Abstract
We present a descriptive analysis of firm-level merchandise trade, focussing on the role of novel exporting behaviour. We document two aspects of the dynamics of trade—the contribution of novel activity to aggregate export growth and, conversely, the substantial exit rates of new trade relationships. The unique contribution of this paper lies in the detailed and comprehensive data we have available on market and product choices. Specifically, we make use of shipment-level goods trade data, linked to information for the universe of economically active New Zealand manufacturers, to examine trade at the firm level and at the product-country-firm nexus. Our growth decomposition and survival analysis suggest several themes: (a) novel market entry is a significant contributor to aggregate export growth; (b) the study of international entrepreneurial behaviour should encompass not just de novo entrants, but the broad range of trade innovations initiated by incumbent exporters; (c) much expansion in trade appears to be incremental in nature; (d) despite this, such innovations appear to be inherently risky; and (e) experience and scale appear to be key factors in overcoming these risks (or at least proxies for such factors).
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
Alternatively, if past export experiences have been unsuccessful, there may be reputational barriers that raise the cost of re-entry or the initial entry costs to new firms.
For a more detailed firm-level study of the role of experience in explaining entry behaviour see Fabling et al. (2009).
The primary threshold for appearing on SNZ’s Business Frame is NZ$30,000 in annual taxable income and the threshold for mandatory Customs filing is NZ$1,000. Unfortunately, similarly comprehensive service trade data do not exist and so the service sector is excluded from the analysis.
From 1998 onwards, over 99 percent of aggregate export value is matched to firms in the LBD. Trade not linked to manufacturers is primarily associated with independent wholesale and retail trade firms (Fabling and Sanderson 2008).
The many revisions (mostly minor) to this system are accounted for by grouping together goods that ever share the same code (using the code accompanying Abowd et al. (2002)). In effect this slightly reduces the resolution at which we observe goods, but has the advantage of consistently classifying goods over time—an important requirement for our subsequent analysis.
We work in financial years to ensure that exports and labour inputs are aligned to the same period. Inland Revenue Department rules mean that most manufacturers report to a 31st March year-end.
Specifically, total employment is the sum of the monthly average employee count as at the 15th of each month plus an annual count of working proprietors. These data come from the Linked Employer-Employee Dataset (LEED), which is based on Pay-As-You-Earn (PAYE) employer and other tax records. Since PAYE non-compliance is likely to be negligible in this population, we assume firms and plants are non-employing if they are absent from LEED.
In fact, we could think of this sort of link as either implying firm continuity or as simply recognising that knowledge (e.g., of how to export) could be maintained in continuing plants.
It turns out that few firms are affected by this issue—without using PBN links we have 2,736 firms that appear not to be exporting between 1996 and 1999, who are exporting over 2004–2006. When we allows for PBN links, this number drops to 2,583. To understand why relationship-level statistics are not affected either, we need simply to observe that the average duration of a firm identifier is much higher than the average duration of a relationship so that, at random, only a very small proportion of relationships would be “cut in two” by firm identifier breaks.
We reiterate, this distribution is visually identical if we include PBN links.
This measure is not available prior to 2000 due to the absence of employment data.
Figures for incumbent firms are net in the sense that the growth in some firms’ trade is offset by declines in trade for firms reducing initial export levels. By way of example, under the no-links method, incumbents contribute 83% of aggregate growth in net terms. The equivalent gross calculation shows that expanding traders contribute 107%, whilst shrinking traders contribute −24%. Similarly at the relationship level, firms shift emphasis between product lines and markets meaning that some pre-existing relationships grow while others decline.
Of course, at some point in time every firm has been an entering exporter. This decomposition, in essence, limits our definition of entrant to events that are new to the last eight years (1999–2006) and have been sustained into the second observation period (2004–2006). Such an approach, which naturally focuses on the early life of most firms in the population, is a common feature of the emerging “international new ventures” literature (Acs et al. 2003).
This disaggregation supports the idea that the PBN links methodology is appropriate. To see this, consider the random linking of two firms (i.e., not based on a real world change in ownership of a manufacturing plant). In such a case, with 10,000 goods and 200 countries, the chance that the two firms share the same product-country mix is minuscule. Thus, if the PBN links did not pick up true firm (or at least plant) continuity we would expect most of the $0.39 billion in trade that is reallocated from entering firms to incumbents to be classified as new relationships. This does not happen—the share of aggregate trade growth in new relationships is almost unchanged.
Consider a hypothetical firm which was exporting apples to Fiji and pears to Australia between 1996 and 1998. A “new combination of existing” trade would be the export of pears to Fiji in 2006.
Besedeš and Prusa use the seven-digit Tariff Schedule to define distinct goods.
Besedeš and Prusa (2006b) go on to show that there are significant differences in the average duration of aggregate spells depending on the level of product differentiation.
Survival estimates are calculated using the Stata (version 9) sts graph command with Greenwood confidence bands. Fabling and Sanderson (2008) estimated similar functions on a weighted basis, accounting for the fact that firms may enter more than once during the observation period and that these observations are not independent. Since their unweighted and weighted results were very similar, we report only unweighted results.
There are clearly a number of additional factors that could be put forward to explain why some trading relationships last longer than others—for example, product-specific characteristics (e.g., some products such as butter may inherently experience more continuous demand). Given the breadth of potential factors, the difficulty in accurately classifying each relationship to different classes, and a preference for brevity, we have restricted ourselves to two.
For example, Fabling and Grimes (2008) demonstrate how size proxies for experience in analyses of foreign currency hedging behaviour.
Tests by SNZ suggest that probabilistic matching is an adequate technique.
To be “economically active” a firm must be observed in our broad-ranging administrative data as either: selling products, purchasing intermediate inputs, employing staff or working proprietors, holding physical capital, or trading (exporting or importing) goods.
References
Abowd J, Creecy R, Kramarz F (2002) Computing person and firm effects using linked longitudinal employer-employee data. LEHD Program Technical Papers TP-2002-06, US Census Bureau
Acs Z, Dana LP, Jones M (2003) Towards new horizons: the internationalisation of entrepreneurship. J Int Entrep 1(1):5–12
Besedeš T, Prusa T (2006a) Ins, outs, and the duration of trade. Can J Econ 39(1):266–295
Besedeš T, Prusa T (2006b) Product differentiation and duration of US import trade. J Int Econ 70(2):339–358
Clerides S, Lach S, Tybout J (1998) Is learning by exporting important? Micro-dynamic evidence from Colombia, Mexico, and Morocco. Q J Econ 113(3):903–947
Fabling R (2007) How innovative are New Zealand firms? Quantifying and relating organisational and marketing innovation to traditional science and technology indicators. In: Science, technology and innovation indicators in a changing world: responding to policy needs. OECD, Paris
Fabling R, Grimes A (2008) Over the hedge? Exporters’ optimal and selective hedging choices. Discussion Papers DP2008/14, Reserve Bank of New Zealand
Fabling R, Sanderson L (2008) Firm level patterns in merchandise trade. Occasional Paper 08/03, Ministry of Economic Development
Fabling R, Sanderson L (2009) Entrepreneurship and aggregate merchandise trade growth in New Zealand. Discussion Papers DP2009/09, Reserve Bank of New Zealand
Fabling R, Grimes A, Sanderson L (2009) Whatever next? Export market choices of New Zealand firms. Discussion Papers DP2009/19, Reserve Bank of New Zealand
Feenstra R, Markusen J, Rose A (2001) Using the gravity equation to differentiate among alternative theories of trade. Can J Econ 34(2):430–447
Greenaway D, Kneller R (2007) Firm heterogeneity, exporting and foreign direct investment. Econ J 117(517):F134–F161
Leonidou L, Katsikeas C (1996) The export development process: an integrative review of empirical models. J Int Bus Stud 27(3):517–551
Melitz M (2003) The impact of trade on intra-industry reallocations and aggregate industry productivity. Econometrica 71(6):1695–1725
Roberts M, Tybout J (1997) The decision to export in Colombia: an empirical model of entry with sunk costs. Am Econ Rev 87(4):545–564
Shaw V, Darroch J (2004) Barriers to internationalisation: a study of entrepreneurial new ventures in New Zealand. J Int Entrep 2(4):327–343
Simmons G (2002) Growing pains: New Zealand qualitative evidence on hurdles to exporting growth. Working Paper 02/10, New Zealand Treasury
Skilling D, Boven D (2006) The flight of the Kiwi: going global from the end of the world. Discussion Paper 2006/01, The New Zealand Institute
Wagner J (2007) Exports and productivity: a survey of the evidence from firm-level data. World Econ 30(1):60–82
Acknowledgements
We are indebted to Ross Johnstone for the initial linking of the trade data, Hamish Hill for helpful discussion on group structures, and Dave Maré, conference participants at the 11th McGill International Entrepreneurship Conference 2008 in Dunedin and two anonymous referees for feedback on the paper.
Author information
Authors and Affiliations
Corresponding author
Additional information
Access to the data used in this study was provided by Statistics New Zealand in accordance with security and confidentiality provisions of the Statistics Act 1975 and the Tax Administration Act 1994. The results in this paper have been confidentialised to protect individual businesses from identification. See an earlier version of this paper, Fabling and Sanderson (2009), for the full disclaimer. The authors are solely responsible for the views expressed.
Appendix: Allocating trade data to manufacturers
Appendix: Allocating trade data to manufacturers
Trade data are linked to the LBD using exact matching on tax identifiers and, in the absence of those, probabilistic matching on business names and addresses. Within group structures (i.e., independent firms with parent-subsidiary ownership relationships), this latter method of linking potentially causes problems since firms can have similar names and/or addresses.Footnote 23 From an economic perspective, enterprise groups could also be organised in a vertically-integrated manner so that trade is actually conducted by wholesale or retail units further up the production chain. In both cases, we wish to identify exports with the manufacturing unit that produced the goods being traded.
These potential issues are further complicated by group restructuring, which can lead to an undesirable “shuffling” of the trade data between group members as reporting lines, Customs details, and/or enterprise identifiers change. Left unaddressed, this issue results in a substantial proportion of aggregate goods trade being allocated to sectors for which no such large-scale trade should exist (primarily in business and financial services), and overplays the role of entering firms in trade growth (see Fabling and Sanderson 2008).
To allocate trade back to the production unit we take the following steps. First, we use the parent-subsidiary relationships present in the LBD to group together all economically active firms that are linked in a year.Footnote 24 We then check whether there is a manufacturer in the group in that year and if there is merchandise trade allocated to a non-manufacturer within that group. If either condition fails to hold then we have no problem as we can credibly assume that any trade has been correctly linked. This leaves us with enterprise groups that have non-manufacturers appearing to be goods exporters and (potentially multiple) manufacturers to allocate trade to. If there is a single manufacturing firm in the group then we allocate all trade to that manufacturer in that year. If we have multiple candidate manufacturers then we choose to first allocate trade to firms that can be clearly identified as traders (because other Customs data are already allocated to them or because their Goods and Service Tax filing indicates they are an exporter). If there are still multiple candidates then we cannot determine which unit produces the exported good, so we merge those manufacturers together, treating them as a single unit across all time, and allocate the trade to that merged unit.
Remaining high value apparent entrants are manually checked to ensure they should not be part of broader group structures and a small number of additional parent-subsidiary relationships are added (typically because an existing group link is present only in a subset of the years in which the firm is observed in the Customs data).
The entire process results in a mere 0.5% of firm-year observations being “merged manufacturers.” In contrast 67% of aggregate trade is in merged manufacturers and 29% of trade has been reallocated from non-manufacturers to manufacturers. From manual investigation of the data it is clear that these numbers primarily reflect the vertically integrated organisation of large New Zealand trading conglomerates.
Rights and permissions
About this article
Cite this article
Fabling, R., Sanderson, L. Entrepreneurship and aggregate merchandise trade growth in New Zealand. J Int Entrep 8, 182–199 (2010). https://doi.org/10.1007/s10843-010-0063-9
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10843-010-0063-9