Journal of International Economics 85 (2011) 317–328

Contents lists available at ScienceDirect

Journal of International Economics

j ourna l homepage: www.e lsev ie r.com/ locate / j i e

Financial liberalization, structural change, and real exchange rate appreciations☆

Felipe Meza ⁎, Carlos UrrutiaCentro de Investigación Económica, ITAM, Mexico

☆ This paper has benefited from comments by EnriqueLipschitz, Rudolfs Bems, Daniel Chiquiar and partInternational Monetary Fund and Banco de Mexico. Tcomments by the editor and two anonymous refereespecially Sangeeta Pratap for giving us access to her dawritten while Urrutia was visiting the IMF Institute.Asociacion Mexicana de Cultura for financial support. Thresearch grant No. 81825 is thankfully acknowledged. A⁎ Corresponding author.

E-mail addresses: felipe.meza@itam.mx (F. Meza), cu

0022-1996/$ – see front matter © 2011 Elsevier B.V. Aldoi:10.1016/j.jinteco.2011.06.001

a b s t r a c t

a r t i c l e i n f o

Article history:Received 15 April 2009Received in revised form 14 March 2011Accepted 10 June 2011Available online 29 June 2011

JEL classification:F3F4

Keywords:Real exchange rateFinancial liberalizationStructural transformation

The last twenty years have witnessed periods of sustained appreciations of the real exchange rate in emergingeconomies. The case of Mexico between 1988 and 2002 is representative of several episodes in Latin Americaand Central and Eastern Europe in which countries opening to capital flows experienced large appreciationsaccompanied by a significant reallocation of workers towards the non-tradable sector. We account for thesefacts using a two sector dynamic general equilibrium model of a small open economy with frictions to laborreallocation and two driving forces: (i) A decline in the cost of borrowing in foreign markets, and (ii)differential productivity growth across sectors. These twomechanisms account together for 60% of the declinein the domestic relative price of tradables in Mexico and for a large fraction of the observed reallocation oflabor across sectors. The decline in the interest rate faced by Mexico in international markets is quantitativelythe most important channel. Our results are robust to the inclusion of terms of trade into the model.

Mendoza, Behzad Diba, Leslieicipants in seminars at thehe paper also benefited fromes. We would like to thanktabases. The paper was partlyWe would like to thank thee support of CONACYT throughll errors are our own.

rrutia@itam.mx (C. Urrutia).

l rights reserved.

© 2011 Elsevier B.V. All rights reserved.

1. Introduction

Between 1988 and 2002 Mexico experienced a substantialappreciation of its real exchange rate (RER). In spite of the 1995crisis, in which the RER briefly depreciated as a result of a sudden stopof loans from abroad, the trend in the whole period shows a 40%appreciation. Similar episodes of RER appreciation have beenobserved in other Latin American and Central and Eastern Europeancountries, coinciding with a period of financial liberalization, capitalinflows and trade deficits. In Mexico, starting in 1988, the opening ofthe capital account increased the ability to borrow in internationalmarkets. The interest rate for loans to Mexico, including the countryrisk premium, fell from about 15% in 1990 to less than 5% in 2002, witha short run jump during the 1995 crisis.

Looking at the Mexican data for the period, we document thefollowing stylized facts: (i) 78% of the RER appreciation correspondsto a decline in the domestic relative price of tradable goods, measuredas the GDP deflator in the tradable goods sectors divided by theoverall GDP deflator; (ii) changes in relative outputs and relativewages across sectors are an important component of the story, but

changes in factor income shares are not; (iii) growth accounting foreach sector reveals an increase in measured TFP in the tradable sector,while TFP remains stagnant in the non-tradable sector; and (iv) thereis a substantial reallocation of resources (capital and labor) from thetradable sector towards the non-tradable sector.

The case of Mexico is representative of a more general trend inemerging economies. We document several episodes of RER appreci-ations in Latin American and European countries following an openingto foreign capital flows. As in Mexico, we observe in these countrieslarge and sustained RER appreciations accompanied by a massivereallocation of labor towards the non-tradable sector. In some episodes,we canalso identify adecline in the cost of foreignborrowingdrivenbyareduction in the country-specific interest rate premium.

In this paper, we use a structural model to analyze the relationbetween the RER appreciation and different shocks affecting theeconomy. Our main objective is to provide a quantitative assessmentof the decline in the cost of foreign borrowing as a mechanism toexplain the stylized facts (i) and (iv) above in a model consistent withfacts (ii) and (iii). For this, we build a two sector, deterministic,dynamic general equilibriummodel of a small open economy that canaccommodate both external interest rate shocks and sectoral TFPchanges. The model is real, abstracting from a monetary side, andconstrained-efficient, in the sense that given the adjustment costs tocapital accumulation and labor mobility the competitive equilibriumis Pareto-optimal. This distinguishes our analysis from alternativestories based on price rigidities, imperfect competition, and so on.

We calibrate the model to some aggregate statistics for theMexicaneconomy, feed it with the exogenous paths for the international interestrate forMexico andmeasured TFP in each sector, and obtain time seriesfor relative prices and other variables of interest. Ourmodel accounts for

80

100

120

318 F. Meza, C. Urrutia / Journal of International Economics 85 (2011) 317–328

60% of the change in the domestic relative price of tradable goodsobserved in the data. We also find that the interest rate channel isquantitatively themost important of the two, accounting on its own fortwo-thirds of the appreciation. In our model, a reduction in the worldinterest rate provides incentives for agents toborrow, increasing currentconsumption and the current account deficit. Depending on howsubstitutable tradable and non-tradable goods are, agents woulddemand more of both goods shifting resources towards the non-tradable sector andbiddingup its price. Themodel is consistentwith thesize of the reallocation of labor towards the non-tradable sector and ourresults are robust to the introduction of international goods differen-tiation and terms of trade shocks.

Calibrated open economy models have been successfully used tounderstand the 1995 crisis in Mexico and its effect on real GDP. A fewrecent examples include Kehoe and Ruhl (2009), Meza (2008) andPratap and Urrutia (2010). Some of these exercises have implicationsfor the evolution of the RER during the sudden stop. In particular,Kehoe and Ruhl (2009) do obtain an RER appreciation after a jump atthe beginning of the crisis, mostly driven by changes in terms of traderather than changes in the domestic relative price of traded goods. Inthis paper we switch the focus of our analysis from the short runeffects of the 1995 crisis to the whole 1988–2002 period and showthat in the long run changes in the domestic relative price of tradedgoods are indeed more important to explain the RER appreciation.Their model also abstracts from sector-specific TFP shocks.

Our analysis also borrows from the structural transformationliterature, which focuses on the long run reallocation of labor acrosssectors. Ngai and Pissarides (2007) study how differences in TFPgrowth rates across sectors lead to structural change in a model withan investment and a consumption sector, while Guerrieri andAcemoglu (2008) study how differences in capital shares acrosssectors lead to more rapid growth of employment in less capital-intensive sectors. In the context of our model, the tradable sectorincludes manufacturing, which is an investment good produced in acapital intensive industry, while the non-tradable sector can bemapped into the consumption, labor intensive sector. Differently tothese papers, we analyze the process of structural transformation inan open economy model and show that the ability to borrow fromabroad is key to understand the size and the speed of laborreallocation across sectors.

Our paper also relates to the empirical literature on the Balassa–Samuelson effect and the long run determinants of the RER (see, forexample, Asea and Mendoza (1994), Canzoneri et al. (1999), andChoudhri and Khan (2005)). The results in this literature offer mixedsupport for the importance of faster productivity growth in thetradable sector as a long run determinant of RER movements. Ourapproach is different, though, in that we use a structural model toevaluate the impact of sectoral TFP shocks measured from the data.1

We find that changes in the cost of accessing foreign credit canamplify the impact of differential productivity growth and thereforeshould be included in any empirical analysis of long run RERdeterminants.

Finally, our work is related to the research on the relation betweenfinancial liberalization and growth. Tornell and Westermann (2005)study the impact of financial liberalization on economic growth andwhether it leads to financial fragility, in an environment in whichthere are credit market imperfections that preclude firms in the non-tradable sector to borrow. Their model displays boom–bust cyclesdriven by changes in the international interest rate and amplified

1 Another difference between our analysis and the literature on the Balassa–Samuelson effect is that we use a different decomposition of the RER, as discussed inSection 2.1.1. Unlike Engel-style decompositions, focusing on the domestic relativeprice of tradable goods allows us to isolate the role of labor reallocation and sectoralproductivity growth in a small open economy from the direct impact of internationalprices.

through a domestic credit channel for which financial frictions arekey. Hence, the real exchange rate displays high volatility ineconomies undergoing a process of financial liberalization. In contrast,our paper abstracts from financial market imperfections and from theshort run crisis episodes, focusing instead on the long run response ofthe economic structure after a process of financial liberalization.

The paper is organized as follows. In Section 2 we discuss theevidence from the 1988–2002 Mexican data and show that thisexperience is similar to other episodes in Latin America and Centraland Eastern Europe. Section 3 introduces the model, while thecalibration and the main quantitative exercise are described inSection 4. In Section 5, we discuss in more detail the mechanismsdriving our results. Section 6 modifies the basic model to allow forinternational good differentiation and terms of trade shocks. Finally,we conclude.

2. Looking at Mexican data: 1988–2002

The first step in our investigation is to look carefully at the RERappreciation inMexico between 1988 and 2002.We show that a fall inthe domestic relative price of tradable goods accounts for about 78% ofthe real appreciation. We also provide a decomposition of the changesin the relative price of tradable over non-tradable goods which guidesour choice of a model in the next section. We perform sectoral growthaccounting exercises for the tradable and non-tradable sectors andidentify TFP shocks (Solow residuals) affecting their relative produc-tivity. Finally, we document other experiences in emerging marketssharing the same characteristics as the Mexican case.

2.1. Real exchange rate and relative prices

We construct the bilateral, GDP based real exchange rate forMexico against the US using the standard definition:

RER≡ eP�

P

where e is the nominal exchange rate (pesos per dollar) and P and P∗

are the GDP deflators in Mexico and the US. Fig. 1 displays the timeseries for this variable between 1988 and 2002, normalized to take thevalue 100 in 1988. Our measure shows a large 40% appreciation in theRER for Mexico between 1988 and 2002 together with a sharp, butshort lived, depreciation during the 1995 crisis. We focus in this paperon the long run negative trend, instead of the short run spike of 1995.

Fig. 1 also compares our measure of the RER against a multilateral,CPI based measure reported by Banco de Mexico. These twomeasuresare very similar and capture the same long run trend. If anything, themultilateral CPI based RER features more volatility, with a largerdepreciation during the 1995 crisis and a bigger appreciation (45%

40

60

1988 1990 1992 1994 1996 1998 2000 2002

GDP bilateral RER CPI multilateral RER

Fig. 1. Evolution of the real exchange rate in Mexico, 1988–2002.

40

60

80

100

120

140

1988 1990 1992 1994 1996 1998 2000 2002

GDP bilateral RER Price Tradables/GDP Residual

Fig. 2. Real exchange rate and the domestic relative price of tradable goods.

60

80

100

120

140

1988 1990 1992 1994 1996 1998 2000 2002

Residual Equation (1) PM/PX (Inverse of TOT)

Fig. 3. Terms of trade and the residual.

319F. Meza, C. Urrutia / Journal of International Economics 85 (2011) 317–328

instead of 40%) over the whole period. We choose to continue theanalysis with our bilateral GDP based RER since it is easy to map intothe NIPA system.

2.1.1. Real exchange rate and the domestic relative price of tradablesMeasuring prices consistently, the following identity holds:

RER≡ eP�

P=

eP�

PT

� �|fflfflffl{zfflfflffl}

residual

PT

P

!: ð1Þ

The second term PT/P is the price of domestic tradable goodsrelative to the domestic aggregate price level. We will refer to thisprice in short as the domestic relative price of tradables. The first termeP∗/PT is a residual which captures deviations from the price ofMexican tradable goods with respect to the foreign price level.

The decomposition is useful because standard neoclassical modelsof the small open economy are silent about this residual. If anything, atwo-sector version with non-tradable goods can generate deviationsbetween the domestic prices of tradable goods and the aggregateprice level: With a weight γ of tradable goods in the aggregate pricelevel, we can approximate the domestic relative price of tradables by:

PT

P≈ PT

PN

!1−γ: ð2Þ

Hence changes in the relative price of the tradable good over thenon-tradable good PT/PN could provide a potential explanation ofmovements in the domestic price of tradables and the RER. However,if the economy is small and markets are competitive, the relationbetween the price of tradables in the domestic market and the foreignprice level is exogenous, so the model has no explanatory power withrespect to it.

It is then relevant to assess the quantitative importance of the twochannels in explaining the RER appreciation inMexico.We construct atime series for the domestic relative price of tradables in Mexicodividing the sectoral value added for tradable sectors by the GDPdeflator, both obtained from NIPA.2Fig. 2 compares this price to theGDP based bilateral RER. As shown, the decline in the domesticrelative price of tradables is the key component to understand the RERappreciation in Mexico. In a crude decomposition, looking only atendpoints, the decline in the domestic relative price of tradables

2 In our data analysis we follow the convention of including manufacturing,agriculture, mining and fishing activities as part of the traded good sector. All otheractivities (in particular, services, construction) are treated as part of the non-tradedsector. Deflators for each sector are computed dividing value added at current pricesby value added at constant (1993) prices.

accounts for 78% of the change in the RER. Changes in the residual asdefined in Eq. (1) are much smaller in the long run, although theyseem to explain the 1995 jump.

2.1.2. Terms of trade and the residualThe residual in Eq. (1) could be capturing different things: Terms

of trade, transportation costs, price of non-tradables abroad, foreignexporters mark-ups, and so on. Perhaps surprisingly, Fig. 3 shows thatthe long run behavior of the residual for Mexico is captured by theinverse of the terms of trade (i.e., the relative price of imports overexports, computed again using deflators from NIPA). The correlationbetween these time series is also high (0.79), although the residualshows more volatility in particular during the 1995 crisis.

Product differentiation by country of origin provides an explanationto differences in the prices of exports and imports even in the context ofcompetitive, small open economymodels. We will add this feature to asecond version of our model to see how robust our results are toexogenous changes in the residual driven by terms of trade shocks.

2.1.3. An alternative decomposition of the RERFollowing Engel (1999), we can also decompose the RER in the

following two components:

RER≡ eP�

PePT�

PT

!|fflfflfflffl{zfflfflfflffl}

deviations LOP

PT= P

PT� = P�

!: ð3Þ

The second term is the domestic relative price of tradables dividedby the foreign relative price of tradables. The first term capturesdeviations in the law of one price in tradable goods. Engel (1999)provides a variance decomposition of the RER for the US and showsthat deviations in the law of one price in tradable goods are moreimportant than previously thought. Mendoza (2005) confirms thisresult for Mexico. Using a time frame comparable to ours, Kehoe andRuhl (2009) calculate that deviations in the law of one price intradable goods account for about 65% of the changes in the RER inMexico.

This result does not contradict our conclusion that most of theaction in explaining long run RER movements in Mexico lies in thedomestic relative price of tradable goods. By construction, Engel-styledecompositions underestimate the role of the domestic relative priceof tradable goods if similar changes in prices are also observed inforeign countries. But for the purpose of our paper, changes in theforeign relative price of tradable goods are irrelevant, as they areexogenous for a small open economy and, contrarily to terms of tradeshocks captured by our residual defined in Eq. (1), they do not affectdomestic decisions.

40

60

80

100

120

1988 1990 1992 1994 1996 1998 2000 2002

Relative Price T/N Rel. Output per Worker N/T

Relative Wage T/N Relative Labor Share N/T

Fig. 4. Decomposing the evolution of the relative price of tradable over non-tradablegoods.

Table 1Decomposition of the relative price of tradable over non-tradable goods.

Contribution (%) Relative price (T/N)

(1) Relative wages (T/N) 24%(2) Relative labor income shares (N/T) 10%(3) Relative output per worker (N/T) 66%(1)+(3) 90%(1)+(2)+(3) 100%

Table 2Sectoral growth accounting.

Annualized growth rate (%) Output Capital Labor Implied A

Tradable sector1988–93 3.5% 2.7% 0.2% 2.1%1993–98 4.4% 4.7% 1.1% 1.5%1998–02 1.5% −0.1% −0.5% 1.9%1988–2002 3.3% 2.6% 0.3% 1.8%

Non-tradable sector1988–93 4.0% 6.2% 4.0% −0.8%1993–98 2.4% 2.3% 2.8% −0.5%1998–02 3.3% 6.5% 1.3% 0.2%1988–2002 3.2% 5.1% 2.8% −0.4%

320 F. Meza, C. Urrutia / Journal of International Economics 85 (2011) 317–328

2.2. More on the relative price of tradable goods

We present now a decomposition of the relative price of tradableover non-tradable goods that will guide our modeling choices:

PT

PN ≡ WT

WN

!|fflfflfflffl{zfflfflfflffl}Relative

Wages

WNLN =PNYN

WTLT =PTYT

!|fflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

Relative Labor

Income Shares

YN=LN

YT =LT

!|fflfflfflfflfflfflffl{zfflfflfflfflfflfflffl}Relative Output

per Worker

:

Note that this identity holds for any economy as long as the datathat we feed into it is collected in a consistent way.

Changes in the relative price of tradable over non-tradable goodscan be accounted for by: (i) movements in the relative average wage,(ii) changes in the relative labor income shares, and (iii) movementsin the relative output per worker. UsingMexican data, Fig. 4 plots eachof the three components against the relative price.3 Table 1 summa-rizes the results of the decomposition, looking only at endpoints.

Notice first that, even though relative labor income shares are farfrom constant over time, they do not display any significative long runtrend. Not surprisingly, their overall impact on relative prices is small.We use this evidence to justify our choice of a Cobb–Douglas productionfunction in the model that follows, instead of a setup in which laborincome shares vary over time. By this choice we lose some action, butour decomposition shows that we miss less than 10% of the change inthe variable that we want to explain, in exchange for tractability.

According to our decomposition, everything else equal relativeprices and relativewages should bedirectly related. In theMexicandata,they are. Between1989 and1996,wages grewat a faster rate in the non-tradable sector, although from 1997 onwards relative wages are largelyflat. Overall, changes in relative wages account for 24% of the fall in therelative price of tradable over non-tradable goods. This evidencesuggests that deviations from wage equalization across sectors play arole in explaining the RER appreciation. Our model will feature a labormarket friction which will be consistent with this property of the data.

The observed decline in relative output per worker of the non-tradable sector against the tradable sector is large and accounts for 66%of the fall in the relative price of tradable over non-tradable goods.Adding the contribution of output per worker and relative wages weaccount for 90% of the decline in this relative price. It is key that over the

3 In these series, output for each sector corresponds to sectoral GDP (value added) atconstant prices, the number of workers is obtained from employment series by sector,and nominal wages for each sector are computed as the ratio of the wage bill (atcurrent prices) divided by the number of workers.

15-year periodanalyzed in thedata labor productivity grewconsistentlyat a faster rate in the tradable sector.We nowanalyzemore deeplywhatis behind these productivity changes using growth accounting.

2.3. Sectoral growth accounting

Inspired by the previous discussion, we continue our analysis byimposing a Cobb–Douglas production function in each sector:

Yit = Ai

t Kit

� �αi Lit� �1−αi

for i=T, N. From this equation, we compute the implied total factorproductivity (TFP) factors as

Ait =

Yit

Kit

� �αi Lit� �1−αi

using data for sectoral output (VA at constant prices), labor (employ-ment) and capital (also at constant prices). Data for capital stocks isobtained fromBanco deMexico surveys, and is consistent at the sectorallevel with the perpetual inventory method. We use the factor sharesαT=0.48 and αN=0.35, whose values will be discussed in detail in thecalibration section.

Table 2 and Fig. 5 report the implied TFP factors obtained from theformula above. TFP in the tradable sector grew on average at a 2.2%annual rate relative to TFP in the non-tradable sector. This rate doesnot change significantly over time, and it is mostly driven by thegrowth in AT.4 At the same time capital and labor reallocate from thetradable sector towards the non-tradable sector, as shown in Fig. 6.The evidence characterizes a period of structural transformation of theMexican economy which is consistent with an explanation of the RERappreciation based on the differential productivity growth betweentradable and non-tradable sectors.

4 Tornell et al. (2004) highlight the stagnation in the non-tradable sector as themain reason for the low growth performance of the Mexican economy followingNAFTA. In their view, the main reason is the lack of credit to non-tradable firms. Theydo not compute TFP by sector, basing their analysis on sectoral output.

80

90

100

110

120

130

140

1988 1990 1992 1994 1996 1998 2000 2002

TFP T Sector TFP N Sector

Fig. 5. Evolution of total factor productivity in tradable and non-tradable sectors.

60

70

80

90

100

110

1988 1990 1992 1994 1996 1998 2000 2002

Relative Labor T/N Relative Capital T/N

Fig. 6. Reallocation of labor and capital between tradable and non-tradable sectors.

321F. Meza, C. Urrutia / Journal of International Economics 85 (2011) 317–328

2.4. The Mexican case in context

Before presenting the model and the quantitative exercise, weprovide some evidence situating the Mexican case in a broadercontext. The Mexican experience shares characteristics with episodesin other emerging economies that have opened to capital flows. In thecase of Latin America we focus on the period post 1988, in whichmany countries regained access to capital markets after the 1982 debtcrisis. In the case of Central and Eastern Europewe focus on the periodpost 1992, after the changes in regimes in these countries.

Table 3Real appreciation and labor reallocation in selected countries.

Latin American countriesAvg. yearly change Mexico (88–02) Venezuela (88–01)Real exchange ratea −3.6% −4.1%Labor reallocationc: LT/LN −2.4% −3.6%

Central and Eastern European countriesAvg. yearly change Bulgaria (92–02) Romania (92–02)Real exchange ratea −7.7% −6.6%Labor reallocationc: LT/LN −2.5% −2.1%d

a CPI based real effective exchange rate from IMF International Financial Statistics, unlessin Fig. 2.

b Trade weighted real exchange rate from JP Morgan. Source: Haver Analytics.c Constructed using ILO Laborstat's employment by industry. For Mexico we used our owd 92–98 for Brazil, 96–01 for Argentina, 94–02 for Romania and Poland, and 93–02 for th

Table 3 reports the percentage change in the RER in selectedcountries. For some countries in Latin America, such as Brazil andArgentina, the period of appreciation is interrupted by financial crises.Overall, the appreciations are of similar magnitude in our sample ofLatin American countries and in Mexico. In Central and EasternEurope the appreciation of the real exchange rate is larger. Table 3 alsoreports the size of the reallocation of labor towards the non-tradablesector, as the average percentage change in the ratio of tradable tonon-tradable workers. The changes observed in Latin American and inCentral and Eastern Europe are of similar magnitude as in Mexico.

The behavior of the interest rates that emerging economies face inthe international markets is consistent with our story. Fig. 7 reportsthe country specific interest rate in international markets for selectedcountries, computed as the real interest rate in the U.S. (3 month T-Bills) plus the JP Morgan's Emerging Markets Bond Index (Global)spread. Unfortunately, indicators for emerging markets country riskare not available for many countries before 1998, so we had to reduceour sample according to the information available. For Mexico, we usea longer series of JP Morgan's Mexican Brady bonds' spread, asdescribed in Meza and Quintin (2007).

Fig. 7 shows a consistent decline in the cost of foreign credit incountries experiencing significant RER appreciations. The interest ratefor loans toMexico fell from about 15% in 1990 to less than 5% in 2002,with a short run jump during the 1995 crisis. Similar declines areobserved in Venezuela and Poland, starting in 1995. It is remarkablethat the decline in the cost of foreign credit in Bulgaria, the countrywith the largest RER appreciation, is also the steepest (from 22% in1995 to about 3% in 2002). Overall, the evidence supports includingchanges in the external interest rate in our analysis as a driving force.

Finally, it is also useful to put the period 1988–2002 in contextwith respect to Mexico's own past experience. A valid concern iswhether choosing 1988 as the starting year for our analysis hidesprevious structural trends in the Mexican economy explaining thesubsequent appreciation. We believe that it is indeed a special periodmarked by a structural break in 1988. Following the 1982 sovereigndefault, for most of the decade Mexico was indeed excluded frominternational capital markets. In 1988 a new Mexican governmentstarted the negotiations that led to the Brady Plan. Mexico regainedaccess to international credit markets, although initially facing a veryhigh interest rate. As seen in Fig. 7 the country risk premium ininternational markets gradually felt over the next five years.

Fig. 8 extends our series for the domestic relative price of tradablegoods and the ratio of employment in tradable sector to non-tradablesector in Mexico back to 1980. The years before 1988 werecharacterized by a mild increase in the price of tradable goods(almost entirely accounted for a jump after 1982, the year of thegovernment default) and almost no labor reallocation across sectors.The sustained appreciation of the real exchange rate in Mexico and itsdriving forces in our analysis do seem to have started by 1988.

Brazil (88–98) Chile (88–00) Argentina (88–01)−3.4%b −2.0% −1.8%b

−4.2%d −3.1% −3.6%d

Czech R. (92–02) Poland (92–02) Hungary (92–02)−5.5% −4.6% −3.0%−2.1%d −4.1%d −3.1%

indicated. A minus sign indicates an appreciation. For Mexico we use the RER presented

n calculations.e Czech Republic.

0%

5%

10%

15%

20%

25%

1990 1992 1994 1996 1998 2000 2002

Mexico Venezuela Bulgaria Poland

Fig. 7. Interest rate for foreign debt in selected countries (including country premium).

60

70

80

90

100

110

1980 1984 1988 1992 1996 2000

Price Tradables/GDP Relative Labor T/N

Fig. 8. Relative price of tradables and labor reallocation in Mexico, 1980–2002.

322 F. Meza, C. Urrutia / Journal of International Economics 85 (2011) 317–328

3. A two-sector model of a small open economy

We build a simple two sector dynamic general equilibrium modelof a small open economy. This model provides a natural laboratory toanalyze the impact of sectoral TFP and interest rate shocks on therelative price of tradable goods and on the allocation of capital andlabor across sectors.

3.1. Production

The economy produces two intermediate goods, one of which istradable while the other cannot be traded with the rest of the world.Each intermediate good is produced combining capital and laborthrough a Cobb–Douglas technology:

Yit = Ai

t Kit

� �αi Lit� �1−αi i = T;N:

Capital and labor are rented from consumers. A final good isproduced using tradable and non-tradable goods as inputs, using theCES aggregator

Yt = γ QTt

� �ρ+ 1−γð Þ QN

t

� �ρh i1ρ

where Yt is the output of the final good (it will also be equal todomestic absorption in our model), Qt

T and QtN are quantities of each

intermediate good and ρ determines the elasticity of substitution,which is 1/(1−ρ).

There is one representative firm in each sector which takes pricesas given. The profit-maximization problem for the intermediateproducer of good i=T, N is

maxYit ;L

it ;K

it

pitYit−wi

tLit−rtK

it

n o

s:to Yit = Ai

t Kit

� �αi Lit� �1−αi

for all t, where the price of tradable and non-tradable goods (ptT andptN), the sector specific wage rates (wt

T and wtN) and the common

rental rate of capital rt are all expressed in units of the final good.Capital is freely mobile across sectors, but there are frictions to laborreallocation that prevent wages to equate across sectors.

Similarly, the producer of the final good solves each period thestatic problem:

maxYTt ;Q

Tt ;Q

Nt

Yt−pTt QTt −pNt Q

Nt

n o

s:to Yt = γ QTt

� �ρ+ 1−γð Þ QN

t

� �ρh i1ρ:

Note that the final good producer does not add any value added tothe economy. GDP at current prices will be given by the sum of valueadded by the tradable and non-tradable intermediate good sectors:

GDPt = pTt YTt + pNt Y

Nt

while real GDP (at constant prices) is constructed using a base year'sprices:

RGDPt = pT0YTt + pN0 Y

Nt :

3.2. Consumption and savings

A representative consumer is endowed with K0 units of initialcapital, B0 units of foreign bonds, and a sequence Lt

∞t = 0 of labor

endowments supplied inelastically to the market. Consumers'intertemporal preferences are summarized by the CRRA lifetimeutility function:

∑∞

t=0βt Ct =Lt� �1−σ−1

1−σ

" #

where Ct represents consumption of the final good and β∈(0, 1) is thediscount factor. Income is obtained from renting labor to each sector,at wage rates in units of the final good wt

T and wtN, and renting capital

at a common rental rate rt. The representative consumer decides howmuch to consume, howmuch to invest in new capital and new foreignbonds, and the fraction of his/her labor endowment θt supplied to thetradable sector.

The budget constraint for each period is:

Ct + Kt + 1 + pTt Bt + 1 = wTt θtLt + wN

t 1−θtð ÞL t+ rt + 1−δð Þ½ �Kt

+ 1 + r∗tð ÞpTt Bt−ψ2

Kt + 1−Kt

Kt

� �2−ϕ

2θt−θt−1ð Þ2

where r∗t ∞

t = 0 is an exogenous sequence of world interest rates,δ∈(0, 1) is the depreciation rate and the parameters ψ, ϕN0 indicatethemagnitude of the quadratic adjustment costs to change the stock ofcapital and tomove labor across sectors, respectively. Our intuition forthe latter is that changing sectors implies for workers some loss ofsector-specific human capital, whose cost is paid by the representative

323F. Meza, C. Urrutia / Journal of International Economics 85 (2011) 317–328

consumer according to this ad-hoc function.5 The initial allocation oflabor across sectors inherited from the past (θ−1) is exogenouslygiven.

The representative consumer maximizes lifetime utility subject tothe budget constraint above. Notice that tradable and non-tradablegoods are combined in the same way to produce consumption andinvestment goods. We choose this specification for simplicity, eventhough it abstracts from changes in the relative price of investmentover consumption goods. Note also that the foreign bond and itsexogenous return are also denominated in units of the tradable good.

3.3. Equilibrium

The model is closed by imposing the following market clearingconditions: (i) for the final good

Ct + Kt + 1− 1−δð ÞKt +ψ2

Kt + 1−Kt

Kt

� �2+

ϕ2

θt−θt−1ð Þ2 = Yt

(ii) for each intermediate good

QTt + NXT

t = YTt

tNQ = YN

t

where NXtT represents net exports of the tradable good, and (iii) for

production factors

KTt + KN

t = Kt LTt = θtLt LNt = 1−θtð ÞLt :

In this setup, the current account can be constructed as the value ofnet exports plus interest payments or as the change in the foreignasset position:

C At ≡ N XTt + r�t Bt = Bt + 1−Bt :

Notice that GDP can be constructed according to NIPA methodol-ogy as:

GDP ≡ pTt YTt + pNt Y

Nt = wT

t θtLt + wNt 1−θtð ÞL t

+ rtKt = Yt + pTt NXTt :

4. Accounting for the Mexican appreciation

This section describes the main exercise in our paper. We computethe transitional path for the small open economy described in theprevious section, starting from an initial stationary equilibrium andgiven exogenous sequences for sectoral TFP {At

T, AtN}, for international

interest rates {rt∗} and total employment Lt

taken from the data. Wethen analyze the resulting sequences for the domestic relative price oftradables in order to assess the ability of the model to generate anappreciation of the RER as the one observed in Mexico. We alsocompare model predictions to data on labor reallocation acrosssectors and discuss its limitations in reproducing the behavior of thecurrent account.

4.1. Calibrating the model

The model is calibrated to Mexican data. A few parameters have adirect empirical counterpart. Others are determined simultaneouslymatching a set of calibration targets. Notice that, although our modelis forced to be consistent with some observations for the Mexican

5 In a recent paper, Pratap and Quintin (2010) argue that labor reallocation can becostly. Using household level data for Mexico, they report that workers who changedsectors and/or occupations during the 1995 crisis experienced sizable losses inearnings. We discuss in more detail this evidence in Section 5.2.

economy, no data on the real exchange rate nor on the relative price oftradable goods is used to calibrate the parameters.

We use the 1980 Mexican input–output matrix to calibrateincome shares in the production functions. Unfortunately, suchmatrices are not computed regularly. However, most of ourcalibration results are consistent with the calibration in Kehoe andRuhl (2009), using an unpublished 1988 matrix. We measurepayments to labor relative to GDP at factor prices for each sector.We adjust the labor income shares by taking into account the incomeof the self-employed, following Gollin (2002) and Garcia-Verdú(2005). We obtain a capital income share in the tradable sector of0.48, and 0.35 in the non-tradable sector. As expected, the tradablesector is capital intensive.

For the final goods aggregator, we choose the value of ρ to have an

elasticity of substitution of12between tradable and non-tradable goods,

similar to Stockman and Tesar (1995). We then calibrate the weight oftradable goods in the production of final goods, γ, using information onfinal domestic demand for tradable and non-tradable sectors in Mexicofrom the 1980 input–outputmatrix. Given the valueofρ, we use thefirstorder conditions of the final goods producer, yielding the relative priceof tradable goods as a function of the ratio of QT toQN. Choosing units tonormalize the 1980 relative price of tradable over non-tradable goods toone, the weight of tradable goods is γ=0.23.

On the consumption side, we choose a standard risk aversioncoefficient implying an intertemporal elasticity of substitution of12.The exogenous sequence of international interest rates {rt*} is

computed as the real interest rate in the US plus a Mexican specificspread, as discussed in Section 2 and reported in Fig. 7. Since the firstobservation available is from the end of 1990, we extrapolate the1988–89 values using information on domestic Mexican interest ratesin dollars, giving us an initial interest rate close to 20% for 1988. Weassume a long run interest rate of 4.5%, that we use to calibrate thediscount factor β. The depreciation rate δ is 5%.

Given these parameters, we jointly calibrate the initial relative TFPbetween the two sectors A0

T/A0N and the initial stock of wealth of the

economy (B0) to obtain in the initial stationary equilibrium of themodel: (i) a fraction of labor allocated to the tradable sector of 40%,and (ii) a fraction of net exports in GDP of 2.3%. These numbers areconsistent with Mexican data for 1988. The sequences for {At

T, AtN} are

constructed given the initial ratio A0T/A0

N and using the observed ratesof growth of TFP for each sector (from Fig. 5).

This leaves us with two parameters (ψ and ϕ), associated to theadjustment costs for capital and labor. Since the adjustment cost forcapital controls the speed of capital accumulation, GDP growth seems anatural target. Similarly, the adjustment cost for labor can be pinneddown by deviations from wage equalization across sectors, which aredue to labor market frictions in our model. Therefore, we jointly choosethe values of these two parameters to minimize the distance betweenthe time series generated by themodel along the equilibrium transitionpath described in the next section and the Mexican 1988–02 data for:(i) real GDPperworker, and (ii) the relativewage between tradable andnon-tradable sectors. The calibration is summarized in Table 4.

4.2. Equilibrium path and the relative price of tradable goods

We compute the transitional equilibrium path of the model asfollows. First, we obtain the initial conditions for capital (K0), bonds(B0), and labor allocation (θ−1) from the stationary equilibrium of themodel given the values for A0

T, A0N, r0∗, and L0. In particular, we compute

this initial steady state assuming a world interest rate r0∗=20% and

adjusting the discount factor accordingly. All other parameters are thesame as in Table 4.

We then feed themodel with the exogenous sequences for sectoralTFP {At

T, AtN}, international interest rates {rt*} and total employment

Table 4Calibration of the model.

Statistic Parameter

Labor income share in tradable sector 0.52 αT 0.48Labor income share in non-tradable sector 0.65 αN 0.35Elasticity of substitution T and N goods 0.5 ρ −1.0Ratio of tradable to non-tradable goodsin domestic demand

0.55 γ 0.23

Long run world interest rate 4.5% β 0.957Intertemporal elasticity of substitution 0.5 σ 2.0Depreciation rate 0.05 δ 0.05Stationary fraction of labor in tradable sector 40% A0

T/A0N 0.516

Stationary fraction of net exports in GDP 2.3% B0 −0.045Minimum distance between data and model

Total real GDP per worker ψ 32.25Relative wage between T and N sectors ϕ 145.64

324 F. Meza, C. Urrutia / Journal of International Economics 85 (2011) 317–328

Lt

constructed from the data.6 We assume that in n=100 periods(years) the economy reaches the new steady state, given the finalvalues for An

T, AnN, rn*, and Ln. These values are assumed to be equal to

their 2002 counterpart in the data, i.e., from 2003 onwardswe assumethey remain constant. Solving the system of first order conditions foreach of the n periods we obtain the equilibrium transition path for theendogenous variables of the model.

Fig. 9 reports the time series obtained from the first fifteenobservations generated by the transitional equilibrium path of ourmodel and compares them to the actual 1988–02 Mexican data. Byconstruction, the model reproduces very well the trends for GDP perworker and relative wages across sectors. A better measure of thesuccess of the model is how well it captures the structural shift oflabor from the tradable to the non-tradable sector, as well as thedownward trend in relative output per worker of the non-tradablesector. The model does this successfully. Moreover, as seen in panel(e) of Fig. 9, the model generates a large decline in the domesticrelative price of tradable goods, defined as pt

T in the model. Lookingonly at endpoints, the model accounts for 60% of the change in thisrelative price which, as discussed in Section 2, is responsible for mostof the RER appreciation in Mexico.

4.3. The 1994–95 crisis and the current account

One limitation of the exercise is that it does not capture theobserved behavior of the current account. Panel (f) of Fig. 9 displaysthe evolution of net exports (as a fraction of GDP), both in the modeland in the Mexican data. Facing a horizon of increasing productivityand declining interest rates, agents in the model borrow too muchfrom abroad, compared to the data. Also, instead of a current accountreversal in 1995 (the year of the Mexican crisis), the benchmarkeconomy displays an even larger trade deficit. The model has sometrouble matching the data around 1995 for other variables as well.

Most of the limitations of the benchmark model are related to oursimplified treatment of the 1995 crisis. First, we are assuming that the1995 interest rate and productivity shocks were perfectly anticipated,so agents start reacting to it in previous periods. Second, wemodel thecrisis itself as a negative TFP shock coupled with an increase in theinternational interest rate for Mexico, not as a sudden stop of loansfrom abroad due to foreign credit rationing.7 The assumptions we use

6 Feeding the model with the observed time series for total employment allows usto take into account the effect of population growth and changes in participation rates.By construction, our experiment is consistent with the observed change in the size ofthe labor force in the Mexican economy from 1988 to 2002. What we do not take intoaccount are changes in the composition of the labor force, both in terms of age and skill,which could be potentially important in explaining the structural change. A seriousanalysis of this channel is beyond the scope of our paper.

7 Kehoe and Ruhl (2009) model the Mexican crisis as an exogenous sudden stop ofcapital inflows. The economy is temporarily unable to borrow from the rest of theworld.

produce a deterioration of the current account due to the temporarynegative income shock.

Our focus in this paper is the long run trend of the RER, not theshort run depreciation of 1995. Still, we test a simple and alternativeway of modeling a sudden stop as a large increase of the internationalinterest rate for Mexico in 1995 and 1996, reaching a value of 40%,which would mimic a two-year long quantitative restriction onforeign borrowing. We report in the first panel of Fig. 10 the ratio ofnet exports over GDP when the shock is perfectly foreseen (since1988) or when it comes in 1995 as a complete surprise. If the suddenstop is anticipated, the average level of borrowing of the economy issmaller before the crisis, even less than the one implied by theobserved trade deficit. As in the benchmark model, the trade balanceworsens during the crisis due to the negative income effect oftemporary TFP shocks. When the sudden stop is unforeseen, theeconomy borrows too much before the crisis but there is a strongreversal of the trade balance in 1995 due to the unexpected fall inpermanent income caused by the interest rate shock.

The implications of these different modeling strategies for the realexchange rate appreciation are important for our exercise. The secondpanel of Fig. 10 displays the evolution of the domestic relative price oftradables. When the sudden stop is anticipated, we account for almostall of the decline in this price, compared to 60% in the benchmarkmodel. In the case when the shock is unforeseen, we account for morethan 50% of the decline in the relative price of tradable goods over thewhole period, even though it overshoots during the crisis.

5. Sources of the Mexican appreciation

In this section, we decompose the decline in the domestic relativeprice of tradable goods generated by the model distinguishingbetween its two main sources, the decline in the internationalinterest rate and sectoral TFP shocks. We also analyze the role of theadjustment cost of labor for our results.

5.1. Sectoral TFP shocks vs. interest rate shocks

Our model economy faces two types of shocks: (i) an interest rateshock, driven by the change in the international interest rate forMexico, and (ii) a productivity shock, the differential TFP growth intradable and non-tradable sectors.

Let us start with the sectoral TFP shocks. As discussed in Section 2,TFP growth has been unequal across sectors. Between 1988 and 2002,TFP grew at a 1.8% yearly rate in the tradable sector, while TFP in thenon-tradable sector remained stagnant (in fact, it declined by 0.4%). Inour model, technological progress in the tradable sector reduces thecost of production in this sector, making tradable goods cheaper andappreciating the RER.

The impact of differential productivity growth across sectors onlabor reallocation is more ambiguous. The direct effect of TFP changesis to switch resources towards themost productive sector. In this case,this is the tradable sector. But there is a second, income effect. Asproductivity growth makes the economy richer, agents demand moreof the two goods. The tradable good can be imported, but the non-tradable good has to be domestically produced so resources movetowards this sector. Depending on how substitutable tradable andnon-tradable goods are in the production of the final good, eithereffect could dominate.

Table 5 shows the results generated by the model shutting downthis channel, this is, without sectoral TFP shocks. In this exercise, westill obtain a decline in the relative price of tradable goods, but ofabout 13% instead of 19% as in the benchmark model. Moreover, inthis version of the model there is also less labor reallocation to thenon-tradable sector compared to what is obtained in the benchmarkmodel and observed in the data.

a) Total GDP per Worker

c) Relative Output per Worker N / T

60

80

100

120

e) Domestic Relative Price of Tradables

60

80

100

120

1988 1990 1992 1994 1996 1998 2000 2002

1988 1990 1992 1994 1996 1998 2000 2002

b) Relative Wage T / N

60

80

100

120

d) Relative Labor T / N

50

70

90

110

f) Net Exports / GDP

-20%

-10%

0%

10%

1988 1990 1992 1994 1996 1998 2000 2002

1988 1990 1992 1994 1996 1998 2000 2002

1988 1990 1992 1994 1996 1998 2000 2002

Data Model

90

100

110

120

130

1988 1990 1992 1994 1996 1998 2000 2002

Fig. 9. Equilibrium transition for the benchmark economy.

325F. Meza, C. Urrutia / Journal of International Economics 85 (2011) 317–328

Most of the decline in the relative price of tradable goods isaccounted for by the decline in the interest rate faced by Mexico ininternational credit markets. In the context of the model, a reductionin the world interest rate provides incentives for agents to borrow,increasing current consumption and the current account deficit.Again, depending on how substitutable tradable and non-tradablegoods are, agents would demand more of both goods shiftingresources towards the non-tradable sector and bidding up its price.This interest rate effect proves to be quantitatively very important.8

To emphasize this point, we carried out an experiment with onlysectoral productivity changes.We fixed the interest rate to 12.25%, theaverage of the whole period, and calibrated the discount factoraccordingly. We then let relative productivity fluctuate as in thebenchmark experiment. The results are reported in Table 5. We foundthat the relative price of tradable goods falls by a much smalleramount, 4.3%, and that the reallocation of labor is also smaller. Noticethat the effects of the two shocks are not additive: The decline ininterest rates also amplifies the effect of sectoral productivity growthin generating a real appreciation.

8 Sensitivity analysis with respect to the intertemporal elasticity of substitution1σ

� �and the elasticity of substitution between tradable and non-tradable goods

11−ρ

� �is available upon request. Our results on the relative price of tradable goods

are robust to changes in each parameter within a reasonable range. Still, we find that

increasing the intertemporal elasticity of substitution or decreasing the elasticity of

substitution between tradable and non-tradable goods slightly ampli.es our results.

This is consistent with the intuition behind the interest rate effect.

5.2. The role of adjustment costs for labor

To analyze the importance of adjustment costs to labormobility forour results, we ran a version of the model in which these costs arereduced to one tenth of their original value (ϕ≈14.5). The results arereported in Table 5. With low adjustment costs for labor, the modelexplains less than half of the observed decline in the relative price oftradable goods, about 5% points less than in the benchmarkexperiment. Moreover, the model greatly overpredicts the size oflabor reallocation towards the non-tradable sector.

Both issues are related. Facing the exogenous sequences ofproductivity and interest rates, agents want to switch resourcesfrom the tradable sector to the non-tradable sector. With adjustmentcosts, this is done at a slower rate, keeping over time an inefficientlyhigh fraction of labor in the tradable sector and bidding down thewage rate in that sector. Producing tradable goods becomes cheaper,and this is reflected in a decline of its relative price with respect tonon-tradable goods, amplifying the RER appreciation.

Adjustment costs for labor are important in our story. We choosethe size of adjustment costs that better matches the evolution ofrelative wages in the data, abstracting from changes in thecomposition of the labor force in the two sectors. In the absence ofdirect microeconomic evidence, our calibration provides a degree ofdiscipline to the exercise. Moreover, although ϕ≈145 seems like alarge number, the amount of resources wasted by reallocating laborrepresents only between 0.5% and 0.7% of GDP in the model.

To provide evidence in favor of the importance and size of laborreallocation costs,we refer to PratapandQuintin (2010).Using individual

Fig. 10. Net exports with foreseen and unforeseen sudden stops in 1995.

Table 5Accounting for the Mexican appreciation.

Change (%) 88–02 Price of tradables Relative labor (T/N)

Data −32.1% −28.9%Benchmark economy −19.1% −24.9%No productivity shocks −12.5% −16.5%No interest rate shocks −4.3% −1.6%Low adjustment costs for labor −14.6% −42.1%

326 F. Meza, C. Urrutia / Journal of International Economics 85 (2011) 317–328

surveydata forMexico, PratapandQuintin (2010) report sizable earningslosses of about 10% for workers who switched jobs during the 1994–95financial crisis. These estimates control for workers' characteristics suchas age, education, etc., and job characteristics such as firm size, formality,etc. and represent the wage losses of movers relative to workers thatstayed at the same sector or occupation. These authors decompose theearnings losses on a fraction attributed to changes in occupation and afraction due to changes in sector of activity, andfind the impact of each tobe roughly equal.

In our quantitative experiment, the calibrated adjustment costs forlabor imply an aggregate loss of resources of less than 1% of GDP, orabout 1.5% of the wage bill. Pratap and Quintin report that in grossterms about 20% of workers in their sample changed sectors duringthe 1994–95 crisis, with an average earnings loss of 8.5% (the averageof their parametric and semi-parametric estimators) compared withworkers who stayed in their jobs. This implies a total earnings loss ofabout 1.7% of the wage bill. Our adjustment costs for labor are in linewith these estimates.9

6. Terms of trade, tariffs and the RER appreciation

In this final section, we modify the basic model by adding inter-national differentiation in tradable goods. This version of the model

9 However, there are a few caveats linking these numbers. One is that theirdefinition of sectors is not ours, as they do not classify industries in tradable and non-tradable sectors. The other is that such large losses are observed during a period ofhigh economic turbulence and might be smaller in more tranquil times. Finally, theselosses correspond to workers already in the labor force who switched sectors, insteadof young new workers joining the labor force.

features an importable good which is an imperfect substitute of thedomestically produced tradable good. The relative price between thetwo defines the terms of trade for this economy which, keeping theassumption of a small open economy, are exogenous. This setupallows us to check the robustness of our previous results with respectto deviations of the price of Mexican tradable goods with respect tothe foreign price level (the residual discussed in Section 2), generatedby terms of trade shocks.

6.1. A model with international differentiation of goods

The basic structure of the model is similar to the one described inSection 4. The main difference is that the final good is now producedaggregating non-tradables and a composite tradable good, itself theresult of aggregating domestically produced tradable goods andimports (Mt)

Yt = γ μ QTt

� �ζ+ 1−μð ÞMζ

t

� �1ζ

0B@

1CA

ρ

+ 1−γð Þ QNt

� �ρ264

3751ρ

with the (Armington) elasticity of substitution 1/(1−ζ). Theassumption is that, because of product differentiation, domesticallyproduced tradable goods and imports are not perfect substitutes.

The producer of the final good solves now each period the staticproblem:

maxYTt ;Q

Tt ;Q

Nt

Yt−pTt QTt − 1 + τtð Þ Mt

pTt =pMt

� ��" #

−pNt QNt

( )

s:to Yt = γ μ QTt

� �ζ+ 1−μð ÞMζ

t

� �1ζ

0B@

1CA

ρ

+ 1−γð Þ QNt

� �ρ264

3751ρ

where (ptT/ptM)∗ are the exogenously given terms of trade for thiseconomy and τt represents an import tariff, rebated to therepresentative consumer as a lump sum transfer.

In equilibrium,markets clear for each domestically produced good,

QTt + XT

t = YTt Q

Nt = YN

t

where XtT represents exports of the tradable good. Tariff collection is

rebated to the consumer as lump sum transfer Tt. Finally, the currentaccount can be constructed as the value of net exports plus interestpayments or as the change in the foreign asset position:

CAt≡ XTt −

Mt

pTt =pMt

� �� !

+ r�t Bt = Bt + 1−Bt :

We will focus on the predictions of the model regarding thedomestic relative price of tradables pt. However, this version of the

Table 6Calibration of the model with terms of trade shocks.

Statistic Parameter

Elasticity of substitution between tradablegoods and imports

2 ζ 0.5

Ratio of imports to tradable goods 0.56 μ 0.57Ratio of tradable to non-tradable goodsin domestic demand

0.55 γ 0.30

Stationary fraction of labor in tradable sector 40% A0T/A0

N 1.391Stationary fraction of net exports in GDP 2.3% B0 −0.126Minimum distance between data and model

Total real GDP per worker ψ 36.55Relative wage between T and N sectors ϕ 178.66

Table 7Accounting using the model with international goods differentiation.

Change (%) 88–02 Price of tradables Relative labor (T/N)Data −32.1% −28.9%Benchmark economy −19.1% −25.2%Model with terms of trade shocks −17.5% −25.3%

Adding import tariffs reduction −16.6% −25.2%

11 See Edwards and Van Wijnbergen (1987) for a detailed discussion on thetheoretical effects of changes in terms of trade and tariffs on relative prices and on theRER, in particular on the income and substitution effects involved.

327F. Meza, C. Urrutia / Journal of International Economics 85 (2011) 317–328

model allows us to construct a real exchange rate which includes theresidual exogenously explained by terms of trade shocks, as

RERt =pTt

pTt =pMt

� �� :

6.2. Revisiting our quantitative results

In order to compute the new version of the model we need first tocalibrate two new parameters in the Armington aggregator, ζ and μ,and recalibrate the parameters γ, A0

T/A0N, B0, ψ and ϕ to be consistent

with the same calibration targets as before. We choose initial terms oftrade pT0 =p

M0

� �∗ = 1 and set the initial import tariff to the level ofτ0=10%. We also set ζ=0.5 in order to have an elasticity ofsubstitution of 2 between imports and tradable goods and jointlycalibrate μ and γ using the 1980 input–output table.10 Finally, wefollow the same strategy as with the basic model in order to calibrateA0T/A0

N and B0 to match the initial labor allocation and net exports, andψ and ϕ to minimize the distance between the model's GDP andrelative wage and the corresponding series in the data. See Table 6. Allthe remaining parameters have the same values reported in Table 4.

6.2.1. The role of terms of tradeAs before, we compute the transitional equilibrium path of the

newmodel as follows. First, we obtain the initial conditions for capital,bonds, and labor allocation from the stationary equilibrium of themodel. Then, we feed the model with the exogenous sequences forsectoral TFP, international interest rates, total employment and termsof trade {(ptT/ptM)∗} observed in the data. The sequence for the terms oftrade corresponds to the (inverse of the) one reported in Fig. 3, whilefor now we keep tariffs constant during the whole period.

The results are similar to the ones in Fig. 9. Once recalibrated tomatch the trends for GDP per worker and relative wages acrosssectors, the model with terms of trade shocks also captures thestructural shift of labor from the tradable to the non-tradable sectorand the change in the composition of output. As reported in Table 7,looking only at endpoints the model with terms of trade shocksaccounts for 54% of the change in the domestic relative price oftradables, compared to 60% in the benchmark model. The model alsoaccounts for 69% of the RER appreciation, although it should benoticed that this number includes the contribution of exogenousterms of trade.

Previous studies (see, for example, De Gregorio and Wolf (1994)and Cashin et al. (2004)) have found an important role for terms oftrade as determinants of real exchange rate movements, especially incommodity exporting countries. Our results are consistent with thesefindings. The small improvement in terms of trade observed inMexico

10 The range of values for this elasticity varies widely in the literature, from less than0.5 to more than 10 (see Ruhl, 2008). It turns that our results are very robust to thechoice of this parameter. A sensitivity analysis on the value of ζ is available uponrequest.

between 1988 and 2002 in fact had a direct impact in the RERappreciation by increasing the price of Mexican tradable goods withrespect to the foreign price level. However, the improvement in termsof trade slightly reduces the decline in the domestic relative price oftradable goods generated by our model.11

6.2.2. The role of import tariffs reductionIn our last experiment we analyze the role of the import tariffs

reduction following the free trade agreements negotiated by Mexicoat the beginning of the 1990s, in particular NAFTA. Following Kehoeand Ruhl (2009), we model the tariff reduction in a simplified way:Starting from a 10% import tariff in 1988, we assume a reduction to 5%in 1994, followed by a 0.5 percentage point per year decline from 1994onwards. As observed in Table 7, the effects of this tariff cut on thelabor allocation across sectors and the domestic relative price oftradables are negligible, probably because import tariffs were alreadylow at the beginning of the period studied.12

7. Conclusions

We identify two main sources of RER appreciation in emergingmarkets using a two sector neoclassical growth model of a small openeconomy: (i) a decline in the real interest rate faced in internationalmarkets, associated to a process of financial liberalization, and (ii)differential TFP growth across tradable and non-tradable sectors. Thesetwo channels explain approximately 60% of the change in the domesticrelative price of tradables in Mexico, although the interest rate channelproves to be quantitatively themost important. The results are robust tothe inclusion of terms of trade and tariffs into the model.

One important question which remains open is: Are the twoidentified channels exogenous and independent of each other? Onecould think of a story in which productivity growth causes anendogenous reduction in the country risk premium by reducing theprobability of default, as inMendoza and Yue (2008). Or even assumingthat the country specific interest rate is exogenous, changes in the cost ofcredit might affect the productivity of firms in a model of financialfrictions. Moreover, as shown in Pratap and Urrutia (2010), thesechanges in the cost of credit affect differently measured TFP in thetradable and non-tradable sectors, providing a potential explanation todifferential productivity growth. A quantitative assessment of thesetransmission mechanisms is an interesting topic for future research.

References

Asea, P.K., Mendoza, E.G., 1994. The Balassa–Samuelson Model: a general-equilibriumappraisal. Review of International Economics 2, 244–267.

Canzoneri, M.B., Cumby, R.E., Diba, B., 1999. Relative labor productivity and the realexchange rate in the long run: evidence for a panel of OECD countries. Journal ofInternational Economics 47, 245–266.

Cashin, P., Cespedes, L.F., Sahay, R., 2004. Commodity currencies and the real exchangerate. Journal of Development Economics 75, 239–268.

12 We also quantify the role of three demand shocks in explaining the decline in therelative price of tradables. These shocks are: (i) migration remittances; (ii) foreignreserves accumulation; and (iii) government spending. Although in theory the threeshocks have potentially an effect on the size of the Mexican appreciation, we show thattheir quantitative impact is also minor.

328 F. Meza, C. Urrutia / Journal of International Economics 85 (2011) 317–328

Choudhri, E.U., Khan, M.S., 2005. Real exchange rates in developing countries: areBalassa–Samuelson effects present? IMF Staff Papers, p. 52.

De Gregorio, J., and H. Wolf, 1994, “Terms of trade, productivity and the real exchangerate,” NBER Working Paper, July.

Edwards, S., Van Wijnbergen, S., 1987. Tariffs, the real exchange rate and the terms oftrade: on two popular propositions in international economics. Oxford EconomicPapers 39, 458–464.

Engel, C., 1999. Accounting for U.S. real exchange rate changes. Journal of PoliticalEconomy 107, 507–538.

Garcia-Verdú, R. (2005), “Factor shares from household survey data,” Banco de MexicoWorking Paper, Number 5, October.

Gollin, D., 2002. Getting income shares right. Journal of Political Economy 110 (2),458–474.

Guerrieri, V., Acemoglu, D., 2008. Capital deepening and non-balanced economicgrowth. Journal of Political Economy 116, 467–498.

Kehoe, T.J., Ruhl, K.J., 2009. Sudden stops, sectoral reallocations and the real exchangerate. Journal of Development Economics 89, 235–249.

Mendoza, E.G., 2005. Real Exchange Rate Volatility and the Price of Nontradable Goodsin Economies Prone to Sudden Stops. Economía. (Fall).

Mendoza, E.G., and V. Yue (2008), “A solution to the default risk — business cycledisconnect”, mimeo.

Meza, F., 2008. Financial crisis, fiscal policy and the 1995 GDP contraction in Mexico.Journal of Money, Credit and Banking 40, 1239–1261.

Meza, F., Quintin, E., 2007. Factor utilization and the real impact of financial crises. TheB.E. Journal of Macroeconomics 7, 1 Advances.

Ngai, L.R., Pissarides, C., 2007. Structural change in a multisector model of growth. TheAmerican Economic Review 97, 429–443.

Pratap, S. and E. Quintin (2010), “Financial crises and labor market turbulence,”mimeo.Pratap, S., and C. Urrutia (2010), “Financial frictions and total factor productivity:

accounting for the real effects of financial crises”, mimeo.Ruhl, K.J. (2008), “The international elasticity puzzle,” mimeo.Stockman, A.C., Tesar, L., 1995. Tastes and technology in a two-country model of the

business cycle: explaining international comovements. The American EconomicReview 85 (1), 168–185.

Tornell, A., Westermann, F. (Eds.), 2005. Boom–Bust Cycles and Financial Liberalization.MIT Press Books, The MIT Press.

Tornell, A., F. Westermann, and L. Martinez (2004), “NAFTA and Mexico's less thanstellar performance,” NBER Working Paper 10289, February.