Pricing to market of Italian exporting firms

Abstract

This article investigates the pricing-to-market (PTM) behaviour of Italian exporting firms, using quarterly survey data by sector and by region over the period 1999q1 to 2005q2. A partial equilibrium imperfect competition model provides the structure according to which the orthogonality of structural shocks is derived. Impulse response analysis shows non-negligible reactions of export-domestic price margins to unanticipated changes in cost competitiveness and in foreign and domestic demand levels, even though these effects appear to be of a transitory nature. For the period 1999 to 2001, a typical PTM behaviour emerges, while, during the most recent years favourable foreign demand conditions allowed firms to increase their export-domestic price margins in face of a strong deterioration of their cost competitiveness. Macroeconomic implications of the observed PTM behaviour are also discussed.

Notes

¹ For a comprehensive survey, see Goldberg and Knetter (1997).

² Currency denomination of export prices, in the presence of nominal rigidities, is another potential source of observed PTM and of violations of the LOP. If firms set in advance (staggering) export prices in foreign currency, differentiation of selling prices to different markets may be the result of exchange rate surprises, rather than the deliberate choice of an optimizing behaviour (Giovannini, 1988). Devereux (1997) provides an alternative explanation based on the assumption of price stickiness in the local currency of the (foreign) buyer.

³ For instance, it is possible for a firm to set prices differently according to the different cyclical demand conditions in the destination markets if the firm has market power and products cannot be arbitraged across borders.

⁴ Recent articles exploring the pricing behaviour of Italian firms are those by Fabiani et al. (2004), de Nardis and Pensa (2004) and Bugamelli and Tedeschi (2005). While the former focuses on the price setting of firms producing mostly for the domestic market, making use of the results of a Bank of Italy survey, the other two works analyse the pricing behaviour of Italian exporters: de Nardis and Pensa aim at detecting the degree of market power of Italian producers of traditional goods in foreign markets and estimate residual demand elasticities on the grounds of the approach of Goldberg and Knetter (1999); Bugamelli and Tedeschi focus on the pricing strategies of Italian firms (across markets and sectors) during the 1990s.

⁵ A former attempt to study export pricing policies of Italian firms using survey information is in Pupillo and Zimmermann (1991).

⁶ It might derive from monopolistic competition and product differentiation or from any other market imperfection.

⁷ If both demand curves in the two markets have constant price elasticities – as in a log-linear demand schedule – then the mark-ups in each destination market are constant; in this case prices practised in H and F are affected by a common marginal cost and the ratio between the price made in F and the price made in H (expressed in the same currency) is invariant to currency movements. This means that there may be a constant wedge between the two prices, but PTM does not change in response to variations of the exchange rate. Whether ERPT is complete or not depends, in this case, on the behaviour of the marginal cost: if the latter is increasing in output (total sales), ERPT is partial (i.e. an appreciation of the domestic currency, negatively affecting total sales, gives rise to a decrease of the marginal cost and, hence, to a reduction of both the domestic and foreign prices expressed in the home currency). As a consequence, the export price expressed in the foreign currency increases less than proportionally to the exchange rate change (Appendix).

⁸ In the model by Caves and Jones (1977), the phenomenon of ‘dumping’ in international trade arises when a monopolistic profit maximizing firm, facing a higher elasticity of demand abroad than at home, is able to discriminate between foreign and domestic markets. As a consequence, it will charge a lower price abroad than at home. Conversely, Brander and Krugman (1983) show how dumping may arise for ‘systematic’ reasons associated with oligopolistic behaviour rather than ‘accidental’ differences in country demands.

⁹ In the empirical specification, we hence follow Marston (1990) who links (permanent) changes in the export price setting to movements of the real (rather than nominal) exchange rate; in our formulation is the (logarithm of the) numerator of the real exchange rate specific to sector i.

¹⁰ In the empirical investigation, firms belonging to the South of Italy (Sicilia, Sardegna, Calabria, Puglia, Basilicata, Molise, Campania, Abruzzo) are not included because of the scant relevance of several sectors in that region.

¹¹ Consider, for instance, Italian footwears. Comparing domestic prices to those charged abroad by the same ‘shoe-firm’ makes undoubtedly more sense than comparing the prices of footwear produced by different firms at home and abroad.

¹² Even though the theoretical model does not explicitly involve non-price obstacles to export, we extend our dataset in order to control for possible effects of these forms of trade costs on the export-domestic price margin.

¹³ The 95% confidence interval of a standard t-test ranges from 3.26 to 5.46, over the full sample, while it goes from 1.63 to 4.97 (from 3.81 to 6.71) over the first (second) sub-sample.

¹⁴ This assumption also holds when measurement errors do not exist at all and implies that and otherwise.

¹⁵ The inclusion of noncompetitiveness factors (or trade costs) in system (9) is modelled by treating that variable as the most exogenous, in order to preserve the causal linkage implied by relations (5)–(8).

¹⁶ In this special case, the number of regressors equals the number of instruments (just-identification of the model); therefore, GMM results are numerically equivalent to those from equation-by-equation 2SLS.

¹⁷ In all models, lagged coefficients, of order >1, turn out to be not statistically significant at the usual significance levels.

¹⁸ The decision to use 12-quarter sub-periods was the result of a compromise between maximizing the number of regressions and maintaining a reasonable sample size for each regression.

¹⁹ Half-life is defined as the number of quarters, which have to pass before the deviation from the steady-state falls to half the size of the initial shock.

²⁰ Lagged reactions to price competitiveness shocks (Figs 3, 4, 5 and 6) might be considered as consistent with models based on sticky prices adjustment due to menu-cost-driven pricing behaviour or other nominal rigidities (see, among others, Ghosh and Wolf, 1994), although in the study conducted in a different context (pricing behaviour of Italian firms in the domestic market), Fabiani et al. (2004) find the price inertia in manufacturing relates mainly to the existence of nominal contracts and to coordination failures.

²¹ Results from the two models (not reported for sake of brevity) indicate that a deterioration of price competitiveness translates into a decrease (increase) of the export-domestic price margin in the first (second) sub-sample, while the time profile of the responses relative to other shocks are consistent with our economic priors. Given that the two samples are independent, IRFs of the differences are equivalent to the difference of IRFs. The same argument holds for the confidence bounds.