| Abstract [eng] |
This dissertation addresses one of the central problems of modern macroeconomics: how to accurately measure and interpret consumers’ inflation perceptions and expectations and what role these expectations play in macroeconomic dynamics. The aim of the dissertation is to assess, from both a methodological and an empirical perspective, how qualitative consumer survey data can be reliably transformed into quantitative indicators of inflation expectations, and to examine how these indicators are related to actual inflation and other macroeconomic variables. The main methodological pillar of the dissertation is the Carlson–Parkin (1975) method, which remains the most widely used approach for transforming qualitative survey responses about the direction of price changes into quantitative measures of inflation expectations and perceptions. The dissertation shows that, although this method is attractive due to its simplicity and practical applicability, its results depend heavily on assumptions about the distribution of expectations, the choice of scaling parameters, and the aggregation procedure. The standard assumption of a normal distribution is not always appropriate, especially in the context of the Baltic countries, where the dynamics of consumer survey results exhibit greater volatility and asymmetry. Therefore, the dissertation systematically compares alternative specifications of the Carlson–Parkin method, including asymmetric and leptokurtic distributions and different choices of scaling parameters. The method is also compared with a simpler approach based on balance statistics. The empirical analysis is conducted at three levels. First, the Baltic countries—Lithuania, Latvia, and Estonia—are examined in detail. These economies are chosen as a particularly informative case for assessing the reliability of the Carlson–Parkin method, as they are characterized by historically higher inflation and inflation volatility, as well as by a weaker correspondence between consumer survey indicators and actual inflation than in aggregated euro area data. The results show that although alternatives to normal distribution can in some cases slightly improve measurement accuracy, these improvements are generally limited. In contrast, the choice of the scaling parameter has a much stronger impact on the final quantitative estimates and on the economic conclusions drawn from them. An analysis of the predictive power of quantified expectations confirms that in most configurations they only rarely lead to a statistically significant improvement in inflation forecasting accuracy compared to simple balance statistics. More substantial improvements are observed only in certain specific periods, in particular after the adoption of the euro in Lithuania and Latvia and during the recent episode of high inflation in the euro area. In the second stage, the analysis is extended to 26 European Union countries using a panel vector autoregression model. This approach makes it possible to assess the dynamic interaction between inflation expectations and key macroeconomic variables, such as inflation, consumption, GDP, unemployment, interest rates, wages, credit, deposits, energy prices, and consumer confidence indicators. One of the main results of the dissertation is that model results about the dynamics between inflation expectations and other macroeconomic variables are sensitive to the chosen specification of the Carlson–Parkin method. The results based on quantified expectations are also compared with those obtained using balance statistics, and differences are found there as well. This implies that technical quantification choices can shape the conclusions of empirical studies regarding the relationship between expectations and other macroeconomic indicators. The analysis also highlights the presence of cross-country heterogeneity, which may be obscured when using aggregated European Union data. The third part of the dissertation is devoted to socio-demographic heterogeneity. Using data from Lithuania and the euro area, inflation perceptions and expectations are analysed across groups defined by age, income, education, and gender. Instead of the usual practice of applying quantification to already aggregated data, the dissertation proposes a new two-step approach: first, the Carlson–Parkin method is applied separately to individual subgroups, and only then are the results aggregated using population weights. It is shown that this approach consistently improves the accuracy of quantified perceptions and expectations, especially in the euro area data, where the largest improvements are observed when disaggregating by education and gender. This indicates that heterogeneity is not merely a descriptive feature of consumer expectations, but a dimension that can be directly integrated into the quantification process itself. The dissertation contributes to the literature in three main ways. First, it provides a critical assessment of the Carlson–Parkin method and its alternatives, highlighting both the possibilities and the limitations of methodological refinements. Second, it expands the empirical evidence for the Baltic countries and offers a detailed analysis of inflation expectations in these economies. Third, it proposes a subgroup-based approach to the quantification of inflation expectations, which allows for a consistent improvement in measurement accuracy. |
