The Effect of the Exchange Rates on Investment in Mexican Manufacturing Industry *

Transcription

1 The Effec of he Exchange Raes on Invesmen in Mexican Manufacuring Indusry * Musafa Caglayan ** Universiy of Sheffield Rebeca I. Muñoz Torres *** Universiy of Warwick Ocober, 2007 ABSTRACT This paper, considering revenue and cos exposure channels, invesigaes he effecs of exchange rae behaviour on fixed capial invesmen in Mexican manufacuring secor over Our analysis also allows for he role of marke srucure and produc differences in ha relaionship. We find ha i) currency depreciaion has a posiive (negaive) effec on fixed invesmen hrough he expor (impor) channel; ii) exchange rae volailiy impacs mosly expor oriened secors; iii) he sensiiviy of invesmen o exchange rae movemens is sronger in non-durable goods secors and indusries wih low mark-up raios. JEL Classificaion: E22 Key Words: srucure Exchange Rae, volailiy, ARCH, invesmen, exernal exposure, marke * We would like o hank Sourafel Girma, Miguel Angel Mendoza, Saqib Jafarey and Rodney Srachan for helpful commens and suggesions. All remaining errors are ours. ** Deparmen of Economics, Universiy of Sheffield, 9 Mappin Sree, Sheffield S1 4DT, England. mcaglayan@shef.ac.uk *** Deparmen of Economics, Universiy of Warwick, Room S2.111b, Covenry, CV4 7AL, England. r.i.munoz-orres@warwick.ac.uk 1

2 1 Inroducion A large body of empirical research has been devoed o undersanding he implicaions of exchange rae movemens on economic aciviy. This lieraure suggess ha a change in he value of he currency may affec he real economy hrough differen channels. A real depreciaion, for example, may have expansionary effecs hrough increasing he operaing profis in he expor secor as well as increasing he cos of he impored goods favouring radable aciviies in he economy (see Nucci and Pozzolo, 2001). Conrarily, large currency depreciaions may deeriorae he firm s ne worh hrough he balance-shee-effec, where depreciaion of he currency may increase he burden of he dollar-denominaed deb (see Carranza e al., 2003; Aguiar, 2004; Parap and Urruia, 2004). Also, i has long been recognised ha invesmen is affeced by exchange rae uncerainy. 1 Several empirical sudies have shown a negaive relaion beween exchange rae volailiy and invesmen. In paricular, Goldberg (1993) and Campa and Goldberg (1995) using a sample of US manufacuring secors conclude ha exchange rae variabiliy has a week depressing effec on invesmen. Darby e al. (1999) find ha alhough exchange rae uncerainy has a significan and negaive impac on invesmen, his effec depends on he paricular indusry involved. Serven (2002) poins ou ha he negaive impac of real exchange rae uncerainy on invesmen is significanly larger in economies ha were highly open and in hose wih less developed financial sysems. More recenly, Harchaoui e al. (2005) show ha depreciaions had a posiive (negaive) effec on oal invesmen when he exchange rae volailiy is low (high), and Fuenes (2006) provides empirical evidence for Chile on he negaive relaionship beween exchange rae volailiy and invesmen under irreversibiliy. This paper invesigaes he linkages beween he level and he volailiy of he exchange rae and firm invesmen behaviour in Mexico using annual manufacuring secoral daa for he period In our invesigaion we consider if he marke 1 There are several compeing heories on he impac of uncerainy on privae invesmen. Several models emphasize a posiive impac from uncerainy on invesmen; see Harman (1972) and Abel (1983). Recen research, following Dixi and Pindyck (1994), highligh a negaive relaionship beween uncerainy and invesmen. See Carruh e al. (2000) for a survey. 2 The Mexican peso was floaed in Also, srucural changes ha were implemened in he Mexican economy lead her o he Norh American Free Trade Agreemen (NAFTA) in These changes ransformed Mexico s economy by creaing an ouward looking and an expor oriened economy (see Messmacher, 2002). 2

3 srucure, exernal exposure measures and produc specific differences affec he linkages beween invesmen and exchange rae movemens. Our resuls can be summarized as follows: we show ha exchange rae depreciaion has a posiive (negaive) effec on invesmen hrough he expor (impor) channel. We also find ha exchange rae volailiy impacs mosly expor oriened secors. Finally, we presen evidence ha he sensiiviy of invesmen o exchange rae movemens is sronger in non-durable goods secors and indusries wih low mark-up raios. The res of he paper is organised as follows. Secion 2 inroduces he empirical specificaion and describes he daa. Secion 3 discusses he empirical resuls and secion 4 concludes he paper. 2 Empirical Specificaion and Daa 2.1 The model We use a simple model, iniially explored by Campa and Goldberg (1995, 1999), o undersand he impac of exchange rae movemens on secoral invesmen in Mexico. Our analysis uilizes he indusry srucure and he exernal exposure of he firms. The model akes he following form: I I s rer vol ir β ( + ε (1) i i i i 1 i 1 = i 0 + β1 + β ex i 2 + β 3 ) + ( β 4 + β 5 ex ) + β 6 1 s 1 rer 2 vol 2 ir 1 i I i i where represens gross domesic invesmen in manufacuring secor i; represens secor sales; rer is he real exchange rae defined as pesos per US dollar; vol sands s for volailiy of he exchange rae and; ir is he annual ineres rae. The level and he volailiy of exchange rae are allowed o ener ino he model in conjuncion wih i exernal exposure, ex, which may have changed significanly over ime. Any such differenial effecs can be deeced hrough he significance of β3 and β 5. Furhermore, we inroduce once lagged level and volailiy of he exchange rae as i has been widely acceped in he lieraure ha movemens in exchange raes akes up o a year o affec he firm s behaviour (see, for insance, Baum e al., 2004). This sraegy is also useful o avoid poenial bias ha may be induced by heir possible correlaion wih curren invesmen (see Campa and Golberg, 1995). 3

4 We esimae he models using he wo-sage leas square (2SLS) mehod, which provides consisen parameer esimaes in he presence of endogeneiy. This is mainly due o he fac ha curren ineres rae affecs he overall manufacuring invesmen as well as invesmen in paricular indusries and, herefore, may be correlaed wih he error erm i ε. The insrumen se includes lagged values of he ineres rae variable. 3 Finally, all equaions ake ino accoun he presence of firmspecific fixed effecs, λ i and he repored sandard errors are correced for heeroskedasiciy following Whie (1980). 2.2 Daa This sudy uses a balanced annual panel daase which provides consisen informaion on invesmen, sales, and price index on Mexican manufacuring indusries over he period Mos of he daa are obained from he Annual Indusry Survey (AIS) conduced by he Naional Insiue of Saisics, Geography and Compuaional Science (INEGI) and he Cenral Bank of Mexico. The survey covers on average 6,224 firms grouped ino 205 hree-digi level manufacuring secors. The AIS is used as an inpu for he Indusrial Census and annual GDP calculaions and he firms in he daase accouns for almos 80 percen of he value added in manufacuring. All variables are repored in real pesos wih 2003 serving as he base year and he producer price index serving as deflaor. Deails of he daa and sources are given in Appendix A. We use he hree-digi secoral daa since INEGI does no provide he firm-level daa informaion. In he following sub-secions, we define he measures of exernal exposure, marke power and exchange rae uncerainy Exernal Exposure Measures Movemens in exchange raes can noably influence he performance of domesic manufacuring indusries. However, he firm s response generally depends on a variey of facors such as he reliance on impored inpus and he share of foreign sales in oal sales. 4 Using informaion on impored inpus and foreign sales repored in he AIS, we compue wo exernal exposure measures (ex). The expor share denoed by 3 We have also experimened wih adding lagged values of sales and exchange raes as insrumens and our resuls were unchanged. 4 Anoher used indicaor of an indusry s exposure o movemens in he exchange rae is he exen o which manufacurers compee wih foreign producers in he domesic marke. For furher discussion see Campa and Goldberg (1997) and Nucci and Pozzolo (2001). 4

5 Xm is calculaed as he raio of expors o oal revenues and he impored-inpu share in a manufacuring secor is approximaed by he raio of impored o oal inpus (Mm). Indusries wih high expor shares are likely o have oal revenues ha are more sensiive o movemens in he exchange rae han producers wih low expor shares. Similarly, a manufacurer ha relies grealy on impored inpus will possibly be more exposed o exchange rae movemens hrough coss han a producer ha relies mosly on domesic inpus (see Campa and Goldberg, 1997). 5 Table 1: Exernal Exposure by manufacuring secors * Expor-share Impored-- inpu share Expor-share Impored-- inpu share Expor-share Impored-- inpu share Manufacuring non-durables 1. Food, Beverages and Tobacco Texile, Apparel and Leaher Indusries Lumber and Wood Producs. Including Furniure Paper and Allied Producs, Prining and Publishing Chemicals, Peroleum, Coal, Rubber and Plasic Producs Manufacuring durables 6. Non-Meallic Mineral Producs, excep Peroleum and Coal Producs Basic Meal Indusries Meal Producs, Machinery and Equipmen. Including Surgical and Precision Insrumens Oher Manufacuring Indusries Toal Manufacuring (*) Mean values over he period Table 1 presens summary informaion for he exernal exposure measures, grouped ino 9-manufacuring indusries and divided by durable versus non-durable manufacuring goods. 6 On average, he indusries mos grealy dependen on expors are meal producs, machinery and equipmen, basic meal indusries, lumber and wood producs and non-meallic mineral producs. The indusries relying mos on impored-inpus ino producion are oher manufacuring indusries followed by a group of non-durable secors, including chemicals, peroleum and plasic producs, paper and allied producs and exiles and leaher indusries. Manufacuring nondurable goods, as a group, have more reliance on impored-inpus ino producion as compared wih manufacuring durable goods and herefore his group of indusries are 5 The erms expor and revenue channel are used inerchangeably in his paper. Likewise, impor and cos channel are aken as synonymous. 6 The breakdown of manufacuring secors ino durables and non-durables goods is based on earlier work by Campa and Golberg (1995). 5

6 expeced o experience higher sensiiviy o movemens in he exchange rae hrough he cos side of is balance shee Measuring Marke Power In our analysis, we explore he role of he compeiive srucure of he indusry in deermining he impac of exchange raes on invesmen. To measure he degree of he secor-specific marke power, a ime varying measure of profis margins on uni price (PCM) is consruced following Domowiz e al. (1986): PCM Value of sales + Δ Invenories Payroll Cos of maerials = (2) Value of sales + Δ Invenories which given he census definiion of value added is idenical o: Value added Payroll markup = Value added + Cos of maerials Table 2 presens summary informaion for a subse of years on our secor specific annual mark-up raios. 7 We observe variabiliy across ime and across manufacuring indusry. In paricular, our resuls show relaively higher variabiliy of he mark-up raios in manufacuring durables and higher values can generally be observed a he end of our sample for all indusries. For example, he non-meallic mineral producs indusry had a mark-up of in 1994, in 1997 and in Among he mos variable indusries are exile, apparel and leaher indusries, oher manufacuring indusries and meal producs, machinery and equipmen. 7 The daa used o consruc his measure is based on our second definiion of mark-up and is aken from he AIS described earlier. 8 A mark-up value of means, ha he indusry ses a price of approximaely 38.7% above is marginal cos. 6

7 Table 2: Indusry Mark-up Raios * Coef. of variaion Manufacuring non-durables 1. Food, Beverages and Tobacco Texile, Apparel and Leaher Indusries Lumber and Wood Producs. Including Furniure Paper and Allied Producs, Prining and Publishing Chemicals, Peroleum, Coal, Rubber and Plasic Producs Manufacuring durables 6. Non-Meallic Mineral Producs, excep Peroleum and Coal Producs Basic Meal Indusries Meal Producs, Machinery and Equipmen. Including Surgical and Precision Insrumens Oher Manufacuring Indusries (*) Mean values over he period Measuring Exchange Rae Volailiy Researchers have proposed a number of ways o generae a proxy for exchange rae volailiy. Noneheless, as Baum e al. (2004) poin ou he choice of a paricular specificaion o generae a proxy for uncerainy may have a considerable impac on he empirical findings. A common approach is o use eiher a moving average sandard deviaion or he coefficien of variaion of he pas monhly exchange raes as a measure of exchange rae volailiy (see Campa, 1993; Campa and Goldberg, 1995; Amuedo and Pozo, 2001 and Harchaoui e al., 2005). However, such a measure gives rise o subsanial serial correlaion in he summary measure. In his sudy, we use GARCH mehodology o proxy for exchange rae volailiy. To compue a proxy for real exchange rae uncerainy, we use monhly daa over he 1988 o 2002 period and esimae an ARCH specificaion for he firs differenced real exchange rae series. We find ou ha he mean equaion ( Δx ) exhibis an AR(2) process and he condiional variance (h) is bes described as an ARCH (1). The resuls from hese esimaions along wih he diagnosics are provided in Appendix B. 9 9 Noe ha we use 12-monh averages of he esimaed condiional variances as a proxy o mach he annual daa ha we use in our final esimaion. 7

8 3 Empirical Resuls In wha follows below we presen empirical resuls on firms invesmen behaviour considering he role of exernal exposure measures, produc ype and he marke srucure Invesmen and he Role of Exposure Measures Table 3 presens he wo-sage leas square regression resuls, where we insrumen he ineres rae up o wo lags, for equaion (1) and allow for fixed effecs. 10 The firs column conains no ineraced erms whereas he remaining columns allow for he presence of ineracions. Those ineracions in he model basically capure he poenial impac of he level and he volailiy of he exchange rae on firm s invesmen behaviour hrough he firm s degree of exernal exposure (i.e. impor versus expor orienaion of he firm). For an expor oriened firm, we expec o find ha exchange rae depreciaion would promoe invesmen as expors increase when he currency depreciaes. This implies β 3 in equaion (1) o ake a posiive sign. For he same ype of firm, an increase in exchange rae volailiy will improve expeced profis enhancing new invesmens. Hence, β 5 is expeced o be posiive. For an impored-inpu oriened firm we expec he signs o swich. Depreciaion will reduce invesmen due o increasing cos of impored goods (β 3 negaive) and expeced profis decline in relaion o volailiy if here is a high reliance on impored inpus. This will in urn imply ha β 5 will be negaive. 11,12 Column 1 provides esimaes of he sandard model where we exclude he level and he volailiy ineracions wih he exernal exposure measure. In line wih earlier findings repored in he lieraure, we find ha invesmen is always posiively and significanly correlaed wih sales as capured by he coefficien β 1. Furhermore, currency depreciaions are associaed wih reducions in invesmen (β 2 coefficien) and volailiy has a posiive impac on our dependen variable (β 4 coefficien), ye neiher coefficien is significan. Similar o he findings repored by Campa and Golberg (1995) for US, he ineres rae has a posiive and significan impac on 10 Summary saisics of he main variables used in he regression analysis are presened in Appendix C. 11 Also see Campa and Goldberg (1995). 12 We should poin ou ha he empirical impac of exchange rae uncerainy on invesmen is an unresolved issue, and i is bes o le he daa speak. 8

9 invesmen. 13 Finally, he Sargan es saisic of overidenifying resricions suggess ha he insrumen se is valid. Nex, we inroduce he ineracion erms ino he model; measures of exernal i exposure, ex, ineraced wih he level and he volailiy of he exchange rae. The resuls from he esimaed coefficiens on he ineraced erms uncover differen and significan responses across manufacuring secors. This finding jusifies our approach and reveals he imporance of using informaion on firms exernal orienaion in undersanding heir invesmen behaviour. Table 3: Two-Sage Leas Squares Resuls from he Pooled Sample Non-ex Xm Mm (1) (2) (3) sales (β 1 ) 1.091* 1.105* 1.100* rer (β 2 ) rer*ex (β 3 ) * vol (β 4 ) * vol*ex (β 5 ) * irae (β 6 ) 0.166* 0.165* 0.165* Overidenificaion es (p-value) No. of observaions: 1,095 Robus sandard errors are repored beneah parameer esimaes * Significan a he 5% level ** Significan a he 10% level Column 2 repors resuls on he expor oriened secors, whereas column 3 concenraes on ha of he impor orienaed secors. 14 From column 2 we see ha he coefficiens associaed wih he ineracion erms, β 3 and β 5 are significan. The 13 Neoclassical models of invesmen suppor he general conclusion ha he effec of sales has he dominan effec on invesmen wih user cos having a minor role (see Chirinko, 1993). Taylor (1999) also aemps o explain he weak link beween ineres raes and invesmen by considering he effecs of differen coss of various ypes of finance available o he firm or he presence of irreversibiliy and uncerainy. 14 To consider he possible influence of he marke srucure under he manufacuring secor operaes, our exernal measures have been adjused yearly by secoral mark-up as follows: ex i *δ, where δ = (1- markup) ; δ=1 describes a perfecly compeiive marke and δ=0 denoes a monopolisic indusry. See Campa and Golberg, 1995 and Nucci and Pozzolo, 2001 for furher discussion. 9

10 ineracion erm β 3 is significan and posiive revealing ha currency depreciaion will have a posiive impac on he invesmen behaviour of he firm hrough he revenue channel. I seems ha depreciaion inducing higher expors and higher firm revenue promoe more invesmen, as he lieraure has widely documened. Ineresingly, volailiy on is own, β 4, has a significan posiive effec on invesmen as some early neoclassical models emphasised. 15 Conrarily, he ineracion beween uncerainy and expor exposure measure coefficien, β 5, akes a negaive sign implying ha as uncerainy increases, invesmen of an expor oriened firm will be hampered. This may be explained by he opion heory ha delaying an invesmen projec may give more accurae view of marke condiions. Hence, exchange rae volailiy, by creaing an uncerain economic environmen, deers invesmen from expor oriened manufacuring secors. 16 In column 3, resuls reveal ha neiher he level nor he volailiy measures have any effec on he invesmen behaviour of he impor oriened firm. Only changes in sales and ineres raes seem o affec such firms. Perhaps, he managers of hese firms mainly concenrae on he inernal marke and as long as sales are on he rise hey keep on invesing and use he impored inermediae goods o saisfy he inernal demand despie movemens on he exchange rae Invesmen and he Role of Durable vs. Non-Durable Goods We hen spli he sample ino wo caegories based on produc durabiliy while considering he degree of firm s exernal exposure. This spli reveals addiional findings as depiced in Table 4. Columns 1 and 2 presen he esimaes for he durable goods secor and columns 3 and 4 lay ou hose for he non-durables secor. For durables goods secor i urns ou ha variaions in he exchange rae do no have any significan impac on invesmen a all. This finding is rue irregardless of he exernal orienaion of he firms. Conrarily, boh he level and he volailiy of exchange rae movemens have significan effecs on invesmen behaviour for non-durable firms (las wo columns in able 4). The level effec of exchange rae is negaive for boh impor and expor oriened firms. Perhaps, due o he naure of he produc, he 15 Harman (1972) and Abel (1983) shows ha if one assumes perfec compeiion and consan reurns o scale as well as symmeric adjusmen coss, an increase in uncerainy may also rise he value of a marginal uni of capial and hence he incenive o inves. 16 Dixi and Pindyck (1994), highligh he imporance of he iming of invesmen decisions. They argued ha irreversibiliy may lead o he posponing of invesmen decisions and herefore makes invesmen adjusmen cos asymmeric. These models assume imperfec compeiion and/or decreasing reurns o scale. 10

11 invesmen behaviour of an expor oriened firm is negaively affeced as he firm looses revenue afer devaluaion and reduces is invesmen. Likewise, an impor oriened firm reduces is invesmens for i ges coslier o run he business afer currency devaluaion. Conrarily, he ineracion erm is posiive and significan for he expor oriened firm. Hence, he oal impac for he expor oriened firm of currency devaluaion seems o be posiive as he heory suggess. When we concenrae on he effecs of exchange rae volailiy on invesmen behavior, i seems ha he expor oriened firm is negaively effeced from volailiy; while he own effec of volailiy is small posiive and significan, he ineracion erm is negaive and significan. The conracionary effec of volailiy on invesmen can be explained by he larger reliance from hese manufacuring indusries on impored inpus ino producion as our prior analysis abou he exernal exposure from manufacuring secors sugges. 17 In conras, for he impor oriened firm, volailiy has an overall posiive effec on invesmen. As we menioned earlier, impor oriened firms possibly compee wih domesic firms and hey increase invesmen o saisfy he inernal demand as long as hey have a srong sales record. Table 4: Two-Sage Leas Squares Resuls by Indusry Group Manufacuring durables Manufacuring non-durables Xm Mm Xm Mm (1) (2) (3) (4) sales (β 1 ) 0.815* 0.796* 1.590* 1.605* rer (β 2 ) ** ** rer*ex (β 3 ) ** vol (β 4 ) * 0.448** vol*ex (β 5 ) * irae (β 6 ) * 0.289* Overidenificaion es (p-value) No. of observaions: 451 (high markup secors); 644 (low markup secors) Robus sandard errors are repored beneah parameer esimaes * Significan a he 5% level ** Significan a he 10% level 17 Empirical sudies focusing on developing counries have found a negaive relaion beween real exchange rae volailiy and expors. See Caballero and Corbo (1991), Serven (2002) and Grier and Smallwood (2007) among ohers. 11

12 3.1.3 Invesmen and he Role of Marke Srucure Nex we explore he role of he marke srucure on he effecs of exchange rae movemens on domesic invesmen. To do so, we spli he sample by he mean markup value ino high and low mark-up manufacuring secors. The aim is o invesigae empirically he role of marke power in deermining he relaionship beween firm invesmen behaviour and he movemens in exchange rae. The resuls in Table 5 reveal differences on he effecs of exchange raes movemens on invesmen when he marke srucure is considered. Columns 1 and 2 show he resuls in secors wih high mark-up and he las wo columns from able 5 presen he esimaes in low mark-up secors. In he case of high mark-up firms, he level of exchange rae is no relevan in deermining he invesmen behaviour of he firm regardless of he orienaion of he firm. However, we see ha in he aggregae exchange rae volailiy effecs he invesmen behaviour of he expor oriened firms negaively; where he own effec is small and posiive; he ineracion erm is negaive and large. Conrarily, in column 3 we observe ha he expor oriened low mark-up firm s invesmen behaviour is more responsive o changes in he exchange rae and is volailiy as boh ineracion erms are significan. Here, depreciaion encourages invesmen hrough he revenue channel (β 3 coefficien) and volailiy has a negaive effec on i (β 5 coefficien). The impored inpu channel, see column 4, does no give us significan resuls. Overall, hese resuls sugges ha low mark-up indusries wih high expor shares are subjeced more o he exchange rae effecs in invesmen Campa and Golberg (1995, 1999), Nucci and Pozzolo (2001) and more recenly Harchahoui e al. (2005) have found similar resuls in which he effec of he real exchange rae on invesmen is sronger for indusries or firms wih low price-cos margin han for firms wih high mark-up. 12

13 Table 5: Two-Sage Leas Squares Resuls by Marke Srucure High Markup secors Low Markup secors Xm Mm Xm Mm (1) (2) (3) (4) sales (β 1 ) * 1.459* rer (β 2 ) rer*ex (β 3 ) * vol (β 4 ) 0.539** vol*ex (β 5 ) ** ** irae (β 6 ) ** 0.167** Overidenificaion es (p-value) No. of observaions: 451 (high markup secors); 644 (low markup secors) Robus sandard errors are repored beneah parameer esimaes * Significan a he 5% level ** Significan a he 10% level 3.2 Exernal Exposures, marke srucure and indusry group Given our findings so far, we inroduce boh exposure measures in he same model and sudy he behaviour of high and low mark-up firms as well as firms ha produce durable and nondurable producs. Table 6 depics hese resuls. On he whole, he sign and significance of all coefficiens are similar o hose resuls we have provided earlier. In almos all cases, sales growh induces higher invesmen. Depreciaion on is own significanly reduces invesmen (β 2 <0) of non-durables secors only. Oherwise, is effecs are insignifican. We observe ha volailiy has a posiive impac (β 5 >0) on invesmen of non-durables secor on is own a he 5% significance level. However, when we consider he ineracions beween he exchange rae and he firm exposure measure, we find ha expor oriened secors significanly increase heir invesmen (β 3 >0) when he domesic currency depreciaes. In he case of impor oriened secors, hey significanly reduce heir invesmen (β 4 <0) when he domesic currency depreciaes. We also find ha volailiy in conjuncion wih he measure of exernal orienaion end o significanly reduce invesmen of expor oriened secors (β 6 <0) and o increase impor oriened secors (β 7 >0). 13

14 Table 6: Two-Sage Leas Squares Resuls using boh Expor and Impored-Inpu Exposures Full sample High Markup Low Markup Durables Non Durables (1) (2) (3) (4) (5) sales (β 1 ) 1.129* * 0.852* 1.599* rer (β 2 ) ** rer*xm (β 3 ) 3.956* * ** rer*mm (β 4 ) vol (β 5 ) * vol*xm (β 6 ) * * * * vol*mm (β 7 ) 1.918** ** irae (β 8 ) 0.164* ** * Overidenificaion es (p-value) Wald es of join significance (p-value) No. of observaions Robus sandard errors are repored beneah parameer esimaes * Significan a he 5% level ** Significan a he 10% level 4. Conclusions This paper invesigaes he effecs of he level and he volailiy of exchange rae on invesmen using a sample of manufacuring secor daa obained from Mexico over he period Our resuls emphasize he role of firm s exernal exposure, he marke srucure and produc characerisics in ha relaionship. We find ha depreciaion affecs invesmen posiively hrough he expor channel and depresses expeced profis if here is a high reliance on impored inpus. When we urn our aenion o exchange rae volailiy, we find ha volailiy has a negaive effec on invesmen for firms ha are expor oriened. We also show ha exchange rae volailiy has a simulaing effec on firm s invesmen hrough he impor channel. Furhermore, our resuls imply ha firms ha operae in low mark-up indusries and ha produce non-durable producs end o be more sensiive o boh he level and volailiy of exchange raes. 14

15 I may be useful o noe some limiaions and possible exensions associaed wih he curren sudy. Currenly, here is only a handful of sudies ha use secor specific daa. Hence, i would be useful o invesigae he linkages beween secoral invesmen and exchange rae movemens using daa from oher counries. We also hink ha firm level analysis migh yield furher ineresing insighs. For insance, i would be quie fruiful o sudy he role of financial fricions along wih exchange rae movemens in deermining he invesmen behaviour of firms implemening a similar model we presen here. 15

16 Appendix A Table A1. Descripion of daa Variable Descripion Source Gross Invesmen (I i ) The sum of gross fixed invesmen and INEGI: variaions in invenories. Gross Fixed Invesmen is defined as purchases minus sales of used and new asses plus improvemens on exising asses plus capial produced for own use. Toal Sales (s i ): Toal Inpus Real Exchange Rae (rer ) Volailiy (vol ) Annual sales per indusry or secor level. The survey provides informaion on foreign and naional sales. Toal inpus per indusry or secor. The survey provides informaion on impored inpus ino producion. The real exchange rae index measures he value of he peso agains more han one hundred rading parners of Mexico. A rise in he index indicaes a real depreciaion of he domesic currency. We compue a proxy for real exchange rae uncerainy using monhly daa and GARCH models. See Appendix B for deails. INEGI: INEGI: Banco de México: Banco de México: Ineres Rae The annual ineres rae (Treasury Bill CETES). Banco de México: Price Index The producer price index measured a he end of he corresponding year is used as deflaor. Banco de México: All variables considered are in erms of 2003 prices. Only secors wih posiive values for invesmen are considered in our sample. 16

17 Appendix B Table B1. Exchange Rae Volailiy ARCH Model* Δx = Δx 0.395Δx + e h (0.003) (0.076) 2 = e 1 (0.000) (0.224) Saisics for he Sandardised Residuals Q 2 -saisic(6) = 2.74 Skewness = 0.33 Kurosis = 6.50 ARCH-LM(6) = 0.44 (0.84) N= (0.073) (*) Bollerslev - Wooldrige robus sandard errors are in parenhesis excep in he ARCH-LM es where he p value is repored. The esimaed model for he change in he log of he real exchange rae is: Δx = β + e 0 + β 2Δx 1 + β 3Δx 2 2 where, e I 1 ~ N(0,h ) and h α + α e denoes he condiional variance. The resuls for he = ARCH (1) specificaion show a mean revering variance process. The value of he ARCH parameer (α1=0.556) indicaes some persisence of volailiy shocks. Diagnosic saisics shows ha he variance equaion has been correcly specified. Specifically, Q-saisics and he Lagrange Muliplier (LM) es ensure no remaining ARCH effecs. Figure B1. Real Exchange Rae and Volailiy in Mexico: Volailiy M01 RER index 1995M M M M M M M M01 Volailiy RER 17

18 Appendix C Table C1: Summary saisics Variable Mean Sd. Dev. Min Max Invesmen rae * Ineres rae Sales rae Exchange rae (*)Esablishmens wih zero or negaive values were eliminaed from he sample. In addiion, secors ha repored values for he invesmen rae wihin 5 sandard deviaions from he sample mean were discarded. 18

19 References Abel, A.B., 1983, Opimal Invesmen under Uncerainy, American Economic Review, 73, pp Aguiar, M., 2004, Invesmen, Devaluaion, and Foreign Currency Exposure: The Case of Mexico, Federal Reserve Bank of Boson, mimeo Amuedo, D.C. and S. Pozo, 2001, Foreign Exchange Raes and Foreign Direc Invesmen in he Unied Saes, The Inernaional Trade Journal, XV(3), pp Baum, C., M. Caglayan and N. Ozkam, 2004, Nonlinear Effecs of Exchange Rae Volailiy on he Volume of Bilaeral Expors, Journal of Applied Economerics, 19(1), pp 1-23 Caballero, R. J. and V. Corbo, 1991, The Effec of Real Exchange Rae Uncerainy on Expors: Empirical Evidence, The World Bank Economic Review, 3(2), pp Campa, J. M., 1993, Enry by Foreign Firms in he Unied Saes under Exchange Rae Uncerainy, Review of Economics and Saisics, 75(4), pp Campa, J. and L. Golberg, 1995, Invesmen in Manufacuring, Exchange Raes and Exernal Exposure, Journal of Inernaional Economics, 38, pp Campa, J. and L. Golberg, 1997, The Evolving Exernal Orienaion of Manufacuring: A Profile of Four Counries, FRBNY Economic Policy Review, July Campa, J. and L. Golberg, 1999, Invesmen, Pass-hrough, and Exchange Raes: A Cross-Counry Comparison, Inernaional Economic Review, 40(2), pp Carranza, L., J.M. Cayo and J.E. Galdon Sanchez, 2003, Exchange Rae Volailiy and Economic Performance in Peru: A Firm Level Analysis, Faculad de Ciencias Economicas y Empresariales, Working Paper 12/03, Universiy of Navarra Carruh, A., A. Dickerson and A. Henley, 2000, Wha do we know abou Invesmen under Uncerainy?, Journal of Economic Surveys, 14(2), pp Chirinko, R. S., 1993, Business Fixed Invesmen Spending: Modelling Sraegies, Empirical Resuls, and Policy Implicaions, Journal of Economic Lieraure, 31, pp Darby, J., A. H. Halle, J. Ireland and L. Piscielli, 1999, The Impac of Exchange Rae Uncerainy on he Level of Invesmen, The Economic Journal, 109, pp Dixi, A. K. and R. S. Pyndick, 1994, Invesmen under Uncerainy, Princenon: Princenon Universiy Press Domowiz, I., G. Hubbard and B. Peersen, 1986, Business Cycles and he Relaionship beween Concenraion and Profis Margins, Rand Journal of Economics, 17, pp

20 Grier, K. and A. D. Smallwood, 2007, Uncerainy and Expor Performance: Evidence from 18 Counries, Journal of Money, Credi and Banking, 39(4), pp Fuenes, O. M., 2006, Exchange Rae Volailiy and Invesmen: Evidence a he Plan-Level, Boson Universiy, mimeo Goldberg, L., 1993, Exchange Raes and Invesmen in Unied Saes Indusry, Review of Economics and Saisics, 75(4), pp Harchaoui, T., F. Tarkhani and T. Yuan, 2005, The Effecs of he Exchange Rae on Invesmen: Evidence from Canadian Manufacuring Indusries, Bank of Canada, Working Paper Harman, R., 1972, The effecs of Price and Cos Uncerainy on Invesmen, Journal of Economic Theory, 5, pp Messmacher, M., 2002, Sabilizaion Policies in Mexico: , Banco de Mexico, mimeo Nucci, F. and A. F. Pozzolo, 2001, Invesmen and he Exchange Rae: An Analysis wih Firm-level Daa, European Economic Review, 45, pp Parap, S. and C. Urruia, 2004, Firm Dynamics, Invesmen, and Deb Porfolio: Balance Shee Effecs on he Mexican Crisis of 1994, Journal of Developmen Economics, 75, pp Serven, L., 2002, Real Exchange Rae Uncerainy and Privae Invesmen in Developing Counries, World Bank, Policy Research Working Paper 2823 Taylor, M., 1999, Real Ineres Raes and Macroeconomic Aciviy, Oxford Review of Economic Policy, 15(2), pp Whie, H., 1980, A Heeroskedasiciy-consisen Covariance Marix Esimaor and Direc Tes for Heeroskedasiciy, Economerica, 48, pp