FS IV 91 - 12
discussion papers
International Comparisons of Pricing to Market Behavior
Michael M. Knetter Dartmouth College
March 1991
ISSN Nr. 0722 - 6748
Forschungsschwerpunkt Marktprozeß und Unter
nehmensentwicklung (II M V ) Research Unit
Market Processes and
Corporate Development ( U M )
W i s s e n s c h a f t s z e n t r u m B e r l i n f ü r S o z i a l f o r s c h u n g
ABSTRACT
International Comparisons of Pricing to Market Behavior
One of the more puzzling phenomenon during the past decade has been the manner in which firms respond to exchange rate
fluctuations. Anecdotal evidence indicates that a wide variety of pricing responses have been implemented, depending both upon the industry as well as the country. The purpose of this paper is to examine the manner by which firms adjust prices in response to exchange rate fluctuation. To compare export price adjustment across a wide variety of industries and nations, a new data base consisting of very precise products for Germany, the United
States, the United Kingdom, and Japan is introduced. The major finding in the paper is that German and Japanese producers tend to engage in what is known as "pricing to market", whereby the firm reduces the price markup to purchasers whose currencies have depreciated. This results in a relatively stable price in the
purchaser's currency. By contrast, firms in the United Kingdom and the United States are found to typically engage in a policy of a constant markup. This results in a greater fluctuation in prices in response to exchange rate variability.
ZUSAMMENFASSUNG
Internationaler Vergleich marktbezogener Preisfestsetzung
Die Preisfestsetzung von Unternehmen als Reaktion auf Wechsel
kursänderungen war eines der interessantesten Phänomene des ver
gangenen Jahrzehnts. Ad-hoc-Befunde deuten auf eine breite Palette von Preissetzungsstrategien hin, die vom jeweiligen Wirtschafts
zweig und Land abhängen. In diesem Beitrag wird die Art und Weise untersucht, in der Unternehmen als Reaktion auf Wechselkursände
rungen ihre Preise anpassen. Um die Exportpreisanpassungen für eine Reihe von Wirtschaftszweigen und Ländern zu analysieren, wurde eine neue Datenbank aufgebaut, die Informationen zu speziel
len Produkten aus der Bundesrepublik Deutschland, Großbritannien, Japan und den Vereinigten Staaten enthält. Das wichtigste Unter
suchungsergebnis besagt, daß bundesdeutsche und japanische Her
steller dazu neigen, "marktbezogene" Preisfestsetzung zu betrei
ben, d.h. sie verringern den Gewinnzuschlagssatz im Rahmen der Preiskalkulation für Käufer, deren Währungen abgewertet wurden.
Dies führt zu relativ stabilen Preisen in der Währung der Käufer
länder. Im Unterschied dazu praktizieren britische und US-amerika
nische Unternehmen eher eine Preispolitik mit einem festen Gewinn
zuschlagssatz. Als Reaktion auf Wechselkursänderungen führt dies dann zu stärkeren Preisschwankungen.
International Comparisons of Pricing to Market Behavior
Michael M. Knetter*Dartmouth College
The optimal response of a firm ’s export price to changes in currency values depends on a variety of factors. These factors operate via two channels: (1) through the impact exchange rates have on marginal cost and (2) through the impact exchange rates have on markups o f price over marginal cost. Destination-specific adjustment of markups in response to exchange rate changes have been referred to in the literature as “pricing to market” (henceforth PTM).* 1 The typical character of PTM is that sellers reduce markups to buyers whose currencies have depreciated against the seller, thereby stabilizing prices in the buyer’s currency relative to a constant markup policy. I will refer to this specific variety of PTM as local currency price stabilization (LCPS).
Theoretically, PTM can arise for many reasons. For a monopolist that price discriminates across export destinations, PTM is a function of the convexity of demand schedules.2 Demand schedules less convex than a constant elasticity schedule imply LCPS, whereas those more convex than a constant elasticity schedule will lead to a perverse relationship-markups increasing as the buyer’s currency depreciates. Rudiger Dornbusch (1987) has examined the impact of market structure on traded goods prices. In general, it appears that the existence of competitors in any market will impose more
discipline on firms in their pricing behavior. In other words, for a given form of the market demand schedule, adding competitors will increase the incentive for LCPS. In dynamic models PTM can arise as a temporary phenomenon due to adjustment costs.3 For
* The author would like to acknowledge financial support from the Lynde and Harry Bradley Foundation and the German Marshall Fund. Part of this work was completed while the author was a Visiting Researcher at Wissenschaftszentrum Berlin.
1 See Paul Krugman (1987).
2 See Robert Fccnstra (1989), Michael Knetter (1990) or Richard Marston (1990).
3 Kenneth Kasa (1990) formulates and estimates a dynamic model in which adjustment costs can generate short run pricing to market behavior.
example, if it is costly for firms to reallocate output across markets, due perhaps to the existence of bottlenecks in marketing or distribution, or to adjust the level of total output, LCPS will be a natural consequence of exchange rate movements.
PTM has been documented by casual and rigorous empiricism in a number of recent studies employing a variety of data sets.4 Based on the movement in 4-digit industry U.S.
import prices relative to a trade-weighted average of foreign production costs, Mann (1986) concluded that foreign profit margins are adjusted to mitigate the impact of exchange rate changes on dollar prices of U.S. imports. Somewhat surprisingly, U.S. exporters showed no tendency to adjust markups in response to exchange rate changes.5 Knetter’s (1989) study of export pricing in U.S. and German 7-digit industries documents strong evidence of LCPS by German exporters to a variety of destinations. Once again, there is no
evidence of LCPS by U.S. exporters. Marston (1990) finds impressive evidence of PTM in a wide range of 4-digit Japanese industries that export primarily to the U.S. Gagnon and Knetter (1990) estimate Japanese auto exporters offset approximately 70% of the effect of exchange rate changes on buyer’s prices through markup adjustment. The comparable number for German auto exports varies by engine size: for small autos, about 40% of the effect o f exchange rate changes are offset by destination-specific markup changes, whereas for large autos adjustment is minimal. There is no evidence o f PTM for U.S. auto exports.
This source country pattern is not evident in their analysis of price adjustment in total merchandise trade.
Taken as a whole, this growing literature supports several beliefs about PTM behavior. First, there is substantial industry-level evidence that Japanese and German exporters use markup adjustment to stabilize local currency prices of exports, although for Japan most of the evidence is based on pricing to the U.S. Second, PTM appears to be
4 See the examples of grey markets mentioned in Krugman (1987) or the empirical studies of Catherine Mann (1986), Joseph Gagnon and Knetter (1990), Kasa (1990), Knetter (1989,1990), and Marston (1990).
5 If anything, the data suggested U.S. export price adjusmtent amplified the effect of exchange rate changes on prices measured in the buyer’s currency.
non-existent for a number of U.S. export industries at various levels o f aggregation.
Third, the degree o f stabilization of local currency prices appears to vary quite widely by industry even within a given exporting country.
This literature lacks an empirical framework capable of nesting competing
explanations of PTM behavior. Gagnon and Knetter (1990) and Marston (1990) allow for both equilibrium price discrimination and disequilibrium dynamics in estimation, although the underlying models are not explicidy dynamic. Knetter (1989,1990) interprets PTM as arising from equilibrium price discrimination, whereas Kasa’s (1990) model emphasizes adjustment costs but ignores the possibility of equilibrium price discrimination.
Short of a framework capable of nesting alternative models of PTM, a
comprehensive study of PTM that examines similar industries across a range o f export source and destination countries is the next best alternative. The evaluation o f industry, source, and destination market patterns of PTM behavior would be of value in determining which explanations o f PTM are most important empirically.
Industry patterns would arise at least in part from differences in adjustment costs in production or distribution that are traceable to underlying differences in technology.
Market power, concentration and conduct also vary by industry and would lead one to expect different patterns of PTM.
Source country patterns may reflect a number of factors. For example, long term employment arrangements (fostered in part by government policy) in Japan and Germany may make separations and layoffs a very costly form of adjustment for firms in many industries. Thus, there may be little scope for quantity adjustments in response to
disturbances o f unknown duration. As a result, firms in these countries will adjust export prices to stabilize prices in the buyer’s currency more than their counterparts in countries in which labor adjustment is less costly. Differences in ± e degree o f investment in foreign production facilities will also be likely to influence PTM behavior. This factor will vary by industry and source country.
Finally, market size and trade policy are two factors likely to determine the overall competitiveness of destination markets. Large economies with few trade barriers are likely to attract more entrants in any given industry. Thus, LCPS may be more common in the U.S. than Japan because the U.S. is a larger and more open market.6
This paper attempts to bridge the gap in the existing literature by analyzing export price adjustment for a variety of U.S., U.K., German, and Japanese 7-digit industries.
There is some overlap of the 7-digit industries across source countries, although exact matches are rare due to the idiosyncracies of industry classifications used by authorities in collecting the export data in the four source countries. The main empirical result is that German and Japanese producers show more tendency to price to market than the U.K. and U.S. industries included in this study. However, for those industries in which exact matches are possible across source countries, behavior is remarkably similar.
Consequently, industry effects appear to be more important than source country effects in explaining the dispersion of PTM behavior across trading relationships. A second important result is that in most cases it is not possible to reject the null hypothesis that the degree of LCPS is identical across destination markets. While this does not imply all destinations are treated the same, it casts some doubt on the existence o f systematic differences across destinations.
The organization of this paper is as follows. The empirical framework is introduced in Section 1. The data used for this study are discussed in Section 2. The results o f estimation are presented and discussed in Section 3. Section 4 concludes the paper.
6 This is certainly a generalization that may or may not have empirical validity. The ranking of the U.S.
and Japan in terms of market size and openness will undoubtedly vary by industry.
1. The Empirical Model
The empirical framework adopted here follows that motivated and outlined in greater detail in Knetter (1989,1990). The motivation for the framework comes from a simple model o f price discrimination by a monopolist selling to several export destinations.
Price changes to any destination will consist of two components: (1) changes in marginal cost and (2) changes in the markup of price over marginal cost. The former effect will be common to all destinations, while the latter may have common and destination-specific components. Marginal costs and markups are assumed to be unobservable, but common movements in prices due to changes in marginal cost or common markup changes are accounted for by a full set of time dummies in the model.7 Destination-specific changes in markups will occur in response to changes that are unique to individual destination
markets.
The crucial destination-specific explanatory variable will be changes in the exchange rate between the exporter’s currency and the currency of the destination market. Other factors, such as changes in income in the destination market, may also play a role, although those effects would be of secondary importance due to the magnitude and variability of exchange rate changes across destinations relative to changes in income.
The general model of export price adjustment I propose to estimate for a 7-digit industry in a given source country can be written as follows:
(1)
7 The interpretation of the time effects as capturing the behavior of marginal cost obviously requires more assumptions when the export sector consists of more than one firm. Although the conditions for exact aggregation are unlikely to hold, the model still controls for underlying changes in industry cost provided all firms costs move together.
where i = 1,...,N and t = 1,...,T index destination of exports and time respectively, p is the log o f export price, x is the log of the destinations-specific exchange rate (expressed as units of the buyer’s currency per unit of the seller’s), and 0t, and ß i are parameters to be estimated. Note that one of the time or country effects must be dropped in estimation to avoid singularity. The error term, £it, is assumed to be independent and identically
distributed with mean zero and variance g
2
c In principle, there could be k seperate /3’s for each destination when the model includes k destination-specific variables.The model given by (1) is an analysis of covariance model in which the intercept term is allowed to vary due to effects that differ across individuals but are constant over time (the A’s) and effects that are constant across individuals but vary over time (the 0’s).
The time and country effects will be treated as fixed, thus inference is conditional on those effects included in the sample. The specific criteria used in selecting the sample of
observations (to be discussed in the next section) make clear the need for conditional inference in this model. As written in equation (1), the model allows for the slope
coefficients to vary across destinations. In estimation I will test whether these coefficients can be constrained across destinations.
The statistical interpretation of the ß ’s is straightforward. A value of zero implies that the markup to a particular destination is unresponsive to fluctuations in the value of the exporter’s currency against the buyer’s. Thus, changes in currency values are fully passed through to the buyer. Negative values of ß imply that markup adjustment is used to
accomplish local currency price stabilization. For example, a value of -.5 means that in response to a 10% appreciation (depreciation) of his currency, the exporter would reduce (increase) his markup by 5%. Assuming constant costs, the price paid in units of the buyer’s currency w ould rise (fall) by only 5%. Positive values of ß correspond to the perverse case in which destination-specific changes in markups amplify the effect of destination-specific exchange rate changes on the price in units of the buyer’s currency.
The economic interpretation o f the /J’s depends on what one assumes about market structure. If the exporter is a monopolist, the value of ß is determined by the convexity of the demand schedule in the destination market. The class of demand schedules having constant elasticity with respect to price imply a value of ß equal to zero, ß is negative provided demand is less convex than a constant elasticity demand schedule. When the export sector consists of multiple firms that compete with firms located in other countries (the typical case), the interpretation of ß is more complicated. Generally speaking, it would be the weighted average over exporting firms of the response of price to changes in
exchange rates. This would depend on the convexity o f residual demands faced by the exporters. Residual demands are in turn a function of market demand for the product in the destination market and the response o f competitors in that market to changes in price
initiated by the exporters. Thus, even if preferences were identical across destinations, market structure or conduct could differ causing price adjustment behavior. For given preferences, a variety of models of imperfect competition predict that more competitors will increase the tendency of exporters to stabilize local currency prices.
The economic interpretations of ß discussed above involve equilibrium responses in the absence of adjustment costs. If exporters face quadratic costs o f adjusting the flow of product to a destination market, then quantity adjustment may take place over a period of time. If quantities were literally fixed in the short run, then one would expect to observe complete LCPS-i.e., a ß o f minus one. As quantity adjustment occurs, local currency prices change and the value of ß would adjust toward zero. With annual frequency data, one would expect much of the adjustment to occur within one period of observation. Even so, part of the contemporaneous response of destination-specific prices to exchange rate changes may reflect adjustment costs. In estimation, lagged values of the change in
exchange rates will be introduced into the model in an attempt to distinguish the importance of adjustment costs. This issue will be taken up later in ± e paper.
It should be noted that all interpretations are based on the idea that exporters view changes in the exchange rate as permanent. The empirical literature on exchange rate determination has yet to offer models that outperform a simple random walk in forecasting future realizations.8 That finding supports the interpretation of exchange rate changes as permanent. It also provides justification for treating exchange rate movements as
exogenous to the firms.
2. Data
The data used in this study are based on the annual value and quantity of exports to selected destination countries for a number of 7-digit industries in four source countries:
U.S., U.K., Japan, and Germany. For each source country-industry pair, data on exports to a number of relatively large (in terms of sales) export destinations are collected. Eligible destination markets are those that have currencies that fluctuate in value against the
exporter’s currency, to the extent possible. The aim in choosing large export destinations is to improve the accuracy o f the unit values (the value o f exports divided by the quantity) as a measure of price and to minimize the number of missing observations. These criteria for data collection imply that sampling over destinations is not random. Therefore the effects in the model will be treated as fixed and inference will be conditional on the effects included in the sample. The specific industries selected and the data sources for the unit value data are listed in the data appendix.
The industries were selected with several factors in mind. One aim was to provide variation in terms the types of products: durables, non-durables, intermediate goods, etc.
Another was to try to choose some products that are important export industries in the source countries being studied. Most important, however, was the attempt to select 8 See Meese and R ogoffs (1983) original paper or the more recent survey by Meese (1990). In a dynamic model that attempts to measure the impact of temporary exchange rate changes on export pricing, Froot and Klemperer (1989) find little evidence to suggest that this is an important phenomenon.
industries that exported from more than one of the source countries in the sample. This task was difficult due to the lack of harmonization of the industry classification codes across source countries. The data set includes a number of chemical products for each source country, although exact matches are rare. Automobiles are exported by all four of the source countries, but the classification by engine size is different for each country.
The data are actually available at higher frequencies in some cases. In the U.S. and Germany, they are available monthly. The choice o f annual frequency reflects primarily the need to economize on data collection effort. All of the data used in the study were
handcopied from government publications of the respective source countries. This creates a trade-off between higher frequency information and more source country and industry variation. The latter seemed to be of greater interest and importance. Previous work (Kasa (1990) and Marston (1990)) has addressed the dynamics o f price responses at higher frequencies. That will not be the focus of this work. There is also a trade-off between frequency and the length of the sample period. The lower frequency information was collected over the entire floating exchange rate period in most cases. Another reason lower frequencies may actually be preferred in constructing unit values is that erratic variation in shipments at high frequencies could increase the amount of noise in the unit value series.
This is particularly likely in cases where there is heterogeneity in the product category.
The exchange rate series used as an explanatory variable is expressed in units of the buyer’s currency per unit of the exporter’s and is based on the annual average nominal exchange rate published in International Financial Statistics. The nominal rate is adjusted by dividing by the wholesale price index in the destination market. The rational for this adjustment is that the optimal export price should be neutral with respect to changes in the nominal rate that correspond to inflation in the destination market. The wholesale price indices are annual averages taken from International Financial Statistics.
3. Estimation and Preliminary Results
For each source country-industry pair, the regression equations for each destination are estimated jointly, imposing the cross-equation restrictions. The errors are assumed to be independent and identically distributed. Errors must be assumed to be uncorrelated across equations, since the presence of a full set of time dummies in the model precludes estimating an unrestricted covariance matrix.
The results of initial regressions based on equation (1) are presented in Tables 1-4 at the end of the paper. Each column of each table reports the results for one industry for one of the source countries. The rows indicate the destination markets. For each
destination equation the table reports the estimate of ß and its standard error, the adjusted R2, and the Durbin-Watson statistic. Estimates of the time effects are not reported since they are of little inherent interest. The destination effects were small in magnitude and seldom statistically significant. Consequently, they were dropped from estimation to conserve on degrees of freedom.
There are several points worth noting about the broad patterns in the estimates of ß.
The results for the U.S. export industries seldom show evidence of significant, negative values of ß. Only the paper industry shows statistically significant evidence of pricing to market behavior that that takes the form of local currency price stabilization to a number of destinations. There is weaker evidence of LCPS in film, aluminum foil and small cars.
For the other six industries, correlation between export price adjustment and exchange rate changes is absent or perverse.
For the other source countries, the evidence of LCPS is much stronger. For the U.K., autos, synthetic dyes, tools, and film all show moderate to strong evidence of LCPS. Of the 14 Japanese industries, only three-fish hooks, golf balls and selenium-lack evidence of LCPS. In some cases, the estimates have rather large standard errors, but they
tend to be uniformly negative. For the 18 German industries, the estimates tend to have much smaller standard errors and again are overwhelmingly indicative o f LCPS.
Tables 5-8 provide a more compact summary o f the results. They present the estimated valued of ß for each source country-industry pair when it is constrained to be the same across destinations. The tables indicate that the null hypothesis of identical values of ß is seldom rejected by the data. Whether one interprets this as evidence that behavior is indeed identical across export destinations depends on prior beliefs. A strong prior of common ß 's would be validated by the data. Thus, there is weak evidence that PTM behavior does not depend critically on destination market. The percentage of point estimates by industry for each source country that imply LCPS is as follows: Germany 89%, Japan 79%, U.K. 67%, and U.S. 45%. Once again, this evidence is suggestive of important differences in behavior that are related to source country.
There is also substantial variation in the estimated values o f ß by industry. The results for German export price adjustment in Table 5 show that LCPS is pervasive in chemical products, which include aluminum oxide, synthetic dyes, preparations for
synthetic dyes, titanium oxide pigments, titanium dioxide, aluminum hydroxide, vitamin A, and vitamin C. Surprisingly little evidence of PTM can be found in German exports of large automobiles, whereas the data indicate LCPS is present in price adjustment on exports of small autos.9 Of the three alcoholic beverage categories, beer and sparkling wine both show evidence of LCPS, while there is no evidence of PTM in white wine. Finally, LCPS characterizes price adjustment in exports of fan belts, steel containers and steel rails.
9 The seemingly puzzling results for large autos deserve some comment. In a previous paper (Knetter (1990)) I have argued that it appears that export prices to the U.S. are quite sensitive to the $/DM exchange rate, but that export prices to all other destinations follow the U.S. price. Hence, the empirical framework here does not detect price discrimination related to exchange rate changes, but yet dollar prices of German autos may move relatively little in comparison with the exchange rate. This may be a credible strategy in light of the fact that about 70% of German exports of large autos go to the U.S., with no other market receiving more than about 3% of total exports. Alternatively, one might also conclude that German automakers have more market power in large autos and as a consequence are better able to pass on exchange rate changes to buyers more easily than they can in small autos, where they may face greater competition.
Arguably the most interesting comparisons invlove those industries for which more than one source country is included in the data sample.10 Since the character of PTM seems to vary a great deal by industry, the source country patterns may be a consequence of the particular export industries chosen for each country. Exports of photographic film are included in the data samples for Japan, U.S., and U.K. The point estimates of ß in the regressions that constrain the coefficients to be the same across destinations (Tables 5-8) are - A l l for the U.K., -.519 for the U.S., and -.940 for Japan. While these numbers do suggest LCPS is pursued more vigorously by Japanese exporters, the behavior is not qualitatively different.
In automobiles, the U.S. does much less LCPS than Japan or the U.K., and somewhat less than Germany in small autos. Gagnon and Knetter (1990) argue that this reflects the fact that exports are a trivial share of total U.S. auto sales abroad. Most U.S.
sales abroad are acheived by production in the market of final sale via multinational operations. Thus, the few autos that are exported tend to be specialty items that are not expected to gain significant shares in the foreign markets. As a result, little attention may be paid to pricing exports. It seems likely that LCPS achieves for foreign producers what foreign production achieves for U.S. producers. As a result, these findings should not be construed as indicating U.S. firms behave differently than other firms faced with the same environment.
In aluminum foil, the point estimates of ß for Japan and the U.S. are -1.38 and -1.15, respectively. It is unlikely the difference is statistically significant given the magnitude of the standard errors. Two paper products are included in the sample, photographic paper exports by Japan and photocopier paper exports by the U.S. The estimates of ß are -.611 for Japan and -.914 for the U.S. In this case, the U.S. exporters engage in a higher degree of LCPS, although these are not identical products. There is no 10 As of this draft, I have yet to pool data for a given industry across source countries to test the restriction of identical slopes, although this will be included in a revised version of the paper.
evidence o f PTM in U.S. or U.K. exports of whiskey, with both estimated coefficients very close to zero. In switches, there is no evidence of PTM in the U.K. data, whereas the U.S. data suggests a perverse result. The positive value of ß implies that markup
adjustment amplifies the effect of exchange rate changes on prices paid by the buyer.
There are several chemical products in which comparisons are possible. The German data show no evidence of PTM in aluminum oxide exports. The U.S. data reveal a positive point estimate, another perverse result. In both cases, the standard errors of the estimated coefficients are quite large, reflecting extremely large standard deviations of the mean changes in the unit values for this product. For synthetic dyes, the point estimates of ß for German and British exports are remarkably close and show strong tendencies for LCPS-the values being -.575 for Germany and -.616 for the U.K.
Exports o f titanium dioxide are included in the U.S., Japanese and German data samples. Here the U.S. appears to be quite different in pricing policy than the other countries. Both Germany and Japan exhibit LCPS, with the estimate for Japan exceeding one in magnitude. The point estimate for the U.S. is positive. Upon closer inspection though, it appears that the U.S. estimate is driven by outliers in the export unit values on shipments to Germany. The standard deviation of the mean price change on shipments to Germany is about three times the magnitude of the same statistic for the other destinations.
When Germany is dropped as a destination, the point estimate of ß for U.S. exports of titanium dioxide is -.57 with a standard error of .23. This is virtually indistinguishable from the estimated behavior of German exporters.
For those industries in which more than one source country is included in the sample, the general finding seems to be that behavior is not terribly different across source countries.11 This finding stands in sharp contrast to the conclusion one might reach without accounting for industry. Indeed, one of the stylized facts of previous work is that 11 In the two cases where U.S. behavior is different, switches and aluminum oxide, the data are extremely noisy indicated by the large estimated standard errors.
U.S. exporters exhibit far less evidence of LCPS than their counterparts in Japan or Germany. To the extent that fact is true in the aggregate, it could well reflect the mix of export industries rather than genuine diffemces in behavior in the same industries.
Tables 9 and 10 report the results of estimating equation (1) on the German and U.S. data with both contemporaneous (ß ß and one-period lagged (/J2) values o f the exchange rate variable included in the regression.12 The coefficients are constrained to be equal across destinations, in part to economize on space in presentation. Recall that quadratic adjustment costs in quantities will lead to a slow response of quantities to exchange rate changes and a greater degree of LCPS in the short run. Thus, we would expect to find to be positive if some LCPS is undone over time as quantities are adjusted. The sum o f the coefficients could be thought of as the “longer run” equilibrium response.
Table 9 presents the results for German export price adjustment. Comparing the sum of the coefficients in Table 9 with the coefficients in Table 5 reveals that the measured degree of LCPS actually increases in 10 of the 18 industries when lagged exchange rates are included. The differences are typically small (less than . 10 in half of the cases), with a few exceptions. The industries in which is positive and reasonably large in magnitude are aluminum oxide, synthetic dyes, Vitamin A, sparkling wine, steel containers, and steel rails. Ex ante, one would think that automobiles might be more likely to fit the costly adjustment story. For each category of autos, the estimate of ß 2 is negative, suggesting that LCPS increases rather than decreases over time.
Turning to Table 10 on U.S. export price adjustment, the measured degree of LCPS (given by the sum of the coefficients) falls in 8 of 11 industries relative to the contemporaneous measures given in Table 7. Notable are the aluminum foil industry in which the initially large degree of LCPS is entirely undone in one period and the
12 Lagged values of the dependent variable were also included in estimation, but the coefficients were small and almost never significant. Consequently, they were dropped to conserve degrees of freedom.
photocopier paper industry in which the reduction in LCPS is fairly large. Only paper and film are characterized by a reasonably large degree of LCPS when lags are included in the model. While most estimates of are positive, in many cases they reinforce perverse markup adjustment to contemporaneous exchange rate changes. In general, Tables 9 and
10 provide little support for the importance of adjustment costs in explaining export price adjustment with annual data.
4. Concluding Remarks and Remaining Work
While this paper is still work in progress, several conclusions can be offered based on the results to date. First, substantial doubt has been cast on the importance of
destination market in determining the extent of LCPS. The destination where one might have expected the most pronounced tendency by sellers to stabilize prices, the U.S., does not appear to receive different treatment based on the results of Tables 1-4. Furthermore, the F-tests for common ß ’s across destinations presented in Tables 5-8 almost never reject the null hypothesis.
Second, comparisons of source country behavior within common industries indicate remarkably little evidence of differences in behavior. This is contrary to the impression generated by previous work on markup adjustment by Mann (1986), Rnetter (1989) and Ohno (1989). To the extent U.S. behavior appears different in the aggregate, it may be due to pattern of industry specialization, rather than behavioral differences by firms in the same industry.
The third (more tentative) finding is that the dynamic pattem of export price adjustment suggests that very few industries are characterized by more LCPS in the short run. This questions the importance of adjustment costs in generating observed patterns of PTM. It should be noted however, that such phenomena may be more important in higher
frequency data. Furthermore, only one-period annual lags are considered here due to the limited number of time periods in the data set.
This study will proceed along a number of dimensions. I will highlight the major tasks of my research agenda in this concluding section.
A. Interpretation o f Results.
One obvious step will be to assemble more qualitative and quantitative information about market structure for the industries studied here. The number of firms in the industry, the location o f major competitors, the importance o f exports in total sales (e.g., the extent of multinational production and the importance of the domestic market), ± e relative importance of particular export destinations (share of total exports going to a destination), and the nature of technology (in terms of the share of fixed costs in total cost) are all factors that may affect the observed price response to exchange rate changes. Armed with better information about industry structure, it should be possible to determine which factors are most important in generating the industry patterns we observe in the data. In addition, I am currently collecting information on 10 additional U.S. export industries. More evidence on the U.S. will help clarify whether U.S. experience is indeed anomolous.
B. Robustness.
There are a number of extensions that can test the robustness of the empirical results. Two general areas that deserve attention are model specification and the handling of outliers in the data. Regarding specification, the first check will be to run a non-linear version of equation (1) that is outlined in Knetter (1990). This model imposes a within equation constraint that forces the impact of unobserved, common disturbances to price (measured by the time effects) to be symmetric with the impact of exchange rate changes on price.
Other possible changes in specification include the inclusion of other destination- specific demand variables, such as real income.13 Another possible adjustment would be to allow the change in the destination market price level to enter the equation seperately, rather than as a part of the exchange rate series. In addition, it may be useful to run the model in logs (rather than first differences of logs) with lagged dpendent and independent variables included to verify that the main findings remain unchanged.
Finally, because the unit value data are known to be rather noisy, it is possible that outliers in the data have undue influence in some of the estimated equations, as was demonstrated in the case o f U.S. exports of titanium dioxide. Indeed, quick checks on the data have revealed that a number of the industries in which estimated coefficients lie outside the range in which LCPS occurs (between zero and minus one) are plagued by a few extreme observations. It has yet to be determined whether these observations reflect error in transcribing data or just noise in the published series due to composition shifts in exports.
Two approaches to dealing with this problem are under consideration. First, one could adopt the conventional method for handling outliers in which the sum of absolute, rather than squared, deviations is minimized. This approach reduces the weight attached to outliers, so they have less effect on the estimated coefficients. Another possibility with the panel data used here is to drop entire destinations for which the price change data appear to be exceptionally noisy. A systematic procedure, such as excluding destinations with unusually large standard deviations of period-to-period changes in price compared to other destinations, could be adopted to guard against data mining. In general it would be
comforting to know that the estimated coefficients are not overly sensitive to those destinations that are included in the sample. If the inclusion of one particular destination has a big impact, it would raise suspicions about robustness.
13 Income variables had little effect on estimated coefficients in previous models that used quarterly data.
This is not surprising given the Mccse and Rogoff (1983) evidence that there is little correlation between exchange rate changes and income.
C. Testing fo r Structural Change and Asymmetries
A good deal of recent theoretical research has emphasized the possiblility that large swings in exchange rates (such as the dollar cycle of the 1980’s) can have permanent effects on market structure and thus alter the equilibrium relationship between prices and exchange rates. Richard Baldwin (1988) and Avinash Dixit (1989) have shown how exchange rate fluctuations can induce hysteresis. There is little empirical evidence on the importance of this issue. Since the data set used here includes information on a relatively large number o f trading relationships over the entire floating exchange rate period, it is well suited to examining the importance of hysteresis in traded goods prices and quantities.
Simple tests for hysteresis involve testing for structural change (e.g., a change in the value of ß) at particular points in the sample, as in Baldwin (1988).
Another issue of interest that can be addressed quite thoroughly with this data set is whether there are asymmetries in the response of prices to exchange rates that depend on the direction o f the exchange rate change. For depreciations of the seller’s currency, marketing and distribution bottlenecks may give rise to adjustment costs that are not present during appreciations. If adjustment costs are one-sided, a different empirical appraoch would be required to evaluate their significance. A simple test would be to split the sample of observations by the direction of currency movements, and then estimate the model on the two subsamples. If asymmetries are important, one would expect the parameter estimates to differ over these subsamples. A more sophisticated approach would take into account past information on quantity flows to determine when bottlenecks are encountered.
REFERENCES
Baldwin, Richard, “Hysteresis in Import Prices: The Beachhead Effect”, American Economic Review, September 1988, 773-785.
Dixit, Avinash, “Hysteresis, Import Penetration, and Exchange Rate Pass-Through”, Quarterly Journal o f Economics, May 1989, 205-228.
Dornbusch, Rudiger, “Exchange Rates and Prices”, American Economic Review, March 1987, 93-106.
Feenstra, Robert, “Symmetric Pass-Through of Tariffs and Exchange Rates Under Imperfect Competition: An Empirical Test”, Journal o f International Economics, forthcoming 1990.
Froot, Kenneth, and Klemperer, Paul, “Exchange Rate Pass-Through When Market Share Matters”, American Economic Review, September 1989, 637-654.
Gagnon, Joseph, and Knetter, Michael, “Pricing to Market in International Trade: Evidence from Panel Data on Automobiles and Total Manufacturing”, Federal Reserve Board of Governors International Finance Discussion Paper #389, October 1990.
Kasa, Kenneth, “Adjustment Costs and Pricing-to-Market: Theory and Evidence”, Cornell University, manuscript, 1990.
Knetter, Michael, “Price Discrimination by U.S. and German Exporters”, American Economic Review, March 1989, 198-210.
Knetter, Michael, “Exchange Rate Pass-Through and Pricing to Market: An Empirical Synthesis”, Dartmouth College Working Paper, 1990.
Krugman, Paul, “Pricing to Market When the Exchange Rate Changes”, in S.W. Arndt and J.D. Richardson, eds., Real-Financial Linkages Among Open Economies,
Cambridge: MIT Press, 1987.
Mann, Catherine, “Prices, Profit Margins, and Exchange Rates”, Federal Reserve Bulletin, June 1986, 366-379.
Marston, Richard, “Pricing to Market in Japanese Manufacturing”, Journal oflnernational Economics, December 1990.
Meese, Richard, “Currency Fluctuations in the Post-Bretton Woods Era”, Journal o f Economic Perspectives, W inter 1990, 117-134.
Meese, Richard and Rogoff, Kenneth, “Empirical Exchange Rate Models of the Seventies:
Do They Fit Out of Sample?”, Journal o f International Economics, February 1983, 3-24.
Ohno, Kenichi, “Export Pricing Behavior of Manufacturing: A U.S.-Japan Comparison”, International Monetary Fund Staff Papers, 550-579.
Table 1 A. German Export Pricing Industries
Destination Vitamin A Vitamin C Sparkling Wine White Wine Fan Belts France
ß (se) .51 (1.26) -.33 (.16) .75 (1.35) -.77 (.40)
R2 .31 .78 .16 .22
DW Canada
2.78 2.60 1.76 2.31
ß (se) -.25 (.57) -.21 (.06) -.18 (.35) -.10 (.11)
R2 .64 .67 .51 .90
DW 3.08 2.57 1.69 2.09
United Kingdom
ß (se) -.07 (.86) -.26 (.11) -.54 (.73) -.33 (.23) -.55 (.30)
R2 .00 .81 .54 .93 .84
DW Japan
2.70 1.88 2.68 1.91 2.65
ß (se) -.96 (.50) -.31 (.15) -.39 (.24)
R2 .31 .45 .76
DW
United States
3.13 2.33 1.77
ß (se) -.56 (.51) -.28 (.13) -.81 (.31) -.18 (.12) -.44 (.20)
R2 .21 .49 .72 .89 .51
DW 1.81 2.55 1.64 1.66 2.54
Denmark
ß (se) -1.29 (.40)
R2 .95
DW Sweden
1.61
ß (se) -.68 (.17)
R2 .59
DW 1.61
Table IB. German Export Pricing
Destination Stl Rails______ Stl Contnrs Autos>3L. 2.1-3L. Autos 1.6-2L Autos France
Industries
ß (se) ■1.50 (1.68) .56 (.81)
R2 .44 .55
DW Canada
2.60 1.22
ß (se) -.39 (.31)
R2 .74
DW
United Kingdom
1.92 ß (se) -.56 (1.51) .04 (.55)
R2 .26 .62
DW Japan ß (se) R2 DW
United States
1.72 2.67
ß (se) -.88 (.93) -.24 (.27)
R2 .39 .80
DW Denmark ß (se) R2 DW Sweden ß (se) R2 DW
1.37 1.28
-.64 (.90) .77 2.55
.05 (.48) .29 2.91
.20 (.22) .03 1.91 .09 (.11)
.22 1.96
.31 (.38) .32 2.60
-.14 (.35) .30 2.07 .31 (.17)
.44 2.22
-.22 (.25) .39 0.83
-.28 (.32) .18 1.66 .01 (.26)
.24 1.98
.26 (.36) .29 2.51
.13 (.23) .25 2.21 -.11 (.09)
.32 2.49
-.12 (.15) .53 0.75
-.63 (16) .77 1.81
-.25 (.63) .28 (.21) -.10 (.16)
.56 .37 .10
1.98 1.79 1.72
Table 1C. German Export Pricing Industries
Destination Alum Oxide 1.1-1.5L Autos Beer Synthetic Dves France
ß (se) -.04 (1.60) .63 (.43) -.92 (.22) -.98 (.17)
R2 .00 .12 .31 .42
DW Canada
2.56 2.20 1.71 2.29
ß (se) .67 (1.71) -.51 (.18) -.69 (.20)
R2 .75 .59 .74
DW 2.38 1.44 1.48
United Kingdom
ß (se) -.44 (.99) .04 (.33) -.17 (.22) -.35 (.14)
R2 .25 .18 .62 .64
DW Japan
1.06 3.05 2.32 2.66
ß (se) -1.07 (1.26) -.45 (.50) -.01 (.20) -.68 (.17)
R2 .13 .49 .16 .60
DW
United States
1.26 1.83 1.44 2.09
ß (se) -.36 (.72) -.63 (.39) -.57 (.11) -.53 (.07)
R2 .16 .59 .84 .79
DW Sweden
1.71 2.95 1.27 2.04
ß (se) 2.00 (1.72) .54 (.30)
R2 .06 .10
DW 2.45 2.79
Table ID. German Export Pricing Industries
Destination Titan Ox Pigmnt Dve Prenrtns Titan Diox Al Hydroxide France
ß (se) -1.41 (.38) -.35 (.95) -1.39 (.48) -.33 (.51)
R2 .73 .40 .74 .51
DW 1.66 1.74 1.15 1.67
Canada
ß (se) -.50 (.18)
R2 .59
DW 2.84
United Kingdom
ß (se) -.90 (.31) -.22 (.55) -.73 (.38) -.76 (1.24)
R2 .61 .40 .36 .62
DW 1.85 2.67 1.75 3.09
Japan
ß (se) -.91 (.50) .01 (.49) -.83 (.74) .22 (.44)
R2 .62 .51 .66 .73
DW 2.34 2.03 1.41 1.32
United States
ß (se) -.81 (.24) .06 (.55) -.64 (.35) -.96 (.28)
R2 .78 .85 .16 .70
DW 1.16 2.83 2.55 1.48
Sweden
ß (se) -1.06 (.40) -.92 (.29) -.02 (.36)
R2 .60 .67 .26
DW 1.73 3.05 1.64
Table 2A. Japanese Export Pricing Industries
Destination Color Film AlumFoil Fish Hooks Tires
Canada
ß (se) -.85 (.15) -2.87 (1.31) .61 (.57) -.28 (.65)
R2 .73 .40 .69 .03
DW Korea
2.20 2.57 2.06 3.12
ß (se) -1.29 (.46) -1.16 (.92) .77 (.78)
R2 .63 .08 .39
DW Sweden
1.69 1.84 2.33
ß (se) -.54 (.32) -.24 (1.24:
R2 .57 .19
DW 2.52 2.84
Germany
ß (se) -1.04 (.31) -.68 (1.57) 2.75 (.98) -1.73 (2.5!
R2 .70 .24 .66 .48
DW
United States
2.71 2.81 2.71 3.17
ß (se) -.96 (.19) -1.48 (.86) 1.16 (.61) -.13 (.53)
k 2 .69 .30 .32 .00
DW Australia
2.71 1.53 2.33 3.38
ß (se) -.45 (.21) -1.24 (1.30) 1.47 (.77) -.10 (.65)
R2 .39 .01 .12 .10
DW 2.46 2.40 2.69 3.26
United Kingdom
ß (se) -.80 (.22) .02 (.52)
R2 .73 .06
DW 2.19 2.64
Phillipines
ß (se) -1.24 (.96) .91 (.59)
R2 .27 .00
DW 1.90 2.01
Table 2B. Japanese Export Pricing Industries
Destination 1.1-2L. Autos A utosclL . Autos>2L. Golf Balls Photo Paper Canada
ß (se) -.68 (.08) -.81 (.19) -.52 (.63)
R 2 .82 .81 .01
DW Korea
1.41 1.30 1.98
ß (se) -.32 (.41)
R2 .84
DW Sweden
3.15
ß (se) -.53 (.13) 3.34 (1.26)
R2 .63 .63
DW Germany
2.17 1.38
ß (se) -.52 (.16) -.25 (.33) -.96 (.36) 1.27 (1.54)
R2 .67 .33 .76 .13
DW
United States
1.39 1.92 1.86 2.32
ß (se) -.71 (.09) .20 (.31) -.74 (.19) .83 (.71) -1.02 (.75)
R2 .81 .39 .73 .84 .78
DW Australia
1.34 1.18 2.42 1.56 2.80
ß (se) -.78 (.50) -.86 (.93)
R2 .60 .62
DW
United Kingdom
1.98 1.23
ß (se) -.35 (.09) .03 (.35) -.39 (.24) 1.92 (.82) -.83 (.83)
R2 .30 .06 .34 .87 .48
DW
Switzerland
2.65 1.66 1.70 1.50 2.71
ß (se) -.27 (.33)
R2 .22
DW Norway ß (se)
2.01
-.70 (.30)
R 2 .70
DW 2.47
Table 2C. Japanese Export Pricing Industries
Destination Iiw er Tubes Imit Pearls Portld Cement Titan Diox Selenium Canada
ß (se) -3.41 (1.62) -.11 (.62) -2.11 (1.31)
R2 .14 .73 .60
DW Australia
2.82 2.66 3.33
ß (se) -.54 (2.59) -1.00 (.69)
R2 .74 .37
DW
United States
2.89 2.67
ß (se) ■2.25 (1.37) -.75 (.36) -.20 (.75) -1.89 (1.66) .47 (.65)
R2 .50 .18 .44 .74 .88
DW 2.53 2.86 3.04 1.56 1.80
United Kingdom
ß (se) 1.88 (1.32) -.71 (.53)
R2 .41 .60
DW 1.81 1.86
Germany
ß (se) -1.77 (.37)
R2 .65
DW Korea ß (se)
2.30
-1.82 (.92) .58 (.72)
R2 .05 .59
DW
Saudi Arabia
2.08 1.88
ß (se) -.48 (.55)
R2 .16
DW 2.40
Phillipines
ß (se) -1.60 (.64)
R2 .59
DW 1.54
India/Netherlands
ß (se) -2.35C90) .88 (1.38)
R2 .61 .79
DW 1.58 1.47
Table 3A. United States Export Pricing Industries
Destination Autos >6cvl Autos<6cvl Alum Oxide Cigarettes Paper Canada
ß (se) .19 (.29) -.37 (.19) 1.80 (.78) -.03 (.25) -1.74 (.62)
R2 .26 .50 .65 .47 .44
DW
Switzerland
2.21 1.73 1.95 1.66 2.02
ß (se) .16 (.31) -.13 (.18) .08 (.19)
R2 .34 .64 .47
DW Sweden
2.35 2.69 2.95
ß (se) .15 (.28) -.44 (.34) .22 (1.28) -.01 (.16) -.85 (.51)
R2 .20 .65 .67 .44 .44
DW 2.05 3.05 1.65 2.15 2.69
Germany
ß (se) .05 (.45) -.21 (.25) 1.26 (1.22) .12 (.24) -.63 (.40)
R2 .02 .46 .34 .41 .30
DW 2.43 1.90 1.46 1.64 2.35
Italy
ß (se) -.45 (.67) 1.69 (.54) -.16 (.11) -.89 (.46)
R2 .28 .20 .58 .58
DW 2.26 1.92 2.33 2.46
Japan
ß (se) .20 (41) -.21 (.24) 3.00 (.81) .10 (.23) -.74 (.50)
R2 .29 .53 .08 .23 .44
DW Australia
2.99 1.97 1.61 1.88 2.40
ß (se) .35 (.67) .39 (.99) .02 (.14) -.87 (1.90)
R2 .06 .44 .49 .37
DW 2.79 2.49 1.35 2.53
United Kingdom
ß (se) -.22 (.49) -.15 (.27) 1.05 (.51) -1.22 (.57)
R2 .14 .01 .39 .00
DW 2.61 3.17 1.86 2.85
Table 3B. United States Export Pricing Industries
Destination tMum Foil Whiskev Yellow Com Titan Diox Film Canada
ß (se) -.89 (1.61) -4.86 (2.92) .07 (1.07) -.63 (1.36)
R2 .21 .80 .10 .18
DW
Switzerland
1.75 1.15 1.28 3.04
ß (se) -1.25 (.64)
R2 .77
DW 2.23
Germany
ß (se) .06 (1.09) -.01 (.48) -.07 (.35) 1.52 (1.55) -.64 (.34)
R2 .44 .28 .85 .78 .46
DW 2.21 2.06 1.60 2.14 2.42
Italy
ß (se) -1.76 (.93) .17 (.58) .27 (.45) -1.28 (.63)
R2 .45 .27 .73 .70
DW 1.60 2.78 1.43 2.62
Japan
ß (se) -.96 (.64) .28 (.71) .04 (.30) .48 (.76) -.16 (.67)
R 2 .18 .21 .80 .38 .00
DW Australia
1.77 3.03 1.24 1.97 1.18
ß (se) -3.06 (1.26) -.01 (.56) 1.02 (1.23) 1.37 (.40)
R2 .11 .76 .04 .26
DW 2.24 3.09 2.15 2.90
United Kingdom
ß (se) -1.26 (.76) -.10 (.29) .02 (.29) .24 (.72) -.43 (.58)
R 2 .21 .43 .84 .05 .70
DW 3.06 3.02 1.22 2.34 3.20
Table 4A. British Export Pricing
Industries
Destination Autos Tractors Whiskey Books
United States
ß (se) -1.07 (.95) -.07 (.19) -.07 (.10) -.08 (.28)
R2 .23 .71 .45 .13
DW 2.27 2.63 2.13 2.45
Canada
ß (se) -1.19 (.80) -.14 (.38) .07 (.17) -.08 (.23)
R2 .50 .50 .45 .74
DW 1.98 2.74 1.82 2.07
Germany
ß (se) -.81 (.73) -.08 (.22) .22 (.20)
R2 .33 .45 .46
DW 2.38 1.45 2.62
Japan
ß (se) -.31 (.73) -.18 (.19) .06 (.06) -.02 (.26)
R2 .06 .55 .17 .10
DW 1.92 2.45 1.47 1.78
Italy
ß (se) -1.50 (1.04) .16 (.20)
R2 .36 .61
DW 2.24 1.30
Australia
ß (se) -.38 (.81) -.17 (.30) .22 (.14) .05 (.20)
R2 .22 .70 .40 .39
DW 2.20 2.24 1.64 2.60
Sweden
ß (se) -.26 (.28) -.05 (.27) .24 (.27)
R2 .65 .39 .57
DW 1.91 2.19 1.72
Switzerland
ß (se) -.69 (.76)
R2 .05
DW 3.07
Table 4B. British Export Pricing Industries
Destination Switches Dyes Tools Film Fabric
United States
ß (se) -.11 (.28) -.61 (.18) -1.76 (.58) -.56 (.34) -.08 (.15)
R2 .44 .39 .44 .43 .67
DW 2.44 1.45 2.30 2.61 1.97
Canada
ß (se) -.04 (.31) -1.31 (.69) -.07 (.16)
R2 .28 .23 .65
DW 3.25 2.90 2.47
Germany
ß (se) .69 (.14) -.53 (.26) -1.31 (.88) -.30 (.98) .11 (.24)
R2 .44 .56 .41 .49 .73
DW 2.00 .235 3.21 3.05 2.72
Japan
ß (se) -.67 (.28) .09 (.17)
R2 .71 .36
DW 2.42 1.98
Italy
ß (se) -.24 (.31) -.71 (.34) -1.29 (.64) -.14 (.22)
R2 .33 .62 .85 .58
DW 2.07 1.49 1.60 2.63
Australia
ß (se) -.15 (.32)
R2 .27
DW 2.58
Sweden
ß (se) .11 (.36) -.95 (.88) -.93 (.34)
R2 .39 .19 .88
DW 3.20 2.73 1.54
Switzerland
ß (se) .35 (.33) -.17 (.40)
R2 .08 .43
DW 2.62 2.56
Table 5. German Exports - Constrained Estimates o f ß
Industry ß js e ) F-Stat
Autos over 3L. .048 (.11) 1.175
Autos 2.1 - 3L. .083 (.23) 1.492
Autos 1.6 - 2L. -.356 (.25) 2.257
Aluminum Oxide -.029 (.98) 0.523
Autos 1.1 - 1.5L. -.542 (.30) 1.684
Beer -.435 (.16) 2.708*
Synthetic Dyes -.575 (.10) 1.679
Titanium Oxide Pigments -.809 (.25) 0.331 Preparations for Syn Dye -.188 (.33) 0.260
Titanium Dioxide -.645 (.30) 0.235
Aluminum Hydroxide -.724 (.42) 0.979
Vitamin A -.488 (.44) 0.265
Vitamin C -.249 (.07) 0.059
White Wine -.021 (.14) 3.925*
Sparkling Wine -.610 (.29) 0.840
Fan Belts -.432 (.23) 0.248
Steel Containers -.359 (.23) 0.770
Steel Rails -.774 (.98) 0.302
* : reject constraint at 5% level.
Table 6. Japanese Exports - Constrained Estimates of ß
Industry F-SttU
Color Film -.940 (.28) 1.500
Photo Paper -.611 (.37) 0.599
Aluminum Foil -1.38 (.81) 0.794
Fish Hooks .854 (.60) 1.678
Tires -.167 (.57) 0.332
Autos 1.1 - 2L -.615 (.10) 3.375*
Autos IL or less -.182 (.24) 1.282
Autos over 2L -.689 (.22) 2.589*
Inner Tubes -2.26 (1.18) 2.938*
Imitation Pearls -.484 (.37) 2.637
Portland Cement -.570 (.64) 1.601
Titanium Dioxide -1.53 (.64) 0.251
Selenium .545 (.62) 0.157
Golf Balls 1.42 (1.16) 10.57*
* : reject constraint at 5% level.
Table 7. U.S. Exports - Constrained Estimates of ß
Industry ß ^sel
EzStat
Autos over 6 cyl .004 (.35) 0.368
Autos 6 cyl or less -.192 (.19) 0.369
Aluminum Oxide 1.45 (.65) 1.339
Cigarettes .032 (.14) 0.554
Paper -.914 (.48) 0.311
Aluminum Foil -1.15 (.85) 1.591
Bourbon .052 (.35) 0.147
Yellow Com -.155 (.13) 11.2*
Titanium Dioxide .725 (.92) 0.695
Switches 1.73 (.61) 0.647
Film -.519 (.59) 2.015
* : reject constraint at 5% level.
Table 8. British Exports - Constrained Estimates of ß
Industry ß^se) F-Stat
Automobiles -.829 (.63) 0.665
Tractors -.146 (.22) 0.137
Whiskey .040 (.08) 0.655
Switches .052 (.15) 0.966
Books .005 (.16) 0.348
Film -.477 (.30) 1.583
Fabric -.017 (.12) 0.486
Synthetic Dye -.616 (.14) 11.04*
Tools -1.47 (.52) 0.315
* : reject constraint at 5% level.
Table 9. German Export Pricing - Constrined Estimates o f Contemporaneous ( ß p and One-Period Lagged (ß2) Exchange Rate Coefficients.
Industry (ss) f c l s e ) ( £ l ± £
Autos 3L and over .06 (.14) -.19 (.15) -.13
Autos 2.1 - 3L .22 (.24) -.24 (.13) -.02
Autos 1.6 - 2L -.34 (.27) -.06 (.15) -.40
Autos 1.1 - 1.5L -.53 (.29) -.25 (.31) -.78
Aluminum Oxide -.'ll (.75) 1.56 (1.21) .79
Beer -.32 (.18) -.33 (.18) -.65
Synthetic Dyes -.63 (.10) .28 (.09) -.35
Titanium Oxide Pigment -.96 (.25) -.01 (.17) -.97 Preparations for Syn Dyes -.22 (.38) -.01 (.31) -.23 Titanium Dioxide -.73 (.24) -.39 (.11) -1.12 Aluminum Hydroxide -.74 (.43) -.01 (.35) -.75
Vitamin A -.79 (.54) .50 (.46) -.29
Vitamin C -.25 (.09) .07 (.07) -.18
White Wine -.01 (.14) .02 (.13) .01
Sparkling Wine -.58 (.21) .29 (.37) -.29
Fan Belts -.32 (.26) -.20 (.28) -.52
Steel Containers -.49 (.23) .17 (.15) -.32
Steel Rails -1.45 (.97) .76 (.49) -.69
Table 10. U.S. Export Pricing - Constrained Estimates o f ß Contemporaneous (P i) and One-Period Lagged $ 2 ) Exchange Rate Coefficients
Industry Hl_£S£l ^ 2 ls e )
Autos over 6 cyl .00 (.34) -.10 (.42) -.10
Autos 6 cyl or less -.22 (.19) .13 (.11) -.09
Aluminum Oxide 1.32 (.64) .44 (.56) 1.76
Cigarettes -.01 (.12) .26 (.11) .25
Paper -1.01 (.53) .41 (.51) -.60
Aluminum Foil -1.28 (.85) 1.36 (.64) .08
Bourbon .03 (.35) .15 (.39) .18
Yellow Com -.09 (.10) .12 (.06) .03
Titanium Dioxide .74 (.88) .05 (.78) .79
Switches 1.86 (.70) -.31 (.81) 1.55
Film -.36 (.64) -.53 (.38) -.89
ZITIERWEISE/CITATION:
Michael M. Knetter:
International Comparisons of Pricing to Market Behavior, Discussion Paper FS IV 91 - 12, Wissenschaftszentrum Berlin für Sozialforschung 1991.
