## 2.1 Survey Data (and related empirical sections)

## Source details

**Canonical URL:** [2.1 Survey Data (and related empirical sections)](https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024040-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2024/english/wpiea2024040-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2024/english/wpiea2024040-print-pdf.pdf.json)

---

### Survey design and variables
- Data source: World Value Surveys — nationally representative, interviews in local languages.
- Main individual preference questions:
  - Question 1: choose most important among (a) maintaining order in the nation, (b) giving people more say in important government decision, (c) fighting rising prices, or (d) protecting freedom of speech; respondents also name the second most important.
  - Question 2: choose most important (and second most important) among (a) a high level of economic growth, (b) making sure this country has strong defense forces, (c) seeing that people have more say about how things are done at their jobs and communities, or (d) trying to make our cities and countryside more beautiful.
- Preference dummies constructed per country-year and individual:
  - “Strong preference” = 1 if the outcome ranked most important.
  - “Moderate preference” = 1 if the outcome ranked second most important.
  - Example: “strong preference for price stability” = 1 if “fighting rising prices” ranked most important in Question 1.
- Demographic and socio-economic controls:
  - age, year of birth
  - male dummy = 1 if individual is male
  - single dummy = 1 if single (as opposed to married, divorced or a widower)
  - education dummies (low education and medium education)
  - income dummies: low income = 1 if individual’s income ≤ 30th percentile of country distribution; middle income = 1 if income between 30th and 70th percentile; high income = 1 if income > 70th percentile
  - occupational status dummies (students, houseworkers, “other” = out of labor force; full-time, part-time, self-employed = employed)
  - town size dummies for size of place of residence

### Inflation data sources and sample
- Objective: assign an inflation state (high inflation or low inflation) to each year between an individual’s beginning of working age and survey year.
- Beginning of working age set at k = 15.
- Historical inflation sources:
  - Advanced economies: Jord`a-Schularick-Taylor Macrohistory database (inflation from mid-1800s).
  - Latin American countries: J´acome and Pienknagura (2022) (starting in the 1920s/1930s depending on country).
  - Other countries: IMF International Finance Statistics (IFS) historical records (mid-1930s) and World Development Indicators (WDI) (starting in the 1960s).
- Final inflation database: un-balanced panel of 42 advanced and emerging markets countries with data starting in most cases in the mid-1930s.
- Descriptive observations (Figure 1):
  - relatively high inflation in late 1940s, spike in mid-1970s, gradual decline from early 1980s to 2020s, post-COVID spike.
  - 75th percentile inflation increased substantially in late 1980s/early 1990s; coefficient of variation spikes in the 1980s.

### High inflation exposure (HIE) — definition and thresholds
- HIE_{c,s,a} = sum_{t=s−a+k}^{s} 1(π_{c,t} > π_{c}) where k = 15 and 1(·) is indicator; counts years of high inflation experienced by individual i age a in country c up to survey year s.
- Limitations: national inflation used as approximation; cannot capture geographic mobility or subnational inflation differences.
- High inflation thresholds used:
  - Sample-wide thresholds (starting in 1936): 25 percent yearly inflation (≈ 90th percentile) and 50 percent (≈ 95th percentile).
  - Country-specific threshold: 75th percentile of country-specific distribution.
  - Minimum threshold for all countries set at 10 percent to avoid type-1 errors.
- Distributional notes (Figure 2):
  - Median exposure increased in the 1990s as EMDEs experienced high inflation especially in the 1970s and 1980s.
  - Recent waves show lower median exposure but high heterogeneity.

### Variations and extensions of HIE
- Longest consecutive spell measure: longest spell of high inflation experienced up to year t for each country-cohort.
- Volatility-based exposure: years where rolling 5-year standard deviation of inflation > thresholds (90th and 95th percentiles of total distribution, and 75th percentile country-specific).
- Acceleration-based classification: a large year-to-year inflation change defined as change exceeding one standard deviation of inflation in the five years prior to the event.
- Discounted (recency) exposure:
  - Discounted HIE_{c,s,a} = sum_{t=s−a+k}^{s} δ^{s−t} 1(π_{c,t} > π_{c}) with δ = 0.95.

### Empirical strategy
- Baseline specification:
  - y_{i,c,s,a} = β * HIE_{c,s,a} + θ' * X_{i,c,s,a} + ε_{i,c,s,a}
  - y = dummy for preferences for price stability (main) or growth.
  - X includes country-year of interview fixed effects, age and year of birth fixed effects, gender, occupation, income, education, marital status dummies; sometimes city-size dummies.
  - Standard errors clustered at the country-year level.
- Identification:
  - Country-year fixed effects control for contemporaneous country factors (current inflation, growth, fiscal variables, inequality).
  - Identification compares, within a year, individuals of same age/cohort across countries and individuals of different ages within a country.
- Robustness and heterogeneity checks:
  - Placebo exposure: high inflation in the 25 years prior to individual’s birth.
  - Timing split: HIE_{15–40}, HIE_{40–65}, HIE_{65+}.
  - Interactions with time-varying country variables z_{c,t} (current inflation, rolling CV of GDP per capita growth, bureaucracy quality).

### Main empirical findings — baseline and extensions
- Demographic correlates (Table 2):
  - Lower income individuals express higher strong and moderate concerns over inflation relative to higher income individuals.
  - Lower education associated with higher inflation concerns.
  - Single individuals are less likely to express concerns over inflation.
  - Men are less likely to express concerns over inflation.
  - Occupational status affects strong concerns.
- Main HIE effects (Table 2):
  - Past exposure to high inflation is positive and statistically significant for both strong and moderate inflation concerns.
  - Point estimate for an additional year of exposure is larger for strong concerns than for moderate concerns.
  - Economic magnitude: a one standard deviation increase in exposure (between 5 and 6 additional years of exposure depending on threshold) increases likelihood of expressing strong and moderate concerns by about 0.8 percentage points — roughly a 3 percent increase in likelihood of expressing strong concerns relative to an ex-ante uniform distribution (from 25 percent to 28 percent).
  - A 0.8 percentage point increase maps to an increase in the share expressing strong concerns from the median value in the 154 country-year surveys to the 75th percentile.
- Persistence (long spells) (Table 3):
  - Length of consecutive high inflation spell has a positive and statistically significant impact on strong inflation concerns, but not on moderate concerns.
  - Marginal effect of an additional consecutive year of high inflation is similar to that of overall HIE.
- Volatility and acceleration (Tables 4 and 5):
  - Exposure to volatile inflation (rolling 5-year SD > thresholds) affects strong preferences for price stability, but not moderate preferences.
  - Exposure to past high and accelerating inflation (year-to-year change > one standard deviation of prior five years) has a more pronounced impact on strong concerns — coefficient for accelerating episodes is about twice as large as for other high inflation episodes across thresholds.
- Recency / discounting (Table 6):
  - Using discounted exposure with δ = 0.95, past exposure remains positive and statistically significant for strong concerns across thresholds.
  - For moderate concerns, impact is not statistically significant for the 50 percent threshold.
  - Discounted exposure point estimate is larger in magnitude than baseline, but discounted measure has less variation (roughly one third of standard deviation), implying similar economic changes in attitudes compared to baseline.

### Key quantitative estimates (selected table highlights)
- Table 2 (Observations 198,674): High Inflation Exposure (HIE) coefficients include 0.0013*** (col 1), 0.0016** (col 2), 0.0013** (col 3), 0.0011** (col 4), 0.0010* (col 5), 0.0014** (col 6). R-squared values: 0.0892 (col 1), 0.0891 (col 2), 0.0891 (col 3), 0.0353 (col 4), 0.0353 (col 5), 0.0354 (col 6).
- Table 3 (Observations 198,674): Consecutive years of HIE coefficients include 0.0012** (col 1), 0.0015** (col 2), 0.0018*** (col 3). R-squared reported up to 0.0892 and 0.0353 across columns.
- Table 4 (Observations 198,674): High Inflation Volatility Exposure (HIVE) coefficients include 0.0017*** (col 1), 0.0019*** (col 2), 0.0015*** (col 3), 0.0009* (col 4).
- Table 5 (Observations 198,674): HIE, large year-on-year change coefficients include 0.0034*** (col 1), 0.0041*** (col 2), 0.0035** (col 3); HIE, small year-on-year change coefficients include 0.0018*** (col 1), 0.0021** (col 2), 0.0014* (col 3).
- Table 6 (Observations 198,674): High Inflation Exposure (HIE), discounted coefficients include 0.0039** (col 1), 0.0040* (col 2), 0.0049*** (col 3), 0.0035** (col 4).

### Heterogeneity by timing of exposure (Section 4.2)
- Exposure split into ages 15–40 (early working age), 40–65 (late working age), and 65+ (post-retirement).
- Key findings (Table 7):
  - Exposure in ages 40–65 is strongly associated with strong and moderate inflation concerns across thresholds.
  - HIE, 40–65 years old coefficients: 0.0022*** (col 1), 0.0026*** (col 2), 0.0031*** (col 3).
  - HIE, 15–40 years old coefficient: 0.0007 (col 1) (not consistently significant).
  - HIE, 65+ years old coefficient: 0.0025 (col 1) (not consistently significant across thresholds).
- Interpretation: episodes that inflict larger economic pain on individuals (e.g., affecting savers or those with dependents) leave stronger marks when they occur in ages 40–65.

### Non-linearities and interactions (Sections 4.3–4.4)
- Interaction with current inflation (Table 8):
  - HIE coefficients: 0.0013*** (col 1), 0.0015** (col 2), 0.0012** (col 3).
  - HIE*inflation coefficient: 0.0001 (col 1) (positive but not statistically significant for strong concerns).
  - For moderate concerns, interaction coefficient positive and statistically significant for the 25 percent and 50 percent thresholds.
- Current inflation correlation (Table 9):
  - Observations 239,971; key coefficient: inflation 0.0016*** (col 1) for strong concerns.
- Interaction with bureaucratic quality (Table 10):
  - Observations 183,579.
  - HIE coefficients: 0.0037*** (col 1), 0.0049*** (col 2), 0.0072*** (col 3).
  - HIE*Bureaucratic Quality: -0.0010** (col 1) (negative and statistically significant: stronger institutions dampen HIE effect).
- Interaction with macroeconomic volatility (Table 11):
  - Observations 197,757.
  - HIE 0.0013*** (col 1); HIE*CV of GDP per capita growth 0.0000*** (col 1) (higher growth volatility amplifies HIE effect).

### Robustness checks and extensions
- Excluding top-five inflation peak countries (Table 12): excludes Argentina, Bolivia, Germany, Nicaragua and Peru; Observations 178,181. Result: HIE point estimate and significance for strong concerns not affected; moderate concerns point estimates and significance fall but remain statistically significant in most cases.
- Sample extensions and town-size controls (Table 13):
  - Observations 239,924 (columns with sample ≥15) and 198,674 (columns with sample ≥25).
  - Including ages 15–25 reduces HIE point estimates and significance (cohorts have HIE close to zero), but HIE remains statistically significant at standard levels.
  - Including town-size dummies leaves HIE impact virtually unchanged.
- Placebo / parallel trends (Table 14):
  - Pre-birth exposure (25 years prior to birth) placebo HIE coefficient: -0.0005 (col 1) (not statistically significant); supports parallel trends assumption.
- Growth concerns (Table 15):
  - Observations 198,674.
  - HIE coefficients for strong growth concerns: -0.0013** (col 1), -0.0009 (col 2), -0.0008 (col 3) — not robustly linked to growth concerns overall.

### Summary of quantitative implications (Section 5)
- Exposure during ages 40–65: strong association with strong and moderate inflation concerns across thresholds.
- Exposure during ages 15–40: significance limited to moderate concerns at the 75th percentile threshold.
- Exposure after 65: significance limited to moderate concerns at the 25 percent and 50 percent thresholds.
- Current inflation amplifies the HIE effect for moderate concerns (25 percent and 50 percent thresholds) but not for strong concerns.
- Bureaucratic quality interaction: negative and statistically significant (dampening effect).
- Growth volatility interaction: positive amplification of HIE effect.
- Robustness checks (excluding top-five hyperinflation countries; extending sample; town-size controls; pre-birth placebo) support the baseline findings.

*Source — wpiea2024040-print-pdf, sections 2.1, 4.2–4.4, and 5.*

### 2.1    Survey Data

### 2.1    Survey Data

### Survey design and variables
- Data source: World Value Surveys — nationally representative, interviews in local languages.
- Main individual preference questions:
  - Question 1 (primary focus): choose most important among (a) maintaining order in the nation, (b) giving people more say in important government decision, (c) fighting rising prices, or (d) protecting freedom of speech; respondents also name the second most important.
  - Question 2: choose most important (and second most important) among (a) a high level of economic growth, (b) making sure this country has strong defense forces, (c) seeing that people have more say about how things are done at their jobs and communities, or (d) trying to make our cities and countryside more beautiful.
- Preference dummies constructed per country-year and individual:
  - “Strong preference” = 1 if the outcome ranked most important.
  - “Moderate preference” = 1 if the outcome ranked second most important.
  - Specific constructs: “strong preference for price stability” = 1 if “fighting rising prices” ranked most important in Question 1.
- Demographic and socio-economic controls:
  - age, year of birth
  - male dummy = 1 if individual is male
  - single dummy = 1 if single (as opposed to married, divorced or a widower)
  - education dummies (low education and medium education)
  - income dummies: low income = 1 if individual’s income ≤ 30th percentile of country distribution; middle income = 1 if income between 30th and 70th percentile; high income = 1 if income > 70th percentile
  - occupational status dummies (students, houseworkers, “other” = out of labor force; full-time, part-time, self-employed = employed)
  - town size dummies for size of place of residence

### Inflation data sources and sample
- Objective: assign an inflation state (high inflation or low inflation) to each year between an individual’s beginning of working age and survey year.
- Beginning of working age set at k = 15.
- Historical inflation sources:
  - Advanced economies: Jord`a-Schularick-Taylor Macrohistory database (inflation from mid-1800s).
  - Latin American countries: J´acome and Pienknagura (2022) (starting in the 1920s/1930s depending on country).
  - Other countries: IMF International Finance Statistics (IFS) historical records (mid-1930s) and World Development Indicators (WDI) (starting in the 1960s).
- Final inflation database: un-balanced panel of 42 advanced and emerging markets countries with data starting in most cases in the mid-1930s.
- Presentation notes (Figure 1 described):
  - Shows median inflation value by year and 25th and 75th percentiles of within-year cross-country distribution.
  - Observations: relatively high inflation in late 1940s, spike in mid-1970s, gradual decline from early 1980s to 2020s, post-COVID spike.
  - 75th percentile inflation increased substantially in late 1980s/early 1990s; coefficient of variation spikes in the 1980s.

### High inflation exposure (HIE) — definition
- HIE operationalized as number of years that individual i, age a, in country c, experienced high inflation from beginning of working age to survey year s:
  - HIE_{c,s,a} = sum_{t=s−a+k}^{s} 1(π_{c,t} > π_{c}) where k = 15 and 1(·) is indicator.
- Limitations noted:
  - Cannot capture individual geographic mobility or subnational inflation differences; uses national inflation as approximation.
- High inflation thresholds used (two approaches):
  - Sample-wide thresholds (starting in 1936): 25 percent yearly inflation (≈ 90th percentile) and 50 percent (≈ 95th percentile).
  - Country-specific threshold: 75th percentile of country-specific distribution.
  - Minimum threshold for all countries set at 10 percent to avoid type-1 errors.
- Distributional notes (Figure 2 described):
  - Median exposure increased in the 1990s as EMDEs experienced high inflation especially in the 1970s and 1980s.
  - Recent waves show lower median exposure but high heterogeneity.

### Variations and extensions of HIE
- Longest consecutive spell measure: for each country-cohort compute the longest spell of high inflation experienced up to year t.
- Volatility-based exposure: identify years where rolling 5-year standard deviation of inflation > thresholds (90th and 95th percentiles of total distribution, and 75th percentile country-specific).
- Acceleration-based classification: define large year-to-year inflation change as change exceeding one standard deviation of inflation in the five years prior to the event.
- Discounted exposure to capture “recency” bias:
  - Discounted HIE_{c,s,a} = sum_{t=s−a+k}^{s} δ^{s−t} 1(π_{c,t} > π_{c}) with δ = 0.95.

### Empirical strategy
- Baseline specification:
  - y_{i,c,s,a} = β * HIE_{c,s,a} + θ' * X_{i,c,s,a} + ε_{i,c,s,a}
  - y = dummy for preferences for price stability (main) or growth.
  - X includes country-year of interview fixed effects, age and year of birth fixed effects, gender, occupation, income, education, marital status dummies; sometimes city-size dummies.
  - Standard errors clustered at the country-year level.
- Identification:
  - Country-year fixed effects control for contemporaneous country factors (current inflation, growth, fiscal variables, inequality).
  - Identification compares, within a year, individuals of same age/cohort across countries and individuals of different ages within a country.
- Placebo and robustness checks:
  - Placebo exposure: high inflation in the 25 years prior to individual’s birth.
  - Heterogeneity by timing of exposure: split HIE into ages 15–40 (early), 40–65 (late working years), and 65+ (post-retirement):
    - y_{i,c,s,a} = β_y * HIE_{15−40,c,s,a} + β_m * HIE_{40−65,c,s,a} + β_o * HIE_{65+,c,s,a} + θ' X + ε
  - Interaction with country-specific variables z_{c,t} (current inflation, time-varying coefficient of variation of GDP per capita growth, bureaucracy quality) to estimate:
    - y_{i,c,s,a} = β * HIE_{c,s,a} + γ * HIE_{c,s,a} * z_{c,t} + θ' X + ε

### Results — baseline and extensions (summary of empirical findings)
- Demographic correlates (Table 2):
  - Lower income individuals express higher strong and moderate concerns over inflation relative to higher income individuals.
  - Lower education associated with higher inflation concerns.
  - Single individuals are less likely to express concerns over inflation.
  - Men are less likely to express concerns over inflation.
  - Occupational status affects strong concerns.
- Main HIE effects (Table 2):
  - Past exposure to high inflation is positive and statistically significant for both strong and moderate inflation concerns.
  - Point estimate for an additional year of exposure is larger for strong concerns than for moderate concerns.
  - Economic magnitude: a one standard deviation increase in exposure (between 5 and 6 additional years of exposure depending on threshold) increases likelihood of expressing strong and moderate concerns by about 0.8 percentage points — roughly a 3 percent increase in likelihood of expressing strong concerns relative to an ex-ante uniform distribution (from 25 percent to 28 percent).
  - A 0.8 percentage point increase in likelihood maps to an increase in the share expressing strong concerns from the median value in the 154 country-year surveys to the 75th percentile.
- Persistence (long spells) (Table 3):
  - Length of consecutive high inflation spell has a positive and statistically significant impact on strong inflation concerns, but not on moderate concerns.
  - The marginal effect of an additional consecutive year of high inflation is similar to that of overall HIE.
- Volatility and acceleration (Tables 4 and 5):
  - Exposure to volatile inflation (rolling 5-year SD > thresholds) affects strong preferences for price stability, but not moderate preferences.
  - Exposure to past high and accelerating inflation (year-to-year change > one standard deviation of prior five years) has a more pronounced impact on strong concerns — coefficient for accelerating episodes is about twice as large as for other high inflation episodes across thresholds.
- Recency / discounting (Table 6):
  - Using discounted exposure with δ = 0.95, past exposure remains positive and statistically significant for strong concerns across thresholds.
  - For moderate concerns, impact is not statistically significant for the 50 percent threshold.
  - Discounted exposure point estimate is larger in magnitude than baseline, but discounted measure has less variation (roughly one third of standard deviation), implying similar economic changes in attitudes compared to baseline.

*Italic: Source — wpiea2024040-print-pdf, section 2.1–4 (Survey Data, Inflation Data, Exposure to High Inflation Episodes, Empirical Strategy, Results).*

### 4.2    Exploring Heterogeneity Based on the Timing of Exposure

### 4.2    Exploring Heterogeneity Based on the Timing of Exposure

### Timing of exposure: methodology
- Exposure broken into three age ranges: exposure between the age of 15 and 40 (early working age), exposure between 40 and 65 (late working age), and exposure after 65.
- For each age range, compute the number of years each cohort experienced high inflation, constructing an index similar to (1) for each age range.
- Estimated specification: equation (3).

### Timing of exposure: key findings
- Exposure in the latter part of an individual’s work life (age 40–65) is strongly associated with strong and moderate inflation concerns, regardless of the high-inflation threshold used.
- Exposure during the initial working years (age 15–40) is statistically significant only for moderate concerns when using the 75th percentile threshold.
- Exposure after retirement age (after 65) is statistically significant only when considering the 25 percent and 50 percent thresholds and moderate concerns.
- Interpretation: high inflation episodes leave a stronger mark when they impose larger economic pain on individuals (e.g., by affecting savers or those with dependents).

### Exploring non-linearities: variables and approach
- Considered interaction effects between past exposure to high inflation and:
  - current inflation levels;
  - macroeconomic volatility (proxied by the 5-year rolling coefficient of variation of GDP per capita growth);
  - quality of a country’s bureaucracy (proxy for institutional strength).
- Extended specification in (2) by adding interaction terms (equation 4). Fixed effects control for the direct impact of each variable on inflation concerns.

### Exploring non-linearities: findings on current inflation
- Table 8: interaction between past inflation exposure and current inflation.
  - For strong concerns: interaction coefficient positive but not statistically significant.
  - For moderate concerns: interaction coefficient positive and statistically significant for the 25 percent and 50 percent thresholds; magnitude larger than for strong concerns.
- Explanation: during periods of high inflation, a larger share of the population may express strong concerns regardless of past experience, while moderate concerns are more sensitive to the amplification from current inflation.
- Supporting evidence: simple regression (Table 9) shows likelihood of expressing strong concerns is strongly correlated with current inflation levels; likelihood of expressing moderate concerns appears uncorrelated with contemporaneous inflation levels.

### Exploring non-linearities: findings on institutions and volatility
- Bureaucratic quality (proxy for institutions):
  - Interaction of the HIE and bureaucratic quality is negative and statistically significant (Table 10).
  - Stronger institutions reduce the impact of past exposure to inflation on current inflation concerns.
  - Interpretation: better institutions/administrative capacity may reassure individuals that authorities can manage inflationary pressures.
- Macroeconomic instability (growth volatility):
  - Interaction results (Table 11) show that higher volatility of real GDP per capita growth amplifies the impact of past inflation exposure on present inflation perceptions.
  - More unstable economies see a stronger effect of past exposure on inflation concerns.

### Extensions and robustness: concerns addressed
- Concern: results driven by a few hyperinflation episodes in emerging markets.
  - Table 12 excludes the five countries with the highest yearly inflation peaks in the sample: Argentina (peak inflation over 3080 percent in 1989), Bolivia (peak inflation over 8170 in 1985), Germany (peak inflation over 1e11 in 1923), Nicaragua (peak inflation over 7750 percent in 1991) and Peru (peak inflation over 7480 percent in 1990).
  - Result: for strong concerns, HIE point estimate and statistical significance not affected; point estimate slightly higher. For moderate concerns, point estimates and significance fall but remain statistically significant in most cases.
- Sample extensions:
  - Included individuals over 15 years of age and included dummies for town size (middle-size town: between 50,000 and 100,000 inhabitants; large-size town: over 100,00 inhabitants).
  - Including ages 15–25 reduces HIE point estimates and statistical significance (these cohorts have HIE close to zero), but HIE remains statistically significant at standard levels.
  - Including town-size dummies leaves HIE impact virtually unchanged.
- Parallel trends / placebo test:
  - Constructed pre-birth “exposure” for each cohort by counting years of high inflation in the 25 years prior to the cohort’s birth (placebo modification of equation (1)).
  - Table 14: pre-birth placebo exposure is not statistically significant and has a substantially smaller point estimate than baseline HIE, supporting the parallel trends assumption.
- Broader economic concerns:
  - Examined whether past inflation exposure affects concerns about growth (using same approach as for inflation).
  - Table 15: exposure to past inflation is not robustly linked to growth concerns. HIE coefficient negative and significant only for one of three thresholds in the case of strong growth concerns and not significant for moderate concerns.
  - Interpretation: inflation exposure affects inflation-focused expectations but not broader macroeconomic concerns.

### Summary of quantitative implications
- Exposure during ages 40–65: strong association with strong and moderate inflation concerns across thresholds.
- Exposure during ages 15–40: significance limited to moderate concerns at the 75th percentile threshold.
- Exposure after 65: significance limited to moderate concerns at the 25 percent and 50 percent thresholds.
- Current inflation amplifies the HIE effect for moderate concerns (25 percent and 50 percent thresholds) but not for strong concerns.
- Bureaucratic quality interaction: negative and statistically significant (dampening effect).
- Growth volatility interaction: positive amplification of HIE effect.
- Robustness checks (excluding top-five hyperinflation countries; extending sample; town-size controls; pre-birth placebo) support the baseline findings.

*Source: Section 4.2–4.4 and 5 of the provided IMF chapter "4.2    Exploring Heterogeneity Based on the Timing of Exposure."*

### References

### References

### Cited works
- Acemoglu, D., N. Ajzenman, C. Girat Aksoy, M. Fiszbein, and C. Molina (forthcoming), “(Successful) Democracies Breed their Own Support,” Review of Economic Studies.
- Allinger, K. and F. Rumler Inflation (2023), “Expectations in CESEE: The Role of Sentiment and Experiences,” Austrian central Bank OeNB working paper.
- Binder, C., and K. Makridis (2022) “Stuck in the Seventies: Gas Prices and Consumer Sentiment,” Review of Economic and Statistics, Vol. 104 (2): 293-305.
- Braggion, F., Von Meyerinck, F., Schaub, N., and Weber, M. (2023). “The long-term effects of inflation on inflation expectations,” mimeo.
- Cavallo, E., G. Cruces, and R. Perez-Truglia (2017) “Inflation Expectations, Learning, and Supermarket Prices: Evidence from Survey Experiments,” American Economic Journal: Macroeconomics, Vol. 9(3): 1-35.
- Conrad, C., Enders, Z., and Glas, A. (2022). “The role of information and experience for households inflation expectations.” European Economic Review, 143, 104015.
- D’Acunto, F., U. Malmendier, and M. Weber (2021) “Gender Roles Produce Divergent Economic Expectations,” Proceedings of the National Academy of Sciences Vol. 118 (21): 1-10.
- D’Acunto, F., U. Malmendier, and M. Weber (2022) “What do the Data Tells Us About Inflation Expectations?” NBER Working Paper 29825.
- D’Acunto, F., U. Malmendier, J. Ospina, and M. Weber (2020) “Exposure to Grocery Prices and Inflation Expectations,” Chicago Booth Paper 19-17.
- Das, S., C. Kuhnen, S. Nagel (2019) “Socioeconomic Status and Macroeconomic Expectations,” Review of Financial Studies, Vol. 33(1): 395-432.
- Erosa, A., and G. Ventura (2002) “On Inflation as a Regressive Consumption Tax,” Journal of Monetary Economics, Vol. 49: 761-795.
- Fischer, S. (1981), “The Costs and Consequences of Inflation,” Carnegie-Rochester Conference Series on Public Policy 15: 5-42.
- Friedman, M. (1969), “The Optimum Quantity of Money and Other Essays,” Chicago.
- Giuliano, P. and A. Spilimbergo (2023), “Aggregate Shocks and the Formation of Preferences and Beliefs,” mimeo.
- Haberger, A. and S. Edwards (1980), “International Evidence on the Sources of Inflation,” mimeo.
- Heymann, D. and A. Leijonhufvud (1995), “High Inflation,” Oxford University Press.
- Jâcome, L. I., S. Pienknagura (2022) “Central Bank Independence and Inflation in Latin AmericaThrough the Lens of History,” IMF Working Papers 2022/186, International Monetary Fund.
- Jordà, O., M. Schularick, and A. M. Taylor (2017) “Macrofinancial History and the New Business Cycle Facts.” in NBER Macroeconomics Annual 2016, volume 31, edited by Martin Eichenbaum and Jonathan A. Parker. Chicago: University of Chicago Press.
- Malmendier, U. (2021), “Exposure, Experience, and Expertise: Why Personal Histories Matter in Economics,” Journal of the European Economic Association, 19 (6), 2857-2894.
- Malmendier, U. and S. Nagel (2011) “Depression Babies: Do Macroeconomic Experiences Affect Risk Taking?” The Quarterly Journal of Economics, Vol. 126: 373-416.
- Malmendier, U. and S. Nagel (2016) “Learning from Inflation Experiences,” Quarterly Journal of Economics, Vol.131: 53-87.
- Salles, I., Y. Gorodnichenko, and O. Coibion (2023) “Lifetime Memories of Inflation: Evidence from Surveys and the Lab”, NBER Working Paper 31996.
- Weber, M., F. DAcunto, O. Coibion, and Y. Gorodnichenko (2022) “The Subjective Economic Expectations of Households and Firms:Measurement, Determinants, and Implications,” Journal of Economic Perspectives, Vol. 36 (3): 157-184.

### Figures and Tables inventory (titles and selected table-level summary statistics preserved exactly)
- Figure 1: Cross-Country Distribution of Inflation over Time—1945-2021
- Figure 2: Distribution of High Inflation Exposure Among Respondents—By Survey Wave
  - Note: Survey waves were conducted in the following years: Wave 1, 1981-84, Wave 2, 1989-93, Wave 3, 1994-98, Wave 4, 1999-2004, Wave 5, 2005-09, Wave 6, 2010-2014, Wave 7, 2017-2022.

- Table 1: Sample of Countries in the Analysis
  - Advanced Economies: Australia, Canada, Finland, France, Germany, Israel, Italy, Japan, Republic of Korea, The Netherlands, Norway, New Zealand, Spain, Sweden, Switzerland, United Kingdom, United States
  - Emerging Markets: Argentina, Bolivia, Brazil, Chile, Colombia, Dominican Republic, Ecuador, Egypt, El Salvador, Guatemala, India, Indonesia, Iraq, Islamic Republic of Iran, Lebanon, Mexico, Malaysia, Nicaragua, Peru, Philippines, South Africa, Thailand, Turkiye, Uruguay, Rep. Bol. of Venezuela

- Table 2: High Inflation Exposure and Concerns over Price Stability
  - Observations 198,674
  - R-squared 0.0892 (columns 1), 0.0891 (2), 0.0891 (3), 0.0353 (4), 0.0353 (5), 0.0354 (6)
  - Key coefficient: High Inflation Exposure (HIE) 0.0013*** (col 1), 0.0016** (col 2), 0.0013** (col 3), 0.0011** (col 4), 0.0010* (col 5), 0.0014** (col 6)

- Table 3: Exposure to Persistent High Inflation and Concerns over Price Stability
  - Observations 198,674
  - R-squared 0.0891 (1), 0.0891 (2), 0.0892 (3), 0.0353 (4), 0.0353 (5), 0.0353 (6)
  - Key coefficient: Consecutive years of HIE 0.0012** (col 1), 0.0015** (col 2), 0.0018*** (col 3)

- Table 4: High Inflation Volatility Exposure and Concerns over Price Stability
  - Observations 198,674
  - R-squared 0.0892 (1), 0.0891 (2), 0.0891 (3), 0.0353 (4), 0.0353 (5), 0.0353 (6)
  - Key coefficient: High Inflation Volatility Exposure (HIVE) 0.0017*** (col 1), 0.0019*** (col 2), 0.0015*** (col 3), 0.0009* (col 4)

- Table 5: High Inflation Exposure and Strong Concerns over Price Stability: High and Accelerating Inflation vs. High and Stable Inflation
  - Observations 198,674
  - R-squared 0.0892 (1), 0.0891 (2), 0.0891 (3), 0.0891 (4), 0.0891 (5), 0.0891 (6)
  - Key coefficients: HIE, large year-on-year change 0.0034*** (col 1), 0.0041*** (col 2), 0.0035** (col 3); HIE, small year-on-year change 0.0018*** (col 1), 0.0021** (col 2), 0.0014* (col 3)

- Table 6: High Inflation Exposure and Concerns over Price Stability, Recency Bias
  - Observations 198,674
  - R-squared 0.0891 (1), 0.0891 (2), 0.0891 (3), 0.0353 (4), 0.0353 (5), 0.0353 (6)
  - Key coefficient: High Inflation Exposure (HIE), discounted 0.0039** (col 1), 0.0040* (col 2), 0.0049*** (col 3), 0.0035** (col 4)

- Table 7: High Inflation Exposure and Concerns over Price Stability, Timing of Exposure
  - Observations 198,674
  - R-squared 0.0892 (1), 0.0892 (2), 0.0893 (3), 0.0354 (4), 0.0353 (5), 0.0354 (6)
  - Key coefficients: HIE, 15-40 years old 0.0007 (col 1); HIE, 40-65 years old 0.0022*** (col 1), 0.0026*** (col 2), 0.0031*** (col 3); HIE, 65+ years old 0.0025 (col 1)

- Table 8: High Inflation Exposure and Concerns over Price Stability, Heterogeneity by Current Inflation
  - Observations 198,674
  - R-squared 0.0892 (1), 0.0891 (2), 0.0891 (3), 0.0354 (4), 0.0353 (5), 0.0354 (6)
  - Key coefficients: HIE 0.0013*** (col 1), 0.0015** (col 2), 0.0012** (col 3); HIE*inflation 0.0001 (col 1)

- Table 9: Strong and Moderate Inflation Concerns and Current Inflation Levels
  - Observations 239,971
  - R-squared 0.0611 (col 1), 0.0212 (col 2)
  - Key coefficient: inflation 0.0016*** (col 1)

- Table 10: Stronger Institutions Reduce the Impact of past Exposure to High Inflation
  - Observations 183,579
  - R-squared 0.0861 (1), 0.0861 (2), 0.0864 (3), 0.0315 (4), 0.0315 (5), 0.0315 (6)
  - Key coefficients: HIE 0.0037*** (col 1), 0.0049*** (col 2), 0.0072*** (col 3); HIE*Bureaucratic Quality -0.0010** (col 1)

- Table 11: Lower Macroeconomic Instability Reduces the Impact of past Exposure to High Inflation
  - Observations 197,757
  - R-squared 0.0896 (1), 0.0895 (2), 0.0895 (3), 0.0355 (4), 0.0355 (5), 0.0355 (6)
  - Key coefficients: HIE 0.0013*** (col 1); HIE*CV of GDP per capita growth 0.0000*** (col 1)

- Table 12: High Inflation Exposure and Concerns over Price Stability: Exclusion of Countries with Top Five Inflation Rates
  - Observations 178,181
  - R-squared 0.0911 (1), 0.0910 (2), 0.0910 (3), 0.0369 (4), 0.0369 (5), 0.0369 (6)
  - Note: Excludes Argentina, Bolivia, Germany, Nicaragua and Peru

- Table 13: High Inflation Exposure and Concerns over Price Stability: Additional Controls and Alternative Samples
  - Observations 239,924 (columns with sample ¿15) and 198,674 (columns with sample ¿25)
  - R-squared reported across models: 0.0871, 0.0892, 0.0352, 0.0354, 0.0353, 0.0892 (various columns)
  - Selected key coefficients: High Inflation Exposure (HIE) ranges in table entries, e.g., 0.0009** (col 1), 0.0013*** (col 2), 0.0009* (col 3), 0.0016** (col 4), 0.0012** (col 5), 0.0013** (col 6)

- Table 14: High Inflation Exposure and Concerns over Price Stability: Placebo Exercise
  - Observations 198,674
  - R-squared 0.0884 (columns 1-3), 0.0346 (columns 4-6)
  - Key coefficient: High Inflation Exposure (HIE) placebo -0.0005 (col 1)

- Table 15: High Inflation Exposure and Growth Concerns
  - Observations 198,674
  - R-squared 0.1582 (strong growth concerns models), 0.0514 (moderate growth concerns models)
  - Key coefficient: High Inflation Exposure (HIE) -0.0013** (col 1), -0.0009 (col 2), -0.0008 (col 3)

*Inflated Concerns: Exposure to Past Inflationary Episodes and Preferences for Price Stability — Working Paper No. WP/2024/040*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024040-print-pdf.pdf_