## 1. NMS and Euro Area: Energy and Food Intensity

## Source details

**Canonical URL:** [1. NMS and Euro Area: Energy and Food Intensity](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2009/_wp0951.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2009/_wp0951.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2009/_wp0951.pdf.json)

---

### Energy and Food Intensity: exposure and contributions to headline inflation
- NMS consume on average about 50 percent more energy than old EU countries (as measured by CPI weight of energy consumption).
- Food share in the CPI basket in NMS ranges from:
  - about 50–60 percent higher (Czech Republic, Slovakia, Estonia, Poland, Slovenia, Hungary),
  - about 100 percent higher (Latvia, Lithuania),
  - over 120 percent higher (Bulgaria, Romania)
  relative to old EU members.
- Table of average 2000–2008 CPI weights (percent):
  - Bulgaria: Energy 14.7, Liquid fuel 4.9, Food 31.6
  - Czech Republic: Energy 13.9, Liquid fuel 3.9, Food 17.7
  - Estonia: Energy 13.9, Liquid fuel 6.1, Food 21.4
  - Hungary: Energy 13.0, Liquid fuel 5.2, Food 18.7
  - Latvia: Energy 12.9, Liquid fuel 3.6, Food 28.5
  - Lithuania: Energy 14.0, Liquid fuel 4.7, Food 28.1
  - Poland: Energy 14.9, Liquid fuel 4.6, Food 22.8
  - Slovak Republic: Energy 16.7, Liquid fuel 3.4, Food 18.8
  - Slovenia: Energy 12.7, Liquid fuel 8.0, Food 17.3
  - Romania: Energy 14.7, Liquid fuel 4.9, Food 31.6
  - Euro Area: Energy 8.8, Liquid fuel 4.8, Food 14.4
- Contribution of food and energy to headline inflation:
  - Euro area: slightly less than 40 percent.
  - NMS: close to 60 percent.
- VAR evidence (tri-variate VAR on monthly seasonally adjusted data, January 2000–June 2008):
  - Commodity prices impact core inflation in most NMS through domestic energy or food prices or both, but indirect effects are generally not large.
  - Country-specific VAR findings:
    - Estonia and Slovenia: commodity prices have little impact on underlying inflation (as in euro area).
    - Bulgaria: food prices are the main indirect driver.
    - Lithuania, Latvia, Hungary, Slovakia: energy prices drive indirect effects.
    - Czech Republic, Poland, Romania: both food and energy drive indirect effects.
- Inflation expectations (pooled regression of EC consumer surveys):
  - Inflation expectations in the NMS are generally well anchored.
  - A 1 percentage point higher food inflation increases inflation expectations by 0.1 percentage points.
  - Energy and core inflation have little impact on inflation expectations.

### Determinants of the Common Component of NMS inflation
- Two-step empirical strategy:
  1. Decompose inflation into a common component and country-specific component using the generalized dynamic factor model (GDFM): x_{t,i} = x^c_{t,i} + θ_{t,i} = L_i(b_i) ν_t + θ_{t,i}.
     - ν_t: vector of size q of common unobservable dynamic factors.
     - θ_{t,i}: mean-zero country-specific component orthogonal to ν_t.
  2. Regress the extracted common and idiosyncratic components on macroeconomic variables capturing hypothesized drivers.
- Pooled regression specification for the common component (notation preserved):
  - x^c_{t,i} = α + β1 P_{1998,i} + β2 Δ r_{t-1,i} + β3 π^e_{t-1,i} + ε_{t,i}
  - Variable definitions:
    - P_{1998,i}: 1998 price level of NMS relative to EU15.
    - Δ r_{t-1,i}: change in interest rates (proxy for EU integration effects).
    - π^e_{t-1,i}: energy inflation (proxy for global commodity price shocks).
- Pooled regression results and interpretation:
  - Model explains over 40 percent of variability of common inflation (Adjusted R-squared: 0.44).
  - Coefficients (Table 2):
    - Relative price level in 1998: Coefficient -0.05 (standard error (0.02))
    - Change in interest rates: Coefficient -0.35 (standard error (0.05)) — interpretation: a country that experiences a 1 percentage point decline in interest rates would have about 1/3 of a percentage point higher common inflation.
    - Energy inflation: Coefficient 0.12 (standard error (0.01)) — interpretation: a 1 percent increase in energy prices accounts for 0.12 percentage points common inflation.
  - Interpretation of the price-level coefficient: in a NMS with a price level 30 percent below the level observed in the EU15, common inflation would be about 1½ percentage points (derived from coefficient -0.05).

### Determinants of Country-specific (Idiosyncratic) Inflation
- Country-specific regression (notation preserved):
  - θ_t = δ + γ1 Δ y_{t-1} + γ2 r_{t-1} + γ3 Δ neer_{t-1} + γ4 π^f_{t-1} + ν_t + r_f
  - Variable definitions:
    - Δ y_{t-1}: output gap (domestic cyclical conditions).
    - r_{t-1}: real interest rate.
    - Δ neer_{t-1}: percent change of nominal effective exchange rate.
    - π^f_{t-1}: foreign inflation.
  - Note: changes in regulated prices and indirect taxes are not explicitly modeled and are captured in the residual term.
- Overall explanatory power:
  - Cyclical factors, financial conditions, and openness explain about half of the variation of the country-specific inflation component.
  - Output gap is statistically significant in most NMS.
  - Real long-term interest rate is a significant determinant across countries.
- Selected country coefficients and fit (Table 3; standard errors in parentheses; "n.s." indicates not statistically significant at 1 percent, 5 percent, or 10 percent):
  - Czech Republic:
    - Output gap: 0.08 (0.03)
    - Real interest rates: -0.90 (0.05)
    - Nominal effective exchange rate: -0.04 (0.01)
    - Adjusted R-squared: 0.84
  - Estonia:
    - Output gap: 0.16 (0.05)
    - Real interest rates: -0.38 (0.10)
    - Nominal effective exchange rate: -0.08 (0.02)
    - Adjusted R-squared: 0.40
  - Hungary:
    - Output gap: 0.12 (0.06)
    - Real interest rates: -0.40 (0.07)
    - Adjusted R-squared: 0.40
  - Latvia:
    - Output gap: 0.1 (0.07)
    - Real interest rates: -0.70 (0.05)
    - Nominal effective exchange rate: 0.50 (0.05)
    - EU inflation: 1.8 (0.50)
    - Adjusted R-squared: -0.70 (as printed in table)
  - Lithuania:
    - Output gap: 0.05 (0.03)
    - Real interest rates: -0.65 (0.06)
    - Nominal effective exchange rate: 0.16 (0.05)
    - Adjusted R-squared: 0.70
  - Poland:
    - Output gap: 0.20 (0.04)
    - Real interest rates: -0.05 (0.03)
    - Nominal effective exchange rate: -0.03 (0.01)
    - EU inflation: 0.6 (0.20)
    - Adjusted R-squared: 0.30
  - Slovak Republic:
    - Output gap: n.s.
    - Real interest rates: -0.45 (0.05)
    - Nominal effective exchange rate: -0.20 (0.05)
    - EU inflation: n.s.
    - Adjusted R-squared: 0.55
  - Slovenia:
    - Output gap: n.s.
    - Real interest rates: n.s.
    - Nominal effective exchange rate: -0.11 (0.04)
    - EU inflation: 1.3 (0.40)
    - Adjusted R-squared: 0.40
  - Bulgaria:
    - Output gap: n.s.
    - Real interest rates: n.s.
    - Nominal effective exchange rate: -0.40 (0.12)
    - EU inflation: 3.00
    - Adjusted R-squared: 0.35
  - Romania:
    - Output gap: n.s.
    - Real interest rates: -0.35 (0.07)
    - Nominal effective exchange rate: -0.16 (0.02)
    - EU inflation: 1.00 (0.06)
    - Adjusted R-squared: 0.45

### Common-factor decomposition and variance explained
- Spectral analysis note:
  - "0.52 on the horizontal axis corresponds to twelve months. The region to the left of that point correspond to periods greater than, while the region to the right to less than twelve months."
  - First common factor (largest eigenvalue):
    - Explains about 65 percent of the data variability at a medium-term horizon.
    - Explains about 70 percent of the data variability at shorter horizons.
  - Marginal contributions decline rapidly across factors; the marginal contribution of the third principal component is considerably smaller than the second.
  - Threshold for identifying common shocks chosen at about 50 percent (following Cristadoro and others (2005)).
  - Spectral analysis conclusion: the first factor is chosen as the single dynamic common factor of inflation in the NMS; remaining factors contribute to country-specific components.
- Static factor selection:
  - Number of static factors chosen using Bai and Ng (2002) algorithm: criterion reaches a minimum at 7 static factors.
  - Given one common shock from spectral analysis, this implies six lags for the dynamic common factor in equation (1).
- Share of inflation variation explained by dynamic common factor(s) (time-domain):
  - With one dynamic factor:
    - Explains about 15 percent in the Baltics.
    - Explains around 45 percent in the Czech Republic and Slovakia.
    - Explains 75 percent in Hungary and Poland.
    - Explains over 80 percent in Bulgaria and Romania.
  - Adding a second dynamic factor increases the share of explained inflation variation to above 80 percent on average for the NMS.

### Methods and data — scope and processing
- First-stage GDFM dataset:
  - 955 predominantly four-digit level HICP series for the NMS, January 1998–June 2008 monthly.
  - Where four-digit not available, next aggregation level used (about 1/4 of dataset).
  - All HICP from Eurostat.
- Data preprocessing:
  - First- and twelve-month differences of HICP components.
  - Seasonal adjustment by X12.
  - Normalization by subtracting mean and dividing by standard deviation for month-on-month and year-on-year regressions.
- Second-stage (non-HICP) data:
  - Industrial production (seasonally adjusted), interest rates, nominal effective exchange rates (monthly), price levels of NMS relative to EU15 (annual).
  - Quarterly and annual models also estimated by averaging monthly data; quarterly models replace industrial production with real GDP.
  - Nominal effective exchange rates, industrial production for most countries, and price levels from Eurostat; remaining industrial production data from Haver Analytics.

### Key quantitative findings and summary implications
- Decomposition and drivers:
  - A significant part of inflation in the NMS over the medium term is driven by common factors — about 65 percent at medium-term horizons (first common factor).
  - The pooled regression indicates relative price level, EU integration forces (proxied by convergence of nominal interest rates), and energy prices explain over 40 percent of variability of common inflation (Adjusted R-squared: 0.44).
  - Idiosyncratic factors (cyclical conditions, financial conditions, exchange rate pass-through) explain about half of the variation of the country-specific component.
- Magnitudes and elasticities:
  - Relative price level (1998) coefficient: -0.05 (s.e. (0.02)).
  - Change in interest rates coefficient: -0.35 (s.e. (0.05)).
  - Energy inflation coefficient: 0.12 (s.e. (0.01)).
  - Interpretation examples:
    - A NMS with a price level 30 percent below EU15: common inflation ≈ 1½ percentage points (computed from coefficient -0.05).
    - A 1 percentage point decline in interest rates → about 1/3 percentage point higher common inflation.
    - A 1 percent increase in energy prices → 0.12 percentage points common inflation.
- VAR/indirect effects:
  - Food and energy account for close to 60 percent of headline inflation in the NMS versus slightly less than 40 percent in the euro area.
  - Indirect (second-round) effects from commodity prices on core inflation are present but limited and heterogeneous across NMS.

### Policy implications and relevance
- For policymakers preparing for euro adoption under the Maastricht inflation criterion, understanding the split between common and country-specific inflation drivers is crucial.
- Common forces identified include:
  - Price-level convergence toward EU15 (Balassa-Samuelson and quality/convergence effects).
  - EU integration effects proxied by convergence of nominal interest rates.
  - Global commodity (energy) price shocks.
- Country-specific policy levers and considerations:
  - Domestic cyclical conditions (output gap) and financial conditions (real long-term interest rates) materially affect idiosyncratic inflation and should be monitored and managed.
  - Exchange rate developments and openness (neer changes) are important for pass-through and country-specific inflation dynamics.
  - Regulated-price and indirect-tax changes are relevant but enter primarily through residuals in country-specific regressions and warrant separate attention in policy analysis.
- Inflation expectations are generally well anchored in the NMS; food-price shocks have a measurable but small effect on expectations (1 percentage point food inflation → 0.1 percentage point higher expectations), suggesting limited de-anchoring risk from commodity shocks in most cases.

*Source: _wp0951 - 1. NMS and Euro Area: Energy and Food Intensity, pages 5–15.*

### 1. NMS and Euro Area: Energy and Food Intensity ..................................................................5

### 1. NMS and Euro Area: Energy and Food Intensity

### Energy and Food Intensity (section)
- Title as listed: "1. NMS and Euro Area: Energy and Food Intensity"
- Location in source: page 5
- Indicates a focused analysis comparing New Member States (NMS) and the Euro Area on energy and food contribution to headline inflation.

### NMS: Determinants of Common Component (section)
- Title as listed: "2. NMS: Determinants of Common Component"
- Location in source: page 14
- Indicates analysis of factors driving the common component of inflation across NMS.

### NMS: Determinants of Country-specific Component (section)
- Title as listed: "3. NMS: Determinants of Country-specific Component"
- Location in source: page 15
- Indicates analysis of factors driving country-specific inflation dynamics within NMS.

### Figures (list)
- Figure 1: "Euro Area: Contribution of Energy and Food to Headline Inflation" — page 5
- Figure 2: "NMS: Contribution of Energy and Food to Headline Inflation" — page 6
- Figure 3: "NMS: Price level, Inflation, and Exchange Rate Regime" — page 8
- Figure 4: "Cumulative Share of Data Variance Explained by Common Factors" — page 12
- Figure 5: "NMS: Headline and One Common Factor Inflation" — page 13

*Source: _wp0951 - 1. NMS and Euro Area: Energy and Food Intensity, pages 5–15*

### References..............................................................................................................

### _wp0951 - References

### I. Introduction — key points and context
- New EU10 member states from Eastern Europe (NMS) are under the obligation to adopt the euro some time in the future.  
- Maastricht inflation criterion: inflation in a member state applying for euro entry should not exceed the average inflation rate of the three countries with the lowest inflation rate in EU plus 1½ percentage points.  
- Over the past ten years, with some exceptions, inflation in the NMS has exceeded the Maastricht inflation criterion most of the time. Exceptions cited:  
  - Poland: from August 2002 to August 2004 and from January 2006.  
  - Slovenia: from January 2006 to July 2007.  
  - Lithuania: from January 2000 to December 2004.  
  - Latvia: from January 2001 to December 2003.  
  - Czech Republic: from August 2002 to November 2007.  
- Slovenia adopted the euro in January 2007; Slovakia scheduled to adopt it in January 2009.  
- Policy relevance: knowing the degree to which inflation is driven by common factors (affecting all NMS) versus country-specific factors (domestic aggregate demand/supply, administered prices, indirect taxes) is useful for policymakers.  
- Summary finding stated: a significant part of headline inflation in the NMS is driven by common factors (price level convergence, EU integration), while country-specific factors (financial conditions, pass-through from foreign prices, demand-supply conditions, administered price adjustments and indirect tax increases associated with EU accession) also played a role.

### II. Related literature — synthesis of arguments and evidence
- Globalization and common factors:
  - Rogoff (2003): globalization lowered inflation via competition and political economy processes increasing anti-inflationary credibility.  
  - Borio and Filardo (2007) and Razin and Binyamini (2007): global economic slack significantly explains domestic inflation, with an increasing role since the 1990s.  
  - Ihrig and others (2007): find no evidence that globalization explains a trend decline in the sensitivity of inflation to the domestic output gap.  
- Studies focused on EU and NMS:
  - Stavrev (2006): GDFM framework separates country-specific and common components of inflation in NMS8 and estimates drivers of country-specific inflation.  
  - Mody and Ohnsorge (2007): across EU25, common cross-border sources of inflation increase, reducing domestically-generated inflation; country-specific inflation in NMS is higher than in old EU members; lower initial price level associated with higher subsequent inflation.  
  - Choueiri and others (2008): in EU25, cross-country differences in common inflation depend on initial price level gaps, changes in the nominal effective exchange rate, institutional quality, and flexibility of economies.

### III. Inflation dynamics in NMS — background, stylized facts, and VAR findings
- Consumption-basket structure and exposure to commodity shocks:
  - NMS are more energy intensive and have a significantly higher share of food prices in their consumer baskets than the old EU members. Table 1 (average for 2000-2008, share in CPI basket, in percent) reports:  
    - Bulgaria: Energy 14.7, Liquid fuel 4.9, Food 31.6  
    - Czech Republic: Energy 13.9, Liquid fuel 3.9, Food 17.7  
    - Estonia: Energy 13.9, Liquid fuel 6.1, Food 21.4  
    - Hungary: Energy 13.0, Liquid fuel 5.2, Food 18.7  
    - Latvia: Energy 12.9, Liquid fuel 3.6, Food 28.5  
    - Lithuania: Energy 14.0, Liquid fuel 4.7, Food 28.1  
    - Poland: Energy 14.9, Liquid fuel 4.6, Food 22.8  
    - Slovak Republic: Energy 16.7, Liquid fuel 3.4, Food 18.8  
    - Slovenia: Energy 12.7, Liquid fuel 8.0, Food 17.3  
    - Romania: Energy 14.7, Liquid fuel 4.9, Food 31.6  
    - Euro Area: Energy 8.8, Liquid fuel 4.8, Food 14.4  
  - On average, NMS consume about 50 percent more energy than old EU countries (as measured by CPI weight of energy consumption). Food share ranges from 50-60 percent higher (Czech Republic, Slovakia, Estonia, Poland, Slovenia, Hungary) to about 100 percent higher (Latvia, Lithuania) to over 120 percent higher (Bulgaria, Romania) relative to old EU members.  
- Contribution of food and energy to headline inflation:
  - Over recent years, food and energy inflation contributed slightly less than 40 percent to headline inflation in the euro area, while contributing close to 60 percent in the NMS.  
- Indirect (second-round) effects on core inflation — VAR evidence:
  - A tri-variate VAR (food, energy, core inflation = HICP excluding energy and all food) estimated using monthly seasonally adjusted data for January 2000 to June 2008 is used to gauge indirect effects. Domestic energy and food prices used to estimate indirect effects on core inflation, filtering short-term volatility in international energy prices.  
  - Findings: commodity prices seem to impact core inflation in most NMS through domestic energy or food prices or both, but effects are not large. Country exceptions:  
    - Estonia and Slovenia: commodity prices seem to have little impact on underlying inflation (as in euro area).  
    - Bulgaria: food prices appear to be the cause of indirect effects.  
    - Lithuania, Latvia, Hungary, Slovakia: energy prices are responsible for indirect effects.  
    - Czech Republic, Poland, Romania: both food and energy prices drive indirect effects on underlying inflation.  
- Inflation expectations:
  - Pooled regression of inflation expectations (EC consumer surveys) on energy, food, and core inflation suggests inflation expectations in the NMS are in general well anchored.  
  - Energy and core inflation have little impact on inflation expectations; food inflation has a significant but relatively small effect: one percentage point higher food inflation increases inflation expectations by 0.1 percentage points.

### IV. Methods and data — two-step empirical approach and datasets
- Two-step approach:
  1. Decompose inflation into a common component and country-specific (idiosyncratic) component using the generalized dynamic factor model (GDFM) of Forni and others (2000). GDFM allows for autocorrelation of dynamic factors and is suited to isolate unobserved common components.  
     - Representation (as given): x_{t,i} = x^c_{t,i} + θ_{t,i} = L_i(b_i) ν_t + θ_{t,i}  (Equation (1) in the source, notation preserved)  
     - ν_t: vector of size q of common unobservable dynamic factors; b_i: vector of loading factors specific to each country; θ_{t,i}: mean-zero country-specific component orthogonal to ν_t.  
  2. Identify determinants of the common and country-specific components by regressing them on macroeconomic variables capturing common and country-specific shocks.  
- Modeling the common component:
  - Pooled regression specification (as in source):  
    x^c_{t,i} = α + β1 P_{1998,i} + β2 Δ r_{t-1,i} + β3 π^e_{t-1,i} + ε_{t,i}  (Equation (2) — variables described below)  
  - Variable definitions:  
    - P_{1998,i}: 1998 price level of NMS relative to EU15 (proxy for initial price level / price-level convergence effects).  
    - Δ r_{t-1,i}: change in interest rates (proxy for EU integration effects).  
    - π^e_{t-1,i}: energy inflation (proxy for global commodity price shocks).  
  - Rationale: initial price levels capture price adjustments toward EU15 levels (Balassa-Samuelson, quality convergence); change in interest rates proxies EU integration effects; energy prices capture global commodity shocks.  
- Modeling the country-specific component:
  - Country-specific regression (as in source):  
    θ_t = δ + γ1 Δ y_{t-1} + γ2 r_{t-1} + γ3 Δ neer_{t-1} + γ4 π^f_{t-1} + ν_t + r_f  (Equation (3) in the source; notation preserved)  
  - Variable definitions:  
    - y_{t-1}: output gap (domestic cyclical conditions).  
    - Δ neer_{t-1}: percent change of nominal effective exchange rate (openness/exchange rate pass-through).  
    - r_{t-1}: real interest rate.  
    - π^f_{t-1}: foreign inflation.  
  - Note: changes in regulated prices and indirect taxes are not explicitly modeled and are captured in the residual term.  
- Data used:
  - First-stage GDFM: 955 predominantly four-digit level HICP series for the NMS, January 1998–June 2008 monthly. Where four-digit not available, next aggregation level used (about 1/4 of dataset). All HICP from Eurostat.  
  - Data preprocessing: first- and twelve-month differences of HICP components; seasonal adjustment by X12; normalization (subtract mean, divide by standard deviation) for month-on-month and year-on-year regressions.  
  - Second-stage (non-HICP) data: industrial production (seasonally adjusted), interest rates, nominal effective exchange rates (monthly), price levels of NMS relative to EU15 (annual). Quarterly and annual models also estimated by averaging monthly data; quarterly models replace industrial production with real GDP. Nominal effective exchange rates, industrial production for most countries, and price levels from Eurostat; remaining industrial production data from Haver Analytics.

### V. Discussion of results — empirical findings and interpretation
- GDFM decomposition:
  - The GDFM extracts a small number of orthogonal common shocks that explain covariation across series; inflation in each country is represented as a weighted sum of the common shocks and an idiosyncratic component.  
  - Covariation across countries is attributed to q common unobservable factors; country-specific variation arises from country-specific shocks plus translated common factors via country-specific loadings and lags.  
- Exchange rate regime and inflation:
  - NMS exchange rate regimes vary from rigid (Baltics, Bulgaria: currency boards or peg with very narrow band) to fully floating/managed (e.g., Hungary).  
  - With only 10 countries, econometric estimation of regime impact on inflation is not feasible; instead, correlation of exchange rate regime with inflation controlling for income level is assessed. Figure 3 result: exchange rate regime does not seem to be a significant driver of inflation in the NMS — countries are approximately randomly distributed around the inflation–income line given income level (regression line: y = -0.3147x + 55.268, R^2 = 0.1449).  
- Indirect pass-through and inflation anchoring:
  - Commodity price shocks have more pronounced direct effects on headline inflation in NMS due to higher food and energy shares; second-round effects on core inflation are generally present but limited in magnitude, with country-specific patterns (see VAR findings above).  
  - Inflation expectations appear generally well anchored in the NMS; food inflation has a measurable but small effect on expectations (1 percentage point higher food inflation → 0.1 percentage point higher inflation expectations).  

*Italic: Source — _wp0951 - References (PDF chapter/section) — content as provided in the supplied document.*

### 0.52 on the horizontal axis corresponds to twelve months. The region to the left of that point

### _wp0951 - 0.52 on the horizontal axis corresponds to twelve months. The region to the left of that point

### Common-factor decomposition and variance explained
- 0.52 on the horizontal axis corresponds to twelve months. The region to the left of that point correspond to periods greater than, while the region to the right to less than twelve months.
- On the vertical axis, the area below the eigenvalues indicates the cumulative share of the variance of the data set explained by each consecutive common factor.
- The first common factor (largest eigenvalue):
  - Explains about 65 percent of the data variability at a medium-term horizon.
  - Explains about 70 percent of the data variability at shorter horizons.
- Marginal contributions decline rapidly across factors; the marginal contribution of the third principal component is considerably smaller than the second.
- Threshold for identifying common shocks chosen at about 50 percent (following Cristadoro and others (2005)) to avoid bias from excessive noise at higher frequencies.
- Conclusion from spectral analysis: the first factor is chosen as the single dynamic common factor of inflation in the NMS; remaining factors contribute to country-specific components.
- Number of static factors chosen using Bai and Ng (2002) algorithm:
  - Criterion reaches a minimum at 7 static factors.
  - Given one common shock from spectral analysis, this implies six lags for the dynamic common factor in equation (1).

### Share of inflation variation explained by dynamic common factor(s)
- In the time domain, with one dynamic factor:
  - Explains about 15 percent in the Baltics (grouping referenced in text).
  - Explains around 45 percent in the Czech Republic and Slovakia.
  - Explains 75 percent in Hungary and Poland.
  - Explains over 80 percent in Bulgaria and Romania.
- Adding a second dynamic factor increases the share of explained inflation variation to above 80 percent on average for the NMS, indicating the two common factors differ in importance across NMS.

### Determinants of common inflation (pooled regression results)
- The pooled regression suggests the relative price level, EU integration forces (proxied by the convergence of nominal interest rates), and energy prices explain over 40 percent of the variability of common inflation in the NMS.
- Table 2 coefficients and fit (Dependent variable: common inflation derived from the GDFM model; Interest rate change in percentage points; relative price level and energy inflation in percent):
  - Relative price level in 1998: Coefficient -0.05 (standard error (0.02))
  - Change in interest rates: Coefficient -0.35 (standard error (0.05))
  - Energy inflation: Coefficient 0.12 (standard error (0.01))
  - Adjusted R-squared: 0.44
- Interpretation of coefficients:
  - Initial relative price level: estimated coefficient negative and significant; implies that in a NMS with a price level 30 percent below the level observed in the EU15 countries, common inflation would be about 1½ percentage points.
  - EU integration (convergence of interest rates): estimated coefficient implies that a country that experiences a 1 percentage point decline in interest rates would have about 1/3 of a percentage point higher common inflation.
  - Energy prices: estimated coefficient implies that a 1 percent increase in energy prices accounts for 0.12 percentage points common inflation.

### Determinants of country-specific inflation (idiosyncratic component)
- Overall, cyclical factors, financial conditions, and openness explain about half of the variation of the country-specific inflation.
- Cyclical factors (captured by the output gap) are statistically significant in most NMS.
- Financial conditions (captured by the real long-term interest rate) are also a significant determinant of the idiosyncratic component.
- Selected significant coefficients from Table 3 (Dependent variable: country-specific component of inflation; Long-term interest rate in percent; other variables year-on-year, in percent; standard errors in parenthesis; n.s.—not statistically significant at 1 percent, 5 percent, or 10 percent):
  - Czech Republic:
    - Output gap: 0.08 (0.03)
    - Real interest rates: -0.90 (0.05)
    - Nominal effective exchange rate: -0.04 (0.01)
    - Adjusted R-squared: 0.84
  - Estonia:
    - Output gap: 0.16 (0.05)
    - Real interest rates: -0.38 (0.10)
    - Nominal effective exchange rate: -0.08 (0.02)
    - Adjusted R-squared: 0.40
  - Hungary:
    - Output gap: 0.12 (0.06)
    - Real interest rates: -0.40 (0.07)
    - Adjusted R-squared: 0.40
  - Latvia:
    - Output gap: 0.1 (0.07)
    - Real interest rates: -0.70 (0.05)
    - Nominal effective exchange rate: 0.50 (0.05)
    - EU inflation: 1.8 (0.50)
    - Adjusted R-squared: -0.70 (as printed in table)
  - Lithuania:
    - Output gap: 0.05 (0.03)
    - Real interest rates: -0.65 (0.06)
    - Nominal effective exchange rate: 0.16 (0.05)
    - Adjusted R-squared: 0.70
  - Poland:
    - Output gap: 0.20 (0.04)
    - Real interest rates: -0.05 (0.03)
    - Nominal effective exchange rate: -0.03 (0.01)
    - EU inflation: 0.6 (0.20)
    - Adjusted R-squared: 0.30
  - Slovak Republic:
    - Output gap: n.s.
    - Real interest rates: -0.45 (0.05)
    - Nominal effective exchange rate: -0.20 (0.05)
    - EU inflation: n.s.
    - Adjusted R-squared: 0.55
  - Slovenia:
    - Output gap: n.s.
    - Real interest rates: n.s.
    - Nominal effective exchange rate: -0.11 (0.04)
    - EU inflation: 1.3 (0.40)
    - Adjusted R-squared: 0.40
  - Bulgaria:
    - Output gap: n.s.
    - Real interest rates: n.s.
    - Nominal effective exchange rate: -0.40 (0.12)
    - EU inflation: 3.00
    - Adjusted R-squared: 0.35
  - Romania:
    - Output gap: n.s.
    - Real interest rates: -0.35 (0.07)
    - Nominal effective exchange rate: -0.16 (0.02)
    - EU inflation: 1.00 (0.06)
    - Adjusted R-squared: 0.45

### Concluding remarks (summary of implications)
- A significant part (about 65 percent) of inflation in the NMS over the medium term is driven by common factors.
- Among common factors, price level convergence and convergence of real interest rates are significant determinants of common inflation, as are commodity (energy) prices.
- Idiosyncratic factors have also played a role in the inflation process:
  - Domestic cyclical conditions and financial conditions play an important role for country-specific inflation in the NMS.
  - Pass-through effects from foreign prices have also played a role.

*Source: IMF staff estimates and calculations (as presented in the provided content).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2009/_wp0951.pdf_