2.1 Variable mark-ups - pricing-to-market and strategic interaction

One widely cited explanation for incomplete pass-through is the Krugman (1987) (and Dornbusch 1987) pricing-to-market (PTM) model of firms' price-setting behaviour in relation to changes in exchange rates. In a monopolistically competitive environment, firms adjust their mark-up depending on the elasticity of demand for their good in the destination market. For example, if firms are reluctant to lose market share, they will lower their mark-up when the exporter's exchange rate appreciates against the importing country.[2] In turn, empirical findings of PTM have been used to provide support for models of imperfect competition and market segmentation (Goldberg & Knetter 1997).

Knetter (1989) proposes a reduced form specification for estimating ERPT, distinguishing between changes in marginal costs and variable mark-ups by exploiting data on the shipments of goods to multiple destinations. If firms use imported goods that are affected by exchange rate movements as inputs, and inputs represent a constant cost increase or decrease across all destinations for a product, the component of a price change due to marginal cost will be the same across destinations, whereas mark-ups are destination-specific. Knetter (1993) applies this methodology to product-level export data finding that, in aggregate, exporters from the UK, Japan and Germany offset between 36 and 48 percent of exchange rate movements by adjusting their mark-ups, while US exporters pass the exchange rate change through to customers. However, Knetter (1993) notes that patterns of PTM are quite similar across source countries for a given industry, and thus that overall differences in PTM may be related to industry composition, rather than country-specific variation in pricing behaviour.

Fitzgerald & Haller (2012) apply a similar approach using microdata from Irish export firms trading the same products domestically and in the UK, with some limited analysis of the role of invoice currency in observed pass-through. They find that when goods are invoiced in local currency (GBP), the relative mark-up across the two markets moves one-for-one with exchange rate fluctuations - that is, exporting firms absorb the full extent of exchange rate changes. In contrast, there is no evidence for mark-ups on goods invoiced in the producer currency (Irish pounds or Euros) being influenced by exchange rate changes. However, the structure of their data prevents robust analysis of producer currency-invoiced trades, as destination data are available only for a cross-section of observations, not the full panel.

Berman et al. (2012) explore the issue of heterogeneous PTM associated with differences in firm performance. They argue that more productive firms are likely to face lower elasticity of demand, leading them to react to exchange rate depreciations by increasing their mark-ups while lower productivity exporters instead pass exchange rate savings through to customers and increase the volume of their exports. Berman et al. (2012) discuss three alternative mechanisms through which this relationship may arise. In a Melitz & Ottaviano (2008) model with linear demand and horizontal product differentiation, the price elasticity of demand increases with the price faced by consumers. As high-productivity firms charge lower prices, these firms face a lower demand elasticity. A real depreciation leads to a fall in the prices faced by consumers, and exporters react by increasing their mark-up.

Atkeson & Burstein (2008) suggest an alternative model in which firms face Cournot competition with nested constant elasticity of substitution across several sectors. If the elasticity of substitution is lower across sectors than within sectors, then the elasticity of demand faced will depend on firms' market share, which is in turn determined by their productivity. In the extreme, a low productivity firm with a market share approaching zero faces a high elasticity of substitution within its own sector, while a high-productivity firm with a market share approaching one faces the lower cross-sectoral elasticity of substitution.

Finally, Berman et al. (2012) develop an extension to the Corsetti & Dedola (2005) model of distribution costs incurred in the local currency. If firms face a per-unit distribution cost payable in the importer currency, a depreciation implies that the production cost accounts for a lower proportion of the consumer price relative to the distribution cost. This reduces the elasticity perceived by the exporter in relation to the export price. High performance exporters again increase their export price more than others. Using detailed data on destination-specific export value and volume for French exporters, Berman et al. (2012) find evidence of heterogeneous PTM, including support for the hypothesis of local currency-denominated distribution costs. Specifically, they use Goldberg & Campa's (2010) estimates of distribution cost by sector and destination interacted with the real exchange rate to show that high distribution costs appear to increase the price elasticity, and decrease the volume elasticity, of exports to exchange rate changes.

In this paper we build on Berman et al.'s (2012) empirical findings, firstly by confirming that heterogeneous PTM is observed among New Zealand exporters, and then by relating this finding to observed invoicing patterns.[3]