Selected Issues


Selected Issues

External Adjustment in Europe: Competitiveness, the Real Exchange Rate, and the Trade Balance1

Large and persistent competitiveness gaps within the euro area (EA), as captured by labor cost and productivity differentials, are often cited as contributing to the large external imbalances of some EA countries. Using a newly constructed dataset, we unpack developments in the real effective exchange rate (REER) on a unit labor cost (ULC) basis, which incorporates wages and labor productivity. We examine the contributions of the nominal effective exchange rate (NEER), own ULCs, trading partner ULCs (within and outside the EA), and underlying ULC components. There were large differences in own ULC inflation across EA countries prior to the crisis, exacerbating cost gaps. Despite some marked adjustments post crisis, gaps remain. Since the euro adoption, changes in trading partners outside the EA—but not within—have dominated trading partner ULC changes. We find evidence that countries’ ULC-based REER appreciations are correlated with lower trade balances, with the relationship stronger for EA countries. Unpacking the REER, different components exhibit different associations with the trade balance. Going forward, further declines in own ULC could enhance competitiveness, supporting external adjustment, but this should occur mainly through rises in total factor productivity (TFP), also boosting income.

A. Introduction

1. Competitiveness gaps between EA countries are often cited as obstacles to their external adjustment (ECB, 2012; Chen, Milesi-Ferretti, and Tressel, 2013). Several EA countries have recently had, or continue to have, large and persistent current account balances, whether deficits (such as Spain in the 2000s) or surpluses (such as Germany in the later 2000s up to today), leading to rising vulnerabilities to either sudden stops or adverse external wealth shocks. In parallel, differences in ULCs between EA countries grew post-euro adoption, with persistent deficit countries often seeing large rises while costs in persistent surplus countries tended to be relatively stable. This experience and the literature suggest that changes in competitiveness can play a role in facilitating external adjustment, helping to shrink or even reverse large and persistent current accounts and thereby reduce external vulnerabilities (IMF, 2015).

2. In this paper, we examine how competitiveness, measured by the relative ULC and its components, has evolved in the EA and how it is associated with the trade balance. To do so, we first construct the ULC-based REER bottom-up, allowing us to decompose each country’s REER and its changes into its bilateral elements (that is, vis-à-vis trading partners or trading partner groups). We are also able to further decompose the REER by component—the NEER, wages (either in common USD or in local currency units, LCUs, if unadjusted by nominal exchange rates and defined as compensation per unit of employment), and labor productivity (defined as real output per unit of employment).2 We then estimate the association of these REER developments (including by component) with the trade balance, which typically accounts for the bulk of the variability in the current account. In some robustness checks, we consider how the estimated association with the relative ULC changes when other cost and non-cost competitiveness controls, such as energy costs (an additional production input) and the investment climate, are included. We emphasize that the results presented here do not attempt to control for potential endogeneity, nor address questions or causation. Instead, the focus is on the simple historical, statistical association between the REER, its underlying components, and the trade balance, for countries either inside or outside the EA.

3. Some of the divergences in ULCs between EA countries have narrowed since the crisis, but the adjustment has relied heavily on wage and job cuts in persistent deficit economies. Since euro adoption the relative ULCs of EA countries have mostly been driven by own ULC developments, but ULC changes among trading partners outside the EA have also contributed importantly. Throughout, we will use the terms “REER,” “REER-ULC,” and “relative ULC” interchangeably. Drivers of the relative ULC have differed markedly across countries. In Germany, the major contributor is lower relative wage growth in both pre- and post-crisis periods. Post-crisis adjustment in some net external debtors (Greece, Portugal, Spain, Italy, and France) benefited, to varying degrees, from lower relative wage growth. Some other countries (such as Portugal and Spain) benefited more from higher relative productivity post-crisis. Relative labor productivity improvements before 2013 mainly reflected falling employment rather than rising output. Over 2014–15, the pace of relative productivity improvement moderated, but with different drivers across countries—in Spain, both relative employment and output rose, while in Italy, output growth continued to be slower than trading partners and there was little rise in relative employment.

4. REER appreciations are associated with lower trade balances. The statistical relationship is stronger for EA countries than for other advanced economies. Unpacking ULCs into wages and labor productivity (real output and employment) reveals that wage moderation and productivity rises (whether due to lower employment or higher real output) are associated with a rise in the trade balance.

5. Going forward, further declines in own ULCs by net debtor countries could enhance competitiveness and facilitate external adjustment. Much of the earlier REER adjustment by net external debtors has been accomplished through wage moderation and job cuts. In countries where nominal wages remain high and labor markets rigid, further wage moderation could help improve competitiveness.3 However, in general, a persistent boost in TFP growth would deliver more continuing competitiveness improvements and have the additional benefit of boosting income growth, making it a more socially desirable way to achieve adjustment over the medium-term. The onus would then lie on structural reforms and productivity-enhancing investments to raise countries’ growth potential, which would help foster convergence as well as potentially contribute to reductions in external imbalances.

B. Data Construction and Stylized Facts

6. The REER/relative ULC is constructed from the bottom-up in order to decompose it into components on a bilateral basis. To maximize coverage and ensure cross-country comparability, we primarily draw upon the Penn World Table (version 9.0) from 1985–2014, splicing data forward where possible through 2015 using a variety of datasets, including the Conference Board’s Total Economy Database, the IMF’s World Economic Outlook, and the European Commission’s (EC) AMECO database (further details are in Appendix I). The REER and NEER are defined as:


where i indexes countries, t indexes years, j indexes trading partners, e is the exchange rate in USD per local currency unit (LCU), P is the relevant price measure (here ULC), * indicates the variable is the weighted average of trading partners, and w is the weight on a given trading partner, such that ΣjWj = 1 and wj ∈ [0,1]∀j. Weights are taken from the IMF’s Statistics Department’s Information Notice System (INS) database, calculated from bilateral trade flows and domestic sales for commodities, manufacturing goods, and services.4 They capture how much a country competes with a given trading partner in particular markets, both at home and abroad. These log measures are additive in their weighted components, enabling us to construct various groupings of the trading partner factors by country (for example, EA and non-EA; EA net creditors, EA net debtors, and others). Recall that the ULC is defined to be the ratio of the nominal wage to real labor productivity. This implies that the log ULC may be further decomposed into the log nominal wage and log real labor productivity components, where the latter is the difference between log real output and log employment (see equation [3]). Thus, another expression for the REER/relative ULC is:


where the terms are the approximate growth rates of the NEER, relative wage, relative output, and relative employment, respectively, and c denotes the wage, Y is real output, and L is labor employed. Since all elements are denominated in common units (2011 PPP-adjusted international dollars), the REER may be compared in levels across countries.5

7. Some EA countries saw large rises in their own ULCs from 1999 to 2008, but these have partly unwound post-crisis. After euro adoption in 1999, ULCs (in LCU) increased in most EA countries (except Germany and the Netherlands), with nominal wages outpacing real labor productivity (Figure 1). ULC rises tended to be larger in the net external debtor countries among the EA-12 (such as Greece, Ireland, Italy, Portugal, and Spain) and other EA countries who joined the monetary union later. This pattern partly reversed, led by various degrees of improvement in labor productivity, mostly during 2009–13. In Greece, Portugal, and Spain, the ULC reduction was driven largely by labor shedding, accompanied by wage declines in the case of Greece. Since 2014, Spain’s ULC has been stable, with wage growth broadly in line with labor productivity (supported by job creation).6 In all countries apart from Greece, wages rose during 2013–15, but at a much slower pace than in previous periods.

Figure 1.
Figure 1.

Decomposition of Own Unit Labor Cost

(log change)

Citation: IMF Staff Country Reports 2017, 236; 10.5089/9781484312353.002.A003

Sources: Penn World Table and IMF staff calculations.Note: “Others” denotes average of 16 advanced economies outside the euro area.Countries are ordered by their 1999 ULC level.

8. Overall REER changes in the EA since 1999 are largely explained by changes in own ULC. Interestingly, among the EA-12, a change in own ULC (in LCU) almost fully translates into a change in the relative ULC. The correlation is much weaker though for other EA countries and other advanced economies, suggesting that, for those economies, the increases (decreases) in ULC may be partly offset by currency depreciation (appreciation) and/or increases (decreases) in foreign ULC (text figure, right).


ULC-based REER and ULC in LCU

(log change, 1999–2015)

Citation: IMF Staff Country Reports 2017, 236; 10.5089/9781484312353.002.A003

Sources: Penn World Table; Eurostat; INS; WEO; and IMF staff calculations.

9. Decomposing the REER into relative wage, output, and employment components reveals significant cross-country variation in underlying drivers since 1999. In Germany, REER developments were heavily affected by relative wages both before and after crisis. In Italy, stagnant growth worsened relative output, driving REER appreciation prior to the crisis and partly offsetting post-crisis moderation in relative wages and employment. In Spain, the pre-crisis REER appreciation reflected growing relative wages and shrinking relative productivity but this pattern is now reversing. In Greece, the appreciation largely reflected increasing relative wages, but now the REER is falling from reductions in relative wages and employment (Figure 2).

Figure 2.
Figure 2.

Decomposition of Relative Unit Labor Cost (REER)

(log level, 1999=0, in percent)

Citation: IMF Staff Country Reports 2017, 236; 10.5089/9781484312353.002.A003

Sources: Penn World Table 9.0; AMECO; Total Economy Database; WEO; IMF staff calculation.

10. Other EA trading partner ULC changes account for little of the overall REER change at the EA country level since the euro adoption. Over this period, all major EA countries (except Italy) experienced a REER depreciation, with Germany and Ireland seeing the largest declines. Both own ULC and the ULC of trading partners outside the EA have tended to be the most important components of REER changes, despite the majority of EA country competition in trade typically occurring between EA countries (see below). This difference in adjustment vis-à-vis EA/non-EA trading partners is there even when the NEER is taken account of (text figure, right).


Real Effective Exchange Rate

(log change, in percent, 1999–2015)

Citation: IMF Staff Country Reports 2017, 236; 10.5089/9781484312353.002.A003

Sources: Penn World Table 9.0; AMECO; Total Economy Database; WEO; staff calculation.

11. Delving into the bilateral aspects of the REER reveals that, apart from Germany and Ireland, the majority of EA countries’ competition in trade is with other EA countries. Focusing on a subset of EA countries and using averages over 1999–2015, trade weights vis-à-vis other EA countries range from about 40 percent (Germany) to closer to 75 percent (Portugal; text figures, below). Importantly, any components of the REER vis-à-vis other EA countries do not benefit from a flexible, bilateral nominal exchange rate. In these cases, bilateral external competitiveness is solely a function of relative prices in local currency.

Sources: IMF Information Notice System (INS) and IMF staff calculations.

12. The scope for simultaneous external adjustment by EA countries may vary with their exposure to net external debtors versus creditors within the EA. Trade weights against other EA countries can be further broken down according to the net foreign asset position of trading partners (text figure, above).7 For Ireland, Italy, France, and the Netherlands, their intra-EA trade weights are mostly vis-à-vis EA net external creditors. By contrast, for Germany, Greece, Portugal, and Spain, it is mostly vis-à-vis EA net external debtors. This complicates the adjustment for Greece, Portugal, and Spain, since other net external debtors will also tend to need to depreciate their REER, dampening each other’s relative price adjustment. Although Germany and the Netherlands are significant trading partners for EA net external debtors, they do not account for the majority of intra-EA trade competition for net external debtors. Hence, although greater inflation or rising ULCs in Germany and the Netherlands would contribute to REER depreciations among EA net external debtors, relative price adjustment vis-à-vis the rest of the EA (such as against other net external debtors) is also important.

C. Econometric Analysis and Findings

13. To gauge how the REER adjustment is related to trade adjustment, we estimate the statistical relationship between REER changes and the trade balance (relative to GDP). There is a large literature using various methods to attempt to get at the underlying causal relationship between the REER and the trade balance or the related current account. For example, using a panel VAR model, Diaz Sanchez and Varoudakis (2013) argue that domestic demand shocks have been the most important drivers of current accounts in Europe, although they also find that cost competitiveness has mattered. Relatedly, Wyplosz (2013) argues that competitiveness changes are endogenous to demand shocks, suggesting that competitiveness is a symptom rather than cause. By contrast, Zemanek, Belke, and Schnabl (2010) find that structural reforms can have a large impact on current accounts in Europe, suggesting that policy-driven productivity changes can improve competitiveness and reduce current account deficits. We do not attempt to disentangle these various channels nor argue for one identification scheme over another to determine causation. Instead, we focus on the simple statistical association between the REER, its components, and the trade balance, for countries either inside or outside the EA.

14. The estimation sample consists of 35 advanced economies over the period from 1985–2015 with data at an annual frequency.8 The baseline linear regression model employed takes the form (equation [4]):


where TB denotes the balance on goods and services (as a percent of GDP), REER denotes the real effective exchange rate on a ULC basis, EA is a dummy variable, taking the value of one if country i in year t is a EA country, and α are a set of country and time fixed effects, controlling for unobserved, time-invariant heterogeneity at the country level and common unobserved shocks, respectively.9 ∊ is a mean-zero, but possibly heteroskedastic and auto- and cross-correlated error term.10 In general, we expect the β coefficients to be negative, implying that an increase in REER (appreciation) is negatively associated with the trade balance. This relationship could differ between currency union members and those outside.

15. In further specifications, we consider decompositions of the REER terms, to estimate how the REER components are related to the trade balance. The breakdowns substitute in the definitions for REER and its components seen in equations [1], [2], and [3], allowing each separate term to have its own estimated coefficient. In addition to the components of the relative ULC/REER, we also attempt to see if the findings are robust to controlling for other cost competitiveness and non-cost competitiveness variables—in particular, we consider local electricity costs (a key input in many industries; from the International Energy Agency) and the investment climate (an amalgam of indices assessing contract enforcement, expropriation risk, ability to repatriate profits, and payments delays; from the PRS Group).

16. The results suggest that a rising relative ULC is associated with shrinking trade balances and that this correlation is stronger for EA countries (Table 1). In the simplest model specification, the estimates indicate that a one percent depreciation in a EA country’s relative ULC is associated with a statistically significant 0.15 percentage points increase in trade balances. By contrast, for non-EA countries, the associated change is less than 0.1 percentage point. NEER depreciation and own ULC are also found to have statistically significant relationships to trade balance changes regardless of EA membership, although the magnitude is slightly larger for EA countries. Consistent with the perspective that it is relative price movements that drive international flows, an increase in foreign ULCs is associated with a rise in the trade balance, but the estimated coefficient is statistically insignificant.

Table 1.

Trade Balance, Real Effective Exchange Rate and Its Components

article image
Note: Sample covers advanced economies as defined by the IMF’s WEO (35 in total, including all current EA-19 member states). For specifications 5 and 6, local electricity cost data is missing for some countries, dropping the sample to 30 countries. The data are annual, spanning 1985–2015 (30 years). The euro area (EA) results are based only on those countries and years in which they were a member of the euro area (for the EA-11, from 1999; others joined later and in different years). Standard errors are in parentheses under the coefficient estimates and are clustered along two dimensions (country and year) to account for heteroskedasticity, autocorrelation within country, and cross-sectional correlation within year. Statistical signficance levels are denoted by * for 10 percent, ** for 5 percent, and *** for 1 percent.

17. Further decomposing the relative ULC suggests that own wage and employment changes are significantly associated with external adjustment. In general, NEER changes are a statistically significant negative correlate for trade balance changes regardless of additional covariates, most consistently for the EA countries. In other words, nominal exchange rate appreciations are associated with shrinking trade balances, in line with priors. Foreign variables tend to have the expected sign, but are typically not statistically significant—higher foreign wages and lower foreign productivity tend to lower the relative ULC, leading to a positive association with the trade balance. Higher own wages (LCU) have a relatively robust negative relationship to the trade balance, with a one percent rise associated with about a 0.2 percentage point decline in the trade balance for EA countries. Own employment rises (which lower labor productivity holding real output constant) also have a relatively robust negative relationship, while own real output (raising productivity and controlling for employment, likely reflecting capital deepening and TFP effects) has a positive, but often statistically insignificant relationship with the trade balance. Overall, own labor productivity improvements appear to be associated with trade balance increases (particularly when controlling for energy costs and the investment climate).

D. Concluding Remarks

18. Since EA countries’ competition in trade is largely with other EA countries, the nominal exchange rate may have a more limited role in external adjustment. However, NEER adjustment was still an important component of REER adjustment vis-à-vis non-EA trading partners in both the pre- and post-crisis periods, accounting for similar amounts of REER changes across EA countries. That said, monetary union members may have to rely more on relative price adjustment (whether via LCU inflation or ULC differentials) vis-à-vis all trading partners, both inside and outside the EA, in order to durably improve their competitiveness and adjust their current accounts.

19. Allowing for greater inflation and productivity growth differentials within the EA would help improve some countries’ competitiveness, but is likely no panacea. The stylized facts suggest that, for the REER, relative price adjustment against non-EA trading partners is often nearly as important as that against EA trading partners. Above EA average inflation in Germany and other net external creditors would mechanically help the REER-ULC adjustment of EA net external debtors, by lowering their relative wage growth (assuming EA net creditors’ price inflation would also be reflected in their wage inflation). However, it should also be accompanied by lower own-country ULC growth in EA net external debtors—wage moderation and higher productivity growth—relative to their trading partners, both inside and outside the EA.

20. Further improvements in own labor productivity by EA net external debtors could have the double impact of encouraging REER adjustment as well as boosting income growth. Although supply-side measures to lift productivity could have an ambiguous impact on the trade balance (and current account), by raising investment and import demand, the results here suggest that on balance, there has been a positive historical association between productivity and the trade balance for the EA countries. Net external debtors’ REER adjustment post-crisis has largely occurred through wage moderation and job cuts. In countries where labor market rigidities remain prevalent, further wage moderation could help improve competitiveness. But, in general, raising TFP growth is preferable to achieve adjustment over the medium term, as it has the scope to deliver continuing competitiveness gains while increasing income growth. In that case, structural reforms and productivity-enhancing investments to raise countries’ growth potential are key, helping to foster convergence as well as potentially contribute to reductions in external imbalances.


  • Bayoumi, Tamim, Jaewoo Lee, and Sarma Jayanthi, 2005, New Rates from New Weights, IMF Working Paper, no. 05/99.

  • Chen, Ruo, Gian Maria Milesi-Ferretti, and Thierry Tressel, 2013, External imbalances in the Eurozone, Economic Policy 28(73): 101142.

    • Search Google Scholar
    • Export Citation
  • Choi, In, 2001, Unit root tests for panel data, Journal of International Money and Finance, 20: 249272.

  • International Monetary Fund (IMF), 2015, Chapter 1: Recent labor market reforms: a preliminary assessment, by Mai Dao, Spain 2015 Article IV Consultation: Selected Issues.

    • Search Google Scholar
    • Export Citation
  • Diaz-Sanchez, Jose L. and Aristomene Varoudakis, 2013, Growth and competitiveness as factors of Eurozone external imbalances,” Policy Research Working Paper, no. WPS6732, World Bank.

    • Search Google Scholar
    • Export Citation
  • European Central Bank, 2012, Competitiveness and external imbalances within the euro area, ECB Occasional Paper Series, no. 139, December.

    • Search Google Scholar
    • Export Citation
  • Feenstra, Robert C., Robert Inklaar, and Marcel P. Timmer, 2015, The Next Generation of the Penn World Table, American Economic Review, 105(10): 31503182.

    • Search Google Scholar
    • Export Citation
  • Goldstein, Morris, and Mohsin S. Khan, 1985, Income and price effects in foreign trade, Handbook of International Economics, Volume 2, p. 10411105.

    • Search Google Scholar
    • Export Citation
  • Im, Kyung So, M. Hashem Pesaran, and Yongcheol Shin, 2003, Testing for unit roots in heterogenous panels, Journal of Econometrics, 115: 5374.

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund, 2015, Exchange rates and trade flows: disconnected?”, World Economic Outlook Chapter 3, October.

  • Spilimbergo, Antonio, Daniel Leigh, Ioannis Halikias, Jörg Decressin, Michael Kumhof, Paulo A. Medas, Prakash Loungani, Raphael A. Espinoza, Susanna Mursula, and Tengteng Xu, 2015, Wage Moderation in Crises: Policy Considerations and Applications to the Euro Area, IMF Staff Discussion Note, no. 15/22, November.

    • Search Google Scholar
    • Export Citation
  • Westerlund, Joakim, 2007, Testing for error correction in panel data, Oxford Bulletin of Economics and Statistics, 69: 709748.

  • Wyplosz, Charles, 2013, the Eurozone crisis and the competitiveness legend, Asian Economic Papers, Fall 12(3): 6381.

  • Zemanek, Holger, Ansgar Belke, Gunther Schnabl, 2010, Current account balances and structural adjustment in the euro area, International Economics and Economic Policy, May 7(1): 83127.

    • Search Google Scholar
    • Export Citation

Appendix I. Data Definition, Sources, and Country Coverage

Data definitions

Balance on goods and services over GDP: (exports of goods and services – imports of goods and services)/GDP.

Electricity cost: total price per MWh, for industry, in current year PPP USD.

Investment profile index: An assessment of factors affecting the risk to investment that are not covered by other political, economic and financial risk components. The risk rating assigned is the sum of three subcomponents, each with a maximum score of four points and a minimum score of 0 points. A score of 4 points equates to Very Low Risk and a score of 0 points to Very High Risk. The subcomponents include: contract viability/expropriation, profits repatriation, payment delays. Log level is used.

Unit labor cost (ULC): The log ULC is constructed as the log labor share of income plus the log of the Penn World Table v. 9.0 (PWT) GDP price level plus the PWT PPP GDP deflator. That is, ln ULC = ln(labsh) + ln(pl_gdpo) + ln(cgdpo) − ln(rgdpo), using PWT variable names.1 See Feenstra, Inklaar, and Timmer (2015) for further details.

  • labsh: share of labor compensation in GDP at current national prices

  • pl_gdpo: price level of GDP (USA = 1 in 2011), equal to the PPP (ratio of nominal GDP to cgdpo) divided by the nominal exchange rate

  • cgdpo: real GDP at current PPPs (in mil. 2011 US$)

  • rgdpo: real GDP at chained PPPs (in mil. 2011 US$)

Data Sources

article image

Country Coverage

article image
Note: Countries with an asterisk (*) do not have data on electricity cost.

Appendix II. Econometric Approach

In the first step, we assessed the stationarity properties in the estimation sample of all the variables in levels via a set of panel unit root tests, under the null that all panels (countries) contain a unit root, including:

  • Im, Pesaran, and Shin (2003) test, accounting for country-specific AR parameters, panel means, and a time trend, based on the Augmented Dickey-Fuller test.

  • A Fisher-type test due to Choi (2001), accounting for country-specific AR parameters, panel means, a time trend, and a single lag, based on the Phillips-Perron test.

Tests for the trade balance (balance on goods and services) relative to GDP, the key dependent variable, fails to reject the null. Similarly, tests of the log REER-ULC and its components in level terms generally fail to reject the null of unit roots. Consequently, we also tested for cointegration of the trade balance with the REER-ULC using the Westerlund (2007) suite of panel cointegration tests, which take no cointegration as their null. These tests generally fail to reject the null, leading us to specify the regression models in first differences. We also experimented with distributed lag and lagged dependent variable specifications for the change in the trade balance to GDP. The estimates generally revealed little statistically significant effect beyond the contemporaneous impact amongst the distributed lag terms (up to two years were considered) and little role for the lagged dependent variable, leading us to make the specification in first differences with the contemporaneous impact solely considered. A unit root for the log of the investment profile measure of institutional quality was rejected, so it was included in log level terms in the specification where it appears.


John Bluedorn and Huidan Lin (EUR). Xiaobo Shao provided outstanding research assistance. We would like to thank staff from the European Commission for their helpful comments and feedback.


Wages and labor productivity are defined on a per worker basis in the core analytic work. We also considered these variables defined on an hourly basis in a robustness check, finding that the results were generally similar to what is presented here.


Spilimbergo and others (2015) illustrate how wage moderation during a crisis can generate the additional benefit of a positive effect on output, helping to buffer the economy against adverse shocks.


See Bayoumi, Lee, and Jayanthi (2005) for full details on the weights construction. In essence, the weights capture how much country i and country j tend to compete with each other in trade. The weight for country i of country j thus depends on the presence of country j in a typical global market for k (for example, commodities, manufacturing goods, and services) and the importance of this market k to country i.


For the EA countries, the correlation between our ULC-based REER measure, constructed vis-à-vis all trading partners, and that from Eurostat relative to 28 trading partners is strong, with the average correlation coefficient of 0.7 across countries during 1999–2015.


There is evidence that the 2012 labor market reforms implemented in Spain contributed to wage moderation and employment growth and made the labor market more resilient to shocks (IMF, 2015).


As of end-2015, based on the External Wealth of Nations II database (updated, 2017), the net external creditor countries in the EA are: Austria, Belgium, Germany, Luxembourg, Malta, and the Netherlands whose net international investment position was positive. The rest of the EA are classed as net external debtors.


The definition of an advanced economy comes from the IMF’s latest World Economic Outlook—see Appendix I for a full listing.


Panel unit root tests and finding a lack of cointegration led to the decision to use differences. Moreover, tests for a distributed lag structure suggest that focus on the contemporaneous relationship is appropriate. We also considered models similar to Goldstein and Khan (1985), where log level exports and imports are considered separately and include external and domestic demand controls respectively. However, given the pervasive endogeneity issues, we decided to focus on the trade balance in a simple and transparent linear regression, emphasizing that the estimated coefficients should be interpreted solely as indicative of statistical associations rather than any causal effect. See Appendix II for further details on the specification choices.


Standard errors are clustered along the country and time dimensions to account for possible heteroskedasticity, autocorrelation within country, and correlation across countries at the same point in time.


The equation is derived from the following: ln(ULC) = ln(wage per person, in USD) − ln(rgdpo) + ln(employment) = {ln(wage per person) + ln(employment) − ln(nominal GDP)} + ln(nominal GDP) − ln(rgdpo) = ln(labsh) + {ln(nominal GDP) – ln(cgdpo)}+ ln(cgdpo) – ln(rgdpo) = ln(labsh) + ln(pl_gdpo) + ln(cgdpo) − ln(rgdpo).

Euro Area Policies: Selected Issues
Author: International Monetary Fund. European Dept.