Greece: Selected Issues and Statistical Appendix

This paper provides model-based projections of inflation, and quantifies the impact of the factors that determine inflation. The reasons behind the poor labor market performance in Greece and the remedial policies are discussed. The new economy is at a relatively early stage of development in Greece compared with most industrial countries, but growing rapidly. The key issues facing the Greek banking sector, against the background of its recent performance and a rapidly evolving market and regulatory environment, is discussed. The statistical data are also presented.

Abstract

This paper provides model-based projections of inflation, and quantifies the impact of the factors that determine inflation. The reasons behind the poor labor market performance in Greece and the remedial policies are discussed. The new economy is at a relatively early stage of development in Greece compared with most industrial countries, but growing rapidly. The key issues facing the Greek banking sector, against the background of its recent performance and a rapidly evolving market and regulatory environment, is discussed. The statistical data are also presented.

I. Inflation Risks in The Wake of Euro-Area Entry1

A. Introduction

7. The process of disinflation embarked upon by Greece in the early 1990s succeeded in bringing headline inflation down to euro-area levels by mid-1999 but was followed by an uptick in inflation last year. Consumer price inflation had gradually been lowered from 20 percent in 1990 to 2½ percent in 1999. Tight monetary and fiscal policies have aided disinflation in recent years, encouraged by the desire to qualify for euro-area entry. In the buoyant economic conditions of 2000, however, the tripling of oil prices and the unwinding of administrative measures that had helped Greece fulfill the Maastricht inflation criterion pushed headline inflation back up toward 4 percent. The rekindling of inflation has coincided with a substantial easing of monetary conditions related to euro entry—significantly larger than in any first-round EMU entrant—and occurred at a time when the output gap may already have closed. A number of other euro countries that entered EMU in similar conditions have since experienced substantial price and cost pressures.

uA01fig01

Consumer Price Inflation 1/

(In percent, year-on-year)

Citation: IMF Staff Country Reports 2001, 057; 10.5089/9781451816136.002.A001

1/ EU-harmonized index.

8. Staff previously considered the impact of monetary conditions and the cyclical position on inflation in Greece, using a variety of approaches ranging from single-equation models to vector autoregressions (VAR).2 This chapter draws on several of these approaches to provide model-based projections of inflation, as well as to quantify the impact of the factors that determine inflation (see Sections B and C for single-equation and VAR models, respectively).

9. Given the fundamental change in the economic and monetary regime engendered by Greece's adoption of the euro as well as recent steps toward structural reform, the “Lucas critique” applies: econometric results based on historical data do not necessarily provide a reliable guide to behavior in the new regime. Indeed, EMU participation does seem likely to affect the way inflationary shocks are propagated through the economy and, moreover, accelerated structural reform has the potential to lower price pressures significantly (these issues are reviewed in Section D). Against this background, the chapter concludes with a brief discussion of near-term inflation prospects and risks in Greece, drawing on the experience from other euro entrants (Section E).

B. Single-Equation Models of Inflation

10. This section presents two single-equation models of inflation and the associated model-based projections for inflation in 2001 and 2002. The first is an update of the simple Phillips curve model employed by staff in a 1998 Selected Issues paper (referenced above). The second is a more elaborate markup model of inflation that embeds features from the Phillips curve. To help evaluate the models’ forecasting properties, the out-of-sample forecasts for the last two years are examined.

A simple Phillips curve specification revisited

11. Expressing the Phillips curve in terms of price inflation instead of wage inflation, and substituting the output gap for the unemployment rate, with the help of a simple Okun relation between output and unemployment, yields the following equation:

π=πe+βgap+Z,

where π and πe stand for the actual and expected year-on-year rate of inflation, respectively, gap is the output gap, and Z represents other shocks, such as commodity prices.

12. Of the two econometric specifications given in the 1998 Selected Issues paper, attention focuses here on the error correction version, which has had the better ex post performance. Inflation expectations are specified as a weighted average of forward-looking elements (e.g., based on foreign consumer price inflation in drachma, πf) and a backward-looking component stemming from less than fully rational expectations or overlapping wage contracts. With the sample updated to 2000, estimation yields the following result (with standard deviations in parentheses below the estimated parameter values):3,4

Δπ=-0.0006+0.18Δπf+0.16gap-2-0.08(π-πf)-1(0.0012)(0.03)(0.06)(0.02)Sample:1981:2-2000:4R2=0.37σ=0.97%DW=1.7

13. The estimated equation suggests a significant role for the output gap in determining inflation. The short-term direct pass-through of foreign inflation in drachmas stands at 18 percent, just below the share of imported goods and services in private consumption which is about 20-25 percent. The overall pass-through converges gradually to one as a result of the long-term cointegrating relation between domestic and foreign inflation (which is equivalent to a relative purchasing power parity condition).

14. Turning to forecasting properties, the model's out-of-sample projection for inflation in 1999 and 2000 (when estimated on data up to 1998) is shown in Table 1. In both years, actual inflation was significantly lower than predicted by the model, but this is due partly to indirect tax cuts and to other inflation-reducing measures whose mechanical effect is estimated to have lowered inflation by 0.9 percentage points in 1999 and by 0.6 percentage points in 2000.5 Even when these effects are taken into account, however, a sizable overestimate of inflation remains for 2000. It is likely that this reflects in part the indirect effects of the lowering of taxes in 1999 (and previously in 1998), which had a subsequent inflation-reducing effect, all else being equal, since they served to limit wage increases triggered by indexation clauses.

Table 1.

Ex Post Analysis: Projections for 1999-2000

(Out-of-sample forecasts; annual average inflation)

article image
Source: Fund staff estimates.

Adjusted for the mechanical impact of indirect tax cuts.

Cost-based model with variable markup

15. A somewhat richer model specification is needed to take these and other recent events into account. In what follows, the domestic price level is modeled as a markup over total unit costs—including unit labor costs, import prices, and energy costs—where the markup responds to the economic cycle in a manner that retains elements from the Phillips curve.6 In contrast to the pure Phillips curve, the inflation rate is tied not only to short-term dynamics but also to the long-run determinants of the price level. The starting point is the following long-horizon relation for the consumer price level:

P=μ(gap,.)(ULCα)(PMδ)(POILγ),

where µ denotes the mark-up factor (which may depend, for example, on the output gap and indirect taxes); ULC nominal unit labor costs; PM import prices (or, in the absence of a quarterly import price series, foreign consumer prices in drachma); and POIL the oil price in drachma. Both the mark-up and costs may vary over the cycle. The equation can be expressed in log-linear form (where lower-case letters denote logs of the variables):

p=log(μ)+αulc+δpm+γpoil.

16. Because standard inference tests can go seriously astray when one or more variables in a regression have unit roots, a first step toward a well-specified model is to determine the order of integration of the variables. The unit root test statistics reported in the Appendix indicate that the log of consumer prices, foreign consumer prices, unit labor costs, and oil prices are nonstationary, and the first three possibly integrated of order two (i.e., I(2)).7 In view of the steady decline in Greek inflation over the last two decades, it is not surprising that the Dickey-Fuller tests should indicate that prices may be I(2). However, this trend reduction is not an innate characteristic of inflation. Rather than basing the analysis on the results of tests that are known to have poor finite-sample properties, the model is specified as if all price and cost series are I(1), bearing in mind that some caveats may apply in estimation.8

17. While the individual series are nonstationary, there may be a stationary linear combination of them reflecting a long-run relationship among the variables (namely, a cointegrating vector). The results of Johansen's maximum likelihood procedure reported in the appendix indicate the presence of one cointegrating vector which corresponds to the following long-run relation:

p=0.54ulc+0.46pm+0.01poil+0.01gap+indtax.(0.09)(0.09)(0.02)(0.005)

The sum of the elasticities on the three cost terms is close to 1, and tests confirm that linear homogeneity cannot be rejected. The somewhat larger weight of foreign prices compared with the direct import content in private consumption may reflect indirect cost effects (as foreign inputs are used in the production of consumption items by Greek firms), as well as price effects, if domestic producers of import substitutes price strategically by maintaining some relation to import prices. The seeming unimportance of oil prices may be due to the impact of oil on other countries’ prices that is already captured in pm; that is, poil merely captures the additional impact in Greece, which consumes relatively much energy per unit of GDP. The long-run equation includes indirect taxes, indtax, since these constitute a wedge between unit costs and consumer prices.9

18. Starting from a general model and simplifying to a statistically acceptable error correction model yields the following equation for the one-quarter change in prices:

Δp=const+0.10Δpm+0.002gap-4-0.08(errorcorrectionterm)+Δindtax(0.04)(0.0007)(0.04)Sample:1985:1-2000:4,R2=0.87σ=0.83%DW=1.7

In free estimation, the coefficient on ∆indtax was (insignificantly) larger than one; in the final specification the coefficient was constrained to one, which is more intuitive. The relatively low significance of the error correction term (which is stationary by construction) reflects the difficulties that may arise when the left-hand-side variable appears nonstationary over the sample. The estimation equation included a linear trend as well as seasonal dummies. The contemporaneous change in unit labor costs was not significant.

19. The model implies an immediate pass-through of exchange rate changes of around 10 percent, reaching almost one-half in the long run by virtue of the cointegrating relation (assuming that unit labor costs are unaffected).10 Excess demand pressure, as reflected in the output gap, raises the markup of prices over costs; in addition, the output gap may affect prices indirectly (and more forcefully) through any impact it may have on unit labor costs. The relatively low coefficient on the error correction term implies a considerable smoothing in the inflation process, consistent with what can be gleaned from an informal examination of the data. The half-life of deviations from long-run equilibrium is about eight quarters.

uA01fig02

Inflation: Actual, Fitted, and Residuals 1/

(In percent)

Citation: IMF Staff Country Reports 2001, 057; 10.5089/9781451816136.002.A001

1/ Fitted and actual values of one-quarter changes have been converted to four-quarter changes. CP Inet of indirect taxes.

20. The model's out-of-sample projections for 1999 and 2000 are Shown in Table 2. The projection is quite close to the outcome in the first year, but clearly above the outcome in the second year. The source of this deviation can be traced predominantly to the role of the exchange rate: the sizable depreciation (on account of euro weakness in 1999–2000) appears not to have been reflected in prices to the extent suggested by earlier behavioral patterns (see below).

Table 2.

Ex Post Analysis: Projections for 1999-2000

(Out-of-sample forecasts; annual average inflation)

article image
Source: Fund staff estimates.

Inflation projections for 2001-02: single-equation results

21. On the basis of the staff's WEO projections for unit labor costs, the output gap, foreign prices, and the exchange rate, the models described in this section can be used to forecast inflation for 2001 and 2002 (Table 311). Both models suggest that inflation could exceed the 2000 level of 3 percent in the coming two years, based on the staff's assessment of the output gap (which is estimated to attain ¾ percent of GDP above potential) and an acceleration in wage increases to some 5 percent a year. The model projection is predicated on the assumption of a broadly unchanged euro exchange rate (the euro/dollar exchange rate is assumed to average 0.94 in 2001 and 0.95 in 2002) and a slight easing in oil prices (to $23 and $21.5 per barrel in 2001 and 2002, respectively).

Table 3.

Inflation Projections for 2001-02

(Annual average inflation)

article image
Source: Fund Staff estimates.

22. The markup model indicates that inflationary risks—based on past behavioral patterns—may be even larger than indicated by the simple Phillips curve model. The long-run relation between the price level, unit labor costs, and import prices implies that a delayed pass-through of exchange rate depreciation could keep inflation close to 4 percent in 2001—unless the euro appreciates markedly relative to present assumptions. As the import price effect tapers off but economy-wide unit labor costs accelerate moderately, the model projects inflation to ease back to around 3¾ percent by 2002.

C. VAR Models

23. In the type of models described above, forecasts of unit labor costs, import prices, and the output gap must be provided as inputs to the inflation projection—and it is by no means obvious that these variables are easier to forecast than inflation itself. Thus, single-equation specifications do not provide a self-contained projection of inflation; in a sense, they transform the nature of the forecasting problem to one of forecasting the explanatory variables in the equation. Structural models or vector autoregressions are required to provide a stand-alone model forecast and this section explores the implications of some simple VAR models.

24. VAR models can be used in two fundamentally different ways to shed light on inflation prospects in Greece. One approach is to estimate and identify a VAR model and use impulse responses to derive the inflation impact of particular shocks. This was the approach taken to gauge the impact of movements in the interest rate and exchange rate in the run-up to EMU in the 1999 Selected Issues paper.12 A second approach is simply to let the VAR system project forecasts for all the endogenous variables, including inflation (in this case, the VAR model does not need to be identified). Each of these approaches is briefly explored below.

25. The response of inflation to standardized shocks to the real interest rate differential (relative to Germany) and changes in the nominal effective in percentage points) exchange rate are shown in Table 4. These impulse responses are based on a structural VAR model with five variables, namely inflation, the real short-term interest rate differential with Germany, and the growth rates of industrial production, nominal wages, and the nominal effective exchange rate. The interactions of the five variables are taken into account in determining the effect of shocks to any of the variables on the others. Identifying restrictions on the contemporaneous interactions of the variables were imposed using the Bernanke (1986) and Sims (1986) methodology. For example, inflation in the current quarter is assumed to be affected directly by activity, wages, and the exchange rate, with a one-quarter delay before shocks to interest rates affect inflation, reflecting the lag in the transmission of monetary policy to the economy.13

Table 4.

VAR Impulse Responses of Inflation

(Deviation of inflation from baseline; in percentage points)

article image
Source: Fund staff estimates.

Response to a one-off shock to the nominal effective exchange rate.

26. The tightening of ECB policy rates since 1999 and the 3½ percent revaluation of the Greek drachma's central parity against the euro in January 2000 serve to lessen the inflationary impact of euro-linked monetary easing. Nonetheless, the reduction in the real short-term interest rate spread over the September 1999–December 2000 period was close to 500 basis points, so that EMU-related interest convergence would entail around 1 percentage point of inflation in 2001–02. This reflects the use in the model of the interest rate paid on bank deposits, which is linked to the short-term policy rate.

uA01fig03

Monetary Conditions in the Run-Up to EMU

Citation: IMF Staff Country Reports 2001, 057; 10.5089/9781451816136.002.A001

Moreover, the trade-weighted drachma exchange rate depreciated by more than 5 percent in 2000, partly as a result of the euro's weakening against other major currencies. The calculation for the effect of monetary easing already includes the inflationary consequences of a depreciation of around 2 percent, implied by the VAR responses of the exchange rate to the easing of real interest rates. The somewhat larger actual appreciation in 1999–2000 is likely to add further to inflationary pressures in 2001–02, although the model responses suggest that the second- and third-year effects are modest. Thus, taking into account the timing of interest rate and exchange rate shocks and their propagation through time, the inflationary impact of EMU-related easing of monetary conditions—much larger than in previous euro entrants (see figure)—is simulated to be above 1 percentage point in 2001–02.

27. The VAR model may also be run as a pure forecasting model—avoiding what is sometimes seen to be overly restrictive identification assumptions and placing minimum economic structure on the quantitative model. The out-of-sample forecasts of the five-variable VAR system14 overshoot actual inflation by a substantial amount (Table 5). As with the simple Phillips curve model reviewed earlier, the sharp reductions in indirect taxes and other measures that helped reduce headline inflation in 1999–2000 account for most of the discrepancy (about two-thirds, considering only the direct impact of tax cuts). Given the “black-box” nature of the unidentified VAR system, it is difficult to pin down exactly what drives the results. The system incorporates the historically observed interactions and mutual repercussions between policy variables (interest rates and exchange rates) and the state of the economy (the output gap, inflation, and wage increases) and projects them assuming unchanged policy reaction functions and behavioral patterns.

Table 5.

VAR Model Projections, 1999-2002

(Annual average inflation)

article image
Source: Fund staff estimates.

Adjusted for the mechanical impact of indirect tax cuts.

28. The VAR specification is also employed to obtain a system projection for inflation in 2001 and 2002 directly. The results imply that inflation could rise further to 4.7 percent in 2002. The projection entails a very buoyant economy with output exceeding estimated potential by more than 2 percent by end-2002, as well as further exchange rate depreciation and a negative real interest rate differential vis-à-vis Germany—all these factors contribute to higher inflation. The results are perhaps best seen as a worst case scenario in which the euro depreciates and demand severely outpaces supply in the Greek economy.

D. Regime Change and Structural Reform: Some Possible Implications

29. The tendency for actual inflation to have undershot the predictions of quantitative models in recent years—even when heterodox anti-inflation measures are explicitly taken into account—raises the question of whether more fundamental changes in the inflation process have occurred. The most obvious candidate for triggering such change is the shift to a stability-oriented macroeconomic policy regime in the run-up to euro adoption. Some structural reforms may also have played a role.

30. Euro-area entry may affect the way inflationary shocks are propagated through the economy, including through changes in the formation of inflation and exchange rate expectations, wage behavior, and increased price competition related to enhanced product and capital market integration. As outlined below, tentative signs of change appear in all three areas.

31. The principal reason that model-based projections overshot actual inflation in 2000 appears to be that exchange rate movements, associated foremost with the depreciation of the euro (and thus the drachma), have not fed through to consumer prices to the same extent as in the past. The literature on exchange rate pass-through as well as the experience of several countries during the 1990s suggests that the extent of exchange rate pass-through may be muted by a number of factors, notably:15

Persistence of exchange rate shock: if businesses and workers perceive a given exchange rate change to be temporary, they are less likely to adjust prices and wages than if they believe it to be permanent;

Credibility of monetary policy: if the central bank succeeds in keeping inflation expectations at bay, second-round effects on wages and prices are less likely to occur;

Market structure: when price competition is intense, foreign producers may find it difficult to pass on higher foreign costs; and

Cyclical conditions: weak demand makes it harder for importers and retailers to pass on higher costs to their customers.

32. In Greece, several of these factors may have been working to reduce the exchange rate pass-through and could continue to do so in future. Historically, depreciations of the drachma were typically not reversed and the drachma has shown a strong trend depreciation throughout much of the last three decades against most major currencies. By contrast, price and wage setters may have been less inclined to view the recent euro weakness as lasting. Thus, whereas, for example, the response of Greek core inflation (as measured by the HICP excluding energy and seasonal food) to the March 1998 devaluation was almost immediate (see figure), core inflation responded relatively late to the euro-induced effective depreciation over the last two years (a sizable fraction of the inflation rise in late 2000 was due to the unwinding of indirect tax cuts that had previously lowered the inflation rate). In addition, the anti-inflation mandate of the European Central Bank has kept inflation expectations in the area low, as gauged, for example, from yields on nominal versus index-linked bonds (where available), as well as from representative economic forecasts. Low inflation expectations are crucial in limiting second-round effects. Mostly, these changes would be expected to slow down the exchange rate impact but not to reduce the long-run effect if the exchange rate were to stay at a depreciated level.

uA01fig04

Exchange Rate Depreciation and Core Inflation

(In percent, year-on-year)

Citation: IMF Staff Country Reports 2001, 057; 10.5089/9781451816136.002.A001

1/ Excluding energy and fresh food

33. The tendency for Greek inflation to have undershot expectations in recent years has also been related to exceptional wage moderation, at least by past Greek standards, as trade unions explicitly supported the country's EMU aspirations. To the extent that this presented a break from past behavior—as captured in the empirical models of Sections B and C—this contributed to the model-based overestimation of inflation in some cases. Looking ahead, it remains to be seen whether wage behavior will change on a sustainable basis, as discussed in more detail in the staff report and Chapter III.

34. Fundamental changes in labor market structures could become part of a broader theme of structural reform with important implications for price behavior. Structural reform of product markets has the potential to lower price pressure directly—a channel most vividly demonstrated by the impact of liberalization on telecommunications and, in some of Greece's EU partners, on electricity prices. More broadly, enhanced competition could set off significant efficiency gains and reduce price pressures throughout the economy. A recent OECD study (Nicoletti and others, 1999) argued that Greece had one of the most burdensome administrative and regulatory business environments among advanced economies (see figure). Progress in product market reform has hitherto lagged behind other countries, but telecommunications liberalization has now caught up with Greece's EU partners, competition policy has been strengthened in the course of 2000, and new initiatives are being considered in a broad range of areas. While product market reforms probably have played a limited role in explaining the model-based inflation overshooting so far, the recent and prospective measures could, if enacted in full, enhance the growth potential and reduce price pressures substantially in the years to come (see the staff report on the 2000 Article IV consultation, in particular Box 3, for a review of product market reform).

uA01fig05

Summary Indicators of Product Market Regulation, 19981/

Citation: IMF Staff Country Reports 2001, 057; 10.5089/9781451816136.002.A001

1/ Higher numbers indicate more restrictive regulations.

E. Inflation Prospects in the Light of Other Euro-Area Entrants’ Experience

35. The experience of those first-round euro-area entrants that were also at a relatively advanced stage of the cycle when they experienced significant euro-related monetary easing—notably Ireland, Portugal, and Spain—may hold lessons for Greece. In all three countries, the resulting demand impetus spurred inflationary pressure, in some cases substantially (see Table 6). Price pressure has intensified not least for nontradables, implying that euro depreciation is not the main culprit behind the acceleration in inflation. Although Greece differs in important respects from this group of countries—especially because of the private sector's large interest income from government bonds, which may reduce the demand effect of interest rate reductions—these experiences highlight potential risks in Greece. Presumably, the full effects of recent monetary easing—much more substantial for Greece than for any first-round EMU entrant—have not yet played out. Moreover, Greece could be more exposed than other euro countries in one important respect: Greece is the only country where backward-looking wage indexation remains in force (although the current two-year wage agreements did not provide for indexation of wages with respect to inflation in 2000). Formal indexation clauses significantly raise the risk of second-round effects of import price increases.

Table 6.

Inflation in Selected Euro Countries

article image
Sources: Eurostat; and IMF, WEO database.

36. In practice, staff and other forecasters’ inflation projections are based on a combination of recent indicators, quantitative economic models, and qualitative judgments. Staff's central forecast for inflation is predicated on improvements in the inflation process relative to past behavioral patterns. However, unless the euro appreciates or oil prices fall more than assumed, lagged effects of the external shocks that hit the Greek economy in 2000 are likely to exert considerable upward pressure on prices in 2001. Moreover, indications have surfaced of substantial wage drift and there are signs that the social partners’ resolve in sustaining wage moderation could be waning. Finally, the inflation catch-up clause in collective wage agreements threatens to be triggered with effect on wages in 2002.

Table 7.

Inflation Forecasts

(Period average; in pereent)

article image
Sources: Ministry of National Economy, Stability and Growth Program; OECD, Annual Review: Greece 2000; and Consensus Forecast

CPL

Private consumption deflator.

HICP.

37. In the absence of accelerated structural reform and consensus on wage restraint to improve competitiveness, unit labor cost growth and inflation are projected to exceed the euro-area average in 2001–02 by a margin somewhat above estimated Balassa-Samuelson effects. On the other hand, if potential output is shifting to a higher trajectory than assumed by staff on account of structural reform and access to finance at low euro rates that boost capital formation, the economy may be better able to sustain rising demand without accompanying inflation pressure. If risks of higher wage increases can be contained, inflation could end up below the staff's projection for 2001 of an average HICP increase of 3.4 percent (well above the approximately 2 percent projected for the euro area).

APPENDIX: Unit Root Tests and Johansen Tests for Cointegration

Unit Root Tests

Augmented Dickey-Fuller (ADF(4)) Test Statistics

article image
Note: ** indicates rejection at 1 percent MacKinnon critical value.

Tests for Number of Cointegrating Vectors

(Johansen Maximum Likelihood test)

Based on trace of the stochastic matrix

article image
Note: Test indicates 1 cointegrating equation at 1 percent significance.

References

  • Bernanke, B., 1986, “Alternate Explanations of the Money-Income Correlation,Carnegie-Rochester Conference Series on Public Policy, Vol. 25.

    • Search Google Scholar
    • Export Citation
  • Campbell, J.Y., and P. Perron, 1991, “Pitfalls and Opportunities: What Macroeconomists Should Know About Unit Roots,NBER Macroeconomics Annual 1991.

    • Search Google Scholar
    • Export Citation
  • Cochrane, J.H., 1991, “Commenton Campbell and Perron (1991), in NBER Macroeconomics Annual 1991.

  • de Brouwer, G., and N.E. Ericsson, 1998, “Modeling Inflation in Australia,Journal of Business & Economic Statistics, No. 4, Vol. 16 (October).

    • Search Google Scholar
    • Export Citation
  • Goldberg, P.K., and M.M. Knetter, 1997, “Goods Prices and Exchange Rates: What Have We Learned?,Journal of Economic Literature (September).

    • Search Google Scholar
    • Export Citation
  • Halikias, I., and N. Sobczak, 1998, “Post-Devaluation Inflation Prospects: An Empirical Investigation,Greece—Selected Issues,IMF Staff Country Report No. 98/100 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Lucas, R.E., Jr. 1976, “Econometric Policy Evaluation: A Critique,” in The Phillips Curve and Labor Markets, ed. by K. Brunner A.H. Meltzer (New York: North-Holland).

    • Search Google Scholar
    • Export Citation
  • Nicoletti, G., S. Scarpetta, and O. Boylaud, 1999, “Summary Indicators of Product Market Regulation with an Extension to Employment Protection Legislation,OECD Economics Department Working Paper No. 226 (Paris: OECD).

    • Search Google Scholar
    • Export Citation
  • Republic of Greece, 2000, Stability and Growth Program 2000-04 (Athens: Republic of Greece).

  • Sims, C., 1986, “Are Forecasting Models Usable for Policy Analysis?,Federal Reserve Bank of Minneapolis Quarterly Review (winter).

    • Search Google Scholar
    • Export Citation
  • Swagel, P., 1999, “Inflationary Implications of EMU related Monetary Easing: An Econometric Estimate,Greece—Selected Issues,IMF Staff Country Report No. 99/138 (Washington: International Monetary Fund)

    • Search Google Scholar
    • Export Citation
1

This chapter was prepared by Mads Kieler.

3

In the absence of quarterly national accounts time series, quarterly data for the output gap have been derived from staff's annual estimates using a numerical technique based on the Hodrick-Prescott filter with residuals added to ensure that the annual averages of the quarterly series equal the annual series.

4

The long-run cointegrating relationship anchors domestic inflation 1:1 to foreign inflation (in drachma); the annual growth rate of the real exchange rate is stationary.

5

Fund staff estimates, based on information supplied by the Bank of Greece.

6

The model and estimation strategy largely follow de Brouwer and Ericsson (1998).

7

Meaning that the series needs to be differenced twice to produce a stationary series (i.e., a series whose mean and variance are independent of time). This would imply that the first difference of the log-series, namely inflation, is nonstationary.

8

Modeling prices as I(1) is standard in the literature; see, for instance, de Brouwer and Ericsson (1998). On the use of unit root tests, see, for example, Campbell and Perron (1991) and Cochrane (1991).

9

If the long-run relation had not included indirect taxes then the error correction term in the short-run model would misleadingly impart upward pressure on inflation after a tax cut—misleadingly because the output gap (and other factors) are controlled for.

10

If unit labor costs do respond to pm, the long-run pass-through may be complete, but this depends on relations outside the current single-equation framework for prices.

11

Note that the results in Table 2, in contrast to Table 1, are based on the inflation outturn through 2000.

12

This approach does not yield an inflation forecast per se but rather the differential impact of particular shocks.

13

For details, see SM/99/255 (10/6/99). Industrial production was used as a measure of economic activity in the absence of quarterly GDP data.

14

The system was amended slightly by substituting the output gap for industrial production.

15

For a review of the literature on exchange rate pass-through, see Goldberg and Knetter(1997).

Greece: Selected Issues and Statistical Appendix
Author: International Monetary Fund