1  Introduction

Empirical micro studies have found heterogeneity in firm performance across industries.^[1] Both theoretical and empirical literature has shown that a reallocation of resources from low to high productivity firms within industries has led to gains in aggregate output.^[2] Such a reallocation generally occurs through policy reforms (mostly trade) or changes in market fundamentals.^[3] In these studies, the discussions of allocation of resources to their most efficient use generally focus on dynamics of firm entry and exit, i.e., whether newly created firms are more productive than exiting ones. Some exceptions are Bernard et al. (2006a, 2008), Nocke and Yeaple (2006), Baldwin and Gu (2006) and Goldberg et al. (2008). Another way of reallocating resources can occur within firms, as firms add and drop products or alter their product mix. According to Bernard et al. (2006a), changes in product mix account for a larger proportion of output growth than firm entry and exit in the U.S. Product churning accounts for a third of the increase in U.S. output between 1972 and 1997. Goldberg et al. (2008) find that in India, changes in product mix contribute about 25% to the aggregate growth in output.

The prototype Longitudinal Business Database (LBD), which is discussed in detail in Section 2, enables us to approach this issue empirically for New Zealand. The existence of comparable studies for a developed economy such as the U.S. (Bernard et al. (2008)) and a developing economy such as India (Goldberg et al. (2008)) helps us put the results for New Zealand in perspective. Compared with these two countries, New Zealand is a smaller but more open economy. For New Zealand, we observe the full set of products each firm exports in each month and analyse how different firms' product mix changes over the period from 1996 to 2007.

We begin by documenting the characteristics of single and multiple product exporters. New Zealand multi-product exporters are similar to U.S. and Indian firms. Overall, multi-product firms perform better, on average, they have larger sales and value added and higher employment and wages.

An analysis of product mix of exporters shows that changes to the product mix are frequent, widespread and affect both aggregate and firm outcomes. The results show that changes in product mix contributed, on net, for 97% of the increase in the value of exports in New Zealand. Looking at the gross changes show that 94% of firms change their product mix.^[4] Decomposing the change in the product mix shows that New Zealand exporters are more likely to drop than add products, which is in contrast to results from Indian firms that are much more likely to add products. These findings suggest that “creative destruction” along the product dimension is prevalent among New Zealand exporters. This might be a reflection that there are no regulatory barriers to such actions by firms. Alternatively, lower trade costs lead to rationalization of extensive margins (production due to existing products) by firms through dropping products with higher marginal costs of production. Theories emphasising the role of “creative destruction” in growth predict that product-dropping behaviour by firms is a part of adjustment to changes in the economy. Furthermore, a majority of firms (79%) simultaneously add and drop products, suggesting that New Zealand exporters are quite dynamic and react to changes in economic conditions.

Recent theoretical models of international trade predict that firms drop products as an adjustment to decline in trade costs. Bernard et al. (2006b) extend the heterogeneous firm model of Melitz (2003) from self-selection of more productive firms into production to selection of products within firms. Their model posits a positive correlation between a firms' extensive (number of products) and intensive margin (output per product), which is observed in the data for New Zealand.

In the analysis of firm behaviour, diversification is considered good as it creates a buffer against different types of shocks. Most studies on diversification focus on the financial portfolio of the firm and its effect on productivity or the value of the firm. For example, Ghironi and Melitz (2005) look at firm entry and exit to explain the U.S. business cycle. Product diversification has largely been ignored due to lack of data. The choice of product mix can be thought as a channel through which price volatility feeds into productivity. On the one hand, if firms choose to have a more varied product mix due to volatility of prices, this leads to a diversification in their production, and increases their resilience to price shocks. On the other hand, such a diversification may lead to a decline in productivity as firms are unable to capture the benefits of comparative advantage and economies of scale. This creates a typical risk vs. return trade-off that can be analysed using firm level data.

This paper analyses the product mix of New Zealand, following closely the work by Bernard et al. (2006a, 2008) for US manufacturing firms. Section 2 describes the data, and Section 3 compares the characteristics of single-product and multi-product exporters. Section 4 extends the analysis to the relative share of new, continuing and exiting firms as well as dropped, added and continued products in the value of exports. Patterns of changes to the product mix as well as impact of product mix changes on firm characteristics are also considered.[5] Section 5 discusses why firms might be engaging in product-switching behaviour. Section 6 concludes.