## 21. ESS Analysis of 90/50 Inequality on Various Social Capital Measures within Subgroups for Ages 25-65

## Source details

**Canonical URL:** [21. ESS Analysis of 90/50 Inequality on Various Social Capital Measures within Subgroups for Ages 25-65](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16176.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16176.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16176.pdf.json)

---

### Introduction
- Observed trends:
  - The share responding that “most people can be trusted” declined from about fifty percent in the early 1970s to thirty-three percent by 2010.
  - Income inequality has increased in the United States and many advanced countries over recent decades.
- Motivation:
  - Tests whether income differences reduce generalized trust given evidence individuals have less trust in people dissimilar to them.
  - Trust links to growth and development through reduced transaction costs, mitigation of principal-agent problems, promotion of trade, financial development, innovation, entrepreneurship, firm productivity, and improved provision of public goods.
- Scope:
  - Main focus: relationship between inequality and generalized trust, plus other social capital variables: helpfulness of others, fairness of others, trust in government, views about redistribution, and life satisfaction.
  - Inequality dimensions examined: 90/10 wage ratio; decomposition into return to education and “residual inequality”; residual P90/P50 and P50/P10; Gini; poverty rate; poverty gap; income shares of top 1 percent, top 5 percent, and top 10 percent.
  - Geographic variation: US states over time (ANES) and European countries over time (ESS).
- Headline:
  - Increase in inequality between 1980 and 2000 explains forty-four percent of the observed decline in trust in the United States.

### Data and Measures
- Data sources:
  - United States: American National Election Surveys (ANES), election years 1980-2012; state-level inequality measures from Census decades 1970-2010 (2010 refers to combined ACS 2009-2011).
  - Europe: European Social Survey (ESS), 2002-2014; OECD Income Distribution Database for inequality measures (equivalent household disposable income).
- Key variables (examples preserved exactly as in source):
  - ANES trust question: “Most people can be trusted” vs “Can’t be too careful”.
  - Other ANES measures: “fairness”, “helpfulness”, trust in government (index 0 to 100), concern about unequal chances (scale 1 to 5), willingness to increase spending on the poor (0/1/2), preference for more/less government.
  - US wage inequality: ratio between the 90th and 10th percentiles of the log wage distribution for individuals ages 25-55, working 30+ hours/week, not self-employed or in group quarters or armed forces.
  - Residual inequality: state-year measure after adjusting for gender, race, age, education, industry, occupation, and returns to observables.
  - ESS social capital measures: “trust people”, “people are fair”, “people are helpful” scaled 0-10; satisfaction in government and with life 0-10; dummy for strong agreement that government should reduce income gaps.
- Descriptive trend examples (plotted values in source):
  - 90/10 ratio of hourly earnings: 1.284, 1.312, 1.396, 1.423, 1.495 (across 1970-2010).
  - College premium: 0.348, 0.470, 0.499, 0.554 (across 1980-2010).
  - ANES Trust People sample means examples: 0.541, 0.451, 0.405, 0.498, 0.504.
  - ESS Trust People sample means examples: 5.064, 4.937, 5.027, 4.681, 4.865, 4.914, 5.242.
  - ESS Disposable Income 90/10 examples: 3.071, 3.249, 3.318, 3.309, 3.396, 3.171, 3.354.
  - ESS Gini - Disposable Income examples: 0.258, 0.306, 0.305, 0.298, 0.301, 0.294, 0.288.

### Empirical Strategy
- Baseline model:
  - Trust_ijt = β1 * Inequality_jt + β2 * X_ijt + μ_j + δ_t + ε_ijt
    - X_ijt: individual controls and time-varying area-level characteristics.
    - μ_j: area fixed-effects (state or country).
    - δ_t: year fixed-effects (ANES uses fixed effects for each survey year).
- Controls:
  - United States (ANES): employment status, family income terciles, age, education, marital status, gender, race, religion; state-level mean wage, property and violent crime rates, education and age distributions, percent foreign born (ages 25-55).
  - Europe (ESS): employment status, age, education, marital status, gender, GDP per capita, percent immigrants.
- Identification and robustness:
  - Exploits geographic variation over time.
  - Robustness checks include many personal and area controls, inclusion/exclusion sensitivity, and clustered standard errors (by state in ANES; by country in ESS).

### United States (ANES) — Core Findings
- Main empirical patterns (Ages 20-45):
  - Residual P90/P10 ratio: negative and significant effect on “trust in people” in full specification.
  - Decomposition:
    - Residual P90/P50 (top-end inequality): no effect on trust.
    - Residual P50/P10 (bottom-end inequality): significantly lowers trust.
- Coefficients and accounting:
  - Full-specification coefficient on residual P90/P10 ratio: -0.909 (standard error 0.415).
  - Actual increase in residual P90/P10 ratio from 1980 to 2000: 0.067.
  - Predicted decline in trust = -0.909 * 0.067 = -0.06.
  - Unexplained decline in trust over 20 years (trend coefficient -0.007 * 20 years) = -0.14.
  - Therefore, inequality increase explains 44 percent of unexplained decline in trust.
- Robustness:
  - Negative effect of residual P50/P10 persists with basic and extended controls.
  - When multiple inequality measures included jointly, residual P50/P10 remains dominant and statistically significant.
- Heterogeneity:
  - Stronger effects for younger individuals (20-45) than older (46-65).
  - Effects much stronger for non-college graduates relative to college graduates.
  - Largest effects for the lowest income tercile; willingness to spend more on the poor reduced mainly for middle and upper income groups.
  - Gender: similar results for men and women.
- Other social capital outcomes (full specification examples):
  - Residual P90/P10 ratio significant and negative for:
    - trust in people,
    - people are fair,
    - people are helpful,
    - trust in government index,
    - willingness to spend more on the poor.
  - Return to education: no effect on trust or other social capital measures.
  - Top income shares (top 1%, top 5%, top 10%): weak, largely insignificant negative effects on trust; some reduction in willingness to spend on the poor.
  - Poverty rate (U.S. Census poverty): not robustly related to trust in people or government.
- Older sample (46-65):
  - Residual P90/P10 generally insignificant.
  - Residual P90/P50 sometimes positive and significant for older individuals.
  - Interpretation: older individuals may have formed trust attitudes earlier in life; current inequality less influential.

### Europe (ESS) — Core Findings
- Overall pattern:
  - Negative relationship between inequality and trust in other people confirmed across advanced European countries.
  - Effects in Europe are more general across the distribution and population subgroups than in the United States.
- Statistically significant measures (examples preserved):
  - Disposable Income Gini: significant and robust negative effects on trust, perceived fairness, perceived helpfulness; increases belief government should reduce income gaps; lowers satisfaction in government, satisfaction with life, and happiness.
    - Example coefficient: Gini - Disposable Income associated with Trust People -5.217*** (1.825) in a specification.
  - Disposable Income 50/10 and poverty rate (relative poverty: incomes < half the median): significant negative effects on trust and other social capital measures.
    - Example coefficient: Disposable Income 50/10 associated with Trust People -0.850*** (0.265).
  - Disposable Income 90/10: negative and significant in several specifications (example: -0.346*** (0.106)).
  - Top income shares (top 1%, top 5%, top 10%): tend to increase view that government should reduce income gaps; little evidence they reduce trust in people (coefficients on trust, fairness, helpfulness are positive and sometimes significant).
- Joint specification and heterogeneity:
  - When multiple inequality components included, P90/P50 and P50/P10 ratios have similar effects.
  - Effects similar across education groups and gender — the negative impact on trust and related measures is broadly shared.
  - Some evidence top-end inequality increases demand for government action to reduce income gaps in Europe.
- Contrast with United States:
  - European effects are more general (affect younger and older, college and non-college).
  - Poverty measures defined relative to the median are more consistently associated with reduced trust in Europe than U.S. absolute poverty measures.

### Mechanisms and Interpretation
- Source of inequality matters:
  - Inequality arising from returns to education (between-group inequality) does not reduce trust.
  - Residual inequality (within age, education, industry, occupation groups) reduces trust.
  - Interpretation: inequality among similar people more likely perceived as unfair and reduces trust; inequality justified by human capital investments is less damaging.
- Distributional location matters:
  - In the United States, inequality at the bottom (P50/P10) is the most robust predictor of reduced trust.
  - In Europe, both bottom and top components (and overall measures like the Gini) matter; top shares increase support for redistribution.
- Behavioral implications:
  - Those most adversely affected by bottom-end inequality (less educated, bottom income tercile) show stronger reductions in trust in the U.S.
  - In Europe, responses to inequality are more diffuse across groups, suggesting broader societal perceptions.

### Policy-Relevant Implications
- Economic and social channels:
  - Rising income inequality can indirectly affect macroeconomic performance by lowering social capital (generalized trust) that supports transactions, cooperation, and institutions.
- Targeted policy priorities indicated by findings:
  - Targeting reductions in bottom-end inequality or addressing residual within-group disparities may be particularly important for preserving or restoring trust.
  - U.S. policy implication: reduce bottom-end residual inequality to improve trust among those most affected (less educated, lower-income).
  - European policy implication: address inequality across the distribution to reduce widespread declines in social capital and respond to increased demand for redistribution.
- Causal interpretation:
  - Robustness to many controls and inclusion of multiple inequality measures supports a causal interpretation, although no single specification can definitively rule out all omitted factors.

### Key empirical coefficients and statistics (selected exact values)
- ANES time trend for Trust People (examples): -0.004***, -0.005***, -0.005***, -0.005***, -0.006***, -0.006***, -0.005***, -0.007***, -0.007*** (standard errors often 0.001).
- ANES residual 90/10 (Ages 20-45): Trust People -0.909** (0.415).
- ANES residual 50/10 (Ages 20-45): Trust People -1.399** (0.616).
- ANES Gini (Ages 20-45): Trust People -3.579** (1.558).
- ESS Disposable Income 90/10 examples: 3.071, 3.249, 3.318, 3.309, 3.396, 3.171, 3.354.
- ESS Gini - Disposable Income examples: 0.258, 0.306, 0.305, 0.298, 0.301, 0.294, 0.288.
- ESS regression examples:
  - Disposable Income 90/10 associated with Trust People -0.346*** (0.106) in one specification.
  - Disposable Income 50/10 associated with Trust People -0.850*** (0.265).
  - Gini - Disposable Income associated with Trust People -5.217*** (1.825).

*Source: 21. ESS Analysis of 90/50 Inequality on Various Social Capital Measures within Subgroups for Ages 25-65 (IMF Working Paper content).*

### REFERENCES_________________________________________________________________________________23

### _wp16176 - REFERENCES_________________________________________________________________________________23

### FIGURES
- 1. Trust has declined sharply in the United States__________________________________________ 6
- 2. The 90/10 ratio of hourly earnings over time in the U.S. ________________________________12
- 3. Change in the college premium in the united states____________________________________ 13
- 4. Residual inequality measures int eh united states_______________________________________ 13
- 5. Residual inequality by state in the u.s.:1980-2010_______________________________________ 14

### TABLES
- 1. ANES Summary Statistics - Means of Demographic Variables___________________________26
- 2. ANES Summary Statistics - Means of Trust and Other Social Capital Variables_________ 27
- 3. ANES Trends in "Trust People" in the United States from 1980-2000___________________ 28
- 4. ANES Trends in Trust and Other Social Capital Variables________________________________29
- 5. Summary Statistics of Demographics and Social Capital Variables in the ESS Sample__ 30
- 6. Summary Statistics of Inequality Measures in the ESS Sample__________________________ 31
- 7. ESS Trends in "Trust People" in Europe from 2002-2014________________________________ 32
- 8. ANES Analysis of Inequality and Trust with Robustness Checks to Control Variables   
20-45 Year Olds__________________________________________________________________________33
- 9. ANES Analysis of Inequality and Trust with Robustness Checks to Control Variables  
46-65 Year Olds__________________________________________________________________________34
- 10. ANES Analysis of Overall Inequality Measures and Social Capital Outcomes  
(Ages 20-45)_____________________________________________________________________________ 35
- 11. ANES Analysis of Inequality at the Bottom End and Social Capital Outcomes  
(Ages 20-45)_____________________________________________________________________________ 36
- 12. ANES Analysis of Inequality at the Top End and Social Capital Outcomes  
(Ages 20-45)_____________________________________________________________________________ 37
- 13. ANES Analysis of Social Capital with Multiple Inequality Variables Together___________  38
- 14. ANES Analysis of Residual 50/10 Inequality on Social Capital by Subgroups___________  39
- 15. ANES Analysis of Residual 90/50 Inequality on Social Capital by Subgroups____________ 40
- 16. ESS Analysis of Inequality and Trust with Robustness Checks____________________________41
- 17. ESS Analysis of Inequality on Various Social Capital Measures for Ages 25-65__________ 42
- 18. ESS Analysis of Inequality on Various Social Capital Measures for Ages 25-65__________ 43
- 19. ESS Analysis of 90/10 Inequality on Various Social Capital Measures within Subgroups  
for Ages 25-65___________________________________________________________________________ 44
- 20. ESS Analysis of 50/10 Inequality on Various Social Capital Measures within Subgroups  
for Ages 25-65___________________________________________________________________________ 45

*Source: _wp16176 - REFERENCES_________________________________________________________________________________23*

### 21. ESS Analysis of 90/50 Inequality on Various Social Capital Measures within Subgroups

### 21. ESS Analysis of 90/50 Inequality on Various Social Capital Measures within Subgroups for Ages 25-65

### Introduction
- Observed trends:
  - The share responding that “most people can be trusted” declined from about fifty percent in the early 1970s to thirty-three percent by 2010.
  - Income inequality has increased in the United States and many advanced countries over recent decades.
- Motivation:
  - Literature shows individuals have less trust in people dissimilar to them; this study tests whether income differences reduce generalized trust.
  - Trust is linked to growth and development through reduced transaction costs, mitigation of principal-agent problems, promotion of trade, financial development, innovation, entrepreneurship, firm productivity, and improved provision of public goods.
- Key research scope:
  - Main focus on relationship between inequality and generalized trust, plus other social capital variables: helpfulness of others, fairness of others, trust in government, views about redistribution, and life satisfaction.
  - Multiple inequality dimensions examined: 90/10 wage ratio, decomposition into return to education and “residual inequality,” residual P90/P50 and P50/P10, Gini, poverty rate, poverty gap, income shares of top 1 percent, top 5 percent, and top 10 percent.
  - Geographic variation exploited: US states over time (ANES) and European countries over time (ESS).
- High-level headline:
  - Increase in inequality between 1980 and 2000 explains forty-four percent of the observed decline in trust in the United States.

### Data and Measures
- Data sources:
  - United States: American National Election Surveys (ANES), election years 1980-2012; state-level inequality measures from Census decades 1970-2010 (2010 refers to combined ACS 2009-2011).
  - Europe: European Social Survey (ESS), 2002-2014; OECD Income Distribution Database for inequality measures (equivalent household disposable income).
- Key sample definitions and variables:
  - ANES trust question: “Most people can be trusted” vs “Can’t be too careful”; also measures of “fairness”, “helpfulness”, trust in government (index 0 to 100), concern about unequal chances (scale 1 to 5), willingness to increase spending on the poor (0/1/2), and preference for more/less government.
  - Wage inequality (US): ratio between the 90th and 10th percentiles of the log wage distribution for individuals ages 25-55, working 30+ hours/week, not self-employed or in group quarters or armed forces.
  - Residual inequality: state-year measure after adjusting for gender, race, age, education, industry, occupation, and returns to observables.
  - ESS social capital measures: “trust people”, “people are fair”, “people are helpful” scaled 0-10; satisfaction in government and with life 0-10; dummy for strong agreement that government should reduce income gaps.
- Notable descriptive trends and measures:
  - Change in the 90/10 ratio of hourly earnings over decades, with plotted values including 1.284, 1.312, 1.396, 1.423, 1.495 across 1970-2010.
  - Change in the college premium (college wage premium) with plotted values 0.348, 0.470, 0.499, 0.554 across 1980-2010.
  - Residual 90/10 ratio lines shown for different control sets; aggregate residual inequality trend shows most overall inequality trend unexplained by observed characteristics and returns.

### Empirical Strategy
- Baseline model:
  - Trust_ijt = β1 * Inequality_jt + β2 * X_ijt + μ_j + δ_t + ε_ijt
    - X_ijt: individual controls and time-varying area-level characteristics.
    - μ_j: area fixed-effects (state or country).
    - δ_t: year fixed-effects (ANES uses fixed effects for each survey year).
- Controls:
  - United States (ANES): employment status, family income terciles, age, education, marital status, gender, race, religion; state-level mean wage, property and violent crime rates, education and age distributions, percent foreign born (ages 25-55).
  - Europe (ESS): employment status, age, education, marital status, gender, GDP per capita, percent immigrants.
- Identification:
  - Exploits geographic variation over time; robustness checks include many personal and area controls and inclusion/exclusion sensitivity.

### United States (ANES) — Core Findings
- Main patterns for 20-45 year olds:
  - Residual P90/P10 ratio: negative and significant effect on “trust in people” in full specification.
  - Decomposition:
    - Residual P90/P50 (inequality at the top): no effect on trust.
    - Residual P50/P10 (inequality at the bottom): significantly lowers trust.
  - Coefficient magnitudes and interpretation:
    - Full-specification coefficient on residual P90/P10 ratio: -0.909.
    - Actual increase in residual P90/P10 ratio from 1980 to 2000: 0.067.
    - Predicted decline in trust = -0.909 * 0.067 = -0.06.
    - Unexplained decline in trust over 20 years (trend coefficient -0.007 * 20 years) = -0.14.
    - Therefore, inequality increase explains 44 percent of unexplained decline in trust.
- Robustness across controls:
  - Negative effect of residual P50/P10 persists when including basic (age, education, gender) and extended controls (race, religion, income, employment status, state-level measures).
  - When multiple inequality measures are included jointly, residual P50/P10 remains dominant and statistically significant while others become insignificant.
- Heterogeneity:
  - Age: stronger effects for younger individuals (20-45) than older (46-65).
  - Education: effects much stronger for non-college graduates relative to college graduates.
  - Income terciles: largest effects for the lowest income tercile; willingness to spend more on the poor reduced mainly for middle and upper income groups.
  - Gender: similar results for men and women.
- Other social capital outcomes (full specification):
  - Residual P90/P10 ratio is significant and negative for:
    - trust in people,
    - people are fair,
    - people are helpful,
    - trust in government index,
    - willingness to spend more on the poor.
  - Return to education: no effect on trust or other social capital measures.
  - Top income shares (top 1%, top 5%, top 10%): weak, largely insignificant negative effects on trust; some reduction in willingness to spend on the poor.
  - Poverty rate (U.S. Census poverty): not robustly related to trust in people or government.
- Older sample (46-65):
  - Residual P90/P10 generally insignificant.
  - Residual P90/P50 sometimes positive and significant for older individuals.
  - Interpretation: older individuals may have formed trust attitudes earlier in life; current inequality less influential.

### Europe (ESS) — Core Findings
- Overall:
  - Negative relationship between inequality and trust in other people is confirmed across advanced European countries.
  - Unlike the United States, the effect in Europe is more general across the distribution and population subgroups.
- Key statistically significant measures:
  - Gini coefficient: significant and robust negative effects on trust, perceived fairness, perceived helpfulness; increases belief government should reduce income gaps; lowers satisfaction in government, satisfaction with life, and happiness.
  - P50/P10 ratio and poverty rate (relative poverty: incomes < half the median): significant negative effects on trust and other social capital measures.
  - P90/P10 ratio: negative and significant only in specifications that include percent immigrants and GDP per capita; otherwise less robust.
  - Top income shares (top 1%, top 5%, top 10%): tend to increase the view that government should reduce income gaps; little evidence they reduce trust in people (coefficients on trust, fairness, helpfulness are positive and sometimes significant).
- Joint specification and heterogeneity:
  - When multiple inequality components included, P90/P50 and P50/P10 ratios have similar effects; employment rates also significant for several outcomes.
  - Effects similar across education groups and gender — the negative impact on trust and related measures is broadly shared.
  - Some evidence that top-end inequality increases demand for government action to reduce income gaps in Europe (contrasts with the United States).
- Contrasts with United States:
  - European effects are more general (affect both younger and older, college and non-college; not confined to those directly harmed).
  - Poverty measures in Europe (relative) more consistently associated with reduced trust than U.S. absolute poverty measure.

### Mechanisms and Interpretation
- Source-specific effects:
  - Inequality arising from returns to education (between-group inequality) does not reduce trust.
  - Residual inequality (within age, education, industry, occupation groups) — potentially reflecting luck, unobserved ability, or unexplained gaps — reduces trust.
  - Interpretation: inequality among similar people appears more likely to be perceived as unfair and reduces trust; inequality justified by human capital investments is less damaging to trust.
- Distributional location matters:
  - Inequality at the bottom (P50/P10) is the most robust predictor of reduced trust in the United States.
  - In Europe, both bottom and top components (and overall measures like the Gini) matter, with top shares increasing demand for redistribution.
- Behavioral implications:
  - Those most adversely affected by bottom-end inequality (less educated, bottom income tercile) show stronger reductions in trust in the U.S.
  - In Europe, responses to inequality are more diffuse across groups, suggesting broader societal perceptions.

### Conclusion
- Main conclusions:
  - Inequality at the bottom of the distribution lowers an individual’s sense of trust in others in both the United States and Europe.
  - For the United States:
    - The increase in inequality between 1980 and 2000 explains forty-four percent of the observed decline in trust.
    - The dominant adverse component is residual P50/P10 (inequality at the bottom within observable groups), with effects concentrated among the less educated and lower-income individuals.
    - Returns to education have no detectable effect on trust.
    - No evidence that increases in top 1 percent or top 5 percent shares have driven declines in trust.
  - For Europe:
    - Inequality reduces trust, fairness, and helpfulness perceptions broadly across subgroups.
    - Both top and bottom parts of the distribution can matter; top shares increase support for government intervention to reduce income gaps.
    - Poverty measures defined relative to the median are consistently associated with lower trust.
- Policy-relevant implications:
  - Rising income inequality can indirectly affect macroeconomic performance by lowering social capital (generalized trust) that supports transactions, cooperation, and institutions.
  - Targeting reductions in bottom-end inequality or addressing residual within-group disparities may be particularly important for preserving or restoring trust.
  - Different institutional contexts and perceptions across the United States and Europe imply that policy responses may need to be tailored: U.S. policies that reduce bottom-end residual inequality could improve trust among those most affected; European policies addressing inequality across the distribution may reduce widespread declines in social capital and increase support for redistribution.
- Causal inference:
  - Robustness to many controls and to inclusion of multiple inequality measures supports a causal interpretation, although no single specification can definitively rule out all omitted factors.

*Source: 21. ESS Analysis of 90/50 Inequality on Various Social Capital Measures within Subgroups for Ages 25-65 (IMF Working Paper content).*

### References

### _wp16176 - References

### Major themes in the referenced literature
- Trust, social capital, and their links to economic outcomes:
  - Alesina A. and E. La Ferrara (2002), “Who Trusts Others?”, Journal of Public Economics, 85, 207–234.
  - Algan, Y. and P. Cahuc (2010), “Inherited Trust and Growth”, American Economic Review, Vol. 100, December 2010: 2060–2092.
  - Guiso, L., Sapienza, P., and L. Zingales (2004), “The Role of Social Capital in Financial Development”, American Economic Review, 94(3), 526–556.
  - Knack, S. and P. Keefer (1997), “Does Social Capital Have an Economic Payoff? A Cross-Country Investigation”, Quarterly Journal of Economics, 112(4), 1252–1288.
  - Zak, P.J. and S. Knack, (2001), “Trust and growth”, Economic Journal, 111 (470), 291–321.

- Culture, institutions, and economic behavior:
  - Alesina, A., and P. Giuliano (2016), “Culture and Institutions”, Journal of Economic Literature, forthcoming.
  - Tabellini, G. (2008), “Institutions and Culture”, Journal of the European Economic Association, 6(2–3), 255–294.
  - Tabellini, G. (2010), “Culture and Institutions: Economic Development in the Regions of Europe”, Journal of the European Economic Association, 8(4), 677–716.

- Inequality, redistribution, and macroeconomic consequences:
  - Galor, O. and J. Zeira (1993). “Income Distribution and Macroeconomics,” Review of Economic Studies, 60(1), 35-52.
  - Dabla-Norris, E., K. Kochhar, N. Suphaphiphat, F. Ricka, E. Tsounta (2015), “Causes and Consequences of Income Inequality : A Global Perspective”, Staff Discussion Notes, No. 15/13.
  - Ostry, J. D., A. Berg, and C. Tsangarides (2014), “Redistribution, Inequality, and Growth.” IMF Staff Discussion Note 14/02.
  - Cingano, F. (2014), “Trends in Income Inequality and its Impact on Economic Growth”, OECD Social, Employment and Migration Working Paper, No. 163.

- Political economy, government quality, and accountability:
  - La Porta R., Lopez-de-Silanes, F., Shleifer, A., and R. Vishny (1999), “The Quality of Government”, Journal of Law Economics and Organization, 15(1), 222–279.
  - Nannicini, T., Stella, A., Tabellini, G., and U. Troiano (2012), “Social Capital and Political Accountability”, American Economic Journal: Economic Policy, 5, 222–250.
  - Rothstein, B. and E.M. Uslaner (2005), “All for All: Equality, Corruption, and Social Trust”, World Politics, Vol.58, pp. 41-72.

### Data sources and empirical materials cited
- Integrated Public Use Microdata Series: Version 3.0 [Machine-readable database]. Minneapolis, MN: Minnesota Population Center [producer and distributor]. (Ruggles et al., 2010)
- ANES summary statistics and regressions tables are included in the document with detailed summary statistics and regression outputs for trust and social capital measures (see Tables 1–21).
  - Table 1: ANES Summary Statistics - Means of Demographic Variables (selected means)
    - Age examples: 39.85, 39.29, 39.872, 40.32, 39.72, 39.77, 39.57, 40.70, 40.83, 40.07, 41.92, 42.76
    - Male proportions: 0.430, 0.459, 0.440, 0.458, 0.436, 0.481, 0.492, 0.470, 0.455, 0.455, 0.450, 0.480
    - Education shares (HS Dropout, HS Graduate, Some College, College Grad, Advanced Degree): e.g., HS Dropout 0.255, 0.224, 0.214, 0.175, 0.167, 0.180, 0.142, 0.115, 0.109, 0.108, 0.091, 0.090; College Grad 0.127, 0.122, 0.118, 0.127, 0.156, 0.151, 0.178, 0.154, 0.178, 0.178, 0.204, 0.206; Advanced Degree 0.044, 0.053, 0.063, 0.081, 0.064, 0.072, 0.080, 0.100, 0.085, 0.079, 0.097, 0.105
    - Minimum Observations examples: 1681, 1684, 1165, 1031, 1542, 1444, 1872, 980, 1275, 1206, 1747, 4516

  - Table 2: ANES Summary Statistics - Means of Trust and Other Social Capital Variables (selected means)
    - Trust People: 0.541, 0.451, 0.405, 0.498, 0.504
    - People are Fair: 0.629, 0.552, 0.670, 0.687
    - People are Helpful: 0.553, 0.589, 0.620, 0.676
    - Trust Index examples: -0.016, -0.005, -0.047, -0.064, -0.132
    - Trust Govt to Benefit All: 0.232, 0.247, 0.207, 0.286, 0.315, 0.247, 0.222, 0.317, 0.356, 0.512, 0.301, 0.194
    - Never Attend Religious Service: 0.209, 0.236, 0.243, 0.220, 0.209, 0.334, 0.342, 0.333, 0.336, 0.346, 0.399, 0.449
    - Minimum/Maximum Observations examples: Minimum Observations 1402, 1879, 1135, 1127, 1415, 1540, 1754, 1007, 715, 534, 827, 2396; Maximum Observations 1775, 1913, 1294, 1144, 1640, 1562, 2002, 1028, 1479, 1214, 1845, 4666

- ESS (European Social Survey) summary statistics and inequality measures (Tables 5–7, 16–21) with many country-level and individual-level sample sizes and means:
  - Table 5 (ESS sample demographics examples, years 2002–2014): Age means: 41.946, 42.063, 42.526, 42.284, 42.931, 43.020, 43.356; Male proportions: 0.480, 0.467, 0.466, 0.463, 0.461, 0.464, 0.473; College Graduate shares: 0.248, 0.258, 0.314, 0.310, 0.324, 0.343, 0.349; Trust People means: 5.064, 4.937, 5.027, 4.681, 4.865, 4.914, 5.242; Satisfied with Life means: 7.029, 6.844, 6.775, 6.544, 6.609, 6.694, 7.397
  - Table 6 (ESS inequality measures examples, years 2002–2014): Disposable Income 90/10: 3.071, 3.249, 3.318, 3.309, 3.396, 3.171, 3.354; Gini - Disposable Income examples: 0.258, 0.306, 0.305, 0.298, 0.301, 0.294, 0.288; Top 1% Income Share examples: 8.830, 9.348, 9.637, 9.084, 8.509, 8.770; Poverty Rate examples: 0.062, 0.103, 0.101, 0.097, 0.100, 0.097, 0.101; Employment Rate examples: 77.999, 79.016, 80.734, 79.489, 78.634, 78.928, 80.237

### Empirical results and reported coefficients (selected highlights from regressions and robustness tables)
- ANES (United States) trends and coefficients (Tables 3–4, 8–15):
  - Time trend (Year) coefficients for "Trust People" in multiple specifications: -0.004***, -0.005***, -0.005***, -0.005***, -0.006***, -0.006***, -0.005***, -0.007***, -0.007*** (standard errors reported as 0.001 in many columns).
  - Age and education coefficients examples (Table 3): HS Graduate 0.181***, Some College 0.259***, College Grad 0.429***, Advanced Degree 0.447*** (standard errors vary by column).
  - Table 4 reported Year coefficients across outcomes: Trust People -0.007*** (0.001); People are Fair -0.001 (0.001); People are Helpful 0.000 (0.001); Trust Index -0.007*** (0.002); Trust Government Index -0.124*** (0.025); Never Attend Religious Service 0.001** (0.000); Increase Spending on Poor -0.016*** (0.002); It is a problem if chances are unequal -0.006*** (0.002); Favor Larger Government -0.010*** (0.001). Observations by dependent variable given (e.g., 3,337 for Trust People).

- ANES inequality and trust regressions (selected coefficients, Tables 10–15):
  - Residual 90/10 coefficients (Ages 20-45, Table 10): Trust People -0.909** (0.415); People are Fair -2.042*** (0.458); People are Helpful -2.118*** (0.510); Trust Index -3.620*** (0.803); Trust Government Index -38.931** (17.102); Increase Spending on Poor 0.516 (0.335); It is a problem if chances are unequal -2.539*** (0.582); Favor Larger Government 1.067 (0.904).
  - Gini coefficients (Table 10): Trust People -3.579** (1.558); People are Fair -6.597*** (2.112); People are Helpful -6.682*** (2.273); Trust Index -13.376*** (3.247); Trust Government Index -118.231** (56.578).
  - Residual 50/10 (Table 10 / 11): Trust People -1.399** (0.616); People are Fair -3.800*** (0.927); People are Helpful -2.713*** (0.754); Trust Index -5.014*** (1.253); Trust Government Index -64.059*** (22.429); Increase Spending on Poor 0.901** (0.444); It is a problem if chances are unequal -2.481** (0.984).
  - Top income shares (Table 12): Top 1% Share coefficients include -0.006 (0.009) for Trust People and -0.022* (0.013) for People are Helpful; Top 5% Share includes coefficients such as -0.008 (0.008) for Trust People and -0.023** (0.011) for People are Helpful.

- ESS (Europe) regressions and robustness checks (Tables 7, 16–21):
  - Time trend (Year) coefficients for "Trust People" in ESS: 0.028*** (0.002) in many columns; some specifications show 0.009***, 0.010***, -0.009***, 0.002 with reported standard errors.
  - Education and gender effects (Table 7): Male 0.087*** (0.017) in some columns; education categories (Educ ISCED 2 to 5) show sizable positive coefficients, e.g., Educ ISCED 5 1.753*** (0.033).
  - Robustness to inequality measures (Table 16): Disposable Income 90/10 coefficient examples: -0.206 (0.132) to -0.368*** (0.094) across columns; Disposable Income Gini coefficients: -5.959** (2.162) to -6.282*** (2.141); Disposable Income 50/10 coefficients: -0.842** (0.355) to -0.904*** (0.316); Top 5% Income Share positive coefficients in some columns: 0.012 (0.009) to 0.025* (0.012).
  - ESS broader social capital outcomes (Table 17, 18): Gini - Disposable Income associated with Trust People -5.217*** (1.825); Disposable Income 90/10 associated with Trust People -0.346*** (0.106); Disposable Income 50/10 associated with Trust People -0.850*** (0.265). Top 1% Income Share sometimes shows positive associations with some outcomes: 0.031* (0.016) for Trust People in one specification.

### Notes on methodology and reporting in the referenced tables
- Reported coefficients are from separate OLS regressions in the ANES and ESS analyses.
- Standard errors are clustered by state (ANES) or by country (ESS) and are reported in parentheses in all tables.
- Significance levels use one, two, or three stars to represent 10 percent, 5 percent, and 1 percent levels, respectively.
- Regressions typically include personal controls (age, education, gender, marital status, race category, religion group, income category, employment status), and fixed-effects (state or country and year) and sometimes state- or country-level controls (mean log income for full-time workers, education distribution, age distribution, percent immigrants, property crime rate, violent crime rate, GDP per capita).

*Source: _wp16176 - References*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16176.pdf_