Exchange Rates and Prices

Questions & answers

Q1. What is the paper’s central question, and what approach does it take to answer it?

The paper asks why an exchange rate change produces large relative-price changes for some groups of goods and negligible changes for others, and answers by building industrial-organization models of price-setting that differ by “the degree of market concentration, the extent of product homogeneity and substitutability, and the relative market shares of domestic and foreign firms” (Introduction). The analysis is explicitly partial equilibrium: it “assumes throughout a given, exogen[e]ous movement in the nominal exchange rate,” combined with “the less than fully flexible money wage,” so that the exchange rate change becomes a cost shock to some firms in an industry (foreign firms selling at home, home firms selling abroad) that forces an industry-wide pricing adjustment. The author explicitly flags that “the assumption of exogen[e]ous exchange rate movements and sticky wages is open to criticism,” defending it only “as a useful working hypothesis for the purpose of investigating relative price issues.”

Q2. What are the two polar benchmark models the paper sets up before building its own?

The paper contrasts a “law of one price” model, in which arbitrage equalizes the home-currency price of a good across countries (p = ep, so relative national price levels are independent of the exchange rate), with an alternative “Keynesian” model in which each country is fully specialized in its own good, goods are imperfectly substitutable, wages are sticky in national currency, and the real exchange rate (the relative price of domestic to foreign GDP deflators) consequently “change[s] one-for-one” with the nominal exchange rate* (Sec. 2). The paper’s stated finding is that “equation (1) [law of one price] would be a useful model of international price relations for materials – say sisal, copper, tea – whereas (2) [the Keynesian polar case] more nearly describes what happens with manufactures,” but that even for manufactures the assumption of a constant price mark-up “is no longer justified when domestic and foreign firms have strategic interactions in their pricing” – motivating the industrial-organization models that follow.

Q3. How does the paper model commodity (“materials”) prices, and what does the empirical test find?

For materials, the paper models world demand as a function of each region’s real commodity price (in terms of its own GNP deflator) and real activity, with commodity supply taken as exogenous, and shows the model implies the elasticity of the US real commodity price with respect to the real exchange rate “should be less than one,” reducing (under equal demand elasticities in both regions) to the US’s share of world commodity absorption (Sec. 3, eq. 3-4). Estimating this relationship on quarterly 1970:1-1985:1 data for metals and agricultural raw materials, the paper reports the real exchange rate “appear[s] as [a] significant determinant[] of changes in the real commodity price” with “the anticipated sign, but the quantitative effect is far larger than the model predicts” – coefficients of about -1.44 (metals) and -1.22 (agricultural materials) against a theoretical bound the paper says “should… certainly not [be] larger than unity in absolute value” (Sec. 3, Table 3). The paper attributes this “overreaction” tentatively to neglected structural/supply-side change (including inventories) and to commodity markets behaving more like asset markets, where anticipated-versus-unanticipated shocks and interest rates matter in ways the simple flow-demand model omits.

Q4. What does the Cournot oligopoly model of manufactures pricing predict about exchange-rate pass-through?

In a Cournot model of a homogeneous good sold by n domestic and n foreign firms in a spatially separated home market, each maximizing profit taking rivals’ quantities as given, the paper derives the elasticity of the home-market equilibrium price with respect to the exchange rate as g = (n/N)(ew*/p), where N = n + n*** (Sec. 4, eq. 8-9). Because both terms in this expression are fractions, “a dollar appreciation will lower [home] price less than proportionally,” with the pass-through larger “the more competitive the industry… and the larger the share of imports in total sales” – so the model spans continuously from the textbook “small country” case (price-taking, full pass-through) to a case of “virtually no impact” on home prices when there are few firms and most are domestic, in which case foreign firms “absorb the dollar appreciation primarily in the form of extra profits rather than increased sales.”

Q5. What do the Dixit-Stiglitz model and its “extended” version with strategic interaction predict?

In the baseline Dixit-Stiglitz model of monopolistic competition (CES preferences over differentiated brands, each firm too small to affect the industry price index), profit maximization yields constant-markup pricing over unit labor cost for every firm (Sec. 4, eq. 16-17), which implies “the relative price of domestic and foreign variants in the home market depends just on relative unit labor costs in a common currency” – so import prices fall in exact proportion to the decline in foreign dollar unit labor costs while domestic prices, and the dollar price of home exports, are predicted to stay unchanged. The paper notes this is “the exact specification of the fixed-price ‘Keynesian’ model,” now derived as an implication of given labor costs and an invariant mark-up rather than assumed directly. When the model is “extended” to let firms be large enough to affect the industry price and to respond to rivals’ prices via a conjectural-variation parameter between zero and one, the resulting equilibrium “differs from the Dixit-Stiglitz one and resembles more nearly the Cournot model” – foreign firms now cut price by less than their full cost reduction, and home firms also cut price, so pass-through becomes partial on both sides rather than the knife-edge Dixit-Stiglitz prediction (Sec. 4, “An Extended Dixit-Stiglitz Model”).

Q6. What does the Salop spatial-competition (“circle”) model add, and what do all the manufactures models agree on?

In the Salop model, consumers are distributed uniformly on a circle and buy from whichever of two adjacent firms (one domestic, one foreign, assumed to alternate) gives higher surplus net of a linear “distance” (taste-mismatch) cost; solving the resulting price-reaction functions for equally spaced, equal-wage firms gives a home-price elasticity to the exchange rate of 1/3 and a foreign-price elasticity of 2/3 (Sec. 4, eq. 26-27a), with both fractions shrinking as the number of firms or the per-unit-distance cost parameter falls, so again pass-through is partial and governed by market structure. Summarizing across the Cournot, Dixit-Stiglitz, and Salop models, the paper states “they all predict that appreciation should lead to a decline in the price of imports,” with “the extent of the decline depend[ing] on a measure of competition and on the relative number of home and foreign firms,” and that for differentiated products “it is always the case that the relative price of the imported brands declines in response to an appreciation” while “export and domestic prices will stay closer in line than import and domestic prices” (Sec. 4, “Summary”).

Q7. What does the paper’s own empirical evidence on manufactures prices show, and how does it line up with the models?

Comparing US to German and Japanese export prices over 1980:4-1984:4 across disaggregated matched series, the paper reports the US lost competitiveness by a mean of 39.3 percent against Germany and 24.9 percent against Japan (Sec. 4, Table 3) – consistent with incomplete pass-through rather than a single world price for each good. Comparing US export to US import prices within the same narrow industries, the paper reports “overwhelming evidence that virtually without exception the dollar appreciation of the past 5 years has been accompanied by an increase in the price of exports relative to imports” (e.g., non-electrical machinery and scientific instruments both show +18 percent cumulative export-price inflation against roughly -10 to -13 percent import-price deflation, 1980:4-1985:1, Table 4) – a pattern the paper notes “would obtain strictly only in the Dixit-Stiglitz model” among the models considered, though it “is a possibility” under the others. But looking at export and import prices each relative to domestic prices (Tables 6-7), export prices “change little relative to domestic prices,” while import prices decline relative to domestic prices by “relatively small” amounts compared to the roughly 40 percent swing in relative unit labor costs – “with a change in relative unit labor costs of more than 40 percent the decline in the relative price is in most cases less than 20 percent” – which the paper interprets as consistent with partial “pricing to the American market,” as the imperfect-competition models (rather than the polar Dixit-Stiglitz or law-of-one-price cases) predict.

Q8. What limitations does the paper place on its own analysis?

The paper explicitly scopes its models to “a relatively short time perspective,” stating “the wage rate is assumed not to react to changes in output and profitability and the number and location of firms in an industry is unaffected” (Concluding Remarks). It states these assumptions “are plausible in the short term” but that “a sustained real appreciation will ultimately show[] its effects in wage cuts in those industries where the loss in competitiveness causes unemployment and wage increases in the expanding sectors,” with firm entry, exit, and relocation across countries as a further longer-run channel the paper’s static, fixed-wage, fixed-market-structure models do not capture. A footnote also flags that the Cournot elasticity result (eq. 9) “is sensitive to the functional form of the demand curve,” reporting a different elasticity expression under constant-elasticity demand than under the linear demand assumed in the main text.