Figure 1: The ECCS respondents per country
Note: This figure shows the monthly
average number of respondents in the European Commission Consumers
Survey (ECCS) in the unfiltered dataset (dashed lines) and in filtered
dataset (solid lines) where consumers answered both inflation-related
questions (Q5.1 and Q6.1). The figure shows results for Germany, France,
Italy, Spain, Belgium, Austria, Finland, and Greece. The sample period
is from January 2004 to December 2024. The frequency is monthly.
Source: European Commission and authors’
calculations.
Note: Figure 1 plots the bar chart that indicates the average number of respondents per month in the European Comission Consumer Survey in the unfiltered data set (indicated by dashed lines) and in the filtered data set, where respondents answered both inflation perception and inflation expectation questions in the survey, questions 5.1 and 6.1, respectively, (indicated by solid bars). Each bar represents the monthly country average for Germany, France, italy, Belgium, Austria,Finland, and Greece. Spain has the highest monthly average of unfiltered responses, 2013. Germany has the highest monthly average of filtered responses, 1608. Finland has the lowest monthly average number of unfiltered responses, 1310. France has the lowest monthy average number of filtered responses, 916. The sample period is from January 2004 to December 2024. The frequency is monthly. Source: European Commission and authors' calculations.
Figure 2: The ECCS distribution of inflation perception responses
Note: This figure shows the histogram of
the ECCS inflation perception responses. The histogram shows percentage
of the responses in the filtered dataset for \(<\)0, [0,1[, [1,2[, [2,3[, [3,4[, [4,5[,
[5,6[, and 6+ annual perceived inflation rate brackets (in percentage)
in the 12 months before the response month. The bucket includes all
negative responses, and the bucket combines responses that are larger or
equal to 6 percent. The results are shown for Germany, France, Italy,
Spain, Belgium, Austria, Finland, and Greece. The sample period is from
January 2004 to December 2024. The frequency is monthly.
Source: European Commission and authors’
calculations.
Note: Figure 2 shows the histogram of the responses by the European Commission Consumer Survey respondents in the filtered dataset for the perceived inflation rate in the previous 12 months relative to the response month. The histogram presents the results in the following brackets for the perceived inflation rate: less than 0 percent, between 0 and 1 percent, between 1 and 2 percent, between 2 and 3 percent, between 3 and 4 percent, between 4 and 5 percent, between 5 and 6 percent, and higher than 6 percent. The figure has 8 panels. The top two panels present the histograms for Germany and France. For Germany, about 20 percent of responses fall into the bracket of between 0 and 1 percent perceived inflation rate nad about 30 percent of responses fall into the bracket of 6 percent perceived inflation rate and more, with the rest of the responses distributed more or less uniformly throughout the remaining brackets. For France, about 20 percent of responses fall into the bracket of between 0 and 1 percent perceived inflation rate and about 40 percent of responses fall into the bracket of 6 percent perceived inflation rate and more, with the rest of the responses distributed more or less uniformly throughout the remaining brackets. The second row of panels present the histograms for Italy and Spain. For Italy, about 35 percent of responses fall into the bracket of between 0 and 1 percent perceived inflation rate, and about 50 percent of responses fall into the bracket of 6 percent perceived inflation rate and more, with the rest of responses distributed more or less uniformly throughout the remaining brackets. For Spain, the responses are concentrated in two brackets: about 20 percent of responses fall into the bracket of between 0 and 1 percent perceived inflation rate, and about 55 percent of responses fall into the bracket of 6 percent perceived inflation rate and more, with the rest of responses distributed more or less uniformly throughout the remaining brackets. The third row of panels present the histograms for Belgium and Austria. For Belgium, about 50 percent of responses fall into the bracket of 6 percent perceived inflation rate and more, with the rest of responses distributed more or less uniformly throughout the remaining brackets. For Austria, about 40 percent of responses fall into the bracket of 6 percent perceived inflation rate and more, with the rest of responses distributed more or less uniformly throughout the remaining brackets. The fourth row of panels present the histograms for Finland and Greece. For Finland, about 25 percent of responses fall into the bracket of between 0 and 1 percent perceived inflation rate, with the rest of responses distributed more or less uniformly throughout the remaining brackets. For Greece, about 20 percent of responses fall into the bracket of between 0 and 1 percent perceived inflation rate and about 65 percent of responses fall into the bracket of 6 percent perceived inflation rate and more, with the rest of responses distributed more or less uniformly throughout the remaining brackets. The sample period is from January 2004 to December 2024. The frequency is monthly. Source: European Commission and authors' calculations.
Figure 3: The ECCS distribution of one-year inflation expectation responses
Note: This figure shows the histogram of
the ECCS inflation expectations responses. The histogram shows
percentage of the responses in the filtered dataset for \(<\)0, [0,1[, [1,2[, [2,3[, [3,4[, [4,5[,
and 6+ annual perceived inflation rate brackets (in percentage). The
bucket includes all negative responses, and the bucket combines
responses that are larger or equal to 6 percent. The results are shown
for Germany, France, Italy, Spain, Belgium, Austria, Finland, and
Greece. The sample period is from January 2004 to December 2024. The
frequency is monthly.
Source: European Commission and authors’
calculations.
Note: Figure 3 shows the histogram of the responses by the European Commission Consumer Survey respondents in the filtered dataset for the expected inflation rate over the 12 months following the response month. The histogram presents the results in the following brackets for the expected inflation rate: less than 0 percent, between 0 and 1 percent, between 1 and 2 percent, between 2 and 3 percent, between 3 and 4 percent, between 4 and 5 percent, between 5 and 6 percent, and higher than 6 percent. The figure has 8 panels. The first row of panels presents the histograms for Germany and France. For Germany, about 30 percent of responses fall into the bracket of between 0 and 1 percent expected inflation rate and about 20 percent of responses fall into the bracket of 6 percent expected inflation rate and more, with the rest of the responses distributed more or less uniformly throughout the remaining brackets. For France, about 35 percent of responses fall into the bracket of between 0 and 1 percent expected inflation rate and about 20 percent of responses fall into the bracket of 6 percent expected inflation rate and more, with the rest of the responses distributed more or less uniformly throughout the remaining brackets. The second row of panels presents the histograms for Italy and Spain. For Italy, about 60 percent of responses fall into the bracket of between 0 and 1 percent expected inflation rate, and about 30 percent of responses fall into the bracket of 6 percent expected inflation rate and more, with the rest of responses distributed more or less uniformly throughout the remaining brackets. For Spain, about 35 percent of responses fall into the bracket of between 0 and 1 percent expected inflation rate, and about 40 percent of responses fall into the bracket of 6 percent expected inflation rate and more, with the rest of responses distributed more or less uniformly throughout the remaining brackets. The third row of panels presents the histograms for Belgium and Austria. For Belgium, about 30 percent of responses fall into the brackets of between 0 and 1 percent expected inflation rate and about 30 percent of responses fall into the bracket of 6 percent expected inflation rate and more, with the rest of responses distributed more or less uniformly throughout the remaining brackets. For Austria, about 20 percent of responses fall into the bracket of between 0 and 1 percent expected inflation rate and about 35 percent of responses fall into the bracket of 6 percent expected inflation rate and more, with the rest of responses distributed more or less uniformly throughout the remaining brackets. The fourth row of panels presents the histograms for Finland and Greece. For Finland, about 25 percent of responses fall into the bracket of between 0 and 1 percent expected inflation rate, with the rest of responses distributed more or less uniformly throughout the remaining brackets. For Greece, about 35 percent of responses fall into the bracket of between 0 and 1 percent expected inflation rate and about 55 percent of responses fall into the bracket of 6 percent perceived inflation rate and more, with the rest of responses distributed more or less uniformly throughout the remaining brackets. The sample period is from January 2004 to December 2024. The frequency is monthly. Source: European Commission and authors' calculations.
Figure 4: The median perceptions, expectations and realized inflation
Note: The figure shows the time series of
the European Commission Consumer Survey (ECCS) inflation perceptions and
expectations for median consumers for Germany, France, Italy, Spain,
Belgium, Austria, Finland, and Greece. It also plots on each chart the
country-specific HICP inflation series. Realized, perceived and expected
inflation series are plotted by matching the twelve-month forecast
horizon. The black dashed line represents zero percent. The sample
period is from January 2004 to December 2024. The frequency is monthly.
Shaded areas represent the Organization for Economic Co-operation and
Development (OECD) recessions.
Source: European Commission, Eurostat, and
authors’ calculations.
Note: Each panel on Figure 4 plots the country-specific time series of the European Commission Consumer Survey median consumer inflation perceptions, median consumer inflation expectations, the country-specific Harmonized Index of Consumer Prices (HICP) series. The 8 panels that correspond to 8 selected countries. The vertical gray bars on each panel represent the recession periods in the euro area as defined by the Organization for Economic Cooperation and Development. The first row of panels presents the time series for Germany and France. The measures of inflation perceptions, expectations, and realized inflation hover around 2 percent with the exception of the period around the Global Financial Crisis (GFC) in 2008 and the COVID-19 pandemic in 2020-2022, where inflation series reached levels of around 10 percent. The second row of panels presents the time series for Italy and Spain. The measures of inflation varied substantially in these countries, especially around the GFC period in 2008, around the sovereign bond crisis period in 2011-2012, and around the COVID-19 pandemic in 2020-2022. The orange lines represent the inflation perception series that reached levels of around 20 percent during the GFC period and the COVID-19 pandemic. The blue lines represent the inflation expectations series that reached levels around 15 percent during the COVID-19 pandemic in Italy, and around 15 percent during the GFC and the COVID-19 periods in Spain. The third row of panels presents the time series for Belgium and Austria. The measures of inflation varied somewhat in these countries. The orange lines represent the inflation perception series that reached levels of around 10 percent during the GFC period and around 15 percent during the COVID-19 pandemic. The blue lines represent the inflation expectations series that hovered around 5 percent but reached levels of around 10 percent during the COVID-19 pandemic in Belgium and Austria. The fourth row of panels presents the time series for Finland and Greece. The measures of inflation were relatively stable in Finland, with soem upticks to around 5 percent during the GFC periods and an increased in perceptions just above 10 percent during the COVID-19 pandemic. In the case of Greece, the inflation perceptions and inflation expectations series, the orange and the blue lines show substantial variation amid some periods of staleness. The inflation perceptions series varied between 10 percent and 20 percent during the GFC period, hovered around 10 percent during the sovereign bond crisis, varied between 5 percent and 10 percent between 2014 and 2020, and reached levels of around 30 percent during the COVID-19 pandemic. The inflation expectations series showed similar variation, except during the COVID-19 pandemic when inflation expectations series reached levels of about 20 percent. The sample period is from January 2004 to December 2024. The frequency is monthly. Source: European Commission, Eurostat, and authors' calculations.
Figure 5: Persistence of median forecast errors
Note: The figure shows results for Eq. (7) where
the forecast errors of a median consumer are regressed on their lag
\(\delta = 1,\ldots, 12\). The purple
dots indicate the estimated coefficient for a particular lag together
with the 95-percent confidence interval. The red line indicates the
\(R^2\) coefficient of a regression
that corresponds to a particular lag. The black dashed line represents
zero percent. The sample period is from January 2004 to December 2024.
The frequency is monthly.
Source: European Commission, Eurostat, and
authors’ calculations.
Note: Each panel on Figure 5 plots the results of the regression in which the inflation forecast errors -- defined as the difference between a median consumer's expected inflation and realized inflation -- are regressed on its own lags from 1 to 12. The x-axis corresponds to the lag and the y-axis corresponds to the estimated coefficient and to the R-square. The purple solid circle indicates the estimated regression coefficient. The vertical black interval for each estimated coefficient indicates the 95-percent confidence interval. The red line plots the corresponding R-square coefficient as a function of the lag. For all countries, the forecast errors, overall, are persistent as indicated by the statistically significant estimated regression coefficients. The first row of panels reports the results for Germany and France. For both countries, the estimated regression coefficients declines with the lag. The estimated coefficients are diminishing with the lag and are statistically significant at the 5-percent level for lags from 1 to 10. For Germany, the R-square coefficients decline from about 90 percent to zero and for France, the R-square coefficients decline from about 75 percent to zero. The second row of panels reports the results for Italy and Spain. For Italy, the estimated regression coefficents are diminishing with the lag and are statistically significant up to the lag 8. The R-square coefficients decline from about 75 percent to zero. For Spain, the estimated regression coefficents are diminishing with the lag and are statistically significant for all lags. The R-square coefficients decline from about 80 percent to zero. The third row of panels reports the results for Belgium and Austria. For Belgium, the estimated regression coefficents are diminishing with the lag, reversing their sign at the lag 10. The coefficients are statistically significant up to the lag 8 and also for lags 11 and 12. The R-square coefficients decline from about 90 percent to zero and then inch up for lags 11 and 12. For Austria, the estimated coefficients are diminishing with the lag and statistically significant up to the lag 10. The R-square coefficients decline from about 90 percent to zero. The fourth row of panels reports the results for Finland and Greece. For Finland, the estimated coefficients are diminishing with the lag and statistically significant up to the lag 10. The R-square coefficients decline from about 90 percent to zero. For Greece, the estimated coefficients are diminishing with the lag and are significant for all lags. The R-square coefficients decline from about 80 percent to to about 10 percent. The sample period is from January 2004 to December 2024. The frequency is monthly. Source: European Commission, Eurostat, and authors' calculations.
Figure C1: Headline and core inflation rates
Notes: This figure shows the Harmonized
Index of Consumer Prices (HICP) (headline and core) for the euro area
from January 2004 to December 2024, and the country-specific HICPs for
Germany, France, Italy, Spain, Belgium, Austria, Finland, and Greece,
from January 2004 to December 2024. The frequency is monthly. Shaded
areas represent the Organisation for Economic Co-operation and
Development (OECD) recessions. Source: Eurostat.
Note: The top panel on Figure C1 presents the time series of the year-on-year headline and core Harmonized Index of Consumer Prices (HICP) inflation. On this and futher panels in this figure the headline and core inflation series are shown in the blue and the red lines, respectively. The measures of inflation hovered around 2 percent during for the most of the sample period with the exception of the period around the Global Financial Crisis (GFC) in 2008 when the headline inflation measure jumped to around 4 percent and of the COVID-19 pandemic in 2020-2022 when both measures increased. During this period, the headline inflation jumpted to about 10 percent, while the core inflation increased to a bit lower level, 7.5 percent. The second row of panels presents country-specific time series of headline and core HICP inflation for Germany on the left and for France on the right. The third row of panels presents country-specific time series of headline and core HICP inflation for Italy on the left and for Spain on the right. The fourth row of panels presents country-specific time series of headline and core HICP inflation for Belgium on the left and for Austria on the right. The fifth row of panels presents country-specific time series of headline and core HICP inflation for Finland on the left and for Greece on the right. Overall, the country-specific measures of inflation hovered around 2 percent during for the most of the sample period with the exception of the period around the GFC in 2008 and of the COVID-19 pandemic in 2020-2022 when both measures increased. On balance, the headline measures of inflation increased by more than the core measures of inflation. The vertical gray bars on each panel represent the recession periods in the euro area as defined by the Organization for Economic Cooperation and Development. The sample period is from January 2004 to December 2024. The frequency is monthly. The sample period is from January 2004 to December 2024. The frequency is monthly. Source: Eurostat.
Figure C2: Energy and food inflation rates
Notes: This figure shows the Harmonized
Index of Consumer Prices (HICP) for food and energy items for the euro
area from January 2004 to December 2024, and the country-specific HICPs
for Germany, France, Italy, Spain, Belgium, Austria, Finland, and
Greece, from January 2004 to December 2024. The frequency is monthly.
Source: Eurostat.
Note: The top panel on Figure C2 presents the time series of the year-on-year food and energy Harmonized Index of Consumer Prices (HICP) inflation. On this and futher panels in this figure food and energy inflation series are shown in the blue and the red lines, respectively. These measures of inflation varied substantially during for the most of the sample period with the exception of the period around the Global Financial Crisis (GFC) in 2008 when the food and energy inflation measures reached around 8 and 5 percent, respectively, and of the COVID-19 pandemic in 2020-2022 when these measures surged to around 15 percent. In addition, the measure of energy inflation briefly turned negative three times in our sample: post the Global Financial Crisis (GFC) period, around 2016, and in 2020. The second row of panels presents country-specific time series of food and energy HICP inflation for Germany on the left and for France on the right. The third row of panels presents country-specific time series of food and energy HICP inflation for Italy on the left and for Spain on the right. The fourth row of panels presents country-specific time series of food and energy HICP inflation for Belgium on the left and for Austria on the right. The fifth row of panels presents country-specific time series of food and energy HICP inflation for Finland on the left and for Greece on the right. Overall, the country-specific measures of food and energy inflation varied somewhat during for the most of the sample period and increased materially during the GFC in 2008 and the COVID-19 pandemic in 2020-2022. The vertical gray bars on each panel represent the recession periods in the euro area as defined by the Organization for Economic Cooperation and Development. The sample period is from January 2004 to December 2024. The frequency is monthly. The sample period is from January 2004 to December 2024. The frequency is monthly. Source: Eurostat.
Figure D1: Inflation perceptions: box plot
Note: The figure shows the box plot of
consumers inflation perceptions by cohorts and countries. The rectangle
represents the interquartile range (P25 for the lower bar and P75 for
the upper bar), the middle bar is the median and the red cross
corresponds to the mean. The sample period is from January 2004 to
December 2024. The frequency is monthly.
Source: European Commission, and authors’
calculations.
Figure D1 presents the box plots of consumers' inflation perceptions across demographic and socioeconomic groups in eight euro-area countries (Germany, France, Italy, Spain, Belgium, Austria, Finland, and Greece). For each country, the figure reports the distribution of perceived inflation by income quartiles, three education levels (primary, secondary, and post-secondary), age groups (16 to 29, 30 to 49, 50 to 64, and over 65 years old), and gender (male or female). The boxes represent the interquartile range defined as the 25th-75th percentile range, the horizontal line indicates the median, and the red cross denotes the mean. Several patterns emerge. First, substantial heterogeneity in inflation perceptions exists across countries. Greece, Spain, and Italy display the highest levels and widest dispersion of inflation perceptions, whereas Finland exhibits consistently lower and more concentrated perceptions. Germany, France, Belgium, and Austria occupy an intermediate position. Second, income-related differences are generally modest. Within most countries, median inflation perceptions vary little across income quartiles, although lower-income households tend to report slightly higher inflation perceptions and greater dispersion than higher-income households. This pattern is most visible in Greece, Spain, and Italy. Third, educational attainment is negatively associated with inflation perceptions in several countries. Individuals with primary education tend to report higher perceived inflation than those with secondary or post-secondary education. The gradient is particularly evident in Spain, Greece, Belgium, and Austria, while it is less pronounced in Finland and Germany. Fourth, age differences appear limited. Inflation perceptions are broadly similar across age groups within each country, although younger respondents often exhibit slightly higher dispersion. No systematic monotonic relationship between age and inflation perceptions is observed. Finally, gender differences are more pronounced than age differences. In all countries, women report higher median and mean inflation perceptions than men. The gender gap is particularly noticeable in Spain, Greece, Belgium, and Italy, while remaining present but smaller in Germany, France, Austria, and Finland. Overall, the figure suggests that inflation perceptions differ more strongly across countries than across demographic groups within countries. Among individual characteristics, education and gender are associated with the most systematic differences, whereas income and age play a comparatively smaller role. The sample period is from January 2004 to December 2024. The frequency is monthly. Source: European Commission and authors' calculations.
Figure D2: Inflation expectations: box plot
Note: The figure shows the box plot of
consumers inflation expectations by cohorts and countries. The rectangle
represents the interquartile range (P25 for the lower bar and P75 for
the upper bar), the middle bar is the median and the red cross
corresponds to the mean. The sample period is from January 2004 to
December 2024. The frequency is monthly.
Source: European Commission, and authors’
calculations.
Figure D2 presents box plots of consumers' inflation expectations across demographic and socioeconomic groups in eight euro-area countries (Germany, France, Italy, Spain, Belgium, Austria, Finland, and Greece). each country, the figure reports the distribution of expected inflation by income quartiles, three education levels (primary, secondary, and post-secondary), age groups (the 16- to 29-year-old group, 30- to 49-year-old group, 50- to 64-year-old group, and over 65 years old), and gender (male or female). The boxes represent the interquartile range defined as the 25th-75th percentile range, the horizontal line indicates the median, and the red cross denotes the mean. Several broad patterns emerge. First, as with inflation perceptions depicted in Figure D1, there is substantial cross-country heterogeneity in inflation expectations. Greece consistently exhibits the highest inflation expectations and the largest dispersion, followed by Spain and, to a lesser extent, Italy. In contrast, Finland displays the lowest inflation expectations and the narrowest distributions. Germany, France, Belgium, and Austria generally lie between these two extremes. Second, inflation expectations tend to decline with income in most countries. Lower-income households report higher expected inflation than higher-income households with the pattern being particularly visible in Germany, France, Belgium, Austria, and Finland. The income gradient is somewhat weaker in Italy and Spain, although lower-income groups still tend to exhibit higher means and wider distributions. Third, educational attainment is negatively associated with inflation expectations. Respondents with primary education generally report higher inflation expectations than those with post-secondary education. This relationship is most apparent in Spain, Belgium, Austria, and Greece, where both the median and the dispersion of expectations decline with education. Finland again shows comparatively small differences across educational groups. Fourth, age-related differences are modest. Inflation expectations are relatively similar across age cohorts within most countries, and no clear monotonic relationship between age and expected inflation emerges. Some countries, such as Greece and Spain, exhibit slightly greater dispersion among older respondents, but the differences in central tendency remain limited. Finally, a pronounced gender gap is evident across all countries. Women consistently report higher inflation expectations than men, both in terms of median and mean values. The gap is especially large in Greece, Spain, Belgium, and Italy, while remaining visible in Germany, France, Austria, and Finland. In addition, female respondents often exhibit greater dispersion in inflation expectations, suggesting more heterogeneity within this group. Overall, Figure D2 indicates that inflation expectations vary considerably across countries and demographic groups. Compared with age, income, and education, gender appears to be one of the most important correlates of inflation expectations. The results also suggest that lower-income and less-educated individuals tend to anticipate higher future inflation, while cross-country differences remain substantially larger than within-country demographic differences. The sample period is from January 2004 to December 2024. The frequency is monthly. Source: European Commission and authors' calculations.
Figure D3: Backward-looking component of the ECCS inflation perceptions and
expectations, median consumer, income
Note: This figure shows regression
results of Eqs. (2)-(5)
estimated by income cohorts and countries using the European Commission
Consumer Survey (ECCS) filtered data set based on median consumers. The
four income cohorts are: first quartile included (assuming lowest
income), second quartile included, third quartile included and fourth
quartile included (assuming highest income). Panel A corresponds to the
results of the changes in inflation perceptions on the changes in
realized inflation in the previous month; Panel B the changes in
inflation expectations on the changes in realized inflation in the
previous month; Panel C the changes in inflation expectations on the
changes in inflation perceptions; and Panel D the changes of inflation
expectations on both the changes in inflation perceptions and the
changes in realized inflation in the previous month. Realized inflation
is the country-specific Harmonized Index of Consumer Prices (HICP) for
all items. The vertical bars correspond to the 95 percent confidence
interval. The sample period is from January 2004 to December 2024. The
frequency is monthly.
Source: European Commission, Eurostat and authors’
calculations.
The first row of four panels in Figure D3 reports the estimated coefficients of the changes in inflation perceptions regressed on the changes in Harmonized Index of Consumer Prices (HICP) headline inflation lagged one month. In this and further rows, from the left to the right, each panel corresponds to the results based on the first quartile that represents the lowest-income households, the second quartile, the third quartile, and the fourth quartile that represents the highest-income households in eight euro-area countries (Germany, France, Italy, Spain, Belgium, Austria, Finland, and Greece). The blue dots represent the estimated coefficients associated with the realized inflation independent variable and the orange dots represent the estimated coefficients associated with the inflation perceptions indepdent variable. Vertical bars around the dots represent 95 percent confidence intervals. As the first row shows, inflation perceptions are generally positively correlated with realized inflation across countries and income groups. The coefficients are positive in most cases, although the magnitude varies considerably across countries. Overall, the strongest correlations are observed in France and Germany whereas the relationship appears weaker and less precisely estimated in Spain and Italy. Importantly, there is no clear monotonic pattern across income quartiles, suggesting that sensitivity of perceptions to realized inflation does not systematically increase or decrease with income. The second row of panels in Figure D3 reports the estimated coefficients of the changes in inflation expectations regressed on the changes in headline inflation lagged one month. The estimated coefficients are again predominantly positive, implying that recent inflation developments influence consumers' views about future inflation. The relationship is particularly strong in Germany, France, and Italy and remains positive in most other countries. Compared with Panel A, the estimates are generally smaller and often accompanied by wider confidence intervals, indicating greater uncertainty surrounding the transmission of realized inflation into expectations. The third row of panels in Figure D3 reports the estimated coefficients of the changes in inflation expectations regressed on the changes in inflation perceptions lagged one month. The coefficients are positive and statistically meaningful across all countries and income groups, providing evidence that consumers rely on their perceptions of current inflation when forming expectations about future inflation. Compared with the previous panels, the estimates are more homogeneous across countries and income cohorts. This result points to a robust backward-looking channel linking perceptions and expectations. The fourth row of panels in Figure D3 reports the estimated coefficients of the changes in inflation expectations regressed on both the changes in headline inflation and changes in inflation perceptions lagged one month. In most countries, inflation perceptions retain a positive and economically significant effect on inflation expectations, while the coefficient on realized inflation becomes smaller and, in several cases, statistically indistinguishable from zero. This suggests that much of the influence of realized inflation on expectations operates indirectly through consumers' perceptions rather than through direct observation of inflation outcomes. Overall, Figure D3 provides evidence that households' inflation expectations are shaped to a larger extent by their perceptions about recent price changes than about realized inflation. The strength of this mechanism varies more across countries than across income groups, suggesting that national factors play a larger role than income heterogeneity in determining how consumers process inflation information. The sample period is from January 2004 to December 2024. The frequency is monthly. Source: European Commission, Eurostat, and authors' calculations.
Figure D4: Backward-looking component of the ECCS inflation perceptions and
expectations, median consumer, education
Note: This figure shows regression
results from Eqs. (2)-(5)
estimated by education cohorts and countries using the European
Commission Consumer Survey (ECCS) filtered data set based on median
consumers. The three education cohorts are: primary, secondary and
post-secondary. Panel A corresponds to the results of the changes in
inflation perceptions on the changes in realized inflation in the
previous month; Panel B the changes in inflation expectations on the
changes in realized inflation in the previous month; Panel C the changes
in inflation expectations on the changes in inflation perceptions; and
Panel D the changes of inflation expectations on both the changes in
inflation perceptions and the changes in realized inflation in the
previous month. Realized inflation is the country-specific Harmonized
Index of Consumer Prices (HICP) for all items. The vertical bars
correspond to the 95 percent confidence interval. The sample period is
from January 2004 to December 2024. The frequency is monthly.
Source: European Commission, Eurostat and authors’
calculations.
The first row of four panels in Figure D4 reports the estimated coefficients of the changes in inflation perceptions regressed on the changes in Harmonized Index of Consumer Prices (HICP) headline inflation lagged one month. In this and further rows, from the left to the right, each panel corresponds to the results based on the education levels of households, namely, the primary education level, the secondary education level, and the post-secondary education level in eight euro-area countries (Germany, France, Italy, Spain, Belgium, Austria, Finland, and Greece). The blue dots represent the estimated coefficients associated with the realized inflation independent variable and the orange dots represent the estimated coefficients associated with the inflation perceptions indepdent variable. Vertical bars around the dots represent 95 percent confidence intervals. As the first row shows, inflation perceptions are positively correlated with the changes in realized inflation across all educational groups, with the magnitude of such correlation varying considerably across countries. The largest coefficients are observed in Greece, France, and Spain, whereas Italy tends to exhibit weaker and less precisely estimated coefficients. Differences across educational groups are relatively limited, suggesting that the transmission from realized inflation to perceptions is broadly similar regardless of educational attainment. The second row of panels in Figure D4 reports the estimated coefficients of the changes in inflation expectations regressed on the changes in headline inflation lagged one month. The estimated coefficients are predominantly positive for all educational categories, indicating that recent inflation developments contribute to the formation of future inflation expectations. As in Figure D3, Greece exhibits the strongest response, while Finland and Austria generally display more moderate effects. The confidence intervals are often wider than in Panel A, reflecting greater uncertainty in the relationship between realized inflation and expectations. No systematic pattern emerges across educational levels, although respondents with post-secondary education appear to display slightly more stable responses across countries. The third row of panels in Figure D4 reports the estimated coefficients of the changes in inflation expectations regressed on the changes in inflation perceptions lagged one month. The coefficients are almost uniformly positive and statistically significant across virtually all countries and education groups. This finding confirms the existence of a strong link between consumers' perceptions of current inflation and their anticipated future inflation. Compared with Panels A and B, the estimates are relatively homogeneous across educational categories. Although consumers with primary education often exhibit somewhat larger coefficients, the differences across education groups are generally modest. The fourth row of panels in Figure D4 reports the estimated coefficients of the changes in inflation expectations regressed on both the changes in headline inflation and changes in inflation perceptions lagged one month. The results indicate that inflation perceptions remain a significant determinant of inflation expectations across all educational groups and countries. In contrast, the direct effect of realized inflation becomes substantially smaller and frequently loses statistical significance once perceptions are taken into account. This pattern suggests that consumers primarily incorporate recent inflation developments into their expectations through the intermediate channel of inflation perceptions. Comparing educational groups, the overall structure of the results is remarkably stable. Consumers with primary education tend to exhibit slightly larger coefficients linking perceptions to expectations, consistent with a somewhat stronger backward-looking component in expectation formation. However, the differences across educational attainment are relatively small compared with the variation observed across countries. Greece consistently displays the strongest backward-looking behavior, while Finland generally exhibits the weakest responses. Germany, France, Belgium, Austria, Spain, and Italy fall between these two extremes. Overall, Figure D4 provides evidence that the backward-looking mechanism underlying inflation expectation formation operates similarly across educational groups. The cross-country heterogeneity in this process appears substantially larger than the heterogeneity associated with educational level of the peer group, suggesting that national factors play a more important role than education in shaping the transmission of inflation experiences into inflation expectations. The sample period is from January 2004 to December 2024. The frequency is monthly. Source: European Commission, Eurostat, and authors' calculations.
Figure D5: Backward-looking component of the ECCS inflation perceptions and
expectations, median consumer, age
Note: This figure shows regression
results from Eqs. (2)-(5)
estimated by age cohorts and countries using the European Commission
Consumer Survey (ECCS) filtered data set based on median consumers. The
four age cohorts are: 16-29y, 30-49y, 50-64y and more than 65y. Panel A
corresponds to the results of the changes in inflation perceptions on
the changes in realized inflation in the previous month; Panel B the
changes in inflation expectations on the changes in realized inflation
in the previous month; Panel C the changes in inflation expectations on
the changes in inflation perceptions; and Panel D the changes of
inflation expectations on both the changes in inflation perceptions and
the changes in realized inflation in the previous month. Realized
inflation is the country-specific Harmonized Index of Consumer Prices
(HICP) for all items. The vertical bars correspond to the 95 percent
confidence interval. The sample period is from January 2004 to December
2024. The frequency is monthly.
Source: European Commission, Eurostat and authors’
calculations.
The first row of four panels in Figure D5 reports the estimated coefficients of the changes in inflation perceptions regressed on the changes in Harmonized Index of Consumer Prices (HICP) headline inflation lagged one month. In this and further rows, from the left to the right, each panel corresponds to the results based on the households' age, namely, the 16 to 29 year-old group, the 30 to 49-year-old group, the 50 to 64-year-old group, and the older than 65 year-old group, in eight euro-area countries (Germany, France, Italy, Spain, Belgium, Austria, Finland, and Greece). The blue dots represent the estimated coefficients associated with the realized inflation independent variable and the orange dots represent the estimated coefficients associated with the inflation perceptions indepdent variable. Vertical bars around the dots represent 95 percent confidence intervals. As the first row shows, inflation perceptions are positively correlated with the changes in realized inflation across all age groups, with the magnitude of such correlation varying considerably across countries. Greece and France tend to display relatively strong correlations, while Germany, Belgium and Austria have lower correlations. The overall relationship between realized inflation and inflation perceptions remains broadly similar across age groups, with no clear evidence that younger or older consumers consistently react more strongly to realized inflation. The second row of panels in Figure D5 reports the estimated coefficients of the changes in inflation expectations regressed on the changes in headline inflation lagged one month. Except for Italy, the coefficients are generally positive, indicating that recent inflation developments influence expectations about future inflation. The estimated effects are highest for France and Greece, and remain positive for most countries and age cohorts. While some variation exists across age groups, the differences are limited relative to the substantial heterogeneity observed across countries. The third row of panels in Figure D5 reports the estimated coefficients of the changes in inflation expectations regressed on the changes in inflation perceptions lagged one month. The coefficients are positive and statistically significant across all age categories, confirming that consumers rely on their preceived past inflation when forming expectations about future inflation. Compared with Panels A and B, the coefficients display less cross-country variation and are remarkably stable across age groups. Younger respondents (the 16 to 29-year-old group) appear to exhibit slightly larger coefficients in several countries, suggesting a somewhat stronger link between perceptions and expectations, but the differences are generally modest. The fourth row of panels in Figure D5 reports the estimated coefficients of the changes in inflation expectations regressed on both the changes in headline inflation and changes in inflation perceptions lagged one month. The results indicate that inflation perceptions retain a positive and economically important effect across all age groups and countries. By contrast, the direct effect of realized inflation becomes considerably smaller once perceptions are controlled for and often loses statistical significance. This finding reinforces the view that realized inflation affects expectations primarily through consumers' perceptions rather than through a direct channel. Comparing age cohorts, the overall structure of the results is highly consistent. The youngest group exhibits somewhat greater variability across countries and occasionally larger coefficients, but there is no systematic age gradient in the backward-looking component of inflation expectations. The coefficients for the other groups are remarkably similar, suggesting that the process through which consumers translate inflation experiences into expectations does not vary substantially over the life cycle. Cross-country differences remain more pronounced than age-related differences. Greece consistently displays the strongest backward-looking behavior, with relatively large coefficients linking realized inflation, perceptions, and expectations. Finland generally exhibits the weakest responses, while Germany, France, Belgium, Austria, Italy, and Spain occupy intermediate positions. Overall, Figure D5 indicates that age plays only a limited role in shaping the backward-looking formation of inflation expectations. Consumers of different ages respond in broadly similar ways to realized inflation and rely to a comparable extent on inflation perceptions when forming expectations. The evidence therefore suggests that country-specific factors are considerably more important than age in explaining heterogeneity in the backward-looking component of inflation expectations. The sample period is from January 2004 to December 2024. The frequency is monthly. Source: European Commission, Eurostat, and authors' calculations.
Figure D6: Backward-looking component of the ECCS inflation perceptions and
expectations, median consumer, gender
Note: This figure shows regression
results from Eqs. (2)-(5)
estimated by gender cohorts and countries using the European Commission
Consumer Survey (ECCS) filtered data set based on median consumers. The
two gender cohorts are: male and female. Panel A corresponds to the
results of the changes in inflation perceptions on the changes in
realized inflation in the previous month; Panel B the changes in
inflation expectations on the changes in realized inflation in the
previous month; Panel C the changes in inflation expectations on the
changes in inflation perceptions; and Panel D the changes of inflation
expectations on both the changes in inflation perceptions and the
changes in realized inflation in the previous month. Realized inflation
is the country-specific Harmonized Index of Consumer Prices (HICP) for
all items. The vertical bars correspond to the 95 percent confidence
interval. The sample period is from January 2004 to December 2024. The
frequency is monthly.
Source: European Commission, Eurostat and authors’
calculations.
Figure D6 text description The first row of four panels in Figure D6 reports the estimated coefficients of the changes in inflation perceptions regressed on the changes in Harmonized Index of Consumer Prices (HICP) headline inflation lagged one month. In this and further rows, from the left to the right, each panel corresponds to the results based on the households' female and male gender in eight euro-area countries (Germany, France, Italy, Spain, Belgium, Austria, Finland, and Greece). The blue dots represent the estimated coefficients associated with the realized inflation independent variable and the orange dots represent the estimated coefficients associated with the inflation perceptions indepdent variable. Vertical bars around the dots represent 95 percent confidence intervals. As the first row shows, for both women and men, inflation perceptions are positively correlated with the changes in realized inflation across nearly all countries and are particularly large in Greece, Spain, and France. The overall pattern is remarkably similar for women and men, suggesting that both genders process information about recent inflation in broadly similar ways. Although women display slightly larger coefficients in some countries, the differences are generally small relative to the variation observed across countries. The second row of panels in Figure D6 reports the estimated coefficients of the changes in inflation expectations regressed on the changes in headline inflation lagged one month. The estimated coefficients are again predominantly positive for both genders, implying that recent inflation developments contribute to the formation of future inflation expectations. Greece exhibits the strongest response among both female and male consumers, while Italy displays the weakest and least precisely estimated effects. As in previous figures, the confidence intervals are often wider than in Panel A, indicating greater uncertainty surrounding the direct influence of realized inflation on expectations. The third row of panels in Figure D6 reports the estimated coefficients of the changes in inflation expectations regressed on the changes in inflation perceptions lagged one month. The coefficients are positive and statistically significant across virtually all countries and for both genders. This finding confirms that consumers rely heavily on their perceptions of current inflation when forming expectations about future inflation. The estimated coefficients are highly similar for women and men. The differences are modest and do not suggest fundamentally different expectation-formation processes by gender. The fourth row of panels in Figure D5 reports the estimated coefficients of the changes in inflation expectations regressed on both the changes in headline inflation and changes in inflation perceptions lagged one month. In both gender groups, inflation perceptions remain positively associated with inflation expectations and retain substantial explanatory power. By contrast, the direct effect of realized inflation is considerably reduced once perceptions are controlled for and frequently becomes statistically insignificant. This result indicates that realized inflation influences expectations primarily through its effect on consumers' perceptions rather than through a direct channel. Comparing women and men, the similarities substantially outweigh the differences. While women occasionally display somewhat stronger coefficients linking perceptions to expectations, the estimated effects overlap considerably across genders. The structure of the backward-looking mechanism is therefore largely common to both groups: consumers first adjust their perceptions in response to realized inflation and subsequently use those perceptions to form expectations about future inflation. As in the previous figures, cross-country differences are more pronounced than demographic differences. Greece consistently exhibits the strongest backward-looking component, with relatively large coefficients throughout all panels, while Finland generally displays weaker responses. Germany, France, Belgium, Austria, Italy, and Spain occupy intermediate positions. Overall, Figure D6 suggests that gender has only a limited influence on the backward-looking formation of inflation expectations. Although women tend to report higher levels of inflation perceptions and expectations in the descriptive statistics (Figures D1 and D2), the mechanisms linking realized inflation, perceptions, and expectations appear broadly similar for women and men. The evidence therefore indicates that gender differences are more apparent in the level of inflation beliefs than in the process through which those beliefs are formed. The sample period is from January 2004 to December 2024. The frequency is monthly. Source: European Commission, Eurostat, and authors' calculations.