Is exchange rate pass-through in pork meat export prices constrained by the supply of live hogs?

(2) For simplicity, it was assumed that all hogs are either sold through contracts or through the spot market. In reality, there is always the possibility that hog producers sell their production through both channels as exemplified in Quebec. This case, however, clutters the analytical model and has no bearing on the empirical model presented in the next section. Moreover, it is implicitly assumed that it is not profitable for processing firms to import additional hogs at a price [e.sup.a] [r.sup.a] + [mu][t.sup.a].

(3) The analysis purposely ignores currency hedging and invoicing strategies because there is no risk in the model when pricing decisions are made. Hedging can be used to significantly reduce the risk exposure of exporting firms when making capacity commitments in the first stage of the game (Donnenfeld and Haug 2003). The available data and the empirical model below are not suited to address this issue.

(4) Gervais and Khraief (2007) detail the unit root testing procedures.

(5) Although the theoretical model involves two different stages with regards to overall pricing strategies of processing firms, there is no explicit dynamics in the model. Dynamics could be investigated by introducing menu costs in revising export prices (Floden and Wilander, 2006) or through the existence of consumers' switching costs a la Klemperer (1995). While the short-run dynamics emerging from these models can be particularly interesting, the current application focuses on the long-run relationship between export prices and exchange rates.

(6) The residuals were pre-whitened using a VAR filter with the order of the VAR selected by the SCB criterion. The truncation parameter of the Bartlett kernel (denoted l) was selected such that l = 4[(T/100).sup.2/9] [nearly equal to] 5.

(7) The results of the pre-cointegration tests are reported in Gervais and Khraief (2007).

(8) A first-order serial correlation coefficient was estimated after correcting the serial correlation in the residuals of the cointegrating vectors to make sure that autocorrelation was completely removed. A test for autoregressive conditional heteroscedasticity (ARCH) of order one did not reject the null hypothesis that the residuals have a constant variance.

(9) A bootstrap procedure was used to investigate whether the empirical distribution of the residuals had any impacts on the qualitative nature of the results. The details of the procedure and results are presented in Gervais and Khraief (2007). In the case of exports to the United States, the bootstrap results are in accordance with the asymptotic theory.

(10) As in the case of the U.S. model, diagnostic checking confirmed that there was no remaining autocorrelation in the residuals while the null hypothesis of no ARCH(1) effect in the residuals could not be rejected by the data for all three equations.

(11) The bootstrap procedure confirms the results of the asymptotic inference in the case of Quebec and Ontario exports to Japan. For Manitoba exports, the bootstrap p-value associated with the capacity coefficient is 0.135 and thus suggests that capacity does not have an impact on export prices. The details of the bootstrap procedure are available in Gervais and Khraief (2007).

(12) The coefficient in Ontario suggests that a depreciation leads to a decrease (increase) in the export price and this behavior is consistent with a number of explanations when there are no predetermined supplies and bottlenecks in upstream markets (e.g., Froot and Klemperer 1989; Hens et al. 1999). It is, however, unclear how market share objectives and various forms of imperfect competition between exporting firms would impact ERPT behavior in the current framework.

(13) From a statistical point of view, the choice of a HAC variance-covariance estimator has important implications on inference. For example, the fully parametric estimator proposed by den Haan and Levin (1997) raises the p-values of the joint null hypothesis that output and the other destination exchange rate are zero in the Manitoba case.

Jean-Philippe Gervais is Canada Research Chair in Agri-Industries and International Trade, Center for Research in the Economics of Agri-Food (CREA), Laval University. Naceur Khraief is graduate research assistant, CREA, Laval University.

Authors wish to thank three anonymous reviewers for providing helpful suggestions as well as seminar participants at North Carolina State University and at the 2005 meetings of the Canadian Agricultural Economics Society. Of course, the standard disclaimer about remaining errors applies. Table 1. Tests of Homogeneity Restrictions for the U.S. and Japan Export Pricing Equations

Wald Null Hypotheses for the U.S. System Test p-value (1) [[beta].sup.QC, US] = [[beta].sup.MB, US] = 5.33 0.070 [[beta].sup.ON, US] (2) [[beta].sup.QC, JP] = [[beta].sup.MB, JP] = 1,178 0.003 [[beta].sup.ON, JP] (3) [[lambda].sup.QC] = [[lambda].sup.MB] = 37.17 0.000 [[lambda].sup.ON] (4) [[phi].sup.QC] = [[phi].sup.MB] = 11.85 0.003 [[phi].sup.ON] (5) [[beta].sup.QC, US] = [[beta].sup.MB, US] = 23,34 0.003 [[beta].sup.ON, US] = [[beta].sup.QC, JP] = [[beta].sup.MB, JP] = [[beta].sup.ON, JP] [[lambda].sup.QC] = [[lambda].sup.MB] = [[lambda].sup.ON]; [[phi].sup.QC] = [[phi].sup.MB] = [[phi].sup.ON] Null Hypotheses for the Japan System (1) [[beta].sup.QC, US] = [[beta].sup.MB, US] = 105.26 0.000 [[beta].sup.ON, US] (2) [[beta].sup.QC, JP] = [[beta].sup.MB, JP] = 54.31 0.000 [[beta].sup.O N, JP] (3) [[lambda].sup.QC] = [[lambda].sup.MB] = 67.81 0.000 [[lambda].sup.ON] (4) [[phi].sup.QC] = [[phi].sup.MB] = 0.421 [[phi].sup.ON] (5) [[beta].sup.QC, US] = [[beta].sup.MB, US] = 0.000 [[beta].sup.ON, US] = [[beta].sup.QC, JP] = [[beta].sup.MB, JP] = [[beta].sup.ON, JP] [[lambda].sup.QC] = [[lambda].sup.MB] = [[lambda].sup.ON]; [[phi].sup.QC] = [[phi].sup.MB] = [[phi].sup.ON] Note: The Wald statistic associated with the first four hypotheses follows a chi-squared distribution with two degrees of freedom. The Wald statistic associated with the fifth set of hypotheses follows a chi-squared distribution with eight degrees of freedom. Table 2. ERPT Equations in the U.S. Market and Hypothesis Testing

Quebec Variables Coefficient t-Statistic Constant 1.644 10.40 U.S. exchange rate, -0.657 -7.31

[[beta].sup.j, US] Japan exchange rate, -0.272 -4.81

[[beta].sup.j, JP] Capacity, [[lambda].sup.j] -0.281 -5.71 Hog price, [[phi].sup.j] 0.493 10.89 Adjusted [R.sup.2] 0.89 Jarque-Bera normality test 8.48 Significance of predetermined Wald test p-value

hog supplies [H.sub.0] : [[beta].sup.j, JP] = 27.70 0.000

[[lambda].sup.j] = 0

Manitoba Variables Coefficient t-Statistic Constant -0.43 -3.05 U.S. exchange rate, -0.471 -8.26

[[beta].sup.j, US] Japan exchange rate, 0.129 1.73

[[beta].sup.j, JP] Capacity, [[lambda].sup.j] 0.018 0.42 Hog price, [[phi].sup.j] 0.709 14.37 Adjusted [R.sup.2] 0.73 Jarque-Bera normality test 168.71 Significance of predetermined Wald test p-value

hog supplies [H.sub.0] : [[beta].sup.j, JP] = 2.35 0.308

[[lambda].sup.j] = 0

Ontario Variables Coefficient t-Statistic Constant 2.242 13.93 U.S. exchange rate, -0.176 -2.58

[[beta].sup.j, US] Japan exchange rate, -0.481 -4.49

[[beta].sup.j, JP] Capacity, [[lambda].sup.j] -0.474 -11.39 Hog price, [[phi].sup.j] 0.482 12.37 Adjusted [R.sup.2] 0.88 Jarque-Bera normality test 0.98 Significance of predetermined Wald test p-value

hog supplies [H.sub.0] : [[beta].sup.j, JP] = 74.11 0.000

[[lambda].sup.j] = 0 Note: The adjusted [R.sup.2] is computed from the first-step unrestricted model estimated with DOLS. The Jarque-Bera statistic is computed using the pre-whitened residuals obtained from the DOLS procedure. The test statistic follows a chi-squared distribution with two degrees of freedom under the null hypothesis that the residuals are normally distributed. The Wald statistics follow a chi-squared distribution with two degrees of freedom. Table 3. ERPT Equations for the Japanese Market and Hypothesis Testing

Quebec Variable Coefficient t-Statistic Constant 3.560 13.18 U.S. exchange rate, 0.170 1.07

[[beta].sup.j, US] Japan exchange rate, -0.821 -11.83