j.1467-9701.2007.01087.x (2)
TRANSCRIPT
-
7/29/2019 j.1467-9701.2007.01087.x (2)
1/24
The World Economy (2008)doi: 10.1111/j.1467-9701.2007.01087.x
2008 The Authors
Journal compilation 2008 Blackwell Publishing Ltd, 9600 Garsington Road,
212 Oxford, OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA
Blackwell Publishing LtdOxford, UKTWECWorld Economy0378-5920 2008 The Author Journal compilation Blackwell Publishers Ltd.2008XXXOriginal ArticlesCONTRIBUTION OF EXPORTINGRICHARD HARRIS AND QIAN CHER LI
Evaluating the Contribution
of Exporting to UK
Productivity Growth: Some
Microeconomic Evidence
Richard Harris and Qian Cher Li
University of Glasgow
1. INTRODUCTION
E
MERGING evidence on industrial restructuring using micro-based data from
the Annual Respondents Database (ARD) has shown that UK productivity
growth is increasingly the result of a market selection process, in which moreproductive entrants replace less productive establishments whilst high pro-
ductivity incumbents gain market share (Oulton, 2000; Disney et al., 2003a).
In particular, the study by Disney et al. suggests that between 1980 and 1992,
50 per cent of labour productivity growth and 8090 per cent total factor produc-
tivity (TFP) growth could be explained by what they term external restructuring
effects (i.e. the impact of market entry and exit as well as inter-firm reallocations
in market share).
1
Using comparable data and a similar approach, Harris (2004)
reports that over the 199098 period, the growth in manufacturing TFP did not
substantially benefit from incumbents improving or a reallocation of market
share from worse to better plants; rather TFP benefited mostly from the
The authors wish to thank UKTI for sponsoring this project and two anonymous referees for theirhelpful comments. However, all views expressed are solely the responsibility of the authors. Thiswork contains statistical data from ONS which is Crown copyright and reproduced with the per-mission of the controller of HMSO and Queens Printer for Scotland. The use of the ONS statisticaldata in this work does not imply the endorsement of the ONS in relation to the interpretation oranalysis of the statistical data. This work uses research datasets which may not exactly reproduceNational Statistics aggregates.
1
Note, and anticipating the results we present below, Disney et al. (2003a) use a different decom-
position approach to the one used here. They also cover a different (non-overlapping) time period,and their data source is the ARD (rather than FAME).
-
7/29/2019 j.1467-9701.2007.01087.x (2)
2/24
CONTRIBUTION OF EXPORTING 213
2008 The Authors
Journal compilation Blackwell Publishing Ltd. 2008
churning of plants whereby plants with higher TFP entered and those with
below-average TFP were more likely to exit.
None of these previous studies have considered the role of exporting firms in
contributing to productivity growth. Exporters are usually larger firms, withhigher productivity, higher capital intensity, faster growth and a more skilled
workforce (see Baldwin and Gu, 2004; Girma et al., 2004; and Greenaway and
Kneller, 2004, for recent evidence). Nevertheless, microeconomic evidence on
the exportingproductivity relationship still remains inconclusive, particularly on
the existence of productivity gains post-export market entry. As noted in a recent
survey:
the micro and macro evidence on exporting and growth would appear to be inconsistent: thereis weak evidence of a causal relationship between exports and productivity growth at the firm
level but a strong correlation at the aggregate level (Greenaway and Kneller, 2005).
However, this evidence is very likely to be reconciled when considering inter-
firm reallocations of resources towards more productive firms, which inter alia
we consider here.
Until now it has not been possible to assess the contribution of exporters to
UK aggregate productivity levels and growth, particularly in terms of restructur-
ing effects, due to data limitations.
2
Consequently, and as noted in DTI (2006),
there has to date been little evidence, particularly for the UK, that substantiates
the overall benefits from international trade, and therefore providing a rationale
for government intervention to help firms develop their exporting activities.However, in this study we are able to use a (weighted) FAME database to
consider, for the first time, the contribution of exporters to aggregate productivity
growth. We have constructed a dataset that distinguishes firms that enter and
exit the database (separately from those that are taken over by others as part of
mergers and acquisitions), and employed the approach taken by Haltiwanger
(1997) to decompose measures of productivity into various components that
represent the impact of resource allocations across surviving exporters and
non-exporters as well as the impact on productivity of the entry and exit of firms.
We consider both labour and total factor productivity, with the TFP measure
being preferred since it takes into account all factor inputs other than labour in
the production process. Labour productivity may increase because of greater
outsourcing or capital deepening, as firms substitute intermediate inputs and cap-
ital for labour, while the efficiency with which all inputs are used may not
necessarily increase (and possibly may even decline).
In summary, we show that in aggregate (and covering all sectors) exporting
firms contributed more to overall UK productivity growth than non-exporting
2
Note the Annual Respondents Database (ARD), the primary source of UK micro-data, does notcontain information on exporting activities.
-
7/29/2019 j.1467-9701.2007.01087.x (2)
3/24
214 RICHARD HARRIS AND QIAN CHER LI
2008 The Authors
Journal compilation Blackwell Publishing Ltd. 2008
firms (i.e. 2.32 per cent compared with 1.55 per cent per annum in terms of
labour productivity, and 1.27 per cent relative to 0.81 per cent per annum in
terms of TFP). This is made up of both intra-firm productivity improvements and
inter-firm reallocations of resources towards more productive exporting firms(which includes the impact of firm entry and exit). Such external restructuring is
especially important in the non-manufacturing sector, suggesting overall that
government policy aimed at boosting exporting needs to take account of the
different channels through which productivity growth occurs across different
sectors.
A major contribution of this paper lies in the dataset used: to obtain data that
are representative of the population of firms operating in the UK (for all market-
based sectors), our analysis is based on a weighted FAME dataset using weights
compiled from the ARD (at the three-digit industry SIC by five size-band levels),
as the unweighted FAME database is unrepresentative of small- to medium-sized
enterprises and therefore cannot produce results that can be generalised to the
UK level.
The rest of the paper is structured as follows. In Section 2, we briefly review
the literature related to export-market dynamics and the linkage between
exporting and aggregate productivity growth. In Section 3 we describe the
weighted FAME dataset used for this study and highlight some data issues. This
is followed by a discussion of methodological issues associated with productivity
decomposition, together with our results (Section 4). The last section concludes.
2. EXPORTING, MARKET DYNAMICS AND AGGREGATE PRODUCTIVITY:
EVIDENCE FROM THE LITERATURE
From a policy perspective, it would seem that there is substantial international
evidence of the benefits from international trade which, at least in part, should
provide a rationale for government intervention to help firms develop their
exporting activities (for the UK see especially Chapter 3 in DTI, 2006).
3
These
benefits are largely linked to the higher productivity of exporters, which leads tohigher overall UK productivity growth through some or all of the following:
(i) exporters achieving relatively higher productivity before they begin selling
overseas which results in them having higher productivity growth post-entry (the
so-called self-selection effect see below); (ii) exporting firms becoming more
productive over time through a learning-by-exporting effect; (iii) intra-industry
(i.e. inter-firm) resource reallocations for firms that remain in operation in a
3
The further issue of whether there are grounds for policy intervention only in the case of market
failures or whether intervention is justified in more general terms is also relevant but discussedelsewhere (see Harris and Li, 2005).
-
7/29/2019 j.1467-9701.2007.01087.x (2)
4/24
CONTRIBUTION OF EXPORTING 215
2008 The Authors
Journal compilation Blackwell Publishing Ltd. 2008
certain period away from non-exporters to higher productivity exporters; (iv) the
entry of higher productivity exporters, i.e. new firm start-ups that immediately
or very soon sell to international markets (what have been called born global
companies in the literature see Oviatt and McDougall, 1995, and the review ofthis literature in Harris and Li, 2005); and (v) the shutdown of lower productivity
firms (which may be exporters but are more likely to be non-exporters who
have the lowest levels of productivity of all, causing them to exit first). The
contribution of (i) + (ii) is the within (or intra-) firm effect of higher produc-
tivity exporters contributing to higher overall UK productivity growth because
they internally improve their levels of efficiency and/or innovativeness; whereas
(iii)(v) are concerned with external restructuring due to the reallocation of
resources between firms that remain in operation throughout and/or through
creative destruction (cf. Schumpeter, 1950).
The earliest studies concerned with the relationship between exports and
productivity growth (and thus the debate on export promotion policies) used a
macroeconomics approach, from the conventional HeckscherOhlin model to
new trade models.
4
However, these macroeconomics-oriented models, arguably,
only provide a limited understanding of firms exporting behaviour and thus have
a limited role in informing policy, which is to a considerable extent targeted at
individual firms. In contrast, the last decade has seen a surge of interest in
studying the microeconomic evidence, taking into account the importance of
heterogeneity amongst firms. This emphasis on the firm-level evidence has
been partly triggered by the availability of data at firm and/or plant level, as wellas recent developments in the use of theoretical modelling and econometric
techniques to exploit these usually more intricate micro-data.
a. An Intra-Firm Perspective: Exporting and Firm Performance
Research on the exportingproductivity relationship was initially empirically
driven, and it is universally found that exporting is positively associated with
firm performance (for recent surveys of the literature, see Greenaway and Kneller,
2005, 2007; Lpez, 2005; Wagner, 2007). Nevertheless, despite this positivelinkage, there is still much controversy about the causal direction of this link,
which is often examined empirically by testing two competing but not mutually
exclusive hypotheses, viz. self-selection and learning-by-exporting.
5
Underlying the self-selection hypothesis are the notions of sunk cost and
firm-level heterogeneity. Since international exposure is associated with more
4
For a review of macroeconomic evidence on trade and growth, see Lpez (2005).
5
In particular, Greenaway and Kneller (2005, Table 1) and Wagner (2007, Table A1) provide
summaries of empirical findings on the exportproductivity nexus and the causality issue usingfirm-level data.
-
7/29/2019 j.1467-9701.2007.01087.x (2)
5/24
216 RICHARD HARRIS AND QIAN CHER LI
2008 The Authors
Journal compilation Blackwell Publishing Ltd. 2008
intensive competition, firms that internationalise are forced to become more
efficient so as to enhance their survival characteristics. Additionally, the existence
of sunk entry costs means exporters have to be more productive to overcome
such barriers to entry before they can realise expected profits, and these fixedcosts have been incorporated into recent theoretical modelling, such as those
undertaken by Clerides et al. (1998), Bernard et al. (2003) and Melitz (2003),
who maintain that exporting firms need to be more productive before entering
the export market in order to overcome the sunk costs of entering overseas
markets. Notably, and perhaps more importantly from a policy perspective, it has
been argued that firms do not naturally consider exporting just because they
have reached a certain necessary productivity threshold; in fact, such self-
selection could be a conscious decision, involving a purposeful improvement of
productivity as a planned strategy to enter export markets (Lpez, 2004, 2005).
For instance, Lpez (2004) develops a simple model in which forward-looking
firms need to invest in new technology in order to become exporters, and the
adoption and assimilation of this technology require them to be more productive
to begin with. Ample empirical evidence has confirmed that this is indeed the
case: in more than 30 studies reviewed in Greenaway and Kneller (2005),
covering a wide range of countries, self-selection is universally found to be
important.
The other direction of this causality concerns the learning effect associated
with exporting: the exposure to more competitive international markets
intensifies firm performance, and therefore gives productivity a further boost, dueto knowledge accumulation, technology transfer, competition effects and the like.
If this fails to hold, there are important policy implications if better firms do
self-select into export markets, and exporting does not further boost productivity,
then export promotion policies could simply be a waste of resources (involving
large-scale dead-weight and possibly even displacement effects given that firms
that export usually also sell to domestic markets).
6
The learning-by-exporting
proposition has, in general, received less support in the literature, with only
significant learning effects found in particular groups, such as younger exporters
(Baldwin and Gu, 2003, 2004, for Canadian manufacturing; Greenaway and Yu,2004, for the UK chemical industry), exporters in small, open and highly export-
oriented economies (Hansson and Lundin, 2004, for the Swedish manufacturing
sector), or firms serving advanced, high-wage export markets (Damijan et al.,
2005).
It is worth noting that the paucity of evidence on this learning effect could
in large part be due to methodological issues involved when testing for the
6
Robust empirical evidence shows that exporters tend to sell very small fractions of their outputabroad (Roberts et al., 1995; Aw et al., 1997).
-
7/29/2019 j.1467-9701.2007.01087.x (2)
6/24
CONTRIBUTION OF EXPORTING 217
2008 The Authors
Journal compilation Blackwell Publishing Ltd. 2008
productivity effect of exporting, such as sample selectivity.
7
The latter suggests
that participants in export markets may possess certain characteristics (e.g.
superior human capital, greater absorptive capacity) such that they would achieve
higher productivity compared with non-participants even if they did not enterexport markets, and thus productivity growth is correlated with the decision to
participate in global markets (but is not caused
by selling in international
markets). Nevertheless, significant post-entry productivity gains in exporting
firms have been reported in recent studies that overcome problems of selectivity
bias, such as Girma et al. (2004) who employ matching techniques and UK
manufacturing data, and Harris and Li (2007a) who use three distinct approaches
(instrumental variable, control function and data-matching techniques) applied to all
market-based sectors in the UK.
b. An Inter-Firm Perspective: Export Market Dynamics and
Aggregate Productivity
Self-selection and the learning-by-exporting hypotheses reviewed above
have attempted to assess only the linkages between exporting and within-firm
productivity growth. Since entry (and exit) in export markets inherently involves
changes in market shares, and thus industrial restructuring, in order to evaluate
or justify trade-promotion policies it is equally important to look at the impact
of firm-level exporting on intra- or inter-industry reallocations of resources and
therefore aggregate productivity growth. This approach provides a holistic viewof the interaction of firms, industries and the aggregate economy.
In the context of exports, the process of entry and exit in export markets
differs from market entry and exit in the conventional sense, since firms can
continue to produce for the domestic market.
8
Moreover, as discussed above,
entrants to export markets invariably have to achieve superior performance
before they enter and could be rewarded with even better performance after they
penetrate foreign markets.
Export market dynamics have been modelled in recent studies by incorporat-
ing intra-industry heterogeneity. General empirical findings in the trade literature
7
See Harris and Li (2007a) for a discussion of the sample selectivity issue in the context ofevaluating the exportproductivity relationship; and more generally, Blundell and Costa Dias(2000) and Heckman and Navarro-Lozano (2004) for reviews of this issue and various solutions.
8
Meanwhile, a persistent transition in and out of exporting has been observed by Bernard andJensen (2004a) a high degree of re-entry by former exporters and a high propensity to stopexporting in former non-exporters. There are at least two competing views to explain this persist-ence the sunk costs argument (i.e. exporting begets exporting) and heterogeneous attributes(certain firms are more export-oriented). Bernard and Jensen attempt to identify the roles of bothsunk costs and plant heterogeneity and confirm the significant presence of both. Nevertheless,
it remains unanswered as to how firms acquire these characteristics that facilitate their entry intoforeign markets.
-
7/29/2019 j.1467-9701.2007.01087.x (2)
7/24
218 RICHARD HARRIS AND QIAN CHER LI
2008 The Authors
Journal compilation Blackwell Publishing Ltd. 2008
tend to suggest that the determinants of a firms entry decision include trade
liberalisation (Baldwin and Gu, 2004), sunk entry costs (Das et al., 2001; Bernard
and Jensen, 2004a; Girma et al., 2004); some firm-level characteristics such as
size (Aw and Hwang, 1995; Roberts and Tybout, 1997; Bleaney and Wakelin,2002; Gourlay and Seaton, 2004); the firms experience, including ex-ante
success (Bernard and Jensen, 1999; Greenaway and Kneller, 2004; Kneller and
Pisu, 2007); export spillovers (Aitken et al., 1997; Greenaway et al., 2004); and
foreign networks (Sjoholm, 2003). In addition, the exit decision depends mainly
on industrial characteristics such as the level of sunk costs, and the minimum
productivity or efficiency needed to secure non-negative profits (Das et al., 2001;
Bernard and Jensen, 2004a).
The impact of export market dynamics on productivity has been studied in an
emerging strand of recent theoretical literature, seeking to explain empirical
findings and then generating testable predictions. Bernard et al. (2003) show that
trade liberalisation expands the market shares of the most productive firms by
providing them with large export markets, while at the same time such liberal-
isation forces firms at the lower end of the productive efficiency distribution to
quit as international competition intensifies. Similarly, in a seminal paper, Melitz
(2003) develops a forward-looking model of steady-state trade with heterogen-
eous firms and imperfect competition to show that trade liberalisation increases
a countrys imports and erodes domestic sales and profits. Firms at the higher
end of the productivity distribution expand their export sales more than they
contract their domestic sales; whereas those non-exporters at the lowest end ofthe productivity distribution have to contract or quit. Consequently, this model
generates a prediction that exporting provides an additional channel for produc-
tivity gains whereby export market dynamics act to increase aggregate produc-
tivity, as better firms expand their market share and the worst firms contract
or exit.
9
A more recent development in the theoretical modelling in this regard is
provided in Bernard et al. (2007). Their approach differs from Melitzs model
in that they focus on comparative advantage. It is found that intra- and
inter-industry reallocations of resources brought about by trade liberalisationimprove average industry productivity and sectoral firm output, but relatively
more so in industries with a comparative advantage than those with comparative
disadvantages.
10
9
In addition, Helpman et al. (2004) provide an extension of Melitzs model to include foreigndirect investment (FDI) as another strategy of firms international expansion. Their model assumesthe (fixed) costs differ amongst domestic firms, exporters and multinationals, in conjunction withfirm-level heterogeneity in productivity.
10
Additionally, their model also has implications for job creation and welfare gains associated withthe creative destruction of heterogeneous firms leading to reallocations of resources.
-
7/29/2019 j.1467-9701.2007.01087.x (2)
8/24
CONTRIBUTION OF EXPORTING 219
2008 The Authors
Journal compilation Blackwell Publishing Ltd. 2008
Empirically, the effect of transitions into and out of export markets on
business performance is often captured by the notion of an export premium,
which measures how much a firms performance changes when it exports (e.g.
Bernard and Jensen, 1999, for the US; Aw et al., 2000, for Korea and Taiwan;Silvente, 2005, for the UK). More specifically, the studies for the US, Korea and
Taiwan conclude that when firms switch from being non-exporters to becoming
exporters, their performance improves, while switching from being exporters to
being domestically-oriented firms retards their performance. Likewise, Silvente
looks at a sample of UK small firms over a seven-year period, showing that there
are symmetric effects on the export premium between entrants and those exiting
new exporters enjoy considerable gains while those exiting from overseas
markets suffer significant losses in terms of employment, wages, sales and pro-
ductivity growth rates.
11
Another strand of the empirical literature compares productivity amongst
distinct sub-groups of exporting firms as well as domestic non-exporters. As
Bernard and Jensen (2004b) report, in US manufacturing new entrants into export
markets are rewarded with a surge in TFP, especially during the first year post-
entry, and thereafter their productivity path becomes flatter, following that of
continuous exporters (although with significantly lower productivity levels).
In contrast, those that exit from exporting are characterised by a substantial
deterioration in productivity to eventually resemble the flat growth trajectory of
continuous non-exporters. In particular, they find increased export opportunities
are associated with both intra- and inter-industry reallocations in the US,accounting for 40 per cent of TFP growth in the manufacturing sector, half of
which can be explained by an intra-industry reallocation of economic activities.
Thus, the higher productivity levels as well as the faster growth rates found in
exporters (in terms of employment and output) offer an additional reallocative
channel for explaining aggregate productivity growth.
12
For Canada, Baldwin and Gu (2003) point to a negative impact for those that
exit, arguing that the ebb and flow induced by international competition culls
some participants from export markets. The least successful entrants have to
withdraw to domestic markets and then lag further behind those that continueserving foreign markets. That is, productivity growth is lower for quitters than
continuers, and substantially lower when compared to new entrants to export
markets.
13
Thus, changing export status is indeed associated with considerable
11
The results from these studies control for the impact of covariates, such as size and industryeffects.
12
Nevertheless, this study does not consider market entry and exit all plants in the dataset existthroughout the period of study. Thus, there is no comparison of the relative importance of creativedestruction, and most importantly how internationalisation interacts with market entry and exit.
13
In addition, the negative impact of exit on the firms efficiency is also captured in Bernard andWagner (1997) and Clerides et al. (1998).
-
7/29/2019 j.1467-9701.2007.01087.x (2)
9/24
220 RICHARD HARRIS AND QIAN CHER LI
2008 The Authors
Journal compilation Blackwell Publishing Ltd. 2008
fluctuations in productivity. Overall, they find exporters accounted for almost
75 per cent of productivity growth in Canadian manufacturing during the 1990s,
28 per cent of which was accounted for by export market entry (both existing
and new entrants).In contrast with the well-documented evidence available for the US and
Canada, there seems to be a paucity of evidence worldwide on aggregate produc-
tivity growth in the context of international trade. Only a limited amount of
evidence of trade-induced productivity growth is available for other countries.
For example, Hansson and Lundin (2004) decompose overall manufacturing
productivity growth in Sweden into within-firm and reallocation effects. They
discover that reallocations due to increased exporting positively impact on aggre-
gate productivity growth, which is then counteracted by reallocations attributed
to changes in domestic shipments. Falvey et al. (2004), again using Swedish
manufacturing data, find that exporting has a sizeable effect on industry pro-
ductivity growth, in terms of increasing market share for higher productivity
exporters.
None of the above studies, however, use nationally representative data that
covers all market-based sectors, to look at the contribution of exporting via
intra-firm, inter-firm and market churning effects. We do this using UK data for
19962004, as set out in the following sections.
3. DATA ISSUES
The FAME dataset used for this study includes all firms operating in the UK
that are required to make a return to Companies House, covering every sector
of the market-based economy. It contains basic information on firm-specific
characteristics, such as turnover, intermediate expenditure on bought-in goods,
services, materials, employment, assets and, most importantly, overseas sales.
We started with over 7,000,000 records covering the period 19962004,
14
and
then omitted all those that lacked the required data (especially on exporting
activities). The result was an unbalanced panel covering 81,819 firms with326,906 observations for 19962004. In this panel, nearly 23 per cent of the
firms are observed throughout the nine-year period.
The resulting FAME dataset (containing the financial information required)
is severely biased towards large enterprises, and is thus unrepresentative of UK
firms. To obtain a representative distribution, we treat the firms in the FAME
14
We therefore have reason to believe that we observe almost all relevant firms who have to makea Companies House return during this period.
-
7/29/2019 j.1467-9701.2007.01087.x (2)
10/24
CONTRIBUTION OF EXPORTING 221
2008 The Authors
Journal compilation Blackwell Publishing Ltd. 2008
dataset as a sample of the ARD population,
15
and consequently weight the FAME
data to produce a database that is representative (by industry and firm size) of the
population of firms operating in Great Britain.
16
Specifically, we use aggregated
turnover data from the ARD subdivided into five size-bands (based on turnoverquintiles
17
) and three-digit industry SICs (resulting in 757 sub-groups).
18
We then
aggregate the FAME data into the same sub-groups, so that we can calculate weights
using ARD total turnover data divided by the comparable data from FAME.
19
Table 1 presents the results from weighting the FAME data, using the weights
derived from the (three-digit by five size-bands) turnover data from the ARD. The
unweighted data from FAME are dominated by the largest firms (defined as firms
with turnover of 7.2 million or above); this sub-group accounts for over 97 per cent
of total turnover. Weighting the FAME data produces a distribution across size-
bands that is comparable to that obtained when using the ARD. This is confirmed
in the final column in Table 1 the ratio of FAME to ARD turnover by size-band is
within a margin of 10 per cent. There is a suggestion that even weighted, the FAME
data slightly underestimate the contribution of the smallest firms (and corre-
spondingly overestimate the importance of the largest firms), but these differences
are not likely to unduly impact on our decomposition of productivity growth using
these weighted data. Further information on the construction of our weighted
FAME dataset and the export market dynamics is provided in Harris and Li (2007b).
Next, the FAME data for 1996 and 2004
20
are subdivided into (i) those firms
that existed throughout 19962004; and (ii) those that first appeared after 1996
15
For a detailed description of the ARD (available at the ONS), see Oulton (1997), Griffith (1999)and Harris (2002, 2005). Analysis using this database for the UK covers a range of areas but notexporting, due to its lack of information on exports. The counterpart to the ARD in the US is theLongitudinal Research Database (LRD) for US manufacturing provided through the US Bureau ofCensus. This US database has been merged with information on exports at the enterprise level toallow studies to be undertaken on the impact of exporting on aggregate productivity growth(see, for example, Bernard and Jensen, 2004a). Other countries have also compiled similar mergeddatabases (e.g. Baldwin and Gu, 2003, for Canada).
16
Efforts have also been made to actually merge FAME into the ARD, but this has been largelyunsuccessful. (See Harris and Li, 2007b, Ch. 2, for details.)
17
Note these are based on dividing the FAME data into five equally sized groups based onconstant-price turnover data, in order to obtain the cut-off points for each size-band.
18
Where there are fewer than 10 enterprises in any sub-group, these data are not used, in order toensure that the ONS data do not potentially disclose confidential information. This results in theloss of some 4 per cent of the total turnover available in the ARD. Note the ARD data have beenweighted to produce population totals for turnover for each size-band by industry (see Harris, 2005,for a discussion of weighting the ARD data).
19
Note we do not weight the FAME data for 34 industries (by three-digit SIC2003), thus allowingeach observation to represent itself; because the FAME data have better coverage in terms of totalturnover than the ARD (because certain industries such as some of financial services are notincluded in the ARD), or (very occasionally) the data do not match the ARD data by size-bandwithin the industry. These 34 industries (out of a total of 215 industries covered) account for just
2.9 per cent of total FAME turnover.
20
Only these two years are needed to decompose productivity growth for 19962004.
-
7/29/2019 j.1467-9701.2007.01087.x (2)
11/24
222 RICHARD HARRIS AND QIAN CHER LI
2008 The Authors
Journal compilation Blackwell Publishing Ltd. 2008
or could not be located in 2004.
21
The second group comprises all the firms that
had experienced some form of changes during the period considered. We then
disaggregate this change group into various sub-groups using information on
when the firm was first and last observed in the database during 19962004, andthe year the firm was incorporated. Details of constructing each sub-group are
provided in Table 2.
21
As previously stated, we started with more than 7,000,000 records from FAME; and we calculatewhich firms belong to the various sub-groups in Table 2 using this full dataset (i.e. we attempt to
avoid interpreting births and deaths of firms based on entries and exits to and from a dataset thatdoes not cover the population of relevant firms).
TABLE 1GBa Turnover (m) in 2003 Based on FAME and ARD by Size-Bands
Size-Bandb FAME ARD % FAME
Unweighted % Weighted % ARD
(1) (2) (3) (2) (3)227 to 1,184 3,578.5 0.2 60,892.8 5.3 63,751.8 5.9 95.5>1,184 to 7,244 49,683.9 2.4 198,417.1 17.4 200,988.3 18.5 98.7>7,244 1,988,609.3 97.4 874,543.9 76.5 812,157.0 74.8 107.5
All 2,042,279.4 100.0 1,142,283.4 100.0 1,085,916.2 100.0 105.0
Notes:a Unweighted FAME data cover the UK.b Size-bands are in 000.
Source: ARD and FAME databases.
TABLE 2Construction of Sub-Groups in FAME
Criteria Sub-Group N a
Observed in both 1996 and 2004 Continuing firm 12,897 (18.7)
Observed in 2004 but not 1996+ Year of incorporation > 1996
Opened post-1996 17,038 (24.7)
Observed in 2004 but not 1996
+ Year of incorporation < 1997
Merger/acquisition 11,322 (16.4)
Observed in 1996 but not in 2004+ Year of incorporation < 1997
Closed before 2004 27,638 (40.1)
Note:a Number of firms observed in FAME database (weighted using ARD weights); figures in parentheses arecolumn percentages.
-
7/29/2019 j.1467-9701.2007.01087.x (2)
12/24
CONTRIBUTION OF EXPORTING 223
2008 The Authors
Journal compilation Blackwell Publishing Ltd. 2008
The main issue in dividing the change group into the various sub-groups is
that firms observed in 2004 but not in 1996 could be either start-ups or subsumed
into other firms through acquisition or merger. We differentiate between these
two sub-groups based on the year of incorporation reported for the firm: thosethat were started in 1996 or earlier are not taken to be new start-ups but rather
first observed post-1996 as part of a newly merged venture.
4. METHODOLOGY AND RESULTS
In order to calculate the total factor productivity, we firstly estimate an aug-
mented production function as follows:
y
it
=
0
+
E
e
it
+
M
m
it
+
K
k
it
+
T
t
+
X
it
+
t
, (1)
where y
, e
, m
and k
refer to the logarithms of real gross output, employment,
intermediate inputs and capital stock in firm i in time t.22 Notably, we have also
included a vector of variables, X, that determine TFP.23 Hence, TFP in growth
terms is defined as (dropping subscripts):
ln TFP=%T+6 %E%M%K . (2)
As shown in equation (2), our preferred approach to estimating TFP is todirectly include the factors determining output (and thus TFP) into the production
function, since this avoids any problems of inefficiency and omitted variable
bias associated with estimating a two-stage model based on typically a growth
accounting approach.24 In short, since TFP is defined as any change in output not
due to changes in factor inputs, these determinants should be included directly
into the model (equation (1)) that determines TFP.
Equation (1) has been estimated for 16 separate industry groups using panel
data for 19962004 and the generalised methods of moments (GMM) systems
approach available in STATA 9.2 (Arellano and Bond, 1998), taking account of
22 With reference to variable definitions in the FAME dataset, output is defined as turnover, inter-mediate inputs as cost of sales less remuneration, and capital stock as tangible assets, all in 000(2000 prices).23 Full details and results are provided in Harris and Li (2007b, Ch. 3). Given data availability, Xcomprised the exporting status of the firm, and which industry and region it belonged to.24 See Harris (2005) for a detailed discussion of the measurement issues surrounding TFP, such asinefficient estimates (Newey and McFadden, 1999, Sec. 6), the omitted variable problem (Wang
and Schmidt, 2002) in a two-stage model, and the endogeneity of inputs and outputs in a productionmodel.
-
7/29/2019 j.1467-9701.2007.01087.x (2)
13/24
224 RICHARD HARRIS AND QIAN CHER LI
2008 The Authors
Journal compilation Blackwell Publishing Ltd. 2008
both self-selection and endogeneity issues.25 Having obtained estimates of TFP
at firm level, the index of aggregate productivity in year tas well as its growth
between tand t kis defined as a geometrically weighted average of individual
firm-level productivity in the economy:
(3)
where it is the share of gross output for firm i in period tfor the economy (in 2000
prices), and P measures productivity (labour productivity26 or TFP). Following
Haltiwanger (1997) and Foster et al. (2001), the decomposition of productivity
growth recognises that there are firms continuing in operation between t and
t k, new firms entering for period t and firms exiting that contributed to
productivity in t k. Thus productivity growth between t and t k (lnPt
) is
defined as:
(4)
Here the first term shows the productivity impact of resource shifts within
firms that are open in both tand t k, depending on how important such firms
are in the base year (in terms of their shares of output in the economy). Term
(ii) also concerns continuing firms, but measures the impact of changing output
shares across firms weighted by the firms ranking in the economy in the base
year. There might be a presumption that the highest ranked firms in the base
year would increase their market shares, denoting a positive redistribution ofresources. However, term (ii) is complemented with term (iii) the covariance
effect that considers whether increases in productivity correspond with increasing
25 Full details and results are provided in Harris and Li (2007b). Note, to account for endogeneitywe instrument the model using lagged values of all the endogenous variables plus two additionalinstruments: the age of the firm and whether it possessed any non-zero intangible assets (accumu-lated through R&D, advertising, etc.). These additional instruments are also included to identifythe model and thus overcome the self-selection problem surrounding which factors determine whoexports (where age and non-tangible assets are highly significant) separately from what factors
determine TFP (where age and non-tangible assets are not significant).26 Here labour productivity is defined as a firms gross output per employee in 2000 prices.
ln ln ln ln ln ,P P P P Pt it it i
t t t k = =
( )
ln (ln ln ) ln
(ln ln ) (ln ln )
i (ii) (iii)
Continuers: Continuers: Continuers:
within-firm between-firm cross-firm
(iv) (v)
Entering firms Exiting firms
it k it it k t k it it it
it it t k it k it k t k
P P P P
P P P P
+ + +
..
-
7/29/2019 j.1467-9701.2007.01087.x (2)
14/24
CONTRIBUTION OF EXPORTING 225
2008 The Authors
Journal compilation Blackwell Publishing Ltd. 2008
market shares. Lastly, terms (iv) and (v) denote the contributions of entering27
and exiting firms, both measured with respect to the economy average in the base
year. Note, term (v) is expected to be negative (if exiting firms have lower
productivity), and thus this term is preceded by a negative sign to allow for apositive impact on aggregate productivity.
Equation (4) has been applied at the total economy level, allowing resources
to be reallocated across industries as well as firms, but separately for exporters
and non-exporters.28 We have also undertaken analyses separately for the manu-
facturing and non-manufacturing sectors, given that exporting is usually more
prevalent in manufacturing (and thus it is of interest to see how the two sectors
differ in terms of their contributions, via exporting, to productivity growth).
Table 3 reports the output shares (the i) for the various sub-groups in 1996 (t k)
27 As explained earlier (Table 2), entrants in the FAME database are separated into firms thatopened post-1996 and firms that were merged/acquired.28 Given that equation (4) is used separately for each sub-group, with respect to within-firm effectsand entering/exiting firms, this is the correct approach since it involves internally consistentdisaggregations of the data. However, our approach implies that the between-firm effects areseparate for each sub-group, and there is no allocation of resources between them. This is unlikelyto be strictly true (although barriers to exporting do exist and seem to be substantial). Allowing for
interactions will involve more covariance terms and the decompositions become significantly morecomplicated. This is something for future consideration.
TABLE 3Output Shares, All Market-Based Sectors, 19962004
ContinuingFirms (t k)
ContinuingFirms (t)
Takeovers/Mergers (t)
EnteringFirms (t)
ExitingFirms (t k)
(a) ManufacturingNon-exporters 8.3 11.3 5.8 5.5 6.7Exporters 47.5 52.6 14.8 10.0 37.5All 55.8 63.9 20.6 15.5 44.2
(b) Non-manufacturingNon-exporters 23.6 19.9 21.0 15.5 33.2Exporters 20.3 21.1 10.3 12.2 22.9All 43.9 41.0 31.3 27.7 56.1
(c) All sectors
Non-exporters 19.6 17.9 17.4 13.1 26.3Exporters 27.4 28.6 11.4 11.7 26.7All 47.0 46.5 28.8 24.8 53.0
Note:Manufacturings share of real gross output in 1996 and 2004 was 26.0 per cent and 23.8 per cent, respectively.
Source: Tabulations from weighted FAME.
-
7/29/2019 j.1467-9701.2007.01087.x (2)
15/24
226 RICHARD HARRIS AND QIAN CHER LI
2008 The Authors
Journal compilation Blackwell Publishing Ltd. 2008
and 2004 (t).29 Notably, the shares of entering (including mergers or acquisitions)
and exiting firms over the period are large (whichever sub-sector of the economy
is considered), which in part is due to opening and closures being cumulative and
the longer the span between t kand t, the greater the effect of churning.30
It is also shown in Table 3(c), covering all sectors, that exporting firms
contributed a significant amount of total output in terms of continuing firms, but
much smaller amounts in terms of mergers or takeovers. However, these figures
for all market-based industries hide differences that exist between the manu-
facturing and non-manufacturing sectors; for continuing firms, output shares are
much higher for exporters (denoting there are both relatively more exporters, and
they tend to be larger firms). In addition, the smaller shares of output for mergers
and takeovers was particularly prevalent for manufacturing non-exporters, and
this sub-group also accounted for relatively much smaller output shares of new
entrants and firms that ceased production.
Table 4 produces initial relative productivity indices for the sub-groups,
confirming that exiting firms tended to have the lowest productivity levels for both
exporters and non-exporters (although this is less pronounced for exporters in
manufacturing), and new firms exhibited relatively high productivity, particularly
for exporters (although again manufacturing exporters are an exception).31
Firms taken-over/merged in both manufacturing and non-manufacturing also had
relatively high productivity (especially if they were also exporting), while non-
manufacturing firms in this sub-group also had high capital intensities (given the
much higher labour productivity indices relative to those for TFP). Exportingfirms that were in operation throughout 19962004 not only had relatively high
productivity in 1996, but experienced a large growth in productivity much
higher than the average for all continuing firms. Overall, Table 4 confirms that
exporting firms are more likely to have higher productivity than non-exporters,
whichever sub-category we consider.
Table 5 and Figure 1 demonstrate that the TFP distribution of exporters
dominates that of non-exporters for each sub-group of firms. 32 Using the
KolmogorovSmirnov test available in STATA (version 9), we are able to reject
29 Note, estimates of productivity growth are known to vary over the business cycle as utilisationof factor inputs vary; however, 1996 and 2004 were neither peaks nor troughs during what wasa relatively stable period of growth in the UK (average GDP growth was 2.7 per cent and 3.2per cent per annum, in 1996 and 2004, respectively).30 Although dealing with the 198092 period, Disney et al. (2003b, Table 6) also report large sharesfor entrants and exiting firms.31 Care needs to be taken when interpreting these relative indices, since they are simple arithmeticmeans that take no account of the relative size of each firm contributing to the overall mean valuereported.32
Separate results for manufacturing and non-manufacturing, while not presented, produced a verysimilar outcome to these results covering all market-based sectors.
-
7/29/2019 j.1467-9701.2007.01087.x (2)
16/24
CONTRIBUTION OF EXPORTING 227
2008 The Authors
Journal compilation Blackwell Publishing Ltd. 2008
TABLE 4Relative Productivity, All Market-Based Sectors, 19962004
Exiting
Firms (t k)
Entering
Firms (t)
Takeovers/
Mergers (t)
Continuing
Firms (t k)
Continuing
Firms (t)
(a) ManufacturingLabour productivityNon-exporters 0.55 1.39 0.96 1.16 1.85Exporters 0.92 0.95 1.24 1.13 1.66All 0.76 1.13 1.22 1.14 1.71
Total factor productivityNon-exporters 0.71 0.89 0.79 0.93 1.01Exporters 1.04 1.03 1.19 1.03 1.28All 0.89 0.97 1.00 1.00 1.21
(b) Non-manufacturing
Labour productivityNon-exporters 0.71 0.90 1.12 0.96 1.02Exporters 1.34 1.78 4.00 1.31 1.38All 0.86 1.09 1.75 1.08 1.14
Total factor productivityNon-exporters 0.83 0.99 1.01 0.99 1.09Exporters 1.14 1.52 1.49 1.18 1.46All 0.90 1.11 1.12 1.06 1.22
(c) All sectorsLabour productivityNon-exporters 0.71 0.96 1.13 1.00 1.12Exporters 1.16 1.58 2.77 1.20 1.47
All 0.86 1.12 1.59 1.09 1.29Total factor productivityNon-exporters 0.79 0.95 0.96 0.96 1.06Exporters 1.12 1.39 1.40 1.15 1.42All 0.90 1.07 1.09 1.05 1.23
Note:Productivity indices are relative to calculated index in 1996 for all firms, based on equation (3) (for eachsub-sector considered).
Source: Tabulations based on weighted FAME data.
TABLE 5
Two-Sample KolmogorovSmirnov Tests on the Distribution of TFP by Sub-Groups, 1996 and 2004
Sub-Group Difference Favourable To:
Exporter Non-Exporter
Continuing firms (t) 0.020 0.247**Mergers/takeovers (t) 0.021 0.269**Entering firms (t) 0.022 0.276**Exiting firms (t k) 0.024* 0.243**
Note:
** Denotes null rejected at 1% level; * null rejected at 5% level. t= 2004; t k= 1996.Source: Calculations based on weighted FAME data.
-
7/29/2019 j.1467-9701.2007.01087.x (2)
17/24
228 RICHARD HARRIS AND QIAN CHER LI
2008 The Authors
Journal compilation Blackwell Publishing Ltd. 2008
the null hypothesis that the distribution is favourable to non-exporters, confirmingthat exporters have a productivity distribution to the right of non-exporters.
Only for firms that exited before 2004 is there some evidence of some significant
cross-over for the two sub-groups (at high values of the empirical cumulative
distributions).
Based on the decomposition of productivity growth set out in equation (4),
Table 6(c) shows firstly that in aggregate, and covering all sectors, exporting
firms contributed more to overall UK productivity growth than non-exporting
firms (i.e. 2.32 per cent compared with 1.55 per cent per annum in terms of
labour productivity, and 1.27 per cent relative to 0.81 per cent per annum interms of TFP33), even though Table 3 shows that in 2004 exporters produced
51.7 per cent of all market-based output. This faster growth was due to relatively
better performance in nearly every category in Table 6(c), except that exporting
firms that exited had relatively high productivity, resulting in a negative impact
on aggregate productivity levels.
With respect to the TFP results in Table 6(c), aggregate productivity for
exporters benefits from a large contribution from continuing firms improving
33
In percentage terms, this amounts to 60.8 per cent of aggregate TFP growth is due to exporters;the latter account for 59.9 per cent of labour productivity growth.
FIGURE 1Productivity-Level Differences between Exporters and Non-Exporters in
Various Sub-Groups, 1996 and 2004
-
7/29/2019 j.1467-9701.2007.01087.x (2)
18/24
CONTRIBUTION OF EXPORTING 229
2008 The Authors
Journal compilation Blackwell Publishing Ltd. 2008
their productivity, while there is little redistribution of resources across exportersthat remain open throughout. The covariance term also indicates that large
continuing exporters in 1996 experienced increases in productivity andincreas-
ing market shares. Exporters that had either entered or been taken-over/merged
also contributed about 38 per cent each of the overall (1.27 per cent per annum)
increase in aggregate productivity. In contrast, most of the TFP improvement
for non-exporters (around 91 per cent of the total) was attributable to lower
productivity firms exiting, rather than from internal improvements or the positive
impact of new firms (or takeovers/mergers).
The labour productivity results in Table 6(c) are broadly similar for exporters,although the contribution to aggregate growth was dominated by the contribution
TABLE 6Decomposition of Productivity Growth, All Market-Based Sectors, 19962004 (cf. equation (4))
Total Within Between Covariance Mergers/
Takeovers
Entering Exit
(a) ManufacturingLabour productivityNon-exporters 1.02 0.15 0.03 0.14 0.03 0.23 0.50Exporters 3.51 1.70 0.36 0.75 0.40 0.06 0.37All 4.53 1.85 0.39 0.89 0.37 0.17 0.87
Total factor productivityNon-exporters 0.07 0.03 0.03 0.03 0.17 0.08 0.29Exporters 1.61 0.83 0.01 0.62 0.32 0.04 0.18All 1.67 0.86 0.04 0.64 0.15 0.05 0.11
(b) Non-manufacturingLabour productivity
Non-exporters 1.76 0.31 0.33 0.24 0.30 0.20 1.41Exporters 1.85 0.34 0.08 0.45 1.79 0.88 0.84All 3.62 0.65 0.25 0.69 2.08 0.68 0.57
Total factor productivity
Non-exporters 1.05 0.04 0.12 0.09 0.05 0.00 0.75Exporters 1.23 0.27 0.03 0.22 0.52 0.64 0.39All 2.28 0.31 0.09 0.31 0.57 0.64 0.36
(c) All sectorsLabour productivity
Non-exporters 1.55 0.19 0.25 0.20 0.26 0.07 1.10
Exporters 2.32 0.19 0.03 0.47 1.45 0.66 0.49All 3.87 0.00 0.27 0.68 1.71 0.59 0.62
Total factor productivity
Non-exporters 0.81 0.04 0.09 0.07 0.07 0.06 0.74Exporters 1.27 0.42 0.01 0.30 0.48 0.48 0.39All 2.09 0.46 0.08 0.38 0.41 0.42 0.35
Note:Figures show the percentage increase per annum.
Source: Tabulations based on weighted FAME data.
-
7/29/2019 j.1467-9701.2007.01087.x (2)
19/24
230 RICHARD HARRIS AND QIAN CHER LI
2008 The Authors
Journal compilation Blackwell Publishing Ltd. 2008
of firms that had merged/been taken over (accounting for over 62 per cent of the
per annum improvement). For non-exporters, there was a relatively large negative
within-firm effect, implying that continuing firms that were relatively large in
1996 had experienced declines in labour productivity between 1996 and 2004.In other respects, the results are similar to those for TFP.
Table 6 also produces results when decomposition of productivity growth is
undertaken separately for manufacturing and non-manufacturing firms.34 Con-
centrating on the TFP results, it can be seen that overall growth per annum was
significantly lower in manufacturing (1.67 per cent compared to 2.28 per cent
per annum in non-manufacturing), due to the poor performance of manufacturing
firms that did not export. The latter mostly reflects the lower TFP contribution
of firms that merged or were taken over, while new firms also contributed
negatively to overall growth. In contrast, manufacturing exporters experienced
significant intra-firm productivity gains (0.83 per cent per annum), with addi-
tional gains for large exporting firms that were also increasing their market shares
(i.e. a covariance effect of 0.62 per cent per annum); while both intra-firm and
covariance effects were much smaller for non-manufacturing exporters. Other
significant differences between exporting firms in manufacturing and non-manu-
facturing were the overall negative impact of churning in manufacturing (with
a net contribution of 0.14 per cent per annum), while new exporters in non-
manufacturing contributed significantly to overall growth in this sub-group.
Finally, Table 7 compares our TFP results with those obtained in other studies
(as discussed in Section 2). Because of the different methods used, the figuresfor various countries and time periods are not strictly comparable; for example,
Bernard and Jensen (2004b) and Hansson and Lundin (2004) used balanced panel
data that do not allow for entry and exit effects. Nevertheless, the table does
show that overall exporters do account for a larger proportion of productivity
growth than do non-exporters in both manufacturing (cf. the US, Canadian and
Swedish figures) and in all market-based sectors (in the UK). However, lack of
comparable information means it is not possible to provide any evidence on
whether productivity growth was mostly due to intra-firm improvements or
external restructuring. The only study (not included in Table 7) that started witha similar approach to ours was Falvey et al. (2004), for Swedish manufacturing
(198094). They used the Haltiwanger-type approach to decompose TFP growth
by industries but did not separate out exporters. Rather, they regressed the net
entry and between-firm contributions (by industry) on export growth (across
industries) and interpreted the results as showing the relationship between
34 These figures cannot be added together to get overall growth across the UK. However, Table 3shows that manufacturings share of real gross output throughout the period was around one-
quarter of total market-based output, which is approximately the weight needed to combine thefigures for manufacturing and non-manufacturing to achieve the overall UK totals.
-
7/29/2019 j.1467-9701.2007.01087.x (2)
20/24
2008TheAuthors
Journalcompilation
B
lackwellPublishingLtd.2008
TABLE 7Decomposition of Productivity Growth, Various Studies
Bernard and Jensen(2004b), Table 8
Hansson and Lundin(2004), Table 10
Baldwin and Gu(2003), Table 13
HarTab
Data and period US LRD, 1983 92 Sweden, 1990 99 Canadian ASM, 1974 96 UKSector Manufacturing Manufacturing Manufacturing All
Decompositionmethodology
Within and betweeneffects only
Within and betweeneffects only
Within + between andentry only
W(me
Methodologicalfeatures
Uses balanced panel soexcludes entry and exit
Uses balanced panel soexcludes entry and exit
Assumes exits replace entrants(thus no separate role for exits)
Userepr
Productivity measures TFP TFP Labour productivity TFPMain results: (% p.a.) (% p.a.) (% of total) (%
Within Between Total Within Between Total Within+Between Entry Total WNon-exporters 0.76 0.48 0.28 1.5 4.2 2.7 3.6 0.9 2.7 0.04Exporters 0.07 1.07 1.14 4.0 2.1 6.1 79.3 18.0 97.3 0.42
Total 0.83 0.59 1.82 5.5 2.1 3.4 82.9 17.1 100 0.46
Note:a The covariance term is added to between, and mergers/takeovers is added to entry from Table 6 to obtain the result
-
7/29/2019 j.1467-9701.2007.01087.x (2)
21/24
232 RICHARD HARRIS AND QIAN CHER LI
2008 The Authors
Journal compilation Blackwell Publishing Ltd. 2008
exporting and productivity growth. However, they ignore intra-firm effects and the
covariance term (in the Haltiwanger decomposition), and mostly their regression
results are statistically insignificant; but they do find evidence to suggest that
export growth raises overall productivity growth through its positive effect on theentry of new firms.
5. CONCLUSIONS
Knowing the contribution of exporting to productivity growth is important;
for example, evidence of the benefits from international trade can support the
UK government when designing policies to help firms develop their exporting
activities (DTI, 2006). We have shown that exporting leads to productivity gains
within firms (confirming the self-selection and/or learning-by-exporting
hypotheses in the exportingproductivity literature), while additionally showing
the importance for UK aggregate productivity growth of inter-firm reallocations
of resources towards more productive exporting firms. Such external restructuring
was especially important in the non-manufacturing sector, suggesting that govern-
ment policy aimed at boosting exporting needs to take account of the different
channels through which productivity growth occurs across different sectors.
Despite some limited micro evidence of trade-induced aggregate productivity
growth for countries like the US, Canada and Sweden, to the best of our knowl-
edge this paper presents the first set of results that assesses the contribution ofexporters to UK aggregate productivity growth between 1996 and 2004, taking
into account restructuring effects. Firstly, and concentrating here on the results
across all market-based sectors, we are able to show that exporting firms con-
tribute a significant amount to the total output of the UK economy, partly due to
their relative size and particularly their better productivity. We have confirmed
that exiting firms tend to have the lowest productivity levels for both exporters
and non-exporters, and new entrants exhibit relatively high productivity (par-
ticularly for exporters). Firms that have been taken over/merged also have
relatively high productivity (especially in conjunction with exporting), and highcapital intensities. Exporting firms in operation throughout 19962004 not only
had relatively high productivity in 1996, but experienced a large growth in
productivity, much higher than the average for all continuing firms. Overall,
these results confirm that exporting firms are more likely to have higher pro-
ductivity than non-exporters, whichever sub-category of firms we consider.
Based on a decomposition of productivity growth we are able to show that,
in aggregate, exporting firms experience faster productivity growth than non-
exporting firms and therefore contribute more to overall productivity growth.
That is, there is a large contribution from continuing firms that exported; and
within productivity gains for exporters were relatively large between 1996 and
-
7/29/2019 j.1467-9701.2007.01087.x (2)
22/24
CONTRIBUTION OF EXPORTING 233
2008 The Authors
Journal compilation Blackwell Publishing Ltd. 2008
2004, with little redistribution of resources across exporters that remained open
throughout. Our results also show that continuing exporters that were initially
large and experienced increases in productivity also experienced increasing
market shares. Exporters that entered global markets after 1996 through beingeither taken over or merged, or as new start-ups, also contributed about 38 per cent
each of the overall growth in aggregate productivity. In contrast, most of the
productivity (labour or TFP) improvement for non-exporters is attributable to
the contribution of lower productivity firms exiting, rather than internal
improvements or new firms entering with high levels of productivity.
REFERENCES
Aitken, B., G. H. Hanson and A. E. Harrison (1997), Spillovers, Foreign Investment, and ExportBehaviour,Journal of International Economics, 43, 10332.
Arellano, M. and S. R. Bond (1998), Dynamic Panel Data Estimation using DPD98 for GAUSS:A Guide for Users, mimeo. Available at http://www.american.edu/academic.depts/cas/econ/gaussres/regress/dpd/dpd98.pdf.
Aw, B. Y. and A. R. Hwang (1995), Productivity and the Export Market: A Firm-Level Analysis,Journal of Development Economics, 47, 2, 31332.
Aw, B. Y., X. Chen and M. J. Roberts (1997), Firm-Level Evidence on Productivity Differentialsand Turnover in Taiwanese Manufacturing, NBER Working Paper No. 6235.
Aw, B. Y., S. Chung and M. J. Roberts (2000), Productivity and Turnover in the Export Market:Micro-Level Evidence from the Republic of Korea and Taiwan (China), The World Bank
Economic Review, 14, 6590.Baldwin, J. R. and W. Gu (2003), Export-Market Participation and Productivity Performance inCanadian Manufacturing, Canadian Journal of Economics, 36, 3, 63457.
Baldwin, J. R. and W. Gu (2004), Trade Liberalisation: Export-Market Participation, ProductivityGrowth, and Innovation, Oxford Review of Economic Policy, 20, 3, 37292.
Bernard, A. B. and J. B. Jensen (1999), Exceptional Exporter Performance: Cause, Effect, orBoth?,Journal of International Economics, 47, 1, 125.
Bernard, A. B. and J. B. Jensen (2004a), Why Some Firms Export, Review of Economics andStatistics, 86, 2, 56169.
Bernard, A. B. and J. B. Jensen (2004b), Exporting and Productivity in the USA, Oxford Reviewof Economic Policy, 20, 34357.
Bernard, A. B. and J. Wagner (1997), Exports and Success in German Manufacturing, Review of
World Economics,133
, 1, 13457.Bernard, A. B., J. Eaton, J. B. Jensen and S. Kortum (2003), Plants and Productivity inInternational Trade,American Economic Review, 93, 126890.
Bernard, A. B., S. Redding and P. K. Schott (2007), Comparative Advantage and HeterogeneousFirms,Review of Economic Studies, 74, 3166.
Bleaney, M. and K. Wakelin (2002), Efficiency, Innovation and Exports, Oxford Bulletin ofEconomics and Statistics, 64, 1, 315.
Blundell, R. and M. Costa Dias (2000), Evaluation Methods for Non Experimental Data,Fiscal Studies, 21, 42768.
Clerides, S. K., S. Lach and J. R. Tybout (1998), Is Learning by Exporting Important?, QuarterlyJournal of Economics, 113, 90348.
Damijan, P. J., S. Polanec and J. Pra8nikar (2005), Does Exporting Increase Productivity? FirmLevel Evidence from Slovenia, paper presented at conference on Globalisation and Firm LevelAdjustment (University of Nottingham).
http://www.american.edu/academic.depts/cas/econ/gaussres/regress/dpd/dpd98.pdfhttp://www.american.edu/academic.depts/cas/econ/gaussres/regress/dpd/dpd98.pdfhttp://www.american.edu/academic.depts/cas/econ/gaussres/regress/dpd/dpd98.pdf -
7/29/2019 j.1467-9701.2007.01087.x (2)
23/24
234 RICHARD HARRIS AND QIAN CHER LI
2008 The Authors
Journal compilation Blackwell Publishing Ltd. 2008
Das, S., M. J. Roberts and J. R. Tybout (2001), Market Entry Costs, Producer Heterogeneity, andExport Dynamics, NBER Working Paper No. 8629.
Disney, R., J. Haskel and Y. Heden (2003a), Restructuring and Productivity Growth in UKManufacturing,Economic Journal, 113, 66694.
Disney, R., J. Haskel and Y. Heden (2003b), Entry, Exit and Establishment Survival in UKManufacturing,Journal of Industrial Economics, 51, 91112.
DTI (2006), International Trade and Investment The Economic Rationale for GovernmentSupport, DTI Economics Paper No. 18.
Falvey, R., D. Greenaway, Z. Yu and J. Gullstrand (2004), Exports, Restructuring and IndustryProductivity Growth, GEP Research Paper No. 2004/40.
Foster, L., J. Haltiwanger and C. Krizan (2001), Aggregate Productivity Growth: Lessons fromMicroeconomic Evidence, in C. R. Hulten, E. R. Dean and M. J. Harper (eds.), New Develop-ments in Productivity Analysis (Chicago: University of Chicago Press).
Girma, S., D. Greenaway and R. Kneller (2004), Does Exporting Increase Productivity? A Micro-econometric Analysis of Matched Firms, Review of International Economics, 12, 5, 85566.
Gourlay, A. and J. Seaton (2004), Explaining the Decision to Export: Evidence from UK Firms,Applied Economics Letters, 11, 15358.Greenaway, D. and R. Kneller (2004), Exporting Productivity in the United Kingdom, Oxford
Review of Economic Policy, 20, 35871.Greenaway, D. and R. Kneller (2005), Exporting and Productivity: Theory, Evidence and Future
Research, Singapore Economic Review, 50, 30312.Greenaway, D. and Kneller, R. (2007), Firm Heterogeneity, Exporting and Foreign Direct Invest-
ment, Economic Journal, 117, F13461.Greenaway, D. and Z. Yu (2004), Firm Level Interactions between Exporting and Productivity:
Industry-Specific Evidence,Review of World Economics, 140, 37692.Greenaway, D., N. Sousa and K. Wakelin (2004), Do Domestic Firms Learn to Export from
Multinationals?, European Journal of Political Economy, 20, 4, 102743.
Griffith, R. (1999), Using the ARD Establishment Level Data to Look at Foreign Ownership andProductivity in the United Kingdom,Economic Journal, 109, F41622.Haltiwanger, J. (1997), Measuring and Analyzing Aggregate Fluctuations: The Importance of
Building from Microeconomic Evidence, Federal Reserve Bank of St. Louis Review, 5578.Hansson, P. and N. Lundin (2004), Exports as an Indicator on or Promoter of Successful Swedish
Manufacturing Firms in the 1990s, Review of World Economics, 140, 41545.Harris, R. I. D. (2002), Foreign Ownership and Productivity in the United Kingdom Some Issues
When Using the ARD Establishment Level Data, Scottish Journal of Political Economy, 49,31835.
Harris, R. I. D. (2004), DTI Industrial Support Policies: Key Findings from New Micro-DataAnalysis, DTI Economics Paper No. 8.
Harris, R. I. D. (2005), Economics of the Workplace: Special Issue Editorial, Scottish Journal ofPolitical Economy, 52, 3, 32343.
Harris, R. I. D. and Q. Li (2005), Review of the Literature: The Role of International Trade andInvestment in Business Growth and Development, Report to UKTI.
Harris, R. I. D. and Q. Li (2007a), Learning-by-Exporting? Firm-Level Evidence for UKManufacturing and Services Sectors, Department of Economics Discussion Paper No. 2007-22(University of Glasgow).
Harris, R. I. D. and Q. Li (2007b), Firm Level Empirical Study of the Contribution of Exportingto UK Productivity Growth. Report to UK Trade and Investment, available at http://www.uktradeinvest.gov.uk/UKTI/fileDownload/FAMEFinalReport2007v2.pdf?cid=401169
Heckman, J. and S. Navarro-Lozano (2004), Using Matching, Instrumental Variables and ControlFunctions to Estimate Economic Choice Models, Review of Economics and Statistics, 86, 3057.
Helpman, E., M. J. Melitz and S. R. Yeaple (2004), Export versus FDI with Heterogeneous Firms,American Economic Review, 94, 30016.
http://www.uktradeinvest.gov.uk/UKTI/fileDownload/FAMEFinalReport2007v2.pdf?cid=401169http://www.uktradeinvest.gov.uk/UKTI/fileDownload/FAMEFinalReport2007v2.pdf?cid=401169http://www.uktradeinvest.gov.uk/UKTI/fileDownload/FAMEFinalReport2007v2.pdf?cid=401169 -
7/29/2019 j.1467-9701.2007.01087.x (2)
24/24
CONTRIBUTION OF EXPORTING 235
2008 The Authors
Journal compilation Blackwell Publishing Ltd 2008
Kneller, R. and M. Pisu (2007), Industrial Linkages and Export Spillovers from FDI, The WorldEconomy, 30, 10534.
Lpez, R. A. (2004), Self-Selection into the Export Markets: A Conscious Decision? IndianaUniversity Paper. Available at http://gemini.econ.umd.edu/cgi-bin/conference/download.cgi?db_
name=mwetit2005&paper_id=22.Lpez, R. A. (2005), Trade and Growth: Reconciling the Macroeconomic and Microeconomic
Evidence,Journal of Economic Surveys, 19, 62348.Melitz, M. J. (2003), The Impact of Trade on Intra-Industry Reallocations and Aggregate Industry
Productivity,Econometrica, 71, 6, 1695725.Newey, W. K. and D. L. McFadden (1999), Large Sample Estimation and Hypothesis Testing,
in D. L. McFadden and R. F. Engle (eds.), Handbook of Econometrics, Vol. 4 (Amsterdam:North-Holland).
Oulton, N. (1997), The ABI Respondents Database: A New Resource for Industrial EconomicsResearch,Economic Trends, 528, 4657.
Oulton, N. (2000), A Tale of Two Cycles, Closure Downsizing and Productivity Growth in UKManufacturing, 197389, National Institute of Economic and Social Research Review, 173,
6679.Oviatt, B. and P. McDougall (1995), Global Start-Ups: Entrepreneurs on a Worldwide Stage,Academy of Management Executive, 9, 3043.
Roberts, M. and J. R. Tybout (1997), The Decision to Export in Colombia: An Empirical Modelof Entry with Sunk Costs, American Economic Review, 87, 54564.
Roberts, M., T. Sullivan and J. Tybout (1995), Micro Foundations of Export Boom, Departmentof Economics Working Paper (Pennsylvania State University).
Schumpeter, J. A. (1950), Capitalism, Socialism and Democracy (New York: Harper & Row).Silvente, F. R. (2005), Changing Export Status and Firm Performance: Evidence from UK Small
Firms,Applied Economics Letters, 12, 56771.Sjoholm, F. (2003), Which Indonesian Firms Export? The Importance of Foreign Networks,
Papers in Regional Science, 82, 33350.
Wagner, J. (2007), Exports and Productivity: A Survey of the Evidence from Firm Level Data,The World Economy, 30, 6082.Wang, H. J. and P. Schmidt (2002), One-Step and Two-Step Estimation of the Effects of
Exogenous Variables on Technical Efficiency Levels, Journal of Productivity Analysis, 18,12944.
http://gemini.econ.umd.edu/cgi-bin/conference/download.cgi?db_name=mwetit2005&paper_id=22http://gemini.econ.umd.edu/cgi-bin/conference/download.cgi?db_name=mwetit2005&paper_id=22http://gemini.econ.umd.edu/cgi-bin/conference/download.cgi?db_name=mwetit2005&paper_id=22