Swedbank Analysis June 27, 2013

Transcription

1 Swedbank Analysis June 27, 213 What drives inflation in the Baltic countries? Consumer price inflation in the Baltic countries has been driven by long-term factors, such as income convergence and higher productivity growth in the traded sector, and a mix of supply-anddemand factors. Global commodity prices are an important driver of inflation in the Baltic countries. The estimated pass-through of international food price shock to domestic food prices is the highest in Lithuania, while that of the crude oil price shock to domestic energy prices is the highest in Latvia. Increases in food and energy price inflation do not lead to persistently rising headline inflation in the Baltic countries, nor do they generate strong second-round effects. The estimation results from the Phillips curve models suggest that the dynamics of headline inflation in the Baltic countries is best explained by the output gap, past inflation, and import price inflation Inflation is a rather inert process in the Baltic countries, which might be due to the inflation expectations of backward-looking agents. Policy actions are necessary in order for Lithuania to increase its chances of meeting the Maastricht inflation criterion and adopting the euro in 215. Promoting competition and price transparency, especially in the regulated sector, fostering productivity growth, and abstaining from further increases in the minimum monthly wage and consumption taxes are crucial for curbing inflation in Lithuania. This paper aims to analyse the factors behind inflation developments in the three Baltic countries from the mid-199s until recently. A better understanding of the factors driving inflation is important for a few reasons. First, high inflation may reduce social welfare, because inflation directly affects people's purchasing power. Second, high inflation may have a negative effect on a country's international competitiveness and thus harm exports, which are an important driver of growth in the Baltic countries. Third, it may result in higher inflation differentials vis-à-vis the euro area and thus become an obstacle for fulfilling the Maastricht inflation criterion and gaining entry to the euro area (Staehr, 21). Estonia adopted the euro in 211 and Latvia will do so in 214. Lithuania has officially committed itself to seeking euro adoption in 215; however, we believe that its probability of meeting the inflation criterion in 215 is not larger than 5%. Swedbank Macro Research. Swedbank AB. SE Stockholm. Phone ek.sekr@swedbank.com Legally responsible publisher: Magnus Alvesson,

2 Factors affecting inflation in the long run Income and price convergence Catching up to the standards of living of those in the developed, old EU member states (real convergence) is one of the main objectives of the lessdeveloped member states, including the Baltic countries. However, countries converging in income levels (real convergence) are likely to experience convergence in price levels as well (nominal convergence). Price level convergence implies that the price of a basket [of goods and services] increases faster than the common currency price of a corresponding basket in the reference area with an initial higher price level (Staehr, 21, p.1). Price level convergence, or nominal convergence, is equivalent to a real exchange rate appreciation, which happens when the nominal exchange rate appreciates or domestic inflation is higher than in the reference area. In the countries with floating-exchange rate regimes, price level convergence materialises mainly through nominal exchange rate appreciation, while in countries with fixed-exchange rate systems domestic prices converge through higher inflation. This is particularly relevant for the Baltic countries, as Estonia is in the currency union, while Latvia and Lithuania have their currencies pegged to the euro at a fixed exchange rate; thus the process of nominal convergence in these countries cannot go through nominal exchange rate appreciation. Thus, price level convergence in the Baltic countries is a potential source of inflation differentials vis-à-vis other EU countries and may create difficulties for the countries in achieving the Maastricht inflation criterion, which is a necessary condition for euro adoption. Countries converging in income levels also go through convergence in price levels In the Baltic countries, nominal convergence emerges through higher domestic inflation In the last decade, the Baltic states experienced rapid real GDP growth, which led to noticeable progress with real convergence as the gap in GDP per capita (in PPS) relative to the EU average narrowed substantially. During the period 2-212, per capita GDP relative to the EU27 average (in PPS) increased from 4% to 7% in Lithuania, from 36% to 62% in Latvia, and from 45% to 68% in Estonia. The process of real convergence was accompanied by nominal convergence, as during that period the relative price level (of GDP) increased from 47.2% to 61.7% in Lithuania, from 51.2% to 68.8% in Latvia, and from 52.4% to 72.4% in Estonia. The price level (of GDP and of household consumption 1 ) relative to the EU average peaked in 28, but then declined during the recession. As the Baltic economies started recovering in 211, so too did their relative price levels. This shows that the processes of real and nominal convergence are highly interrelated. 1 Comparative price level of final consumption by private households, including indirect taxes. 2 Swedbank Analysis June 27, 213

3 Relative income and price levels in Baltic countries, , EU27 = Price level of GDP Estonia Latvia Lithuania GDP per capita in PPS Source: Eurostat To measure this relationship, we estimated two simple regressions linking relative price and income levels, using annual panel data for the Baltic countries from 1995 to 211. The regressions of relative price levels (of GDP and of household consumption) on the per capita GDP relative to the EU27 average (all variables in natural logarithms) estimated an elasticity between.76 and.81, depending on the price level index used. This means that an increase in GDP per capita (in PPS) relative to the EU average of 1% tends to be accompanied by an increase in the relative price level of %. The null hypothesis of no correlation between price level and income variables can be rejected at a 1% significance level. The regressions account for 8-83% of the variability in the price levels. Despite the good overall fit of the simple regressions, some important explanatory variables are not included in them; thus, the results obtained should be interpreted with caution. The processes of real and nominal convergence are highly interrelated Dependent variable Intercept β R-squared Price level of GDP.79*** (.2).81*** (.5).83 Price level of household consumption 1.11*** (.21).76*** (.5).8 Notes: *** significant at 1% level; standard errors in parentheses. In all estimates, the explanatory variable is GDP per capita (in PPS) relative to the EU27 average. All variables are in natural logarithms. The estimation method used is groupwise WLS (iterated). Source: Swedbank estimation As an exercise to illustrate that the Baltic countries still need to go a long way to converge fully to the EU average price level, we took the estimated elasticities as constant. Assuming that the GDP per capita (in PPS) in the Baltic countries will grow at a constant 4.5% rate (average growth was 7.7% during ), and that the average per capita GDP in the EU27 (in PPS) will grow at a 2.5% rate in annual terms (average growth was 3.4% during ), 2 we calculated how long it will take to close the gap between the price levels in the Baltic countries and the EU. We project that Estonia will reach 1% of the EU average price level of GDP in 234; Latvia, in 237; and Lithuania not sooner that in 242. Meanwhile, in the Baltic countries the relative price levels of final consumption by private households are higher than the price levels of GDP; thus, despite an estimated lower 2 For the period , the IMF forecasts average annual GDP per capita in PPP growth of 5.6% in Lithuania, 5.1% in Latvia, and Estonia and 3% in the EU (source: EconStats). Swedbank Analysis June 27, 213 3

4 elasticity for the price level of household consumption, it should take, according to our projections, a shorter period of time for this price level to converge fully to the EU average price level. We project that the price level of household consumption in Estonia will reach the average EU price level by around 228, and that Latvia will converge fully in 232 and Lithuania only in 24. Relative price level of GDP and of household consumption (EU27 = 1), p 219p 225p 231p 237p PL of GDP - EE PL of GDP - LV PL of GDP - LT PL of cons. - EE PL of cons. - LV PL of cons. - LT It will take a long time for the Baltic countries to converge fully to the EU price level Source: Swedbank calculation Meanwhile, we project that, in terms of income, Estonia will converge fully to the EU average level in 232; Lithuania, in 233; and Latvia only in 239. Relative GDP per capita in PPS (EU27 = 1), p 219p 225p 231p 237p Estonia Latvia Lithuania Source: Swedbank calculation However, the convergence projections are based on rather rough income growth assumptions, given such a long projection period; thus the actual convergence will be more volatile and may take longer or shorter, depending on the pace of income growth in the Baltic countries and on the average in the EU, as well as other factors. Moreover, the projections are based on estimated elasticities of very simple regressions, which may suffer from the omitted variable bias; thus, these projections are rather crude and should be interpreted with care. Price convergence has not been even across different consumer product categories in the Baltic countries. In all three Baltic countries clothing and footwear are more expensive than on average in the EU. The full conver- Relative price levels differ across different product categories 4 Swedbank Analysis June 27, 213

5 gence in the price level of clothing and footwear in the Baltic countries can be explained by the fact that these goods are highly tradable; thus their prices to a large extent are determined in international markets (the law of one price), although transportation costs and the market size matter as well. The same applies to prices of household furnishings, which are at about 75-85% of the EU average. The transport category consists of personal transport equipment and transport services. As personal transport equipment consists of highly tradable goods, their prices are about 85-9% of the EU average price level; meanwhile, as transport services are subsidised and more wage sensitive, their prices in the three Baltic countries are still at only 6-7% of the EU average. The prices of food have been converging steadily to the EU average level; however, they still remain 13-23% below the EU average price level because of the limited tradability of some food products. Due to short conservation periods as well as local tastes, some products, such as bread, are very often produced and consumed in the home country; thus, their prices are to some extent determined by demand factors in the specific country and have not yet fully converged to the EU average price level. Prices of tradable goods are closer to the EU average price level than those of nontradables The price levels of some nontradable services, such as health and education, remain well below the EU price level, and the prices of communications, due to very intense competition and large investments, are even diverging from the EU average price level. Interestingly, Lithuania has the lowest price level of communications among all EU member states. Surprisingly, housing prices are only 46% of the EU average in Lithuania, compared with 63% and 77% in Latvia and Estonia, respectively. The lower housing and utility prices in Lithuania can be explained by the provision of housing expense discounts to poor households by the Lithuanian government. Furthermore, Lithuania and Latvia, unlike Estonia, apply a reduced value-added tax (VAT) on heating services. Relative price level indices by product categories, 212, EU27=1 Actual individual consumption Clothing and footwear Food and non-alc.beverages Household furnishings Alc.beverages and tobacco Transport Recreation and culture Communication Restaurants and hotels Housing and utilities Health Education Estonia Latvia Lithuania Source: Eurostat The impact of real convergence on inflation is likely to diminish over time. However, the Baltic countries still display productivity and income levels that are considerably below those of the old EU member states. Thus, together with real convergence processes in the Baltic countries, there will be a further convergence in the price levels. Nominal convergence will also result in inflation differentials vis-à-vis the euro area, which may even get in the way of Lithuania's adopting the euro in 215. Swedbank Analysis June 27, 213 5

6 Balassa Samuelson effect and other culprits There are a few different channels through which real convergence may affect the price levels of the catching-up economies, such as the Baltic countries. The first channel is productivity growth, which is most commonly related to the Balassa-Samuelson (BS) effect. According to the BS model, the economy has a traded and a nontraded sector. It is assumed that the prices of tradable goods are determined from abroad and wages are paid in accordance to labour productivity. Stronger productivity growth in the traded sector pushes up wages in this sector. Under the assumption of perfect labour mobility across sectors, wage growth is carried over into the nontraded sector; thus the wages equalise across sectors. The mismatch between labour productivity and wages in the nontraded sector translates into higher prices of nontradable goods and services and, hence, an increase in the overall price level. The BS effect implies that higher, faster productivity growth in the traded sector ultimately causes faster price growth in the nontraded sector Assuming that the factor intensities are the same in the traded and nontraded sectors, this domestic version of the BS effect implies that the difference in inflation rates of nontradables and tradables, i. e., the growth rate of the relative price of nontradables, should equal the difference in the growth rates of labour productivity in tradable and nontradable sectors. The BS effect in the new EU member states has been widely analysed, and mixed results have been obtained. Staehr (21) finds that higher productivity growth in the traded than in the nontraded sector has exerted upward pressure on inflation in the CEE countries. According to Mihaljek and Klau (29), the domestic BS effects are positive in Estonia, but negative in Latvia and Lithuania-- i.e., the BS effects subtract from rather than add to overall inflation. These authors also find evidence that higher productivity growth in tradable relative to nontradable industries has contributed to higher inflation in these countries vis-à-vis the euro area (the international BS effect). However, the evidence on the size of the BS effect is rather controversial, as the model is based on many assumptions and there is no accepted criterion for the classification of the sectors into tradable and non-tradable. In this paper, to obtain labour productivities, we classified agriculture, forestry and fishing, and manufacturing as tradables, while classifying construction, wholesale and retail trade, transport, accommodation and food service activities, information and communication, financial and insurance activities, and real estate activities as nontradables. To obtain the relative price of nontradables, the HICP inflation of total goods was used as a proxy for the inflation of tradables, and that of total services was used as a proxy for the inflation of nontradables. In the Baltic countries, average labour productivity growth in the traded sector tends to outpace productivity growth in the nontraded sector (see Appendix 1), which allows the BS effect to kick in. In line with the BS hypothesis, until 21 the inflation of nontradables tended to move in line with inflation of tradables, or was even higher (see Appendix 2). However, in 21 the inflation of tradables started to outpace that of nontradables. The negative gap between the inflation of nontradables and that of tradables opened up because wages contracted during the recession; since the nontraded sector is more labour intensive, this allowed it to lower the prices of nontradable goods and services. Price increases were also impossible due to very weak Average labour productivity growth in the traded sector tends to outpace productivity growth in the nontraded sector 6 Swedbank Analysis June 27, 213

7 domestic demand during this period. Meanwhile, the inflation of tradable goods was fuelled by rising commodity prices. This suggests that, despite higher growth of productivity in the tradable than in the nontradable sector, the positive BS effects were unlikely during the period. However, until 21, the productivity differentials could have contributed positively to sectoral inflation differentials and, thus, to overall inflation in the Baltic countries. Moreover, the inflation of tradables started to decelerate in 211 and reached the inflation level of non-tradables in Thus it might be the case that the BS effect is starting to kick in, and it might be an important factor driving inflation as the Baltic economies continue to recover. Productivity growth and inflation differentials, period average, pp Lithuania Latvia Estonia Lithuania Latvia Estonia Positive BS effects were unlikely during the period of growth in the non-traded sector, due to contracting wages and weak domestic demand Productivity grow th differential (T-NT) Inflation differential (NT-T) Source: Swedbank calculation Income convergence can also affect inflation through changes in the consumption pattern. With rising income, the demand for nontradable goods and services relative to tradable goods tends to increase, which has been the case in the Baltic countries during the past decades. For example, in the Baltic countries the weight of services in the consumer basket has risen from 12-18% in 1998 to 25-29% in 213. Meanwhile, the weights of some tradable goods--e.g., food and clothing--have decreased substantially. Growing income in the Baltic countries will further lead to a greater weight of nontradable goods and services in the consumption basket in these countries, which will amplify the upward pressure of the BS effect on consumer prices. Moreover, higher income might make the demand for both tradable and nontradable goods less price elastic. This may increase markups for sellers, who operate in markets with little competition, and, thus, to higher inflation (Staehr, 21). The real convergence process is in many cases accompanied by more trade openness. However, this channel may affect prices in both directions. On one hand, inflation may increase because the prices of tradable goods catch up with international prices through the law of one price. On the other hand, it may also lead to lower price levels, as greater openness may increase competition, which, in turn, may reduce markups and lower prices. Chen et al (24) empirically confirm that greater openness lowers prices in the EU countries by both reducing markups and raising productivity. Meanwhile, Staehr (21) does not find any evidence that increased openness has an effect on inflation in the CEE countries. Rising income leads to higher demand for nontradable goods and services Increasing trade openness can affect prices in both directions 3 Inflation of tradables and nontradables may also be affected by tax changes; thus we also looked at inflation differentials using inflation at constant taxes. However, this did not change the situation significantly in the Latvian or Estonian cases (no such data were available for Lithuania). Swedbank Analysis June 27, 213 7

8 Openness to trade imports plus exports, % of GDP 2% 18% 16% 14% 12% Lithuania 1% Latvia 8% Estonia 6% 4% 2% % Source: Eurostat, Swedbank calculation Openness to trade, which is measured as the sum of imports and exports as a percent of GDP, has been increasing rapidly in the Baltic countries since 21; last year, it reached all-time highs. The indicator has been increasing mainly due to booming exports. Although the inflation of tradable goods has been above that of non-tradables, the price level of consumer goods relative to the average price level in the EU has not changed significantly since 21. Thus, it is difficult to say whether the increase in trade integration has had any effect on domestic inflation recently. Openness to trade has been increasing since 29 due to growing exports Factors affecting inflation in the short run Demand factors The position of the output gap is an important factor affecting inflation. Economic theory suggests that when the economy operates above its potential, i.e., the output gap is positive, there will be excess demand for output, which will lead to rising prices. Conversely, when the output gap is negative, there will be an excess supply of resources, which, in turn, will be reflected in downward pressure on prices. A positive output gap demand exceeding capacity - leads to accelerating prices Annual Inflation and output gap, counterclockwise Latvia Lithuania Estonia Inflation, % Output gap, % 2-4 Source: Eurostat, EC (AMECO) Historical data suggest that there is a positive relationship between the output gap and inflation in the Baltic countries. In Lithuania, Latvia, and Estonia, 8 Swedbank Analysis June 27, 213

9 the gap between actual and potential output as a share of potential output 4 started to open up in 23 and was expanding until it reached 11-13% in 27. As a result, the annual inflation rate reached double digits in 28. The collapse of economic activity in resulted in a contraction in the positive output gap, which became negative in 29; this has not yet returned to its potential level in Latvia and Lithuania. Meanwhile, Estonia recovered faster after the economic recession and has had small positive output gaps for the last two years and slightly higher inflation. In addition, it is obvious that the negative output gap is closing up in Latvia and Lithuania and that economic growth will soon approach its potential level; thus, inflationary pressures will start to rebuild when the output gap becomes positive again. The credit growth is another important factor that may have a significant effect on consumer demand. During the economic boom, consumer demand was fuelled by cheap and easily receivable credit. Due to the high uncertainty in the market, credit growth halted and interest rates increased significantly during the years of economic recession. The household loan portfolio has been contracting since 29 in the Baltic countries, but, due to improving expectations about the future, growing wages, and record-low interest rates, the trend is likely to be reversed sometime soon. However, we do not believe that credit growth will reach the levels seen before the crisis; thus it will not drive domestic demand enough for it to have a large effect on inflation. During the economic boom in the Baltic countries, the increasing demand for labour, combined with high emigration rates after accession to the EU, resulted in an overheated labour market. The unemployment rates dropped below the natural level and to all-time lows of % in 27, which led to increasing competition among companies for the labour force; this, in turn, pushed real wage growth above the growth of productivity. The rapid wage growth fuelled consumer demand and made it easier for companies to increase their markups, which, again, put additional pressure on inflation.5 During the recession, the level of unemployment surged and wages declined. The purchasing power of households decreased significantly, as did prices. Although domestic demand has been weak recently, the further growth of real wages will stimulate household consumption. In Estonia wage growth is likely to lead to higher prices of services, however, we do not believe that it will create inflationary pressures in Latvia and Lithuania in the near future, at least not from the demand perspective. In the Baltic countries, negative output gaps are turning positive again Credit growth was an important inflationary factor in the past but is unlikely to drive inflation in the future Wage growth will not create strong inflationary pressures in the near future 4 Output gap estimates are subject to great uncertainty since potential output is unobservable. 5 For more on product market competition, see Swedbank Analysis, published on June 14, /index.csp Swedbank Analysis June 27, 213 9

10 Annual real net wage growth 25% 2% 15% 1% 5% % -5% Estonia Latvia Lithuania -1% -15% Source: National Statistics An increase in inflation expectations can have a direct effect on domestic demand and, thus, on inflation. If the households believe that inflation will rise in the near future, they may reduce their future consumption and increase current consumption, which would boost domestic demand and thus lead to an increase in prices. Inflation expectations (balance of answers) have been rather stable in recent years in Lithuania, but still are higher than in the other two Baltic countries. In Latvia, due the improving macroeconomic situation, as well as to the plans to adopt the euro in 214, consumer expectations about inflation have been rising recently. Meanwhile, Estonian consumers have started lowering their inflation expectations since the end of 212. Consumer expectations of a price increase at the end of 212 in Estonia came from the announcement of the liberalisation of the electricity market in January 213, through which consumers were informed about an increase in the electricity price. However, in the past we have found that consumers usually are not very good at predicting prices. 6 Inflation expectations can have a direct effect on both domestic demand and inflation Consumer inflation expectations over next 12 months, balance of answers Lithuania Latvia Estonia Source: DG ECFIN Inflation expectations differ significantly across different business sectors and tend to be lower than consumer inflation expectations. The higher consumer inflation expectations may to a certain extent be explained by consumers' backward-looking behaviour - if one thinks that prices have been rising, one most probably expects them to rise further. Business representa- 6 See Swedbank Discussion Paper, published on June 14, Swedbank Analysis June 27, 213

11 tives, on the other hand, are much better acquainted with their cost drivers and their intentions to change prices in the future. Nevertheless, business inflation expectations have only a very limited power to predict actual price developments. Inflation expectations over next 12 months across sectors, Lithuania, balance of answers Construction Services Retail Industry Source: Statistics Lithuania Supply factors In the short run, inflation can be affected not only by demand factors, but also by supply factors, or simply by the increase in the costs of production. As mentioned above, rapid wage growth may create inflationary pressures through an increase in domestic demand. Moreover, wage growth affects companies production costs, which may be passed on to consumers through higher prices. During the period, nominal wage growth was outpacing real productivity growth in all three Baltic countries, which resulted in rising nominal unit labour costs (ULCs). 7 The increase in labour costs pushed up total production costs and contributed positively to rising consumer prices. The sharp drop in wages and rising labour productivity during the recession resulted in a fall in the nominal ULC and inflation; however nominal ULCs have been growing in Lithuania and Estonia since the end of 211. Rising nominal ULCs may create inflationary pressures in the Baltic countries, as well as undermine the countries cost competitiveness in the near future. Wage growth's overshooting productivity growth may create inflationary pressures 7 Nominal ULC is the ratio of total nominal costs per employee to real labour productivity. Swedbank Analysis June 27,

12 Annual growth of nominal unit labour costs, % Lithuania Latvia Estonia Source: Eurostat, Swedbank Inflation in the Baltic countries depends largely on the changes in import prices. Meanwhile, the import prices are heavily influenced by the changes in global commodity prices. Thus, a global commodity price shock puts upward pressure on domestic inflation, especially in such small and open economies like the Baltic countries'. Moreover, the value of the euro has an effect on import prices, as the euro is the official currency in Estonia, and Lithuania and Latvia have their currencies pegged to the euro. In the Baltic countries, the depreciation of the euro with respect to the US dollar translates into higher prices of imported commodities (gas, oil, etc.), which are being paid for in US dollars, and causes inflation to rise. Global commodity prices and the value of the euro affect import prices and thus inflation Commodity prices have been increasing rapidly in the last decade they rose substantially during the period of 23-28, fell sharply during the global recession, and increased again in 21 and 211, reaching and sometimes even exceeding pre-crisis peaks. An increase in commodity prices, especially those of food and oil, fed through import prices into local consumer prices. Such direct (first-round) effects on domestic inflation are determined by the weights of these commodities in the consumer basket and their usage as inputs in the production of final goods. However, increases in commodity prices may also generate indirect (second-round) effects on inflation due to higher inflation expectations and, thus, higher wage demands. Global commodity price (in EUR) indices, 24 = Crude oil Food Industrial raw materials Source: HWWI institute 12 Swedbank Analysis June 27, 213

13 The direct effect of commodity price swings on domestic inflation depends largely on the composition of the consumer basket the larger the share of the commodity-related components in the consumption basket, the more vulnerable is the country to global commodity price developments. The share of food in the HICP consumption basket is rather large in the Baltic countries, where it makes up 21-25% of the consumer basket, compared with just 15% in the euro area countries. The difference is less striking when it comes to energy in the Baltic countries, the share of energy in the consumption basket is 14-16%, while in the euro area only around 11%.Of the Baltic countries, Lithuania has the largest share of both energy and food in its consumption basket. In Lithuania, the share of energy and food in its consumption basket is higher than in Latvia and Estonia Weights of food, energy, and other items in consumer basket, 213 EA Food Estonia Energy Latvia Non-food non-energy Lithuania % 2% 4% 6% 8% 1% Source: Eurostat Gelos and Ustyugova (212) empirically confirm that economies with large weights of food in the consumer basket and high oil intensities are more prone to experience strong inflationary pressures from food and fuel price shocks. Energy intensity, measured as inland consumption of energy per unit of real GDP, is significantly higher in the Baltic countries than the EU average. Of the Baltic countries, Lithuania consumes the lowest amount of oil equivalent in relation to its GDP. Latvia s energy intensity is slightly higher; however, the extremely high consumption of energy by Estonia remains unrivalled. For example, the energy intensity in Estonia is almost four times the average intensity in the euro area countries. Despite this high energy intensity, Estonia is less dependent on imported energy resources, i.e., natural gas, than Latvia or Lithuania. This is because Estonia, for heating and production of electricity, relies to a large extent on domestic oil shale reserves, while in Latvia and Lithuania mainly imported natural gas is used. Since oil shale mining is very energy intensive, this to some extent explains the very high energy intensity in Estonia. Energy intensity is highest in Estonia Swedbank Analysis June 27,

14 Energy intensity of the economy, kilogram of oil equivalent per 1 euro EU27 Euro area Sw eden Lithuania Latvia Estonia Source: Eurostat Another important factor that affects an economy s vulnerability to the inflationary impact of a commodity price shock is the credibility of its monetary policy. A fully operational central bank has the power to affect inflation and its expectations by altering the monetary base and bank reserve requirements, adjusting interest rates, and raising public awareness of the issue. However, in the Baltic countries monetary policy responses regarding inflation are constrained because monetary policy is largely exogenous and has only a very remote and indirect impact on inflation dynamics. Lithuania has a currency board the strictest form of a currency peg, which strips the central bank of most monetary policy instruments and, most important, forbids it from issuing domestic currency uncovered by foreign currency reserves. This monetary choice in itself has been a very good safeguard against speculation and inflation expectations. Changes in regulated prices are another important factor that affects inflation in the Baltic countries. The prices in the monopolistic sectors (gas and electricity, and heating) or certain sectors that provide public services (e.g., public transport, health and education services, and postal services) are usually influenced or approved by the regulatory institutions. The growth of administered prices has eased from the record highs reached in 28, when they were inflated mainly by rapidly rising commodity prices and public sector wages. Since administered prices constitute a significant share of the consumer basket (1-16% in 212), they add significantly to the overall inflation in the Baltic countries. Inflation and its expectations in Baltic countries are managed by the lack of active monetary policy Changes in administrated prices have a significant impact on overall inflation 14 Swedbank Analysis June 27, 213

15 Annual changes in administered prices, % Estonia Latvia Lithuania Source: Eurostat Increases in indirect taxes, such as the value-added tax (VAT) and excise tax, directly affect the prices of the goods on which these taxes are imposed. The excise taxes on alcohol, tobacco, and fuel have been continually increased in the Baltic countries in order for them to meet the minimum requirements of the EU; this resulted in higher inflation of these items since the accession to the EU. The VAT was increased in the Baltic countries during the crisis; however, due to the harsh economic situation, it was most likely not fully transferred to consumer prices but resulted instead in a squeeze in companies profit margins. Commodity price pass-through Realising the vulnerability of the Baltic countries to global commodity price developments, it is important to quantify the effect of international price increases on inflation in the Baltic countries. This brings us to the next section, in which we perform a pass-through analysis. We estimate the spillover (pass-through) from international commodity prices to domestic commodity prices in the Baltic countries, following the IMF (211) estimation methodology. In particular, we estimate the effect of a 1% increase in international food and oil prices on domestic food and energy prices, respectively. The pass-through analysis is based on country-by-country bivariate regressions of monthly domestic food or energy price inflation on current and 12 lags of monthly international food or oil price inflation, respectively, controlling for 12 lags of domestic food or energy price inflation: domestic domestic world t = α + β jπ t j + γ kπ t k j= 1 k= π + ε t, domestic where π denotes the annualised month-over-month log difference in the domestic food or energy component of HICP, 8 world and π stands for the 8 For food, we use food and nonalcoholic beverages (COICOP 1), and energy includes electricity, gas and other fuels (COICOP 4.5), plus fuels and lubricants (COICOP 7.2.2) Swedbank Analysis June 27,

16 annualised month-over-month log difference in international food 9 or Brent crude oil prices. In order to control for seasonal fluctuations, the regressions include monthly seasonal dummies. International oil and food prices used in the analysis are expressed in euros in order to account for exchange rate fluctuations. For this analysis, we use monthly data from 21 to 212. The pass-through coefficient, which reflects the full, long-term pass-through from international to domestic prices, is computed in the following way: PT 12 k k= = 12 1 γ j= 1 β j The unit root tests (ADF) have been carried out prior to the estimation to test for the stationarity of the variables in the model above. The null hypothesis of the series' containing a unit root was rejected at the 1% level for all three Baltic countries; thus, all the variables in the model are stationary. The estimated long-term pass-through of a 1% food price shock to domestic food prices is.32% in Lithuania the highest in the Baltic countries. The pass-through on domestic food inflation is lower in Latvia and Estonia, where it stands at.2% and.17%, respectively. These findings are in line with the ones by the IMF (211), which performed a similar pass-through analysis and found that the median pass-through to domestic food inflation was only.18% in advanced economies and.34% in emerging and developing economies. The estimated effect of a 1% increase in international oil prices on domestic energy prices is the highest in Latvia, where it stands at.36%. Meanwhile, Lithuania and Estonia seem to be less vulnerable to world oil price fluctuations, as the pass-through to domestic energy prices is around.2% in these countries. The pass-through of food price shock to domestic food prices is the highest in Lithuania The pass-through of oil price shock to domestic energy prices is the highest in Latvia Pass-through from world commodity inflation to domestic food and energy inflation, % Lithuania Latvia Estonia.5. International food prices to domestic food inflation International crude oil prices to domestic energy prices Source: Swedbank calculation 9 The HWWI food index is used, which includes cereals (barley, maize, wheat, and rice), oilseeds (soybeans and soybean meal), oils (soybean oil, coconut oil, palm oil, and sunflower oil), and tropical beverages (coffee, cocoa, and tea). 16 Swedbank Analysis June 27, 213

17 One might be surprised by such a small effect of world commodity inflation on domestic inflation, but a number of factors explain this incomplete passthrough. First, the cost share of commodities in the final product is below 1%, as labour and other administrative costs account for a large part of the final price of the product. Thus, commodity prices affect only a part of the input costs of food and energy. Second, international commodity price indices do not necessarily reflect domestic consumption patterns. For example, an increase in international prices of certain food items that have a large weight in the food index would raise the overall food price index, but the effect of this increase on domestic food inflation might be rather small if the consumption of those food items were relatively small in the Baltic countries. In order to minimise the compositional effects, we repeat the pass-through analysis by focusing on a more tightly defined consumer product category, such as fuel. Thus, we estimate the pass-through of world crude oil prices to domestic fuel prices. We find that a 1% increase in international crude oil prices leads to.32% and.37% increases in domestic fuel prices in Lithuania and Latvia, respectively. World oil prices have the highest effect on domestic fuel prices in Estonia, which has an estimated pass-through of.45%. These findings are again in line with the ones by the IMF (211) their estimated pass-through of world crude oil prices to gasoline prices in EU27 countries is.34%. This incomplete pass-through (lower than 1%) can to a large extent be explained by the large share of excise duties and the VAT in the final fuel price in the EU (unlike, for example, in the US). The pass-through of oil price shock to domestic fuel prices is the highest in Estonia Is headline reverting to core and vice versa? The impact of rising commodity prices on domestic inflation dynamics can be transitory or permanent, depending largely on the nature of the commodity price shock. If commodity price hikes are due to adverse supply shocks, then it is more likely that the impact on inflation will be more transitory an inflationary impulse from commodity price hikes will soon die out and headline inflation will reverse to core inflation. 1 On the contrary, commodity price increases may have a more persistent effect on inflation if the upside shocks to commodity prices are more persistent (e.g., due to rising demand for energy or food products), or if they lead to second-round effects. We now turn to the question of whether headline inflation returns to core inflation in the Baltic countries, once the gap between headline and core inflation has been opened (see Appendix 3). If it does, it means that increases in commodity prices do not lead to persistently rising domestic headline inflation. Following Cecchetti and Moessner (28) and OECD (25), we test this hypothesis by considering this regression: π π + headline t headline headline core t 12 = α + β ( π t 12 π t 12 ) ε t, headline core where π denotes headline inflation and π is the core inflation. Headline inflation is calculated as the year-over-year log-difference in the HICP, while core inflation as the year-over-year log-difference in HICP, excluding energy and unprocessed food. The regressions are estimated for Lithuania, Latvia, and Estonia separately, using monthly data from 23 to 212. If headline reverts to core, the coefficient β on the gap between head- 1 Headline inflation is the total consumer price inflation in the economy. Core inflation is a measure that excludes certain components with volatile price movements. In our case, core inflation excludes energy and unprocessed food. Swedbank Analysis June 27,

18 line and core inflation should be negative; if β =, then headline inflation does not revert to core. After estimating the regressions, we find that the coefficient β is negative in all three Baltic countries. 11 This suggests that headline inflation tends to revert to core inflation in the Baltic countries; thus, increases in food and energy price inflation are mostly temporary and do not lead to persistently rising headline inflation. Interestingly, in Estonia the headline inflation has been reverting to core inflation faster than in Latvia and Lithuania. This could be due not only to a difference in price structure, but also to greater price transparency and stronger competition in Estonia. Increases in food and energy price inflation do not lead to persistently rising headline inflation We further study the converse situation - whether core inflation reverts to headline inflation. The reversion of core inflation to headline inflation is consistent with a worrying consequence of second-round effects - higher food and energy prices cause inflation and wage expectations to rise and, thereby, core inflation to catch up with headline inflation. Once again following Cechetti and Moessner (28), we study this issue by estimating this regression for the three Baltic countries separately: π π + ε core t core core headline t 12 = γ + δ ( π t 12 π t 12 ) t If core inflation reverted to headline, the coefficient δ would be negative, indicating the presence of second-round effects, while δ = would imply that core does not revert to headline inflation. The estimation results suggest that there is no evidence for second-round effects of higher commodity prices on inflation in the Baltic countries, as, although all the estimated coefficients are positive, they are not statistically significant. This may be because inflationwage growth expectations were feeding themselves for only a short period before the crisis. Estimation results are provided in Appendix 4. No evidence exists for strong second-round effects of higher commodity prices on inflation Phillips curve estimation In this section, we perform a simple empirical analysis in order to assess which short-run factors have been affecting inflation in the Baltic countries. This is done using the traditional, backward-looking Phillips curve model. Such a model explains inflation with demand-and-supply factors, such as the output gap and cost-push terms, thereby capturing productivity, input prices, and labour market developments. Our empirical analysis is based on quarterly data for the three Baltic countries from the second quarter of1997 to the third quarter of 212. Following Eickmeier and Moll (29) and Benkovskis et al. (29), we use the traditional, backward-looking Phillips curve model specification: 1 1 t = α + βπ t 1 + γgapt 1+ δ j impt j + ϑ j ulct j ε t, j= j= π + where π t is inflation, calculated as the quarter-on-quarter log difference in the HICP, which is regressed on past inflation ( π t 1 ), the lagged output gap ( 11 The coefficient is statistically significant (the null hypothesis that the coefficient is zero is rejected) at a 1% significance level for Lithuania and Estonia, and at a 1% level for Latvia. 18 Swedbank Analysis June 27, 213

19 gap t 1 ), quarter-on-quarter log-differences of import prices ( impt j ulct j ). unit labour costs ( ), and Past inflation captures the persistence of inflation trends that could be caused by inflation inertia or inflation expectations of backward-looking agents. If this is the case, then we should get a positive-sign coefficient on the past inflation variable. The cyclical demand conditions are reflected by the output gap, which is measured as the Hodrick-Prescott (HP) 12 -filtered real GDP. If inflation behaves procyclically, then a coefficient with a positive sign on the output gap is expected. Meanwhile, the supply shocks are reflected by the changes in unit labour costs and import price inflation, and we again expect positive-sign coefficients on these two variables. The model is estimated for Lithuania, Latvia, and Estonia separately. The unit root tests (ADF) have been carried out prior to the estimation of the model to test for the stationarity of the variables. The null hypothesis of the unit root was rejected for all the variables for all three Baltic countries, except for unit labour costs for Estonia. Thus, this variable was excluded from the regression model in the Estonian case. The estimation results suggest that inflation dynamics in the Baltic countries are explained by the output gap, past inflation, and import price inflation. These variables appear to be statistically significant and with expected positive signs. Meanwhile, the unit labour cost variable enters the regression significantly only in the Latvian case; it remains insignificant for Lithuania (it is excluded from the regression for Estonia). This is in line with our previous observations that the increases in unit labour costs during the economic boom were more pronounced in Latvia than in the other Baltic countries, and therefore had a significant effect on inflation in Latvia (see Appendix 5). However, it should be kept in mind that the traditional Phillips curve framework is a reduced-form model; thus the results obtained by estimating it should be interpreted with care. Headline inflation is best explained by the output gap, past inflation, and import price inflation A few studies have tackled the issue of inflation dynamics and the factors affecting inflation in the Baltic countries by estimating a Phillips curve model. Bitans et al (21) have estimated the traditional Phillips curve for Latvia; they report a statistically significant effect of the output gap, unit labour costs, and nominal effective exchange rate (used instead of import prices) on core inflation rates. Benkovskis et al (29) estimate the traditional, backward-looking Phillips curve model for all three Baltic countries and conclude that the dynamics of core inflation in the Baltic countries are best explained by the output gap; meanwhile, the authors find little evidence that supply-side shocks explain the fluctuations of core inflation rates in the Baltic region. Masso and Staehr (25), Melihovs and Zasova (27), Dabusinskas and Kulikov (27), and Virbickas (212) estimate forwardlooking new Keynesian Phillips curve models; thus, their results are not comparable to ours. Fulfilling the Maastrich inflation criterion in Lithuania The Baltic countries committed themselves to adopting the euro as their currency when they joined the EU. Estonia successfully adopted the euro in 211. Latvia has been positively assessed on its compliance with the Maastricht convergence criteria and soon will be officially invited to join the euro area on 1 January 214. Lithuania has an official target of adopting the euro 12 The HP filter is a statistical method used to decompose time series into a trend component and a cyclical component. For quarterly data, λ=16 is used. Swedbank Analysis June 27,