## wpiea2022033-print-pdf

## Source details

**Canonical URL:** [wpiea2022033-print-pdf](https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022033-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2022/english/wpiea2022033-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2022/english/wpiea2022033-print-pdf.pdf.json)

---

### Major findings and interpretation
- Core empirical conclusion: "ideology trumps self-interest in explaining attitude towards reform, and ideology can change with new (expert) information."
- Policy implications:
  - Politicians need to forge a political consensus within existing ideologies or try to alter these; appealing to self-interest alone is not enough.
  - Altering ideology through appropriate communication strategies may be effective, as expert opinion had a "large and significant impact" in the research.
- Future research avenues highlighted:
  - Whether ideology plays a similar role for sectoral reforms (e.g., domestic product market deregulation, international trade liberalization) versus economy-wide policy issues.
  - The role of different types of new information beyond expert opinion (example: covid-19 increased public opinion’s sensitivity to climate change).
  - Cross-country variation: "Does ideology play a larger role in some countries, and if so why?"

### Quantitative decomposition of opinion dispersion (Appendix I, Table A.1)
- Contributions of individual variables to dispersion in opinions about EPL deregulation (Method 3):
  - Female: 0.25%
  - Young: 0.74%
  - Have one or more children: 0.34%
  - Upper class: 0.35%
  - Industry FE: 0.33%
  - Occupation FE: 0.48%
  - Little role of gov. regulating economy: 0.29%
  - Major role of gov. regulating economy: 0.85%
  - Politically right: 0.28%
  - Politically left: 0.60%
  - Lack of trust: 0.64%
  - Competition is good: 0.22%
  - Knowledgeable in EPL: 0.55%
  - Economic policy is important: 0.68%
  - Knowledgeable in economic policy: 0.26%
  - Country FE: 6.87%
  - Error Terms: 86.27%
- Notes: "This table shows the contributions of the each of significant variables only from Table 2 in the main text. Contributions of Industry, Occupation and Country FEs show the sum of each dummy variable within the group."

### Regression evidence on personal vs societal reasons for opposing EPL reform (Appendix II, Table A2)
- Dependent variable: equals 1 if respondent is somewhat or strongly opposed to EPL reforms, 0 if indifferent or supports. Standard errors clustered at country level. Significance levels: * 0.1, ** 0.05, *** 0.01.
- Selected coefficients (standard errors in parentheses):
  - Female: -2.1* (1.10) in (1); -1.0 (1.33) in (2)
  - Young: 0.2 (1.93) in (1); -0.8 (1.84) in (2)
  - Senior: -6.3*** (1.87) in (1); -5.9** (2.12) in (2)
  - Foreign born: 4.7 (3.67) in (1); 5.1 (3.81) in (2)
  - Married: 1.2 (1.53) in (1); 1.5 (1.79) in (2)
  - Have one or more children: -0.2 (1.91) in (1); -0.5 (2.01) in (2)
  - Low skilled: 2.5 (1.69) in (1); 0.8 (1.96) in (2)
  - Lower class: -1.6 (3.04) in (1); -1.6 (4.00) in (2)
  - Upper class: -2.1 (2.21) in (1); -3.4 (2.07) in (2)
  - Unemployed: -7.2** (2.88) in (1); -8.2** (3.09) in (2)
  - Out of labor force: -10.2*** (1.81) in (1); -11.9*** (2.03) in (2)
  - Temporary contract: -1.1 (3.35) in (1); -0.8 (3.35) in (2)
  - Self employed: -11.5*** (2.57) in (1); -11.3*** (2.48) in (2)
  - Not worked previously: -1.4 (4.14) in (1); 0.0 (.) in (2)
  - Little role of gov. regulating economy: 4.9 (2.96) in (1); 4.7 (3.23) in (2)
  - Major role of gov. regulating economy: -2.3** (0.91) in (1); -1.5 (1.12) in (2)
  - Politically right: 3.8 (2.49) in (1); 2.7 (2.37) in (2)
  - Politically left: -3.5* (1.92) in (1); -3.9* (1.86) in (2)
  - Lack of trust: 2.1* (1.05) in (1); 2.4* (1.16) in (2)
  - Competition is good: 1.7 (2.47) in (1); 1.8 (2.30) in (2)
  - Competition is high: -0.3 (1.75) in (1); -0.6 (1.74) in (2)
  - Objective knowledge of EPL: 8.0* (4.10) in (1); 7.9 (4.92) in (2)
  - Economic policy is important: -3.5 (2.41) in (1); -3.4 (3.18) in (2)
  - Self-declared knowledge of economics: -1.6 (1.52) in (1); -2.5 (1.74) in (2)
- Model controls and fit:
  - Country FE: Yes in both columns
  - Industry FE and Occupation FE: included in column (2) only
  - Observations: 3175 in (1); 2901 in (2)
  - R-squared: 0.05 in both columns
- Note: "Dependent variable equals 1 if respondent is opposing EPL reforms due to personal reasons, 0 if due to societal reasons."

### Notes on Appendix III tables A3–A8 (formats and dependent variables)
- Common notes across Tables A3–A8:
  - Dependent variable in many specifications: equals 1 if respondent is somewhat or strongly opposed to EPL reforms, 0 if indifferent or supports.
  - Table A8 dependent variable: raw score from -2 to 2, where -2 = strongly opposed, -1 = somewhat opposes, 0 = indifferent, 1 = somewhat supports, 2 = strongly supports EPL reform.
  - Table A3: "with Weights controlling for demographics but summing to one within each country."
  - Table A4: "with Weights controlling for demographics and country differences in population size."
  - Table A5: Logit with the same baseline regression.
  - Table A6: Probit with the same baseline regression.
  - Table A7: Flip the version with support instead of oppose with no weights; dependent variable equals 1 if respondent is somewhat or strongly supportive, 0 if indifferent or opposing.
  - Standard errors are clustered at country level.
  - Significance levels are consistently: * 0.1, ** 0.05, *** 0.01.
  - "Number of observations in column (3) and (4) drop due to occupation/industry fixed effects because those who did not work previously do not have occupation/industry information."

### Survey instrument: objective, core questions, and treatment
- Survey objective and scope:
  - Elicits respondents' opinions about the effects of making it easy for employers to lay off permanent workers.
  - Distinguishes permanent vs. temporary contract workers.
  - Uses conditional and randomized question flows, sliders, multiple response, and ranking tasks.
- Core questions and response formats:
  - Q22: Perceived impact if it becomes easy for employers to layoff permanent workers.
    - ROWS (randomize options): 1 Layoffs; 2 Hires; 3 How often workers change jobs; 4 Time it takes for an unemployed person to find a job; 5 Chances that a worker with a temporary contract finds a permanent contract job; 6 Chances that an unemployed person find a permanent contract job; 7 Economy's GDP (“increases” means economy is richer)
    - COLUMNS: 1 Reduces; 2 No effect; 3 Increases; 4 Don’t Know
  - Q23: Overall support for making it easy for employers to lay off permanent workers for economic reasons (excluding discriminatory layoffs).
    - INSERT SLIDING SCALE – 1 TO 5: 1 Strongly oppose; 2 Somewhat oppose; 3 Indifferent; 4 Somewhat support; 5 Strongly support
  - Q24: Reasons for support (ASK IF Q23 = 4/5; MULTIPLE RESPONSE)
    - 1 Makes the economy richer and/or creates jobs; 2 Reduces inequality in society; 3 Benefits me personally; 88 Other reasons
  - Q25: Reasons for opposition (ASK IF Q23 = 1/2; MULTIPLE RESPONSE)
    - 1 Makes the economy poorer and/or does not create jobs; 2 Increases inequality in society; 3 Hurts me personally; 4 Might benefit me personally but, on balance, there is a higher chance it will hurt me; 5 Is more likely to benefit me personally rather than hurt me, but even so I would rather not take the chance; 6 Damages the community I live in; 88 Other reasons
- Ranking and conditional follow-ups:
  - Q26: Rank reasons selected at Q25 (ASK IF MORE THAN 1 OPTION IS SELECTED AT Q25)
  - Q27–Q32: Conditional follow-ups probing whether compensatory measures or certainty of personal benefit would change support, with binary responses 1 Yes; 2 No, and follow-up questions on reasons if No.
  - Q33: If Q25 = 88 (other reasons), probe specific “other” reasons (MULTIPLE RESPONSE)
    - 1 Employers have no right to lay off workers freely, even for economic reasons; 2 Free-market (“neo-liberal”) policies are bad; 3 The quest for economic growth is bad; 4 This would be disruptive for the community; 88 Other – Please Specify
- Randomized information treatment (D):
  - One of two participants randomly receives informational text: “Economists have found that making it easy for employers to lay off permanent workers when they feel the economic need to do so is beneficial for the overall economy (productivity goes up, GDP and average income go up; many workers on temporary contracts and the unemployed can get permanent jobs; and unemployed people can find new jobs more quickly).”
  - Q34: Repeat of Q23 after treatment (INSERT SLIDING SCALE – 1 TO 5)
  - Q35: For treated respondents who still oppose (ASK IF TREATED AND Q37 = 1/2 AND Q23 = 1/2; MULTIPLE RESPONSE)
    - 1 I don’t trust the experts’ conclusions.; 2 It still hurts me personally.; 3 It might benefit me personally but, on balance, there is a higher chance it will hurt me.; 4 It is more likely to benefit me personally rather than hurt me, but even so I would rather not take the chance.; 88 Other reasons
  - Q36: Rank reasons selected at Q35 (IF MORE THAN 1 OPTION IS SELECTED AT Q35)
- Instrument features and implementation notes:
  - Multiple question types: grid with sliding arrow, sliding scale 1 to 5, multiple response, single response, rank ordering.
  - Randomization applied to row order in Q22 and option order in multiple response questions.
  - Conditional branching based on previous answers (numerous ASK IF conditions).
  - Special codes: 88 used to denote “Other” responses in multiple items.
  - Treatment group receives economists’ summary; follow-up asks whether treatment changes stated preferences and elicits reasons for continued opposition.

*Appendix III Table A3–A8, as presented in the source PDF.*

### Appendix III Table A3-A8.

### Appendix III Table A3-A8.

### Major findings and interpretation (from V. Conclusion and Appendix results)
- Core empirical conclusion: "ideology trumps self-interest in explaining attitude towards reform, and ideology can change with new (expert) information."
- Policy implications:
  - Politicians need to forge a political consensus within existing ideologies or try to alter these; appealing to self-interest alone is not enough.
  - Altering ideology through appropriate communication strategies may be effective, as expert opinion had a "large and significant impact" in the research.
- Future research avenues highlighted:
  - Whether ideology plays a similar role for sectoral reforms (e.g., domestic product market deregulation, international trade liberalization) versus economy-wide policy issues.
  - The role of different types of new information beyond expert opinion (example: covid-19 increased public opinion’s sensitivity to climate change).
  - Cross-country variation: "Does ideology play a larger role in some countries, and if so why?"

### Quantitative decomposition of opinion dispersion (Appendix I, Table A.1)
- Contributions of individual variables to dispersion in opinions about EPL deregulation (Method 3):
  - Female: 0.25%
  - Young: 0.74%
  - Have one or more children: 0.34%
  - Upper class: 0.35%
  - Industry FE: 0.33%
  - Occupation FE: 0.48%
  - Little role of gov. regulating economy: 0.29%
  - Major role of gov. regulating economy: 0.85%
  - Politically right: 0.28%
  - Politically left: 0.60%
  - Lack of trust: 0.64%
  - Competition is good: 0.22%
  - Knowledgeable in EPL: 0.55%
  - Economic policy is important: 0.68%
  - Knowledgeable in economic policy: 0.26%
  - Country FE: 6.87%
  - Error Terms: 86.27%
- Notes: "This table shows the contributions of the each of significant variables only from Table 2 in the main text. Contributions of Industry, Occupation and Country FEs show the sum of each dummy variable within the group."

### Regression evidence on personal vs societal reasons for opposing EPL reform (Appendix II, Table A2 — full results)
- Dependent variable: equals 1 if respondent is somewhat or strongly opposed to EPL reforms, 0 if indifferent or supports. Standard errors clustered at country level. Significance levels: * 0.1, ** 0.05, *** 0.01.
- Selected coefficients (column labels as in the source; standard errors are reported in parentheses in the source):
  - Female: -2.1* (1.10) in (1); -1.0 (1.33) in (2)
  - Young: 0.2 (1.93) in (1); -0.8 (1.84) in (2)
  - Senior: -6.3*** (1.87) in (1); -5.9** (2.12) in (2)
  - Foreign born: 4.7 (3.67) in (1); 5.1 (3.81) in (2)
  - Married: 1.2 (1.53) in (1); 1.5 (1.79) in (2)
  - Have one or more children: -0.2 (1.91) in (1); -0.5 (2.01) in (2)
  - Low skilled: 2.5 (1.69) in (1); 0.8 (1.96) in (2)
  - Lower class: -1.6 (3.04) in (1); -1.6 (4.00) in (2)
  - Upper class: -2.1 (2.21) in (1); -3.4 (2.07) in (2)
  - Unemployed: -7.2** (2.88) in (1); -8.2** (3.09) in (2)
  - Out of labor force: -10.2*** (1.81) in (1); -11.9*** (2.03) in (2)
  - Temporary contract: -1.1 (3.35) in (1); -0.8 (3.35) in (2)
  - Self employed: -11.5*** (2.57) in (1); -11.3*** (2.48) in (2)
  - Not worked previously: -1.4 (4.14) in (1); 0.0 (.) in (2)
  - Little role of gov. regulating economy: 4.9 (2.96) in (1); 4.7 (3.23) in (2)
  - Major role of gov. regulating economy: -2.3** (0.91) in (1); -1.5 (1.12) in (2)
  - Politically right: 3.8 (2.49) in (1); 2.7 (2.37) in (2)
  - Politically left: -3.5* (1.92) in (1); -3.9* (1.86) in (2)
  - Lack of trust: 2.1* (1.05) in (1); 2.4* (1.16) in (2)
  - Competition is good: 1.7 (2.47) in (1); 1.8 (2.30) in (2)
  - Competition is high: -0.3 (1.75) in (1); -0.6 (1.74) in (2)
  - Objective knowledge of EPL: 8.0* (4.10) in (1); 7.9 (4.92) in (2)
  - Economic policy is important: -3.5 (2.41) in (1); -3.4 (3.18) in (2)
  - Self-declared knowledge of economics: -1.6 (1.52) in (1); -2.5 (1.74) in (2)
- Model controls and fit:
  - Country FE: Yes in both columns
  - Industry FE and Occupation FE: included in column (2) only
  - Observations: 3175 in (1); 2901 in (2)
  - R-squared: 0.05 in both columns
- Note: "Dependent variable equals 1 if respondent is opposing EPL reforms due to personal reasons, 0 if due to societal reasons."

### Notes on Appendix III tables A3–A8 (regression formats and dependent variable definitions)
- Common notes across Tables A3–A8:
  - Dependent variable in many specifications: equals 1 if respondent is somewhat or strongly opposed to EPL reforms, 0 if indifferent or supports.
  - Table A8 dependent variable: raw score from -2 to 2, where -2 = strongly opposed, -1 = somewhat opposes, 0 = indifferent, 1 = somewhat supports, 2 = strongly supports EPL reform.
  - Table A3: "with Weights controlling for demographics but summing to one within each country."
  - Table A4: "with Weights controlling for demographics and country differences in population size."
  - Table A5: Logit with the same baseline regression.
  - Table A6: Probit with the same baseline regression.
  - Table A7: Flip the version with support instead of oppose with no weights; dependent variable equals 1 if respondent is somewhat or strongly supportive, 0 if indifferent or opposing.
  - Standard errors are clustered at country level.
  - Significance levels are consistently: * 0.1, ** 0.05, *** 0.01.
  - "Number of observations in column (3) and (4) drop due to occupation/industry fixed effects because those who did not work previously do not have occupation/industry information."

### Survey instrument details relevant to interpretation
- Key dependent variable framing used across tables:
  - Opposition/support question coded as binary in many regressions (somewhat or strongly opposed/supportive vs indifferent or opposite).
  - Raw attitudinal scale used in Table A8: -2 to 2 (see above).
- Selected socio-demographic and attitudinal question formats used as regressors (exact response categories preserved in the questionnaire):
  - Gender: 1 Male; 2 Female; 3 Other/nonbinary.
  - Age: categories including "18-21", "22-29", "30-39", "40-49", "50-59", "60-65".
  - Household income brackets (US example): "Below $20,000", "$20,001-$40,000", "$40,001-$50,000", "$50,000-$75,000", "$75,001-$100,000", "$100,000-$150,000", "More than $150,000".
  - Role of government in regulating economy: 1 Little/some role; 2 Moderate role; 3 Huge/major role.
  - Political position: 1 Left; 2 Center-Left; 3 Center-Right; 4 Right.
  - Knowledge questions: Q16 self-assessed knowledge on economic policies (1 Not knowledgeable at all, 2 Not very knowledgeable, 3 Somewhat knowledgeable, 4 Highly knowledgeable).
  - Trust questions: variety of 4-point scales for interpersonal and institutional trust.
  - Occupation and industry classification categories listed verbatim in the questionnaire.

*Appendix III Table A3–A8, as presented in the source PDF.*

### Introduction: Labor laws govern layoff procedures f or workers. Different procedures may apply for

### wpiea2022033-print-pdf - Introduction: Labor laws govern layoff procedures f or workers. Different procedures may apply for

### Survey objective and scope
- Elicits respondents' opinions about the effects of making it easy for employers to lay off permanent workers.
- Distinguishes permanent vs. temporary contract workers.
- Uses conditional and randomized question flows, sliders, multiple response, and ranking tasks.

### Core questions and response formats
- Q22: Perceived impact if it becomes easy for employers to layoff permanent workers.
  - ASK ALL; SHOW GRID – SLIDING ARROW; SINGLE RESPONSE PER ROW
  - ROWS (randomize options):
    - 1 Layoffs
    - 2 Hires
    - 3 How of ten workers change jobs
    - 4 Time it takes f or an unemployed person to find a job
    - 5 Chances that a worker with a temporary contract finds a permanent contract job
    - 6 Chances that an unemployed person find a permanent contract job
    - 7 Economy's GDP (“increases” means economy is richer)
  - COLUMNS:
    - 1 Reduces
    - 2 No ef fect
    - 3 Increases
    - 4 Don’t Know

- Q23: Overall support for making it easy for employers to lay off permanent workers for economic reasons (excluding discriminatory layoffs).
  - ASK ALL; SLIDING SCALE; SINGLE RESPONSE
  - INSERT SLIDING SCALE – 1 TO 5
    - 1 Strongly oppose
    - 2 Somewhat oppose
    - 3 Indif ferent
    - 4 Somewhat support
    - 5 Strongly support

- Q24: Reasons for support (ASK IF Q23 = 4/5; MULTIPLE RESPONSE; RANDOMIZE OPTIONS)
  - 1 Makes the economy richer and/or creates jobs
  - 2 Reduces inequality in society
  - 3 Benef its me personally
  - 88 Other reasons

- Q25: Reasons for opposition (ASK IF Q23 = 1/2; MULTIPLE RESPONSE; RANDOMIZE OPTIONS)
  - 1 Makes the economy poorer and/or does not create jobs
  - 2 Increases inequality in society
  - 3 Hurts me personally
  - 4 Might benefit me personally but, on balance, there is a higher chance it will hurt me
  - 5 Is more likely to benefit me personally rather than hurt me, but even so I would rather not take the chance
  - 6 Damages the community I live in
  - 88 Other reasons

### Ranking and conditional follow-ups
- Q26: Rank reasons selected at Q25 (ASK IF MORE THAN 1 OPTION IS SELECTED AT Q25; RANK ORDER; SHOW ALL OPTIONS (1-6) SELECTED AT Q25; AUTOCODE AS RANK 1 IF ONLY 1 OPTION IS SELECTED AT Q25)
  - Desktop: drag and drop; RANDOMIZE OPTIONS
  - Mobile: dropdown ranking; RANDOMIZE OPTIONS

- Q27: If opponent cites increased inequality (Q25 = 2), would compensatory measures that ensure losers become winners change support?
  - ASK IF Q25 = 2; SINGLE RESPONSE
    - 1 Yes
    - 2 No
  - Possible compensatory measures example: higher social benefits

- Q28: If answer to Q27 = 2, reason for continued opposition:
  - ASK IF Q27 = 2; SINGLE RESPONSE
    - 1 Do not trust government
    - 2 Measures are not enough
    - 88 Others- Please specify

- Q29: If opponent cited personal hurt (Q25 = 3), would compensatory measures targeted to you change support?
  - ASK IF Q25 = 3; SINGLE RESPONSE
    - 1 Yes
    - 2 No

- Q30: If Q29 = 2, reason for continued opposition:
  - ASK IF Q29 = 2; SINGLE RESPONSE
    - 1 Do not trust government
    - 2 Measures are not enough
    - 88 Others- Please specify

- Q31: If Q25 = 4 (might benefit but likely hurt), would certainty of personal benefit change support?
  - ASK IF Q25 = 4; SINGLE RESPONSE
    - 1 Yes
    - 2 No

- Q32: If Q25 = 5 (more likely to benefit but prefer not to take the chance), would certainty of personal benefit change support?
  - ASK IF Q25 = 5 AND Q25=/=4; SINGLE RESPONSE
    - 1 Yes
    - 2 No.

- Q33: If Q25 = 88 (other reasons), probe for specific “other” reasons (MULTIPLE RESPONSE; RANDOMIZE OPTIONS)
  - 1 Employers have no right to lay off workers f reely, even for economic reasons
  - 2 Free-market (“neo-liberal”) policies are bad
  - 3 The quest for economic growth is bad
  - 4 This would be disruptive for the community
  - 88 Other – Please Specify

### Randomized information treatment (D)
- One of two participants randomly receives informational text:
  - “Economists have found that making it easy for employers to lay off permanent workers when they feel the economic need to do so is beneficial for the overall economy (productivity goes up, GDP and average income go up; many workers on temporary contracts and the unemployed can get permanent jobs; and unemployed people can find new jobs more quickly).”

- Q34: Repeat of Q23 after treatment (ASK ALL; SLIDING SCALE; SINGLE RESPONSE)
  - INSERT SLIDING SCALE – 1 TO 5
    - 1 Strongly oppose
    - 2 Somewhat oppose
    - 3 Indif ferent
    - 4 Somewhat support
    - 5 Strongly support

- Q35: For treated respondents who still oppose (ASK IF TREATED AND Q37 = 1/2 AND Q23 = 1/2; MULTIPLE RESPONSE; RANDOMIZE OPTIONS)
  - 1 I don’t trust the experts’ conclusions.
  - 2 It still hurts me personally.
  - 3 It might benefit me personally but, on balance, there is a higher chance it will hurt me.
  - 4 It is more likely to benefit me personally rather than hurt me, but even so I would rather not take the chance.
  - 88 Other reasons

- Q36: Rank reasons selected at Q35 (IF MORE THAN 1 OPTION IS SELECTED AT Q35; RANK ORDER; SHOW ALL OPTIONS SELECTED AT Q35; AUTOCODE AS RANK 1 IF ONLY 1 OPTION IS SELECTED AT Q35)
  - Desktop: drag and drop; RANDOMIZE OPTIONS
  - Mobile: dropdown ranking; RANDOMIZE OPTIONS

### Instrument features and implementation notes
- Multiple question types: grid with sliding arrow, sliding scale 1 to 5, multiple response, single response, rank ordering.
- Randomization applied to row order in Q22 and option order in multiple response questions.
- Conditional branching based on previous answers (numerous ASK IF conditions).
- Special codes: 88 used to denote “Other” responses in multiple items.
- Treatment group receives economists’ summary; follow-up asks whether treatment changes stated preferences and elicits reasons for continued opposition.

*Working Paper No. WP/22/33*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022033-print-pdf.pdf_