Abstract
Inflation in Estonia and the Baltics surged over 2021–22 and has been well above euro area average, against the backdrop of a sharp rise in global commodity prices. This paper conducts an empirical analysis of the inflation drivers in the three Baltic countries compared to the euro area. It finds that the passthrough to inflation from global commodity prices in Estonia has been higher than in the other Baltics and the euro area. While Estonia’s inflation has so far been largely driven by external factors, domestic factors such as wage growth also appear to be statistically significant drivers of prices of food and several services components of the CPI. The large size of the inflation surge calls for broad-based policy response to prevent an entrenching of high inflation and the associated economic consequences.
Recent Drivers of Inflation in Estonia: A Comparative Perspective with the Baltics and Euro Area1
Inflation in Estonia and the Baltics surged over 2021–22 and has been well above euro area average, against the backdrop of a sharp rise in global commodity prices. This paper conducts an empirical analysis of the inflation drivers in the three Baltic countries compared to the euro area. It finds that the passthrough to inflation from global commodity prices in Estonia has been higher than in the other Baltics and the euro area. While Estonia’s inflation has so far been largely driven by external factors, domestic factors such as wage growth also appear to be statistically significant drivers of prices of food and several services components of the CPI. The large size of the inflation surge calls for broad-based policy response to prevent an entrenching of high inflation and the associated economic consequences.
A. Introduction
1. The surge in inflation in Estonia and the other Baltics has been large both in absolute terms and relative to the euro area average. Year-on-year harmonized consumer price inflation reached 22 percent in Estonia through June 2022, with Lithuania (20.5 percent) and Latvia (19 percent) being in a similar range (Figure 1). While the same measure of inflation in the euro area has also been quite elevated at 8.6 percent relative to the 2 percent ECB target, it has been substantially lower than in the Baltics. Estonia’s inflation perceptions and expectations have also been on the rise. The high level and the wide gap of inflation in the Baltics versus the euro area highlights challenges of the policymakers’ responses/toolkits against the backdrop of euro area’s common monetary policy. Moreover, much of the recent surge in the consumer prices has been unanticipated, with inflation surprises disrupting expectations-setting processes and complicating macroeconomic analysis, forecasting, and overall policy planning.
Inflation and Inflation Expectations in the Baltics and the Euro Area 2021–22
Citation: IMF Staff Country Reports 2022, 290; 10.5089/9798400220357.002.A001
Sources: Eurostat, ECB, and IMF staff estimates.Inflation and Inflation Expectations in the Baltics and the Euro Area 2021–22
Citation: IMF Staff Country Reports 2022, 290; 10.5089/9798400220357.002.A001
Sources: Eurostat, ECB, and IMF staff estimates.Inflation and Inflation Expectations in the Baltics and the Euro Area 2021–22
Citation: IMF Staff Country Reports 2022, 290; 10.5089/9798400220357.002.A001
Sources: Eurostat, ECB, and IMF staff estimates.2. A key element in the recent rise in inflation has been the concurrent upsurge in global commodity prices. The fuel price composite index increased, in US dollar terms, by almost 5 times over a 2-year period in 2022:Q2 (relative to the bottom reached in 2020:Q2 during the pandemic) and by 2½ times relative to the pre-pandemic level of 2019:Q4 (Figure 2). In parallel, international food prices increased by over 60 percent relative to pre-pandemic levels. Other global commodity prices, including for metals, cotton, and fertilizers, have also exhibited strong growth during this period. The global prices are likely to be reflected in the consumer prices as the affected products are direct inputs to the specific components of the consumer basket, as well as through spillovers to other prices. The war in Ukraine has represented a further upward shock to commodity prices, particularly for natural gas, whose prices increased 10 times in 2022:Q2 relative to 2020:Q2.
Prices for Global Commodities and Products, 2020–22Q2
Citation: IMF Staff Country Reports 2022, 290; 10.5089/9798400220357.002.A001
Source: IMF staff estimates.Prices for Global Commodities and Products, 2020–22Q2
Citation: IMF Staff Country Reports 2022, 290; 10.5089/9798400220357.002.A001
Source: IMF staff estimates.Prices for Global Commodities and Products, 2020–22Q2
Citation: IMF Staff Country Reports 2022, 290; 10.5089/9798400220357.002.A001
Source: IMF staff estimates.3. Estonia’s characteristics and past empirical estimates suggest that its inflation passthrough from global commodity prices could be relatively high. Small open economies have more limited possibilities to reallocate resources, mitigate external shocks, and substitute imported inputs. While the general relationship between trade openness and the commodity price passthrough is mixed (see Neely and Rapach (2011) and Gelos and Ustyugova (2012)), openness alone is an imperfect control for small size and other characteristics of Estonia.2 In this context, the literature suggests that economies with higher food shares in CPI baskets, fuel intensities, and pre-existing inflation levels (all of which are Estonia’s characteristics) are more prone to experience inflationary effects from commodity price shocks (Gelos and Ustyugova (2012)). Rigobon (2010) confirms that Estonia’s commodity price passthrough was significantly higher in the 1990s and 2000s than that for larger euro area economies across sectors computed by the author (e.g., oil, natural gas, and wheat price passthrough). However, Estonia’s economic structure and conditions have significantly changed since the estimates were made.
4. The passthrough from global commodity prices needs to be assessed in combination with other inflation drivers. The unprecedented size of the surge of both commodity prices and general inflation calls for a more holistic evaluation of external and domestic factors. In particular, there have been several other factors with substantial potential for inflationary effects, including (i) supply chain disruptions; (ii) large policy support packages that were implemented to offset the effects of the COVID-19 shock; (iii) the emergence of labor shortages during the COVID-19 crisis and (iv) sustained expansionary monetary policy stances in advanced economies. It would be desirable to assess these factors in tandem with the passthrough from global prices with a view to gauging the potential for the emergence of sustained inflationary pressures. A better understanding of the current episode of high inflation would also help throw light on its competitiveness and distributional effects, which are important in terms of policy implications.
5. This paper studies Estonia’s inflation drivers in comparison to the other Baltics and the euro area, with a particular focus on the role of global commodity prices. Section B presents several global and regional stylized facts regarding Estonia’s inflation and its recent surge in a comparative perspective. Section C computes and compares the differences in the global commodity price passthroughs for Estonia, other Baltics, and the euro area average. Section D estimates a more holistic model of Estonia’s inflation that combines external and domestic drivers. Section E further assesses the role of some domestic drivers. Section F discusses policy implications and concludes.
B. Some Comparative Stylized Facts on Estonia’s Inflation
6. In 2011-20, inflation in Estonia was well-anchored overall, although it was moderately higher than in the euro area. For the first decade after its accession to the euro area, Estonia’s inflation averaged less than 0.2 percent per month and 2¼ percent annually (see Figure 3, left panel). While this level was the highest among the Baltic countries and higher than that in the euro area (by about 1 percentage point on an annual basis), while also exceeding the 2 percent ECB target, it was within the margin that was considered reasonable given the still-substantial room for convergence with the euro area’s price level. Estonia’s core inflation (1.6 percent annual rate) was lower than the headline inflation for the same period, being higher than in the euro area and Latvia but lower than in Lithuania.3
Inflation in the Baltics and the Euro Area, 2011–2022Q2
Citation: IMF Staff Country Reports 2022, 290; 10.5089/9798400220357.002.A001
Source: Eurostat and IMF staff estimates.Inflation in the Baltics and the Euro Area, 2011–2022Q2
Citation: IMF Staff Country Reports 2022, 290; 10.5089/9798400220357.002.A001
Source: Eurostat and IMF staff estimates.Inflation in the Baltics and the Euro Area, 2011–2022Q2
Citation: IMF Staff Country Reports 2022, 290; 10.5089/9798400220357.002.A001
Source: Eurostat and IMF staff estimates.7. Over 2021 and the first half of 2022, Estonia’s average monthly inflation rate increased 8-fold relative to the previous 10-year average. Between January 2021 and June 2022, monthly inflation averaged 1.3 percent, which was slightly higher than in Lithuania and Latvia (1.2 percent each) and more than double the euro area average, which also increased perceptibly (to around 0.6 percent monthly). Moreover, inflation in the Baltics and the euro area was on a pronounced accelerating trend during this period. During this period, Estonia’s inflation was more variable than that of the euro area and the other Baltic countries, as reflected in a higher standard deviation of the monthly inflation rate (Figure 3, right panel).
8. Energy prices increased sharply and accounted for a large share of the inflation surges, with Estonia seeing the highest rate of energy price growth. The energy component of Estonia’s HICP increased by 97 percent between December 2020 and May 2022, which was four times the pace of headline inflation for the same period.4 The same components for Latvia (65 percent), Lithuania (72 percent), and the euro area (52 percent) also experienced large increases, though they were not as high as in Estonia. Within the energy price component of the HICP, Estonia’s increases in electricity tariffs have been a major contributor to the difference with other comparators. After adjusting for the weight of the energy in the consumer basket, this component accounted for over one-half (57 percent) of the overall inflation in Estonia over this period (Figure 4). Such direct contribution has also been significant (if somewhat lower) for Latvia and the euro area (about 50 percent) and Lithuania (about 40 percent). All these contributions were well above the weight of energy in the HICP, which ranged from 10 to 15 percent for these countries.
Energy Inflation in the Baltics and the Euro Area, 2021–2022
Citation: IMF Staff Country Reports 2022, 290; 10.5089/9798400220357.002.A001
Source: Eurostat and IMF staff estimates.Energy Inflation in the Baltics and the Euro Area, 2021–2022
Citation: IMF Staff Country Reports 2022, 290; 10.5089/9798400220357.002.A001
Source: Eurostat and IMF staff estimates.Energy Inflation in the Baltics and the Euro Area, 2021–2022
Citation: IMF Staff Country Reports 2022, 290; 10.5089/9798400220357.002.A001
Source: Eurostat and IMF staff estimates.9. Food price inflation has been a more modest but growing contributor to the overall inflation, with direct contribution being larger than in the euro area but smaller than in the other Baltic countries. Food price inflation in Estonia accelerated to 17 percent y/y in May 2022 (Figure 5). This was double the euro area average but smaller than the growth of food prices in Latvia (19 percent) and Lithuania (25 percent). Taking account of the weight of the food price component in the HICP, they directly accounted for about 17 percent of Estonia’s inflation, compared to around 25 percent for Latvia and Lithuania. For now, the rate of growth of food prices has been similar to that of overall inflation, slightly below this level in Estonia and above it in Lithuania. However, food price growth has been appreciably accelerating for all countries in the first few months of 2022.
Food Price Inflation in the Baltics and the Euro Area, 2021–2022
Citation: IMF Staff Country Reports 2022, 290; 10.5089/9798400220357.002.A001
Source: Eurostat and IMF staff estimates.Food Price Inflation in the Baltics and the Euro Area, 2021–2022
Citation: IMF Staff Country Reports 2022, 290; 10.5089/9798400220357.002.A001
Source: Eurostat and IMF staff estimates.Food Price Inflation in the Baltics and the Euro Area, 2021–2022
Citation: IMF Staff Country Reports 2022, 290; 10.5089/9798400220357.002.A001
Source: Eurostat and IMF staff estimates.10. Inflation has been becoming increasingly broad-based both in the Baltics and the euro area. Core inflation in Estonia (Figure 6) increased more than four-fold during the recent surge: from 0.14 percent per month in 2011–20 to 0.6 percent per month in 2021–22 (through May). This increase was close to the rates of growth observed in the Baltic peers and higher than on average in the euro area. Lithuania’s core inflation was the highest among the Baltics, reaching 0.7 percent per month during the recent surge. Euro area’s core inflation has been more contained, being at around one-half of Estonia’s rate during the inflation surge and increasing slightly less (by about 3½ times relative to the 2011-20 clip). Core inflation continued to experience a further acceleration during the first five months of 2022: the average monthly pace reached 0.9 percent in Estonia, being below the 1.1 percent observed both in Latvia and Lithuania but well above the 0.4 percent observed in the euro area.
Core Inflation in the Baltics and the Euro Area, 2021–2022
Citation: IMF Staff Country Reports 2022, 290; 10.5089/9798400220357.002.A001
Source: Eurostat and IMF staff estimates.Core Inflation in the Baltics and the Euro Area, 2021–2022
Citation: IMF Staff Country Reports 2022, 290; 10.5089/9798400220357.002.A001
Source: Eurostat and IMF staff estimates.Core Inflation in the Baltics and the Euro Area, 2021–2022
Citation: IMF Staff Country Reports 2022, 290; 10.5089/9798400220357.002.A001
Source: Eurostat and IMF staff estimates.C. Analyzing the Global Commodity Price Passthrough
11. A simple regression analysis was performed to assess the relative magnitudes of the global commodity pass-through effect on HICP inflation in the Baltics and the euro area. Country-specific time series OLS regressions were run on monthly data from November 2010 through March 2022, aiming to size up characteristics of the passthrough during a period when Estonia and the other Baltic countries were effectively anchored by euro area membership. Regarding the key variables, the domestic HICPs (or their components) were regressed on the following explanatory variables, all expressed in euros: (i) Brent oil prices; (ii) natural gas prices; and (iii) the global food price index.5 Parsimonious specifications were obtained using general-to-specific modeling, starting with nine lags for all explanatory variables, sequentially eliminating statistically insignificant variables at the standard pre-set criteria. Seasonal dummies, and where relevant, time dummies were used as additional controls. The approach of relying on global explanatory variables as dominant drivers in such regressions includes the advantage of high cross-country comparability at a time of large effects of external shocks. However, this approach has several problems, including (i) potential omitted variable bias for other inflation drivers; (ii) usual reverse causality problems; (iii) overfitting concerns, and (iv) difficulty in disentangling the passthrough effects from concurrent demand and supply shocks. That said, there are several mitigants to these problems.6
12. The results suggest that Estonia’s global commodity passthrough to the headline inflation has been higher than in the comparators, essentially reflecting passthrough from food prices. Table 1 presents the results of the regressions, with the coefficients for all three commodity prices being consistently positive and statistically significant. Overall, the fit of the models is relatively good with an adjusted R-squared hovering at around 0.6-0.65. The combined passthrough from changes in oil and gas prices is estimated to be comparable between Estonia and the other Baltics, but it is generally higher for the Baltics than for the euro area.7 Significantly, Estonia has the highest estimated passthrough from global food prices to general consumer prices, which is consistent with the fact that, over the last decade Estonia’s food price inflation was higher than that in the Baltic countries and the euro area. Recursive analysis of the regression results suggests that the passthrough may not be fully symmetric, with its absolute size appearing to be smaller during episodes when the commodity prices are falling.8
Estonia: Comparative Passthrough from Global Prices to Headline HICP 1/ (2010–22 (March)), monthly data
Results based on OLS regressions of HICP on global commodity price benchmarks (in Euros).
In months, weighted average based on coefficients.
Sums of statistically significant coefficients at 5 percent.
Estonia: Comparative Passthrough from Global Prices to Headline HICP 1/ (2010–22 (March)), monthly data
| Oil | Natural gas | Food | Total | ||||
|---|---|---|---|---|---|---|---|
| Speed 2/ | Size 3/ | Speed 2/ | Size 3/ | Speed 2/ | Size 3/ | Size 3/ | |
| Estonia | 3.0 | 0.018 | 8.0 | 0.008 | 3.9 | 0.054 | 0.080 |
| Latvia | 0.9 | 0.026 | 9.0 | 0.006 | 4.0 | 0.025 | 0.057 |
| Lithuania | 0.3 | 0.022 | 5.0 | 0.004 | 5.0 | 0.026 | 0.052 |
| Euro-area | 0 | 0.014 | 5.0 | 0.005 | 7.0 | 0.019 | 0.038 |
Results based on OLS regressions of HICP on global commodity price benchmarks (in Euros).
In months, weighted average based on coefficients.
Sums of statistically significant coefficients at 5 percent.
Estonia: Comparative Passthrough from Global Prices to Headline HICP 1/ (2010–22 (March)), monthly data
| Oil | Natural gas | Food | Total | ||||
|---|---|---|---|---|---|---|---|
| Speed 2/ | Size 3/ | Speed 2/ | Size 3/ | Speed 2/ | Size 3/ | Size 3/ | |
| Estonia | 3.0 | 0.018 | 8.0 | 0.008 | 3.9 | 0.054 | 0.080 |
| Latvia | 0.9 | 0.026 | 9.0 | 0.006 | 4.0 | 0.025 | 0.057 |
| Lithuania | 0.3 | 0.022 | 5.0 | 0.004 | 5.0 | 0.026 | 0.052 |
| Euro-area | 0 | 0.014 | 5.0 | 0.005 | 7.0 | 0.019 | 0.038 |
Results based on OLS regressions of HICP on global commodity price benchmarks (in Euros).
In months, weighted average based on coefficients.
Sums of statistically significant coefficients at 5 percent.
13. Estonia’s passthrough from global to domestic energy prices is broadly in line with that of the other Baltics. Regressing the respective domestic energy components of the HICP on global energy prices puts Estonia in the middle-of-the-pack among the Baltics in terms of the passthrough within the energy sector (Table 2). Interestingly, such energy passthrough is relatively high for the euro area, which would still be consistent with the smaller passthrough to broader HICP given the lower weight of energy in the euro area (around 10 percent) relative to the Baltics. By contrast, Estonia’s energy component share in the HICP is relatively high, having risen substantially in 2022 to almost 16 percent, thereby entailing a larger impact of the same increase in domestic energy prices on broader inflation.9
Estonia: Comparative Passthrough from Global Prices to Energy HICP 1/
Based on OLS regressions of energy HICP on global energy price benchmarks.
In months, weighted average based on coefficients.
Sums of statistically significant coefficients at 5 percent.
Estonia: Comparative Passthrough from Global Prices to Energy HICP 1/
| Oil | Natural gas | Total | |||
|---|---|---|---|---|---|
| Speed 2/ | Size 3/ | Speed 2/ | Size 3/ | Size 3/ | |
| Estonia | 3.1 | 0.139 | 1.6 | 0.075 | 0.213 |
| Latvia | 1.9 | 0.158 | 2.2 | 0.028 | 0.186 |
| Lithuania | 0.7 | 0.160 | 2.6 | 0.088 | 0.248 |
| Euro-area | 1.2 | 0.214 | 2.0 | 0.017 | 0.231 |
Based on OLS regressions of energy HICP on global energy price benchmarks.
In months, weighted average based on coefficients.
Sums of statistically significant coefficients at 5 percent.
Estonia: Comparative Passthrough from Global Prices to Energy HICP 1/
| Oil | Natural gas | Total | |||
|---|---|---|---|---|---|
| Speed 2/ | Size 3/ | Speed 2/ | Size 3/ | Size 3/ | |
| Estonia | 3.1 | 0.139 | 1.6 | 0.075 | 0.213 |
| Latvia | 1.9 | 0.158 | 2.2 | 0.028 | 0.186 |
| Lithuania | 0.7 | 0.160 | 2.6 | 0.088 | 0.248 |
| Euro-area | 1.2 | 0.214 | 2.0 | 0.017 | 0.231 |
Based on OLS regressions of energy HICP on global energy price benchmarks.
In months, weighted average based on coefficients.
Sums of statistically significant coefficients at 5 percent.
14. Estonia’s estimated higher passthrough to overall inflation from global food prices is confirmed by the specific investigation of the domestic food price component. Regressing domestic food price inflation on global commodity prices results in Estonia’s coefficients being significantly higher than for other comparators (Table 3). Interestingly, for all comparators the increase in global natural gas prices has a statistically significant association with the rise in domestic food prices, thereby suggesting a spill-over effect between natural gas and food prices, which however seems to be quantitatively small.
Estonia: Comparative Passthrough from Global Prices to Food HICP 1/
(2010-22 (March)), monthly data
Based on OLS regressions of food HICP on global commodity price benchmarks.
In months, weighted average based on coefficients.
Sums of statistically significant coefficients at 5 percent.
Estonia: Comparative Passthrough from Global Prices to Food HICP 1/
(2010-22 (March)), monthly data
| Food prices | Natural gas | Total | |||
|---|---|---|---|---|---|
| Speed 2/ | Size 3/ | Speed 2/ | Size 3/ | Size 3/ | |
| Estonia | 1.3 | 0.156 | 7 | 0.024 | 0.180 |
| Latvia | 0 | 0.052 | 8 | 0.017 | 0.069 |
| Lithuania | 3.4 | 0.089 | 7 | 0.014 | 0.103 |
| Euro-area | 1.4 | 0.051 | 7 | 0.006 | 0.057 |
Based on OLS regressions of food HICP on global commodity price benchmarks.
In months, weighted average based on coefficients.
Sums of statistically significant coefficients at 5 percent.
Estonia: Comparative Passthrough from Global Prices to Food HICP 1/
(2010-22 (March)), monthly data
| Food prices | Natural gas | Total | |||
|---|---|---|---|---|---|
| Speed 2/ | Size 3/ | Speed 2/ | Size 3/ | Size 3/ | |
| Estonia | 1.3 | 0.156 | 7 | 0.024 | 0.180 |
| Latvia | 0 | 0.052 | 8 | 0.017 | 0.069 |
| Lithuania | 3.4 | 0.089 | 7 | 0.014 | 0.103 |
| Euro-area | 1.4 | 0.051 | 7 | 0.006 | 0.057 |
Based on OLS regressions of food HICP on global commodity price benchmarks.
In months, weighted average based on coefficients.
Sums of statistically significant coefficients at 5 percent.
15. The estimated passthrough from global commodity prices to core inflation is moderately less significant for Estonia. Overall, the size of the effect is significantly smaller for core inflation than for headline inflation for all countries, with Estonia’s passthrough coefficient estimated to be less than half of that for the headline passthrough (Figure 7). In contrast to the results for headline inflation, Estonia’s estimated passthrough to core inflation is no longer the highest among the Baltics (e.g., is lower than Lithuania’s).
Passthrough from Global Prices to Headline and Core Inflation
Citation: IMF Staff Country Reports 2022, 290; 10.5089/9798400220357.002.A001
Source: IMF staff estimates.Passthrough from Global Prices to Headline and Core Inflation
Citation: IMF Staff Country Reports 2022, 290; 10.5089/9798400220357.002.A001
Source: IMF staff estimates.Passthrough from Global Prices to Headline and Core Inflation
Citation: IMF Staff Country Reports 2022, 290; 10.5089/9798400220357.002.A001
Source: IMF staff estimates.D. A More Holistic Analysis of Estonia’s Inflation
16. A mark-up model has been used to assess the drivers of Estonia’s inflation more comprehensively. As per De Brouwer and Ericsson (1995) and Ericsson (2009), this model provides a consistent theoretical framework for the inflation process by proxying prices as mark-ups over unit labor costs as well as imported and fuel prices. Such model is particularly appealing for an open economy as Estonia’s that relies on imported inputs. The implementation of the model entails a prior analysis of the order of integration of the time series and of possible co-integrating relationships. In the event of the latter, a comprehensive and relatively parsimonious model could be estimated whereby inflation reflects a combination of long-term drivers of the price level and short-term deviations from the path based on cyclical and short-term shock-related factors. This framework combines both external and domestic drivers while helping imbed the global commodity price passthrough within a consistent framework.10
17. The mark-up model was fitted to Estonia’s monthly and quarterly data. The use of high-frequency data is needed to focus on the relatively short period of interest (e.g., Estonia being anchored by membership in the euro area). Employing both quarterly and monthly data is a useful check on the robustness of the results. It also entails a trade-off of better power of the empirical tests of the monthly data and improved menu of variables for the quarterly data. For example, the data on unit labor costs are only available for Estonia in a quarterly format, so a proxy based on monthly wages had to be used instead in the monthly model. Similarly, the quality of the output gap data is better in a quarterly format as it can be constructed from GDP statistics, which are not available in a monthly format. As in the passthrough analysis of the previous section, a general-to-specific methodology was used, with seasonal and time dummies being additional controls.
18. The empirical analysis of inflation points to the high role of global commodity drivers and is broadly consistent with the passthrough analysis of the previous section. The key results of the inflation regressions are presented in Table 4. The cointegration analysis has permitted to establish long-term relationships between inflation, unit labor costs, and import prices. The cointegrating vector enters the short-term inflation equation as a lagged error correction term (ECM), which is statistically significant and negative as expected. The estimated coefficients of the commodity price passthrough are similar to those that were estimated in the passthrough analysis: the size of the combined passthrough from the global commodity prices is also estimated at 0.08 in the benchmark monthly HICP regression model. Alternative specifications put it in a similar range: 0.1 when domestic CPI is used instead of the HICP. The quarterly model has very similar coefficients for the passthrough from global energy prices but fails to detect a statistically significant passthrough for the food prices, likely highlighting the reduced power of the small quarterly sample.
Estonia: Regression Estimates from the Mark-up Model
WTI price is used instead of Brent price in the quarterly model.
Estonia: Regression Estimates from the Mark-up Model
| Dependent Variable: | HICP Monthly 2010(12)-2022(2) |
Dom CPI Monthly 2010(12)-2022(2) |
HICP 1/ Quarterly 2010(4)-2021(2) |
|---|---|---|---|
| ECM t-1 | -0.06*** | -0.05*** | -0.26*** |
| -3.57 | -3.25 | -7.04 | |
| ΔBrent t | 0.008*** | 0.010*** | 0.031*** |
| 3.17 | 3.87 | 7.51 | |
| ΔBrent t-1 | 0.009*** | 0.011*** | 0.009** |
| 3.45 | 4.14 | 2.16 | |
| ΔBrent t-6 | 0.006** | ||
| 2.39 | |||
| ΔBrent t-7 | 0.006** | ||
| 2.36 | |||
| ΔGas prices t-1 | 0.007** | 0.006** | 0.012*** |
| 2.59 | 2.41 | 3.74 | |
| ΔGas prices t-2 | |||
| ΔGas prices t-8 | 0.006** | ||
| 2.27 | |||
| ΔGlob. food prices t-7 | 0.029** | ||
| 2.54 | |||
| ΔGlob. food prices t-8 | 0.040*** | 0.038*** | |
| 3.31 | 3.29 | ||
| Constant | -0.000 | -0.000 | 0.002 |
| -0.20 | -0.20 | 1.67 | |
| output gap/slack | insignificant | insignificant | 0.021 |
| 0.83 | |||
| ΔRGSM Rate t | |||
| Number of observations | 135 | 135 | 43 |
| R-squared overall | 0.72 | 0.71 | 0.82 |
| R-squared adj. | 0.68 | 0.67 | 0.78 |
| Seasonal/time dummies | yes | yes | yes |
WTI price is used instead of Brent price in the quarterly model.
Estonia: Regression Estimates from the Mark-up Model
| Dependent Variable: | HICP Monthly 2010(12)-2022(2) |
Dom CPI Monthly 2010(12)-2022(2) |
HICP 1/ Quarterly 2010(4)-2021(2) |
|---|---|---|---|
| ECM t-1 | -0.06*** | -0.05*** | -0.26*** |
| -3.57 | -3.25 | -7.04 | |
| ΔBrent t | 0.008*** | 0.010*** | 0.031*** |
| 3.17 | 3.87 | 7.51 | |
| ΔBrent t-1 | 0.009*** | 0.011*** | 0.009** |
| 3.45 | 4.14 | 2.16 | |
| ΔBrent t-6 | 0.006** | ||
| 2.39 | |||
| ΔBrent t-7 | 0.006** | ||
| 2.36 | |||
| ΔGas prices t-1 | 0.007** | 0.006** | 0.012*** |
| 2.59 | 2.41 | 3.74 | |
| ΔGas prices t-2 | |||
| ΔGas prices t-8 | 0.006** | ||
| 2.27 | |||
| ΔGlob. food prices t-7 | 0.029** | ||
| 2.54 | |||
| ΔGlob. food prices t-8 | 0.040*** | 0.038*** | |
| 3.31 | 3.29 | ||
| Constant | -0.000 | -0.000 | 0.002 |
| -0.20 | -0.20 | 1.67 | |
| output gap/slack | insignificant | insignificant | 0.021 |
| 0.83 | |||
| ΔRGSM Rate t | |||
| Number of observations | 135 | 135 | 43 |
| R-squared overall | 0.72 | 0.71 | 0.82 |
| R-squared adj. | 0.68 | 0.67 | 0.78 |
| Seasonal/time dummies | yes | yes | yes |
WTI price is used instead of Brent price in the quarterly model.
19. While the analysis is not able to detect a significant role of short-term domestic factors in the inflation process, it at least points to the role of long-term domestic factors. Measures of the output gap, as well as other domestic variables that were attempted to be included (wage growth and fiscal and monetary policy variables) proved to be statistically insignificant in the above regressions for the headline HICP. On the one hand, this result highlights the nature of Estonia as a small open economy, where external factors would be particularly impactful. On the other hand, a significant influence of the domestic factors cannot be ruled out. For one, domestic factors enter through the long-term cointegrating relationship through unit labor costs, which in turn depend on the evolution of wages. The difficulty of establishing a short-term relationship also partly reflects data volatility and weaknesses in Estonia’s small economy superimposed on the intrinsic data challenges that are not Estonia-specific. The latter for example include uncertainty over: (i) real-time estimates of the output gap and (ii) appropriate metrics of monetary policy impact in a currency union.
E. Assessing the Role of Wages in Driving Estonia’s Inflation
20. A more granular analysis of inflation subcomponents could throw light on an important question of whether the recent spike in prices could morph into sustained inflationary pressure. In this regard, a key factor to consider is that of the wage price spiral. However, comprehensively assessing the potential for such a spiral would involve an analysis of the determinants of not only prices but wages. The latter goes beyond the scope of this paper, which is limited to the narrower investigation of the determinants of prices. Within this more limited scope, this section zooms in on the relationship between wages and different subcomponents of Estonia’s consumer price index.11
21. While the regression analysis of the preceding section does not to detect a significant relationship between Estonia’s wage growth and headline inflation, there is some, if mixed, indication of a positive relationship for core inflation.12 Regressing core inflation on lagged wage growth, changes in global commodity prices, and several other controls, reveals that the relationship between core inflation and wage growth is likely positive and could be statistically significant, at least in some samples. However, the results generated are not very robust to changes in samples and specifications and should therefore be treated with caution.
22. There is evidence of a positive relationship between wage growth and food price inflation in Estonia. Regressing the growth of domestic food prices on the lagged wage growth, global food price growth, and seasonal dummies yields a positive and highly significant coefficient on the lagged wage growth, both in the monthly and quarterly data samples (Table 5). This evidence is consistent with the literature findings in other countries, including India and the US, where wages or general income growth are associated with higher food price inflation (see Lee et al. (2000)) and Samal et al. (2022)).
Estonia: Relationship Between Food CPI Components and Wage Growth
Estonia: Relationship Between Food CPI Components and Wage Growth
| Dependent Variable: | Food HICP monthly | Food domestic CPI quarterly |
| lagged wage growth | 0.064*** | 0.31*** |
| 5.01 | 4.09 | |
| key controls | growth in global food prices | growth in global food prices |
| growth in global gas prices | ||
| other controls | seasonals & time dummies | seasonals |
| number of observations | 135 | 43 |
Estonia: Relationship Between Food CPI Components and Wage Growth
| Dependent Variable: | Food HICP monthly | Food domestic CPI quarterly |
| lagged wage growth | 0.064*** | 0.31*** |
| 5.01 | 4.09 | |
| key controls | growth in global food prices | growth in global food prices |
| growth in global gas prices | ||
| other controls | seasonals & time dummies | seasonals |
| number of observations | 135 | 43 |
23. Regressions also detect a relationship between lagged wage growth and some services subcomponents of Estonia’s domestic inflation. Regressing services components of Estonia’s consumer price index on lagged wage growth and other controls points to a positive relationship between wage and price increases for (i) transportation, (ii) communications, and (iii) “miscellaneous services” prices (Table 6). This evidence is consistent with the literature findings that rising wages could trigger increases in services prices relative to those of goods prices due to limited competition involved in their provision (see Brauer (1997)).
Estonia: Relationship Between Selected Services Prices and Wage Growth
Represents sum of statistically significant coefficients at 5 percent; t-statistics reported for the highest value.
Estonia: Relationship Between Selected Services Prices and Wage Growth
| Dependent Variable: rate of growth of a respective services component of the CPI | |||
| Sample: monthly 2010(11)–2022(3) | |||
| Services component: | transportation 1/ | communications | miscellaneous |
| lagged wage growth | 0.349*** | 0.030** | 0.024** |
| 3.11 | 2.67 | 2.55 | |
| other controls | global commodity prices, lagged dependent variable | ||
| Number of observations | 137 | 137 | 137 |
| R-squared overall | 0.73 | 0.56 | 0.35 |
| R-squared adj. | 0.69 | 0.48 | 0.32 |
| Seasonal/time dummies | yes | yes | yes |
Represents sum of statistically significant coefficients at 5 percent; t-statistics reported for the highest value.
Estonia: Relationship Between Selected Services Prices and Wage Growth
| Dependent Variable: rate of growth of a respective services component of the CPI | |||
| Sample: monthly 2010(11)–2022(3) | |||
| Services component: | transportation 1/ | communications | miscellaneous |
| lagged wage growth | 0.349*** | 0.030** | 0.024** |
| 3.11 | 2.67 | 2.55 | |
| other controls | global commodity prices, lagged dependent variable | ||
| Number of observations | 137 | 137 | 137 |
| R-squared overall | 0.73 | 0.56 | 0.35 |
| R-squared adj. | 0.69 | 0.48 | 0.32 |
| Seasonal/time dummies | yes | yes | yes |
Represents sum of statistically significant coefficients at 5 percent; t-statistics reported for the highest value.
F. Conclusions and Policy Recommendations
24. The empirical analysis of Estonia’s inflation points to a sizable role of the passthrough from global commodity prices, which has been a key source of variation in its inflation rate. Estonia’s estimated passthrough has been broadly comparable to—but generally higher than—that of its Baltic neighbors on account of the higher food price passthrough. Furthermore, such passthrough in the Baltic countries is higher than for the euro area. However, some of these numerical estimates of the passthrough effects need to be treated with caution as they are sensitive to sample selection and specification methods, against the backdrop of a relatively short sample sizes, rapid structural change, and data series limitations in the small economies of the Baltic region. Another reason for caution is a possibility of a large structural break in the economic and policy landscape that is marked by the recent upshift in the inflation levels in Estonia and more globally.
25. The role of domestic factors in Estonia’s recent inflation developments is more elusive to assess and could well be underestimated. Conventional domestic factors of inflation, such as output gaps, wages, labor market developments, and policy stances and measures do not seem to have a robust (statistical) association with Estonia’s headline inflation in recent years. That said, some links can be detected in terms of the longer-term relationships between wages and prices and short-term relationships between the growth of wages and prices for food and some services components of the CPI. The absence of a stronger relationship between domestic factors and broad inflation measures may however reflect data volatility and weaknesses in a period of much lower inflation that was characterized by well-anchored expectations and a relative absence of inflationary shocks. The real time uncertainty over the level of the output gap may be hampering the assessment of the domestic factors that have been contributing to the inflation surge.
26. The recent (and still ongoing) inflation spike has created significant policy challenges. These include: (i) risks of reduced economic and investor confidence because of increased economic, real income, and policy uncertainty; (ii) disruptions to private and public investment planning, including due to increases in construction costs; (iii) adverse impact on income distribution since the high inflation has a particularly deleterious impact on the poor and the most vulnerable; and (iv) potential damage to the credibility of economic policies and frameworks, particularly if some of the other adverse effects of inflation are not well-contained.
27. It is critical to implement policies that prevent an entrenching of elevated inflation while mitigating the effect of the increase in the price level in an efficient manner. The forthcoming normalization of the ECB’s monetary policy with a credible strategy to attain the 2 percent euro area-wide target will help steer inflation dynamics in Estonia in the right direction. In parallel, the reversal of the commodity price surge that is currently expected by the futures markets, will provide a powerful disinflationary push if it materializes. The inflation dampening effect from the declines in global commodity prices may be particularly strong given Estonia’s high estimated passthrough, although caution is warranted due to the possibility of an asymmetric passthrough. Still, due to extremely high current inflation levels, additional policy action may also be needed to amplify the disinflationary forces, including tighter fiscal policies to restrain demand and structural policies in sectors where greater competitive forces or supply-side measures may help further moderate price pressures, notably in the energy sector. Well-targeted and efficient measures of social support to the most vulnerable will be critical components of such a policy package.
References
Brauer D. (1997) “Do Rising Labor Costs Trigger Higher Inflation?” Current Issues in Economics and Finance, Volume 3, Number 11, September.
Chen S. (2009) Oil price pass-through into inflation, Energy Economics, 31, 126–33.
De Brouwer G. and N. Ericsson (1995) “Modeling Inflation in Australia” Board of Governors of the Federal Reserve System, International Finance Discussion Papers, 530, Washington DC, November.
Ericsson N. (2009) “Constructive Data Mining: Modeling Australian Inflation,” mimeo, Federal Reserve Board, Washington DC, June.
Gelos G. and Y. Ustyugova (2012) Inflation Responses to Commodity Price Shocks – How and Why Do Countries Differ?, IMF Working Paper, WP/12/225, September.
Knotek E. and S. Zaman (2014) “On the Relationships between Wages, Prices, and Economic Activity,” Economic Commentary, Federal Reserve Bank of Cleveland, Number 2014-14, August.
Lee C., Schluter. G. and B. O’Roark (2000) “Minimum wage and food prices: an analysis of price pass - through effects,” International Food and Agribusiness Management Review 3 (2000) 111–128.
Neely, C. and D. Rapach, 2011, “International Co-movements in Inflation Rates and Country Characteristics,” Journal of International Money and Finance, 30, pp. 1471–1490.
Rigobon R. (2010) “Commodity Prices Passthrough” Central Bank of Chile Working Paper, 572.
Samal, A., Ummalla, M. & Goyari, P. (2022) “The impact of macroeconomic factors on food price inflation: an evidence from India.” Futur Bus J 8, 15.
Sekinea A. and T. Tsuruga (2018) “Effects of commodity price shocks on inflation: a cross-country analysis” Oxford Economic Papers, pp. 1108-1135, Vol. 70, Issue 4, October.
Prepared by Bogdan Lissovolik (EUR/STA). The analysis benefitted from discussions with the authorities, and the comments received on the presentation that took place during the 2022 Article IV consultation mission.
With respect to these structural characteristics, the literature finds that economies with higher food shares in CPI baskets, fuel intensities, and pre-existing inflation levels are more prone to experience sustained inflationary effects from commodity price shocks (Gelos and Ustyugova (2012)).
This paper uses core inflation measure preferred by the ECB (e.g., excluding energy, food, alcohol, and tobacco).
In this context, the Bank of Estonia has recently highlighted measurement problems of electricity prices that may possibly overstate increases in those prices is recent months (link to Estonian language blog).
Controls also include the lagged dependent variable. Additional global commodity price indexes were tested as candidates for control variables, but they proved less significant relative to the three commodity price variables that were chosen.
In particular: (i) the omitted variable does not appear to be large for Estonia when additional (domestic) controls are included (see section below); this in part reflects data quality problems for other determinants (e.g., output gap); (ii) concerns over reverse causality or similar issues seem to be mitigated in small open economies (such as the Baltics) and (iii) overfitting can be kept in check through robustness analysis and tests.
Public compensation mechanisms of energy (and possibly food) price increases may be different across countries and therefore might affect comparisons of impact on HICP. For example, gas price hikes could be compensated directly to the gas provider or through transfers to households.
This potential asymmetry is not pursued further in this paper, as it merits a separate in-depth investigation.
Estonia’s relatively high energy intensity could be an additional underlying explanation for the high share in the HICP and CPI and passthrough to inflation.
It is important to note that the OLS regressions reported below do not by themselves test for causal relationships but rather should be interpreted as evidence of correlation, unless supplemented by theoretical arguments or additional statistical tests.
Since the mark-up model and the related co-integrating framework apply only to headline inflation but not inflation subcomponents, the analysis below is based on simple OLS regressions that follow an eclectic approach and try to balance theory and empirics.
So far, there has been limited evidence that wage growth is associated with overall inflation in an advanced economy like the US (see Knotek and Zaman (2014)).
