3  Characteristics of Multi-Product Firms

The extent and the variability provided by the data allows us to test the predictions of the models in Bernard et al. (2006b) and Nocke and Yeaple (2006) about the characteristics of multi-product firms and firms' extensive margins. We find that firms exporting multiple products, industries and sectors dominate New Zealand exporters. This dominance is shown in Table 2, which shows the number of firms that export a single product, which is 6663 and their share of all exporting firms, which is 28 percent. The number of firms producing more than one good is 17412, which refers to 72 percent of all exporting firms. Multi product firms also account for 99 percent of total value of exports. The last column of Table 2 shows that multi-product exporters export on average 22 products, compared to 3.5 (3) produced by U.S. (Indian) manufacturing firms. One potential reason the mean number of products per firm is so much higher in New Zealand is that the data used in this study is much more disaggregated (HS10 codes) than the other two studies (SIC5).

Note: Breakdown of firms according to whether they produce multiple products or in multiple sectors or industries. Pooled sample. Numbers rounded for confidentiality reasons.

Firms which export in multiple industries make up 61 percent of all exporters, whereas the percentage for multiple sectors is 56 percent. In comparison, a similar study that analyses US manufacturing firms finds that single-product firms are 59 percent of all firms and multiple industry firms are 29 percent and multiple sectors are 13 percent. This suggests that New Zealand multi-product firms are more diverse than U.S. firms. This higher degree of diversification could be a response to lack of well-functioning capital markets.

One potential cause is that New Zealand firms try to diversify by exporting more than one product as a way of hedging against uncertainty in export markets. In this case, one would expect the percentage of multi-product exporters to increase when the exchange rate is more volatile. Another reason could be that due to the small size of New Zealand exporters in world markets, firms are forced to export a variety of products. In this case, the percentage of multi-product firms should not vary as the market share of New Zealand exporters does not change significantly over time.

In summary, multi-product exporters dominate exports in New Zealand and they exhibit considerable differences from single-product ones.

3.1 Multi-product firms are stronger performers than single-product firms

Theory predicts that due to the presence of fixed costs “better” firms in terms of productivity are expected to self-select into being multi-product exporters. Firms with higher productivity are able to cover the fixed costs of a larger number of products due to their higher revenues per product. Table 3 compares the characteristics of firms that export multiple products (industries and sectors) with those exporting only single ones.

Note: Table shows results from OLS regressions of log of firm characteristics on a dummy indicating firm status. Product 2 is equal to 1 if the firm exports in multiple sectors, Product 4 is equal to 1 if the firm exports in multiple industries and Product 10 is equal to 1 if the firm exports multiple products. All regressions include industry fixed effects and standard errors clustered at the industry level. Number of observations rounded to nearest five for confidentiality reasons. ** represents significance at 1%.

The logs of firm characteristics in 2005 are regressed on a dummy variable indicating the firm's status as a single or a multiple product/industry/sector exporter. The dummies are constructed as follows: Product10 is equal to 1 if a firm exports more than one product at the ten digit code and 0 if only a single product is exported. Likewise, Product4 is equal to 1 if a firm exports products in more than one industry, and Product2 is equal to 1 if a firm exports products in more than one sector. The firm characteristics considered are sales, employment, wages, value added (defined as the difference of their sales and purchases), productivity (defined as the ratio of value added to employment) and value of exports. These regressions use industry-fixed effects, and the standard errors are clustered at the industry level as well.^[11]

Multi-product firms on average have larger sales, higher employment, wages, value of exports and value added than single product ones. The coefficient on the multi-product dummy in column 1 shows that multi-product firms have on average 17% (e^2.94-1) higher value of sales than single product ones.

According to Bernard et al. (2006b), multi-product firms are more productive since they are able to cover the fixed costs of producing a larger range of products. On the other hand, Nocke and Yeaple (2006) assume that marginal costs for each product are increasing in the total number of products, implying that in equilibrium, multi-product firms have lower productivity for their marginal products and lower overall productivity than single product ones, although they are ex-ante better. At the industry and product level, productivity for multi-product exporters is statistically significant and higher than single product exporters. The coefficient for productivity is negative at the sector level although it is insignificant. This is the same result as for the US, suggesting that measuring productivity in firms with products that span several sectors is difficult.

An interesting point to note is that the coefficients are larger than the results reported for U.S. firms, suggesting the differences between multi- and single-product exporters in New Zealand are larger than those in the US.

Another way of comparing multi-product and single-product firms is to analyse i) the prior performance of single product firms that switch to multi-product firms to those that do not switch and ii) the performance of multi-product firms that add products to those that do not add. In order to do so, firms' log sales and log value of exports in 2000 are regressed against a dummy that equals 1 if the firm added a product between 2001 and 2006. Initial industry fixed effects are included and the standard errors are clustered at the industry level. The results in Table 4 show that there is indeed selection among exporters. Firms that added products were stronger performers even before they added products. The difference is larger than that for Indian firms.

Note: Regression of log sales and exports in 2000 on a dummy Add which is equal to 1 if the firm added at least one product between 2000 and 2006. Initial industry fixed effects included. Standard errors are clustered at the industry level. Columns 2 and 4 include only multi-product firms.