## wpiea2024216-print-pdf

## Source details

**Canonical URL:** [wpiea2024216-print-pdf](https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024216-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2024/english/wpiea2024216-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2024/english/wpiea2024216-print-pdf.pdf.json)

---

### Introduction — research scope and rationale
- Focus: individuals’ beliefs and preferences toward product market regulation reforms aimed at fostering private participation and competition in network sectors (PMR reforms).
- Objectives:
  - (1) understanding the drivers of skepticism toward PMR reforms;
  - (2) examining whether research-based information on the need for and potential effects of policy changes can influence public support for reforms;
  - (3) assessing whether complementary or compensatory measures can foster greater reform support.
- Survey: large-scale online surveys of 6,300 individuals across three countries—Mexico, Morocco, and South Africa—with 2,100 respondents per country; half of respondents in each country randomly assigned to a questionnaire focusing exclusively on one sector (electricity or telecommunications).

### Key empirical findings on determinants of support
- Socioeconomic characteristics explain only 6% of individuals’ support for PMR reforms.
- Beliefs and perceptions explain about 80% of support for policies; within this:
  - Knowledge and perceptions of policies explain 37% of support.
  - Market-oriented beliefs explain 35% of support.
- Individuals more likely to support reform if they believe private competition will lead to lower prices, higher quality, or broader access.
- Views on trust and corruption significantly correlate with policy support but account for a smaller fraction of variability.
- Views about the role of government or companies in ensuring equitable access do not significantly correlate with reform support.

### Experimental evidence — treatments and their effects
- Treatment arms:
  - Control group: no information.
  - Status Quo treatment: sector-specific information benchmarking national cost, quality, and access to the United States (sources: ITU, IEA, World Bank Enterprise Surveys, GSMA). Treatment does not convey information on private participation.
  - Status Quo + Effect of Policies treatment: Status Quo information plus cross-country evidence and a simple description of findings on how private participation led to improvements in cost, quality, and access, with emphasis on stronger benefits when combined with strong regulation.
- Effects on standardized support for reforms (OLS with controls):
  - Status Quo treatment:
    - Electricity sector: coefficient 0.080** (standard error 0.040); expressed as a 0.08 increase in the standard deviation of the policy support variable. This corresponds to a 30% reduction in the policy support gap between left-wing and right-wing respondents on this issue.
    - Telecommunications sector: coefficient -0.003 (0.037), not significant.
    - Both sectors pooled: coefficient 0.031 (0.027), not significant.
  - Status Quo + Effect of Policies treatment:
    - Both sectors: coefficient 0.335*** (0.027).
    - Electricity: coefficient 0.407*** (0.041).
    - Telecommunications: coefficient 0.271*** (0.035).
    - Effects equivalent to 1.5 times the policy support gap between left-wing and right-wing respondents for electricity and 1.7 times for telecommunications when compared to the gap.
- Effects on petition behavior (petition index and individual petitions):
  - Petition index construction: -1 for respondents willing to sign petition limiting private entry; 1 for respondents willing to sign petition facilitating entry; 0 otherwise.
  - Status Quo treatment: no statistically significant impact on petition outcomes.
  - Status Quo + Effect of Policies treatment:
    - Petition index (Both): 0.235*** (0.029).
    - Electricity: increases willingness to sign petition facilitating private entry by 11.3% (an increase of 20.9% relative to control mean of 54.1%) and decreases willingness to sign petition limiting private entry by 9% (a reduction of 17.3% relative to control mean of 52.1%).
    - Telecommunications: increases willingness to sign petition facilitating private entry by 6.7% (an increase of 10.1% relative to control mean of 66.6%) and decreases willingness to sign petition limiting private entry by 5.4% (a reduction of 12.8% relative to control mean of 42.2%).

### Mechanism: treatment effects on policy beliefs (perceptions)
- Dependent perceptions: overall perceived benefit for consumers and perceived effects on cost, quality, and access.
- Status Quo treatment:
  - No significant change in overall perception in either sector.
  - Electricity: marginal increase in perception that competition lowers prices (significant at 10%).
  - Telecommunications: reductions in beliefs that competition improves quality and access (possible benchmark-to-US effect).
- Status Quo + Effect of Policies treatment:
  - Significant positive impacts on overall perception and on perceived effects on cost, quality, and access in both sectors.
  - Magnitudes (standard deviations):
    - Electricity: effects range from 0.295 to 0.410 standard deviations across the four perception measures.
    - Telecommunications: effects range from 0.189 to 0.289 standard deviations across the four perception measures.
  - Sectoral differences reported between 33.7% higher (for access perceptions) and 63.5% higher (for overall perception) in electricity relative to telecommunications.
- Link: improved perceptions under the Status Quo + Effect of Policies treatment explain its stronger effect on support and petition behavior.

### Reasons for non-support and role of mitigating measures
- Distribution of reasons for nonsupport in control group:
  - Societal concerns (affordability and access for the poorest, community impacts): 54% of responses.
  - Personal concerns (self-interest: price increases, quality, job loss): 22% of responses.
  - Other concerns (national security, uncertainty about benefits, lack of trust in evidence, unspecified): approximately 24% of responses.
- Prevalence of non-support after treatments:
  - Status Quo treatment: 60.5% (electricity) and 50.2% (telecommunications) would not support the reform.
  - Status Quo + Effect of Policies treatment: 48.5% (electricity) and 37.3% (telecommunications) would not support the reform.
- Mitigating measures offered (tailored to concerns):
  - Independent regulatory agency to ensure competition, quality, fair prices, and national security.
  - Government commitment to ensure price affordability for poorest households and nationwide coverage including remote rural areas.
  - Temporary job protection and job-training programs for affected workers.
- Indicative effectiveness (control-group respondents who initially opposed reforms):
  - 50–80% indicate they would change stance toward support if mitigating measures addressed their concerns.
  - Across all complementary and compensatory measures, the share who would opt to support the reforms exceeds 50%.
- Ultimate non-supporters (reasons and shares from Table 1):
  - Both sectors:
    - Don’t trust the private sector: 43.8
    - Don’t want private sector or foreign investors to control provision of services: 35.9
    - Don’t trust government’s willingness or ability to implement good reforms: 18.2
    - Other Reasons: 2.1
  - Electricity:
    - Don’t trust the private sector: 44.4
    - Don’t want private sector or foreign investors to control provision: 36.1
    - Don’t trust government’s willingness or ability: 17.6
    - Other Reasons: 1.9
  - Telecommunications:
    - Don’t trust the private sector: 42.2
    - Don’t want private sector or foreign investors to control provision: 35.1
    - Don’t trust government’s willingness or ability: 19.9
    - Other Reasons: 2.8

### Heterogeneity: country- and individual-level
- Country-level heterogeneity:
  - Status Quo treatment:
    - No significant variation across countries overall; exception: Mexico shows a positive and statistically significant effect (Status Quo (Mexico) coefficient 0.131* (0.068) in electricity), suggesting pooled Status Quo effect may be driven by Mexico.
  - Status Quo + Effect of Policies treatment:
    - Positive and significant across countries.
    - Electricity sector magnitude ranking: Mexico (highest), Morocco, South Africa.
    - Telecommunications sector magnitude ranking: South Africa (highest), Mexico, Morocco.
- Individual-level heterogeneity by socioeconomic status and beliefs:
  - Electricity sector:
    - Wealthier respondents (above the 75th percentile) show less sensitivity to Status Quo treatment.
    - Urban residents more responsive to Status Quo + Effect of Policies treatment.
    - Individuals in the 25th to 50th income percentiles show lower sensitivity to Status Quo + Effect of Policies treatment.
    - Individuals working in, or knowing someone working in, public utility companies are less sensitive to both treatments.
  - Telecommunications sector:
    - Female respondents and those with incomes between the 25th and 50th percentiles are more sensitive to Status Quo + Effect of Policies treatment.
    - Individuals working in, or knowing someone working in, public utility companies are less sensitive to Status Quo + Effect of Policies treatment.
  - Beliefs/perceptions heterogeneity (section 6.3):
    - Electricity: respondents who believe utility companies “should play a key role in ensuring equity and fair access” are less sensitive to both treatments.
    - Telecommunications: respondents with experiences of lack of access, perceiving improved relative deprivation, and perceiving lower corruption in utility companies are more sensitive to Status Quo + Effect of Policies treatment.
    - Double interactions: no significant double-interaction differences in electricity; in telecommunications, trust in government/institutions combined with low perceived corruption and lack of access increases sensitivity to Status Quo + Effect of Policies treatment.

### Robustness checks and data quality procedures
- Survey fieldwork: online between June 7 and July 12, 2024, conducted by YouGov.
- Sampling and representativeness:
  - Baseline sample: 6,300 respondents; 2,100 per country.
  - YouGov recruited via web advertisements, member referrals, press coverage, marketing; representative samples aligned with country demographic distributions using sampling and sample matching methodology.
  - Over-representation of tertiary education: South Africa 6%, Mexico 10%, Morocco 22%.
  - Over-representation of urban residents: South Africa 5%, Mexico 27%, Morocco 23%.
  - Under-representation of unemployed workers: South Africa around 10%; Mexico and Morocco maximum deviations 2-4 times higher than South Africa’s largest deviation.
  - Supplementary sample collected: 692 in Mexico, 336 in Morocco, 405 in South Africa.
  - Re-weighting: weighted-OLS re-weighting to match within-country distributions of age, gender, employment status, education, and geographic location; Status Quo + Effect of Policies remains statistically significant and only "about 7% lower" in the weighted sample compared to unweighted.
- Data quality and attention checks:
  - Questionnaire translated and administered in local languages (Spanish; French and Arabic; Afrikaans, English, Xhosa, and Zulu).
  - Anti-bot checks: disposable email domains, duplicate answers, suspicious email similarities.
  - Two attention checks: sports interest multiple-choice and favorite color instruction ("puce"); qualitative reviews to remove duplicates/nonsense.
  - Median time spent on survey: 21 minutes; average is 122 minutes driven by outliers; after removing top and bottom 5% of respondents, average drops to 24.5 minutes. Appendix A.2 Figure A2 shows distribution.
- Balance tests:
  - Joint likelihood ratio test p-value for Status Quo = 0.151; Tables A2 and A4 joint p-values = 0.603 and 0.866 respectively.
  - Appendices report balance tables and robustness tables (Tables A2–A11) with sample sizes and exact coefficients preserved.

### Treatment texts (illustrative excerpts used in experiments)
- Electricity — Status Quo (South Africa example):
  - "South Africans pay 68% more for electricity than people in the United States*."
  - "About 92% of the companies in South Africa report that they face electricity power cuts for about 2 hours and 20 minutes each time.*"
- Electricity — Status Quo + Effect of Policies:
  - Repeats Status Quo points above.
  - Academic findings: "Production costs and consumer tariffs fell by 10 to 30% in some Latin American countries when private firms entered the market." Example: "one study showed a 28% reduction in power outages."
  - Importance of regulation and Guatemala example: "280,000 new residential connections by private electricity distributors."
- Telecommunications — Status Quo (South Africa example):
  - "South Africans spend five times more of their monthly income on mobile data plans than people do in the United States.*"
  - "Downloading on mobile phones is 74% slower compared to the United States, and about 25% of people still can't use the internet.*"
- Telecommunications — Status Quo + Effect of Policies:
  - Repeats Status Quo points above.
  - Example quantitative finding: "In Uganda, after private mobile companies entered in the late 1990s, within just four years of a second company joining the market, mobile subscriptions increased more than twentyfold."
  - Peru example on payphones in rural areas after private investment and regulation.

### Key methodological notes and variable construction
- Outcome transformation (z-scores): two-step process following Dechezleprêtre et al. (2022):
  - First subtract control group mean and divide by control group standard deviation within each country-sector.
  - Second, standardize that index across the entire sample by subtracting the mean and dividing by the standard deviation within each sector.
  - Resulting z-scored outcomes have mean 0 and standard deviation 1.
- Support for policies coding: 1 to 5 where 1 = "Strongly oppose", 3 = "Neither oppose nor support", 5 = "Strongly support".
- Petition index coding: -1 if support petition limiting private entry, 1 if support petition facilitating private entry, 0 otherwise.
- Appendix A.5 provides precise variable definitions and questionnaire codings; Appendix A.3 lists treatment scripts and sources.

### Policy-relevant takeaways
- Messaging that explains how policies work and communicates empirical effects of PMR reforms (Status Quo + Effect of Policies) is more effective at increasing public support and generating willingness to act (petitions) than messaging that highlights only the costs of the status quo.
- Tailored complementary and compensatory measures (regulatory safeguards, affordability commitments, job protections and retraining) can substantially convert initial opponents: indicative evidence suggests 50–80% of initially opposed respondents could change to support if concerns are addressed.
- Persistent opposition is rooted mainly in lack of trust in private firms or in government’s ability/willingness to implement reforms; trust-building measures and credible institutions are therefore central to wider reform acceptance.
- Communication strategies should account for heterogeneous beliefs and perceptions across groups (e.g., views on equity, trust in government, perceptions of corruption, urban/rural status, employment ties to utilities), given their moderating role on information responsiveness.

*Italic: Source — wpiea2024216-print-pdf (excerpted content).*

### References.  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . 

### References.

### Appendix contents
- A   Appendix: page 47
- A.1   Sample Statistics: page 47
- A.2   Duration: page 48
- A.3   Treatment scripts: page 49
- A.4   Data sources used to design the treatments: page 54
- A.5   Variables definitions: page 55
- A.6   Additional descriptive statistics: page 61
- A.7   Balance Analysis: page 62
- A.8   Additional figures: page 66
  - A.8.1    Who holds positive perceptions about competition?: page 66
  - A.8.2    Role of socioeconomic characteristics and beliefs and perceptions: page 68
  - A.8.3    Characteristics of the ultimate non-supporters for reforms: page 74
- A.9   Additional tables: page 76
- A.10 Questionnaire: page 80
  - A.10.1  Survey links: page 80
  - A.10.2  Benchmark questionnaire: page 80

### Introduction — research scope and rationale
- Focus: individuals’ beliefs and preferences toward product market regulation reforms aimed at fostering private participation and competition in network sectors (PMR reforms).
- Objectives:
  - (1) understanding the drivers of skepticism toward PMR reforms;
  - (2) examining whether research-based information on the need for and potential effects of policy changes can influence public support for reforms;
  - (3) assessing whether complementary or compensatory measures can foster greater reform support.
- Survey: large-scale online surveys of 6,300 individuals across three countries—Mexico, Morocco, and South Africa—focusing on two network sectors: electricity and telecommunications. Half of respondents in each country randomly assigned to a questionnaire focusing exclusively on one sector.
- Hypothesized drivers of attitudes studied:
  - individuals’ socioeconomic characteristics (e.g., sector of employment, education level);
  - beliefs about government versus markets;
  - perceptions about the role of network sectors for equity;
  - beliefs about trust and corruption;
  - views about and satisfaction with utility services;
  - perceptions about the effect of reforms to increase competition on price, quality, and access.

### Key empirical findings on determinants of support
- Socioeconomic characteristics explain only 6% of individuals’ support for PMR reforms.
- Knowledge and perceptions of policies explain the largest share of support, followed by market-economy beliefs.
- Individuals more likely to support reform if they believe private competition will lead to lower prices, higher quality, or broader access to services.
- Views on trust and corruption significantly correlate with policy support but account for a smaller fraction of variability.
- Views about the role of government or companies in ensuring equitable access do not significantly correlate with reform support.

### Experimental evidence — information treatments and effects
- Treatments:
  - Status quo treatment: factual information on cost, quality, and access to electricity or telecommunications in respondents’ country with comparisons to the United States (sources include International Telecommunication Union, International Energy Agency, World Bank Enterprise Surveys, Global System for Mobile Communications).
  - Complementary treatment: status quo information plus research-based evidence on effects of PMR reforms on price, quality, and access (emphasizing improvements when combined with strong regulation ensuring competition and customer protection).
- Effects:
  - Status quo treatment: positive impact on support in the electricity sector; support increases by 0.08 standard deviation, corresponding to a 30% reduction in the policy support gap between left-wing and right-wing respondents on this issue. Effect for telecommunications positive but not statistically significant.
  - Complementary treatment (status quo + research evidence): stronger and statistically significant effects in both sectors when pooled; effect economically meaningful—equivalent to 1.6 times the policy support gap between left-wing and right-wing respondents when respondents on the two sectors are pooled.
  - Real-stakes validation: information treatment significantly increased likelihood to endorse a petition supporting the reform and decreased likelihood to sign a petition against the reform.

### Reasons for nonsupport and role of mitigating measures
- Main reasons for nonsupport: societal concerns (consistent with models of social preferences).
- Mitigating measures: tailored complementary and compensatory measures can significantly foster support.
  - Indicative (not strictly causal) evidence: 50–80% of respondents in the control group initially opposed to PMR reforms indicate they would change stance toward support if mitigating measures addressed their concerns.
  - Mitigating measures particularly effective for individuals fearing job losses (e.g., workers in public utility companies or those with close connections to them).
- Persistent opposition: lack of trust is the prevailing motive among those who would not support reforms even with mitigating measures; continued opposers mostly cite trust in government and the private sector and doubts about institutions’ ability to implement reforms or mitigating measures effectively.

### Literature linkages and contribution
- Connects to political economy of reforms literature (e.g., Tompson, 2009; OECD, 2010) on the “cost of the status quo” obstacle.
- Builds on evidence-based information literature on individual policy preferences (e.g., Rodriguez Chatruc et al., 2021; Douenne and Fabre, 2022; Duval et al., 2024).
- Distinction from prior privatization skepticism studies: uses a dedicated, multi-country survey focused specifically on hypothetical PMR reforms and experimentally evaluates information effects and mitigating measures.

### Survey implementation and sample details
- Fieldwork: online between June 7 and July 12, 2024, conducted by YouGov.
- Countries: Mexico, Morocco, South Africa.
- Sample: baseline sample of 6,300 respondents, with 2,100 respondents per country; respondents required to be residents and 18 years or older.
- Recruitment and representativeness: YouGov panel recruited via web advertisements, member referrals, press coverage, marketing; representative samples aligned with country demographic distributions (age, gender, education, employment status, regional distribution) using sampling and sample matching methodology.
- Incentives: respondents rewarded upon successful completion, varying by country.
- Pre-analysis plan and ethics: pre-analysis plan registered on AsPredicted (#178272) and an ethics certificate granted by the German Association for Experimental Economic Research under their expedite review procedure.
- Regulatory stance context: regulatory stance in the tighter network sector in the three countries ranks between the 6th and 52nd percentiles among 46 emerging market peers according to the IMF Structural Reform Database; for Mexico and South Africa, regulatory stance (entry barriers) ranks between the 12th and 24th percentiles among 67 advanced and emerging market peers per World Bank-OECD database.
- Time spent on survey: median time is 21 minutes; average is 122 minutes driven by outliers; after removing top and bottom 5% of respondents, average drops to 24.5 minutes. Appendix A.2, Figure A2 shows the cumulative density function of time spent.

*Source: wpiea2024216-print-pdf - References. (Content extracted from the provided PDF excerpt.)*

### 2.2    Sample

### 2.2    Sample

### Key points on sampling and representativeness
- YouGov’s online targeting approach aims to identify a representative sample but its online nature results in deviations from demographic targets, especially in emerging market economies and developing countries where network infrastructures are unevenly developed.
- Degree of deviation from population distribution varies by country; the sample best represents the population in South Africa.
- The sample tends to over-represent individuals with tertiary education and urban residents, and under-represent those living in rural areas, consistent with patterns documented for middle-income countries.

### Exact sample deviations reported
- Over-representation of respondents with tertiary education:
  - South Africa: 6%
  - Mexico: 10%
  - Morocco: 22%
- Over-representation of urban residents:
  - South Africa: 5%
  - Mexico: 27%
  - Morocco: 23%
- Under-representation of unemployed workers:
  - South Africa: around 10%
  - Mexico and Morocco: maximum deviations are 2-4 times higher than South Africa’s largest deviation

### Procedures and planned adjustments
- The sampling and sample matching approach is a two stage process: first a random sample of the target population is chosen (sampling frame) and, in a second step, survey respondents are matched with the sampling frame using propensity scores.
- A supplementary sample was collected to augment baseline analysis; it is described later in the robustness checks section.
- In Section 7, main results are reproduced by re-weighting the samples within each country to match the distribution in the population.

*Italic: Source — wpiea2024216-print-pdf (section 2.2–3.1).*

### 2.3    Data Quality

### Measures to ensure comprehension and contextual relevance
- Questionnaire translated and administered in local languages in each country.
  - Languages used: Spanish in Mexico; French and Arabic in Morocco; Afrikaans, English, Xhosa, and Zulu in South Africa.
  - Respondents could select preferred language at the beginning of the survey.
- Treatment slideshow text and illustrations tailored to each country; status quo information based on country-specific data; illustrative images included each country’s national flag and map.
- Two pre-tests of the questionnaire were conducted with small groups: once on Prolific and once on YouGov.

### Measures to minimize bias and priming
- Survey administered by YouGov without disclosing the institutional source of the questionnaire to prevent institutional influence on responses.
- Post-treatment randomization implemented to reduce priming:
  1. The order of the two petitions to increase or limit private entry in the network sectors.
  2. The choices in questions regarding reasons for not supporting private operations in the network sectors.

### Screening and attention checks
- Anti-bot checks by YouGov: checks for disposable email domains, duplicate answers in open-ended questions, and suspicious similarities between email addresses.
- Two attention check questions unrelated to survey content:
  - First attention check: follow specific instructions to answer a multiple-choice question about level of interest in sports (used for robustness analysis).
  - Second attention check: state favorite color according to given instructions (used to screen out inattentive respondents in the baseline sample).
- YouGov conducted qualitative reviews to remove duplicate entries, nonsense responses, and other suspicious entries.

*Italic: Source — wpiea2024216-print-pdf (section 2.3).*

### 2.4    Real-stakes Questions

### Rationale and inclusion
- Recognizing that self-reported preferences may not align with actual behavior, the survey includes real-stakes questions: respondents were asked if they would be willing to sign a hypothetical petition either in support of or against policies to foster private participation in utility sectors.

### Correlations between petition signing and attitudes
- Figure 1 presents regression coefficients of petition signing on policy support and perceptions, controlling for socioeconomic characteristics and country fixed effects; analyses conducted on the entire sample and sub-samples for the electricity and telecommunications sectors.
- Key associations:
  - Respondents showing stronger support for PMR reforms are about 20% more likely to sign a petition in support of policies fostering private participation.
  - Those respondents are around 15% less likely to sign a petition limiting private participation.
  - Correlations are qualitatively similar but less pronounced for respondents having more positive perceptions about the effects of private participation.

*Italic: Source — wpiea2024216-print-pdf (section 2.4).*

### 2.5    Questionnaire Structure

### Overview of survey sections (7 main sections)
- Socioeconomic Questions:
  - Collected age, gender, race, region of residence, employment, language, education, and urban/rural status.
  - Demographics used for analysis and quota targeting; placed at beginning.
- Pre-treatment Beliefs:
  - Four groups of questions: societal trust and perceptions about corruption, views about market economy and government intervention, sector-specific knowledge and satisfaction, and perceptions of private participation and pre-treatment support.
  - Example questions: “Would you say the governments in Mexico/Morocco/South Africa can or cannot be trusted, no matter which political party is in charge?” and “How much do you think the government in Mexico/Morocco/South Africa should intervene in how private companies set prices for the goods and services they sell?”
  - At the start of the third set of pre-treatment belief questions, half of respondents in each country were randomly assigned to focus exclusively on either the electricity or telecommunications sector.
- Treatments:
  - Random assignment into one of two information treatments or a control group; treatment scripts and sources reported in Appendix A.3.
- Post-treatment beliefs, support for policies, and willingness to act:
  - Elicited post-treatment beliefs about effects of competition on consumers (overall and on cost, quality, access).
  - Gauged support for private participation and included real-stakes petition questions (two petitions: one facilitating entry of private firms, one limiting entry).
- Reasons behind support or opposition:
  - Respondents could select personal reasons, societal reasons, or other reasons (including national security concerns with an open box).
  - Follow-up: supporters asked if they’d still support if government sold its own businesses; opponents asked if they’d reconsider if government committed to complementary and compensatory measures; persistent opponents asked to specify reason.
- Additional socioeconomic questions, perceptions of the survey, and links:
  - Additional controls for heterogeneous treatment effects placed toward the end.
  - Final section on respondents’ perceptions of the survey (perceived political bias, trustworthiness, easiness to understand, length) used for robustness analysis.
  - Treated respondents who expressed interest were provided links to sources and studies referenced in treatments.

### Data definitions and balance checks
- All variables used and constructed from the questionnaire are defined in Appendix A.5.
- Descriptive statistics for main variables (support for policies, petitions) are in Appendix A.6.
- Appendix A.7 shows socioeconomic characteristics are broadly balanced between control and treatment groups.

*Italic: Source — wpiea2024216-print-pdf (section 2.5).*

### 3    What Explains Support for PMR reforms?

### Research focus
- Investigates respondents’ perceptions of allowing private companies to compete to provide electricity and telecommunications, including effects on service price, quality, and access.
- Explores the role of these policy-effect perceptions in explaining support for PMR reforms, comparing them with other beliefs (e.g., trust and distributional concerns) and individual characteristics (economic self-interest).

*Italic: Source — wpiea2024216-print-pdf (section 3).*

### 3.1    Perceptions about the effects of private participation and competition

### Measurement approach
- Respondents’ overall perceptions of more competition’s impact on consumers were elicited on a scale between highly detrimental and highly beneficial.
- Similar questions asked about competition’s effect on price, quality, and access.
- Results for the control group are reported in Figure A5 (Panel 3a pooled across sectors; Panels 3b and 3c for electricity and telecommunications respectively).

### Cross-country and sectoral heterogeneity (exact figures preserved)
- Mexico:
  - Electricity sector survey: 39% express positive perceptions of the overall role of private participation and competition.
  - Telecommunications sector survey: 54% express positive perceptions.
- South Africa and Morocco:
  - Electricity sector survey: over 50% hold positive views.
  - Telecommunications sector survey: over 60% hold positive views.
- By service aspect:
  - Cost impact: only around 40% express favorable perceptions.
  - Quality and access: around 60% express favorable perceptions.
- Favorable perceptions are consistently higher in the telecommunications sector than in the electricity sector across dimensions, except for perception of cost impact in South Africa.

### Who views competition more favorably?
- Regression analysis (standardized perceptions variable on socioeconomic characteristics and views/satisfaction with the sector, controlling for country fixed effects and treatment indicators) yields:
  - More positive perceptions among:
    - Males
    - Older individuals (aged 35 or above)
    - Higher-income respondents (those in the upper 50th percentile)
    - Those who perceive limited private participation in network sectors or have experienced limited access to services
  - More negative perceptions among:
    - Respondents with lower levels of education (secondary)
    - Those working in utility companies
    - Individuals with left-wing political orientations

### Additional notes on results and appendices
- Figure 4 reports coefficients from the whole sample (both sectors). Regression tables are in Appendix A.9.
- Alternative results for sector sub-samples are in Appendix A.8.1 (Figures A6 and A7).

*Italic: Source — wpiea2024216-print-pdf (section 3.1).

### 3.2    Predicting policy support: Beliefs and perceptions versus socioeco-

### 3.2    Predicting policy support: Beliefs and perceptions versus socioeconomic characteristics

### Model specification and estimation
- Main regression specification:
  - y_{i,c} = β2 S_{i,c} + β3 B_{i,c} + β1 T_{i,c} + θ_c + u_{i,c}
  - y_{i,c} is support for PMR reforms for individual i in country c.
  - S_{i,c} captures individuals’ socioeconomic characteristics (five categories: demographics, income, education, employment, and political leaning).
  - B_{i,c} captures beliefs and perceptions (five categories: market economy, distribution and equity, trust and corruption, views and satisfaction with sector, and knowledge and effect of policies).
  - T_{i,c} are treatment indicators; θ_c are country fixed effects.
- Estimation method: Ordinary Least Squares (OLS).

### Socioeconomic determinants of support (results excluding beliefs/perceptions)
- Key associations reported (Figure 5):
  - Support for PMR reforms is notably lower among individuals working in, or who know someone working in, public utility companies.
  - Support is lower among women and respondents with children.
  - Respondents who self-identify as politically left-leaning demonstrate lower support for reforms.
  - Support tends to increase with age, income, and education levels.
- Interpretation:
  - Lower support for those linked to public utilities may reflect self-interest concerns about costs of introducing competition in network sectors.
  - Lower support among women and respondents with children is consistent with lower risk tolerance documented in prior literature.
  - Political orientation and higher social status (age, income, education) correlate with greater openness to reform in network sectors.

### Beliefs and perceptions (results conditional on socioeconomic controls)
- Pro-market beliefs:
  - Respondents who believe the government should play a minimal role in market activities or view foreign capital as beneficial are more supportive of PMR reforms.
- Distribution and equity perceptions:
  - Respondents who perceive income distribution in their country as unfair are less supportive of reforms.
  - Belief that government should ensure everyone has access to essential utility services, or that companies providing these services have a key role in reducing inequality and ensuring fair access, are not statistically significant predictors of opposition to private participation.
- Trust and corruption:
  - Trust in others and in public institutions (e.g., courts and parliament) is significantly and positively associated with support for private participation.
  - Low trust in the government and perceptions of corruption are associated with a favorable view toward private participation (likely reflecting skepticism about letting governments perceived as untrustworthy or corrupt handle utility services).
- Satisfaction and personal experience:
  - Respondents dissatisfied with public utility services, or those who experienced a lack of access for themselves or someone close to them, tend to be more supportive of reforms.
- Beliefs about expected effects:
  - Respondents who perceive that competition among companies can lead to lower prices, improved quality, or broader access are more likely to support PMR reforms.

### Relative role of socioeconomic characteristics versus beliefs/perceptions (dominance analysis)
- Overall shares of explained variance in support for PMR reforms (both sectors combined):
  - Individuals’ socioeconomic characteristics account for only 6% of individuals’ support for reforms.
  - Beliefs and perceptions explain about 80% of support for policies.
  - Within beliefs and perceptions:
    - Knowledge and perceptions of policies explain 37% of support.
    - Market-oriented beliefs explain 35% of support.
  - Distributional concerns, together with trust and perceptions on corruption, weigh as much as individual characteristics in explaining support.
- Note: Findings hold at the sector level (electricity and telecommunications) as reported in appendices.

### Randomized information experiment — treatments
- Treatment arms:
  - Control group: no information.
  - Status Quo treatment:
    - Provides sector-specific information about the cost of the status quo (opportunity cost of maintaining nonreform scenario).
    - Electricity: unit cost of electricity (in purchasing power) compared to the US; quality measured by frequency and duration of power cuts (data from International Energy Agency and World Bank’s Enterprise Surveys).
    - Telecommunications: cost of mobile phone subscription and mobile download speeds relative to the US; percentage of individuals without internet access (data from International Telecommunication Union and Global System for Mobile Communications).
    - Treatment does not convey information on private participation.
  - Status Quo + Effect of Policies treatment:
    - Includes Status Quo information plus cross-country evidence and a simple description of findings on how private participation led to improvements in cost, quality, and access, especially when combined with strong regulation ensuring competition and customer protection.
    - Provides illustrative examples and links to cross-country studies (World Bank; Inter-American Development Bank).
    - Objective: test whether explaining how policies work affects support.

### Effects of the treatments on support for policies (Figure 8)
- Status Quo treatment:
  - Positive and statistically significant effect on support for reforms in the electricity sector.
  - Statistically insignificant effects for both sectors combined and for the telecommunications sector.
  - Effect size in electricity: a 0.08 increase in the standard deviation of the policy support variable.
    - This 0.08 increase represents a 30% reduction in the policy support gap between left-wing and right-wing respondents on this issue.
  - Interpretation: raising awareness of costs of not reforming primarily influences support in the electricity sector.
- Status Quo + Effect of Policies treatment:
  - Positive and statistically significant effects for both sectors combined, and for the electricity and telecommunications sectors individually.
  - Effect is particularly strong in the electricity sector; effect size is approximately 50% larger in electricity than in telecommunications.
  - Quantitative comparisons:
    - Effects are equivalent to 1.5 times the policy support gap between left-wing and right-wing respondents for the electricity sector.
    - Effects are equivalent to 1.7 times the policy support gap between left-wing and right-wing respondents for the telecommunications sector.
- Sectoral differences in treatment effects may reflect:
  - Higher perceived initial level of private participation in telecommunications versus electricity.
  - Higher satisfaction with telecommunications services relative to electricity.
  - Underlying differences in beliefs and perceptions regarding benefits of competition in each sector.

### Effects of the treatments on willingness to sign petitions (Figure 9)
- Outcome measure:
  - Petition index: -1 for respondents willing to sign petition limiting private entry; 1 for respondents willing to sign petition facilitating entry; 0 for respondents who declined or signed both.
- Status Quo treatment:
  - No statistically significant impact on willingness to sign any petitions.
  - Estimated effects in electricity have expected direction but are not statistically significant.
  - Interpretation: Status Quo treatment does not generate demand for private actions.
- Status Quo + Effect of Policies treatment:
  - Statistically and economically significant impacts on the petition index and on willingness to facilitate or limit private entry in both sectors.
  - Increases willingness to sign petition asking government to facilitate private entry by:
    - 11.3% in the electricity sector (an increase of 20.9% relative to control group mean of 54.1%).
    - 6.7% in the telecommunications sector (an increase of 10.1% relative to control group mean of 66.6%).
  - Decreases willingness to sign petition to limit private entry by:
    - 9% in the electricity sector (a reduction of 17.3% relative to control group mean of 52.1%).
    - 5.4% in the telecommunications sector (a reduction of 12.8% relative to control group mean of 42.2%).
  - Interpretation: Status Quo + Effect of Policies treatment generates demand for both public policy reform and private action; effects are attributable to specific information content rather than priming about utility services.

*Source: IMF working paper chapter 3.2 (section entitled "Predicting policy support: Beliefs and perceptions versus socioeconomic characteristics").*

### 4.4    Uncovering the mechanism: Treatment Effects on policy beliefs

### 4.4    Uncovering the mechanism: Treatment Effects on policy beliefs

### Overview
- Purpose: Estimate impact of information treatments on individuals’ perceptions of the effects of private participation and competition in the electricity and telecommunications sectors.
- Dependent variables: four perceptions of the effect of private participation and competition — overall perceived benefit for consumers, and perceived effects on cost (prices), quality, and access to services in each sector.
- Estimation method: Ordinary Least Squares regressions controlling for socioeconomic characteristics, beliefs and perceptions, and country fixed effects.

### Effects of the Status Quo treatment
- Overall perception:
  - The Status Quo treatment does not significantly change individuals’ perceptions about the overall effect of private participation and competition for consumers in either sector.
- Individual dimensions:
  - Electricity sector:
    - The treatment increases (marginally) the perception that competition leads to lower prices for consumers; the estimated impact is statistically significant at only the 10% level.
  - Telecommunications sector:
    - The Status Quo treatment reduces respondents’ beliefs that competition leads to improvements in quality and access.
    - Possible explanation: benchmarking domestic quality and access to the US in the treatment may induce a negative shift in perceptions for telecommunications, given pre-treatment differences in satisfaction between sectors.

### Effects of the Status Quo + Effects of Policies treatment
- Overall and dimension-specific perceptions:
  - The Status Quo + Effects of Policies treatment significantly improves respondents’ overall perception of the effects of competition for consumers in both sectors.
  - It also significantly improves perceptions that competition leads to lower prices, better quality, and better access in both the electricity and telecommunications sectors.
- Magnitude of impacts (by sector and measure):
  - Electricity sector: treatment effects on the four perception measures range from 0.295 to 0.410 standard deviations.
  - Telecommunications sector: treatment effects on the four perception measures range from 0.189 to 0.289 standard deviations.
  - Sectoral difference magnitudes: differences between sectors are reported as between 33.7% higher (for perceptions about access) and 63.5% higher (for the overall perception variable) in the electricity sector relative to telecommunications.

### Link between perceptions and policy support
- Perceptions about policy effects are strong predictors of policy support (Section 3.2).
- The insignificance of the Status Quo treatment on overall perceptions is consistent with its small effect on policy support (Section 4.2).
- The Status Quo + Effects of Policies treatment’s positive impact on perceptions explains its stronger effect on policy support.

### Reasons for non-support and mitigating measures
- Prevalence of non-support after treatments:
  - Status Quo treatment: among respondents who received it, 60.5% (electricity) and 50.2% (telecommunications) indicated they would not support the reform.
  - Status Quo + Effects of Policies treatment: these figures drop to 48.5% (electricity) and 37.3% (telecommunications).
- Grouping of reasons for nonsupport (control group distribution):
  - Societal concerns (worries about affordability and access for the poorest, community impacts): account for 54% of responses.
  - Personal concerns (self-interest: price increases, quality, job loss): represent 22% of responses.
  - Other concerns (national security, uncertainty about benefits, lack of trust in evidence, unspecified): approximately 24% of responses.
- Mitigating measures offered (tailored to stated concerns):
  - For cost/quality/national security concerns: creation of an independent regulatory agency to ensure competitive delivery of high-quality services at fair prices.
  - For access/affordability concerns for the poorest: government commitment to ensuring price affordability for the poorest households and adequate coverage nationwide, including remote rural areas.
  - For job-loss concerns: temporary job protection and job-training programs for affected workers.
- Effectiveness of mitigating measures (control group indicative results):
  - Offering tailored complementary and compensatory measures can substantially increase support: 50–80% of control-group respondents who initially opposed PMR reforms indicated they would change to support if mitigating measures addressed their concerns.
  - Across all complementary and compensatory measures, the share who would opt to support the reforms exceeds 50%.

### Ultimate non-supporters: reasons and characteristics
- When asked why they would still oppose reforms despite mitigating measures, respondents mainly cite:
  - Lack of trust in the private sector.
  - Opposition to private sector or foreign investor control of service provision.
  - Lack of trust in government’s willingness or ability to implement reforms or mitigating measures.
- Reported ultimate reasons for non-support (percentages reported in Table 1):
  - Both sectors:
    - Don’t trust the private sector: 43.8
    - Don’t want the private sector or foreign investors to control the provision of services: 35.9
    - Don’t trust the government’s willingness or ability to implement good reforms: 18.2
    - Other Reasons: 2.1
  - Electricity:
    - Don’t trust the private sector: 44.4
    - Don’t want the private sector or foreign investors to control the provision of services: 36.1
    - Don’t trust the government’s willingness or ability to implement good reforms: 17.6
    - Other Reasons: 1.9
  - Telecommunications:
    - Don’t trust the private sector: 42.2
    - Don’t want the private sector or foreign investors to control the provision of services: 35.1
    - Don’t trust the government’s willingness or ability to implement good reforms: 19.9
    - Other Reasons: 2.8
- Characteristics of respondents less likely to remain opposed once mitigating measures are in place (sectoral differences):
  - Electricity sector:
    - Mitigating measures reduce opposition among individuals working in utility companies.
    - Mitigating measures gain support among initially skeptical respondents with pro-market views (those favoring minimal government role, seeing foreign companies in utility services as beneficial, trusting institutions, or viewing competition as beneficial for prices and access).
    - Respondents with left-leaning political views remain more likely to oppose reform despite mitigating measures.
  - Telecommunications sector:
    - Mitigating measures help gain support from individuals in the 25th to 50th income percentiles and those who perceive foreign companies as beneficial to the country’s economy.
    - Respondents who believe utility companies play an important role in ensuring equity are more likely to oppose the reform even after mitigating measures.

### Heterogeneity in treatment effects
- Country-level heterogeneity (Mexico, Morocco, South Africa):
  - Status Quo treatment:
    - No significant variation across countries in general; effects largely statistically non-significant.
    - Exception: Mexico — the estimated effect is positive and statistically significant, suggesting the pooled-sample Status Quo effect may be driven by Mexico.
  - Status Quo + Effects of Policies treatment:
    - Positive and statistically significant impact on support across countries.
    - Magnitude ranking for the electricity sector: Mexico (highest), followed by Morocco, then South Africa.
    - Magnitude ranking for telecommunications: South Africa (highest), followed by Mexico, then Morocco.
    - Quantitatively significant heterogeneity in magnitudes across countries.
  - Potential interpretation for heterogeneity:
    - For Status Quo treatment, Mexico’s proximity to the US could make respondents more responsive to benchmarking versus the US.
    - For South Africa, longstanding prominence of electricity reform (including during the May 2024 general elections) may reduce responsiveness to treatments about policy effects.

- Individual-level heterogeneity: socioeconomic characteristics
  - Electricity sector:
    - Wealthier respondents (above the 75th percentile of country income) show less sensitivity to the Status Quo treatment.
    - Urban residents are more responsive to the Status Quo + Effects of Policies treatment.
    - Individuals in the 25th to 50th income percentiles demonstrate lower sensitivity to the Status Quo + Effects of Policies treatment.
    - Individuals working in, or knowing someone employed by, public utility companies are less sensitive to both treatments.
  - Telecommunications sector:
    - Female respondents and those with incomes between the 25th and 50th percentiles show greater sensitivity to the Status Quo + Effects of Policies treatment (the latter contrasts with the electricity-sector finding).
    - Individuals working in, or who know someone working in, public utility companies are less sensitive to the Status Quo + Effects of Policies treatment (similar to electricity sector).
  - Interpretation notes:
    - Less supportive attitude among upper class may reflect better alternative options under a poor status quo.
    - Urban residents and women may have more to gain from PMR reforms (e.g., reduced home production burden for women and labor-market opportunities).
    - Lower sensitivity of low-income respondents in the electricity sector is surprising; may reflect skepticism that private companies will expand services to remote areas at a loss and aligns with findings of limited impacts from rural electrification in recent literature.

*Source: wpiea2024216-print-pdf - 4.4    Uncovering the mechanism: Treatment Effects on policy beliefs*

### 6.3    Individual-level heterogeneity: beliefs and perceptions

### 6.3    Individual-level heterogeneity: beliefs and perceptions

### Heterogeneous treatment effects by beliefs and perceptions
- Electricity sector:
  - Respondents who believe utility companies “should play a key role in ensuring equity and fair access” are less sensitive to both treatments.
  - Respondents who believe the government should ensure access to utilities are less sensitive to the Status Quo + Effect of Policies treatment.
  - Respondents perceiving improved relative deprivation are more responsive to the Status Quo + Effect of Policies treatment.
- Telecommunications sector:
  - Respondents with experiences of lack of access, perceiving improved relative deprivation, and perceiving lower corruption in utility companies are more sensitive to the Status Quo + Effect of Policies treatment.
- Interpretation:
  - The authors note the theoretical grounding for some findings is less clear but describe the results as “intuitive and aligned with the expected pay-off of the reforms for respondents with different sets of beliefs and perceptions.”

### Double interactions (joint heterogeneity)
- Overall:
  - No significant double-interaction differences observed in the electricity sector.
- Telecommunications sector:
  - Heterogeneity emerges: respondents who both (a) trust their government or key national institutions, (b) believe corruption in utility companies is low, or (c) perceive a high level of private company participation in the sector, and simultaneously (d) experience a lack of access to services, are more sensitive to the Status Quo + Effect of Policies treatment.

### Robustness checks: control variables and experimenter demand effect
- Specification checks (Table 2):
  - Columns 1-3: estimate treatment impacts excluding control variables.
  - Columns 4-6: control for a limited set of pre-specified control variables.
  - Columns 7-9: randomize the order of the support for policies question; treatment impacts remain unchanged.
  - Columns 10-12: limit the sample to men; treatment impacts remain unchanged.
- Experimenter demand effects:
  - Authors reference prior work (Quidt et al., 2018) that such biases are likely modest.
  - Based on columns 7-12 of Table 2, the authors conclude experimenter demand effects “are not significant in our survey.”
- Key coefficients from Table 2 (selected, preserving reported values):
  - Status quo coefficients (example entries): 0.031, 0.103**, -0.015, 0.037, 0.091**, -0.011, 0.077**, 0.132**, 0.044, 0.029, 0.098*, -0.020.
  - Status quo + effect of policies coefficients (example entries): 0.345***, 0.381***, 0.314***, 0.326***, 0.379***, 0.274***, 0.368***, 0.443***, 0.306***, 0.271***, 0.355***, 0.177***.
  - Notes: Robust standard errors in parentheses; *, **, *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

### Robustness checks: data quality
- Approaches (Table 3):
  - Columns 1-3: delete respondents who failed the second attention check.
  - Columns 4-6: augment baseline sample with additional data, excluding those who failed the second attention check.
  - Columns 7-9: drop respondents who felt the survey was biased.
  - Columns 10-12: drop respondents in the top and bottom 5% of the distribution of time spent on the survey.
- Findings:
  - The main results on support remain unchanged across these checks.
- Sample augmentation detail:
  - Alongside the baseline sample of 2,100 respondents per country, additional data collected: 692 in Mexico, 336 in Morocco, and 405 in South Africa.

### Robustness checks: weighted-OLS (re-weighting to population characteristics)
- Re-weighting procedure:
  - Weights match the within-country distribution of age, gender, employment status, education, and geographic location.
  - Regressions control for individuals’ socioeconomic characteristics, beliefs and perceptions, and country fixed effects.
- Main findings (Table 4, Panels A–C):
  - The Status quo + effect of policies treatment has a statistically significant positive effect on support for policies, irrespective of sector.
  - For both sectors combined, the impact of the Status quo + effect of policies treatment in the weighted sample is “only about 7% lower” than the estimate in the unweighted sample.
  - The Status quo + effect of policies treatment is significantly higher for the electricity sector and lower for the telecommunications sector in the weighted sample.
  - The Status quo treatment becomes statistically insignificant for reform support in the weighted sample, irrespective of sector, and is largely insignificant for petitions and first-stage variables.
- Selected coefficient examples from Table 4 (preserving reported values):
  - Panel A (Both sectors), Status quo + effect of policies: 0.311***, 0.244***, -0.289***, -0.235***, -0.289***, -0.289***, -0.289***, -0.289***.
  - Panel B (Electricity), Status quo + effect of policies: 0.128***, 0.373***, -0.289***, -0.289***, 0.128***, 0.373***, -0.289***, 0.373***.
  - Panel C (Telecommunications), Status quo + effect of policies: 0.176***, 0.158***, 0.0370, 0.217***, 0.176***, 0.158***, 0.125***, 0.125***.
  - Panel C (Telecommunications), Status quo examples: -0.049, -0.113**, -0.041, 0.034, -0.035, -0.035, -0.161**, -0.211***.

### Implications for interpretation
- The heterogeneity analysis indicates beliefs and perceptions moderate how respondents react to information about the status quo and to evidence about the effects of PMR reforms.
- Re-weighting and multiple data-quality and specification checks do not overturn the core finding that the Status quo + effect of policies treatment increases support, while the Status quo treatment is less robust.

### Policy-relevant takeaways (drawn from this section and linked robustness evidence)
- Messaging that explains how policies work and communicates the effects of PMR reforms is more effective at increasing public support than messaging that highlights only the costs of the status quo.
- Policy communication should account for heterogeneous beliefs and perceptions across groups (e.g., views on equity, trust in government, perceptions of corruption, experience of access), as these factors influence responsiveness to information.
- Robustness and re-weighting checks suggest the positive effect of explaining policy effects (Status quo + effect of policies) on support for reforms is not an artifact of sample composition or attention biases.

*Source: IMF staff analysis — section 6.3, “Individual-level heterogeneity: beliefs and perceptions,” including Figures and Tables referenced in the original chapter.*

### References

### wpiea2024216-print-pdf - References

### References list (bibliography)
- Contains full bibliography of works cited in the chapter, including (sample entries as presented):
  - Alfaro, Laura, Maggie Chen, and Davin Chor (2023). Can evidence-based information shift preferences towards trade policy? Tech. rep. National Bureau of Economic Research.
  - Alsan, Marcella, Luca Braghieri, Saray Eighmeyer, Minjeong Joyce Kim, Stefanie Stantcheva, and David Y Yang (Oct. 2023). “Civil Liberties in Times of Crisis”. In: American Economic Journal: Applied Economics 15.4, pp. 389–421.
  - Andrés, Luis A, J Luis Guasch, Thomas Haven, and Vivien Foster (2008). The impact of private sector participation in infrastructure: lights, shadows, and the road ahead. World Bank Publications.
  - Dechezleprêtre, Antoine, Adrien Fabre, Tobias Kruse, Bluebery Planterose, Ana Sanchez Chico, and Stefanie Stantcheva (2022). “Fighting climate change: International attitudes toward climate policies”.
  - Duflo, Esther (Dec. 2012). “Women Empowerment and Economic Development”. In: Journal of Economic Literature 50.4, pp. 1051–1079.
  - Megginson, William L and Jeffry M Netter (2001). “From state to market: A survey of empirical studies on privatization”. In: Journal of economic literature 39.2, pp. 321–389.
  - World Bank (2006). Peru: Rethinking Private Sector Participation in Infrastructure: Towards Effective Public Private Partnerhips/concessions in the Provision of Infrastructure Services. Vol. Latin America and the Caribbean Regional Office. Finance, Private Sector and Infrastructure Sector Management Unit. World Bank.
- The references section spans pages that include the bibliography and transitions into Appendix A (page numbers as shown: 44–46).

### Appendix A — Contents and structure
- Appendix A includes the following labeled subsections and materials:
  - A.1 Sample Statistics
    - Figure A1: Difference between Sample Averages and Population Means (panels for Mexico, Morocco, South Africa). Note: 95% confidence intervals illustrated.
  - A.2 Duration
    - Figure A2: Distribution of Time Spent on the Survey. Note: ZA = South Africa, MX = Mexico, MA = Morocco.
  - A.3 Treatment scripts
    - English versions of treatment scripts for each country listed (Mexico, Morocco, South Africa).
    - Benchmark two treatment scripts for South Africa (electricity sector): Figures A3 and A4 labeled “Status quo treatment” and “Status quo + effect of policies treatment”.
  - A.4 Data sources used to design the treatments
    - Electricity — Status quo and Status quo + effect of policies: data sources include International Energy Agency (IEA) Energy Prices dataset; World Bank Enterprise Surveys; Academic Studies including World Bank: Harris (2003) and IDB: Balza et al. (2020a).
    - Telecommunication — Status quo and Status quo + effect of policies: data sources include International Telecommunication Union (ITU) - World Telecommunication/ICT Indicators Database; Global System for Mobile Communications (GSMA) - Mobile Connectivity dataset; Academic Studies including World Bank: Harris (2003) and IDB: Balza et al. (2020a).
  - A.5 Variables definitions
    - Detailed definitions for treatment indicators, outcome variables, socio-economic characteristics, beliefs and perceptions variables, and how variables are transformed.
    - Key methodological note on outcome transformation:
      - Support for policies and beliefs about effect of policies are transformed into z-scores using a two-step process following Dechezleprêtre et al. (2022): first subtract control group mean and divide by control group standard deviation within each country-sector; second, standardize that index across the entire sample by subtracting the mean and dividing by the standard deviation within each sector. Resulting z-scored outcomes have mean 0 and standard deviation 1.
    - Exact questionnaire codings preserved (scales and anchor labels), for example:
      - Support for policies: coded 1 to 5 where 1 = "Strongly oppose", 3 = "Neither oppose nor support", 5 = "Strongly support".
      - Pre-treatment support for policies: coded 1 to 5 where 1 = "Fully handled by public companies or the government", 3 = "Equally handled by public and private companies", 5 = "Fully handled by private companies".
      - Perception, trust, fairness, satisfaction, and knowledge variables include precise scale codings (1–4, 1–5, 1–10) and anchor text as listed.
    - Petition index construction: takes value -1, 0, or 1 with exact definitions preserved.
  - A.6 Additional descriptive statistics
    - Figure A5: Share of respondents supporting the reform or willing to sign the petition (panels: Both Sectors, Electricity, Telecommunications). Note describes control-group proportions and countries MA, MX, ZA.
    - Table A1: Differences between electricity and telecommunications sectors for sector-specific variables (columns and statistics presented, with reported differences, t-test statistics and p-values).
  - A.7 Balance Analysis
    - Narrative summary: balance tests compare treatment and control groups across socioeconomic characteristics and pre-treatment perceptions; no differences statistically significant across all three scenarios; for status quo treatment the joint likelihood ratio test p-value = 0.151; for Tables A2 and A4 p-values reported as 0.603 and 0.866 respectively.
    - Table A2: Balance Test of Covariates between Treatment and Control Groups (columns: Treated, Control, t, p-value). Example entries: Age (35-54) Treated 0.41, Control 0.41, t 0.55, p-value 0.579; Observations and full variable list included in table.
    - Table A3: Balance Test between Status Quo Treatment and Control Groups (example: Perception of private participation in utilities Treated -0.23, Control -0.16, t -2.34, p-value 0.019).
    - Table A4: Balance Test between Status Quo + Effect of Policies Treatment and Control Groups (example: Above 75th percentile Treated 0.20, Control 0.17, t 1.76, p-value 0.079).
  - A.8 Additional figures (subsections)
    - A.8.1 Who holds positive perceptions about competition?
      - Figure A6: Perceptions of overall effect of competition — Electricity Sector.
      - Figure A7: Perceptions of overall effect of competition — Telecommunications Sector.
    - A.8.2 Role of socioeconomic characteristics and beliefs and perceptions
      - Figure A8: Correlation between socioeconomic characteristics and support for policies — Electricity sector.
      - Figure A9: Correlation between socioeconomic characteristics and support for policies — Telecommunications sector.
      - Figure A10: Correlation between beliefs and perceptions and support for policies — Electricity sector.
      - Figure A11: Correlation between beliefs and perceptions and support for policies — Telecommunications sector.
      - Figure A12: Drivers of PMR reforms support — Electricity sector (dominance analysis).
      - Figure A13: Drivers of PMR reforms support — Telecommunications sector (dominance analysis).
    - A.8.3 Characteristics of the ultimate non-supporters for reforms
      - Figure A14: Socioeconomic characteristics of the ultimate non-supporters for reforms.
      - Figure A15: Beliefs and perceptions of the ultimate non-supporters for reforms.
  - A.9 Additional tables
    - Table A5: Perceptions of the overall effect of competition — regression results (columns (1) Both sectors, (2) Electricity, (3) Telecom). Reported Observations: 6,295; 3,187; 3,108. R-squared values: 0.293, 0.093, 0.115. Includes coefficient estimates and standard errors; significance markers *** p<0.01, ** p<0.05, * p<0.1.
    - Table A6: Correlation between socioeconomic characteristics and support for policies — regression results (columns (1) Both, (2) Electricity, (3) Telecom). Reported Observations: 6,300; 3,190; 3,110. R-squared values: 0.133, 0.165, 0.118. Coefficients and standard errors reported with controls for treatments and country fixed effects.

### Key methodological and data notes (preserved exactly)
- Z-score transformation for outcomes follows a two-step process and references Dechezleprêtre et al. (2022).
- Petition index coding: -1 if support petition limiting private entry, 1 if support petition facilitating private entry, 0 otherwise.
- Country codes used in figures: ZA = South Africa, MX = Mexico, MA = Morocco.
- Balance test joint p-values reported explicitly: status quo joint likelihood ratio test p-value = 0.151; Tables A2 and A4 joint p-values reported as 0.603 and 0.866 respectively.
- Exact sample sizes and table statistics reported in Tables A5 and A6: Observations 6,295 / 3,187 / 3,108 (Table A5); Observations 6,300 / 3,190 / 3,110 (Table A6). R-squared and R-squared adjusted values included per table.

*Content summarized from the file: wpiea2024216-print-pdf - References.*

### Appendix A.5 for the precise definitions of the variables.  Standard errors are in

### Appendix A.5 for the precise definitions of the variables. Standard errors are in parentheses.

### Tables A7–A11: Regression results and treatment effects (selected estimates and statistics)
- Table A7: Correlation between beliefs and perceptions and support for policies (OLS; controls for socioeconomic characteristics, treatment indicators and country fixed effects)
  - Sample sizes and fit:
    - Observations: 6,281 (Both), 3,186 (Electricity), 3,095 (Telecom)
    - R-squared: 0.398 (Both), 0.432 (Electricity), 0.364 (Telecom)
  - Selected coefficients (coefficient (standard error)) for columns (1) Both, (2) Electricity, (3) Telecom:
    - Government role in economic activity: 0.098*** (0.013); 0.083*** (0.019); 0.110*** (0.017)
    - Government role in regulating the economy: 0.022* (0.011); 0.034** (0.016); 0.013 (0.015)
    - Government role in price setting: 0.046** (0.011); 0.040* (0.016); 0.042** (0.015)
    - Perception of foreign companies: 0.068*** (0.013); 0.064** (0.019); 0.087*** (0.019)
    - Perception of foreign companies in utilities: 0.241*** (0.013); 0.235*** (0.018); 0.185*** (0.020)
    - Fairness: 0.032** (0.013); 0.053*** (0.019); 0.015 (0.017)
    - Trust in people: 0.028** (0.014); 0.024 (0.020); 0.035** (0.016)
    - Trust in government: -0.050*** (0.016); -0.051** (0.023); -0.055** (0.020)
    - Trust in institutions: 0.047*** (0.015); 0.044* (0.023); 0.046** (0.019)
    - Corruption in the country: -0.058*** (0.016); -0.070*** (0.023); -0.048** (0.020)
    - Lack of access to utility services: 0.034*** (0.012); 0.010 (0.017); 0.032** (0.016)
    - Knowledge about the utility service’s regulator: 0.066*** (0.012); 0.059** (0.019); 0.040** (0.015)
    - Perception of the effect of competition on price: 0.177*** (0.015); 0.256*** (0.023); 0.082** (0.020)
    - Perception of the effect of competition on quality: 0.098*** (0.018); 0.080*** (0.026); 0.104*** (0.024)
    - Perception of the effect of competition on access: 0.126*** (0.018); 0.118*** (0.026); 0.146*** (0.026)

- Table A8: Effect of the treatments on policy support (OLS; controls include socioeconomic characteristics, beliefs and perceptions, country fixed effects)
  - Coefficients (standard error) for treatments, columns (1) Both, (2) Electricity, (3) Telecom:
    - Status quo: 0.031 (0.027); 0.080** (0.040); -0.003 (0.037)
    - Status quo + effect of policies: 0.335*** (0.027); 0.407*** (0.041); 0.271*** (0.035)
  - Observations: 6,287 (Both), 3,186 (Electricity), 3,101 (Telecom)
  - R-squared: 0.353 (Both), 0.368 (Electricity), 0.319 (Telecom)
  - Control group’s outcome mean: -0.309 (Both), -0.521 (Electricity), -0.101 (Telecom)

- Table A9: Effect of the treatments on willingness to sign a petition (multiple petition outcomes; OLS)
  - Both sectors (columns 1–3), Electricity (4–6), Telecom (7–9); Controls and country fixed effects included
  - Status quo treatment coefficients are generally small and not significant:
    - Example: Status quo (Both, Petition index): -0.004 (0.030)
  - Status quo + effect of policies shows positive and significant effects on facilitating petitions and negative (significant) effects on limiting petitions:
    - Status quo + effect of policies (Both): 0.235*** (0.029) for Petition index; 0.089*** (0.014) for Petition to facilitate; -0.071*** (0.014) for Petition to limit
    - Electricity: 0.302*** (0.042); 0.113*** (0.019); -0.090*** (0.021)
    - Telecom: 0.176*** (0.040); 0.067*** (0.019); -0.054** (0.019)
  - Observations vary by outcome; example: Petition index (Both) Observations: 6,287; R-squared: 0.199
  - Control group’s outcome means reported, e.g., Petition index (Both): -0.224; Petition to facilitate (Both): 0.604; Petition to limit (Both): 0.471

- Table A10: Effect of the treatments on beliefs about the effects of policies (Both, Electricity, Telecom; outcomes: Overall, Cost, Quality, Access)
  - Status quo treatment generally not significant; selected exceptions:
    - Status quo (Both, Access): -0.069** (0.029)
    - Status quo (Telecom, Cost): -0.104** (0.041); Status quo (Telecom, Access): -0.139*** (0.041)
  - Status quo + effect of policies shows consistent positive and significant effects across outcomes:
    - Example (Both): Overall 0.242*** (0.028); Cost 0.344*** (0.029); Quality 0.238*** (0.028); Access 0.234*** (0.028)
    - Electricity (Overall): 0.309*** (0.041); Cost 0.410*** (0.041); Quality 0.295*** (0.041); Access 0.270*** (0.041)
    - Telecom (Overall): 0.189*** (0.038); Cost 0.289*** (0.040); Quality 0.190*** (0.039); Access 0.202*** (0.039)
  - Observations and R-squared vary by column; example: Observations 6,286 (Both, Overall); R-squared 0.247

- Table A11: Effect of the treatments on support for policies — Country-level heterogeneity (OLS by country)
  - Electricity sector (columns 1–3): Mexico, Morocco, South Africa
    - Status quo (Mexico): 0.131* (0.068)
    - Status quo + effect of policies (Mexico): 0.468*** (0.067)
    - Status quo + effect of policies (Morocco): 0.430*** (0.073)
    - Status quo + effect of policies (South Africa): 0.327*** (0.074)
    - Observations: 1,089 (Mexico), 1,068 (Morocco), 1,029 (South Africa)
    - R-squared: 0.384 (Mexico), 0.265 (Morocco), 0.305 (South Africa)
  - Telecommunications sector (columns 4–6): Mexico, Morocco, South Africa
    - Status quo + effect of policies (Mexico): 0.249*** (0.057)
    - Status quo + effect of policies (Morocco): 0.188*** (0.061)
    - Status quo + effect of policies (South Africa): 0.352*** (0.063)
    - Observations: 1,006 (Mexico), 1,028 (Morocco), 1,067 (South Africa)
    - R-squared: 0.395 (Mexico), 0.299 (Morocco), 0.271 (South Africa)
  - Control group’s outcome means reported by country and sector (examples):
    - Electricity (Mexico): -0.818; Morocco: -0.766; South Africa: 0.0375
    - Telecommunications (Mexico): -0.195; Morocco: -0.254; South Africa: 0.144

### Appendix A.10: Questionnaire (benchmark English version for South Africa) — structure and key items
- Survey administration and consent
  - Survey languages offered: Zulu / Zoeloe / IsiZulu / Zulu; Afrikaans / IsiBhunu; Xhosa / IsiXhosa; English
  - Drop-down for year of birth: years from 1900-2024
  - Consent response options:
    - Yes: "Yes, I would like to take part in this study, and I confirm that I LIVE IN SOUTH AFRICA, and I am 18 or older."
    - No: "No, I would not like to participate."
  - Incentive: "In exchange for your participation, you will receive 1500 credited to your YouGov account."
  - Contact: surveypeople2024@gmail.com
  - Eligibility: "You must live in South Africa to participate in this survey."
- Socioeconomic and demographic modules (examples of items and response categories preserved exactly)
  - Gender (question 3): 1. Man; 2. Woman; 3. Non-binary; 4. Other
  - Race (question 4): 1. Asian / Indian; 2. Black; 3. Coloured; 4. White
  - Province (question 5): Eastern Cape; Free State; Gauteng; KwaZulu-Natal; Limpopo; Mpumalanga; North West; Northern Cape; Western Cape
  - Urban/rural (question 7): 1. Urban (city or town, population 2,001 or more); 2. Rural (village, population 2,000 or less)
  - Employment status (question 8): 1. Self-employed or independent; 2. Salaried employee in a state company; 3. Salaried employee in a private company; 4. Salaried employee in a government department; 5. Temporary out of work; 6. Retired or pensioner; 7. Don't work, responsible for housework; 8. Student
  - Education (question 12): enumerated options including "Primary school" through "Master’s Degree or Doctoral Degree"; codes: 96. None of these; 98. Don't know; 95. Prefer not to say
  - Age finished full-time education (question 13): categories 15 or under; 16; 17-18; 19; 20+; 6. Still at school/Full time student; 7. Can't remember; 97. Not applicable
  - Years of formal education (question 14): drop down 0-22 years; Not sure also an option
- Trust, corruption, distribution, and well-being (selected items)
  - Trust in people (question 15): 1. People cannot be trusted; 2. Most people cannot be trusted; 3. Most people can be trusted; 4. People can be trusted
  - Trust in government (question 16): same 4-point scale wording applied to "Governments in South Africa"
  - Views on corruption (question 18): 10-point slider where "1" = "there is no corruption in this country" and "10" = "there is abundant corruption in this country"
  - Fairness of income distribution (question 19): 1. The distribution of income is very unfair; 2. The distribution of income is unfair; 3. The distribution of income is fair; 4. The distribution of income is very fair
  - Standard of living compared to parents/grandparents (question 20): 1. My standard of living is worse; 2. My standard of living is slightly worse; 3. My standard of living is about the same; 4. My standard of living is slightly better; 5. My standard of living is better
- Market economy and role of government / firms (selected items)
  - Who should handle productive activities (question 21): 1. Government-owned companies; 2. Government-owned companies, but with some private involvement; 3. Equally private companies and government-owned companies; 4. Private companies, but with some government involvement; 5. Private companies
  - Government role in regulating the economy (question 22): 1. Extensive role; 2. Moderate role; 3. Some role; 4. Minimal role
  - Government price intervention (question 23): 1. The government should always intervene; 2. The government should often intervene; 3. The government should occasionally intervene; 4. The government should rarely intervene
  - Perception of foreign companies’ effect on the economy (question 24): 1. They greatly harm our economy; 2. They somewhat harm our economy; 3. They neither improve nor harm our economy; 4. They somewhat improve our economy; 5. They greatly improve our economy
  - Government ensuring access to essential services (question 25): 1. Strongly agree; 2. Agree; 3. Neither agree nor disagree; 4. Disagree; 5. Strongly disagree
- Knowledge, satisfaction, and perceptions about the sector and regulators (selected items)
  - Awareness of regulator (question 26): response options include "No, I have never heard that name"; "Construction company"; "Sport company"; "Electricity regulator"; "Industrial activities"; "Water services"; "Telecommunications regulator"
  - Satisfaction with services (question 27): 1-10 slider
  - Cost, quality, access evaluations (question 28): cost/quality/access questions with ordered categorical options (e.g., cost: Very expensive to Very cheap; quality: Very bad to Very good; access: Access is very limited to Access for almost everyone/everywhere)
  - Role of companies in reducing inequality (question 29): 1. A major role; 2. A moderate role; 3. A limited role; 4. No role at all
  - Whether companies work for a few powerful people or everyone's benefit (question 30): enumerated responses including "They work for the benefit of a few powerful people" through "They work for everyone's benefit"
  - Perceived impact of foreign companies in utilities (question 31): 1. Very negative; 2. Somewhat negative; 3. Neither positive nor negative; 4. Somewhat positive; 5. Very positive
- Attention check (question 32): instruction to select both “Extremely interested” and “Not interested at all” to indicate attention
- Pre-treatment perceptions and support (selected items)
  - Provision of services public vs private (question 33): slider 1–10 where 1 = fully handled by public companies and 10 = fully handled by private companies
  - Attitudes toward private companies providing services (question 34): prompt begins "How do you feel about private companies being allowed to [generate and sell electricity/sell telecommunication services] in South Africa? The provision of [electricity/telecommunication services] in South Africa should be:" (question text preserved; response scale not shown in excerpt)

*Source: wpiea2024216-print-pdf (Appendix A.5, Tables A7–A11; Appendix A.10 questionnaire excerpts).*

### 1. Fully handled by public companies or the government

### 1. Fully handled by public companies or the government

### Treatment and experimental design
- Note: No content is shown to the control group. The respondents are randomly allocated to three groups and received different treatments.
- Separate treatments depending on whether the respondent was chosen to answer questions on Electricity or Telecommunications.

### Electricity – Status quo (presentation text shown to respondents)
- The cost of electricity is higher in South Africa than in rich countries. South Africans pay 68% more for electricity than people in the United States*.
  - *Source: Prices in national currency from the International Energy Agency (IEA) are adjusted by the purchasing power of the South African Rand vis-à-vis the US dollar.
- Electricity is also less reliable in South Africa than in richer countries. About 92% of the companies in South Africa report that they face electricity power cuts for about 2 hours and 20 minutes each time.*
  - *Source: World Bank.

### Electricity – Status quo + effect of policies (presentation text shown to treatment group)
- Status-quo points repeated:
  - South Africans pay 68% more for electricity than people in the United States*.
  - About 92% of the companies in South Africa report that they face electricity power cuts for about 2 hours and 20 minutes each time.*
- Academic findings presented to respondents:
  - When private companies help deliver electricity services, more people get access, quality improves, and consumers often pay less.
    - *Sources: World Bank; Inter-American Development Bank (IDB).
  - Example quantitative findings summarized:
    - Production costs and consumer tariffs fell by 10 to 30% in some Latin American countries when private firms entered the market.
    - Frequency and duration of service interruptions decreased significantly; one study showed a 28% reduction in power outages.
  - Importance of regulation:
    - Benefits are greater if there is a strong and independent regulator to ensure companies truly compete, deliver on promises, and protect customers.
    - Guatemala example: combining private investment with effective regulation enhanced rural electricity access, with 280,000 new residential connections by private electricity distributors.
    - *Sources: World Bank; Inter-American Development Bank (IDB).

### Telecommunication – Status quo (presentation text shown to respondents)
- Cost and reliability in South Africa:
  - South Africans spend five times more of their monthly income on mobile data plans than people do in the United States.* 
    - *Source: International Telecommunication Union (ITU).
  - Downloading on mobile phones is 74% slower compared to the United States, and about 25% of people still can't use the internet.*
    - *Sources: International Telecommunication Union (ITU); Global System for Mobile Communications. (GSMA).

### Telecommunication – Status quo + effect of policies (presentation text shown to treatment group)
- Status-quo points repeated:
  - South Africans spend five times more of their monthly income on mobile data plans than people do in the United States.* 
  - Downloading on mobile phones is 74% slower compared to the United States, and about 25% of people still can't use the internet.*
- Academic findings presented to respondents:
  - When private companies help deliver phone services and internet, more people get access, quality improves, and consumers often pay less.
    - *Sources: World Bank; Inter-American Development Bank (IDB).
  - Example quantitative finding:
    - In Uganda, after private mobile companies entered in the late 1990s, within just four years of a second company joining the market, mobile subscriptions increased more than twentyfold.
  - Importance of regulation:
    - Benefits are greater if there is a strong and independent regulator to ensure competition, delivery, and customer protection.
    - Peru example: combining private investment with effective regulation in the 1990s enhanced provision of payphones in rural areas.
    - *Sources: World Bank; Inter-American Development Bank (IDB).

### Post-treatment materials and participant choice
- Question 35: Option to receive links to mentioned sources and studies at the end of the questionnaire.
  - 1. Yes, I would like to receive the links to the mentioned studies.
  - 2. No, I don't want to receive the links to the mentioned studies.

### Post-treatment beliefs about effects of policies
- Question 36 (perception of competition): Do you view it as beneficial or detrimental for consumers when several companies compete to provide services such as [electricity/telecommunication services]?
  - 1. Highly detrimental
  - 2. Somewhat detrimental
  - 3. Neither beneficial nor detrimental
  - 4. Somewhat beneficial
  - 5. Highly beneficial
- Question 37 (perceived changes across dimensions): How do you think private companies competing to provide [electricity/telecommunication services] changes things? (Cost, Quality, Access)
  - Response scale for each dimension:
    - Cost: Significantly pricier / Somewhat pricier / No change / Somewhat cheaper / Significantly cheaper
    - Quality: Significantly worse / Somewhat worse / No change / Somewhat better / Significantly better
    - Access: Significantly worse / Somewhat worse / No change / Somewhat better / Significantly better

### Post-treatment support for policies
- Question 38: Support or oppose allowing private companies to produce and sell [electricity/telecommunication services] in South Africa?
  - 1. Strongly oppose
  - 2. Somewhat oppose
  - 3. Neither oppose nor support
  - 4. Somewhat support
  - 5. Strongly support

### Post-treatment willingness to act (petitions)
- Questions 39-40 (random order): Would you sign one of the following petitions?
  - 39. Petition asking the government to facilitate the entry of private firms in the [electricity/telecommunication services] sector.
    - 1. I want to sign this petition.
    - 2. I do not want to sign this petition.
  - 40. Petition asking the government to limit the entry of private firms in the [electricity/telecommunication services] sector.
    - 1. I want to sign this petition.
    - 2. I do not want to sign this petition.

### Reasons behind support (conditional on support responses)
- Questions 41–44 (appear if respondent selects support in Question 38; question 42 for control group):
  - Reasons for support (select ALL that apply):
    - 1. Would benefit me personally.
    - 2. Would benefit the economy.
    - 3. Would reduce inequality by allowing more people to have access.
    - 4. I was not aware of how much the [electricity/telecommunication services] could improve in South Africa. (treatment groups)
    - 5. Other reasons
  - Question 43: If "Other reasons" selected — short answer field.
  - Question 44: Would you still support this if it meant the government would sell its own [electricity/telecommunication] business to private companies?
    - 1. Yes, I would still support
    - 2. No, I would not support

### Reasons behind opposition (conditional on oppose responses)
- Questions 45–56 (appear if respondent selects do not support in Question 38):
  - Question 45: Open-ended reason for opposition.
  - Question 46: Which describe reasons for not supporting (select all that apply)? (Answer choices appear in random order)
    - 1. It would hurt me personally as I could lose my job
    - 2. It would hurt me personally as I would need to pay more for utility services
    - 3. It would hurt me personally as the quality of services will deteriorate and/or cannot be guaranteed
    - 4. It may benefit me personally, but I am not certain and I would rather not take the chance
    - 5. It would affect my community, as many jobs would be destroyed
    - 6. It would affect my community, as poorer households will need to pay more or will lose access
    - 7. It would affect my community, as private firms would cut service immediately to poor households if payment is delayed
    - 8. It would affect my community, as private firms would not provide coverage in remote areas
    - 11. It would pose national security risks
    - 9. I don't trust the evidence provided
    - 10. Other reasons
  - Question 47: Rank selected reasons in order of importance (conditional).
  - Question 48 (conditional if choices 2, 3, or 7 selected in Q46): Government commits to create an independent regulatory agency with delegated powers to ensure competition, quality, fair price, and national security — would you then support private entry?
    - 1. Yes
    - 2. No
  - Question 49 (if answered No to Q48): Which reasons better describe continued opposition? (select from random-order list)
    - 1. I do not trust the private sector can improve the provision of utility services
    - 2. I do not trust the government can implement good reforms to allow private firms to produce and sell [electricity/telecommunication services]
    - 3. I do not trust the private sector
    - 4. I do not want the private sector to control the provision of utility services
    - 5. I do not want foreign investors to control the provision of utility services
    - 6. Other reasons
  - Question 50: If "Other reasons" selected — short answer field.
  - Question 51 (conditional if choices 6, 7, or 8 selected in Q46): Government commits to measures ensuring affordability for poorest households and nationwide provision, even in remote rural areas — would you then support private entry?
    - 1. Yes
    - 2. No
  - Question 52 (if answered No to Q51): Which reasons better describe continued opposition? (select all that apply; random order)
    - 1. I do not trust the private sector can improve the provision of utility services
    - 2. I do not trust the government can implement good reforms to allow private firms to produce and sell [electricity/telecommunication services]
    - 3. I do not trust the private sector
    - 4. I do not want the private sector to control the provision of utility services
    - 5. I do not want foreign investors to control the provision of utility services
    - 6. Other reasons
  - Question 53: If "Other reasons" selected — short answer field.
  - Question 54 (conditional if choice 1 or 5 selected in Q46): Government commits to adopt measures to protect jobs in affected firms (e.g., prohibits firing for some time) and provide training opportunities — would you then support private entry?
    - 1. Yes
    - 2. No
  - Question 55 (if answered No to Q54): Which reasons better describe continued opposition? (select all that apply; random order)
    - 1. I do not trust the private sector can improve the provision of utility services
    - 2. I do not trust the government can implement good reforms to allow private firms to produce and sell [electricity/telecommunication services]
    - 3. I do not trust the private sector
    - 4. I do not want the private sector to control the provision of utility services
    - 5. I do not want foreign investors to control the provision of utility services
    - 6. Other reasons
  - Question 56: If "Other reasons" selected — short answer field.

### Second block of socioeconomic and attention-check questions
- Question 57: Attention check — "To show that you've read this much, when asked for your favorite color, please enter the word "puce" in the text box below." What is your favorite color?
- Question 58: Have you, or anyone you know, ever experienced a lack of access to [electricity services or experienced loadshedding (planned cuts)/ telecommunication services or systematic interruptions of service], either now or in the past?
  - 1. Yes
  - 2. No
  - 3. Not sure
- Question 59: Does anyone in your close circle, including yourself, work for a public utility company?
  - 1. Yes
  - 2. No
  - 3. Not sure
- Question 60 (asked if 1, 2, 3 or 4 are selected in Question 8): Which sector of the economy better describes your main job (single answer)?
  - 1. Agriculture, Forestry and Fishing
  - 2. Mining and Quarrying
  - 3. Manufacturing
  - 4. Electricity, Gas, Water Supply
  - 5. Construction
  - 6. Wholesale Trade
  - 7. Retail Trade (including, among others, stores and retailers)

*Source: wpiea2024216-print-pdf - 1. Fully handled by public companies or the government*

### 8. Transportation and Storage (including, among others, air, rail and road transport, and

### 8. Transportation and Storage (including, among others, air, rail and road transport, and postal and courier activities)

### Survey sections and skip logic
- Questions cover sector classification, employment status, household characteristics, demographics, political behavior, and perceptions of the survey.
- Conditional prompts:
  - "Ask if 5, 6, or 7 are selected in Question 8."
  - "Ask if 5, 6, or 7 are selected in Question 8 AND Question 61’s selection is different from 5."

### Industry / sector classification (used in multiple questions)
- 1. Agriculture, Forestry and Fishing
- 2. Mining and Quarrying
- 3. Manufacturing
- 4. Electricity, Gas, Water Supply
- 5. Construction
- 6. Wholesale Trade
- 7. Retail Trade (including, among others, stores and retailers)
- 8. Transportation and Storage (including, among others, air, rail and road transport, and postal and courier activities)
- 9. Accommodation and Food Activities (including, among others, hotels and restaurants)
- 10. Information and Communication (including, among others, IT, telecommunications, publishing and broadcasting activities)
- 11. Finance and Insurance
- 12. Real Estate
- 13. Professional, Scientific, Technical, Administrative and Support Service Activities (including, among others, lawyers, accountants, architects, notaries...etc)
- 14. Community, Social and Personal Services (including, among others, public administration, education health, social services)
- 15. Arts and Entertainment

### Employment status (Question 61)
- 1. Self-employed or independent
- 2. Salaried employee in a state company
- 3. Salaried employee in a private company
- 4. Salaried employee in a government department
- 5. Didn't work, responsible for housework

### Latest sector of main job (Question 62)
- Repeats the Industry / sector classification list (items 1–15 as above).

### Household head and income (Questions 63–64)
- 63. Are you the head of the household?
  - 1. Yes
  - 2. No
- 64. Gross HOUSEHOLD income (combined income of all earners in household from all sources, before tax):
  - 1. Under R2,000 per month
  - 2. R2,000 - R3,999 per month
  - 3. R4,000 - R5,999 per month
  - 4. R6,000 - R7,999 per month
  - 5. R8,000 - R9,999 per month
  - 6. R10,000 - R11,999 per month
  - 7. R12,000 - R13,999 per month
  - 8. R14,000 - R15,999 per month
  - 9. R16,000 - R17,999 per month
  - 10. R18,000 - R19,999 per month
  - 11. R20,000 - R39,999 per month
  - 12. R40,000 - R59,999 per month
  - 13. R60,000 - R79,999 per month
  - 14. R80,000 - R99,999 per month
  - 15. R100,000 - R149,999 per month
  - 16. R150,000 or more per month
  - 95. Prefer not to answer
  - 98. Don't know

(Note: the original listing contains repeated and out-of-order numeric labels including "11 if 0." and "17. R20,000 – R24,999 per month", "18. R25,000 – R29,999 per month", "19. R30,000 – R34,999 per month", "20. R35,000 – R39,999 per month". The preserved canonical list above follows the primary enumerated items provided in the source content.)

### Housing tenure (Question 65)
- 1. Own – outright without a loan/mortgage
- 2. Own – with a mortgage/loan (i.e I have borrowed money from a bank or similar to buy a house)
- 3. Rent
- 4. Neither – I live with my parents, family or friends but pay some rent to them
- 5. Neither – I live rent-free with my parents, family or friends
- 97. Other
- 95. Prefer not to say

### Marital status and parental status (Questions 66–67)
- 66. What is your marital status?
  - 1. Married
  - 2. Separated
  - 3. Divorced
  - 4. Widowed
  - 5. Never married
  - 6. Domestic / civil partnership
- 67. Are you a parent or guardian? Please select all that apply
  - 1. Yes, of at least one child younger than 18
  - 2. Yes, of at least one child 18 years old or older
  - 3. No, I am neither a parent or guardian
  - 98. Don't know/Prefer not to say

### Religion and religiosity (Questions 68–69)
- 68. Do you regard yourself as belonging to any particular religion, and if so, to which of these do you belong?
  - 2. Yes - Anglican
  - 3. Yes - Baptist
  - 4. Yes - Catholic
  - 5. Yes - Uniting Church
  - 6. Yes - Presbyterian
  - 7. Yes - Greek Orthodox
  - 8. Yes - Buddhism
  - 9. Yes - Islam
  - 10. Yes - Jewish
  - 11. Yes - Hinduism
  - 12. Yes - Other Christian
  - 13. Yes - Methodist Church
  - 14. Yes - Zion Christian Church
  - 15. Yes - Pentecostal
  - 16. Yes - Other African traditional religion
  - 97. Yes - Other
  - 1. No, I do not regard myself as belonging to any particular religion.
  - 95. Prefer not to say
- 69. How important is religion in your life?
  - 1. Very important
  - 2. Somewhat important
  - 3. Not too important
  - 4. Not at all important

### Political engagement and affiliation (Questions 70–74)
- 70. How often do you follow government and public affairs?
  - 1. Most of the time
  - 2. Some of the time
  - 3. Only now and then
  - 4. Hardly at all
  - 7. Don't know
- 71. With which political party do you most closely identify?
  - 1. African National Congress (ANC)
  - 2. Democratic Alliance (DA)
  - 3. Economic Freedom Fighters (EFF)
  - 4. Inkatha Freedom Party (IFP)
  - 5. Freedom Front Plus (VF+)
  - 6. African Christian Democratic Party (ACDP)
  - 7. United Democratic Movement (UDM)
  - 8. African Transformation Movement (ATM)
  - 9. Good
  - 10. National Freedom Party (NFP)
  - 11. African Independent Congress (AIC)
  - 12. Congress of the People (COPE)
  - 13. Pan Africanist Congress of Azania (PAC)
  - 14. Al Jama-ah (ALJAMA)
  - 15. ActionSA
  - 16. Patriotic Alliance (PA)
  - 17. Forum for Service Delivery (F4SD)
  - 18. Abantu Batho Congress (ABC)
  - 19. MAP16 Civic Movement
  - 20. Independent Civic Organisation of South Africa (ICOSA)
  - 21. African People's Convention (APC)
  - 22. National Coloured Congress (CCC)
  - 23. African People's Movement (APEMO)
  - 24. United Christian Democratic Party (UCDP)
  - 25. Team Sugar South Africa (TSSAP)
  - 26. Concerned Local Residents - Plaaslike Besorgde Inwoners (PBI)
  - 27. Bolsheviks Party of South Africa (BPSA)
  - 28. United Independent Movement (UIM)
  - 29. African Democratic Change (ADeC)
  - 30. Better Residents Association (BRA)
  - 31. Build One South Africa (BOSA)
  - 32. uMkhonto we Sizwe (MK)
  - 98. Other
  - 97. None
  - 99. Don't know
- 72. Self-placement on left-right political scale
  - Slider question that allows respondent to select where their political alignment falls
- 73. Did you vote in the elections for National Assembly held on 29 May 2024?
  - 1. Yes, I voted
  - 2. No, I did not vote
  - 3. Don't want to answer
- 74. Which party did you vote for in the National Assembly?
  - 1. Hope4SA
  - 2. ActionSA
  - 3. African Christian Democratic Party (ACDP)
  - 4. African Independent Congress (AIC)
  - 5. African National Congress (ANC)
  - 6. African Transformation Movement (ATM)
  - 7. Al Jama-ah
  - 8. Build One South Africa (BOSA) With Mmusi Maimaine
  - 9. Congress of the People (COPE)
  - 10. Democratic Alliance (DA)
  - 11. Economic Freedom Fighters (EFF)
  - 12. Freedom Front Plus (FF Plus)
  - 13. Good
  - 14. Inkatha Freedom Party (IFP)
  - 15. National Freedom Party (NFP)
  - 16. Pan Africanist Congress (PAC)
  - 17. Patriotic Alliance (PA)
  - 18. Rise Mzansi (RISE)
  - 19. uMkhonto we Sizwe (MK)
  - 20. United Democratic Movement (UDM)
  - 21. Independent candidate
  - 22. Other
  - 23. Not sure
  - 24. Prefer not to say

### Perceptions of the survey and information treatments (Questions 75–79)
- 75. Do you think this survey was politically biased?
  - 1. Yes, very left-wing biased
  - 2. Yes, somewhat left-wing biased
  - 3. No, it did not feel bias
  - 4. Yes, somewhat right-wing biased
  - 5. Yes, very right-wing biased
- 76. How would you qualify the information we provided you with?
  - 1. Very untrustworthy
  - 2. Somewhat untrustworthy
  - 3. Neither trustworthy nor untrustworthy
  - 4. Somewhat trustworthy
  - 5. Very trustworthy
- 77. How easy was it to understand this survey?
  - 1. Very difficult
  - 2. Difficult
  - 3. Neither difficult nor easy
  - 4. Easy
  - 5. Very easy
- 78. How did you find the length of the survey?
  - 1. Too short
  - 2. About the right length
  - 3. Too long
- 79. Open feedback: "Please feel free to give us any feedback you may have regarding this survey." (Short answer response)

### Sources and references shown to respondents in treatment groups
- Electricity – Status quo
  - • Cost of electricity: International Energy Agency (IEA).
  - • Electricity reliability: World Bank.
- Electricity – Status quo + effect of policies
  - • Cost of electricity: International Energy Agency (IEA).
  - • Electricity reliability: World Bank.
  - Academic Studies:
    - • World Bank.
    - • Inter-American Development Bank (IDB).
- Telecommunication – Status quo
  - • Cost of telecommunication services: International Telecommunication Union (ITU).
  - • Telecommunication services reliability: International Telecommunication Union (ITU); Global System for Mobile Communications (GSMA).
- Telecommunication – Status quo + effect of policies
  - • Cost of telecommunication services: International Telecommunication Union (ITU).
  - • Telecommunication services reliability: International Telecommunication Union (ITU); Global System for Mobile Communications (GSMA).
  - Academic Studies:
    - • World Bank.
    - • Inter-American Development Bank (IDB).

*Private Participation and its Discontents: Insights from Large-Scale Surveys Working Paper No. WP/2024/216*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024216-print-pdf.pdf_