Publications

The goal of this paper is to determine the heterogeneity in communication among the ECB Governing Council members, employing the GPT language model, and assess its impact on financial markets with high-frequency data since 2014. Our findings uncover trends in sentiments and topics in line with economic developments and similar studies. We highlight a marked divergence between hawkish and dovish members. Using regression analysis, we demonstrate that the communication from the Governing Council members significantly influences various euro area asset classes, particularly Italian sovereign bonds. These effects are consistently significant across different periods. Our study also highlights specific topics that exert a more pronounced influence on market dynamics, notably unconventional monetary policy since 2014 and, more recently, interest rate discussions. Hawkish communication is observed to have a more substantial effect on sovereign bond yields. Furthermore, speeches by Executive Board members and the Governors of Germany, France and Austria are found to have the most significant overall impact on euro area markets.

Keywords: ECB, monetary policy, communication, sentiment analysis, euro area, financial markets.

JEL codes: C80, E43, E44, E58, G14.

We provide evidence regarding how European and U.S. macroeconomic and monetary policy events affect euro area markets. By analyzing over 170 macroeconomic indicators from 2002 to 2024Q1, we assess the impacts across different events, countries, and time periods. We rely on a high-frequency impact identification strategy and estimate the effects of releases on market variability and the directional effects of surprises across different markets. We find that, compared to European data releases, U.S. events tend to have a more pronounced effects on euro area markets. On average, monetary and employment events induce the strongest repricing in euro area financial markets. In most of the specifications, long-term sovereign yields were more sensitive to macroeconomic releases than other instruments, while short-term bonds and stock prices were impacted much less frequently.

Keywords: macroeconomic news, macroeconomic releases, monetary policy, euro area, financial markets.

JEL codes: E43, E44, F40, G12, G14, G15

After a decade of muted consumer price growth, inflation has picked up again, with the price increase of many goods and services spiking in 2022. Two extraordinary events – the COVID-19 pandemic and the Russian aggression against Ukraine – played a leading role in the jump in inflation. The high risk of deep recession in 2020-2021 forced governments and central banks to implement various supportive measures. As the economies adapted to the pandemic, recoveries followed unexpectedly quickly with the help of expansive monetary and fiscal policies. Nevertheless, pandemic-induced supply-chain disruptions have resulted in delivery delays and increased production and transportation costs across the globe. Thus, recovering demand faced a still-constrained supply. In 2021, recovering economies were hit by another shock – the Russian war against Ukraine. The war contributed to a rise in energy prices (notably, that of natural gas, which Europe was especially dependent on). All these factors – expansionary fiscal and monetary policies, rapidly recovering economies, residual post-pandemic disruptions in supply chains, and increases in energy prices – have led to an unexpected rise in inflation throughout the world, including Lithuania.

In this occasional paper, we analyse various topics related to inflation in Lithuania, predominantly focusing on the recent inflationary episode. The latter rise of inflation was unprecedented. In 2022, average annual inflation reached 18.9 per cent in Lithuania, a level which had not been seen for more than two decades. We analyse the nature of the recent inflation shock, duration, underlying causes, and consequences. While this study mainly deals with Lithuania, it also addresses the question of whether its inflation dynamics differs from that in the rest of the euro area, and if so, how. The study thus contributes to a more nuanced understanding of inflationary process in Lithuania. While integrated within the general topic, each of the chapters in the study can be seen as separate analytical notes focusing on distinct topics.

In Section 1.1. (“Stylized facts of consumer price dynamics”) and Section 1.2 ("Dynamics of consumer, producer and input prices”), we provide an overview of the dynamics of inflation and its components in Lithuania over the past two decades. During the economic boom of 2004 to 2008, Lithuania experienced an upward pressure in consumer prices. This ended in 2009 with the global financial crisis, which triggered a significant downturn in the Lithuanian economy. Afterwards, a period of relative stability in inflation took place until the COVID-19 pandemic. At its start in 2020, consumer price inflation decelerated, but price growth picked up in 2021-2022. Since reaching its peak in 2022, the annual inflation rate has been steadily declining. Historically, energy prices in Lithuania have been characterized by especially high volatility. During periods of higher inflation, they have been one of the main drivers of inflation, while during periods of lower inflation, energy prices have been an important factor reducing it.

In Section 1.3. (“Inflation expectations of Lithuanian households”) and Section 1.4. (“Inflation expectations of Lithuanian firms”), we use existing survey data on Lithuanian households’ and firms’ inflation expectations to better understand their evolution in the recent high inflation environment. A clear upward bias can be observed in households’ and firms’ inflation expectations. However, there is also a significant co-movement between actual inflation and inflation expectations. As inflation started to decline in 2023, similar trends can be observed in inflation expectations. 

In Section 2.1. (“Effects of energy supply shocks on price inflation along the production chain”), we assess the impact of energy supply shocks on price inflation along the production chain in Lithuania. The energy shocks are identified in two independent monthly BVAR models (Messner and Zorner (2023)). Producer price inflation for energy and food reacts at half the rate of equivalent international inflation in the month of the shock and then continues to rise for a year or year and a half. Consumer food price inflation reacts to a similar extent as producer food price inflation, while consumer energy price inflation reacts to a lesser extent than producer energy price inflation. More importantly, these reactions occur with a lag of about one year after the shock. Finally, the impact at the bottom of the production chain, i.e. on core consumer price inflation, is quite limited. Overall, this section shows that energy supply shocks propagate gradually through the supply chain over time and are not passed on, on a one-by-one basis, to the final consumer.

In Section 2.2. (“Wage and price responses to aggregate and labour market shocks”), we assess how global and labour market shocks affect wages and consumer prices in Lithuania, and how wage responses in turn affect prices in a quarterly BVAR. Aggregate demand, aggregate supply, labour supply and wage markup shock s are identified following Foroni et al. (2018). The impulse response functions (IRFs) show that global macroeconomic shocks have a persistently higher impact on wages (hourly earnings) and consumer prices than labour-specific shocks. Typical price and wage reactions have their maximum effects after about a year, underlining their rigidity to change. Counterfactual scenarios, in which wages do not react to shocks, reveal that such wage-price spirals can be significant after aggregate supply and demand shocks. Following a demand shock, wage reactions fuel price reactions in the medium term. Following a supply shock, wage reactions counterbalance price reactions over time.

In Section 2.3. (“Energy price inflation shocks in Lithuania and the Euro area”), we analyse how the energy price shocks affect economies in Lithuania and the Euro area. We estimate two separate BVAR models (one for Lithuania, the other for the EA), including respective time series from 2002Q1 to 2022Q4 of yoy energy, food, and core HICP inflation, as well as the unemployment rate and yoy total compensation per employee, following Corsello and Tagliabracci (2023). The IRFs show that Lithuania was more vulnerable to, and more affected by, energy price inflation shocks than the EA on average over the period. For an equivalent energy shock, the effects on HICP consumer price and wage inflation were larger and more persistent.

In Section 2.4. (“What has driven the surge in inflation in Lithuania? A production-side decomposition.”), using input-output tables, we decompose the inflation into its four drivers – prices of energy, prices of other imported products, wages and gross operating surplus. In our analysis, we focus on the period from 2021Q1 to 2023Q2 and find that all these supply-side factors contributed significantly to the increase in price level. We show that wage increases accounted for 40% of the calculated increase in price level, while the remaining increase was accounted for in broadly similar proportions by higher energy costs, more expensive imports of non-energy goods and services, and an increase in non-energy sector gross operating surplus (profit). The analysis also indicates that the recent increase in production costs has not yet been fully passed on to consumer prices in 2023Q2.

In Section 2.5. (“Lithuania’s nominal effective exchange rate fluctuations and domestic inflation.”), we analyse whether changes in nominal effective exchange rates have played a significant role in the recent surge of inflation. A relatively large share of Lithuanian imports is denominated in foreign currency, implying that inflation can be at least partially explained by currency depreciation. To determine the exchange rate pass-through to prices, a simple VAR analysis is conducted. The results of analysis indicate that exchange rate pass-through to import prices is incomplete in Lithuania, meaning that there is no tit-for-tat increase in import prices following currency depreciation. The pass-through for producer and consumer prices is even lower. Nominal exchange rate developments explain slightly more than 10% of import price variability, yet only about 1% of producer and consumer price variability. It follows that although the depreciation of the euro contributed to increasing inflation in the most recent inflation period (2021–2022) in Lithuania, its impact on producer and consumer prices was very limited.

In Section 2.6 (“A comparison of consumption basket item weights and price levels in Lithuania and the euro area”), we analyse if the differences in the composition of consumption baskets in Lithuania and euro area can explain a significant portion of inflation differentials. While gradually converging, the structure of the Lithuanian consumption basket still differs somewhat from that in the EA average. The greatest differences exist in the weights of services and food. In countries with a higher standard of living, households tend to spend less on basic needs and more on services. The same trends are observed in the development of the Lithuanian economy; as the standard of living approaches the EU average, the price level also converges, and services become a more prominent part of the consumption basket. Different weights of various goods and services in consumption baskets lead to different item weights for inflation calculation. Our calculations show that if in Lithuania we had HICP weights equal to those in the EA, our average annual headline inflation rate would have been about 1.6 percentage points lower than the factual in 2022.

In Section 2.7. (“Can price level convergence explain longer-term differences in inflation rates across euro area countries?”), following Honohan and Lane (2003), we provide evidence that in a monetary union, remaining price level differences lead to higher inflation in countries with lower price levels. For every single percentage point (pp) deviation below the average price level, countries experience around 0.02-0.036 pp higher inflation. In 2022, the price level in Lithuania was still 26 percent below the EU average. This would imply that the annual inflation in Lithuania could be about 0.5-0.9 pp higher than EA average due to the price level convergence in 2022.

The unexpectedly high inflation has affected government finances substantially. In Section 3 (“Implications of temporary acceleration in inflation for public finances”), we analyse the implications of higher inflation on the fiscal position of the general government sector. We break down recent general government revenue growth into four explanatory factors: real economic activity, price growth, the effect of government’s discretionary decisions (fiscal measures) and the unexplained component or tax residual. Our decompositions show that in 2021-2022, the observed increase in tax revenue was significantly affected by the strong growth of the macroeconomic bases and implemented fiscal measures. As regards the impact of inflation, more than half of the increase in receipts from VAT, personal income tax and social contributions can be attributed to the rise in the price component. As inflation decelerates, there will be a corresponding slowdown in the nominal GDP growth and the deceleration in goods and services inflation. This would naturally slow the growth of general government revenue.

All in all, this analysis implies that during the periods of a temporary increase in inflation fiscal policy should resist using the inflation-induced proceeds to finance permanent increases in spending. In the near future, inflation could show persistence and respond more slowly to changing trends in import and producer prices (compared to the upswing) since not all of the increased costs were fully passed through to consumer prices by the middle of 2023. In the longer term, inflation prospects in Lithuania will depend not only on economic policy but also future changes in the energy sector and climate-related developments and their impact, as well as the ability to adapt to those shifts.

Keywords: inflation, convergence, price level, supply shock

JEL Codes: C25, E61, G18, G21, G51

The views expressed are those of the author(s) and do not necessarily represent those of the Bank of Lithuania.

In this paper we examine multiple dimensions of ECB monetary policy communication by identifying its sentiment and relation with the economic environment and financial markets. We quantify communication sentiment using transcripts from official ECB communication events – press conferences, accounts and Executive Board speeches – as well as media reactions that highlight the key messages of those events. Importantly, we create unique lexicons for both of those communication types. We find that the overall trends in the sentiment indices for the analysed communication events closely resemble the movements of monetary policy stance as well as macroeconomic indicators in the euro area, both before and after the COVID-19 shock period. The communication tone generally shifts in advance of actual monetary policy actions. Using regression analysis, we find some expected, statistically significant effects of press confer-ence sentiment on bank stock prices (information-type shock) and identify the impact of Executive Board speeches on euro area fiscal borrowing costs (short-term OIS rates). Fragmentation issues among euro area member states do not seem to be negatively affected by the sentiments of the ECB’s communication. Still, policy makers should be aware that the tone of their communication events is likely to affect particular financial markets.

Keywords: ECB, monetary policy, communication, sentiment analysis, euro area, financial markets.

JEL Codes: C80, E43, E44, E58, G14.

The views expressed are those of the author(s) and do not necessarily represent those of the Bank of Lithuania.

Communication issues in central banking are important for maintaining the transparency of decisions and preparing financial markets for future changes in monetary policy. This study aims to determine what impact ECB monetary policy communication has on sovereign yields in the euro area on an intraday basis. We analyze different types of ECB monetary policy communication events: ECB monetary policy decisions, press conferences, accounts, and speeches made by Executive Board members. With the help of OLS and panel regression, we study how these communication events and control variables affect the intraday yield changes of major euro area sovereign and overnight index swap markets since 2014. The results from the baseline regression reveal that all four types of analyzed ECB monetary policy communication events have been affecting yields of the largest euro area sovereigns, with ECB decisions and press conferences showing the most substantial impact. Countries with the highest debt levels (such as Italy, Spain, and France) experienced the most robust changes in fiscal costs from ECB communication events, while the German bund market seems less affected. However, the period encompassing the economic shock induced by the Covid-19 pandemic shows much weaker effects, while Executive Board members who have been in charge since the start of the sample period of 2014 seem to have a much more substantial impact on euro area yields than more recent members. Sovereign yields bear the most decisive impact from media articles covering speeches’ topics of unconventional monetary policy measures and, to a smaller extent, interest rates and monetary policy targets.

JEL Codes: C80, E43, E44, E58, G12.

The views expressed are those of the author(s) and do not necessarily represent those of the Bank of Lithuania.

This paper examines changing ECB communication and how it has impacted euro area financial markets over the past two decades. We applied a combination of topic modelling and sentiment analysis for over 2000 public ECB Executive Board member speeches, as well as over 200 ECB press conferences. Topic analysis revealed that the ECB’s main focus has shifted from strategy and objectives, at the inception of the euro area, to various policy actions during the global financial crisis and, more recently, to instruments and economic developments. Sentiment analysis showed an expected trend of a more negative communication tone during periods of turmoil and a gradual shift to a more dovish monetary policy tone over time. Regression analysis revealed that sentiment indices had the expected impact on financial market indicators, while press conferences showed substantially stronger effects than speeches.

JEL Codes: C80, E43, E44, E58, G12.

The views expressed are those of the author(s) and do not necessarily represent those of the Bank of Lithuania.

Using two different methodologies, we estimate time-varying natural real rates of interest for a majority of euro area (EA) countries, including Lithuania. We find that natural real rates have been declining, particularly since 2008, albeit to different extent across EA countries. Lower rates could (at least partly) be explained by lower productivity and population growth. In line with previous literature, we find evidence of a substantial dispersion of the natural interest rate across EA economies. This became especially evident during the financial crisis of 2008-2009 and the sovereign debt crisis of 2010-2012, while estimates of natural rates tend to converge during "calm" periods. Estimates of natural rates for Lithuania were significantly above the estimates of core EA countries over 2002-2008, but this has changed after the crisis. From 2011 the estimates of natural rates for Lithuania tend to be close to the average for EA countries.

JEL Codes: C32, E32, E43, E52.

The views expressed are those of the author(s) and do not necessarily represent those of the Bank of Lithuania.

This paper examines the impact of the euro adoption on the economy of Lithuania during its first five years (2015-2019) as a member of the euro area. First, it assesses the impact of the euro adoption in Lithuania on interest rates and real exports, after which it investigates the impact on Lithuanian macroeconomic indicators with a LTDSGE model, using impulse response functions obtained in 2013 in research conducted by the Bank of Lithuania. The paper further estimates the impact of the euro adoption on Lithuanian macroeconomic indicators using the synthetic control method (SCM). The results of this paper confirm the main conclusions of the aforementioned 2013 study, namely, that the long-term benefits of the euro adoption were much higher than the costs, which were mainly short-term or could even be considered as valuable investments. 

JEL Codes: E17, E52, F33, F45

The views expressed are those of the author(s) and do not necessarily represent those of the Bank of Lithuania.

Available only in Lithuanian