This Selected Issues paper analyzes the main factors underlying the evolution of competitiveness, export performance, and labor market developments in Poland. The paper seeks to identify key questions of relevance for policymaking. The overall evolution of competitiveness compared with Poland’s trading partners and competitors since 1995 is analyzed using standard measures of the real effective exchange rate (REER). This paper also examines the role of foreign investors in financing domestically issued public debt in Poland.

Abstract

This Selected Issues paper analyzes the main factors underlying the evolution of competitiveness, export performance, and labor market developments in Poland. The paper seeks to identify key questions of relevance for policymaking. The overall evolution of competitiveness compared with Poland’s trading partners and competitors since 1995 is analyzed using standard measures of the real effective exchange rate (REER). This paper also examines the role of foreign investors in financing domestically issued public debt in Poland.

III. Exchange Rate Pass-Through in Poland15

The paper finds that Poland exhibits a moderate exchange rate pass-through to consumer prices. Estimates based on a recursive VAR model suggest that the pass-through to consumer prices from a change in the value of the zloty vis-à-vis a basket of currencies is negligible in the short-term (at most 7 percent within three months) and reaches at most 22 percent within one year. The pass-through seems to have declined relative to the mid-to-late 1990s.

A. Introduction

27. Recent significant depreciation of the zloty vis-à-vis the euro raises questions as to its implications for consumer price inflation and hence for the conduct of monetary policy. The zloty has been depreciating since mid-2001, loosing more than 14 percent of its value against the euro in 2003 and another 2½ percent in the first quarter of 2004 (Figure 1). At the same time, headline inflation edged up from ½ percent in January 2003 to 1.7 percent by end-March 2004. Although inflation remains low, continued zloty depreciation raises the question of whether pressures that could fuel inflation are building up and, if so, within what time frame. Answering this question would provide useful guidance for monetary policy decisions since the objective of Poland’s Monetary Policy Council is to maintain a continuous target of annual inflation within a two-percentage-points wide interval centered at 2½ percent.

Figure 1.
Figure 1.

Poland: Exchange Rate, Euro Per zloty

November 1999- March 2004

Citation: IMF Staff Country Reports 2004, 218; 10.5089/9781451831924.002.A003

28. Several factors suggest that the exchange rate pass-through in Poland could be non-negligible. First, the exchange rate and prices seem to move together, with producer price changes seemingly more closely linked to exchange rate movements than consumer price changes (Figure 2). Second, Poland is a relatively open economy with few trade barriers—trade amounts to nearly two-thirds of GDP.16 Finally, a sizable share of private domestic demand is allocated to imported goods, which suggests that exchange rate shocks could transmit to domestic producer and consumer prices. Indeed, the share of imported goods in total investment is estimated at 70 percent; also, the breakdown of CPI components and associated weights suggests that the share of tradables in the consumption basket is about 55 percent.

Figure 2.
Figure 2.

Poland: Exchange Rates and Prices

January 1996 - March 2004

Citation: IMF Staff Country Reports 2004, 218; 10.5089/9781451831924.002.A003

1/ Weighted average of the euro/zloty and USD/zloty rates, with respective weights 65 percent and 35 percent. A decline in the NEER or Basket ER indicates a depreciation of the zloty.

29. The pass-through of exchange rate fluctuations to domestic prices is the subject of a large body of empirical research. Choudhri and Hakura (2001) estimate the pass-through in a single equation model for a set of 71 countries, and find that it is strongly positively correlated with average inflation. McCarthy (1999) estimates a recursive VAR model for nine industrialized countries, for which he finds that the exchange rate pass-through is modest and may have declined in recent years. Most recently, a number of Fund studies used McCarthy’s model to estimate the pass-through in Fund member countries as an input for advice on monetary policy.17

30. Previous studies have found an exchange rate pass-through in Poland within the range of 20 to 30 percent. In its Inflation Report for the first quarter of 2003, the National Bank of Poland (NBP) states that, based on monthly data for 1998–2002, the long-run passthrough coefficient is 28 percent for the nominal effective zloty rate. Based on quarterly data for 1993–2000, Darvas (2001) finds zero instantaneous pass-through in Poland but between 20 and 30 percent pass-through in the long run.18

31. In line with the above results, this study finds that Poland exhibits a moderate exchange rate pass-through to consumer prices. The empirical methodology adopted here closely follows McCarthy (1999). Based on data available as of the first quarter of 2004, the results indicate that a depreciation of the zloty vis-à-vis a basket of currencies—representing Poland’s trading partner currencies—does not have a significant instantaneous effect on consumer prices. After twelve months, however, 22 percent of that depreciation will have translated into a permanent price change—in other words, a temporary increase in the inflation rate. This result is also close to recent NBP staff findings, as indicated by discussions held in Warsaw during the 2004 Article IV consultation mission.

32. The rest of this paper is organized as follows. Section B describes the data and methodology used to estimate the pass-through. Sections C through E summarize the empirical results. Section F concludes.

B. Methodology and Data

33. The paper estimates an empirical model that mirrors the pricing chain in the economy, allowing for external and domestic shocks to filter through to prices. Following McCarthy (1999), the model is a recursive VAR system that includes three types of domestic prices: imported goods’ prices (IP), producer prices (PPI) and consumer prices (CPI). The pricing chain flows from import prices first, to producer prices and finally to consumer prices. In addition, it is assumed that prices can be affected by four different shocks: exogenous supply shocks, embodied in shocks to international commodity prices Pcom); domestic demand shocks, proxied by the output gap (Ygap); domestic supply shocks, captured by total labor costs (W)19; and shocks to the exchange rate (XR).

34. The shocks are identified through a Choleski decomposition. The variables are ordered according to the following assumptions: (a) shocks to the commodity price index are exogenous; (b) exchange rate shocks do not have a contemporaneous effect on the output gap; (c) domestic supply shocks do not have contemporaneous effects on the exchange rate or on the output gap; (d) shocks to import prices have a contemporaneous effect on both producer and consumer prices but not on any other variable; (e) shocks to the PPI only have a contemporaneous effect on consumer prices; (f) shocks to the CPI do not have a contemporaneous effect on any of the other variables.20

35. The VAR system can therefore be written as follows:

Ptcom = Et1Ptcom + μtcomYtgap = Et1Ytgap + α1μtcom + μtgapXRt = Et1XRt + β1μtcom + β2μtgap + μtXRWt = Et1Wt + δ1μtcom + δ2μtgap + δ3μtXR + μtwIPt = Et1IPt + γ1μtcom + γ2μtgap + γ3μtXR + γ4μtw + μtIPPPIt = Et1PPIt + λ1μtcom + λ2μtgap + λ3μtXR + λ4μtw + μtIP + μtPPICPI = Et1CPIt + ρ1μtcom + ρ2μtgap + ρ3μtXR + ρ4μtw + ρ5μtIP + ρ6μtPPI + μtCPI

E refers to the expectations operator. In each equation, the contemporaneous shocks affecting the endogenous variable are captured by the µ’s, whose superscripts indicate the source of the shock.

36. Careful consideration should be given to the choice of the exchange rate measure, as it is expected to influence the results. The euro/zloty exchange rate could be a reasonable choice: Direction of Trade statistics confirm the importance of the euro in Poland’s trade patterns, suggesting that a stronger pass-though to domestic prices could be expected from the euro-zloty rate than from, say, the dollar-zloty rate. Indeed, about 55 percent of Polish imports originate from the Euro area, while an additional 6 to 8 percent come from other new EU members whose currencies seem more closely linked to the euro than to other currencies.21 However, focusing on a single exchange rate could be misleading: it does not take into account price effects of other exchange-rate movements, especially in periods when the zloty is simultaneously appreciating against one major currency and depreciating against another—as was the case for the zloty in 2003 and early 2004. Accordingly, the nominal effective exchange rate seemed the most appropriate measure to use. The model was also estimated using a weighted average of the zloty/euro and zloty/dollar exchange rates, with respective weights 65 and 35 percent—thereafter referred to as “bivariate basket rate”.22 This is a simpler measure of the average exchange rate and can be more readily used in forecasting exercises that do not assume constant exchange rates.

C. Estimation Results: Price Effects of External and Domestic Shocks

37. The model was estimated on transformed monthly data for the period extending from 1996:1 to 2003:9. Table 1 summarizes the data sources and variable definitions. All series were log-transformed and seasonally adjusted—except for the output variable constructed as specified in Table 1. Unit root tests indicated that while the output gap series is stationary, all other series are integrated of order one.23 The model was therefore estimated in first order differences.24,25 In each equation, the expectation operators were replaced with lags of the endogenous variables—three lags were sufficient to generate white noise residuals. Impulse response functions and variance decompositions were calculated to characterize the role of domestic and external shocks in price fluctuations.

Table 1.

Poland: Data Sources and Description

Sample Period: 1996:1–2003:9

article image

38. Figures 3 and 4 show the dynamic effects on producer and consumer price inflation of external and domestic shocks as well as shocks to the exchange rate. In each chart, the solid line depicts the estimated percentage year-on-year change in the PPI or the CPI over time (the horizontal axis indicates time in months), and for a horizon of two years, following a 10 percent shock to either of the above listed variables—keeping everything else constant. The dotted lines represent one-standard deviation bands around the point estimates.26 The responses are of the expected signs and statistically significant for at least the first year after the shock.27

Figure 3.
Figure 3.

Impulse Response Functions-Model with NEER

Responses, in percent, to shocks of magnitude 10 percent 1/

Citation: IMF Staff Country Reports 2004, 218; 10.5089/9781451831924.002.A003

1/ A positive shock to the NEER indicates average zloty appreciation.
Figure 4.
Figure 4.

Impulse Response Functions-Model with Bivariate Basket Rate

Responses, in percent, to shocks of magnitude 10 percent 1/

Citation: IMF Staff Country Reports 2004, 218; 10.5089/9781451831924.002.A003

1/ A postive shock to the basket rate indicates a zloty appreciation.

39. While commodity price shocks have comparable effects on producer and consumer price inflation, exchange rate shocks have larger—but somewhat less persistent—effects on producer price inflation. Indeed, an increase in commodity prices by 10 percent leads to an immediate increase in year-on-year producer and consumer price inflation, by 1½–2 percent; both inflation measures remain around that level for nearly one year before declining to insignificant levels. An effective (average) appreciation of the zloty28 by 10 percent, however, triggers an immediate jump in year-on-year PPI inflation by about 2.3 percent but no contemporaneous reaction in CPI inflation. PPI inflation peaks 6 months later, at close to 4 percent, while year-on-year CPI inflation only increases by just above 2 percent. However, the effects of the exchange rate shock on producer price inflation die out after about a year and a half, six months before the same happens to consumer price inflation.

40. Domestic demand and supply shocks have both larger and more persistent effects on consumer price inflation. An increase in the output gap by 10 percent, capturing a demand shock, has a negligible instantaneous impact on the PPI but immediately raises year-on-year consumer price inflation by nearly 1 percent. In the month following the shock, producer price inflation peaks at about 1 percent, before declining back to zero (and becoming statistically insignificant). Consumer price inflation peaks at about 2.2 percent three months after the demand shock, before gradually declining back to zero within the course of the year. Supply shocks have a larger and far more persistent inflationary effect than demand shocks on both producer and consumer prices. Twelve months following a 10 percent increase in total labor costs, year-on-year producer and consumer price inflation peak at about 2.2 percent and 3¼ percent respectively, before starting to decline. The producer price inflation effect is wound down by the end of the following year whereas it takes about 6 more months for the effect on consumer price inflation to dissipate.

41. Exchange rate shocks play the most significant role among the shocks analyzed above in explaining producer and consumer price inflation fluctuations. Tables 2.a. and 2.b. present the variance decomposition of the forecast error variance for each of producer and consumer price inflation into the parts due to each of the seven innovations of the model. These results show that, in the case of both the NEER and the bivariate basket rate, and excluding the first few months after a shock occurs, the exchange rate is the most important determinant of PPI and CPI inflation variance—even more so than the price series’ own innovations. Indeed, shocks to the NEER can explain up to half of the variance in the PPI and CPI forecast errors. The corresponding share explained by shocks to the bivariate basket rate is even greater, at about two-thirds.

Table 2.a.

Poland: Variance Decomposition of the Forecast Errors *

Results for the Model with the Bivariate Basket Rate

article image
Table 2.b.

Poland: Variance Decomposition of the Forecast Errors *

Results for the Model with the Bivariate Basket Rate

article image

The forecasts of the PPI and CPI refer to dynamic forecasts calculated across the full sample.

D. Estimation Results: Exchange Rate Pass-Through to Domestic Prices

42. For a given price variable, the j periods-ahead pass-through coefficient for an exchange rate shock at time t is defined as the ratio of cumulative impulse responses of the price variable and the exchange rate variable between t and t+j. Formally, the pass-through is calculated as follows:

PTt,t+j=i=tt+jIRFiCPIi=tt+jIRFiXR,

where IRFjz indicates the impulse response of variable z to an exchange rate shock in period i. This formula ensures that all secondary effects of the initial exchange rate shock on exchange rate dynamics will be taken into consideration, which is not the case when calculating the impulse response functions.

43. It turns out that the pass-through to producer prices is much faster and initially significantly larger than the pass-through to consumer prices (Table 3). This result was already foreshadowed in the impulse response functions depicted in Figures 3 and 4. The instantaneous pass-through to the CPI is negligible, whereas it exceeds 10 percent in the case of the PPI. Within a year, the pass-through from a given exchange rate shock to producer prices is about 75 percent larger than the counterpart pass-through to consumer prices. Eventually, however, and within two years, the pass-through to consumer prices, at 39 or 44 percent depending on the exchange rate definition, is just below the pass-through to producer prices. The difference is that most of the pass-through to the PPI had cumulated within one year of the shock, a period over which only about half of the total pass-through to the CPI occurs.

Table 3.

Poland: Exchange Rate Pass-Through, 1996–2003

(In percent)

article image

The long-run refers to the maximum period for which the impulse response function remains statistically significant. It corresponds to 15 months for the PPI in both models; for the CPI, it is either 19 months (bivariate basket rate model) or 21 months (NEER model).

E. Role of the Inflationary Environment in Determining the Pass-Through

44. The degree of pass-through is believed to depend on the inflationary environment. A lower inflationary environment is associated with less persistent price shocks, which could induce producers to pass-on a smaller share of exchange rate shocks to prices (Taylor, 2000). Indeed, as previously noted, Choudhri and Hakura (2001) find evidence for a positive association between the degree of exchange-rate pass-through and the average inflation rate in a sample of 71 countries (excluding Poland). In order to assess the influence of inflation conditions on the pass-through in Poland, the sample is split in two sub-periods: 1996–99 and 2000–03. Inflation averaged 13½ percent during 1996–99, having declined from 21 percent in January 1996 to just below 10 percent at end 1999. The second sample corresponds to a low inflation period, in which inflation averaged 4½ percent.

45. The model was slightly modified to accommodate the smaller samples. In order not to loose too many degrees of freedom and considerably reduce the efficiency of coefficient estimates, the import price variable—which had the least important influence on producer and consumer prices—was eliminated from the model, but the ordering of the remaining six variables was kept unchanged.29 Two lags were used to estimate the resulting six-equation model, which was enough to yield uncorrelated residuals.

46. The results support the view that the pass-through is higher in high-inflation periods (Table 4). This is particularly the case for longer horizons. Indeed, the immediate pass-through of exchange rate changes to consumer price inflation has remained insignificant across time. The pass-through cumulated after a period of six or twelve months has diminished, however—its size in the sample covering the most recent four years is less than half of the estimated size in 1996–99 for the model using the NEER.

Table 4.

Poland: Exchange Rate Pass-Through to the CPI in Sub-Sample Periods

(In percent)

article image

F. Conclusion

47. This paper finds that the exchange rate pass-through to consumer prices in Poland is moderate. At about 7 percent after the first quarter and 22 percent within a year, this pass-through is comparable to that estimated for Belgium or the Netherlands, but somewhat higher than that estimated for France, Germany, and Italy (as reported in Choudhri and Hakura, 2001Table 5). Section D also showed that the pass-through has been declining over time, with the change in inflation environment. Therefore, the pass-through results obtained from the full sample (see Table 3) should be viewed as upper limits on the pass-though that one should expect to prevail in Poland today, given the particularly low level of inflation in the past two years.

Table 5.

Poland: Exchange Rate Pass-Through in Poland and Selected EU countries+

In percent

article image

The results for countries other than Poland are from Choudhri and Hakura (2001).

References

  • Belaisch A., 2003, “Exchange Rate Pass-Through in Brazil,IMF Working Paper 03/141 (Washington: International Monetary Fund).

  • Bhundia A., 2002, “An Empirical Investigation of Exchange Rate Pass-Through in South Africa,IMF Working Paper 02/165 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Billmeier A., and L. Bonato, 2002, “Exchange Rate Pass-Through and Monetary Policy in Croatia,IMF Working Paper 02/109 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Choudhri E., and D. Hakura, 2001, “Exchange Rate Pass-Through to Domestic Prices: Does the Inflationary Environment Matter?IMF Working Paper 01/194 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Coricelli F., and others, 2003, “Exchange Rate Pass-Through in Candidate Countries,CEPR Discussion Paper No. 3894 (London: Centre for Economic Policy Research).

    • Search Google Scholar
    • Export Citation
  • Darvas Z., 2001, “Exchange Rate Pass-Through and Real Exchange Rate in EU Candidate Countries,Bundesbank Discussion Paper 10/01 (Frankfurt: Deutsche Bundesbank).

    • Search Google Scholar
    • Export Citation
  • Gueorguiev, N, 2003, “Exchange Rate Pass-Through in Romania,IMF Working Paper 03/130 (Washington: International Monetary Fund).

  • Leigh, D. and M. Rossi, 2002, “Exchange Rate Pass-Through in Turkey,IMF Working Paper 02/204 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • McCarthy J., 1999, “Pass-Through of Exchange Rates and Import Prices to Domestic Inflation in Some Industrialised Economies,BIS Working Paper No. 79 (Basel: Bank For International Settlements).

    • Search Google Scholar
    • Export Citation
  • National Bank of Poland, 2003, “Inflation Report for the 1st Quarter 2003,” available at http://www.nbp.pl/en/publikacje/raport_inflacja/iraport_1_03_gb.pdf

    • Search Google Scholar
    • Export Citation
  • Przystupa J., 2002, “The exchange rate in the monetary transmission mechanism,” National Bank of Poland Mimeo (www.nbp.pl/publikacje/materialy_i_studia/25_en.pdf)

    • Search Google Scholar
    • Export Citation
  • Taylor J., 2000, “Low Inflation, Pass-Through, and the Pricing Power of Firms,European Economic Review Vol. 44, pp.13891408.

15

Prepared by Nada Choueiri.

16

The share of trade in GDP has risen from about 50 percent in the early 1990s to about 70 percent in 2003.

17

See the References section for a list of the IMF working papers that cover this topic.

18

Using monthly data for 1993-2002 for Poland, Coricelli et al. (2003) find that 80 percent of an equilibrium change in the exchange rate will transmit to inflation. However, their methodology is suitable mainly for longer time periods and in countries where the underlying equilibrium exchange rate-price relationship is stable.

19

The addition of domestic supply shocks, following Gueorguiev (2003), is a departure from McCarthy’s model which only included the other six variables.

20

While there could be other reasonably justifiable identifying restrictions, the above ordering was supported by results of Granger causality tests.

21

This link is mainly due to policy management in these countries. Estonia and Lithuania have fixed exchange rates vis-à-vis the euro. Hungary and Slovenia’s respective exchange rates are pegged to the euro and allowed to fluctuate within horizontal bands. Slovakia closely manages its exchange rate to minimize volatility against the euro. As for the Czech republic, although it has no formal exchange rate policy, the value of its koruna is more stable versus the euro than the dollar (as can be verified by plots of the koruna/euro and koruna/dollar exchange rates.) Poland’s trade with the remaining accession countries is negligible.

22

The weights were chosen based on Direction of Trade statistics, which suggested that about 65 percent (35 percent) of Polish imports are priced either in euro (U.S. dollar) or in a currency closely linked to the euro (U.S. dollar).

23

Evidence for the CPI was mixed, with some tests suggesting it could be I(2) but others indicating it was I(1).

24

In other words, the cointegration of the variables was ignored—although cointegration tests indicated the presence of several cointegrating vectors. Three reasons justify this decision. First, the time-period covered by the study is too short to factor in an equilibrium relationship among the variables at hand that one could reasonably be confident with. Second, the transition process in the Polish economy in this short time period implies continuous changes in the underlying equilibrium. Third, the study’s purpose is to understand the pass-through over short horizons, for which ignoring the underlying long-run equilibrium of the economy should not significantly undermine the results.

25

For any variable X, the 12-period difference XtXt-12 was used in the estimation. The results should therefore be interpreted in terms of year-on-year price and exchange rate changes.

26

The model was estimated using RATS. The standard deviations of the impulse responses were calculated by 2000 bootstrap replications of the model, using pseudo-historical data created by drawing with replacement from the empirical distribution of the VAR innovations. All responses are to one unit shock in the exchange rate, which approximates a year-on-year percentage increase in the average monthly level of the exchange rate.

27

PPI and CPI responses to a shock in import prices (not shown) are only statistically significant for the first three to four months.

28

An appreciation of the zloty is captured in the model by an increase in the NEER or in the bivariate basket rate.

29

Even with this modification the small sample sizes affected the efficiency of the results as impulse response functions for all shocks were generally only significant for a few months after a given shock. Therefore the results of these smaller models are only provided to illustrate the link between inflation conditions and the exchange rate pass-through.

Republic of Poland: Selected Issues
Author: International Monetary Fund
  • View in gallery

    Poland: Exchange Rate, Euro Per zloty

    November 1999- March 2004

  • View in gallery

    Poland: Exchange Rates and Prices

    January 1996 - March 2004

  • View in gallery

    Impulse Response Functions-Model with NEER

    Responses, in percent, to shocks of magnitude 10 percent 1/

  • View in gallery

    Impulse Response Functions-Model with Bivariate Basket Rate

    Responses, in percent, to shocks of magnitude 10 percent 1/