## CHAPTER 3 ONLINE ANNEXES

## Source details

**Canonical URL:** [CHAPTER 3 ONLINE ANNEXES](https://www.imf.org/-/media/files/publications/weo/2024/october/english/ch3onlineannexes.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/weo/2024/october/english/ch3onlineannexes.pdf.md)
- [Structured JSON version](/-/media/files/publications/weo/2024/october/english/ch3onlineannexes.pdf.json)

---

### Narrative dataset construction and methodology
- Information sources and extraction:
  - Information extracted from the EIU reports using OpenAI’s GPT-4o model.
  - The model’s performance was evaluated iteratively against human-generated answers until achieving an overall accuracy of approximately 80 percent in the test sample.
  - The model was tasked with answering questions related to six aspects of reforms: (i) scope and direction of reform; (ii) timing and duration; (iii) actors; (iv) consultations/communications strategies; (v) complementary/compensatory measures; and (vi) social resistance and government response.
- Definitions and episode rules:
  - A reform episode is defined as a continuous period of reform efforts within the same reform area and with a direction consistent with promoting economic growth.
  - A reform attempt is considered implemented if at least one EIU report during the reform episode mentions an implementation of the relevant reform measures.
  - Distinct reform episodes are separated by periods of at least two consecutive quarters with no discussion, adoption, or implementation within that reform area.
  - The narrative database records whether consultation/communications strategies or complementary/compensatory measures were prominent enough to be discussed in the EIU reports; absence in the narrative database does not necessarily indicate these measures were not employed, nor does it indicate intensity or adequacy.

### Descriptive counts and episode inventory
- Online Annex Table 3.2.1. Number of Reform Episodes, 1996–2023 (as presented):
  - AEsEMEsLICsTotal
  - PMR-Electricity761798 5340
  - Elder LP77482127
  - Migrant Integration1147019203

### Social resistance, reversals, and implications
- Definition of reversal:
  - A reversal is the implementation of a policy in the opposite direction of a previously identified reform implementation (examples provided in the annex).
- Key descriptive finding:
  - Among PMR-electricity and elder LP reforms, a significantly higher share of reforms that were implemented with resistance were subsequently reversed within five years, compared with those implemented without resistance (Online Annex Figure 3.2.1).
  - Statistical significance markers in the figure: ***, **, and * denote significance at the 1, 5, and 10 percent levels, respectively, for t-tests comparing reversal rates between resisted and unresisted implemented reforms.

### Econometric framework and controls
- Model:
  - Multinomial logistic regression framework specified in equations (3.2.1) and (3.2.2).
  - Three outcome categories (j = 0,1,2):
    - j = 0: reform episodes that ended without a reform implementation (discussed but never implemented) — base outcome.
    - j = 1: reforms implemented but resisted.
    - j = 2: reforms implemented without resistance.
  - Z_iut is the vector of determinants; X_iut denotes controls; α_g(i) denotes country income group fixed effects where g(i) ∈ {AEs, EMEs, LICs}.
- Controls and data sources (Online Annex Table 3.2.2 summary):
  - Population age shares — World Bank, World Development Indicators
  - Real GDP growth — IMF, World Economic Outlook database
  - Financial crises — Laeven and Valencia (2020); Nguyen and others (2022)
  - Inflation — International Monetary Fund, World Economic Outlook database
  - Fiscal space — Kose and others (2022)
  - Exchange rate arrangements — Ilzetzki and others (2019)
  - Fiscal rules — Davoodi and others (2022)
  - Monetary framework — Romelli (2022)
  - Globalization index — Gygli and others (2019)
  - Trade flows and trade partners — Conte and others (2023)
  - IMF program — IMF, Monitoring of Fund Arrangements (MONA)
  - Political factors — Scartascini, Cruz, and Keefer (2021); Herre (2023)
  - Income inequality — Solt (2020)
  - Electoral democracy index — Coppedge and others (2024)

### Main econometric findings (Annex Table 3.2.3 highlights)
- General:
  - Variables grouped into seven categories: macroeconomic context, macroeconomic framework, external factors, political factors, reform strategies, and control variables.
  - Each coefficient indicates the relative risk ratio (exponent of γ_j) for a one-unit increase in the explanatory variable; values greater (lower) than 1 indicate increased (decreased) odds of implementation relative to the base outcome.
- Selected coefficients and statistical significance (odds ratios with robust standard errors in parentheses, as reported):
  - Macroeconomic context:
    - GDP growth deviation from 10-year average: 0.910** (0.042) for Imp (R) in column (1).
    - Recession: 0.220** (0.155) for Imp (R) in column (1).
    - Financial crisis: 2.090** (0.624) for Imp (R) in column (1).
  - Macroeconomic framework:
    - Inflation deviation from 1-year average: 1.001 (0.001) in column (1) and 0.998*** (0.001) in column (2).
    - Strong fiscal position: 1.084 (0.305) in column (1).
    - Strong fiscal/monetary framework: 1.354 (0.333) in column (1) and 1.425* (0.276) in column (2).
  - External factors:
    - Share of countries implementing the reform in the past 5 years: 1.042 (0.610) in column (1) and 3.245** (1.593) in column (2).
    - Globalization index: 0.892*** (0.027) in column (1).
    - Engaged in regional trade agreements: 1.000 (0.006) in column (1).
    - IMF program: 0.789 (0.337) in column (1) and 1.130 (0.304) in column (2).
  - Political factors:
    - Chief Executive, years in office: 1.045* (0.025) in column (1).
    - Year of election: 0.371** (0.159) in column (1) and 0.534*** (0.128) in column (2).
    - Years until next election: 1.166 (0.145) in column (1).
    - Large margin of majority: 1.395 (0.382) in column (1) and 1.595*** (0.258) in column (2).
    - Leader ideology (left): 0.689** (0.100) in column (1).
  - Reform strategies (notable strong associations with implementation):
    - Consultation/communication strategy: 3.477*** (1.055) in column (1); 1.295 (0.271) in column (2); 4.306** (2.702) in column (3).
    - Complementary/compensatory measures: 3.652*** (1.096) in column (1); 0.982 (0.201) in column (2); 3.723** (2.007) in column (3); 3.402*** (1.340) in column (4); 2.162* (0.882) in column (5); 1.627* (0.415) in column (6).
- Interpretation note:
  - Coefficients report changes in the odds ratio of reform implementation outcomes. Value greater (smaller) than 1 indicates increase (decrease) in the odds ratio relative to the baseline outcome (reforms that were discussed but never implemented).
  - Estimations include controls for lagged reform index for PMR-Electricity, exports of goods and services as a share of GDP, share of population over 65 years old, deviation of Gini coefficient from 10-year average, electoral democracy index, and country income group fixed effect.
  - ***, **, and * denote significance at the 1, 5, and 10 percent level, respectively.

### Marginal effects and role of reform strategies (Annex Figure 3.2.2 and dominance analysis)
- Marginal effects (average marginal effects, calculated at the observation level then averaged):
  - Consultation and communications strategies:
    - Associated with a 6-8 percentage points higher probability of implementation in the case of resisted reforms (reported in the text summarizing Annex Figure 3.2.2).
  - Compensatory and complementary measures:
    - Associated with a 5-8 percentage point increase in the probability of resisted reforms being implemented.
    - Associated with a 6 percentage point increase of implementation in the case of unresisted elder LP and migrant integration reforms.
  - The marginal effects are shown with 90 percent confidence intervals in the figure.
- Dominance analysis:
  - Reform strategies remain the most important in predicting reform outcomes.
  - Joint explanatory power of reform strategies ranges from 25 to 45 percent across reform areas (as reported in the annex text).
  - Robustness check: dominance analysis re-estimated by including determinants individually (rather than in clusters) confirmed reform strategies’ prominence.

### Survey experiments on public attitudes (Online Annex 3.3) — overview
- Purpose and coverage:
  - Purpose: study drivers of people’s attitudes towards reforms using large-scale online surveys with embedded randomized control trial experiments.
  - Survey 1 (PMR reforms): Mexico, Morocco, and South Africa (focus on reforms to ease PMR and network sectors).
  - Survey 2 (Migrant integration policies): Canada, Italy, and the United Kingdom (focus on policies to integrate foreign-born workers).
- Execution and sample:
  - Conducted between June 7 and July 12, 2024, by YouGov.
  - Total respondents: 12,600 (2,100 per country).
  - Respondent selection: representative preselected pool based on each country’s demographic profile along age, gender, education, employment status, and regional distribution; propensity score matching applied.
  - Mode: online surveys with residents aged 18 years and older; administered in local languages.
  - Data validation: standard checks and protocols employed.
- Context notes and pre-registration:
  - PMR survey countries: regulatory stance in network sectors relatively tight compared to peers.
  - Migrant integration survey countries: 2020 Migrants Integration Policy Index scores cited — Canada 76/100 in immigrants labor market mobility and 80/100 overall; Italy 67/100 and 58/100; United Kingdom 48/100 and 56/100.
  - Both surveys pre-registered in AsPredicted and certified by the German Association for Experimental Economic Research.

### Survey design and outcome measures
- Common questionnaire features:
  - Collects socioeconomic information; elicits beliefs about societal values, institutions, and policies; randomly assigns respondents to information treatments or control; constructs three main outcome variables: support for policies; perception of the effects of the policies; willingness to sign a petition in support or against the reform.
- Policy support measures:
  - PMR questionnaire: “Do you support or oppose allowing private companies to produce and sell [electricity / telecommunication services] in the country” on Likert scale.
  - Migrant integration questionnaire: four policies (work authorization on arrival; quicker recognition of foreign qualifications; job training and search assistance; support for immigrant entrepreneurship) asked on Likert scale; answers combined into a country-specific index following Anderson 2008, then standardized across countries.

### Perceptions, stereotypes, and belief elicitation
- PMR perceptions:
  - Questions on whether competition is beneficial or detrimental for consumers and on cost, quality and access effects of private competition.
- Migrant integration perceptions:
  - Question on how integration measures would affect: (i) jobs for people born in the country, (ii) money the government gets from taxes, (iii) crime immigrants commit. Responses used to create an index.
- Stereotypes and narratives:
  - Open-ended elicitation on immigrants’ traits (hard-working, lazy, dangerous, friendly, refugees, illegal immigrants) and perceived economic impact; analyzed using large language models.

### Empirical model for survey experiments
- Main regression specification (Equation 3.3.1):
  - y_iou = β1 T_iou + β2 S_iou + β3 B_iou + θ_u + ε_iou  (Equation 3.3.1; presented in the annex).
  - y_iou is the outcome for individual i in country u (support, beliefs, petition willingness).
  - T_iou are treatment indicators testing three information strategies.
  - S_iou captures individuals’ socioeconomic characteristics.
  - B_iou includes beliefs and perceptions.
  - θ_u denotes country fixed effects; ε_iou is the error term.

### Role of beliefs versus socioeconomic characteristics
- Dominance analysis (Luchman, 2021) used to decompose variance explained by predictors.
- Clustering of belief/perception variables:
  - PMR: Market Economy; Distribution and Equity; Trust and Corruption; Views and Satisfaction with the Sector; Knowledge and Effect of Policies.
  - Migrant integration: Stereotypes; General Trust and Universalism; Contact with other cultures; Knowledge and Effect of Policies.
- Key quantitative result:
  - Beliefs and perceptions explain about 80 percent of support for policies.
  - Sensitivity (PMR restricted to control group): beliefs and perceptions account for 81 percent of policy support while individuals’ characteristics explain 9 percent (compared to 80 percent and 6 percent, respectively, in the baseline); knowledge and perceptions about the effect of policies explain 29 percent of policy support (compared to 37 percent in the baseline). Variance decomposition for migrant integration remains essentially unchanged when restricted to the control group.

### Correlates of policy support (summary)
- PMR reforms:
  - Lower support among those working in, or who know someone working in, public utility companies.
  - Support tends to increase with age, income, and education.
  - Support tends to decrease among women, respondents with children, and those identifying as left-leaning.
- Migrant integration policies:
  - Support tends to increase with education and with familiarity with immigrants and other cultures.
  - Politically right-leaning respondents are more likely to oppose immigration policies.
  - Respondents working in low-skilled occupations are more likely to oppose migrant integration policies.

### Information strategies and treatment effects
- Three illustrated slideshow treatments:
  - Status-quo treatment: raises awareness about costs of maintaining the status quo.
  - Effect of Policies treatment: provides information about beneficial effects of policies.
  - Immigrants’ stories (counter-stereotype treatment): presents real-life immigrant experiences to counter cultural/value difference narratives.
- Treatment coverage:
  - PMR: Mexico, Morocco, South Africa.
  - Migrant integration: Canada, Italy, United Kingdom.
- Main empirical findings:
  - Information strategies can significantly boost support for reforms.
  - Explaining the effect of policies (Effect of Policies; Status quo + Effect of Policies) on outcomes people care about can significantly increase reform support.
  - The Effect of Policies + Mechanism treatment yields even stronger effects, especially among individuals with misperceptions about immigrants and politically right-leaning respondents.
  - Results broadly consistent at the country level (Annex Table 3.3.1 and 3.3.2 summaries).

### Examples of treatments shown
- Effect of Policies (migrant integration survey snapshot, United Kingdom):
  - Notes that immigrants can wait from a few months to several years before receiving permission to work; research finds when immigrants start working soon after arrival, the economy gets bigger and stronger (studies cited include Harvard University and the International Monetary Fund).
- Immigrants’ Stories (migrant integration survey snapshot):
  - Example: Kabir, a software engineer from India who immigrated to Canada, applied for more than 100 IT jobs over two years with no success due to lack of local experience; after attending night courses, he found a job as IT technical support in a bank.

### Experimental evidence on information treatments and mechanisms
- Regression framework: Equation 3.3.1 estimated with y capturing respondents’ beliefs about policy effects (price, quality, access for PMR; jobs, public finances, crime for migrant integration).
- PMR survey (Annex Table 3.3.1, columns 7–9):
  - The Status quo + Effect of Policies treatment significantly shifts perceptions regarding the effect of competition on cost, quality and access.
  - Example: share of respondents who believe fostering competition positively impacts access increases by about 3 percentage points.
- Migrant integration survey (Annex Table 3.3.2, columns 6–8):
  - All treatments shift beliefs about impact on public finances, crime rates, and jobs for native workers.
  - Particularly large effect on the share who believe providing immigrants with tools to work can reduce crime rates.
- Channel conclusion:
  - Explaining the effect of policies and how they work is a crucial channel for increasing reform support across policy areas.

### Effects on stated willingness to act (petition questions) and behavioral proxy
- Petition outcomes:
  - PMR survey: Information treatment significantly increased the share inclined to endorse a petition supporting the reform (column 10) and decreased the share willing to sign a petition against the reform (column 11).
  - Migrant integration survey: Only the share prepared to sign a petition against integration reforms significantly decreased because of the treatment (column 10).
  - Treatment more effective among respondents initially opposed to integration policies, notably those with biased perceptions and politically right-leaning respondents.
- Caution:
  - Gap between intentions and actions exists; skepticism about petition impact, privacy concerns, need for more information, and reluctance to engage in follow-up actions can affect translation to real-world behavior.

### Sample, estimation choices, weighting, and robustness
- Baseline estimation: unweighted Ordinary Least Squares (OLS) used to estimate Equation 3.3.1.
  - Rationale: respondents selected from representative preselected pools; no clear endogenous sampling problem; using weights can reduce precision (reference Solon, Haider and Wooldridge 2015).
- Deviations from demographic targets:
  - PMR survey overrepresents higher education, urban residents, and those employed in some countries.
  - Migrant integration survey: representativeness high in Italy; oversamples employed and high-income people in Canada and United Kingdom.
- Robustness check: Weighted Least Squares (WLS) matching within-country distributions; main results remain broadly unchanged.
- Heterogeneity exploration:
  - Divergence dimensions: education, urbanicity and unemployment for PMR; unemployment and income for migrant integration. Analysis does not reveal significant changes.

### Mitigating measures tested to address respondents’ concerns (Annex 3.3.3)
- Experimental design: respondents asked why they do not support policy change and whether they would support reform if government committed to hypothetical mitigating measures.
- PMR mitigating measures:
  - Creation of an independent regulatory agency for cost, quality, and national security concerns.
  - Government commitment to ensuring price affordability for the poorest households and adequate coverage across the country.
  - Temporary jobs protection and job-training programs for affected workers.
- Migrant integration mitigating measures:
  - (i) active labor market policies to facilitate reallocation of workers;
  - (ii) temporarily lower tax rates or social security payments to provide financial relief for negatively affected workers;
  - (iii) measures to improve skill and geographical allocation of migrant workers, including migrant-tailored online platforms for job search and mentoring programs and cross-country coordination.
  - Commitment to adequate expansion of public infrastructure to address overcrowding concerns.
  - Public support to facilitate construction of new housing or improve geographical allocation of immigrants.

### Persistent opposition and primary reasons for continued resistance (Annex Table 3.3.3)
- Among respondents who would still oppose reform even with mitigating measures, most cited reasons relate to trust and skepticism about implementation ability.
- PMR reforms — “Ultimate Reason for Non-Support” (Percent):
  - Don’t want the private sector or foreign investors to control the provision of services: 43.83
  - Don’t trust the private sector: 35.86
  - Don’t trust the government’s willingness or ability to implement good reforms: 18.19
  - Other reasons: 2.12
- Migrant integration policies — “Ultimate Reason for Non-Support” (Percent):
  - Don’t trust the government's willingness or ability to implement good reforms: 53.94
  - Don’t want foreign workers in the country: 14.90
  - Doubt effectiveness or feasibility of policies or mitigating measures: 10.97
  - Concerns about jobs: 7.28
  - Fiscal constraints: 6.86
  - Other reasons: 6.05

### Country case selection for EPL reform episodes (Online Annex 3.4)
- Selection approach:
  - Major EPL reforms identified where change in aggregate EPL index exceeded two standard deviations of the average change in the IMF structural reform database.
- Sample of historical EPL reform episodes and characteristics (entries as presented):
  - Bolivia (1985): LIC; Trigger: Economic crisis with hyperinflation; EPL Change: Major; Other Labor Policy Focus: Wage bargaining; Other Major Reforms within Two Years: Currency, fiscal, trade, SOE privatization, business deregulation, financial, etc.; Reform Status: Reversed in 2006.
  - Brazil (2017): EME; Trigger: Economic crisis, new government mandate to reduce high labor cost and rigidity; EPL Change: Minor; Other Labor Policy Focus: Labor litigation, collective bargaining, outsourcing; Other Major Reforms within Two Years: Progress made in SOE privatization; Reform Status: Implemented with some resistance.
  - Denmark (1990s): AE; Trigger: High unemployment rate; EPL Change: Minor; Other Labor Policy Focus: ALMP, Unemployment benefits; Other Major Reforms within Two Years: Education, housing, healthcare, pension; Reform Status: Implemented and sustained.
  - France (2015–17): AE; Trigger: High unemployment rate; EPL Change: Minor; Other Labor Policy Focus: Collective bargaining, temporary work; Other Major Reforms within Two Years: Education, training, housing, unemployment insurance, corporate tax; Reform Status: Implemented and sustained; further planned reforms in the area slightly delayed.
  - Georgia (2006): LIC; Trigger: New government mandate; EPL Change: Major; Other Labor Policy Focus: Working hours and leaves; Other Major Reforms within Two Years: Tax and customs, business deregulation, judicial, education, healthcare, land rights, anti-corruption; Reform Status: Partially unwound in 2013.
  - Germany (2003–05): AE; Trigger: High unemployment rate; EPL Change: Minor; Other Labor Policy Focus: ALMP, unemployment benefits, temporary work; Other Major Reforms within Two Years: Education, healthcare, pension, corporate tax; Reform Status: Implemented with some resistance.
  - India (2014–20): EME; Trigger: New government mandate to address long-standing structural weaknesses; EPL Change: N/A; Other Labor Policy Focus: Minimum wage, working conditions; Other Major Reforms within Two Years: Tax system, demonetization, insolvency, financial inclusion, housing, agricultural, health insurance; Reform Status: Legislated in 2020 but not yet fully implemented.
  - Korea (2016): AE; Trigger: Need to boost growth and competitiveness in the face of ageing population; EPL Change: Minor; Other Labor Policy Focus: Minimum wage, working hours, youth employment; Other Major Reforms within Two Years: Corporate governance, education, environmental, and energy; Reform Status: Largely withdrawn due to resistance.
  - Mexico (2012): EME; Trigger: New government mandate; EPL Change: Minor; Other Labor Policy Focus: Collective bargaining; Other Major Reforms within Two Years: Energy, telecommunication, fiscal, financial, political, and electoral; Reform Status: Implemented and sustained.
  - Peru (2008): EME; Trigger: Desire to reduce informal employment by pro-reform government; EPL Change: Minor; Other Labor Policy Focus: Flexible working hours and easing regulation; Other Major Reforms within Two Years: Fiscal transparency, education, healthcare, pension, judicial, anti-corruption; Reform Status: Implemented with adjustments.
  - Vietnam (2012): LIC; Trigger: Economic considerations as part of transition to market economy; EPL Change: Minor; Other Labor Policy Focus: Regulations on labor contracts and labor protection; Other Major Reforms within Two Years: Education, anti-corruption, business deregulation, social security, and welfare; Reform Status: New Labor Code enacted in 2012 and sustained.

### Ancillary empirical details and table diagnostics (selected figures from Annex Table 3.3.1)
- Number of observations (columns): 3,186; 3,101; 6,287; 2,095; 2,096; 2,096; 6,286; 6,286; 6,282; 6,283; 6,2816,287 (as presented in table).
- R2 (columns): 0.37; 0.32; 0.35; 0.39; 0.29; 0.27; 0.16; 0.16; 0.17; 0.17; 0.15; 0.33 (as presented in table).
- Coefficient examples (as presented with significance markers and standard errors in parentheses):
  - Status quo: 0.08** (0.04); -0.00 (0.04); 0.03 (0.03); 0.09** (0.05); -0.00 (0.05); 0.01 (0.05); -0.02 (0.03); -0.07** (0.03); 0.00 (0.03); 0.00 (0.01); 0.00 (0.01); 0.00 (0.04).
  - Additional coefficient block: 0.41*** (0.04); 0.27*** (0.03); 0.34*** (0.03); 0.36*** (0.04); 0.31*** (0.05); 0.34*** (0.05); 0.34*** (0.03); 0.24*** (0.03); 0.23*** (0.03); 0.09*** (0.01); -0.07*** (0.01); 0.31*** (0.05).
- Table notes:
  - Regressions control for individual characteristics, beliefs, country fixed effects, and pre-treatment support; robust standard errors in parentheses.
  - Symbols *, **, and *** indicate significance at the 10 percent, 5 percent, and 1 percent levels, respectively.

*Source: IMF staff calculations and text from CHAPTER 3 — UNDERSTANDING THE SOCIAL ACCEPTABILITY OF STRUCTURAL REFORMS (Online Annexes).*

### CHAPTER 3  UNDERSTANDING THE SOCIAL ACCEPTABILITY OF STRUCTURAL

### CHAPTER 3  UNDERSTANDING THE SOCIAL ACCEPTABILITY OF STRUCTURAL REFORMS — Online Annexes

### Narrative dataset construction and methodology
- Information sources and extraction:
  - The information was extracted from the EIU reports using OpenAI’s GPT-4o model.
  - The model’s performance was evaluated iteratively against human-generated answers until achieving an overall accuracy of approximately 80 percent in the test sample.
  - The model was tasked with answering questions related to six aspects of reforms: (i) scope and direction of reform; (ii) timing and duration; (iii) actors; (iv) consultations/communications strategies; (v) complementary/compensatory measures; and (vi) social resistance and government response.
- Definitions and episode rules:
  - A reform episode is defined as a continuous period of reform efforts within the same reform area and with a direction consistent with promoting economic growth (examples: increasing labor supply among older workers for elder LP; enhancing competition and private participation for PMR; improving integration of foreign-born workers for Migrant Integration).
  - A reform attempt is considered implemented if at least one EIU report during the reform episode mentions an implementation of the relevant reform measures.
  - Distinct reform episodes are separated by periods of at least two consecutive quarters with no discussion, adoption, or implementation within that reform area.
  - The narrative database records whether consultation/communications strategies or complementary/compensatory measures were prominent enough to be discussed in the EIU reports; absence in the narrative database does not necessarily indicate these measures were not employed, nor does it indicate intensity or adequacy.

### Descriptive counts and episode inventory
- Online Annex Table 3.2.1. Number of Reform Episodes, 1996–2023 (as presented):
  - AEsEMEsLICsTotal
  - PMR-Electricity761798 5340
  - Elder LP77482127
  - Migrant Integration1147019203

### Social resistance, reversals, and implications
- Definition of reversal:
  - A reversal is the implementation of a policy in the opposite direction of a previously identified reform implementation (examples: decreasing minimum retirement age for elder LP; introducing entry barriers for PMR-electricity; obstructing labor market integration of foreign-born workers for Migrant Integration).
- Key descriptive finding:
  - Among PMR-electricity and elder LP reforms, a significantly higher share of reforms that were implemented with resistance were subsequently reversed within five years, compared with those implemented without resistance (Online Annex Figure 3.2.1).
  - Statistical significance markers in the figure: ***, **, and * denote significance at the 1, 5, and 10 percent levels, respectively, for t-tests comparing reversal rates between resisted and unresisted implemented reforms.

### Econometric framework and controls
- Model:
  - Multinomial logistic regression framework specified in equations (3.2.1) and (3.2.2).
  - Three outcome categories (j = 0,1,2):
    - j = 0: reform episodes that ended without a reform implementation (discussed but never implemented) — base outcome.
    - j = 1: reforms implemented but resisted.
    - j = 2: reforms implemented without resistance.
  - Z_iut is the vector of determinants; X_iut denotes controls; α_g(i) denotes country income group fixed effects where g(i) ∈ {AEs, EMEs, LICs}.
- Controls and data sources (Online Annex Table 3.2.2 summary):
  - Population age shares — World Bank, World Development Indicators
  - Real GDP growth — IMF, World Economic Outlook database
  - Financial crises — Laeven and Valencia (2020); Nguyen and others (2022)
  - Inflation — International Monetary Fund, World Economic Outlook database
  - Fiscal space — Kose and others (2022)
  - Exchange rate arrangements — Ilzetzki and others (2019)
  - Fiscal rules — Davoodi and others (2022)
  - Monetary framework — Romelli (2022)
  - Globalization index — Gygli and others (2019)
  - Trade flows and trade partners — Conte and others (2023)
  - IMF program — IMF, Monitoring of Fund Arrangements (MONA)
  - Political factors — Scartascini, Cruz, and Keefer (2021); Herre (2023)
  - Income inequality — Solt (2020)
  - Electoral democracy index — Coppedge and others (2024)

### Main econometric findings (Annex Table 3.2.3 highlights)
- General:
  - Variables are grouped into seven categories: macroeconomic context, macroeconomic framework, external factors, political factors, reform strategies, and control variables.
  - Each coefficient in the table indicates the relative risk ratio (exponent of γ_j) for a one-unit increase in the explanatory variable; values greater (lower) than 1 indicate increased (decreased) odds of implementation relative to the base outcome.
- Selected coefficients and statistical significance (as reported in the table; values report odds ratios with robust standard errors in parentheses):
  - Macroeconomic context:
    - GDP growth deviation from 10-year average: 0.910** (0.042) for Imp (R) in column (1).
    - Recession: 0.220** (0.155) for Imp (R) in column (1).
    - Financial crisis: 2.090** (0.624) for Imp (R) in column (1).
  - Macroeconomic framework:
    - Inflation deviation from 1-year average: 1.001 (0.001) in column (1) and 0.998*** (0.001) in column (2).
    - Strong fiscal position: 1.084 (0.305) in column (1).
    - Strong fiscal/monetary framework: 1.354 (0.333) in column (1) and 1.425* (0.276) in column (2).
  - External factors:
    - Share of countries implementing the reform in the past 5 years: 1.042 (0.610) in column (1) and 3.245** (1.593) in column (2).
    - Globalization index: 0.892*** (0.027) in column (1).
    - Engaged in regional trade agreements: 1.000 (0.006) in column (1).
    - IMF program: 0.789 (0.337) in column (1) and 1.130 (0.304) in column (2).
  - Political factors:
    - Chief Executive, years in office: 1.045* (0.025) in column (1).
    - Year of election: 0.371** (0.159) in column (1) and 0.534*** (0.128) in column (2).
    - Years until next election: 1.166 (0.145) in column (1).
    - Large margin of majority: 1.395 (0.382) in column (1) and 1.595*** (0.258) in column (2).
    - Leader ideology (left): 0.689** (0.100) in column (1).
  - Reform strategies (notable strong associations with implementation):
    - Consultation/communication strategy: 3.477*** (1.055) in column (1); 1.295 (0.271) in column (2); 4.306** (2.702) in column (3).
    - Complementary/compensatory measures: 3.652*** (1.096) in column (1); 0.982 (0.201) in column (2); 3.723** (2.007) in column (3); 3.402*** (1.340) in column (4); 2.162* (0.882) in column (5); 1.627* (0.415) in column (6).
- Interpretation note:
  - Coefficients report changes in the odds ratio of reform implementation outcomes. Value greater (smaller) than 1 indicates increase (decrease) in the odds ratio relative to the baseline outcome (reforms that were discussed but never implemented).
  - Estimations include controls for lagged reform index for PMR-Electricity, exports of goods and services as a share of GDP, share of population over 65 years old, deviation of Gini coefficient from 10-year average, electoral democracy index, and country income group fixed effect.
  - ***, **, and * denote significance at the 1, 5, and 10 percent level, respectively.

### Marginal effects and role of reform strategies (Annex Figure 3.2.2 and dominance analysis)
- Marginal effects (average marginal effects, calculated at the observation level then averaged):
  - Consultation and communications strategies:
    - Associated with a 6-8 percentage points higher probability of implementation in the case of resisted reforms (reported in the text summarizing Annex Figure 3.2.2).
  - Compensatory and complementary measures:
    - Associated with a 5-8 percentage point increase in the probability of resisted reforms being implemented.
    - Associated with a 6 percentage point increase of implementation in the case of unresisted elder LP and migrant integration reforms.
  - The marginal effects are shown with 90 percent confidence intervals in the figure.
- Dominance analysis (Figure 3.4 referenced in the chapter):
  - Reform strategies remain the most important in predicting reform outcomes.
  - Joint explanatory power of reform strategies ranges from 25 to 45 percent across reform areas (as reported in the annex text).
  - Robustness check: dominance analysis re-estimated by including determinants individually (rather than in clusters) confirmed reform strategies’ prominence.

### Survey experiments on public attitudes (Online Annex 3.3)
- Overview of the two surveys:
  - Purpose: study drivers of people’s attitudes towards reforms using large-scale online surveys with embedded randomized control trial experiments.
  - Survey 1: PMR reforms (product market regulation) — covers Mexico, Morocco, and South Africa (focus on reforms to ease PMR and network sectors; sample selection conditioned on feasibility of representative online surveys).
  - Survey 2: Migrant integration policies — covers Canada, Italy, and the United Kingdom (focus on policies to integrate foreign-born workers into the labor market; participants required to be born in the country).
- Survey execution and sample:
  - Conducted between June 7 and July 12, 2024, by YouGov.
  - Institutional source of questionnaire not revealed to respondents to prevent response biases.
  - Total respondents: 12,600 (2,100 per country).
  - Respondent selection: representative preselected pool based on each country’s demographic profile along age, gender, education, employment status, and regional distribution; propensity score matching applied to improve alignment with the population sample.
  - Mode: all surveys conducted online with residents aged 18 years and older; administered in local languages.
  - Data validation: standard checks and protocols employed to validate data and ensure reliability.
- Survey context notes:
  - PMR survey countries: regulatory stance in network sectors is relatively tight compared to peers; referenced databases include IMF Structural Reform Database and World Bank-OECD database.
  - Migrant integration survey countries: each has experienced significant inflows of different types of immigrants; 2020 Migrants Integration Policy Index scores cited for context (Canada 76/100 in immigrants labor market mobility and 80/100 overall; Italy 67/100 and 58/100; United Kingdom 48/100 and 56/100).
  - Pre-registration and certification: both surveys pre-registered in AsPredicted and certified by the German Association for Experimental Economic Research.

*Source: IMF staff calculations and text from CHAPTER 3 — UNDERSTANDING THE SOCIAL ACCEPTABILITY OF STRUCTURAL REFORMS (Online Annexes).*

### CHAPTER 3  UNDERSTANDING THE SOCIAL ACCEPTABILITY OF STRUCTURAL

### CHAPTER 3  UNDERSTANDING THE SOCIAL ACCEPTABILITY OF STRUCTURAL REFORMS

### Survey design and outcome measures
- Common questionnaire features:
  - Collects socioeconomic information on respondents.
  - Elicits beliefs about societal values, institutions, and policies related to the reform context.
  - Randomly assigns respondents to one of multiple information treatments or to a control group (no information).
  - Gathers responses to construct three main outcome variables: support for policies; perception of the effects of the policies; and willingness to sign a petition in support or against the reform.
- Policy support measures:
  - PMR reforms questionnaire: Respondents asked “Do you support or oppose allowing private companies to produce and sell [electricity / telecommunication services] in the country” using a Likert scale from strongly support to strongly oppose.
  - Migrant integration policies questionnaire: Respondents given four policies (work authorization on arrival; quicker recognition of foreign qualifications; job training and search assistance; support for immigrant entrepreneurship). Asked “Would you support or oppose the following policies to help immigrants find jobs in the country?” for each policy on a scale from strongly support to strongly oppose. Answers combined into a country-specific index following Anderson 2008, further standardized across countries.

### Perceptions and belief elicitation
- PMR perceptions:
  - Overall perception: “Do you view it as beneficial or detrimental for consumers when several companies compete to provide services such as [electricity / telecommunications] ?” with responses from highly beneficial to highly detrimental.
  - Three-dimension question: “How do you think private companies competing to provide [electricity / telecommunication] services change things? Think about the cost, the quality and access to the [electricity / telecommunication] services?” with options from significantly cheaper or better, to significantly pricier or worse for each dimension.
- Migrant integration perceptions:
  - Question: “Imagine that the government implemented measures to integrate immigrants in the labor market. How do you think this would affect the following elements in your country?” Elements: (i) jobs for people born in the country, (ii) money the government gets from taxes, (iii) crime immigrants commit. Responses range from strongly decrease to strongly increase. An index is created based on these answers.
- Stereotypes and narratives:
  - Stereotypes elicited via open-ended questions about immigrants’ traits (hard-working, lazy, dangerous, friendly, refugees, illegal immigrants) and perceived economic impact; analyzed using large language models. Annex Figure 3.3.1 provides an overview by country.

### Empirical model
- Main regression specification (Equation 3.3.1):
  - 푦푦
    푖푖표표
    =  훽훽
    1
    푇푇
    푖푖표표
    +  훽훽
    2
    풔풔풔풔풔풔
    풊풊표표
    +  훽훽
    3
    풃풃풃풃풃풃
    풊풊표표
    +  휃휃
    표표
    +  휀휀
    푖푖표표
    (3.3.1)
  - 푦푦
    푖푖표표
    is the main variable of interest for individual 푖푖 in country 푢푢 (support for PMR reforms or migrant integration policies; beliefs about policy effects; willingness to sign a petition).
  - 푇푇
    푖푖표표
    are treatment indicators testing three information strategies.
  - 풔풔풔풔풔풔
    푖푖표표
    captures individuals’ socioeconomic characteristics (demographics, income, education, employment, social engagement, political leaning).
  - 풃풃풃풃풃풃
    푖푖표표
    includes beliefs and perceptions (trust in government and institutions; views on government’s role in the economy; stereotypes about immigrants).
  - 휃휃
    표표
    denotes country fixed effects; 휀휀
    푖푖표표
    is the error term.

### Role of beliefs versus socioeconomic characteristics
- Dominance analysis (Luchman, 2021) used to decompose the share of variance in policy support explained by predictors (socioeconomic characteristics; treatment indicators; country fixed effects; beliefs and perceptions), implementing the approach described in Dechezleprêtre and others (2022) and Dabla-Norris and others (2023).
- Clustering of belief/perception variables:
  - For PMR reforms: Market Economy; Distribution and Equity; Trust and Corruption; Views and Satisfaction with the Sector; Knowledge and Effect of Policies.
  - For migrant integration policies: Stereotypes; General Trust and Universalism; Contact with other cultures; Knowledge and Effect of Policies.
- Key quantitative result:
  - Overall, beliefs and perceptions explain about 80 percent of support for policies.
  - Sensitivity: Restricting the dominance analysis for PMR reforms to the control group yields beliefs and perceptions account for 81 percent of policy support while individuals’ characteristics explain only 9 percent (compared to 80 percent and 6 percent, respectively, in the baseline regression); knowledge and perceptions about the effect of policies explain 29 percent of policy support (compared to 37 percent in the baseline regression). The variance decomposition results for migrant integration policies remain essentially unchanged when restricted to the control group.

### Correlates of policy support (summary of Annex Figures 3.3.2 and text)
- PMR reforms:
  - Lower support among those working in, or who know someone working in, public utility companies.
  - Support tends to increase with age, income, and education.
  - Support tends to decrease among women, respondents with children, and those identifying as left-leaning.
- Migrant integration policies:
  - Support tends to increase with education and is higher for people familiar with immigrants and other cultures (immigrant friends, lived abroad, charity work).
  - Politically right-leaning respondents are more likely to oppose immigration policies.
  - Respondents working in low-skilled occupations are more likely to oppose migrant integration policies (potential labor market competition concerns).

### Information strategies and treatment effects
- Three information strategies tested via illustrated slideshow treatments:
  - Status-quo treatment: raises awareness about costs of maintaining the status quo (costs of non-reforming) in electricity and telecommunications — includes information on price, quality, and access.
  - Effect of Policies treatment: provides information about beneficial effects of policies (for PMR and migrant integration).
  - Immigrants’ stories (counter-stereotype treatment): presents real-life immigrant experiences and struggles in the labor market to counter cultural/value difference narratives.
- Treatment coverage by survey:
  - PMR survey covers Mexico, Morocco, and South Africa.
  - Migrant integration survey covers Canada, Italy, and United Kingdom.
- Main empirical findings (Annex Tables 3.3.1 and 3.3.2):
  - Information strategies can significantly boost support for reforms.
  - Explaining the effect of policies (Effect of Policies; Status quo + Effect of Policies) on outcomes people care about (e.g., price of utility services, level of crime) can significantly increase reform support.
  - The Effect of Policies + Mechanism treatment (explaining how policies work) yields even stronger effects.
  - This deeper-mechanism information is particularly important to shift support among individuals with misperceptions about immigrants and their economic effects, as well as politically right-leaning respondents (illustrated in Annex Table 3.3.2, column 2).
  - Results are broadly consistent at the country level (Mexico, Morocco, South Africa analyses reported in Annex Table 3.3.1, columns (4)-(6)).

### Examples of treatments shown in questionnaires
- Effect of Policies (migrant integration survey snapshot, United Kingdom):
  - Text notes that in the UK, immigrants can wait from a few months to several years before receiving permission to work. Research studies find that when immigrants start working soon after they arrive, the economy gets bigger and stronger. (Studies cited include Harvard University and the International Monetary Fund.)
- Immigrants’ Stories (migrant integration survey snapshot):
  - Example story: Kabir, a software engineer from India who immigrated to Canada, applied for more than 100 IT jobs over two years with no success due to lack of local experience; after attending night courses to enhance English and IT skills, he found a job as IT technical support in a bank and is now delighted to reside in Canada with his family.

*Source: CHAPTER 3 ONLINE ANNEXES — CHAPTER 3 UNDERSTANDING THE SOCIAL ACCEPTABILITY OF STRUCTURAL REFORMS (International Monetary Fund | October 2024).*

### CHAPTER 3  UNDERSTANDING THE SOCIAL ACCEPTABILITY OF STRUCTURAL

### CHAPTER 3  UNDERSTANDING THE SOCIAL ACCEPTABILITY OF STRUCTURAL REFORMS

### Experimental evidence on information treatments and mechanisms
- Regression framework: Equation 3.3.1 estimated with 푦푦
푖푖표표 capturing respondents’ beliefs about the effect of policies on variables (price, quality, access to utility services in the PMR survey; jobs for native workers, public finances, crime rates in the migrant integration survey).
- PMR survey (Annex Table 3.3.1, columns 7–9):
  - The Status quo + Effect of Policies treatment significantly shifts perceptions regarding the effect of competition on the cost, quality and access to electricity and telecommunication services.
  - Example: the share of respondents who believe that fostering competition positively impacts access to electricity and telecommunication services increases by about 3 percentage points.
- Migrant integration survey (Annex Table 3.3.2, columns 6–8):
  - All treatments shift beliefs about the impact of policies on public finances, crime rates, and jobs for native workers.
  - Particularly large effect on the share of people who believe that providing immigrants with the tools to work can reduce crime rates.
- Channel conclusion: Explaining the effect of policies and how they work is a crucial channel for increasing reform support across very different policy areas.

### Effects on stated willingness to act (petition questions) and behavioral proxy
- Survey included hypothetical petition questions (real-stakes framing) after eliciting policy support.
- PMR survey (Annex Table 3.3.1):
  - Information treatment significantly increased the share inclined to endorse a petition supporting the reform (column 10).
  - Information treatment decreased the share willing to sign a petition against the reform (column 11).
- Migrant integration survey (Annex Table 3.3.2):
  - Only the share prepared to sign a petition against foreign-born workers’ integration reforms significantly decreased because of the treatment (column 10).
  - Treatment is more effective among respondents initially opposed to integration policies, notably:
    - Respondents with biased perceptions and negative stereotypes about immigrants.
    - Politically right-leaning respondents.
  - Annex Table 3.3.2, column 11 reports the marginal effect for individuals with negative stereotypes about immigrants (results also hold for other categories).
- Caution: gap between intentions and actions exists in literature; factors such as skepticism about petition impact, privacy concerns, need for more information, and reluctance to engage in follow-up actions can affect translation to real-world behavior.

### Sample, estimation choices, weighting, and robustness
- Baseline estimation: unweighted Ordinary Least Squares (OLS) used to estimate Equation 3.3.1.
  - Rationale: respondents selected from representative preselected pools; no clear endogenous sampling problem; using weights can reduce precision.
  - Reference: Solon, Haider and Wooldridge 2015.
- Deviations from demographic targets noted:
  - PMR survey in some countries tends to overrepresent individuals with higher education levels, urban residents, and those currently employed.
  - Migrant integration survey: representativeness high in Italy; oversamples employed and high-income people in Canada and United Kingdom.
- Robustness check: Weighted Least Squares (WLS) using weights matching within-country distribution of education, urbanicity, employment, gender, age, and region of residence.
  - Main results in last columns of Annex Tables 3.3.1 and 3.3.2 remain broadly unchanged.
- Heterogeneity exploration:
  - Dimensions examined where divergence occurs between weighted and unweighted samples: education, urbanicity and unemployment for PMR reforms; unemployment and income for migrant integration policies.
  - Analysis does not reveal significant changes, supporting appropriateness of unweighted OLS due to gain in precision.

### Mitigating measures tested to address respondents’ concerns (Annex 3.3.3)
- Experimental design: respondents asked why they do not support policy change and whether they would support reform if government committed to hypothetical mitigating measures tailored to their concerns.
- PMR reforms — mitigating measures:
  - For concerns about cost and quality, or national security risks if sector handled by private companies: creation of an independent regulatory agency (to ensure, for instance, that companies compete to deliver high-quality services at fair prices).
  - For concerns about access and affordability for the poorest: government commitment to ensuring price affordability for the poorest households and adequate coverage across the country, including in remote rural areas.
  - For concerns about effect on jobs: temporary jobs protection and job-training programs for affected workers.
- Migrant integration policies — mitigating measures:
  - For concerns about job and wage losses or fairness toward native workers:
    - (i) active labor market policies to facilitate reallocation of workers;
    - (ii) temporarily lower tax rates or social security payments to provide financial relief for negatively affected workers;
    - (iii) measures to improve skill and geographical allocation of migrant workers, including migrant-tailored online platforms for job search and mentoring programs and cross-country coordination.
  - For concerns about overcrowding of public services: commitment to adequate expansion of public infrastructure (schools, hospitals, transportation).
  - For concerns about cost of housing: public support to facilitate construction of new housing or improve geographical allocation of immigrants.

### Persistent opposition and primary reasons for continued resistance (Annex Table 3.3.3)
- Among respondents who would still oppose reform even with mitigating measures, most cited reasons relate to trust and skepticism about implementation ability.
- PMR reforms — “Ultimate Reason for Non-Support” (Percent):
  - Don’t want the private sector or foreign investors to control the provision of services: 43.83
  - Don’t trust the private sector: 35.86
  - Don’t trust the government’s willingness or ability to implement good reforms: 18.19
  - Other reasons: 2.12
- Migrant integration policies — “Ultimate Reason for Non-Support” (Percent):
  - Don’t trust the government's willingness or ability to implement good reforms: 53.94
  - Don’t want foreign workers in the country: 14.90
  - Doubt effectiveness or feasibility of policies or mitigating measures: 10.97
  - Concerns about jobs: 7.28
  - Fiscal constraints: 6.86
  - Other reasons: 6.05

### Country case selection for EPL reform episodes (Online Annex 3.4)
- Selection approach: identifies major employment protection legislation (EPL) reforms as those where the change in the aggregate EPL index in the IMF structural reform database exceeded two standard deviations of the average change in the database (approach of Alesina and others (2023a)).
- Selected country cases span different income groups and geographic regions; reforms differ in comprehensiveness, bundling with other reforms, and implementation status (allowing study of approval, implementation, and sustainability).
- Sample of historical EPL reform episodes and characteristics (exact entries as presented):
  - Bolivia (1985): LIC; Trigger: Economic crisis with hyperinflation; EPL Change: Major; Other Labor Policy Focus: Wage bargaining; Other Major Reforms within Two Years: Currency, fiscal, trade, SOE privatization, business deregulation, financial, etc.; Reform Status: Reversed in 2006.
  - Brazil (2017): EME; Trigger: Economic crisis, new government mandate to reduce high labor cost and rigidity; EPL Change: Minor; Other Labor Policy Focus: Labor litigation, collective bargaining, outsourcing; Other Major Reforms within Two Years: Progress made in SOE privatization; Reform Status: Implemented with some resistance.
  - Denmark (1990s): AE; Trigger: High unemployment rate; EPL Change: Minor; Other Labor Policy Focus: ALMP, Unemployment benefits; Other Major Reforms within Two Years: Education, housing, healthcare, pension; Reform Status: Implemented and sustained.
  - France (2015–17): AE; Trigger: High unemployment rate; EPL Change: Minor; Other Labor Policy Focus: Collective bargaining, temporary work; Other Major Reforms within Two Years: Education, training, housing, unemployment insurance, corporate tax; Reform Status: Implemented and sustained; further planned reforms in the area slightly delayed.
  - Georgia (2006): LIC; Trigger: New government mandate; EPL Change: Major; Other Labor Policy Focus: Working hours and leaves; Other Major Reforms within Two Years: Tax and customs, business deregulation, judicial, education, healthcare, land rights, anti-corruption; Reform Status: Partially unwound in 2013.
  - Germany (2003–05): AE; Trigger: High unemployment rate; EPL Change: Minor; Other Labor Policy Focus: ALMP, unemployment benefits, temporary work; Other Major Reforms within Two Years: Education, healthcare, pension, corporate tax; Reform Status: Implemented with some resistance.
  - India (2014–20): EME; Trigger: New government mandate to address long-standing structural weaknesses; EPL Change: N/A; Other Labor Policy Focus: Minimum wage, working conditions; Other Major Reforms within Two Years: Tax system, demonetization, insolvency, financial inclusion, housing, agricultural, health insurance; Reform Status: Legislated in 2020 but not yet fully implemented.
  - Korea (2016): AE; Trigger: Need to boost growth and competitiveness in the face of ageing population; EPL Change: Minor; Other Labor Policy Focus: Minimum wage, working hours, youth employment; Other Major Reforms within Two Years: Corporate governance, education, environmental, and energy; Reform Status: Largely withdrawn due to resistance.
  - Mexico (2012): EME; Trigger: New government mandate; EPL Change: Minor; Other Labor Policy Focus: Collective bargaining; Other Major Reforms within Two Years: Energy, telecommunication, fiscal, financial, political, and electoral; Reform Status: Implemented and sustained.
  - Peru (2008): EME; Trigger: Desire to reduce informal employment by pro-reform government; EPL Change: Minor; Other Labor Policy Focus: Flexible working hours and easing regulation; Other Major Reforms within Two Years: Fiscal transparency, education, healthcare, pension, judicial, anti-corruption; Reform Status: Implemented with adjustments.
  - Vietnam (2012): LIC; Trigger: Economic considerations as part of transition to market economy; EPL Change: Minor; Other Labor Policy Focus: Regulations on labor contracts and labor protection; Other Major Reforms within Two Years: Education, anti-corruption, business deregulation, social security, and welfare; Reform Status: New Labor Code enacted in 2012 and sustained.

### Ancillary empirical details and table diagnostics (selected figures from Annex Table 3.3.1)
- Number of observations (columns): 3,186; 3,101; 6,287; 2,095; 2,096; 2,096; 6,286; 6,286; 6,282; 6,283; 6,2816,287 (as presented in table).
- R2 (columns): 0.37; 0.32; 0.35; 0.39; 0.29; 0.27; 0.16; 0.16; 0.17; 0.17; 0.15; 0.33 (as presented in table).
- Coefficient examples (as presented with statistical significance markers and standard errors in parentheses):
  - Status quo: 0.08** (0.04); -0.00 (0.04); 0.03 (0.03); 0.09** (0.05); -0.00 (0.05); 0.01 (0.05); -0.02 (0.03); -0.07** (0.03); 0.00 (0.03); 0.00 (0.01); 0.00 (0.01); 0.00 (0.04).
  - Additional coefficient block: 0.41*** (0.04); 0.27*** (0.03); 0.34*** (0.03); 0.36*** (0.04); 0.31*** (0.05); 0.34*** (0.05); 0.34*** (0.03); 0.24*** (0.03); 0.23*** (0.03); 0.09*** (0.01); -0.07*** (0.01); 0.31*** (0.05).
- Table notes: regressions control for individual characteristics, beliefs, country fixed effects, and pre-treatment support; robust standard errors in parentheses; symbols *, **, and *** indicate significance at the 10 percent, 5 percent, and 1 percent levels, respectively.

*Source: IMF staff calculations based on IMF-YouGov survey.*

---


_Source: https://www.imf.org/-/media/files/publications/weo/2024/october/english/ch3onlineannexes.pdf_