BBoowwddooiinn CCoolllleeggee BBoowwddooiinn DDiiggiittaall CCoommmmoonnss Economics Department Working Paper Series Faculty Scholarship and Creative Work 12-15-2010 DDooeess PPrroodduuccttiivviittyy RReessppoonndd ttoo EExxcchhaannggee RRaattee AApppprreecciiaattiioonnss?? AA TThheeoorreettiiccaall aanndd EEmmppiirriiccaall IInnvveessttiiggaattiioonn Yao Tang Bowdoin College, [email protected] Follow this and additional works at: https://digitalcommons.bowdoin.edu/econpapers Part of the Economics Commons RReeccoommmmeennddeedd CCiittaattiioonn Tang, Yao, "Does Productivity Respond to Exchange Rate Appreciations? A Theoretical and Empirical Investigation" (2010). Economics Department Working Paper Series. 2. https://digitalcommons.bowdoin.edu/econpapers/2 This Working Paper is brought to you for free and open access by the Faculty Scholarship and Creative Work at Bowdoin Digital Commons. It has been accepted for inclusion in Economics Department Working Paper Series by an authorized administrator of Bowdoin Digital Commons. For more information, please contact [email protected]. Does Productivity Respond to Exchange Rate Appreciations? A Theoretical and Empirical Investigation Yao Tang∗ Bowdoin College December 15, 2010 Abstract Although real currency appreciations pose direct difficulties for exporters and import-competing firms as they will face more intense competition, is it possible that such competition spurs firms to improve productivity? To answer this question, the paper first constructs a theoretical model to show how the competitive pressures of currency appreciations induce firms to improve productivity by adopting new tech- nologies. In addition, the model predicts that during appreciations there will be a positive relation between market concentration and improvements in productivity for industries highly exposed to trade, because the marginal benefits of productivity improvement will be bigger for firms with a larger market share. The paper then examines Canadian manufacturing data from1997 to 2006, and finds evidence consis- tent with model predictions. I find that growth rates of labor productivity were on average higher during the Canadian dollar appreciation between 2002 and 2006, after controlling for industry characteristics. Within the group of highly traded Canadian industries,themoreconcentratedonesexperiencedlargergrowthinlaborproductivity. JEL Classification: F3, F4 Keywords: exchange rate appreciation, productivity, technology adoption. ∗ 9700 College Station, Brunswick, Maine 04011-8497, USA. Email: [email protected]. Web page: http://www.bowdoin.edu/˜ytang/. I am grateful for comments from Werner Antweiler, Paul Beaudry, Michael Devereux, Viktoria Hnatkovska, Ross Hickey, Haifang Huang, Amartya Lahiri, Vadim Marmer, AnjiRedish,andHenrySiu. IalsothankThomasLemieuxandAndreyStoyanovfortheirhelpwithdata compilation. All errors are mine. 1 1 Introduction Substantial exchange rate movements over the last few decades have raised a question: what are the impacts of a major real exchange rate appreciation on economic perfor- mance? Conventional wisdom suggests that such appreciation worsens terms of trade and weakens the competitiveness of home firms. Meanwhile, the possibility remains that to maintain competitiveness, firms will be forced to raise productivity by reducing their costs. Some scholars and economic commentators argue that a “hard currency”, meaning a currency less prone to depreciation, can contribute to higher productivity growth. For instance, (Porter, 1990, p.640) suggests that the appreciations of the Yen in the 1980s had spurred the Japanese industry to become more competitive. Harris (2001) argues that the Canadian dollar depreciation in the 1990s was partially responsible for the Canadian productivity decline. To answer the question of whether manufacturing productivity responds to real ap- preciations,Ifirstconstructamodelinwhichcurrencyappreciationscanprovideincentives for firms to improve productivity if they are in industries highly exposed to trade. The model also predicts that among highly traded industries, the highly concentrated ones will invest more in productivity improvements since the marginal benefits of productivity gain will be greater for firms with a larger market share. Second, I test the predictions empiri- cally by using Canadian manufacturing data from 1997 to 2006. The results suggest that manufacturing productivity growth responded positively to the appreciation of the Cana- dian dollar between 2002 and 2006. Within industries exposed to a substantial amount of trade, the highly concentrated ones experienced a larger gain in labor productivity during the appreciation period. Inaneoclassicalframework, profit maximizationby firms automaticallyimpliescost minimization. However, some economists have long argued that product market compe- tition forces firms to lower costs and thus improve productivity. Nickell (1996) contains a 2 review of earlier contributions along this line of thinking. Some of the theoretical models are based on contract theory, for example Hart (1983) and Raith (2003). Vives (2008) examines a wide variety of industrial organization models, and concludes that, in general, increased competition encourages product and process innovations. Holmes, Levine and Schmitz (2008) provide a simple setup to explain the positive relation between competi- tion and adoption of new technology, based on the empirical observation that technology changes are often disruptive in the sense that the transition to higher productivity often features initially higher marginal costs. This paper adapts the Holmes et al. (2008) assumption of disruptive technological changetoclarifytheeffectofincreasedcompetitionduetorealexchangerateappreciations on productivity. In the model, one of the costs of adopting a cost-reducing technology is profit loss due to a temporarily high marginal cost of production during the transition. When the exchange rate appreciates, there is less profit to be made, and so profit loss due to adopting new technology is also smaller. However, if firms in an industry are shielded by high trade costs, then their profitability is less influenced by appreciations, and the incentive to improve productivity provided by appreciations is smaller. Compared to Holmes et al. (2008) and other previous papers which focus on when firms are likely to adopt new technologies to improve productivity, this paper also studies what types of firms are likely to invest more in productivity improvement. The model predicts firms will invest to achieve bigger productivity gain if they are in industries with a low trade cost and a high level of concentration. In industries with fewer firms, since the marginal benefits of productivity improvements are greater, firms in these industries are likely to invest more in productivity improvements. There are a number of studies that provide evidence of a positive correlation be- tween competition and productivity improvement, with competitive pressure measured as the number of competitors, concentration ratio, trade barriers, or the effect of competi- 3 tion policy. MacDonald (1994) finds that import competition improved productivity in highly concentrated US industries. Nickell (1996) suggests that an increase in the number of competitors was associated with total factor productivity (TFP) gain in a sample of 700 firms in the UK. Symeonidis (2008) exploits the variation arising from the introduc- tion of anti-cartel laws in UK industries, and finds that collusion reduced industry-level productivity growth. Galdon-Sanchez and Schmitz (2002) and Syverson (2004) are two papers that focus on individual industries. The former paper investigates Canadian and American iron ore producers, which doubled labor productivity, and increased material efficiency by 50% in response to intense price competition from Brazilian firms. The latter paper examines ready-mixed concrete plants in the US, and finds that an increase in local competition led to higher average productivity and lower productivity dispersion. A few recent firm-level studies examine the effect of exchange rate appreciation on firm performances. Fung (2008) examines the productivity responses of Taiwanese firms to a major currency appreciation and finds productivity gain mainly due to the exit of less efficient firms and bigger production scale of surviving firms after the appreciation. Using a micro data set from Norway, Ekholm, Moxnes and Ulltveit-Moe (2008) find that net- exporting manufacturing firms experienced productivity gain after the appreciation of the Norwegian Krone in the early 2000s. They argue that the gain in productivity came from technologicalimprovement,andemploymentcuts. Baggs,BeaulieuandFung(2009)study the relation between firm performances and exchange rate in Canada between 1986 and 1997. They suggest that an appreciation decreased sales, profitability and survival rate, while a depreciation strengthened them. Studying the Canadian agricultural implements industry, Tomlin (2010) also reports evidence that an appreciation reduces the survival probability of plants, especially the less productive ones. To test the predictions of the theoretical model, this paper use data on 237 Cana- dian manufacturing industries between 1997 and 2006 to study how industry-level labor 4 productivity growth interacts with exchange rate movements, concentration, and trade costs. The Canadian dollar experienced substantial movements in the period, allowing me to investigate the productivity response to a major appreciation. I find that growth rates of labor productivity, measured as value added per production worker, were on av- erage higher during the Canadian dollar appreciation between 2002 and 2006. Within the industries with a high trade-to-revenue ratio, the highly concentrated ones experi- enced greater growth in labor productivity. The empirical analysis controls for energy use growth, material use growth, R&D expenditure growth, productivity growth in corre- sponding US industries, industry fixed effects, and year fixed effects. My empirical work, based on industry-level data, complements the firm-level studies by building a model that links industry features to the size of productivity gain, and providing supporting evidence. Relative to the aforementioned papers, this paper makes two contributions. Theo- retically, it studies whether firms will improve productivity by adopting new technologies to counter the effect of appreciations, and what type of firms will invest more in new technologies. Empirically, the estimates in this paper suggest the productivity responses of highly-traded and concentrated Canadian manufacturing industries to the Canadian dollar appreciation between 2002 and 2006 were positive and significant. The next section lays out the basic modeling environment. Section 3 introduces the technological opportunity for home firms to improve productivity, and examines how home firms’ choices interact with an appreciation. Section 4 tests the model predictions on Canadian manufacturing data and section 5 concludes. 2 Basic Model Setup There are two countries, the home (h) and the foreign (f), and each has a representative household. The two households have the same given wealth 𝑊 and consume a continuum ofgoodsindexedby𝑖with𝑖 ∈ [0,1]. laborsuppliesinbothcountriesareperfectlyinelastic. 5 The home household’s problem is to maximize 2 1 𝛽𝑡−1 𝑙𝑜𝑔(𝐶 )𝑑𝑖 𝑖𝑡 𝑡=1 ∫0 ∑ subject to the life-time budget constraint 2 1 𝛽𝑡−1 𝑃 𝐶 𝑑𝑖 ≤ 𝑊 (1) 𝑖𝑡 𝑖𝑡 𝑡=1 ∫0 ∑ 𝐶 denotes the quantity of good 𝑖 and 𝑃 is its price. Similarly the foreign household 𝑖𝑡 𝑖𝑡 maximizes 2 1 𝛽𝑡−1 𝑙𝑜𝑔(𝐶∗)𝑑𝑖 𝑖𝑡 𝑡=1 ∫0 ∑ subject to the life-time budget constraint 2 1 𝛽𝑡−1 𝑃∗𝐶∗𝑑𝑖 ≤ 𝑊∗ (2) 𝑖𝑡 𝑖𝑡 𝑡=1 ∫0 ∑ Followingthe convention ininternational economics, the superscript ∗denotes variablesin the foreign country. The household preferences determine the demand functions for good 𝑖 in both countries 𝑊/(1+𝛽) 𝐶 = (3) 𝑖𝑡 𝑃 𝑖𝑡 𝑊/(1+𝛽) 𝐶∗ = (4) 𝑖𝑡 𝑃∗ 𝑖𝑡 where 𝑊/(1+𝛽) is normalized to be 1. For each good 𝑖, there are 𝑛 home firms and 𝑛 foreign firms who can produce it. I 𝑖 𝑖 will refer to these firms as firms in industry 𝑖. In both periods, all home firms are endowed with a constant marginal cost of 𝑐 = 𝑐 unit of labor and the foreign firms are endowed 𝑖ℎ𝑡 ℎ with a constant marginal cost of 𝑐 = 𝑐 . Thus in the model, home and foreign labor 𝑖𝑓𝑡 𝑓 productivities in any industry are 1 and 1 . labor is the only input and is not mobile 𝑐ℎ 𝑐𝑓 across countries. Every good is tradable, subject to an iceberg trade cost 𝜏 for good 𝑖, 𝑖 6 meaning that for each 𝜏 unit of good 𝑖 shipped to the other country only one unit will 𝑖 arrive. 𝜏 and𝑛 aredrawnfromthejointCDF𝐹(𝜏,𝑛)withsupport[1,∞)×[1,2,⋅⋅⋅ ,𝑛].1 𝑖 𝑖 The market structure within each industry is similar to that found in Brander and Krugman (1983). The home firms and foreign firms of industry 𝑖 produce using labor in theirrespectivecountries. However,theyarefreetoselltheirproductioninbothcountries. For a given period, the home and foreign firms of industry 𝑖 play a Cournot game in the home market to determine the quantities of good 𝑖 produced by each firm for the home market. Simultaneously, the same firms also compete in a Cournot game in the foreign market. Asmentionedbefore,inallperiodsboththehomeandforeignfirmfaceaniceberg trade cost 𝜏 when they sell in the non-native market. Figure 2.1 illustrates the market 𝑖 structure. The problem of home Firm 𝑗 of industry 𝑖 is 𝑗 𝑗 𝑗∗ 𝑗 𝑗∗ max Π = 𝜋 +𝑒 𝜋 +𝛽(𝜋 +𝑒 𝜋 ) (5) 𝑥𝑗 ,𝑥𝑗∗ ,𝑥𝑗 ,𝑥𝑗∗ 𝑖ℎ 𝑖ℎ1 1 𝑖ℎ1 𝑖ℎ2 2 𝑖ℎ2 𝑖ℎ1 𝑖ℎ1 𝑖ℎ2 𝑖ℎ2 𝑗 𝑗∗ where 𝑥 and 𝑥 are the quantities it produces for home and foreign markets in period 𝑖ℎ1 𝑖ℎ1 𝑗 𝑗∗ 𝑗 1, and 𝑥 and 𝑥 are the quantities for home and foreign markets in period 2. 𝜋 and 𝑖ℎ2 𝑖ℎ2 𝑖ℎ1 𝑗 𝑗∗ 𝑗∗ 𝜋 are profits from the home market in periods 1 and 2. 𝜋 and 𝜋 are profits from 𝑖ℎ2 𝑖ℎ1 𝑖ℎ2 the foreign market, measured in the foreign currency. 𝑒 and 𝑒 are the real exchange 1 2 rates in the two periods. They are defined as the price of one unit of real foreign money balance in terms of real home money balance, so a decrease in 𝑒 is a real appreciation of 𝑡 the home currency. The exchange rates are determined exogenously and known to all firms at the begin- ning of period 1. This assumption may appear surprising for economists familiar with the macroeconomic models of exchange rate determination. However, given that my interests are on the effect of exchange change rate on firms’ behavior and that the macroeconomic 1Inthismodel,thenumberoffirmsinanindustryisexogenouslygiven. Thistreatmentcanbeviewed as a simplification of the case where firms can enter and exit an industry freely and the number of firms in equilibrium is determined by the exogenous fixed cost of entry. 7 models of exchange rate have enjoyed limited empirical success, I argue that treating the exchange rate as exogenous is appropriate in this paper.2 In setting up the firm’s problem, I assume firms will discount future at the rate of time preference of the household, who is also the owner of the firms. In reality, firms may differ in the discount factor. For firms who place little value on future, there is very little incentive for them to adopt a technology that will bring a future benefit, holding other factors constant. The objective function also features no expectation operator, as I assume firms have perfect foresight of future. While expectation plays an important role in decision, I choose to suppress it here so as to focus discussion on how exchange rate lowers opportunity cost of adopting new technology. On empirical section, it is argued that firms in Canada have a good idea about the path of exchange rate since appreciations tend to be persistent and commodity prices are a good forecaster of exchange rate of the Canadian dollar. At the beginning of period 1 all firms observe each other’s marginal costs for all times. Then all firms in industry 𝑖 play a game to determine quantities of output in the fourmarkets(homeandforeignmarketsinperiod1and2). Thestrategyofhomefirm𝑗 in 𝑗 𝑗∗ 𝑗 𝑗∗ industry 𝑖 is the set of quantities 𝑥 ,𝑥 ,𝑥 ,𝑥 , and the strategy of foreign firm 𝑗 𝑖ℎ1 𝑖ℎ1 𝑖ℎ2 𝑖ℎ2 { 𝑗 𝑗∗ 𝑗 }𝑗∗ in industry 𝑖 is the set of quantities 𝑥 ,𝑥 ,𝑥 ,𝑥 . There are four subgames, one 𝑖𝑓1 𝑖𝑓1 𝑖𝑓2 𝑖𝑓2 { } for each market in each period. I focus on the subgame perfect equilibrium in which firms in industry 𝑖 of each country play symmetric strategies. I assume firms have to determine simultaneously the quantities in both markets in a period, hence in each period, the two subgames for the two markets are independent. In period 2, firms have to play a Nash equilibrium in the subgames. By the standard backward induction principle, they will also have to play a Nash equilibrium in the subgames in period 1. Thus all four subgames are independent, so the subgame perfect equilibrium involves firms playing the symmetric 2InTang(2008),Iendogenizetheexchangerateandtheincomeofthehouseholdsinatheoreticalmodel and find that firms have incentive to improve productivity when the exchange rate appreciates. 8 Nashequilibriumineachsubgame. Theoutputquantitiesineachsubgamearedetermined as the symmetric Nash equilibrium quantities in that subgame. We can calculate in the maximized total profit as the sum of maximized profits from each subgame. Normalizing home wage to be 1, the profit of the home firm 𝑗 of industry 𝑖 in the home market at time 𝑡 is 1 𝑗 𝑗 𝑗 𝜋 = (𝑃 −𝑐 )𝑥 = ( −𝑐 )𝑥 (6) 𝑖ℎ𝑡 𝑖𝑡 ℎ 𝑖ℎ𝑡 𝑛𝑖 𝑥𝑘 + 𝑛𝑖 𝑥𝑘 ℎ 𝑖ℎ𝑡 𝑘=1 𝑖ℎ𝑡 𝑘=1 𝑖𝑓𝑡 where 𝑥𝑘 and 𝑥𝑘 are the quantities∑of good 𝑖 p∑roduced by home firm 𝑘 and foreign 𝑖ℎ𝑡 𝑖𝑓𝑡 firm 𝑘 for the home market. The last equality follows from (3) and the market clearing condition 𝐶 = 𝑛𝑖 𝑥𝑘 + 𝑛𝑖 𝑥𝑘 . When the home firm 𝑗 chooses 𝑥𝑗 to maximize 𝑖𝑡 𝑘=1 𝑖ℎ𝑡 𝑗=1 𝑖𝑓𝑡 𝑖ℎ𝑡 (6), the first orde∑r condition ∑is 𝑥𝑘 + 𝑛𝑖 𝑥𝑘 𝑘∕=𝑗 𝑖ℎ𝑡 𝑘=1 𝑖𝑓𝑡 −𝑐 ≤ 0 (7) ( 𝑛𝑖 𝑥𝑘 + 𝑛𝑖 𝑥𝑘 )2 ℎ ∑𝑘=1 𝑖ℎ𝑡 ∑𝑘=1 𝑖𝑓𝑡 Similarly the profit of fore∑ign firm 𝑗 of∑industry 𝑖 in the home market at time 𝑡 is 1 𝑗 𝑗 𝑗 𝜋 = (𝑃 −𝑒 𝜏 𝑐 )𝑥 = ( −𝑒 𝜏 𝑐 )𝑥 (8) 𝑖𝑓𝑡 𝑖𝑡 𝑡 𝑖 𝑓 𝑖𝑓𝑡 𝑛𝑖 𝑥𝑘 + 𝑛𝑖 𝑥𝑘 𝑡 𝑖 𝑓 𝑖𝑓𝑡 𝑘=1 𝑖ℎ𝑡 𝑘=1 𝑖𝑓𝑡 𝑗 ∑ ∑ When the foreign firm 𝑗 chooses 𝑥 to maximize (8), the first order condition is 𝑖𝑓𝑡 𝑛𝑖 𝑥𝑘 + 𝑥𝑘 𝑘=1 𝑖ℎ𝑡 𝑘∕=𝑗 𝑖𝑓𝑡 −𝑒 𝜏 𝑐 ≤ 0. (9) ( 𝑛𝑖 𝑥𝑘 + 𝑛𝑖 𝑥𝑘 )2 𝑡 𝑖 𝑓 ∑𝑘=1 𝑖ℎ𝑡 ∑𝑘=1 𝑖𝑓𝑡 (7) and (9) implici∑tly define th∑e best responses functions of the home 𝑗 and foreign firm𝑗 toquantitiesproducedbyotherfirms. Combining (7)and(9)andimposingsymme- try among all home firms and symmetry among all foreign firms, we have the equilibrium relation between outputs of home and foreign firms 𝑛 𝑐 −(𝑛 −1)𝑒 𝜏 𝑐 𝑗 𝑖 ℎ 𝑖 𝑡 𝑖 𝑓 𝑗 𝑗 𝑥 = 𝑥 = 𝛼 (𝑡,𝑖)𝑥 (10) 𝑖𝑓𝑡 𝑛 𝑒 𝜏 𝑐 −(𝑛 −1)𝑐 𝑖ℎ𝑡 1 𝑖ℎ𝑡 𝑖 𝑡 𝑖 𝑓 𝑖 ℎ where 𝛼 (𝑡,𝑖) = 𝑛𝑖𝑐ℎ−(𝑛𝑖−1)𝑒𝑡𝜏𝑖𝑐𝑓. A careful examination of (7) suggests that if 𝑐 is large, 1 𝑛𝑖𝑒𝑡𝜏𝑖𝑐𝑓−(𝑛𝑖−1)𝑐ℎ ℎ then the home firms will produce zero quantities, and foreign firms will produce large 9