40
©Richard Harris and Q Cher Li
44
Learning-by-Exporting Hypothesis
3.12 Turning now to the learning-by-exporting hypothesis, export-oriented firms are
assumed to experience an acceleration in productivity growth following entry. If
this is not true, this has important policy implications – if better firms do self-
select into export markets, and exporting does not further boost productivity,
then export subsidies could simply be a waste of resources (involving large-
scale dead weight and possibly even displacement effects given that firms that
export also usually sell to domestic markets as well
26
).
3.13 The learning-by-exporting proposition has, unfortunately, received somewhat
less support in the literature. Many early empirical studies raised doubts about
the causality running from exporting to productivity, since they found
productivity growth did not increase post entry, notwithstanding that exporting
firms on average experienced significantly higher growth in terms of
employment and wages. (Aw and Hwang, 1995, for Taiwan; Bernard and
Jensen, 1995 and 1999a, for the US; Bernard and Wagner, 1997, for Germany;
Clerides et al., 1998, for Columbia, Mexico and Morocco; Delgado et al., 2002,
for Spain; Wagner, 2002, for Germany). For example, applying a novel non-
parametric analysis of productivity distributions for Spanish firms, Delgado et
al. (2002) failed to find significant differences between new exporters and
continuing exporters by analysing the post entry productivity growth
distribution. Analogically, exporters were found to be no different from non-
exporters, although limited learning effects could be found among younger
exporters.
3.14 Nevertheless, some of the literature covered in Chapter 2, particularly in the
business management field, emphasises the importance of exporting (or
internationalisation in general) as a learning process. The process of going
international is perceived as a sequence of stages in the firm’s growth trajectory,
which involves substantial learning (and innovating) through internal and
external channels, so as to enhance its competence base and improve its
performance. Thus, the learning-by-exporting proposition is consistent with
26
Robust empirical evidence shows that exporters tend to sell very small fractions of their output
abroad (Aw et. al., 1997; Campas, 1999; Sullivan et. al., 1995). Note also, UK government policy is
not to provide subsidies to exporters but to rather increase export market entry through combating
market failures – see Chapter 4. However, the issue of deadweight and possible displacement is still
relevant – see the discussion in par. 4.32 (chapter 4).
38
©Richard Harris and Q Cher Li
45
other areas of literature on business internationalisation. Indeed, positive
learning effects for firms engaged in exporting have been identified, particularly
for some of the economics literature and where different econometric
methodologies are adopted.
3.15 For instance, in an attempt to examine the learning-by-exporting hypothesis,
Kraay (1999) finds (using data for a sample of Chinese industrial enterprises)
that past export is significantly associated with better total factor and labour
productivity performance and he further shows that these learning effects are
most pronounced among established exporters although they can be
insignificant and occasionally negative in new entrants to export markets..
Moreover, in a firm-level survey on manufacturing productivity in five East
Asian economies, Hallward-Driemeier et. al. (2002) not only identify higher
productivity post export-market entry but go one step further to explore the
sources and mechanisms of this productivity growth – it is in aiming for export
markets that firms consistently make a series of decisions that consequently
accelerate their productivity, with regard to their investment, training,
technology, selection of inputs etc. This is consistent with the notion of
absorptive capacity and the resource-based view discussed in Chapter 2.
27
3.16 What’s more, there’s also a strand of literature documenting evidence on the co-
existence of selection and learning effects. Baldwin and Gu (2003) explore the
export-productivity linkage in Canadian manufacturing and find evidence that
productivity improves following export-market participation; in contrast to
Kraay (1999) they find learning effects of export are stronger for younger
businesses. Using data for the UK chemical industry, Greenaway and Yu (2004)
test both hypotheses and find strong evidence that firms self-selected into export
markets; they however also report more varied learning effects dependent on the
age of establishments – significant and positive for new entrants, less significant
for more experienced exporters and negative for established exporters. More
recently, Girma et al. (2004) use ‘propensity score matching’ techniques to
overcome problems of selectivity bias when evaluating the causal effect of
exporting on performance characteristics, and thus suggest that firms do self-
27
Castellani (2002) also reports a positive relationship between labour productivity and exporting
intensity for Italian firms between 1989-1994– only firms substantially involved in exporting have
significantly faster productivity growth.
36
©Richard Harris and Q Cher Li
46
select into export markets but that exporting also further boosts firm-level
productivity.
3.17 Arguably the empirical evidence still remains inconclusive regarding the causal
mechanisms underlying the well-established empirical association between
export orientation and productivity growth, in particular whether the learning-
by-exporting hypothesis holds. Nevertheless, there may be several explanations
to account for such discrepancies amongst the empirical literature in this area.
3.18 To begin with, there are structural differences between the various databases
used when testing for learning effects. Baldwin and Gu (2004) put forward a
convincing explanation as to why there should be different learning effects in
Canadian and US plants: learning from international best practices was more
important for productivity growth in Canadian plants that export vis-à-vis US
plants, whose principal source of raising productivity is technology developed
domestically. In addition, given a smaller market size in Canada where
competition is not as intense as in the US, exposure to international competition
is more likely to induce participants to become more productive and
competitive. Thirdly, expanding into much larger foreign markets relative to
domestic market, Canadian producers will benefit from greater product
specialisation and longer production runs, which is more likely to have an
impact on productivity; whereas this is less of an issue in US firms given the
already bigger domestic market. all of these will contribute to a greater export
impact on productivity growth in Canada.
3.19 Similar mechanisms of raising productivity may also apply in the UK. For
instance, learning benefits are likely to be less in the US firms that export vis-à-
vis UK firms, since the US firms are overall likely to be closer to technological
frontier (which is set by the US), and they are also exposed to a more
competitive market (Girma et al., 2004). In contrast, Sweden has a high
participation rate for firms involved in export markets and high degree of
openness, which to some extent resembles more the US economy. This may
partly explain the similar performance profiles found between Swedish
exporters and non-exporters (Greenaway et. al., 2003).
3.20 In addition to these country-specific differences associated with the learning
process, firm performance characteristics may well differ both within and across
industries as well. From a resource-based viewpoint, in order to learn when
44
©Richard Harris and Q Cher Li
47
operating in foreign markets, and in order to internalise international knowledge
spillovers, a firm needs to invest more in training and innovation so as to
enhance its absorptive capability to exploit and assimilate (often tacit)
knowledge that is obtained externally. This argument is substantiated by the
evidence of significant learning effects uncovered in the UK chemical industry,
which is a typical high-tech sector that undertakes a large amount of R&D
expenditure. (Greenaway and Yu, 2004).
3.21 Secondly, the heterogeneity of export markets may also play a role in
determining the extent to which participants will gain higher productivity from
exporting. For instance, Damijan et al. (2005) suggest that learning from
exporting is crucially dependent on the degree of competitive pressures facing
firms in different foreign markets – exporting per se does not warranty
productivity gains; rather, productivity only improves significantly when firms
are serving advanced, high-wage export markets.
3.22 Lastly and most importantly, there are also certain methodological issues
involved when testing for productivity effect of exporting. A problem usually
encountered in microeconometric evaluation studies is sample selectivity.
28
.
This problem arises when making comparisons between a ‘treatment group’
(e.g. export-market entrants) and the rest of the population, when it is known (or
at least suspected) that the treatment group are not drawn randomly from the
whole population. This issue is of paramount importance when interpreting the
results obtained from comparing exporters and non-exporters, and upon which
policy conclusions are then based.
29
3.23 More specifically, participants in export markets may posses certain
characteristics such that they achieve better performance (in terms of higher
productivity) vis-à-vis non-participants even when they do not enter export
markets, and this productivity gain is correlated with the decision to participate
28
Another possible econometric problem may arise when most of the empirical studies tend to pool
information across all firms with heterogeneous export histories to examine these learning effects of
exporting. In fact, distinct learning effects are uncovered amid firms of different age (Krray, 1999;
Delgado et. al., 2002; Baldwin and Gu, 2003; Greenaway and Yu, 2004). For instance, Krray (1999)
allows export history to have an effect on learning effects (by allowing the coefficient on lagged export
to vary with the export history of the plant), and finds significantly positive effects of exporting merely
in more established Chinese firms.
29
See Harris (2005) for a brief survey; Blundell and Costa Dias (2000) and Heckman and Navarro-
Lozano (2004) for a comprehensive review of the sample selectivity issue and various approaches to
this. More technicalities of the selection problem are treated in the Appendix for interested readers.
39
©Richard Harris and Q Cher Li
48
in the global market. This will mean that standard estimation techniques lead to
biased results. These characteristics would likely include superior managerial
capability, organisational skills, absorptive capacity, etc. They are associated
both with achieving higher productivity and the decision to self-select into
export markets.
3.24 That ignoring selectivity problem leads to biased results is indeed reinforced by
the theoretical evidence of the heterogeneity of firm productivity prior to entry
(Head and Ries, 2003; Melitz, 2003; Helpman et. al., 2004) and the unanimous
empirical evidence of significant differences between exporters and non-
exporters (in terms of productivity, employment, capital-intensity, R&D, etc.),
but similar characteristics between new exporters and continuing exporters.
3.25 Several standard approaches have been proposed in the literature to combat this
selection problem. One approach is ‘matching’ i.e. selecting a valid ‘control’
group to compare exporters’ performance with only those non-exporters with
similar characteristics to those that export are chosen for the control group. This
approach therefore assumes that the treated and non-treated groups are
effectively the same in all relevant respects (except the non-treated group do not
export) so that the productivity outcome that would prevail in the absence of
treatment is the same in both cases. Using a propensity score matching
approach, Girma et. al., (2004) found significantly positive post-entry learning
effects for UK exporters.
3.26 Another technique for eliminating selectivity bias is the difference-in-difference
estimator. For example, in conjunction with matching, Greenaway and Kneller
(2004) use a difference-in-difference approach to control for changes in other
observable determinants of productivity post entry, and find that there are
significant productivity gains from exporting in the unmatched sample but these
disappear when they use a matched sample. Other approaches suggested in the
literature to deal with self-selection bias include instrumental variable
estimation and Heckman two-stage estimation, which are closely linked in a
way.
30,31
30
For instance, Kneller and Pisu (2005) provide an example of deploying Heckman selection process to
model two decisions of whether to export or not and how much to export, but in a different setting -
export spillovers from FDI. To our knowledge, there are few studies utilising instrumental variable
estimation to examine the causality between export and productivity, possibly due to lack of
appropriate instruments.
41
©Richard Harris and Q Cher Li
49
3.27 In a nutshell, although the empirical literature presents compelling evidence in
favour of the self-selection hypothesis, the findings are less conclusive with
respect to the learning-by-exporting hypothesis. The results obtained in the
literature are of great importance for policy marking and their policy
implications will be discussed in Chapter 4.
Other Characteristics of Firm-Level Adjustment
3.28 In order to get a complete picture of firm-level adjustment to globalisation, we
also need to examine some other factors characterising firm behaviour in a
global market. This will help to put the export-productivity relationship into
context, and explain the underlying resources for such a relationship.
Innovation
3.29 First and foremost, innovation is generally perceived as the major driving force
behind exporting in trade theories (Vernon, 1966; Krugman, 1979, 1995). From
a firm perspective, exporters need to invest in R&D and training to develop
internally by absorbing, assimilating and managing technologies and ideas
obtained in foreign markets. Innovation facilitates a firm’s competency
development and brings about scale and scope economies. The resulting greater
production efficiency enables firms to expand their domestic market share
through import substitution, and most importantly, to penetrate new foreign
markets and increase their exports share.
32
This is in line with the notion of
absorptive capacity and the crucial role of R&D in developing such capacity,
thereby allowing firms to internalise external knowledge (Cohen and Levinthal,
1989, 1990). This may help to explain how differences in productivity effect
export-market participation as observed in heterogeneous firms, industries and
countries.
33
Empirically, Bleaney and Wakelin (2002) and Roper and Love
(2002) have reported significant differences in terms of R&D expenditures at
plant level between exporters and non-exporters in UK manufacturing, and thus
31
A comparison of relative merits of all approaches is available in the Appendix.
32
Note, firms that export usually only sell a small proportion of their output in foreign markets.
Therefore when they expand due to efficiency gains, they can capture additional shares in both
domestic and foreign markets.
33
See Aw et. al. (2000) for a comparative study between Taiwan and South Korea.
32
©Richard Harris and Q Cher Li
50
the moderating effect of innovation on the export-productivity nexus; similar
findings are also suggested for the US (Bernard and Jensen, 2001), and Canada
(Baldwin and Gu, 2004). In particular, Baldwin and Gu (2004) made use of data
for Canadian manufacturers to test whether exporters had higher levels of R&D.
The results show that undertaking R&D is 10% higher (after controlling for
other relevant covariates such as size) for exporters (but there is no statistically
significant differential in favour of exporters prior to their
internationalisation).Thus, they show some evidence for increased innovation
activity after internationalising, which is consistent with their arguments that
benefits from export-market entry are not ‘automatic’ – in order to achieve post-
entry productivity gains, exporters invest more in R&D and human capital to
acquire more foreign technologies and develop enhanced absorptive capacities.
Industrial/Spatial Agglomeration
3.30 Others concentrate on the role of certain structural factors in increasing the
probability of export market entry. Firstly, the importance of geographic factors
is captured in Overman et. al.’s (2003) survey of the literature on the economic
geography of trade flows and the location of production. If information on
foreign market opportunities and costs is asymmetric, then it is reasonable to
expect firms to cluster within the same industry/region so as to achieve
information sharing and therefore minimise entry costs. Co-location may help
improve information about foreign markets and tastes so as to provide better
channels through which firms distribute their goods (Aitken et. al., 1997). There
are usually two dimensions to these agglomeration effects – a regional effect
and an industrial effect. The former comprises the spatial concentration of
exporters (from various industries). Whereas the industry effect is where
exporting firms from the same industry co-locate. Greenaway and Kneller
(2004) provide empirical evidence that shows that the industrial dimension of
agglomeration would appear to be more important for the UK while Bernard
and Jensen (2001) found it to be insignificant in explaining the probability of
exporting in the US. The benefits brought about by the co-location of firms on
36
©Richard Harris and Q Cher Li
51
the export decision have also been documented in other empirical studies, for
instance, Aitken et al. (1997) for Mexico.
34
Market Concentration
3.31 In a similar way, market concentration is also expected to positively impact
upon a firm’s propensity to export and its performance post entry. A high level
of Concentration of exporters within an industry may improve the underlying
infrastructure that is necessary to facilitate access to international markets or to
access information on the demand characteristics of foreign consumers.
Therefore, we might expect a higher propensity for non-participants to go
international in a market with a higher degree of concentration of export
activity. Evidence for UK manufacturing covering the 1988 to 2002 period is
provided by Greenaway and Kneller (2003).
Export Spillovers
3.32 Alongside these location effects is the impact of export spillovers, i.e.
knowledge spillovers from foreign firms that impact on the export decision of
domestic firms. These spillovers take place if there is a transfer of knowledge
from about foreign markets to domestic firms. This linkage is derived from the
literature on international knowledge diffusion. International trade is argued to
be a conduit for the transfer of knowledge and thus conducive to productivity
growth (Grossman and Helpman, 1991). From a firm perspective, participation
in international markets brings firms into contact with international best
practices and facilitates learning and competency development. Following Coe
and Helpman’s (1995) seminal piece on international R&D spillovers, there has
been an increasing interest on the impact of spillovers. It is widely felt that such
spillovers provide positive information externalities (Aitken et. al., 1997), and
as a public good these knowledge spillovers cane help domestic recipients to
achieve higher technological standards with less effort.
3.33 The positive effect of export spillovers result from both supply and demand
side impacts. The supply side argument is derived from the existence of sunk
entry costs as discussed in Chapter 2. Export market entry costs arise as a result
34
In contrast, in a recent study for US plants, Bernard and Jensen (2004) find negligible spillovers
resulting from the export activities of other plants; nevertheless, this discrepancy between other studies
35
©Richard Harris and Q Cher Li
52
of imperfect information when establishing foreign marketing channels,
developing new packaging/product varieties, and learning bureaucratic
procedures, etc. By their very nature, information spillovers can significantly
reduce any problems of information asymmetry and therefore lower start up
costs, so allowing rational firms to enter export market when the present value
of their anticipated profits exceed current fixed costs. In contrast, there may also
be a demand-side impact associated with export spillovers: following the
establishment of a presence in overseas market, foreign awareness of (and thus
demand for) domestically produced goods may also rise, pulling more domestic
firms into export markets.
3.34 In addition, Kneller and Pisu (2005) examined the role of FDI industrial
linkages in explaining export activity at the firm-level. They found that the
decision to enter an export market was positively related to the presence of
foreign plants in the same industry and region; the decision concerning how
much to export was affected positively by the presence of foreign firms in
downstream industries. In a recent study using a large panel of UK firms,
Greenaway et. al. (2004) also find evidence of positive spillover effects from
multinational enterprises (henceforth MNEs) on the decision to export of
domestic (UK) firms, and on their export propensity.
International outsourcing
3.35 Finally, we detect a growing interest in the literature of the impact of
international outsourcing on productivity in globalised firms. The rationale for
expecting a positive effect from outsourcing in international markets is
consistent with the notion of learning and absorptive capacity as discussed in
Chapter 2. As pointed out by Görg, et. al. (2005), in the short run domestic
plants that are engaged in international outsourcing may have greater access to
internationally traded inputs at lower costs/higher quality than is available
domestically; in the long run, such outsourcing activity may also bring about a
reallocation of factor shares, and consequently a further impact upon
productivity. Therefore we might expect the increasing use of internationally
traded inputs to boost productivity in these ‘extroverted’ plants.
may be explained by their sample selection criteria (restrictive to large plants only) and measures of
industry (2 digit level) and regions (measured by states).
Documents you may be interested
Documents you may be interested
