A. Introduction

1. In early and mid-2005, policymakers and market participants were focused on the possible breach of the “strong” edge of Hungary’s exchange rate band with the euro. The Czech koruna, the Hungarian forint, the Polish zloty, and the Slovak koruna—the currencies of the four central European (CE-4) new members of the European Union—had appreciated markedly since January 2004 (Figure 1).² The forint was moving in the range of Ft 241-250 per euro. There was even a concern that speculators may employ strategies to gain from temporary breaches of the strong edge of the band at Ft 240.01 per euro. The Magyar Nemzeti Bank (MNB), in its August 2005 Quarterly Report on Inflation, inferred that markets viewed the prospects of the new members with a high degree of optimism (p. 15):

CE-4: Currency Appreciation vs. Euro, Jan. 2004 - June 2005

(January 1, 2004 = 100; increase denotes depreciation)

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Source: Blomberg

• Download Figure
• Download figure as PowerPoint slide

View Full Size

CE-4: Currency Appreciation vs. Euro, Jan. 2004 - June 2005

(January 1, 2004 = 100; increase denotes depreciation)

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Source: Blomberg

• Download Figure
• Download figure as PowerPoint slide

CE-4: Currency Appreciation vs. Euro, Jan. 2004 - June 2005

(January 1, 2004 = 100; increase denotes depreciation)

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Source: Blomberg

• Download Figure
• Download figure as PowerPoint slide

• “Trends in the global market also affected CEE markets. On certain occasions, there was strong co-movement between the Czech, Polish, Hungarian and Slovak currencies, whilst country-specific factors were pushed into the background. Price and yield data as well as investment banks’ analyses suggest that the risk perception of the new EU Member States has been even more favourable than the general perception of emerging markets.”

The implication was that country-specific features and fundamentals were overwhelmed by the sanguine view of the region, to the point that markets were willing to discount well-known weaknesses, such as Hungary’s large fiscal and current account deficits.

2. Since September 2005, however, the forint has weakened considerably, and its volatility has increased. Between September 2005 and June 2006, the forint weakened from Ft 244 per euro to Ft 282 per euro, a depreciation of 13½ percent (Figure 2). The forint’s volatility also increased, particularly in the months of March and June 2006, when the coefficient of variation (standard deviation/mean) increased to 2 percent and 3 percent, respectively, compared with an average of 0.6 percent in the previous two years (Figure 3). The forint’s weakening, especially relative to the other CE-4 currencies, has been attributed, in part, to Hungary’s weak fundamentals: the country-specific factors are, apparently, no longer “in the background.” The MNB’s May 2006 Quarterly Report on Inflation reflected this shift in sentiment (p. 10):

CE-4: Currencies vs. Euro, Sept 2005 - June 2006

(September 1, 2004 = 100; increase denotes depreciation)

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Source: Bloomberg.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

CE-4: Currencies vs. Euro, Sept 2005 - June 2006

(September 1, 2004 = 100; increase denotes depreciation)

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Source: Bloomberg.

• Download Figure
• Download figure as PowerPoint slide

CE-4: Currencies vs. Euro, Sept 2005 - June 2006

(September 1, 2004 = 100; increase denotes depreciation)

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Source: Bloomberg.

• Download Figure
• Download figure as PowerPoint slide

CE-4: Historical Currency Volatility, 2004-06 ^1/

(Coefficient of variation, 30-day moving window, in percent)

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Sources: Bloomberg; and IMF staff calculations.^1/ Defined as the 30-day standard deviation divided by the 30-day average, times 100.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

CE-4: Historical Currency Volatility, 2004-06 ^1/

(Coefficient of variation, 30-day moving window, in percent)

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Sources: Bloomberg; and IMF staff calculations.^1/ Defined as the 30-day standard deviation divided by the 30-day average, times 100.

• Download Figure
• Download figure as PowerPoint slide

CE-4: Historical Currency Volatility, 2004-06 ^1/

(Coefficient of variation, 30-day moving window, in percent)

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Sources: Bloomberg; and IMF staff calculations.^1/ Defined as the 30-day standard deviation divided by the 30-day average, times 100.

• Download Figure
• Download figure as PowerPoint slide

• “From September 2005 ...financial markets were concerned about the developments and fundamental risks and the high budget and current account deficits. Hungarian and regional trends began to diverge at that point.”

The spikes in the forint’s volatility also coincided with heightened global emerging market (EM) volatility. The benefit of EU membership proved insufficient to insulate the CE-4 currencies—and particularly the forint—from weakness.

3. This selected issues chapter has three objectives. First, it aims to statistically separate the common and country-specific components of CE-4 currency movements. Second, it further decomposes the CE-4’s common factor into the part explained by global EM sentiment, and the part that is unique to the CE-4. Finally, it analyzes the trends in these three components, to understand past currency developments, as well as to reflect on near-term prospects.

4. Currency fluctuations are decomposed using two different approaches. The first approach, the state space model, assumes that CE-4 currency movements are formed of two unobserved components: one that is common to all four countries and another that is country specific. These unobserved components are derived using Kalman filter and maximum likelihood estimation techniques following specification of their dynamics and expressing the model in state space form. The second approach, the global factor model, takes the set of global emerging markets and assumes that currency movements are composed of three parts: one that is common to all emerging markets, one that is common to each region, and one that is country specific. The advantage this approach holds over the first model is that it can shed light on the extent to which the comovement in CE-4 currencies is due to region-specific, as opposed to global, factors.

5. The paper reaches three conclusions. First, the study confirms a strong comovement—specifically, a common trend toward appreciation—among the CE-4 currencies. However, the importance of this common component varies for the four countries. While it accounts for about 90 percent of the variation in the Czech and Slovak koruny and the Polish zloty, it explains less than half of the variation in the forint. Second, although country-specific movements have weakened the forint, it has benefited from being in the CE-4 region: if it had not taken part in the regional appreciation, the currency would have weakened past the central parity as early as March 2006, and would have been close to Ft 300 per euro by end-June 2006. Third, the CE-4’s common appreciation over the past three years can be segmented into two subperiods. From mid-2003 to end-2004, there was a strong, region-specific appreciation, the tail end of a long and steady appreciation that began in mid-2000 and that can be dubbed an “EU accession effect.” But since 2005, this region-specific appreciation has ended, and any subsequent appreciation in CE-4 currencies has been due to the strength of emerging markets more generally vis-à-vis the euro.

6. The rest of the chapter proceeds as follows. Section B lays out the state space model of CE-4 currency movements, describes the data used, and analyzes the estimation results. Section C presents the global factor model EM currency movements, compares the CE-4 estimates with those obtained from the state space model, and decomposes the CE-4 common component into its regional and global parts. Section D takes a brief look at the possible factors that move the various components, and Section E concludes.

B. The State Space Model of CE-4 Currencies

7. The first model used in this chapter, the state space model, focuses solely on decomposing CE-4 currency movements into common and country-specific components. The “unobserved components” dynamic factor analysis used here is based on maximum likelihood Kalman filtering (Engle and Watson 1981; Harvey 1989; and Cuthbertson, Hall, and Taylor 1992). This methodology has frequently been used to identify unobserved components, including by Stock and Watson (1991) to identify the state of the business cycle, by Fama and Gibbons (1982) to study the behavior of the ex ante real interest rate, and by Mody and Taylor (2006) to identify East Asian “common vulnerabilities.” Let y_it be the exchange rate of country i at time t, expressed in local currency units per euro, where i = 1 (Czech Republic), 2 (Hungary), 3 (Poland), or 4 (Slovak Republic). There is an unobserved component, K_t, that is common to all CE-4 currencies, and a second component, η_it, that is idiosyncratic or country specific:

where γ⁽ⁱ⁾ is the country-specific sensitivity of country i to the common component; that is, the influence of the common component may vary from country to country.

8. Both the common and country-specific unobserved components are assumed to follow autoregressive processes. The persistence of the country-specific factors is allowed to vary across countries:

The innovations ω_t and ε_it are assumed to be Gaussian and orthogonal, and the variance of ω_t is normalized to unity in order to achieve identification of the common component:

The system expressed in equations (2)-(4) can be written in state space form as follows:

9. The Kalman filter provides an algorithm for estimating the parameters. Once a system has been expressed in state space form as in (7)-(8), the Kalman filter can be used to calculate the exact likelihood function from which the parameter estimates, as well as the unobserved components, can be derived. A full description of state space models and the Kalman filter can be found in Hamilton (1994).

10. The model is estimated using daily CE-4 exchange rates versus the euro, covering the period from 2003 to end-June 2006. Daily exchange rate data were taken from Bloomberg, and days with missing data (e.g., holidays) were filled in with the previous day’s value. The estimation was performed on demeaned exchange rate series, using the S-space object in EViews. To ensure that the variances are nonnegative, ${\sigma }_1^2$ through ${\sigma }_4^2$ were reparameterized, using ${\rho }_i\equiv In\left({\sigma }_i^2\right),i=1,2,3,4.$

11. The state space model estimates show very persistent unobserved components. Both the common and country-specific components in the CE-4 currencies are highly persistent, with autoregressive parameters very close to one (Table 1). This result is not surprising in light of the well-known result, first documented by Meese and Rogoff (1983a and 1983b), that exchange rates are best modeled as a random walk.

Table 1.

State Space Parameter Estimates

Source: IMF staff calculations.

^1/The country-specific variance, σ_i², is computed as σ_i ² = exp(ρ_i).

Table 1.

State Space Parameter Estimates

Autoregressive Parameters			Country Sensitivities			Innovation Variances
Parameter	Estimate	Sth. error	Parameter	Estimate	Sth. error	Parameter	Estimate	Sth. error	Implied σ21/
ϕ	0.999	0.002	-	-	-	-	-	-
ψ(1)	0.964	0.016	γ(1)	0.061	0.004	ρ1	-4.977	0.052	0.083
ψ(2)	0.999	0.002	γ(2)	0.644	0.042	ρ2	0.122	0.042	1.063
ψ(3)	0.985	0.006	γ(3)	0.013	0.001	ρ3	-7.869	0.051	0.020
ψ(4)	0.995	0.003	γ(4)	0.071	0.003	ρ4	-5.141	0.065	0.076

Source: IMF staff calculations.

^1/The country-specific variance, σ_i², is computed as σ_i ² = exp(ρ_i).

View Table

Table 1.

State Space Parameter Estimates

Autoregressive Parameters			Country Sensitivities			Innovation Variances
Parameter	Estimate	Sth. error	Parameter	Estimate	Sth. error	Parameter	Estimate	Sth. error	Implied σ21/
ϕ	0.999	0.002	-	-	-	-	-	-
ψ(1)	0.964	0.016	γ(1)	0.061	0.004	ρ1	-4.977	0.052	0.083
ψ(2)	0.999	0.002	γ(2)	0.644	0.042	ρ2	0.122	0.042	1.063
ψ(3)	0.985	0.006	γ(3)	0.013	0.001	ρ3	-7.869	0.051	0.020
ψ(4)	0.995	0.003	γ(4)	0.071	0.003	ρ4	-5.141	0.065	0.076

Source: IMF staff calculations.

^1/The country-specific variance, σ_i², is computed as σ_i ² = exp(ρ_i).

12. The common component explains a large portion of the variation in CE-4 currencies, but its importance varies across countries, with the Hungarian forint being the outlier. Because the common and country-specific components are not constructed to be orthogonal, they have a nonzero correlation, and a perfect variance decomposition is not possible. We measure the importance of the common component in two different ways. First, we perform the variance decomposition $\mathrm{var}\left(y_{it}\right)=\mathrm{var}\left({\gamma }^{\left(i\right)}{k}_t\right)+\mathrm{var}\left({\eta }_{it}\right)+\mathrm{cov}({\gamma }^{\left(i\right)}{k}_t,{\eta }_{it}),$ but we ignore the covariance term (or equivalently, reallocate it to the first two terms based on their relative variances), so that the share of the variance explained by the common component is given by $\mathrm{var}\left({\gamma }^{\left(i\right)}{k}_t\right)/(\mathrm{var}({\gamma }^{\left(i\right)}{k}_t)+\mathrm{var}({\eta }_{it}\left)\right)$ (Table 2, first column). Second, we regress each CE-4 currency on the common component and a constant, and get the R-squared from the regression (Table 2, second column).³ By both measures, the common component explains a very high proportion of the variation in the Czech koruna, the Polish zloty, and the Slovak koruna. But the common component accounts for a much smaller share of the variance of the Hungarian forint.

Table 2.

CE-4 Currencies: Share of Variance Explained by the Common Component

Source: IMF staff calculations.

^1/Defined as $\mathrm{var}\left({\gamma }^{\left(i\right)}{k}_t\right)/(\mathrm{var}({\gamma }^{\left(i\right)}{k}_t)+\mathrm{var}({\eta }_{it}\left)\right)$

^2/Defined as the R -squared in a regression of a country’s exchange rate on the extracted common component.

Table 2.

CE-4 Currencies: Share of Variance Explained by the Common Component

	Share in Total Variance 1/	R-squared of Common Component 2/
Czech koruna	0.97	0.97
Hungarian forint	0.45	0.01
Polish zloty	0.90	0.90
Slovak koruna	0.86	0.84

Source: IMF staff calculations.

^1/Defined as $\mathrm{var}\left({\gamma }^{\left(i\right)}{k}_t\right)/(\mathrm{var}({\gamma }^{\left(i\right)}{k}_t)+\mathrm{var}({\eta }_{it}\left)\right)$

^2/Defined as the R -squared in a regression of a country’s exchange rate on the extracted common component.

View Table

Table 2.

CE-4 Currencies: Share of Variance Explained by the Common Component

	Share in Total Variance 1/	R-squared of Common Component 2/
Czech koruna	0.97	0.97
Hungarian forint	0.45	0.01
Polish zloty	0.90	0.90
Slovak koruna	0.86	0.84

Source: IMF staff calculations.

^1/Defined as $\mathrm{var}\left({\gamma }^{\left(i\right)}{k}_t\right)/(\mathrm{var}({\gamma }^{\left(i\right)}{k}_t)+\mathrm{var}({\eta }_{it}\left)\right)$

^2/Defined as the R -squared in a regression of a country’s exchange rate on the extracted common component.

13. The common component contains a trend appreciation that has supported the CE-4 currencies during 2004-06. After weakening somewhat in 2003, the common component appreciated steadily from February 2004 until March 2006 (Figure 4, bottom line in each panel).⁴ This regional appreciation strengthened the Czech koruna by 13 percent, the forint by 11 percent, the zloty by 18 percent, and the Slovak koruna by 8 percent during this period.

CE-4: Decomposition of Currency Movements into Common and Country-Specific Components versus the Euro, 2003-06

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Sources: Bloomberg; and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

CE-4: Decomposition of Currency Movements into Common and Country-Specific Components versus the Euro, 2003-06

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Sources: Bloomberg; and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

CE-4: Decomposition of Currency Movements into Common and Country-Specific Components versus the Euro, 2003-06

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Sources: Bloomberg; and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

14. Over the past three years, country-specific factors modestly strengthened the Czech and Slovak koruna, weakened the zloty slightly, and weakened the forint substantially. The country-specific components of the Czech and Slovak koruna have added 2½ percent and 1¼ percent, respectively, to the total appreciation since 2003. In contrast, country-specific factors have weakened the zloty by 7 percent over the same period. For Hungary, the country-specific component of the forint has weakened by 22 percent over that period.

15. Yield differentials can explain country-specific currency movements only for Hungary, and even there only for forint movements before September 2005. Scatterplots of country-specific currency movements against three-month interest rate differentials vis-à-vis the euro area show at most a weak link between the two (Figure 5), and regressions of the former on the latter yield R-squareds of less than 10 percent for the Czech Republic, Poland, and the Slovak Republic. For Hungary, the link is stronger—the regression R-squared is 54 percent—suggesting that the tight monetary policy stance adopted in 2004 played a role in keeping the forint strong, and that the gradual easing in early 2005 played some role in the forint’s weakening. However, the upper-right scatterplot of Figure 5 also makes clear that the weakening of the forint between September 2005 and June 2006 was unrelated to any narrowing of the interest rate differential.

CE-4: Scatterplots of Country-Specific Currency Movements Against Interest Rate Differentials vis-à-vis the Euro Area, 2003-06

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Sources: Bloomberg; and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

CE-4: Scatterplots of Country-Specific Currency Movements Against Interest Rate Differentials vis-à-vis the Euro Area, 2003-06

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Sources: Bloomberg; and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

CE-4: Scatterplots of Country-Specific Currency Movements Against Interest Rate Differentials vis-à-vis the Euro Area, 2003-06

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Sources: Bloomberg; and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

C. The Global Factor Model of Emerging Market Currencies

16. The state space model proves unwieldy in further disentangling the global and regional factors driving the common component in the CE-4. To identify global and regional components of currency movements one needs to move beyond the CE-4 and include the wider set of emerging market currencies. While in principle a state space model could also be applied to this larger universe of currencies by specifying the dynamics of the global, regional, and country factors, the large number of parameters makes maximum likelihood estimation of this larger state space model unreliable. Moreover, the alternative approach adopted here to achieve that decomposition, the global factor model, provides a point of comparison with the results of the state space model, and overlapping findings enhance confidence that the results are not driven by the method employed.

17. To identify global and regional factors, we use the following global factor model:

where Δ log(y_it) is the change (log difference) in the exchange rate y_it of country i at time t, α_t is a global factor common to all EM currencies, ${\beta }_t^{\left(j\right)}$ is the regional factor common to countries in region j, and ε_it is the country-specific component. Similar factor models—often run on industry- or firm-level data and containing industry factors in addition to country factors—have been used to model comovements in industrial output (Stockman, 1988), employment (Marimon and Zilibotti, 1998), and stock returns (Brooks and Catao 2000; and Brooks and Del Negro 2004).

18. The global factor model is estimated using cross-sectional ordinary least squares regressions, without imposing any dynamic specification on the factors. Specifically, for each time period t the exchange rates for 35 emerging markets are regressed on a constant (α_t) and a full set of regional dummies D_j (which generates ${\beta }_t^{\left(j\right)}$); the residual provides the estimate of the country-specific component, ε_it. Because a full set of dummies will be perfectly collinear with the constant term, instead of dropping one of the regional dummies to achieve identification we constrain the coefficients ${\beta }_t^{\left(j\right)}$ to sum to zero in each period $(\underset{i\in region\left(j\right)}{\mathrm{\Sigma }}{\beta }_t^{\left(j\right)}=0,\forall t)\mathrm{.}$ This allows us to interpret the magnitude of the regional component ${\beta }_t^{\left(j\right)}$ as region’s over- or underperformance relative to the global average.

19. The global factor model is estimated using daily exchange rates for the 35 emerging market countries listed in Table 3. As before, daily exchange rates were obtained from Bloomberg, and days with missing data (e.g., holidays) were filled in with the previous day’s value. The period covered is 1997 to 2006 to provide a long-term perspective. To allow comparisons with the state-space model results in the previous section, we continue to use the euro as our benchmark currency. Switching to U.S. dollar exchange rates only affects the global factor, since the euro per U.S. dollar cross rate will be common to all the currencies; all the estimates of the regional and country-specific factors will be identical.

Table 3.

Emerging Markets Included in Global Factor Model

Table 3.

Emerging Markets Included in Global Factor Model

Argentina	Indonesia	Philippines
Brazil	Israel	Poland
Bulgaria	Jordan	Romania
Chile	Korea	Russia
China	Latvia	Slovak Republic
Colombia	Lebanon	Slovenia
Croatia	Lithuania	South Africa
Czech Republic	Malaysia	Sri Lanka
Egypt	Mexico	Thailand
Estonia	Morocco	Turkey
Hungary	Pakistan	Venezuela
India	Peru

View Table

Table 3.

Emerging Markets Included in Global Factor Model

Argentina	Indonesia	Philippines
Brazil	Israel	Poland
Bulgaria	Jordan	Romania
Chile	Korea	Russia
China	Latvia	Slovak Republic
Colombia	Lebanon	Slovenia
Croatia	Lithuania	South Africa
Czech Republic	Malaysia	Sri Lanka
Egypt	Mexico	Thailand
Estonia	Morocco	Turkey
Hungary	Pakistan	Venezuela
India	Peru

20. The global factor model’s estimates of the common and country-specific components for the CE-4 currencies are very similar to those obtained from the state- space model. A plot of the CE-4 common component over the 2003-06 period shows the same common factor as derived from the state space model: weakness in 2003, followed by a steady appreciation from February 2004 to March 2006, with a brief interruption in March-April 2005 (Figure 6). Similarly, the global factor model’s estimates of the country-specific components show the forint to be the outlier, as in the state space model (Figure 7). The magnitudes are slightly different in the two models—the country-specific depreciations are smaller (and the appreciations larger) in the global factor model—which is likely due to the assumption of equal weights on the common component, whereas this is allowed to vary in the state-space model (recall, the γ weighting on the common factor was country-specific).

CE-4: Common Component (Regional + Global), 2003-06

(January 1, 2003 = 100; increase denotes depreciation)

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Source: IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

CE-4: Common Component (Regional + Global), 2003-06

(January 1, 2003 = 100; increase denotes depreciation)

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Source: IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

CE-4: Common Component (Regional + Global), 2003-06

(January 1, 2003 = 100; increase denotes depreciation)

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Source: IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

CE-4: Country-Specific Components, 2003-06

(January 1, 2003 = 100; increase denotes depreciation)

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Source: IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

CE-4: Country-Specific Components, 2003-06

(January 1, 2003 = 100; increase denotes depreciation)

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Source: IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

CE-4: Country-Specific Components, 2003-06

(January 1, 2003 = 100; increase denotes depreciation)

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Source: IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

21. The global model shows that region-specific factors strengthened the CE-4 currencies until end-2004, and, thereafter, the positive sentiment toward emerging markets supported these currencies. Over the past decade, the CE-4’s common component has depreciated by 11 percent (Figure 8). The period from 1997 to 2003 was a difficult period for emerging markets, with currencies weakening by almost 60 percent on average (Figure 9). Once we filter out this global weakening, we are left with the CE-4’s region-specific component (Figure 10). The relative strength of the Central European currencies—the EU accession effect—is now evident, as seen in the steady appreciation of these currencies by almost 50 percent between late 2000 and early 2005. It appears, however, that this process has come to an end: since early 2005, the CE-4’s region-specific component has been flat. Instead, more broad-based EM appreciation vis-à-vis the euro supported the CE-4 currencies in 2005 and the first quarter of 2006. It is still unclear whether the weakening since March 2006 is a temporary blip, or a sign that this period of EM strength is ending.

CE-4: Common Component (Regional + Global), 1997-2006

(January 1, 1997 = 100; increase denotes depreciation)

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Source: IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

CE-4: Common Component (Regional + Global), 1997-2006

(January 1, 1997 = 100; increase denotes depreciation)

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Source: IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

CE-4: Common Component (Regional + Global), 1997-2006

(January 1, 1997 = 100; increase denotes depreciation)

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Source: IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

D. Correlates of the Hungary-Specific, Regional, and Global Components

22. Surprises in monetary and exchange rate policy moved the Hungary-specific component in 2003. Here we focus on the movements in the forint that are not attributable either to changes in the common component nor to yield differentials. We do this by regressing the Hungary-specific component on the three-month interest rate differential vis-à-vis the euro area, and taking the residual; this is plotted in Figure 11, along with selected news events. The first sharp drop, in June 2003, was due to the MNB’s surprise devaluation of the central parity by 2.2 percent. The forint promptly fell by more than 9 percent, from Ft 246/euro to Ft 266/euro, forcing the MNB to hike the policy rate by 300 basis points. The second sharp drop, in late November and early December, followed a period where concerns about Hungary’s imbalances and slowing growth weakened the forint outside the MNB’s publicly declared “target range” of Ft 250-260/euro. The MNB hiked the policy rate by another 300 basis points between rate-setting meetings, but the forint weakened even further, to Ft 273/euro, before eventually recovering.

Hungary: Selected News Events that Depreciated the Forint, 2003-06

(Hungarian forints per euro)

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Hungary: Selected News Events that Depreciated the Forint, 2003-06

(Hungarian forints per euro)

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

• Download Figure
• Download figure as PowerPoint slide

Hungary: Selected News Events that Depreciated the Forint, 2003-06

(Hungarian forints per euro)

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

• Download Figure
• Download figure as PowerPoint slide

23. The forint remained in a tight range from mid-2004 until September 2005, when fiscal disappointments and increased global sensitivity to EM imbalances weakened the forint. In September-October 2005, a succession of negative fiscal news items—including upward revisions of the 2004 deficit outcome and 2005 deficit projection, and an EU request for an updated Convergence Programme with a more realistic consolidation path—weakened the forint, and it has not recovered to its previous levels since. Increased global investor sensitivity to imbalances in selected emerging markets, initiated by the fall of the Icelandic krona in March 2006 and the Turkish lira in June 2006, have further weakened the forint.

24. At the regional level, the CE-4 component shows a strong correlation with other “convergence indicators,” suggesting that EU accession probably played a central role in the region’s steady appreciation since 2001. We look at two such indicators: the first is the simple average of CE-4 5-year local currency bond yield spreads vis-à-vis German bunds, an indicator of current interest rate convergence with the euro area; the second is the simple average of CE-4 5-year/5-year forward swap spreads vis-à-vis the euro, an indicator of future (i.e., five years forward) interest rate convergence that is frequently used as an indicator of euro adoption prospects. As seen in Figure 12, the appreciation of the CE-4 common component has mirrored the convergence of both current and expected interest rates.

CE-4: Region-Specific Component and Selected “EU Accession” Indicators, 1997-2006

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Sources: Bloomberg; JP Morgan; and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

CE-4: Region-Specific Component and Selected “EU Accession” Indicators, 1997-2006

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Sources: Bloomberg; JP Morgan; and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

CE-4: Region-Specific Component and Selected “EU Accession” Indicators, 1997-2006

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Sources: Bloomberg; JP Morgan; and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

25. Finally, at the global level, since the U.S. dollar is the primary benchmark currency for most EM currencies, it is useful to decompose our global (EM/euro) component into two parts: the global EM/U.S. dollar component, and the U.S. dollar/euro exchange rate. While the analysis in this chapter has focused on exchange rates vis-à-vis the euro, as this is the natural benchmark currency for the forint and the other CE-4 currencies, most emerging market currencies are primarily benchmarked against the U.S. dollar. To understand what drives global EM currency trends, therefore, the global EM/euro component shown in Figure 9 can be decomposed into two parts: the U.S. dollar per euro exchange rate (Figure 13, bottom-left panel), and a second part that measures the performance of global EM currencies against the U.S. dollar (Figure 13, bottom-right panel).

Decomposing the Global EM/Euro Component into the Global EM/US Dollar Component and the US Dollar/Euro Exchange Rate

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Sources: Bloomberg; and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Decomposing the Global EM/Euro Component into the Global EM/US Dollar Component and the US Dollar/Euro Exchange Rate

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Sources: Bloomberg; and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

Decomposing the Global EM/Euro Component into the Global EM/US Dollar Component and the US Dollar/Euro Exchange Rate

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Sources: Bloomberg; and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

26. The global EM/U.S. dollar component shows a strong correlation with indicators of global risk appetite, such as the U.S. high-yield spread. The U.S. high-yield spread is often used as a proxy for international investors’ attitude towards risk, but is also a leading indicator of U.S. economic activity (Gertler and Lown 1999; Mody and Taylor 2003). As can be seen in Figure 14, there is a strong correlation between the global EM/U.S dollar component and the U.S. high-yield spread.

The Global EM/US Dollar Component and the US High-Yield Spread, 1997-2006

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Sources: Bloomberg; and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

The Global EM/US Dollar Component and the US High-Yield Spread, 1997-2006

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Sources: Bloomberg; and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

The Global EM/US Dollar Component and the US High-Yield Spread, 1997-2006

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A001

Sources: Bloomberg; and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

E. Conclusions

27. The two legs that have held the forint up in recent years—a strong “EU accession effect” and positive sentiment toward emerging markets—may no longer be strong enough to offset Hungary’s weak fundamentals. The CE-4’s common appreciation over the past two years can be segmented into two subperiods. From mid-2003 through 2004, there was a strong, region-specific appreciation—the so-called “convergence play,” reflecting the optimism associated with EU accession. Since 2005, this region-specific appreciation has ended, and CE-4 currencies have been influenced by global sentiment towards emerging markets. This link to emerging markets sentiment provided strong support through 2005 and early 2006 but, since March, has been a source of weakness and volatility.

28. Within the CE-4, the Czech and Slovak koruna and the Polish zloty have largely moved together, but the Hungarian forint has moved substantially on its own. Regional factors helped strengthen all the CE-4 currencies, to varying degrees. The Hungary-specific component of the forint has been particularly weak. The earlier weakening of the forint (in 2005) was related to monetary easing and the consequent narrowing of interest rate differentials vis-à-vis the euro area. However, the most recent period of forint weakness has been due to other factors, the most obvious candidate being the persistent slippages in fiscal outcomes and worries about Hungary’s growing public and external debt, especially in an environment of greater global volatility. With the increased global weakness and volatility since March 2006, the likelihood is increasing that fundamentals, in Hungary and elsewhere, will play a larger role in currency movements.