## wpiea2023049-print-pdf

## Source details

**Canonical URL:** [wpiea2023049-print-pdf](https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023049-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2023/english/wpiea2023049-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2023/english/wpiea2023049-print-pdf.pdf.json)

---

### 1. INTRODUCTION — Motivation and context
- Inflation surged to levels not seen in multiple decades in many countries due to a strong recovery of demand after the pandemic, supply chain disruptions, and commodity price increases, exacerbated by Russia’s war in Ukraine.
- In the Western Balkan region, inflation increased to 14.5 percent in December year-on-year, as compared to 9.2 percent in the Euro Area.
- Western Balkans: Albania, Bosnia and Herzegovina, Kosovo, Montenegro, North Macedonia, and Serbia.
- Upswing in inflation is broad-based, raising questions about drivers, persistence, and appropriate policy responses.

### 1. INTRODUCTION — Empirical strategy
- Two complementary approaches:
  - Reduced-form Phillips curve model augmented with international food and energy prices; estimated on panel quarterly data using robust regressions (Huber (1964) M-estimators).
  - Structural VAR (SVAR) model including real GDP growth, inflation, domestic monetary or fiscal policy, and a measure of the external price environment; identification via sign restrictions to isolate supply, demand, monetary/fiscal policy, and external price shocks.
- Rationale for panel approach: similarities of shocks in the region and limited country-specific data.

### 1. INTRODUCTION — Key empirical findings (both approaches)
- International food prices:
  - Important determinant of inflation in the Western Balkans.
  - Affect headline inflation through domestic food prices and also affect core inflation.
  - Domestic food inflation plays a role in determining inflation expectations.
  - High persistence of inflation implies international food price shocks can have a large and lasting impact; returning inflation to pre-shock levels would be gradual.
- Other determinants of headline inflation:
  - Nominal effective exchange rate (NEER) and international energy prices are statistically significant determinants.
  - Core inflation is also affected by inflation expectations.
- Slack measures:
  - Output or unemployment gaps do not show consistent direct effects on inflation; they play a statistically significant role in determining inflation expectations (measurement limitations in the WB due to informality).
- SVAR summary:
  - Inflation in the WB is not just driven by external shocks; domestic factors play an important role.
  - "About two-thirds of the inflation variation in the short term and about half of the inflation variation in the medium term is attributed to domestic shocks in average for the WB."
  - Recent surge in inflation cannot be fully attributed to external shocks; domestic factors, including strong domestic demand, contributed.

### 1. INTRODUCTION — Policy implications highlighted
- From Phillips curve results:
  - Rationale may exist for temporary policies to moderate immediate and complete pass-through of international to domestic food prices during upswings; careful calibration required to ensure benefits outweigh costs and to minimize distortions and fiscal costs.
  - Priority on increasing credibility of macroeconomic stabilization policies to reduce inflation persistence in the WB.
- From SVAR results:
  - Domestic macroeconomic stabilization policies have an important role in keeping inflation low and stable in the WB.
  - Any measures to curb pass-through from international food prices require careful design to minimize fiscal costs and avoid fueling demand pressures.

### Stylized facts (contextual)
- Since mid-2022, WB economies experienced inflation rates unseen in the recent past; prior low-inflation environment after 1990s hyper-inflation episodes.
- Exchange rate regimes vary: unilateral euroization (Montenegro and Kosovo), currency board (Bosnia and Herzegovina), stabilized arrangement (Serbia and North Macedonia), floating regime (Albania). With formal/informal euroization, average WB inflation tracked Euro Area inflation until accelerating faster and exceeding it in 2022.
- Food shares in CPI:
  - Simple average weight of food in CPI basket for the region: around 42 percent.
  - Staple foods (cereals and vegetable oils) combined share in food basket averaging around 9.4 percent in the WB, as compared to 3.2 percent in the EU.
  - Public spending on agriculture (percent of agricultural GDP): about 10% on average in the WB versus 21% in the EU (proxy for government regulation).
- Energy share in CPI similar to the EU; lower overall energy price inflation in most WB countries relative to the EU attributed to higher reliance on domestically generated energy and low reliance on natural gas.
- Inflation breadth and persistence:
  - Producer price growth in the WB lower than in the EU (partly due to relatively lower energy inflation).
  - Wages increasing; average wage growth picking up and catching up with headline inflation in recent months.
  - Minimum wage increases and recovery-driven demand pressures contributed to cost and demand-side inflationary pressures.
- Inflation expectations rose:
  - Central bank survey-based expectations in Albania and Serbia exceeded targets (Albania from February 2022; Serbia from December 2021), though they lag actual inflation.
  - One-year ahead Consensus Forecast data for 4 of 6 WB economies show a pick-up from February 2022.
- Policy responses:
  - Independent monetary policy countries raised policy rates several times since early 2022 (effectiveness limited by high euroization).
  - Fiscal and targeted support: one-off assistance in Albania, Bosnia and Herzegovina, Kosovo, and Serbia; sizable public wage and pension increases in Bosnia and Herzegovina.
  - Energy tariff adjustments and subsidies: partial household electricity tariff adjustments with subsidies in Kosovo, North Macedonia, Serbia; subsidies to energy sector in Bosnia and Herzegovina and North Macedonia.
  - VAT and excise cuts for some food and fuel items in Montenegro, North Macedonia, Serbia; price caps and profit margin limits for critical staples and fuel in all WB countries; export bans/quotas in Serbia and Bosnia and Herzegovina.

### 3. LITERATURE REVIEW — Overview and context
- Renewed interest due to global inflation increase; relates to literature on inflationary impacts of COVID-19, supply chain disruptions, and commodity price surges.
- Historical pandemics tended to lower trend inflation, but COVID-19 was different due to unprecedented policy response, fast vaccination, and adaptation to virtual work.

### 3. LITERATURE REVIEW — Key findings from recent literature
- Supply bottlenecks major contributors:
  - Celasun et al. (2022): in the Euro Area half of producer price inflation of manufactured goods in 2021 explained by supply bottlenecks, half by higher demand.
  - Carriere-Swallow et al. (2022): shocks to global shipping costs create supply bottlenecks with more persistent impact on domestic inflation than global commodity price shocks.
- Commodity prices and cross-country heterogeneity:
  - IMF (2022a): rising commodity prices explain around two thirds of the current surge of inflation in Europe.
  - Heterogeneity explained by food and energy weights, energy mix, price regulations, policy measures, and exchange rate developments.
- Policy response and expectations:
  - Initially timid policy responses may have contributed to underestimating the strength of the 2021 recovery and persistence of shocks.
- Second-round effects and pass-through:
  - Baba and Lee (2022): pass-through of commodity price shocks to wages in Europe larger if overall inflation is higher when the shock hits; larger for emerging than advanced European economies.
  - Boranova et al. (2021): wage growth historically led to high inflation but post-Global Financial Crisis impact weakened; pass-through higher in periods of high inflation.
- Western Balkans specific:
  - World Bank (2022): headline inflation in the WB started to increase through supply-side pressures, with strong consumer demand and high passthrough effects contributing to higher core inflation.

### 3. LITERATURE REVIEW — Methodological approaches and this paper’s empirical approach
- Augmented Phillips curve widely used; augmentations include global commodity prices, foreign price developments, exchange rates, global energy and food prices.
- SVARs use sign and zero restrictions; sign restriction approach follows Rubio-Ramirez et al. (2010); Bayesian estimation follows Uhlig (1994).
- This paper’s Phillips curve formal variables: 휋휋푖푖,푡푡; 휋휋푖푖,푡푡퐶퐶퐶퐶; 푦푦�푖푖,푡푡; NEER; international food price; international energy price; country fixed effects; error term.
- Data: Quarterly 2007Q1-2022Q3 for six WB economies; robust panel regression with country fixed effects preferred; data limitations noted (quality, short span, measures of slack).

### 3. LITERATURE REVIEW — Empirical results (Phillips curve estimates summary)
- Persistence:
  - Lagged dependent variable coefficient averages around 0.5 across specifications for all measures of inflation (headline, core, trimmed-mean 10% and 30%).
- International food prices:
  - Statistically significant impact on all measures of inflation, including core inflation in most specifications.
- NEER:
  - Important driver of inflation; reflects exchange rate pass-through or relative price pass-through.
- International energy prices:
  - Statistically significant in all regressions.
- Inflation expectations:
  - Significant for core inflation.
- Output gap:
  - Coefficient not statistically significant in any equations; attributed to measurement difficulty of output gaps in the WB.

### 3. LITERATURE REVIEW — Determinants of domestic food and energy prices
- Unprocessed food prices: affected by NEER and international food prices.
- Processed food prices: primarily affected by international food prices.
- Domestic energy price inflation: international energy prices are statistically significant determinants.
- High and statistically significant persistence coefficients in processed food, unprocessed food, and domestic energy regressions.

### 3. LITERATURE REVIEW — Inflation expectations formation (Table 3 results)
- Column 1:
  - Higher headline inflation increases one-year ahead inflation expectations (backward-looking).
  - NEER coefficient statistically significant and positive.
  - Output gap coefficient statistically significant and positive.
- Column 2 (components model):
  - Core inflation coefficient positive and statistically significant.
  - Domestic food inflation (processed and unprocessed) associated with higher inflation expectations.
- Implication: increases in domestic food prices could have longer-lasting impacts via inflation expectations.

### 3. LITERATURE REVIEW — Summary conclusions and policy implications
- Headline inflation in the WB driven by:
  - Its own persistence.
  - International food and energy prices.
  - NEER.
- Core inflation also influenced by inflation expectations.
- Output gap affects inflation mainly through inflation expectations.
- Policy implications:
  - Pass-through mitigation: policies to curb pass-through from international to domestic food prices may merit consideration but should be temporary, targeted, and designed to minimize fiscal costs and distortions.
  - Demand stabilization and expectations: macroeconomic policies stabilizing aggregate demand may help stabilize inflation via expectations.
  - Credibility and macro stabilization: increase credibility of macro policies; for countries with independent monetary policy maintain appropriately tight monetary policy; all WB countries should keep fiscal policy in check to cool domestic demand and anchor inflation expectations.

### 5. STRUCTURAL VAR ANALYSIS — SVAR setup and identification
- Quarterly data through 2022 Q3; country-specific SVARs.
- Common variables: real GDP growth, headline CPI inflation rate, external price pressure index (or Euro Area inflation).
- Albania, Serbia, North Macedonia: include proxy for monetary policy rate.
- Montenegro, Kosovo, Bosnia and Herzegovina: include government spending (percent of GDP) as fiscal policy measure.
- Estimation: reduced-form VAR with four variables estimated using a Bayesian approach; structural shocks identified via sign restrictions (Rubio-Ramirez et al., 2010).
- EPPI definition: import-weighted producer price index of countries from which the subject country imports, converted to local currency using NEER, and relative to percent change of subject country’s GDP deflator.

### 5. STRUCTURAL VAR ANALYSIS — Identified shocks by country grouping
- Independent monetary policy countries:
  - Supply shock: increases output and suppresses inflation.
  - Non-monetary demand shock: increases output and inflation.
  - Domestic monetary shock: increases policy rate and suppresses inflation.
  - Foreign (external) shock: increases EPPI and domestic inflation.
- Countries without independent monetary policy (Kosovo, Montenegro, BiH):
  - Supply shock; private demand shock; fiscal policy shock; external price shock (Euro Area inflation used).
  - Fiscal shock: increases CPI inflation, real GDP growth, and government spending.
  - Private demand shock: increases output and inflation but not government spending.

### 5. STRUCTURAL VAR ANALYSIS — Key SVAR results
- Variance decompositions:
  - Domestic factors (demand and macro policy shocks) play significant roles alongside external price shocks.
  - "About two-thirds of the inflation variation in the short term and about half of the inflation variation in the medium term is attributed to domestic shocks in average for the WB."
- Country-specific dominant domestic shocks:
  - Albania and North Macedonia: aggregate demand and monetary policy shocks most relevant.
  - Serbia: supply shocks larger role in the short term.
  - Kosovo and Bosnia and Herzegovina: fiscal shocks most important domestic driver.
  - Montenegro: private aggregate demand shocks matter more.
- Caveat for Albania: limited two-way variability of monetary policy rate may overestimate monetary policy impact; IMF staff and Bank of Albania estimate exchange rate fluctuations played a more prominent role recently.
- Historical decompositions:
  - External shock impact is clear for recent surge, but domestic factors contributed in all countries to varying degrees.
  - Collapse of aggregate demand during COVID-19 dragged inflation down.

### 5. STRUCTURAL VAR ANALYSIS — Policy implications
- Maintain adequately tight macroeconomic policies as external commodity price shocks moderate to help reduce inflation.
- Domestic macroeconomic stabilization policies and credibility are important to anchor inflation and reduce persistence, even where monetary policy autonomy is limited.

### Appendix I — Stylized facts charts and tables (selected numeric snapshots)
- CPI basket composition (average level between 2008M1 and 2022M11):
  - Processed food: 23.8
  - Unprocessed food: 18.3
  - Energy: 8.7
  - Core PCE: 49.1
- Country averages (selected):
  - Albania (2008M1–2022M11): Processed Food 23.4; Unprocessed Food 13.4; Energy 21.4; Core PCE 41.7
  - Bosnia and Herzegovina (2016M1–2022M11): Processed food 30.3; Unprocessed 16.1; Energy 13.2; Core PCE 40.4
  - Kosovo (2007M1–2022M11): Processed Food 24.2; Unprocessed Food 17.2; Energy 17.7; Core PCE 40.9
  - Montenegro (2011M1–2022M11): Processed Food 27.9; Unprocessed Food 16.6; Energy 14.9; Core PCE 40.6
  - North Macedonia (2010M1–2022M11): Processed Food 25.1; Unprocessed Food 15.4; Energy 18.3; Core PCE 41.2
- Share of Basic Commodities in CPI, 2022 (shares in CPI basket):
  - Cereal: ALB 2.0; BIH 2.2; MNE 1.7; EU 0.4
  - Wheat: ALB 1.1; BIH 1.6; MNE 1.2; EU 0.1
  - Vegetable oil: ALB 2.1; BIH 1.6; MNE 1.5; EU 0.3
  - Overall basic commodities: ALB 4.0; BIH 3.8; MNE 3.2; EU 0.7
- Share of Basic Commodities in “Food and Beverage” category, 2022:
  - Overall: ALB 10.0; BIH 9.9; MNE 8.2; EU 3.2
- Public Spending on Agriculture (percent of Agricultural GDP):
  - ALB 3.0; BIH 8.0; KOS 9.3; MKD 16.2; MNE 8.1; SRB 12.5; EU 20.5

### Appendix III — Descriptive statistics (selected)
- Headline inflation: Obs. 370; Mean 3.4; Stand. Dev. 5.2; Min -7.7; Max 28.6
- Core inflation: Obs. 290; Mean 3.4; Stand. Dev. 5.8; Min -8.2; Max 40.8
- Trimmed-mean 10%: Obs. 308; Mean 0.7; Stand. Dev. 0.9; Min -1.0; Max 4.9
- Trimmed-mean 30%: Obs. 308; Mean 0.2; Stand. Dev. 0.5; Min -0.4; Max 2.7
- Processed food inflation: Obs. 323; Mean 5.5; Stand. Dev. 10.2; Min -8.3; Max 74.0
- Unprocessed food inflation: Obs. 323; Mean 11.5; Stand. Dev. 22.0; Min -27.8; Max 144.9
- Energy inflation: Obs. 327; Mean 5.5; Stand. Dev. 14.6; Min -33.3; Max 97.3
- Inflation expectation: Obs. 224; Mean 2.3; Stand. Dev. 2.4; Min -1.0; Max 10.7
- Output gap: Obs. 354; Mean 0.0; Stand. Dev. 2.9; Min -19.6; Max 10.4
- NEER: Obs. 371; Mean -0.6; Stand. Dev. 5.1; Min -17.8; Max 32.1
- International food price: Obs. 371; Mean 5.6; Stand. Dev. 21.2; Min -47.8; Max 60.2
- International energy price: Obs. 371; Mean 6.5; Stand. Dev. 54.8; Min -178.3; Max 109.8

### Appendix — Robustness and key regression coefficient highlights
- Robustness (Unemployment gap instead of Output gap; Table A3.2):
  - Lagged dependent variable: 0.648*** (0.054) in column (1); 0.664*** (0.041) in column (2).
  - 1y inflation expectation (t-1): 0.031 (0.099) in column (1); 0.515*** (0.090) in column (2).
  - Unemployment gap: 0.119 (0.116) in column (1); 0.088 (0.093) in column (2).
  - NEER: 0.074** (0.033) in column (1); 0.069** (0.029) in column (2).
  - International food price inflation: 0.022** (0.009) in column (1); 0.021*** (0.008) in column (2).
  - International energy price inflation: 0.011*** (0.003) in column (1); 0.013*** (0.003) in column (2).
  - Observations: 172 (col 1); 252 (col 2). R-squared: 0.654 (col 1); 0.655 (col 2). Number of countries: 4.
- Using External Price Pressure instead of NEER (Table A3.4 — selected coefficients):
  - Column (1) Headline: Lagged dependent variable 0.720*** (0.067); Int.l' food price inflation 0.031*** (0.012); Int.l' energy price inflation 0.007* (0.004); Observations 135; R-squared 0.673.
  - Column (2) Core PCE: Lagged dependent variable 0.645*** (0.042); 1y inflation expectation (t-1) 0.528*** (0.100); Int.l' food price inflation 0.026*** (0.008); Int.l' energy price inflation 0.013*** (0.003); Observations 236; R-squared 0.662.
  - Across columns: lagged dependent variable consistently positive and significant; international food price inflation positive and often significant; international energy price inflation smaller but significant in several specifications; External Price Pressure coefficients generally small and imprecise.

_Italic: IMF Working Paper — Inflation Dynamics in the Western Balkans (selected sections from wpiea2023049-print-pdf)._

### 1. INTRODUCTION ________________________________________________________________ 5

### 1. INTRODUCTION

### Motivation and context
- Inflation surged to levels not seen in multiple decades in many countries due to a strong recovery of demand after the pandemic, supply chain disruptions, and commodity price increases, exacerbated by Russia’s war in Ukraine.
- In the Western Balkan region, inflation increased to 14.5 percent in December year-on-year, as compared to 9.2 percent in the Euro Area.
- The Western Balkans include Albania, Bosnia and Herzegovina, Kosovo, Montenegro, North Macedonia, and Serbia.
- The current upswing in inflation is broad-based, raising questions about drivers, persistence, and appropriate policy responses.

### Empirical strategy
- Two complementary approaches are used:
  - Reduced-form Phillips curve model augmented with international food and energy prices, estimated on panel quarterly data using robust regressions (Huber (1964) M-estimators).
  - Structural VAR (SVAR) model including real GDP growth, inflation, domestic monetary or fiscal policy, and a measure of the external price environment; identification via theoretically motivated sign restrictions to isolate shocks to aggregate supply, aggregate demand, monetary or fiscal policy, and external price pressure.
- Rationale for panel approach: similarities of shocks in the region and limited country-specific data.

### Key empirical findings (from both approaches)
- International food prices are an important determinant of inflation in the Western Balkans:
  - They affect headline inflation through domestic food prices (food accounts for a large share of the consumption basket) and also affect core inflation.
  - Domestic food inflation plays a role in determining inflation expectations.
  - High persistence of inflation implies international food price shocks can have a large and lasting impact; bringing inflation back to pre-shock levels would be gradual.
- Other determinants of headline inflation:
  - The nominal effective exchange rate and international energy prices are statistically significant determinants of headline inflation.
  - Core inflation is also affected by inflation expectations.
- Slack measures (output or unemployment gaps) do not show consistent direct effects on inflation, though they play a statistically significant role in determining inflation expectations (with caution due to measurement limitations in the WB, where informality is high).
- SVAR results:
  - Inflation in the WB is not just driven by external shocks; domestic factors play an important role.
  - About two-thirds of the inflation variation in the short term and about half of the inflation variation in the medium term can be attributed to domestic shocks on average for the WB over the entire sample period.
  - The recent surge in inflation cannot be fully attributed to external shocks; domestic factors, including strong domestic demand, contributed to it.

### Policy implications highlighted
- Two-fold implications from the Phillips curve results:
  - There may be a rationale for temporary policies aiming to moderate the immediate and complete pass-through of international to domestic food prices during upswings. Careful calibration is key to ensure benefits outweigh costs and to minimize distortions to price signals and fiscal costs.
  - Increasing the credibility of macroeconomic stabilization policies should be prioritized to reduce inflation persistence in the WB.
- SVAR-based implications:
  - Domestic macroeconomic stabilization policies have an important role in keeping inflation low and stable in the WB.
  - Any measures to curb pass-through from international food prices to domestic inflation require careful design to minimize fiscal costs and avoid fueling demand pressures that could further raise inflation.

### Additional contextual stylized facts summarized
- Since mid-2022, WB economies experienced inflation rates unseen in the recent past; prior to this, the region enjoyed a relatively low-inflation environment after the 1990s hyper-inflation episodes.
- Exchange rate regimes in the region vary: unilateral euroization (Montenegro and Kosovo), currency board (Bosnia and Herzegovina), stabilized arrangement (Serbia and North Macedonia), and floating regime (Albania). With formal or informal euroization, average WB inflation tracked Euro Area inflation increasingly closely until accelerating faster and exceeding it in 2022.
- Food plays a relatively large role in WB CPIs:
  - The simple average weight of food in the CPI basket for the region is around 42 percent.
  - Staple foods (cereals and vegetable oils) have a combined share in the food basket averaging around 9.4 percent in the WB, as compared to 3.2 percent in the EU.
  - Public spending on agriculture in percent of agricultural GDP is about 10% on average in the WB versus 21% in the EU (used as a proxy for government regulation).
- Energy share in CPI is similar to the EU; lower overall energy price inflation in most WB countries relative to the EU is attributed to higher reliance on domestically generated energy (especially electricity) and low reliance on natural gas. Higher international gasoline prices did pass through, increasing transport-sector prices.
- Inflation has become increasingly broad-based and persistent:
  - Producer price growth in the WB has been lower than in the EU (partly due to relatively lower energy inflation).
  - Wages have been increasing in the region, with average wage growth picking up and catching up with headline inflation in recent months.
  - Minimum wage increases and recovery-driven demand pressures contributed to cost and demand-side inflationary pressures.
- Inflation expectations rose:
  - Central bank survey-based expectations in Albania and Serbia exceeded targets (Albania from February 2022; Serbia from December 2021), though they lag actual inflation.
  - One-year ahead Consensus Forecast data for 4 of 6 WB economies show a pick-up from February 2022.
- Policy responses in the region:
  - Countries with independent monetary policy raised policy rates several times since early 2022 (effectiveness limited on the credit channel due to high euroization).
  - Fiscal and targeted support measures: one-off assistance packages in Albania, Bosnia and Herzegovina, Kosovo, and Serbia; sizable public wage and pension increases in Bosnia and Herzegovina.
  - Energy tariff adjustments and subsidies: household electricity tariffs partially adjusted in Kosovo, North Macedonia, and Serbia with accompanying subsidies; subsidies to the energy sector in Bosnia and Herzegovina and North Macedonia to support stable tariffs.
  - Other measures: VAT and excise tax cuts for some food and fuel items in Montenegro, North Macedonia, and Serbia; price caps and profit margin limits for critical staple food and fuel items in all WB countries; export bans or quotas for some critical items in Serbia and Bosnia and Herzegovina.

*IMF Working Paper — Inflation Dynamics in the Western Balkans (1. INTRODUCTION section)*

### 3.   Literature Review

### 3.   Literature Review

### Overview and context
- Increasing inflation worldwide has renewed interest in explaining the determinants of inflation and shifted focus from low and stable inflation or deflation to high inflation.
- This paper relates to the literature studying the inflationary impacts of the multiple shocks of COVID-19, supply chain disruptions, and commodity price surges, especially in Europe.
- Historical analysis suggests past pandemics led to a significant decline in trend inflation (Bonam and Smadu, 2021), but COVID-19 appeared different given the unprecedented policy response, fast vaccination, and adaptation to virtual work in advanced and partly emerging market economies.

### Key findings from recent literature
- Supply bottlenecks were major contributors to inflation during the pandemic:
  - Celasun et al. (2022) estimate that in the Euro Area half of the producer price inflation of manufactured goods in 2021 was explained by supply bottlenecks due to disruptions in production facilities and shortage of intermediate inputs while the other half was explained by higher demand.
  - Carriere-Swallow et al. (2022) find that shocks to global shipping costs that create supply bottlenecks tend to have a more persistent impact on domestic inflation compared to shocks to global commodity prices.
- Commodity prices and cross-country heterogeneity:
  - IMF (2022a) estimates that rising commodity prices explain around two thirds of the current surge of inflation in Europe.
  - Cross-country heterogeneity in inflation dynamics can be explained by differences in food and energy weights in CPI baskets, differences in the energy mix, price regulations, policy measures, and exchange rate developments.
- Policy response and expectations:
  - Initially timid policy responses may have played a role as the strength of the 2021 recovery may have been underestimated, supply shocks were seen as temporary, and inflation expectations were believed anchored (Reis, 2022).
- Second-round effects and pass-through:
  - Baba and Lee (2022) find that the pass-through of commodity price shocks to wages in Europe is larger if overall inflation is higher when the commodity price shock hits, and the pass-through is larger for emerging than advanced European economies.
  - Boranova et al. (2021) show wage growth historically led to high inflation but since the Global Financial Crisis the impact weakened; pass-through is significantly higher in periods of high inflation.
- Western Balkans specific observations:
  - World Bank (2022) notes headline inflation in the Western Balkans started to increase through supply side pressures, with strong consumer demand and high passthrough effects also contributing to higher core inflation.

### Methodological approaches in the literature
- Augmented Phillips curve:
  - Widely used to study drivers of the recent inflation surge; augmentations include global commodity prices, foreign price developments, exchange rates, and global energy and food prices.
  - IMF (2022a) estimates Phillips curves for 24 advanced and 7 emerging European economies, finding inflation strongly associated with lower economic slack and higher past and expected future price developments; emerging European economies respond more strongly to foreign price developments than advanced economies.
  - In that analysis, the Phillips curve could explain up to 60 percent of the recent surge in inflation, leaving the remaining 40 percent unexplained. Regressing residuals on additional variables, IMF (2022a) finds around 30 percent of the recent surge could be associated to pandemic-driven supply bottlenecks.
- Vector autoregression (VAR) and structural VAR (SVAR):
  - SVARs allow endogenous interdependencies and use sign and zero restrictions to identify structural shocks; sign restriction approach dates to Faust (1998) and Uhlig (2005).
  - The sign restriction approach used in this paper follows Rubio-Ramirez et al. (2010) and the Bayesian estimation method follows Uhlig (1994).
  - IMF (2022c) finds in New Zealand domestic shocks have a slightly larger role in determining inflation but external shocks contributed more to the recent surge.
  - Celasun et al. (2022) use SVAR and find supply bottlenecks had an important impact on inflation.
- Local projections:
  - Baba and Lee (2022) and Binici et al. (2022) use local projection method (econometrically equivalent to VAR) and find inflation in Europe has become more responsive to both domestic and global shocks in the post-pandemic period.

### Empirical approach used in this paper
- A reduced-form Phillips curve augmented to cover foreign price developments relevant for the Western Balkans is estimated:
  - Formal equation variables (as specified): 휋휋푖푖,푡푡 (quarter-over-quarter annualized inflation), 휋휋푖푖,푡푡퐶퐶퐶퐶 (expected inflation), 푦푦�푖푖,푡푡 (output gap or unemployment gap), 푁푁푁푁푁푁푁푁푖푖,푡푡 (quarterly change in nominal effective exchange rate), 퐹퐹퐹퐹퐹퐹퐹퐹푡푡 (quarterly growth in international food price), 푁푁퐸퐸퐸퐸퐸퐸퐸퐸 푦푦𝑡 (quarterly growth in international energy price), plus country fixed effects 퐹퐹푁푁푖푖 and error term 휀휀푖푖,푡푡.
- Data and estimation:
  - Quarterly data covering the period of 2007Q1-2022Q3 for six economies of the Western Balkans.
  - Estimation method: robust panel regression correcting standard errors for heteroscedasticity and autocorrelation; preferred specification includes country fixed effects but not time fixed effects.
  - Data limitations noted: quality issues, short span of time series; measures of slack (output gap and unemployment gap) are particularly challenging in the Western Balkans due to large informality and difficulty assessing potential output and unemployment; inflation expectations data from Consensus Forecast are available for four out of six countries (leading some specifications to exclude Kosovo and Montenegro).

### Empirical results — Phillips curve estimates (summary)
- Persistence:
  - For all measures of inflation (headline, core, trimmed-mean 10% and 30%), strong persistence is found, with the lagged dependent variable coefficient averaging around 0.5 across specifications.
- International food prices:
  - International food prices have a statistically significant impact on all measures of inflation, including core inflation (non-food, non-energy inflation) in most specifications.
  - Possible mechanisms: food as an input to services (restaurants, tourist accommodation); large share of food in CPI influencing price-setting and wages.
- Nominal effective exchange rate (NEER):
  - NEER is an important driver of inflation; for autonomous currency countries this reflects exchange rate pass-through, for fixed/no separate legal tender it suggests relative price passthrough.
- International energy prices:
  - International energy price inflation is statistically significant in all regressions.
- Inflation expectations:
  - Inflation expectations appear to be significant for core inflation.
- Output gap:
  - Coefficient associated with the output gap is not statistically significant in any of the equations; attributed to difficulty of accurately measuring output gaps in the Western Balkans.

- Robustness:
  - Results robust to different specifications.
  - Appendix Table A3.2 (unemployment gap used instead of output gap): unemployment gap not statistically significant; NEER not consistently significant in those specifications.
  - Appendix Table A3.3 (fixed effect OLS): results in line with Table 1, except NEER not consistently statistically significant.
  - Appendix Table A3.4 (external price pressure index (EPPI) instead of NEER): EPPI coefficient not consistently statistically significant in most specifications.

### Determinants of domestic food and energy prices
- Separate analysis of processed and unprocessed food prices:
  - Unprocessed food prices are affected by the NEER and international food prices.
  - Processed food prices are primarily affected by international food prices.
- Domestic energy price inflation:
  - International energy prices are statistically significant determinants.
- Persistence:
  - In all three regressions summarized in Table 2 (processed food, unprocessed food, domestic energy), the coefficient of persistence of inflation is high and statistically significant.

### Inflation expectations formation (results)
- Two regressions summarized in Table 3:
  - Column 1: Higher headline inflation leads to an increase in one-year ahead inflation expectations, indicating a backward-looking component.
    - NEER coefficient is statistically significant and positive, implying agents expect more inflation when NEER depreciates.
    - Output gap has a statistically significant positive coefficient, suggesting when the economy is above potential agents expect higher inflation.
  - Column 2: Alternative model using components of headline inflation rather than aggregate:
    - Core inflation coefficient is positive and statistically significant.
    - Domestic food inflation (both processed and unprocessed) is also associated with higher inflation expectations.
- Implication: current increases in domestic food prices could have a more long-lasting impact on domestic inflation through inflation expectations.

### Summary of empirical conclusions
- In the Western Balkans, headline inflation is driven by:
  - Its own persistence.
  - International food and energy prices.
  - The nominal effective exchange rate.
- For core inflation, inflation expectations also play a role.
- All measures of inflation are highly persistent.
- Current inflation levels, especially food inflation, play a role in the formation of inflation expectations.
- The only role found for the output gap in determining inflation operates through inflation expectations.

### Policy implications
- Pass-through mitigation:
  - High pass-through of international food prices to headline inflation, and their importance for core inflation and inflation expectations, means international food price shocks can have a large and lasting impact.
  - Policies aiming to curb pass-through from international to domestic food prices may have merits because reducing high inflation triggered by large food price shocks solely through conventional monetary and fiscal policies may be more costly and difficult.
  - Caveats: such policies can have high fiscal costs and distort price signals. Measures aiming to moderate pass-through of international food and energy prices to end users should remain as temporary and as targeted as possible and be designed efficiently to minimize distortions to price signals.
- Demand stabilization and expectations:
  - Difficulty measuring economic slack notwithstanding, the statistically significant impact of the output gap on inflation expectations suggests macroeconomic policies aiming to stabilize aggregate demand may help stabilize inflation via inflation expectations.
- Credibility and macro stabilization:
  - High persistence of inflation and backward-looking inflation expectations imply increasing the credibility of macroeconomic policies is a priority in the Western Balkans.
  - For countries with independent monetary policy: need to maintain appropriately tight monetary policy in the current environment of elevated inflation driven by commodity price shocks.
  - For all Western Balkans countries: keep fiscal policy in check to cool domestic demand and help anchor inflation expectations.

*Source: 3. Literature Review — IMF Working Paper (chapter/section text provided).*

### 5.   Structural VAR Analysis

### 5.   Structural VAR Analysis

### SVAR setup and estimation approach
- Quarterly data through 2022 Q3 are used for country-specific SVARs.
- Country SVAR specifications vary by monetary policy autonomy:
  - Common variables for all countries: real GDP growth, headline CPI inflation rate, external price pressure index (or Euro Area inflation in alternative specifications).
  - Albania, Serbia, and North Macedonia (independent monetary policies): include a proxy for the monetary policy rate as one of the variables in the system.
  - Montenegro and Kosovo (unilaterally euroized) and Bosnia and Herzegovina (currency board): include government spending (in percent of GDP) as a measure of fiscal policy.
- Rationale and limitations:
  - Model assumes an interdependent system of real GDP growth, inflation rate, a measure of external inflation, and monetary policy or fiscal policy.
  - Parsimonious model choice due to short time series; results may be affected by the short span of the time series.
  - Advantage: SVAR does not rely on measures of the output gap or on inflation expectations, which are difficult to estimate for the WB economies.
- Estimation and identification:
  - Reduced-form VAR with four variables estimated using a Bayesian approach.
  - Structural shocks identified using theoretically motivated sign restrictions (Rubio-Ramirez et al., 2010 approach).
  - Procedure: draw random structural decompositions, check impulse responses against sign restrictions, retain iterations that satisfy restrictions, interpret results from a large number of draws.

### Identification of structural shocks (country groupings)
- Countries with independent monetary policy (Table 4 summary in text):
  - Identified shocks: Supply shock, Demand shock, Monetary policy shock, External price shock.
  - Supply shock: increases output and suppresses inflation.
  - Non-monetary demand shock: increases output and inflation.
  - Domestic monetary shock: increases the domestic monetary policy rate and suppresses inflation.
  - Foreign (external) shock: increases the external price pressure index (EPPI) and increases domestic inflation.
  - EPPI definition: import-weighted producer price index of countries from which the subject country imports, converted to local currency using the nominal effective exchange rate, and relative to the percent change of the subject country’s GDP deflator.
- Countries without independent monetary policy (Table 5 summary in text: Kosovo, Montenegro, BiH):
  - Identified shocks: Supply shock, Private demand shock, Fiscal policy shock, External price shock.
  - External price variable chosen: Euro Area inflation (due to exchange rate regimes).
  - Fiscal shock: increases CPI inflation, real GDP growth, and government spending.
  - Domestic private demand shock: increases output and inflation but does not affect government spending (no impact restriction on government spending to distinguish from fiscal shock).

### SVAR results: variance decompositions and historical decompositions
- Forecast error variance decompositions (Figure 3, described in text):
  - Notable result: domestic factors, especially domestic demand and macro policy shocks, play a significant role in determining inflation alongside the external price shock.
  - Exact phrasing from the source: "About two-thirds of the inflation variation in the short term and about half of the inflation variation in the medium term is attributed to domestic shocks in average for the WB."
  - Country-specific patterns:
    - Albania and North Macedonia: aggregate demand and monetary policy shocks appear as the most relevant domestic shocks.
    - Serbia: supply shocks play a larger role in the short term.
    - Kosovo and Bosnia and Herzegovina: fiscal shocks stand out as the most important domestic driver of inflation.
    - Montenegro: private aggregate demand shocks appear to matter more.
  - Caveat for Albania: "Given the lack of two-way variability of monetary policy rate in the sample, results might overestimate the impact of monetary policy on inflation dynamics in Albania. Both IMF staff research and Bank of Albania estimate a more prominent role of exchange rate fluctuations than of monetary policy in Albania in the recent past."
- Historical decompositions (Figure 4, described in text):
  - Historical decompositions show contributions from different factors to inflation deviations from trend at each point in history.
  - For the recent inflation surge: external shock impact is clear, but domestic factors also contribute to inflation in all countries to different degrees.
  - Common feature: collapse of aggregate demand during the COVID-19 pandemic dragged inflation down.

### Key findings and interpretation
- Inflation in the Western Balkans is not driven solely by external shocks; domestic factors are important contributors.
- Domestic macroeconomic stabilization policies matter for keeping inflation low and stable even where monetary policy autonomy is limited.
- Implication: while commodity and external price shocks contributed to the recent surge in inflation, maintaining adequately tight macroeconomic policies will be important as those shocks moderate.

### Policy implications (from SVAR results and broader analysis)
- Maintain adequately tight macroeconomic policies as external commodity price shocks moderate to help reduce inflation.
- Domestic macroeconomic stabilization policies play an important role in anchoring inflation; increasing credibility of these policies should be prioritized to reduce inflation persistence and improve policy effectiveness.

*Source: 5.   Structural VAR Analysis, wpiea2023049-print-pdf.*

### Appendix I. Stylized Facts Charts and Tables

### Appendix I. Stylized Facts Charts and Tables

### Contributions to headline inflation (components and trends)
- Figures present contributions to headline CPI inflation (y/y change) by component: Processed food, Unprocessed food, Energy, Core PCE, and Headline for Albania, Bosnia and Herzegovina, Kosovo, Montenegro, North Macedonia, and Serbia (series run from Jan-19 to Nov-22).
- Sources: Eurostat, WB central banks, Haver Analytics, and IMF staff calculations.

### CPI basket composition (average shares)
- Overall (average level between 2008M1 and 2022M11):  
  - Processed food: 23.8  
  - Unprocessed food: 18.3  
  - Energy: 8.7  
  - Core PCE: 49.1
- Albania (average between 2008M1 and 2022M11):  
  - Processed Food: 23.4  
  - Unprocessed Food: 13.4  
  - Energy: 21.4  
  - Core PCE: 41.7
- Bosnia and Herzegovina (average between 2016M1 and 2022M11):  
  - Processed food: 30.3  
  - Unprocessed: 16.1  
  - Energy: 13.2  
  - Core PCE: 40.4
- Kosovo (average between 2007M1 and 2022M11):  
  - Processed Food: 24.2  
  - Unprocessed Food: 17.2  
  - Energy: 17.7  
  - Core PCE: 40.9
- Montenegro (average between 2011M1 and 2022M11):  
  - Processed Food: 27.9  
  - Unprocessed Food: 16.6  
  - Energy: 14.9  
  - Core PCE: 40.6
- North Macedonia (average between 2010M1 and 2022M11):  
  - Processed Food: 25.1  
  - Unprocessed Food: 15.4  
  - Energy: 18.3  
  - Core PCE: 41.2
- Serbia (average between 2007M1 and 2022M11): (values printed but not repeated separately in the source figure text)

Sources: Eurostat, WB central banks, Haver Analytics, and IMF staff calculations.

### Food and energy weights; share of basic commodities
- Table A1.1. Share of Basic Commodities in CPI, 2022 (shares in CPI basket)
  - Cereal: ALB 2.0; BIH 2.2; MNE 1.7; EU 0.4
  - Wheat: ALB 1.1; BIH 1.6; MNE 1.2; EU 0.1
  - Vegetable oil: ALB 2.1; BIH 1.6; MNE 1.5; EU 0.3
  - Overall: ALB 4.0; BIH 3.8; MNE 3.2; EU 0.7
- Table A1.1. Share of Basic Commodities in CPI, 2022 (shares in “Food and Beverage” category)
  - Cereal: ALB 4.9; BIH 5.8; MNE 4.4; EU 1.6
  - Wheat: ALB 2.7; BIH 4.1; MNE 3.1; EU 0.5
  - Vegetable oil: ALB 5.1; BIH 4.2; MNE 3.7; EU 1.5
  - Overall: ALB 10.0; BIH 9.9; MNE 8.2; EU 3.2

Sources: Haver Analytics, WB central banks, and IMF staff calculations

### Public spending on agriculture (average for recent years)
- Table A1.2. Public Spending on Agriculture (In percent of Agricultural GDP):  
  - ALB 3.0  
  - BIH 8.0  
  - KOS 9.3  
  - MKD 16.2  
  - MNE 8.1  
  - SRB 12.5  
  - EU 20.5

Sources: IMF FAD, Haver Analytics, WB central banks, and IMF staff calculations

### Inflation heat maps (item-level distribution across inflation brackets)
- Figures show CPI share of items in different inflation brackets across time (Jan-19 to Oct-22) for each country with brackets: <-1.5; [-1.5, -0.5); [-0.5, 0); [0, 0.5); [0.5, 1); [1, 1.5); [1.5, 2); [2, 2.5); [2.5, 5); [5, 10); [10, 20); >=20.  
- Sources: Haver Analytics, WB central banks and IMF staff calculations.

### Core inflation heat maps (item-level distribution within core category)
- Figures mirror the inflation heat maps but restricted to core category items over Jan-19 to Oct-22.  
- Sources: Haver Analytics, WB central banks and IMF staff calculations.

### Different measures of underlying inflation (time series comparison)
- Figure A1.6 compares Headline, Core PCE (paper's definition), Trimmed mean 10%, Trimmed mean 30% from Jan-17 to Nov-22.
- Sources: Haver Analytics, WB central banks, and IMF staff calculations.

### Inflation and wage growth (headline inflation vs. wages and minimum wage)
- Figure A1.7 shows Headline and Average wage (private) and Minimum wage (RHS) time series (quarterly) from Mar-17 to Sep-22 for Albania, Bosnia and Herzegovina, Kosovo, Montenegro, North Macedonia, and Serbia.

### Headline vs. inflation expectations
- Figure A1.8 compares Headline inflation with 12-month ahead inflation expectations (Consensus Forecast, BoA, NBS where available) from Jan-17 to Nov-22. Note: inflation expectation shows 12-month ahead situation.  
- Sources: Haver Analytics, WB central banks, Consensus Forecast Database, and IMF staff calculations.

### Appendix II — Data definitions and sources (selected variables and sources)
- Headline inflation: Annualized quarter-over-quarter, seasonally adjusted (panel) and year-on-year (SVAR) inflation, all CPI items, 3-month average — Haver Analytics, WB central banks (Panel; SVAR).
- Core inflation: Annualized q/q, seasonally adjusted inflation, non-food and non-energy CPI items, 3-month average — Haver Analytics, WB central banks (Panel).
- Trimmed-mean 10% and 30%: Annualized q/q, seasonally adjusted inflation excluding top/bottom 10% (or 30%) yoy CPI items, 3-month average — Haver Analytics, WB central banks (Panel).
- Processed food inflation: Annualized q/q, seasonally adjusted inflation, processed food CPI items, 3-month average — Haver Analytics, WB central banks (Panel).
- Unprocessed food inflation: Annualized q/q, seasonally adjusted inflation, unprocessed food CPI items, 3-month average — Haver Analytics, WB central banks (Panel).
- Energy inflation: Annualized q/q, seasonally adjusted inflation, energy-related CPI items — Haver Analytics, WB central banks (Panel).
- Inflation expectations: One year ahead fixed horizon expectations on headline inflation — Consensus Forecast (Panel).
- Output gap: Deviation from the Hodrick-Prescott filtered, seasonally adjusted GDP, in percent — World Economic Outlook, IMF staff estimates (Panel).
- NEER: Annualized q/q, seasonally adjusted growth rate of Nominal Effective Exchange Rate, Trade Partners by CPI — IMF STA EER database (Panel).
- International food price: Annualized q/q growth rate of Commodity Food and Beverage Price Index — Global Assumptions database (Panel).
- International energy price: Annualized q/q growth rate of Commodity Fuel (energy) Index — Global Assumptions database (Panel).
- External Price Pressure Index: Constructed following Chapter 2 of the October 2021 World Economic Outlook and Chapter 2 of the October 2022 Regional Economic Outlook for EUR — IMF staff calculations (Panel; SVAR).
- Euro Area inflation: Year-on-year growth CPI inflation, 3-month average — Haver (SVAR).
- GDP growth: Year-on-year growth, quarterly data — Haver (SVAR).
- Monetary policy rate: Country-specific measures (Albania Overnight Credit Rate; North Macedonia 1-Month Interbank Offer Rate; Serbia Belgrade Overnight Index Average) — Haver (SVAR).
- Government spending: Central or General government expenditures as a share of GDP, quarterly — Haver, IMF staff estimates (SVAR).

### Appendix III — Descriptive statistics (Table A3.1)
- Headline inflation: Obs. 370; Mean 3.4; Stand. Dev. 5.2; Min -7.7; Max 28.6
- Core inflation: Obs. 290; Mean 3.4; Stand. Dev. 5.8; Min -8.2; Max 40.8
- Trimmed-mean 10%: Obs. 308; Mean 0.7; Stand. Dev. 0.9; Min -1.0; Max 4.9
- Trimmed-mean 30%: Obs. 308; Mean 0.2; Stand. Dev. 0.5; Min -0.4; Max 2.7
- Processed food inflation: Obs. 323; Mean 5.5; Stand. Dev. 10.2; Min -8.3; Max 74.0
- Unprocessed food inflation: Obs. 323; Mean 11.5; Stand. Dev. 22.0; Min -27.8; Max 144.9
- Energy inflation: Obs. 327; Mean 5.5; Stand. Dev. 14.6; Min -33.3; Max 97.3
- Inflation expectation: Obs. 224; Mean 2.3; Stand. Dev. 2.4; Min -1.0; Max 10.7
- Output gap: Obs. 354; Mean 0.0; Stand. Dev. 2.9; Min -19.6; Max 10.4
- NEER: Obs. 371; Mean -0.6; Stand. Dev. 5.1; Min -17.8; Max 32.1
- International food price: Obs. 371; Mean 5.6; Stand. Dev. 21.2; Min -47.8; Max 60.2
- International energy price: Obs. 371; Mean 6.5; Stand. Dev. 54.8; Min -178.3; Max 109.8

### Robustness and regression findings (selected results from Tables A3.2 and A3.3)
- Using the Unemployment Gap instead of Output Gap (Table A3.2): key coefficients (robust standard errors in parentheses; significance stars preserved)
  - Lagged dependent variable: 0.648*** (0.054) in column (1); 0.664*** (0.041) in column (2).
  - 1y inflation expectation (t-1): 0.031 (0.099) in column (1); 0.515*** (0.090) in column (2).
  - Unemployment gap: 0.119 (0.116) in column (1); 0.088 (0.093) in column (2).
  - NEER: 0.074** (0.033) in column (1); 0.069** (0.029) in column (2).
  - International food price inflation: 0.022** (0.009) in column (1); 0.021*** (0.008) in column (2).
  - International energy price inflation: 0.011*** (0.003) in column (1); 0.013*** (0.003) in column (2).
  - Observations: 172 (col 1); 252 (col 2). R-squared: 0.654 (col 1); 0.655 (col 2). Number of countries: 4 (each specification).
  - Note: Columns correspond to different dependent variables (Headline inflation, Core PCE, Trimmed-mean 10%, Trimmed-mean 30% as labeled in the table).
- Using FE OLS regressions instead of robust regression (Table A3.3): reported coefficients mirror Table A3.2 for key variables and significance levels (same numerical entries and standard errors reported).

- Robust standard errors shown in parentheses. Significance: *** p<0.01, ** p<0.05, * p<0.1.

*Sources: IMF Working Paper — Appendix I, II, III (figures and tables as provided). *

### Appendix Table A3.4. Using Measure of External Price Pressure Instead of NEER

### Appendix Table A3.4. Using Measure of External Price Pressure Instead of NEER

### Regression specification and dependent variables
- Table presents eight regressions labeled (1) through (8).
- Column groups are associated with the following labels as they appear in the source:
  - Headline inflation
  - Core PCE
  - Headline inflation
  - Trimmed-mean 10%
  - Headline inflation
  - Trimmed-mean 30%
- Common regressors reported:
  - Lagged dependant variable
  - 1y inflation expectation (t-1)
  - Output gap
  - External Price Pressure
  - Int.l' food price inflation
  - Int.l' energy price inflation
  - Constant

### Key coefficient estimates (by column)
- Column (1)
  - Lagged dependant variable: 0.720*** (0.067)
  - 1y inflation expectation (t-1): 0.094 (0.114)
  - Output gap: 0.044 (0.096)
  - External Price Pressure: 0.044 (0.030)
  - Int.l' food price inflation: 0.031*** (0.012)
  - Int.l' energy price inflation: 0.007* (0.004)
  - Constant: 0.501 (0.423)
  - Observations: 135
  - R-squared: 0.673
- Column (2)
  - Lagged dependant variable: 0.645*** (0.042)
  - 1y inflation expectation (t-1): 0.528*** (0.100)
  - Output gap: -0.016 (0.049)
  - External Price Pressure: 0.022 (0.023)
  - Int.l' food price inflation: 0.026*** (0.008)
  - Int.l' energy price inflation: 0.013*** (0.003)
  - Constant: 0.671* (0.400)
  - Observations: 236
  - R-squared: 0.662
- Column (3)
  - Lagged dependant variable: 0.233*** (0.084)
  - 1y inflation expectation (t-1): 0.007 (0.024)
  - Output gap: -0.018 (0.083)
  - External Price Pressure: 0.017 (0.025)
  - Int.l' food price inflation: 0.020* (0.010)
  - Int.l' energy price inflation: 0.004 (0.004)
  - Constant: 1.077*** (0.372)
  - Observations: 134
  - R-squared: 0.474
- Column (4)
  - Lagged dependant variable: 0.496*** (0.049)
  - 1y inflation expectation (t-1): 0.030*** (0.008)
  - Output gap: -0.003 (0.061)
  - External Price Pressure: 0.052* (0.029)
  - Int.l' food price inflation: 0.010 (0.011)
  - Int.l' energy price inflation: 0.010*** (0.004)
  - Constant: 1.021** (0.489)
  - Observations: 213
  - R-squared: 0.465
- Column (5)
  - Lagged dependant variable: 0.786*** (0.070)
  - 1y inflation expectation (t-1): (not reported)
  - Output gap: 0.007 (0.019)
  - External Price Pressure: 0.008 (0.006)
  - Int.l' food price inflation: 0.010*** (0.002)
  - Int.l' energy price inflation: 0.000 (0.001)
  - Constant: 0.111 (0.085)
  - Observations: 135
  - R-squared: 0.702
- Column (6)
  - Lagged dependant variable: 0.662*** (0.045)
  - 1y inflation expectation (t-1): (not reported)
  - Output gap: -0.004 (0.009)
  - External Price Pressure: 0.004 (0.004)
  - Int.l' food price inflation: 0.006*** (0.002)
  - Int.l' energy price inflation: 0.001* (0.001)
  - Constant: 0.157** (0.076)
  - Observations: 229
  - R-squared: 0.633
- Column (7)
  - Lagged dependant variable: 0.194*** (0.040)
  - 1y inflation expectation (t-1): (not reported)
  - Output gap: -0.006 (0.006)
  - External Price Pressure: 0.003 (0.002)
  - Int.l' food price inflation: 0.002** (0.001)
  - Int.l' energy price inflation: 0.001** (0.000)
  - Constant: 0.107*** (0.027)
  - Observations: 135
  - R-squared: 0.567
- Column (8)
  - Lagged dependant variable: 0.686*** (0.032)
  - 1y inflation expectation (t-1): 0.030*** (0.008) [appears in column (4) labeling; source shows a 0.030*** under column grouping]
  - Output gap: 0.000 (0.004)
  - External Price Pressure: 0.002 (0.002)
  - Int.l' food price inflation: 0.001* (0.001)
  - Int.l' energy price inflation: 0.000** (0.000)
  - Constant: 0.069** (0.029)
  - Observations: 229
  - R-squared: 0.778

### Inferential and sample details
- Number of countries: 46 (reported for all columns)
- Country fixed effects: Yes
- Time fixed effects: No
- Missing countries reported across specifications: KOS; MNEKOS; MNEKOS; MNEKOS; MNE (as listed)
- Robust standard errors reported in parentheses
- Significance legend:
  - *** p<0.01
  - ** p<0.05
  - * p<0.1

### Notable patterns visible in the estimates
- Lagged dependent variable coefficients are positive and statistically significant in all columns reported (values range from 0.194*** to 0.786***).
- International food price inflation shows positive and often statistically significant coefficients across specifications (examples: 0.031***, 0.026***, 0.020*, 0.010***, 0.006***, 0.002**).
- International energy price inflation coefficients are smaller but reach statistical significance in several specifications (examples: 0.007*, 0.013***, 0.010***, 0.001*, 0.001**).
- External Price Pressure coefficients are generally small and imprecisely estimated in these specifications (examples: 0.044, 0.022, 0.017, 0.052*, 0.008, 0.004, 0.003, 0.002), with one coefficient reaching * p<0.1 in the table.

*Appendix Table A3.4, as presented in the source PDF.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023049-print-pdf.pdf_