would permit this type
of analysis. We need to develop such data so that both within-country studies and cross-country
comparisons become feasible. With such data a variety of open questions could be investigated including:
quantifying the magnitude and timing of reallocation, analysing the connection between the pace and
timing of reallocation with market and institutional structure, and analysing the connection between the
pace and timing of reallocation and key macro outcomes like output, consumption, employment and
productivity growth. Putting together the requisite data for within-country and cross-country study of these
issues should be a high priority.
5.
Concluding remarks
50. In this study we have focused on the contribution of the reallocation of activity across individual producers in accounting for aggregate productivity growth. A growing body of empirical analysis reveals striking patterns in the behaviour of establishment-level reallocation and productivity. First, there is a large ongoing pace of reallocation of outputs and inputs across establishments in market economies. Second, the pace of reallocation varies secularly, cyclically and by industry. Third, there are large and persistent
16
ECO/WKP(2000)40 productivity differentials across establishments in the same industry even in well functioning market economies. Fourth, entering establishments tend to have higher productivity than exiting establishments. Large productivity differentials and substantial reallocation are the necessary ingredients for an important role for reallocation in aggregate productivity growth. The emerging evidence suggests that the process of economic growth at the micro level is incredibly noisy and complex - there is a vast amount of churning as businesses and workers seek to find the best methods, products, locations and matches. This churning is an inevitable and vital component of economic growth. However, a number of conceptual and measurement issues remain. We don't have a clear understanding of the sources of within and between-country variation in the nature and magnitude of this churning, we don't have a clear understanding of the sources of within-industry heterogeneity in productivity levels and growth rates, and in turn we don't have a clear understanding of all of this variation for within and between-country outcomes like economic growth. A key obstacle for current work is that the requisite data development is still in early stages. 17
ECO/WKP(2000)40
Table 1a. Gross reallocation of employment, output, equipment and structures
Ten-year changes from 1977-87 for U.S. manufacturing
Measure
Creation (expansion) rate
Share of creation (expansion) due to entrants
Destruction (contraction) rate
Share of destruction (contraction) due to exits
Fraction of excess reallocation within 4-digit industry
Real gross output
49.4
0.44
34.4
0.61
0.80
Employment
39.4
0.58
45.8
0.62
0.75
Capital equipment
46.1
0.42
37.1
0.51
0.71
Capital structures
44.9
0.44
48.4
0.42
0.69
Table 1b. Gross reallocation of employment and output for automobile repair shops
Panel A: Five-year changes from 1987-92, employment
Real gross output
51.8
0.76
40.3
0.62
0.81
Employment
50.9
0.76
44.2
0.64
0.88
Notes: See text for details of construction of output, equipment and structures measures.
Source: Foster et al. (2000).
18
ECO/WKP(2000)40
Table 2. A comparison of decompositions of aggregate productivity growth
A. Multi-factor productivity decompositions
Country Frequency Sample period
Sectoral coverage
Weight used to calculate within establishment changes1
Average fraction from within establishment changes
Fraction of activity2 from entrants (t)
Fraction of activity from exits (t-k)
Relative productivity of births (t) to deaths (t-k)
Average fraction from net entry
Study
United States
5-year
1972-87 Selected mfg Output (t-k)
0.37
industries (23)
N/A
N/A
N/A
N/A
Baily, Hulten and
Campbell (1992)
United States
5-year
1977-92
All mfg
Output (t-k)
0.12
industries
N/A
N/A
N/A
0.25 Foster, Haltiwanger,
and Krizan (2000)
United States
10-year 1977-87
All mfg
Output (t-k)
0.48
industries
0.21
0.22
1.14
Taiwan3
5-year
1981-91 Selected mfg Output (avg. of (t- 0.94
N/A
N/A
N/A
industries (9)
k) and t)
(Median =
0.63)
Columbia
Annual
1978-86 Selected mg Input Index4 (avg
1.00
industries (5) of (t-k) and t)
N/A
0.05
1.05
0.26 N/A
Foster, Haltiwanger, and Krizan (2000) Aw, Chen and Roberts (1997)
N/A
Liu and Tybout
(1996)
19
ECO/WKP(2000)40
Table 2 (continued)
B. Labour productivity growth decompositions
Country Frequency Sample period
Sectoral coverage
Weight used to calculate within establishment changes
Average fraction from within establishment changes
Fraction of activity from entrants (t)
Fraction of Relative
Average
activity from productivity of fraction from
exits (t-k) births (t) to net entry
deaths (t-k)
Study
United States
5-year
1977-92
All mfg.
Man-hours (t-k)
1.00
Industries
N/A
N/A
N/A
0.20 Foster, Haltiwanger,
and Krizan (2000)
United States
5-year
1987-92
Selected Employment (t-k)
0.99
service sector
industries
0.32
0.39
1.11
1.12 Foster, Haltiwanger,
and Krizan (2000)
United
10-year 1977-87
All mfg.
Employment
0.77
States
Industries
(t-k)
0.21
0.25
1.34
0.31 Foster, Haltiwanger,
and Krizan (2000)
United States
Annual
1972-88
All mfg.
Man-hours (t-k)
1.20
Industries
0.01
0.02
1.03
0.02 Baily, Bartelsman
and Haltiwanger
(1997)
Israel
3-year
1979-88
All mfg. Employment (avg
0.83
0.08
0.06
1.20
Industries
of (t-k)
and t)
Griliches and Regev (1995)
1. Within contribution is measured as the weighted sum of establishment-level productivity growth as a fraction of aggregate index of productivity growth. In all cases, output above refers to gross output. 2. Activity is measured in the same units as weight (e.g. employment or output). 3. Simple average (and simple median) of industry-based results reported. 4. The input index is a geometric mean of inputs using estimated factor elasticities.
20
ECO/WKP(2000)40
Table 3. Sensitivity of decomposition results to business cycle and sector
Five-year frequency
1977-1982
1982-1987
1987-92
Sectoral coverage
Multi-factor productivity growth
Fraction from within establishment changes
Multi-factor productivity growth
Fraction from within establishment changes
Multi-factor productivity growth
Fraction from within establishment changes
Study
All mfg industries
2.43
-0.12
8.26
0.58
3.30
-.06
Foster, Haltiwanger, and Krizan (2000)
Selected mfg
2.39
industries (23)
-0.46
15.63
0.87
N/A
N/A
Baily, Hulten and Campbell (1992)
Notes: Weight for within calculation from both studies is initial gross output share for the establishment in each industry. Results aggregated across industries are based upon weighted average with weight for this purpose equal to the average of nominal gross output for the industry.
21
ECO/WKP(2000)40 BIBLIOGRAPHY AGHION, Philippe and Peter HOWITT (1992), "A model of growth through creative destruction", Econometrica, 60(2), pp. 323-351. AW, Bee Yan, Xiaomin CHEN and Mark J. ROBERTS (1997), "Firm-level evidence on productivity differentials, turnovers, and exports in Taiwanese manufacturing", NBER Working Paper No. 6235. BAILY, Martin Neil, Eric J. BARTELSMAN and John HALTIWANGER(1996), " Downsizing and productivity growth: myth or reality?" Small Business Economics, 8, pp. 259-278. BAILY, Martin Neil, Eric J. BARTELSMAN and John HALTIWANGER(1997), "Labor productivity: structural change and cyclical dynamics," NBER Working Paper No. 5503 (revised). BAILY, Martin Neil, Charles HULTEN and David CAMPBELL (1992), "Productivity dynamics in manufacturing establishments", Brookings Papers on Economic Activity: Microeconomics, pp. 187249. BALDWIN, John R. (1995), "Turnover and productivity growth" in Baldwin (ed.), The Dynamics of Industrial Competition, Cambridge: Cambridge University Press, pp. 208-38. BALDWIN, John R., Tim DUNNE and John HALTIWANGER (1995), "Establishment turnover in Canada and the United States," in Baldwin (ed.), The Dynamics of Industrial Competition, Cambridge: Cambridge University Press, pp. 119-52. BARTELSMAN, Eric J. (1995), "Using microdata for economic research and policy analysis," unpublished manuscript, Central Planning Bureau, Netherlands. BARTELSMAN, Eric J. and Mark DOMS (1997), "Understanding productivity: lessons from longitudinal micro datasets," mimeo. BARTELSMAN, Eric J. and Wayne GRAY (1996), "The NBER manufacturing productivity database," NBER Technical Working Paper No. 205. BARTELSMAN, Eric J. and Phoebus J. DHRYMES (1994), "Productivity dynamics: U.S. manufacturing establishments, 1972-86," Board of Governors of the Federal Reserve Board, Finance and Economics Discussion Series, No. 94-1. BERNARD, Andrew B. and J. Bradford JENSEN (1995), "Exceptional exporter performance: cause, effects, or both?" mimeo. BERNDT, Ernst R. and Jack E. TRIPLETT (eds.), "Fifty years of economic measurement", Studies in Income and Wealth, Vol. 54, Chicago: The University of Chicago Press. 22
BLANCHARD, Olivier and Michael KREMER (1997), "Disorganization", Quarterly Journal of Economics. CABALLERO, Ricardo and Mohamad HAMMOUR (1994), "The cleansing effects of recessions," American Economic Review, 84(5), pp. 1356-68. CABALLERO, Ricardo and Mohamad HAMMOUR (1996), "On the timing and efficiency of creative destruction," Quarterly Journal of Economics, 111(3), pp. 805-852. CABALLERO, Ricardo and Mohamad HAMMOUR (2000), "Institutions, restructuring, and macroeconomic performance," NBER Working Paper No. 7720 (May). CAMPBELL, Jeffrey R (1997), "Entry, exit, technology, and business cycles", NBER Working Paper No. 5955. CAVES, Richard E (1997), "Industrial organization and new findings on the turnover and mobility of firms", Working Paper, Harvard University. CHARI, V.V. and Hugo HOPENHAYN (1991), "Vintage human capital, growth, and the diffusion of new technology," Journal of Political Economy, 99 (6), pp. 1142-65. COOLEY, T., J. GREENWOOD and M. YORUKOGLU (1996), "The replacement problem", Working Paper, University of Rochester. COOPER, Russell, John HALTIWANGER and Laura POWER (1997), "Machine replacement and the business cycle: lumps and bumps", NBER Working Paper No. 5260 (revised). DAVIS, Steven J. and John HALTIWANGER (1991), "Wage dispersion between and within U.S. manufacturing establishments, 1963-86", Brookings Papers on Economic Activity, Microeconomics, pp. 115-20. DAVIS, Steven J. and John HALTIWANGER (1999), "Gross job flows", in ASHENFELTER and CARD (eds.), Handbook of Labor Economics, North-Holland, pp. 2711-2805. DAVIS, Steven J., John HALTIWANGER and Scott SCHUH (1996), Job Creation and Destruction, Cambridge: MIT Press. DEAN, Edwin R. and Kent KUNZE (1992), "Productivity measurement in Service industries", in Output Measurement in the Service Sector, Chicago: University of Chicago Press for NBER, pp. 73-101. DIAL, Jay and Kevin J. MURPHY (1995), "Incentives, downsizing, and value creation at general dynamics," Journal of Financial Economics, 37, pp. 261-314. DOMS, Mark and Timothy DUNNE (1994), "Capital adjustment patterns in manufacturing establishments", Center for Economic Studies Working Paper, CES 94-11. DOMS, Mark, Timothy DUNNE and Kenneth TROSKE (1996), "Workers, wages, and technology", Quarterly Journal of Economics, 112, pp. 253-90. DWYER, Douglas (1995), "Technology locks, creative destruction, and non-convergence in productivity levels", Center for Economic Studies Working Paper, CES 95-6. 23
ECO/WKP(2000)40 DWYER, Douglas (1997), "Productivity races I: are some productivity measures better than others?", Center for Economic Studies Working Paper, CES 97-2. DUNNE, Timothy, Mark ROBERTS and Larry SAMUELSON (1989), "Establishment turnover and gross employment flows in the U.S. manufacturing sector," Journal of Labor Economics, 7(1), pp. 48-71. ERICSON, Richard and Ariel PAKES (1989), "An alternative theory of firm and industry dynamics", Discussion Paper 445, Columbia University (September). ERICSON, Richard and Ariel PAKES (1995), "Markov-perfect industry dynamics: a framework for empirical work", Review of Economic Studies, 62(1) pp. 53-82. FOSTER, Lucia, John HALTIWANGER and C.J. KRIZAN (1998), "Aggregate productivity growth: lessons from microeconomic evidence", NBER Working Paper No. 6803 (mimeo., revised June 2000). GRILICHES, Zvi (1994), "Productivity, R&D, and the data constraint", American Economic Review, 84(1), pp. 1-23. GRILICHES, Zvi (ed.) (1992), "Output measurement in the service sectors", Studies in Income and Wealth, Vol. 56, Chicago: The University of Chicago Press. GRILICHES, Zvi and Haim REGEV (1995), "Productivity and firm turnover in Israeli industry: 1979-1988", Journal of Econometrics, 65, pp. 175-203. GROSSMAN, Gene M. and Elhanan HELPMAN (1991), Innovation and Growth in the global economy, Cambridge: MIT Press. HALTIWANGER, John (1997), "Measuring and analyzing aggregate fluctuations: the importance of building from microeconomic evidence", Federal Reserve Bank of St. Louis Economic Review, January/February. HALTIWANGER, John and Milan VODOPIVEC (2000), "Gross worker and job flows in a transition economy: an analysis of Estonia", mimeo. (June). HOPENHYN, Hugo (1992), "Entry, exit, and firm dynamics in long run equilibrium", Econometrica, 60(5), pp. 1127-50. HOPENHYN, Hugo and Richard ROGERSON (1993), "Job turnover and policy evaluation: a general equilibrium approach", Journal of Political Economy, 101(5), pp. 915-38. JOVANOVIC, Boyan (1982), "Selection and the evolution of industry," Econometrica, 50(3), pp. 649-70. JOVANOVIC, Boyan and Glenn M. MacDONALD (1994), "Competitive diffusion", Journal of Political Economy, 102 (1) pp. 24-52. JOVANOVIC, Boyan and Rafael ROB (1989), "The growth and diffusion of knowledge", Review of Economic Studies, 56, pp. 569-82. KENDRICK, John W. and Beatrice N. VACCARA (eds.) (1980), "New developments in productivity and measurement analysis, Studies in Income and Wealth, Vol. 44, Chicago: The University of Chicago Press. 24
LAMBSON, Val E. (1991), "Industry evolution with sunk costs and uncertain market conditions", International Journal of Industrial Organization, 9(2), pp. 171-96. LIU, Lili and James R. TYBOUT (1996), "Productivity growth in Chile and Columbia: the role of entry, exit and learning", in ROBERTS and TYBOUT (eds.), Industrial Evolution in Developing Countries: Micro Patterns of Turnover, Productivity and Market Structure, New York: Oxford University Press for the World Bank, pp. 73-103. LUCAS, Robert E. (1977), "On the size distribution of business firms", Bell Journal of Economics, 9(Autumn), pp. 508-23. MANSFIELD, Edwin, Mark SCHWARTZ and Samuel WAGNER (1981), "Imitation costs and patents", Economic Journal, 91, p. 907-918. McGUCKIN, Robert H (1995), "Establishment microdata for economic research and policy analysis: looking beyond the aggregates", Journal of Economics and Business Statistics, 13(1), pp. 121-6. MURPHY, Kevin J (1997), "Pay, performance, and executive compensation", in ASHENFELTER and CARD (eds.), Handbook of Labor Economics. NASBETH, Lars and George RAY (eds.) (1974), The Diffusion of New Industrial Processes: An International Study, Cambridge: Cambridge University Press. NELSON, Richard R. and Sidney G. Winter (1982), An Evolutionary Theory of Economic Change, Cambridge, MA: Harvard University Press. OLLEY, G. Steven and Ariel PAKES (1996), "The dynamics of productivity in the telecommunications equipment industry", Econometrica, 64(6), pp. 1263-1297. PAKES, Ariel and Mark SCHANKERMAN (1984), "The rate of obsolescence of patents, research gestation lags, and the private rate of return to research resources", in GRILICHES (ed.), R&D, Patents, and Productivity, Chicago: University of Chicago Press for NBER. ROBERTS, Mark J. and Dylan SUPINA (1997), "Output price and markup dispersion in micro data: the roles of producer and heterogeneity and noise," Center for Economic Studies Discussion Paper, CES 97-10. ROBERTS, Kevin and Martin L. WEITZMAN (1981), "Funding criteria for research, development, and exploration projects", Econometrica, 39, pp. 1261-88. ROGERS, Everett M (1983), Diffusion of Innovations (3rd ed.), New York: Free Press. SCHUMPETER, J.A. (1942), Capitalism, Socialism, and Democracy, New York: Harper and Brothers. SIEGEL, Donald and Zvi GRILICHES (1991), "Purchased services, outsourcing, computers and productivity in manufacturing", NBER Working Paper No. 3678. STEIN, Jeremy (1997), "Waves of creative destruction: firm-specific learning-by-doing and the dynamics of innovation", Review of Economics and Statistics, 4(2), pp. 265-288. 25
ECO/WKP(2000)40 TRAGER, Mitchell L. and Richard A. Moore (1995), "Development of a longitudinally-linked establishment based register, March 1993 through April 1995", U.S. Bureau of the Census Working Paper. TYBOUT, James R. (1996), "Heterogeneity and productivity growth: assessing the evidence" in ROBERTS and TYBOUT (eds.) (1996), Industrial Evolution in Developing Countries: Micro Patterns of Turnover, Productivity, and Market Structure, New York: Oxford University Press for the World Bank, pp. 43-72. 26
ECONOMICS DEPARTMENT WORKING PAPERS 266. Determinants of Long-term Growth: A Bayesian Averaging of Classical Estimates (BACE) Approach (October 2000) Gernot Doppelhofer, Ronald I. Miller and Xavier Sala-i-Martin 265. The Great Reversals: The Politics of Financial Development in the 20th Century (October 2000) Raghuram G. Rajan and Luigi Zingales 264. Trade and Growth: Still Disagreement about the Relationship (October 2000) Robert Baldwin 263. Growth Effects of Education and Social Capital in the OECD countries (October) Jonathan Temple 262. Human Capital in Growth Regressions: How Much Difference Does Data Quality Make? (October 2000) Angel de la Fuente and Rafael Domйnech 261. Raising the Speed Limit: US Economic Growth in the Information Age (October 2000) Dale W. Jorgenson and Kevin J. Stiroh 260. Summary of an Informal Workshop on the Causes of Economic Growth (October 2000) Jonathan Temple 259. Knowledge, Technology and Economic Growth: Recent Evidence from OECD Countries (October 2000) Andrea Bassanini, Stefano Scarpetta and Ignazio Visco 258. Average Effective Tax Rates on Capital, Labour and Consumption (October 2000) David Carey and Harry Tchilinguirian 257. The health care System in Poland (September 2000) Nathalie Girouard and Yutaka Imai 256. Public Expenditure Reform: The Health Care Sector in the United Kingdom (August 2000) Vincent Koen 255. Regulatory Reform in Road Freight and Retail Distribution (August 2000) Olivier Boylaud 254. Regulation, Market Structure and Performance in Air Passenger Transportation (August 2000) Rauf Gonenc and Giuseppe Nicoletti 253. Policy Interdependence during Economic Transition: the Case of Slovakia 1999-2000 (June 2000) Joaquim Oliveira Martins and Tristan Price 252. E-Commerce: Impacts and Policy Challenges (June 2000) Jonathan Coppel 251. The Implementation and the Effects of Regulatory Reform: Past Experience and Current Issues (June 2000) Rauf Gonenc, Maria Maher and Giuseppe Nicoletti 27
ECO/WKP(2000)40 250. The Concept, Policy Use and Measurement of Structural Unemployment: Estimating a Time Varying NAIRU across 21 OECD Countries (June 2000) Pete Richardson, Laurence Boone, Claude Giorno, Mara Meacci, David Rae and David Turner 249. Options for Reforming the Spanish Tax System (June 2000) Isabelle Joumard and Aristomиne Varoudakis 248. Economic Growth in the OECD Area: Recent Trends at the Aggregate and Sectoral Level (June 2000) Stefano Scarpetta, Andrea Bassanini, Dirk Pilat and Paul Schreyer 247. Economic Effects on the 1999 Turkish Earthquakes: an Interim Report (June 2000) Alexandra Bibbee, Rauf Gonenc, Scott Jacobs, Josef Konvitz and Robert Price 246. Policy Influences on Economic Growth in OECD Countries: an Evaluation of the Evidence (June 2000) Sanghoon Ahn and Philip Hemmings 245. The Tax System in the Czech Republic (May 2000) Chiara Bronchi and Andrew Burns 244. The Tax System in Norway: Past Reforms and Future Challenges (May 2000) Paul van den Noord 243. A Changing Financial Environment and the Implications for Monetary Policy (May 2000) Paul Mylonas, Sebastian Schich, Gert Wehinger 242. Carbon Emission Leakages: a General Equilibrium View (May 2000) Jean-Marc Burniaux and Joaquim Oliveira Martins 241. The Healthcare System in Hungary (April 2000) Eva Orosz and Andrew Burns 240. Comparing Semi-Structural Methods to Estimate Unobserved Variables: the HPMV and Kalman Filters Approaches (April 2000) Laurence Boone 239. New Issues in Public Debt Management: Government Surpluses in Several OECD Countries, the Common Currency in Europe and Rapidly Rising Debt in Japan (April 2000) Paul Mylonas, Sebastian Schich, Thorsteinn Thorgeirsson and Gert Wehinger 238. Regulation, Industry Structure and Performance in the Electricity Supply Industry (April 2000) Faye Steiner 237. Regulation, Market Structure and Performance in Telecommunications (April 2000) Olivier Boylaud and Giuseppe Nicoletti 236. Predicting the Evolution and Effects of the Asia Crisis from the OECD Perspective (April 2000) Pete Richardson, Ignazio Visco and Claude Giorno 235. Modelling Manufacturing Export Volumes Equations A System Estimation Approach (April 2000) Keiko Murata, David Turner, David Rae and Laurence Le Fouler 234. The Polish Tax Reform (March 2000) Patrick Lenain and Leszek Bartoszuk 28

J Haltiwanger

File: aggregate-growth-what-have-we-learned-from-microeconomic-evidence.pdf
Author: J Haltiwanger
Published: Thu Oct 12 15:21:27 2000
Pages: 28
File size: 0.14 Mb


Life-span development, 8 pages, 0.41 Mb

, pages, 0 Mb
Copyright © 2018 doc.uments.com