## wpiea2023035-print-pdf

## Source details

**Canonical URL:** [wpiea2023035-print-pdf](https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023035-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2023/english/wpiea2023035-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2023/english/wpiea2023035-print-pdf.pdf.json)

---

### Background and Motivation
- Purpose: use life satisfaction data to estimate UK residents’ willingness to pay (WTP) to abate air pollution (NO2, PM2.5, PM10) to inform climate mitigation and environmental policy design in the United Kingdom.
- Context:
  - Major local pollutants in the UK: Nitrogen Dioxide (NO2) and Particulate Matter PM2.5 and PM10.
  - Levels of NO2, PM2.5, and PM10 have been significantly higher than national and international targets (EU Ambient Air Quality Directive) and remain above WHO recommended guidelines despite a decreasing trend 2010–2019.
  - Environmental Audit Committee estimates health costs from air pollution range between £8.5 billion and £20.2 billion every year.
  - DEFRA (2015) estimates a £2.7 billion loss to the UK economy from declines in productivity due to air contamination.
- Observed relationship with wellbeing: negative relationship between pollution levels and life satisfaction ranking after a threshold; study emphasizes capturing benefits and costs of pollution abatement for policymaking.
- Policy context: Environment Bill (2021) and Clean Air Strategy aim to set targets, action plans, measures, and deadlines to improve air quality; WTP estimates inform cost-benefit analysis and green tax design (Lindahl tax concept referenced).

### Objective of the Study
- Primary objective:
  - Estimate UK residents’ valuation of air quality for NO2, PM2.5, and PM10 abatement by using short-term and long-term happiness measures.
- Policy relevance:
  - Marginal WTP determines marginal private benefits and total benefits from abating a pollutant; can be compared against abatement costs in cost-benefit analysis (Srinivasan and Stewart, 2004).
  - Provides inputs for drafting the Environmental Bill (2021) and designing green tax schemes (Marrouch and Sinclair-Desgagné, 2012).
- Conceptual channels linking pollution to wellbeing (study focuses exclusively on neuropsychological channel and controls for health and income):
  - Health channel: pollution → deteriorated physical health → decreased psychic well-being.
  - Income/productivity channel: pollution → lower health → reduced productivity/income → lower happiness.
  - Neuropsychological channel: exposure produces “environmental stress,” disturbing autonomic nervous system functioning and increasing negative feelings (disgust, depression).
- Methodological approach:
  - Micro-econometric life satisfaction regression from an indirect utility function including income, pollutant, and covariates.
  - Compute Marginal Rate of Substitution (MRS) between income and pollutant to proxy WTP (Oswald and Wu, 2010). Formal model in Section 5.

### Previous Literature (overview and empirical findings)
- Broad findings:
  - Many studies find a negative relationship between air pollutants and life satisfaction across cross-sectional, finer spatial matching, and panel-fixed effects approaches.
  - Notable estimates and contexts (preserve reported figures and methodologies):
    - Luechinger (2010): willingness to pay ≈ 1.1 percent of income for a unit abatement using 2SLS (vs 0.6 percent using OLS).
    - Cuñado and Pérez de Gracia (2012): mean-income citizen implicitly willing to pay 1.4 percent of yearly salary to decrease PM10 exceedance days by 1 percent.
    - Jidong and Yiran (2015): one 휇gm-3 decline in NO2 valued at approximately US$189 (China).
    - Ferreira et al. (2013): one 휇gm-3 increase in SO2 decreases life satisfaction by 0.016–0.030 points on an 11-point scale.
    - Liu et al. (2018): individuals willing to pay US$130 annually to decrease NO2 by one 휇gm-3 (China).
    - Welsch (2003): abating NO2 by a standard deviation valued at US$900; lead at US$1400 (10 European countries).
    - Knight and Howley (2017): a standard deviation increase in NO2 negatively affects happiness by ≈ 0.015 standard deviations (UK).
  - Some studies report neutral or indeterminate effects; endogeneity (including reverse causality and subjective pollution measures) can alter conclusions (e.g., Lin et al. (2019); Goetzke and Rave (2015)).

### Contribution to the Literature
- Addresses gaps:
  - Provides updated UK valuation covering period 2004 – 2019.
  - Targets NO2, PM10, and PM2.5 (PM2.5 especially underrepresented in prior valuation literature).
  - Uses panel methods to control for unobserved time-invariant characteristics.
  - Geo-matches pollution at the LSOA level (LSOA: 1,000 to 3,000 residents; study restricted to England and Wales).
  - Distinguishes life satisfaction (long-term evaluation) and hedonic happiness (short-term); hypothesizes short-term happiness more elastic to current pollution.
  - Assesses heterogeneity across sub-groups (education, urban residence) to inform equity and policy targeting.

### Theoretical Model and Welfare Interpretation
- Functional form:
  - Cobb-Douglas assumed: LS_it = ∏_c D_itc^(β_c) (equation (1)).
  - Log-linear form: lLS_it = Σ_c β_c lD_itc (equation (2)), where lLS_it = ln(LS_it) and lD_itc = ln(D_itc).
- Life satisfaction as proxy for utility supported by literature (Diener et al., 1999; DiTella et al., 2003).
- Marginal Rate of Substitution (MRS) derived:
  - MRS = − (∂ lLS_it / ∂ D_itp) / (∂ lLS_it / ∂ D_itI) (equation (3)); interpretable as monetary WTP to offset marginal pollution change.
- Welfare use:
  - MRS can construct marginal benefit curves for pollutant abatement; intersection with marginal cost yields socially optimal abatement Q*.
  - MRS informs design of Pigouvian taxes via marginal external cost (MEC), although some firm-side parameters remain unknown.

### Data: coverage, sampling, and pollutants
- Dataset:
  - Period: 2004 – 2019 (panel, yearly frequency).
  - Main sources: merged British Household Panel Survey (BHPS) and United Kingdom Household Longitudinal Survey (UKHLS / Understanding Society).
    - BHPS: 1991–2008 (~10,000 individuals).
    - UKHLS: started 2009 (~50,000 individuals).
  - Geo-identification: LSOA of each individual in each wave; geographic scope restricted to England and Wales.
- Survey sampling notes:
  - ~20 percent face-to-face interviews; remainder by other methods including telephone.
  - UKHLS General Population Sample (GPS): ≈ 25,000 households in first wave; proportionally stratified and clustered.
  - Ethnic Minority Boost Sample (EMBS) included to oversample five key ethnic groups (EMBS focuses on areas dense with these groups and "has at least 1,000 surveyed adults per ethnic group").
- Dependent variables:
  - Life Satisfaction: ordinal, values 1–7.
  - Hedonic happiness: ordinal, 0–36 (sum of 12 mental health covariates scaled 0–3); higher hedonic score = more distressed.
- Covariates: age (and age^2), gender, marital status, education, labour force status, empirical income (equivalized by OECD scale, deflated, logged, and squared logged included), health satisfaction, household size, children, GOR, year.
- Sample descriptive statistics (selected):
  - Respondents aged 15+; average age = 47.
  - Male:Female ratio = 46:54.
  - Average household size = three.
  - 75 percent completed high school; ~33 percent have university degree.
  - Labour force: ≈ 5 percent unemployed, 55 percent employed, 40 percent not in labor force.
  - Average monthly deflated income (OECD scale) = 2,086 British Pounds (£).
  - Average monthly income per household without OECD scale or inflation = £3,679.
  - Median monthly household income = £3,035.
  - Median life satisfaction = 5; median health satisfaction = 6.
  - Urban residence ≈ 80 percent.
- Pollutants data:
  - Pollutants: NO2, PM2.5, PM10; unit μg/m3.
  - Source: DEFRA modelled 1x1 km2 grids, aggregated to LSOA by selecting pollutant value closest to LSOA population-weighted centroid (ESRI ArcGIS Spatial Join).
  - Regulatory annual maximum safe levels reported:
    - NO2 and PM10: 40 μgm-3.
    - PM2.5: 25 μgm-3.
  - Stylized facts:
    - Lowest recorded levels for NO2, PM10, PM2.5 in South East of England (2018, 2017, 2017 respectively).
    - Highest recorded levels in London (2005, 2004, 2004 respectively).
    - Pairwise correlations: correlation(PM2.5, PM10) = 0.94; correlation(NO2, PM2.5) = 0.733.
    - Due to high correlations, each pollutant considered separately in regressions.
    - LSOA-mean ambient levels declined over time:
      - NO2: 17.32 μg/m3 in 2004 and 12.03 μg/m3 in 2020.
      - PM2.5: 13.163 μg/m3 in 2004 and 8.788 μg/m3 in 2020.
    - Decline partly attributed to policy acts including the Climate Change Act (2008) and Planning and Energy Act (2008).

### Empirical Model and Estimation Strategy
- Core model (expanded log-linear specification):
  - H_ijt = β_0 + β_1 P_jt + β_2 age_it + β_3 age_it^2 + β_4 Income_it + β_5 Income_it^2 + β_6 sex_it + β_7 health_it + β_8 educ_it + β_9 emp_it + β_10 marital_it + θ_1 GOR_j + θ_2 year_t + μ_ijt
  - H_ijt = life satisfaction or hedonic happiness of respondent i in LSOA j at time t; P_jt = ambient pollutant in LSOA j at t.
- Implementation notes:
  - Dependent variables and pollutants logged; GOR dummies included for regional fixed effects; year dummies for time effects.
  - Weather variables considered but excluded to avoid data loss and aggregation bias.
- Estimation techniques:
  - Primary: ordinal logit with random or fixed effects (random effects usually preferred in ordinal logit literature).
  - Robustness: pooled logit (standard errors clustered at LSOA), random effects probit.
  - Clustering at LSOA addresses serial correlation (pollutant varies at LSOA; happiness at individual level).
- Standardization:
  - Standardized regressions (dependent and independent variables) used to compare pollutant effects across happiness measures and relative importance versus other life factors.
- WTP / MRS computation:
  - Applying MRS yields:
    - MRS_ijt = − β_1 / (β_4 + 2 β_5 Incomē)
  - Interpret MRS as monetary valuation (happiness price) of pollution; invariant to monotone transformations of life satisfaction.
- Subgroup analyses:
  - Re-estimate and compute MRS for: (i) individuals with high education; (ii) urban residents.
- Endogeneity considerations:
  - Reverse causality from happiness to pollution considered unlikely; potential reverse causality between happiness and income addressed via sample restriction (housewives approach following Pischke (2011)) to test robustness. Results interpreted cautiously.

### Results and Discussion — Key Findings
- Estimation methods: random effects ordinal logit (primary), pooled ordinal logit and random effects ordinal probit (robustness). Results consistent across models.

Effects on long-term Life Satisfaction
- Main pollutant elasticities (Table 6, controlling for health assessment):
  - 1 percent increase in NO2 → odds of higher (log) life satisfaction decreases by approximately 9 percent.
  - 1 percent increase in PM10 → odds decreases by approximately 9.5 percent.
  - 1 percent increase in PM2.5 → odds decreases by approximately 10.7 percent.
  - PM2.5 has the largest impact on likelihood of being happy (effects net of controlled health assessment; interpreted as psychic/neuropsychological channel).

Covariate effects (life satisfaction)
- Age: negative but diminishing effect.
- Unemployment (vs employed or not in labour force): negative effect.
- Married (vs widowed/divorced/separated): positive effect.
- Income: non-linear (initially higher income associated with lower happiness; at higher levels associated with higher happiness).
- Education: higher education associated with lower odds of long-term life satisfaction (similar to Clark (2003)).

Effects on short-term Hedonic Happiness
- Main pollutant elasticities (Table 9):
  - 1 percent increase in NO2 → odds of higher (log) hedonic unhappiness increase by approximately 7.9 percent.
  - 1 percent increase in PM10 → odds increase by approximately 9.9 percent.
  - 1 percent increase in PM2.5 → odds increase by approximately 9.9 percent.
  - PM10 and PM2.5 have larger short-run impacts on hedonic happiness.
- Covariate effects (short-term):
  - Income: positive, non-linear effect.
  - Men: higher odds of short-term happiness.
  - Education: having some education increases short-term happiness (GCSE > A-level > degree).
  - Employment: employed, retired, or student (vs unemployed) increases short-term happiness.
  - Married (vs other statuses): increases short-term happiness.
  - Self-perceived health: better perceived health increases odds of short-term happiness.
- Note: higher values of hedonic happiness variable represent higher levels of short-term unhappiness (more distress).

Relative and Standardized Magnitudes
- Standardized regressions (Tables 12 and 13):
  - Life satisfaction standardized effects:
    - 1 standard deviation (SD) increase in NO2 or PM2.5 decreases odds of a higher SD of life satisfaction by 1.1 percent.
    - For PM10: 0.9 percent.
  - Hedonic happiness standardized effects:
    - 1 SD increase in any pollutant decreases odds of a higher SD of hedonic happiness by roughly 1 percent.
  - Conclusion: pollutants have similar intensity effects on both happiness measures; PM2.5 slightly stronger—expected given high pairwise correlations.

Comparisons with Major Life Events
- Life satisfaction (standardized comparisons):
  - Unemployment (vs paid employment) decreases odds of 1 SD increase in happiness by ≈ 9 percent (≈ 9 times pollutant disutility).
  - Divorce (vs marriage) decreases odds by ≈ 4 percent (≈ 4 times pollutant disutility).
- Hedonic happiness:
  - Unemployment (vs in-paid employment) ≈ 14 times as important as pollution.
  - Divorce plays almost the same effect as air pollution.
- Interpretation: life events have stronger impacts than pollution, but pollution still materially affects wellbeing.

Happiness Price / Willingness to Pay (MRS results)
- WTP evaluated at mean income (entire sample, Table 14):
  - Avoid one 휇g/m3 increase in PM10: almost £60 monthly.
  - Avoid one 휇g/m3 increase in PM2.5: £103 monthly.
  - Avoid one 휇g/m3 increase in NO2: £62.5 monthly.
- Subsamples:
  - Urban sample (≈ 66 percent of full-sample amounts):
    - PM10: £40.5
    - PM2.5: £74
    - NO2: £40.5
  - Highly educated sample:
    - PM10: £60.66
    - NO2: £83.19
    - PM2.5: £83
  - Note: variance in distribution of payments across pollutants ~ 3.75 times smaller in highly educated sample.
- Comparative literature references:
  - Welsch (2003): maximum US$900 per year to abate one 휇g/m3 of NO2 (ten European countries).
  - Ferreira and Modo (2010) for Ireland: US$870 annually to abate one 휇g/m3 of PM10.

Endogeneity of Income
- Reverse causality concern addressed using Pischke (2011) approach (restricted sample of married housewives with partner, unemployed/students/long-term sick):
  - Relationship income → happiness remains U-shaped and significant.
  - Pollutant coefficients remain similar and significant.
  - Interpretation: positive unidirectional causality from income to happiness present; other variables (e.g., marriage) may still be endogenous and not instrumented here.

### Conclusions and Policy Implications
- Empirical summary:
  - 1 percent higher levels of NO2, PM10, PM2.5 significantly decrease odds of log happiness by 9, 9.5 and 10.7 percent respectively (controlling for demographics and socioeconomic conditions).
  - Short-term hedonic effects: 7.9, 9.9, and 9.9 percent respectively.
  - Standardized results: pollutants have roughly the same effect on life satisfaction and hedonic happiness.
  - Disutility comparisons: a 1 휇g/m3 increase in pollutants generates disutility approximately 4 and 9 times less important than divorce and unemployment, respectively (life satisfaction); for hedonic happiness, unemployment is 14 times more important while divorce is roughly equal to pollution.
  - WTP (evaluated at mean income, monthly per 1 휇g/m3 abated): NO2 = £62.5; PM10 = £60; PM2.5 = £103. Urban dwellers willing to pay less; highly educated individuals willing to pay more for most pollutants.
- Policy implications:
  - Welfare increases from pollution control are substantial; efficiency grounds for environmental policy supported.
  - Equity and heterogeneity matter: policies should be education- and city-specific.
  - Transport sector targeted measures recommended (transportation substantive source of pollutants and greenhouse gases).
  - Environmental (green) tax reforms can be informed by happiness-based WTP estimates (indirect green consumption taxes suggested).
  - Public perception of government efforts matters for happiness; media should improve information delivery on government efforts to reduce air pollution.
  - Education increases environmental valuation and pro-environmental behavior; recommend public education campaigns.
- Methodological note and future research:
  - Happiness-regression approach provides valuations without stated-preference biases, but potential endogeneity of happiness and other explanatory variables remains a limitation; future work should explore instrumental variables and novel empirical strategies.

_italic: Source — IMF Working Paper — wpiea2023035-print-pdf (sections 1.1–1.2, 2.1, 3–6, Appendix I)._

### 1.1 Background and Motivation.......................................................................... 4

### 1.1 Background and Motivation

### Context and motivation
- The paper uses life satisfaction data to help design climate mitigation policies in the United Kingdom by estimating the willingness to pay of UK citizens to abate air pollution.
- Abating air pollution has been an ongoing UK policy concern since the industrial revolution (Environmental Protection UK, 2021).
- The major local pollutants in the UK identified are Nitrogen Dioxide (NO2) and Particulate Matter PM2.5 and PM10.
- Levels of NO2, PM2.5, and PM10 have been significantly higher than targets set nationally and internationally (EU Ambient Air Quality Directive) and remain above World Health Organization (WHO) recommended guidelines despite a decreasing trend over the past decade (Figure 1: Trend mean pollutants level 2010-2019).

### Measured economic and health impacts cited
- Environmental Audit Committee estimates the value of health costs from air pollution to range between £8.5 billion and £20.2 billion every year.
- Air contamination with pollutants resulted in approximately a £2.7 billion loss to the UK economy due to declines in productivity (DEFRA, 2015).

### Observed relationship with wellbeing
- Figure 2 shows a negative relationship between pollution levels and life satisfaction ranking after a certain threshold.
- The study emphasizes that outputs are important for researchers and policymakers to capture benefits and costs of pollution abatement in decision-making.

### Policy context
- Following Brexit, the UK government is drafting and enforcing the Environment Bill (2021) to set targets, action plans, measures, and deadlines to improve air quality in line with the Clean Air Strategy and WHO guidelines (DEFRA, 2021).
- Estimating willingness to pay can inform cost-benefit analyses for environmental proposals and assist design of green tax schemes; an estimate of the tax generated from marginal willingness to pay is referred to in the literature as a Lindahl tax.

---

### 1.2 Objective of the Study

### Primary objective
- Estimate UK residents’ valuation of air quality in terms of NO2, PM2.5, and PM10 abatement by tapping into their short-term and long-term happiness.

### Policy decision relevance
- Estimating the marginal willingness to pay for environmental quality determines marginal private benefits and total benefits from abating a pollutant (the demand for environmental quality).
- These monetary valuations can be compared against abatement costs in cost-benefit analysis to evaluate environmental proposals (Srinivasan and Stewart, 2004).
- The estimates provide input for policymakers drafting the Environmental Bill (2021) and for designing green tax schemes to correct for air pollution externalities (Marrouch and Sinclair-Desgagné, 2012).

### Conceptual channels linking pollution to wellbeing
- The effect of environmental quality on well-being can be divided into three channels:
  - Health channel: pollution causes deteriorations in physical health, leading to decreased psychic well-being (documented in epidemiological literature; Sørensen et al., 2014; Babisch et al., 2014; WHO, 2013).
  - Income/productivity channel: lower health from pollution affects productivity, income, and eventually happiness (Ostblom and Samakovlis, 2004).
  - Neuropsychological channel: exposure to air pollution may produce “environmental stress,” disturbing autonomic nervous system functioning and lowering emotional happiness (Evans and Cohen, 2004; Stenlund et al., 2009). Exposure is also affiliated with increased negative feelings and psychological states, including disgust or depression (Ferreira et al., 2013).
- This study focuses exclusively on the third channel and controls for health and income to isolate neuropsychological effects.

### Methodological approach
- The study uses a micro-econometric life satisfaction regression starting from an indirect utility function with arguments including income, the air pollutant, and other personal covariates.
- After estimating the function, the marginal rate of substitution (MRS) between income and the air pollutant is calculated to determine how much income a consumer is willing to give up to abate a unit of pollution; this MRS serves as a proxy for willingness to pay.
- Conditional on appropriate functional form and econometric specification, this method is argued to be robust and practical (Oswald and Wu, 2010). The formal model is presented in Section 5.

*Source: IMF Working Paper — 1.1 Background and Motivation and 1.2 Objective of the Study.*

### 2.1  Previous Literature

### wpiea2023035-print-pdf - 2.1  Previous Literature

### 2.1  Previous Literature — overview
- The first step in estimating benefits of pollution abatement within a life satisfaction micro-econometric regression framework is to compute the estimate of additional units of pollution on happiness.
- A summary of the literature is provided in Table 1 of the Appendix (not reproduced here).

### 2.1.1  Negative Effects — empirical findings
- A significant proportion of the literature finds a negative relationship between air pollutants and life satisfaction.
- Cross-sectional or pooled cross-sectional studies:
  - Luechinger (2010) (EU Eurobarometer): SO2 negatively impacts happiness. OLS then 2-Stage Least Squares (instrumenting local pollution with foreign sources) to account for endogeneity. Willingness to pay: approximately 1.1 percent of income to abate a unit of the pollutant (compared to 0.6 percent using OLS).
  - Cuñado and Pérez de Gracia (2012) (Spain, ESS 2008): OLS at regional level. Negative association. Mean-income citizen implicitly willing to pay 1.4 percent of yearly salary to decrease number of days PM10 exceeds 50 micrograms per cubic meter (휇gm-3) by 1 percent.
  - Jidong and Yiran (2015) (China, Chinese General Social Survey 2010): Ordered Probit with NO2 varying at city level. A one 휇gm-3 decline valued at approximately US$189.
    - Note: this estimate is considerably lower than European estimates; possible explanations include empirical differences, lower standards of living in China, and the Environmental Kuznets Curve "technique effect" (Copeland and Taylor, 2004).
  - Ferreira et al. (2013) (multiple ESS waves, regional up to NUTS-3): OLS. A one 휇gm-3 increase in SO2 decreases life satisfaction by a range of 0.016-0.030 points on an 11-point scale.
  - Liu et al. (2018) (China, World Values Survey pooled 2001, 2007, 2012, regional level): Ordered Probit and Feasible Generalized Least Squares. Negative relationship for NO2 and PM10. Individuals willing to pay US$130 every year in taxes to decrease NO2 by one 휇gm-3.
- Studies with finer spatial matching:
  - Ambrey et al. (2014) (Australia, HILDA): geo-matched at Collection District level; Ordered Probit; negative relationship for PM10 exceedances on life satisfaction.
  - Orru et al. (2016) (Estonia, ESS 2010 and 2012): 1x1 km2 grid geo-matching; OLS; negative association for PM10.
- Panel data and fixed-effects studies:
  - Welsch (2003) (10 European countries, World Database of Happiness, 1990-1997): panel FE GLS. Valuation: abating NO2 by a standard deviation valued at US$900; lead at US$1400.
  - Zhang et al. (2017) (China Family Panel Studies, city-level, panel with fixed effects): API effect not significant on overall life satisfaction but significant negative relationships with mental health and short-term hedonic happiness; effects asymmetric and larger for those working outdoors, lower salaries, or living in more polluted areas.
  - Knight and Howley (2017) (UK, Understanding Society 2002-2014): panel FE, NO2 at LSOA level. A standard deviation increase in NO2 negatively affects happiness by approximately 0.015 standard deviations. Authors compare this estimate with coefficients of other regressors in life satisfaction regression.

- Additional note: For physiological and psychological mechanisms linking air pollution to well-being, see Lin et al. (2019). Lin et al. (2019) also examine pollution’s impact on health (random effects ordinal logit) and find negative effects on odds ratios of satisfaction with health for NO2, PM2.5, and PM10, with PM2.5 strongest.

### 2.1.2  Neutral / Indeterminate Effects — empirical findings and endogeneity
- Some studies find no significant effects of pollution on happiness:
  - Lin et al. (2019) (Taiwan): multilevel latent growth modeling (LGM); city-level geo-matching; variation in happiness not significantly explained by NO2, PM2.5, or PM10.
- Endogeneity concerns can alter conclusions:
  - Goetzke and Rave (2015) (Germany, German Socioeconomic Panel Data): IV-probit using subjective perceptions of NO2, PM10, and SO2. They find endogeneity from reverse causality biases results—note this arises because they use subjective rather than objective pollution measures.

### 2.2  Contribution to the Literature — how this paper extends prior work
- Gaps this paper seeks to address:
  - Limited valuation studies for the UK; existing studies often cover outdated periods. This study provides new insights for the UK for the period 2004 – 2019.
  - Targets three pollutants: NO2, PM10, and PM2.5 — with PM2.5 severely lacking in literature despite being among the most dangerous pollutants (European Environmental Agency, 2009).
  - Uses panel data methods to control for unobserved time-invariant characteristics affecting life satisfaction, improving on many cross-sectional studies.
  - Models air pollution at the LSOA level to reduce aggregation bias and the ecological fallacy (LSOA: 1,000 to 3,000 residents; as of 2011 there are 34,753 LSOAs in England and Wales). The study is restricted to England and Wales.
  - Distinguishes life satisfaction (overall evaluation) from hedonic (short-term) happiness; models both, hypothesizing short-term happiness is more elastic to current pollution.
  - Assesses heterogeneous effects across sub-groups to inform policymakers.

### 3. Theoretical Model — key specifications and welfare interpretation
- Life satisfaction functional form assumed Cobb-Douglas:
  - LS_it = ∏_c D_itc^(β_c)  (equation (1))
  - Log-linear form: lLS_it = Σ_c β_c lD_itc  (equation (2)), where lLS_it = ln(LS_it) and lD_itc = ln(D_itc).
- Life satisfaction measures are treated as proxies for utility, supported by multidisciplinary evidence (Diener et al., 1999 and DiTella et al., 2003).
- Marginal Rate of Substitution (MRS) between pollution and income derived from total differential of (2):
  - MRS = − (∂ lLS_it / ∂ D_itp) / (∂ lLS_it / ∂ D_itI)  (equation (3)), where D_itp is ambient pollutant level and D_itI is income.
- Uses MRS as a welfare-cost (or benefit) measure useful in cost-benefit analyses:
  - MRS can aid in constructing marginal benefit curves for pollutant abatement; intersection with marginal cost yields socially optimal abatement Q* (Figure 1, Appendix).
  - MRS can partially inform the marginal external cost (MEC) used to design Pigouvian taxes that move firm output from profit-maximizing Qp to socially optimal Qs (Figure 2, Appendix).
  - Notes some firm-side parameters (MPC and MB) remain unknown; willingness to pay remains helpful (Callan and Thomas, 2013).

### 4. Data — dataset coverage and sampling
- Dataset period: 2004 – 2019 (panel, yearly frequency).
- Main sources: merged British Household Panel Survey (BHPS) and United Kingdom Household Longitudinal Survey (UKHLS / Understanding Society).
  - BHPS: 1991 to 2008 (18 waves), ~10,000 individuals.
  - UKHLS: started 2009, ~50,000 individuals.
  - Merged to generate longer panel; final dataset limited to 2004 – 2019 due to pollutant data availability constraints.
- Surveys are nationally representative of the UK population and focus on socioeconomic conditions, health, values, housing, well-being, environment, etc.
- Geo-identification: merged dataset identifies LSOA of each individual in each wave. LSOAs: designed for small area statistics, 1,000 to 3,000 residents; no boundary changes like wards or postcodes.
- Geographic scope: restricted to England and Wales (Scotland and Northern Ireland use different geographic units).
- Sampling design:
  - Approximately 20 percent of each survey’s sample receives face-to-face interviews; remainder via other methods including telephone.
  - BHPS began 1991-1992 (England only initially), expanded to Wales and Scotland in 1999 and Northern Ireland in 2000; BHPS incorporated into UKHLS (Lynn, 2009).
  - UKHLS General Population Sample (GPS): approximately 25,000 households in first wave; proportionally stratified and clustered sample of addresses in England, Scotland, and Wales.

_italics: Source — wpiea2023035-print-pdf - 2.1  Previous Literature (IMF Working Paper content provided) _

### section is the Ethnic Minority Boost Sample (EMBS), designed specifically to create an oversampling of five key ethnic

### Ethnic Minority Boost Sample (EMBS), designed specifically to create an oversampling of five key ethnic groups

### Sampling design and components
- The EMBS section focuses only on areas dense with these groups and "has at least 1,000 surveyed adults per ethnic group."
- The study dataset is composed of four sections:
  - Ethnic Minority Boost Sample (EMBS) — oversample of five key ethnic groups.
  - General Population Sample (GPS).
  - General Population Comparison Sample (GPCS) — a subsample of the GPS whose interviews include all questions asked in the EMBS to allow comparative analysis between the general population and specific minority groups.
  - British Household Panel Survey Sample (ex-BHPS) — originally introduced in 1991, expanded geographically (Wales and Scotland in 1999 and Northern Ireland in 2001), ending with over 8,000 households.
- GPS and BHPS use a stratified clustered design extracted from postcode sectors; postcode sectors were grouped based on addresses, then stratified through a list of characteristics (region, the percentage of pensionable age, the percentage of agriculturally employed individuals (in urban regions), and the percentage of single households with no pensions (in non-urban regions)).

### Dependent variables and covariates used
- Dependent variables:
  - Life Satisfaction: ordinal, values 1 (least satisfied) to 7 (most satisfied).
  - Hedonic happiness: ordinal, possible values 0 to 36; sum of 12 mental health covariates each scaled 0 (never) to 3 (most of the time). Higher hedonic score = more distressed.
- Covariates included (following the literature):
  - age (include squared age to capture U-shaped relationship)
  - gender (male/female)
  - marital status
  - level of education
  - labor force status
  - state of employment
  - income (absolute household income, equivalized by OECD scale, deflated by ONS official inflation rates, logged, and squared logged value included)
  - health satisfaction (used as proxy for subjective health assessment)
- Rationale and notes on covariates:
  - Education, labor force status, and income have mixed findings in literature; income is equivalized by the OECD scale and deflated, then logged; squared logged income included to allow for nonconstant marginal utility effects.
  - Health satisfaction included because severe health problems and disabilities negatively affect subjective well-being in prior literature.

### Key descriptive statistics (sample characteristics)
- Sample composition and central tendencies:
  - Respondents aged 15 and over; average age = 47.
  - Male:Female ratio = 46:54.
  - Slightly more than half of respondents are married; almost one-third have never been married.
  - Average household size = three.
  - Education: 75 percent have at least completed their high school education; almost one third have a university degree.
  - Labor force status: approximately 5 percent are unemployed, 55 percent are employed, and 40 percent are not part of the labor force (disabled, full-time student, retired, on government training scheme, etc.).
  - Average monthly deflated income (after OECD scale) = 2,086 British Pounds (£).
  - Average monthly income per household without OECD scale or inflation = £3,679.
  - Median monthly household income = £3,035.
  - Median ratings (scale 1 to 7): life satisfaction = 5; health satisfaction = 6.
  - Urban residence: roughly 80 percent live in urban areas.
- Source and period: Based on authors’ own calculations relying on data from BHPS and UKHLS years 2004 – 2019.

### Pollutants data: dataset, measures, and stylized facts
- Pollutants and measurement:
  - Pollutants used: NO2, PM2.5, PM10 (main regressors).
  - Measurement unit: microgram per cubic meter (μg/m3).
  - Data extracted for period 2004 – 2019 from DEFRA; modelled for 1x1 km2 grids and aggregated to LSOA level by selecting pollutant value closest (in Euclidean distance) to LSOA’s population-weighted centroid using ESRI ArcGIS Spatial Join Tool.
- Regulatory annual maximum safe levels (as reported):
  - NO2 and PM10: 40 μgm-3.
  - PM2.5: 25 μgm-3.
- Sources and health risks (as summarized from EPA and DEFRA): NO2 mainly from transportation, electricity, industrial production; PM sources include transportation, fires, powerplants. Literature disagrees which pollutant is most damaging; NO2 linked to smog, low-level ozone, pulmonary effects; PM linked to cardiovascular and respiratory effects.
- Stylized facts and statistics:
  - Lowest recorded levels of NO2, PM10, and PM2.5 were in the South East of England (registered in 2018, 2017, and 2017 respectively).
  - Highest recorded levels for these pollutants were registered in London (recorded in 2005, 2004, and 2004 respectively).
  - Several LSOA regions surpassed the UK’s legal annual mean ambient levels (Table 4 in Appendix referenced in source).
  - Pairwise correlations among pollutants (Table 5 referenced):
    - correlation(PM2.5, PM10) = 0.94
    - correlation(NO2, PM2.5) = 0.733
  - Given high pairwise correlations, each pollutant is considered separately in regressions to avoid multicollinearity.
  - LSOA-mean ambient levels declined over time:
    - NO2: 17.32 μg/m3 in 2004 and 12.03 μg/m3 in 2020.
    - PM2.5: 13.163 μg/m3 in 2004 and 8.788 μg/m3 in 2020.
  - Decline attributed in part to increased eco-friendly awareness and policy acts, including the Climate Change Act (2008) and Planning and Energy Act (2008).

### Empirical model specification and estimation strategy
- Core estimated model (expanded equation (2) into (4)):
  - H_ijt = β_0 + β_1 P_jt + β_2 age_it + β_3 age_it^2 + β_4 Income_it + β_5 Income_it^2 + β_6 sex_it + β_7 health_it + β_8 educ_it + β_9 emp_it + β_10 marital_it + θ_1 GOR_j + θ_2 year_t + μ_ijt
  - Where H_ijt stands for life satisfaction or hedonic happiness of respondent i in LSOA j at time t; P_jt denotes ambient pollutant (NO2, PM2.5, PM10) in LSOA j at time t; educ, emp, marital are vectors of dummies; GOR and year are fixed effects.
- Model notes and implementation choices:
  - Dependent variables and pollutants are in logs.
  - Dummy variables for GOR are included for regional fixed effects at GOR level rather than LSOA due to high number of LSOA regions.
  - Year dummy variables included to capture time fixed effects.
  - Weather variables were considered but not included to avoid significant data loss and aggregation bias.
- Estimation techniques and robustness:
  - Primary estimator: ordinal logit model using maximum likelihood with either fixed or random effects; random effects usually preferred in the ordinal logit literature due to potential inconsistency of fixed effects.
  - Results presented using odds ratios for interpretation.
  - Robustness checks: pooled logit (with standard errors clustered at LSOA level) and random effects probit.
  - Clustering at LSOA level addresses serial correlation because pollutant varies at LSOA while happiness varies at individual level.
- Standardization and comparative analyses:
  - Standardized (dependent and independent) variables will be estimated to:
    - Compare effect of pollutant on life satisfaction versus hedonic happiness.
    - Inspect relative importance of pollution versus other life factors on life satisfaction and hedonic happiness.
    - Examine relative importance of different pollutants.
- Willingness-to-pay (WTP) and marginal rate of substitution (MRS) computation:
  - Applying (3) to (4) yields (5) for MRS for mean income (Incomē):
    - MRS_ijt = − β_1 / (β_4 + 2 β_5 Incomē)
  - MRS interpreted as monetary valuation (happiness price) of pollution: amount individuals are willing to pay to compensate for a marginal increase in ambient pollutant while keeping life satisfaction constant.
  - MRS is invariant to monotonically increasing transformations of life satisfaction.
- Subgroup analyses:
  - Estimations and MRS calculations repeated for:
    - (i) individuals with high levels of education
    - (ii) urban residents
  - Rationale: provide policymakers information relevant for equity and efficiency.
- Endogeneity and reverse causality considerations:
  - Reverse causality from happiness to pollution considered highly unlikely and possibly insignificant because:
    - happiness measured at individual level while pollution at LSOA level;
    - any effect of happiness on pollution likely occurs with a lag;
    - happiness is subjective, pollution is objective.
  - Potential reverse causality from happiness to income is considered; methodology to test endogeneity follows Pischke (2011) by restricting sample to housewives who (i) are currently with a partner/husband, (ii) earn a household income, and (iii) are unemployed, study full time, or are unable to work due to sickness — effectively reducing direct income generation by the woman and testing whether coefficients remain similar. Any conclusions from this restriction should be interpreted cautiously.

*Italic: Based on content from IMF Working Paper section describing the Ethnic Minority Boost Sample (EMBS) and accompanying variables, descriptive statistics, pollutants data, and empirical model.*

### 6. Results and Discussion

### 6. Results and Discussion

### 6.1 Effects of Local Air Pollutants on Life Satisfaction
- Regression method: random effects ordinal logit (Table 6); pooled ordinal logit (Table 7) and random effects ordinal probit (Table 8) used as robustness checks; results similar across models.
- Main pollutant effects (from Table 6, controlling for health assessment):
  - When NO2, PM10, and PM2.5 increase by 1 percent, the odds of higher (log) life satisfaction decreases by approximately 9, 9.5, and 10.7 percent respectively.
  - PM2.5 has the biggest impact on the likelihood of being happy.
  - Interpretation: effects reported net of the controlled health assessment; thus captured changes reflect psychic and psychological channels (unless perceptions of health diverge from actual health outcomes).
- Covariate effects:
  - Age: negative, but diminishing, effect on happiness.
  - Unemployment (vs employed or not in labor force): negative effect on life satisfaction.
  - Married (vs widowed, divorced or separated): positive effect.
  - Income: non-linear (initially higher income associated with lower happiness; eventually higher income associated with higher happiness).
  - Education: higher levels of education associated with lower odds of having long-term life satisfaction (similar to Clark (2003)).

### 6.2 Effects of Local Air Pollutants on Hedonic Happiness
- Regression approach: random effects ordinal logit reported (Table 9); pooled ordinal logit (Table 10) and random effects ordinal probit (Table 11) as robustness; models yield similar results.
- Main pollutant effects (from Table 9):
  - When NO2, PM10, and PM2.5 increase by 1 percent, the odds of higher (log) hedonic unhappiness increase by approximately 7.9, 9.9, and 9.9 percent respectively.
  - PM10 and PM2.5 have the bigger short-run impact on hedonic happiness.
- Covariate effects for short-term happiness:
  - Income: positive, non-linear effect on short-term happiness.
  - Men: higher odds of being happy in the short-term.
  - Education: having some education (vs no education) increases short-term happiness; effect strongest for GCSE > A-level > degree.
  - Employment status: employed, retired, or student (vs unemployed) increases likelihood of short-term happiness.
  - Marital status: being married (vs never married, divorced, retired, or widowed) increases short-term happiness.
  - Self-perceived health: believing oneself healthier increases odds of higher hedonic happiness.
- Note: higher values of hedonic happiness variable represent higher levels of short-term unhappiness (more distress).

### 6.3 Relative Magnitudes of Pollutants on Both Happiness Measures
- Standardized regressions (dependent variables and pollutants in level) reported in Table 12 (life satisfaction) and Table 13 (hedonic happiness).
- Standardized effects:
  - A one standard deviation increase in NO2 or PM2.5 decreases the odds of a higher standard deviation of life satisfaction by 1.1 percent.
  - For PM10, the odds are 0.9 percent.
  - For hedonic happiness, a one standard deviation increase in any pollutant decreases the odds of a higher standard deviation of hedonic happiness by roughly 1 percent.
- Conclusion: all three pollutants play almost the same role and have similar intensity effects on both happiness measures, although PM2.5 has slightly stronger effects.
- Explanation: similarity expected given high levels of pairwise correlations among pollutants.

### 6.4 Relative Magnitudes of Pollutants and Major Life Events on Happiness
- Purpose: standardized results allow comparison across variables measured in different units (pollution vs tangible life events).
- Comparisons on life satisfaction:
  - Unemployment (vs paid employment) decreases the odds of a 1 standard deviation increase in happiness by approximately 9 percent.
  - Divorce (vs marriage) decreases the odds by approximately 4 percent.
  - Therefore, in standard deviation terms, estimated disutility from unemployment is 9 times pollutant disutility; divorce disutility is almost 4 times pollutant disutility.
  - Conclusion: life events have stronger effects, but pollution still plays a significant role; life events between 4 and 9 times as important as pollution.
- Comparisons on hedonic happiness:
  - Unemployment (vs in-paid employment) is almost 14 times as important as pollution.
  - Divorce plays almost the same effect as air pollution.
  - Rationale: unemployment causes stronger short-term distress; divorce effects may attenuate over time due to coping strategies; interpersonal problems may matter more for hedonic unhappiness.

### 6.5 The Happiness Price of Air Pollution
- Method: estimate marginal rate of substitution (MRS) to derive willingness to pay (WTP); results in Table 14; focus on life satisfaction for valuations.
- WTP estimates (entire sample, evaluated at mean income):
  - Avoid one 휇g/m3 increase in PM10: almost £60 monthly.
  - Avoid one 휇g/m3 increase in PM2.5: £103 monthly.
  - Avoid one 휇g/m3 increase in NO2: £62.5 monthly.
- Subsample results:
  - Urban sample (column 2): urban citizens willing to pay around 66 percent of the overall sample amounts:
    - PM10: £40.5
    - PM2.5: £74
    - NO2: £40.5
  - Highly educated sample (column 3): average earners of the more educated subsample willing to pay:
    - PM10: £60.66
    - NO2: £83.19
    - PM2.5: £83
  - Note: for the highly educated sample, variance in the distribution of payments across pollutants is approximately 3.75 times smaller.
- Comparative literature:
  - Welsch (2003): maximum of US$900 per year to abate one 휇g/m3 of NO2 (ten European countries).
  - Ferreira and Modo (2010) for Ireland: US$870 annually to abate one 휇g/m3 of PM10.

### 6.6 The Endogeneity of Income
- Concern: reverse causality between income and happiness (happier people may earn higher incomes).
- Strategy: following Pischke (2011), sample restricted to married housewives and housewives with a partner who are unemployed, full-time students, long term sick, or on government training schemes—assumption that personal income is absent and household income exogenous to individual's happiness.
- Findings from restricted sample:
  - Relationship between income and happiness remains U-shaped and significant.
  - Coefficients comparable to original regressions (Tables 6–13).
  - Pollutant coefficients remain similar and significant.
  - Interpretation: positive and significant unidirectional causality from income to happiness present.
- Caveat: other explanatory variables may be endogenous (e.g., marriage and happiness); instrument selection for all such variables lacking in literature (Van Praag, 2007); not pursued further here.

### 7. Conclusion (Key Results and Policy Implications)
- Main empirical results:
  - 1 percent higher levels of NO2, PM10 and PM2.5 significantly decrease the odds of the log of happiness by 9, 9.5 and 10.7 percent respectively (controlling for demographics and socioeconomic conditions).
  - Effects on short-term hedonic happiness: 7.9, 9.9, and 9.9 percent respectively.
  - Standardized results: pollutants have roughly the same effect on life satisfaction and hedonic happiness.
  - Disutility comparisons: a 1 휇g/m3 increase in pollutants generates disutility approximately 4 and 9 times less important than divorce and unemployment, respectively (life satisfaction); for hedonic happiness, unemployment is 14 times more important while divorce is roughly equal to pollution.
  - Evaluated at mean income, households willing to pay monthly amounts of £62.5 (NO2), £60 (PM10), and £103 (PM2.5) per 1 휇g/m3 abated; urban dwellers pay less; highly educated individuals pay more for most pollutants.
- Policy implications:
  - Welfare increase from pollution control is substantial; supports efficiency grounds for environmental policy.
  - Equity and heterogeneity matter: policies should be education- and city-specific.
  - Transport sector targeted measures recommended, as automobiles and transportation are substantive sources of examined local air pollutants and also contribute to greenhouse gases.
  - Governments can implement environmental (green) tax reforms through indirect green consumption taxes informed by happiness/utility regression estimates.
  - Public perception of government efforts matters for happiness; media should enhance information delivery about government efforts to reduce air pollution.
  - Education increases environmental valuation and encourages pro-environmental behavior; public education campaigns recommended.
- Methodological note:
  - Happiness-regression approach provides valuations without directly asking respondents their WTP, avoiding stated-preference biases; however, potential endogeneity of happiness and other explanatory variables remains a limitation and an avenue for future research (instrumental variables, novel empirical strategies).
- Overall conclusion: air pollution significantly affects happiness and willingness to pay for cleaner air is non-negligible; results useful for policymakers designing efficient, effective, and equitable pro-environmental interventions.

*IMF Working Paper section: 6. Results and Discussion*

### Appendix I.

### Appendix I.

### Literature Summary
- Table 1: Summary of the Literature — papers, year, countries, pollutants, nature of data, model, result, aggregation level (selected entries from source):
  - Luechinger 2010 | Europe | SO2 | Pooled 2SLS | Negative | Country
  - Cuñado and Pérez de Gracia 2012 | Spain | NO2, CO2, PM10 | Cross Section OLS | Negative | Region
  - Jidong and Yiran 2015 | China | NO2 | Cross Section Ordered probit | Negative | City
  - Ferreira et al. 2013 | Europe | SO2 | Pooled OLS | Negative | (Up to) NUTS3
  - Liu et al. 2018 | China | NO2, PM10 | Pooled Ordered probit | Negative | Region
  - Ambrey et al. 2014 | Australia | PM10 | Pooled Ordered probit | Negative | Collection District
  - Orru et al. 2016 | Estonia | PM10 | Pooled OLS | Negative | 1x1 km2 grid
  - Welsch 2003 | Europe | NO2 | Panel FE GLS | Negative | Country
  - Zhang et al. 2017 | China | API | Panel FE | Negative | City
  - Knight and Howley 2017 | UK | NO2 | Panel FE | Negative | LSOA
  - Lin et al. 2019 | Taiwan | NO2, PM2.5, PM10 | Panel Multilevel LGM | No effect | City
  - Goetzke and Rave 2015 | Germany | Subjective Perception | Cross Section IV probit | Indeterminate | N/A

### Variable Definitions (Table 2)
- Dependent Variables:
  - Life Satisfaction: Subjective report of individual’s life satisfaction. Values: 1 (not satisfied) – 7 (very satisfied)
  - Hedonic Happiness: How distressed is the individual? Values: 0 (not distressed) – 36 (very distressed)
- Main Regressors:
  - NO2: Annual mean concentration of NO2 for each LSOA. In μg/m3
  - PM2.5: Annual mean concentration of PM2.5 for each LSOA. In μg/m3
  - PM10: Annual mean concentration of PM10 for each LSOA. In μg/m3
- Objective Individual Covariates (binary unless specified): Male; Age; Urban
- Labour Force Status (binary indicators; “unemployed” is the base category)
- Highest level of education (binary indicators; “No qualification” is the base category)
- Marital Status (binary indicators; “married” is the base category)
- Subjective Individual Covariates: Health Satisfaction (1–7)
- Objective Household Covariates: Household Size; Children; Equivalent Income (Gross household deflated monthly income divided by the equivalence OECD scale)
- Geographic Covariates: GOR; LSOA
- Time Covariates: Year

### Summary Statistics (Table 3 — selected)
- Dependent Variables:
  - Life Satisfaction: N = 341,833; Mean = 5.173; Std. Dev = 1.453; Min = 1; Max = 7
  - Hedonic Happiness: N = 350,769; Mean = 11.124; Std. Dev = 5.517; Min = 0; Max = 36
- Main Regressors (N = 241,000 each):
  - NO2: Mean = 15.202; Std. Dev = 6.692; Min = 2.409; Max = 66.036
  - PM2.5: Mean = 15.398; Std. Dev = 3.015; Min = 6.468; Max = 33.272
  - PM10: Mean = 10.304; Std. Dev = 2.102; Min = 4.319; Max = 24
- Selected covariates (N reported in table):
  - Male: N = 408,000; Mean = 0.461; Std. Dev = 0.498
  - Age: N = 410,000; Mean = 47.359; Std. Dev = 18.679; Min = 15; Max = 104
  - Urban: N = 351,000; Mean = 0.795; Std. Dev = 0.404
  - Equivalent Income: N = 395,162; Mean = 2086.6; Std. Dev = 1412.155; Min = 0; Max = 62288

### Description of Pollutants (Table 4)
- Nitrogen Dioxide (NO2)
  - Source: vehicle emissions, electricity, industrial operations
  - Maximum Acceptable Concentration: 0.053 ppm (1 year)
  - Health Risks: inflammation and irritation in respiration system
  - Maximum Annual Mean Ambient Level for Human Health Protection: 40 μg/m3
- Particulate Matter (PM2.5 and PM10)
  - Source: fires, caliche-topped roads; chemical reactions inside powerplants and vehicles
  - Maximum Acceptable Concentration: 150 μg/m3 (1 day for particles <10 μm); 35 μg/m3 (1 day for particles <2.5 μm)
  - Health Risks: irritation in respiratory system, asthma, irregular cardiac activity
  - Maximum Annual Mean Ambient Level for Human Health Protection:
    - PM2.5: 25 μg/m3
    - PM10: 40 μg/m3
- Source cited in table: United States Environmental Protection Agency. (2016). National Ambient Air Quality Standards Table.

### Pollutant Correlations (Table 5)
- Correlation matrix:
  - NO2 with NO2 = 1.000
  - NO2 with PM10 = 0.691
  - NO2 with PM2.5 = 0.733
  - PM10 with PM10 = 1.000
  - PM10 with PM2.5 = 0.940
  - PM2.5 with PM2.5 = 1.000

### Regression Results — Life Satisfaction (selected tables)
- Table 6: Pollutants on Life Satisfaction: Random Effects Ordinal Logit (N = 201129)
  - Log(Pollutant) coefficients:
    - NO2: -0.094*** (0.019)
    - PM10: -0.099*** (0.038)
    - PM2.5: -0.113*** (0.038)
  - Log(EquivalentIncome): -0.349*** (0.052)
  - Log2(EquivalentIncome): 0.040*** (0.004)
  - Gender: -0.110*** (0.017)
  - Age: -0.046*** (0.003); Age2: 0.001*** (0.000)
  - Health Satisfaction: 0.747*** (0.004)
  - Number of Obs. = 201129
  - Notes: ***, **, * indicate significance at the 1, 5, and 10% levels respectively. Year and regional fixed effects included.

- Table 7: Pollutants on Life Satisfaction: Pooled Ordinal Logit (N = 201129)
  - Log(Pollutant) coefficients:
    - NO2: -0.056*** (0.019)
    - PM10: -0.108*** (0.041)
    - PM2.5: -0.124*** (0.040)
  - Log(EquivalentIncome): -0.422*** (0.052)
  - Log2(EquivalentIncome): 0.047*** (0.004)
  - Number of Obs. = 201129
  - Notes: Year and regional fixed effects included.

- Table 8: Pollutants on Life Satisfaction: Random Effects Ordinal Probit (N = 201129)
  - Log(Pollutant) coefficients:
    - NO2: -0.049*** (0.011)
    - PM10: -0.052** (0.021)
    - PM2.5: -0.063*** (0.021)
  - Log(EquivalentIncome): -0.057*** (0.009)
  - Log2(EquivalentIncome): -0.024*** (0.001)
  - Number of Obs. = 201129
  - Notes: Year and regional fixed effects included.

- Table 12: Pollutants on Life Satisfaction: Pooled Ordered Logit (Standardized) (N = 199839 / 201129 depending on panel)
  - Pollutant (standardized) coefficients:
    - NO2: -0.011*** (0.001)
    - PM10: -0.009*** (0.002)
    - PM2.5: -0.011*** (0.003)
  - Log(EquivalentIncome): -0.120*** (0.044)
  - Log2(EquivalentIncome): 0.193*** (0.003)
  - Health Satisfaction: -0.192*** (0.002)
  - Number of Obs. = 201129 (depending column)

### Regression Results — Hedonic Happiness (selected tables)
- Table 9: Pollutants on Hedonic Happiness: Random Effects Ordinal Logit (N = 199839)
  - Log(Pollutant) coefficients:
    - NO2: 0.076*** (0.020)
    - PM10: 0.094** (0.037)
    - PM2.5: 0.094*** (0.036)
  - Log(EquivalentIncome): 0.06 (0.049)
  - Log2(EquivalentIncome): -0.012*** (0.004)
  - Gender: -0.558*** (0.019)
  - Age: 0.034*** (0.003); Age2: -0.000*** (0.000)
  - Health Satisfaction: -0.333*** (0.003)
  - Number of Obs. = 199839

- Table 10: Pollutants on Hedonic Happiness: Pooled Ordinal Logit (N = 199839)
  - Log(Pollutant) coefficients:
    - NO2: 0.054*** (0.019)
    - PM10: 0.136*** (0.041)
    - PM2.5: 0.131*** (0.041)
  - Log(EquivalentIncome): 0.115** (0.045)
  - Log2(EquivalentIncome): -0.017*** (0.003)
  - Health Satisfaction: -0.408*** (0.004)
  - Number of Obs. = 199839

- Table 11: Pollutants on Hedonic Happiness: Random Effects Ordinal Probit (N = 199839)
  - Log(Pollutant) coefficients:
    - NO2: 0.042*** (0.011)
    - PM10: 0.050** (0.021)
    - PM2.5: 0.053*** (0.020)
  - Log(EquivalentIncome): 0.033 (0.028)
  - Log2(EquivalentIncome): -0.007*** (0.002)
  - Gender: -0.298*** (0.010)
  - Age: 0.019*** (0.002); Age2: -0.000*** (0.000)
  - Health Satisfaction: -0.344*** (0.003)
  - Number of Obs. = 199839

- Table 13: Pollutants on Hedonic Happiness: Pooled Ordered Logit (Standardized) (N = 200412)
  - Pollutant (standardized) coefficients:
    - NO2: 0.010*** (0.001)
    - PM10: 0.011*** (0.002)
    - PM2.5: 0.011*** (0.002)
  - Log(EquivalentIncome): 0.035*** (0.040)
  - Log2(EquivalentIncome): -0.076*** (0.003)
  - Health Satisfaction: -0.344*** (0.003)
  - Number of Obs. = 200412

### Marginal Rate of Substitution (MRS) between Happiness and Pollutants: Heterogeneity Differences (Table 14)
- Full Sample / Urban Sample / High Education Sample (selected parameters and MRS)
  - PM10:
    - 훽1: 0.0002362 | 0.0001608 | 0.0003056
    - 훽4: -4.44E-06 | -4.25E-06 | -5.92.E-06
    - 훽5: 1.15E-10 | 7.12E-11 | 1.65E-10
    - Income bar (퐼푛푐표푚푒̅̅̅̅̅̅̅̅̅̅): 2079.3 | 2018.2 | 2674.3
    - MRS PM10 for Income: 59.62 | 40.58 | 60.66
  - PM2.5:
    - 훽1: 0.0004078 | 0.0002924 | 0.0004189
    - 훽4: -4.44E-06 | -4.25E-06 | -5.93E-06
    - 훽5: 1.15E-10 | 7.11E-11 | 1.65E-10
    - Income bar: 2079.3 | 2018.2 | 2674.3
    - MRS PM2.5 for Income: 102.93 | 73.78 | 82.99
  - NO2:
    - 훽1: 0.0002458 | 0.0001596 | 0.0004158
    - 훽4: -4.41E-06 | -4.23E-06 | -6.E-06
    - 훽5: 1.14E-10 | 7.06E-11 | 1.63E-10
    - Income bar: 2079.3 | 2018.2 | 2674.3
    - MRS NO2 for Income: 62.45 | 40.46 | 83.19
- Source note for Table 14: Based on authors’ own calculations by relying on data from BHPS and UKHLS years 2004 – 2019.

*Source: wpiea2023035-print-pdf - Appendix I.*

### References

### References

### Happiness, subjective well-being, and life satisfaction literature
- Alesina,  A.,  Di  Tella,  R.,  &  MacCulloch,  R.  (2004).  Inequality  and  happiness:  Are  Europeans  and  Americans different? Journal of Public Economics, 88, 2009–2042.  
- Berman, W. H., & Turk, D. C. (1981). Adaptation to divorce: Problems and coping strategies. Journal of Marriage and the Family, 179-189. 
- Blanchflower, D. G., & Oswald, A. J. (1998). What makes an Entrepreneur? Journal of Labor Economics, 16(1), 26–60. 
- Blanchflower,  D.G.,  &  Oswald,  A.J.  (2004a). Money, gender, and happiness: An empirical study. Scandinavian Journal of Economics, 106(3), 393–415. 
- Blanchflower,  D.G.,  &  Oswald,  A.J.  (2004b).  Well-being  over  time  in  Britain  and  the  USA. Journal  of  Public Economics, 88, 1359–1386. 
- Diener,  E.,  Suh,  E.M.,  Lucas,  R.E.  and  Smith,  H.L.  (1999)  Subjective  well-being:  three  decades  of  progress. Psychological Bulletin 125: 276–302.  
- DiTella, R., MacCulloch, R., Oswald, A. (2003). The macroeconomics of happiness. Review of Economics and Statistics 85, 809–827.  
- Dolan, P., Peasgood, T., & White, M. (2008). Do we really know what makes us happy? A review of the economic literature on the factors associated with subjective well-being. Journal of economic psychology, 29(1), 94-122. 
- Frey,  B.S.,  Luechinger,  S.  &  Stutzer,  A.  (2010)  The  Life  Satisfaction  Approach  to  Environmental  Valuation. Annual Review of Resource Economics, 2, 139–160.  
- Goetzke, F., & Rave, T. (2015). Regional air quality and happiness in Germany. International Regional Science Review, 38(4), 437-451. 
- Graham,  C.,  &  Pettinato,  S.  (2001).  Happiness,  markets  and  democracy:  Latin  America  in  comparative perspective. Journal of Happiness Studies, 2, 237–268. 
- Graham,  C.,  Eggers,  A.,  &  Sukhtankar,  S.  (2004).  Does  happiness  pay?  An  exploration  based  on  panel  data from Russia. Journal of Economic Behavior and Organization, 55(3), 319–342. 
- Helliwell, J. F. (2003). How’s life? Combining individual and national variables to explain subjective well-being. Economic Modelling, 20, 331–360. 
- Kahneman,  D.,  &  Deaton,  A.  (2010).  High  income  improves  evaluation  of  life  but  not  emotional  well-being. Proceedings of the national academy of sciences, 107(38), 16489-16493. 
- Khan et al. entries on life satisfaction and family structure: Flouri,  E.  (2004).  Subjective  well-being  in  midlife:  The  role  of  involvement  of  and  closeness  to  parents  in childhood. Journal of Happiness Studies, 5, 335.; Louis,  V.  V.,  &  Zhao,  S.  (2002).  Effects  of  family  structure,  family  SES,  and  adulthood  experiences  on  life satisfaction. Journal of Family Issues, 23(8), 986-1005. 
- Oswald, A. J., & Wu, S. (2010). Objective confirmation of subjective measures of human well-being: Evidence from the USA. Science, 327(5965), 576-579. 
- Pischke, J.S. (2011). Money and happiness: evidence from the industry wage structure (No. w17056). National Bureau of Economic Research. 
- Van Praag, B. M. (2007). Perspectives from the happiness literature and the role of new instruments for policy analysis. CESifo Economic Studies, 53(1), 42-68. 

### Air pollution, health impacts, and annoyance
- Ambrey, C. L., Fleming, C. M., & Chan, A. Y. C. (2014). Estimating the cost of air pollution in South East Queensland: An application of the life satisfaction non-market valuation approach. Ecological Economics,97, 172-181. 
- Babisch, W., Wolf, K., Petz, M., Heinrich, J., Cyrys, J., & Peters, A. (2014). Associations between traffic noise, particulate air pollution, hypertension, and isolated systolic hypertension in adults. Environmental Health Perspectives, 122(5), 492–498. 
- Brunekreef,  B.,  Künzli,  N.,  Pekkanen,  J.,  Annesi-Maesano,  I.,  Forsberg,  B.,  Sigsgaard,  T.,  Keuken,  M., Forastiere,  F.,  Barry,  M.,  Querol,  X.  &  Harrison,  R.M.  (2015)  Clean  air  in  Europe:  beyond  the  horizon? The European Respiratory Journal, 45, 7–10.  
- Harrison, R. (2018). Associations of long-term average concentrations of nitrogen dioxide with mortality. 
- Shah, A. S. V., Lee, K.K., McAllister, D.A., Hunter, A., Nair, H., Whiteley, W., Langrish, J.P., Newby, D.E. & Mills, N.L.  (2015)  Short  term  exposure  to  air  pollution  and  stroke:  systematic  review  and  meta-analysis. BMJ,  350, h1295–h1295. 
- Sørensen, M., Lüdorf, P., Ketzel, M., Andersen, Z. J., Overvad, K., & Raaschou-Nielsen, O. (2014). Combined effects of road traffic noise and ambient air pollution in relation to risk for stroke? Environmental Research, 133, 49–55.  
- Stenlund,  T.,  Lidén,  E.,  Andersson,  K.,  Garvill, J.,  &  Nordin, S.  (2009).  Annoyance  and  health  symptoms  and their influencing factors: A population-based air pollution intervention study. Public Health, 123, 339–345. 
- World Health Organization  (WHO).  (2013).  Review  of  evidence  on  health  aspects  of air  pollution:  REVIHAAP Project Technical Report. Copenhagen: World Health Organisation. 
- Lin, W.H., Pan, W.C., & Yi, C.C. (2019). “Happiness in the air?” the effects  of  air  pollution  on  adolescent happiness. BMC Public Health, 19(1), 795. 
- Zhang, X., Zhang, X., & Chen, X. (2017). Happiness in the air: How does a dirty sky affect mental health and subjective well-being?. Journal of Environmental Economics and Management, 85, 81-94
- Liu, N., Liu, R., Huang, J., & Chen, L. (2018). Pollution, Happiness and Willingness to Pay Taxes: The Value Effect of Public Environmental Policies. Problemy Ekorozwoju, 13(1). 
- Jidong, Y., & Yiran, Z. (2015). Happiness and Air Pollution. China Economist, 10(5), 62. 
- Knight, S. J., & Howley, P. (2017). Can clean air make you happy? Examining the effect of nitrogen dioxide (NO2) on life satisfaction. Health, Econometrics and Data Group (HEDG) Working Papers, 17(08). 
- Orru, K., Orru, H., Maasikmets, M., Hendrikson, R., & Ainsaar, M. (2016). Well-being and environmental quality: Does pollution affect life satisfaction?. Quality of Life Research, 25(3), 699-705. 
- Ferreira, S., Akay, A., Brereton, F., Cuñado, J., Martinsson, P., Moro, M., & Ningal, T. F. (2013). Life satisfaction and air quality in Europe. Ecological Economics, 88, 1-10. 
- Welsch, H. (2003). Environment and happiness: Valuation of air pollution in ten European countries (No. 356). DIW Discussion Papers. 
- Ambiguous/related empirical studies: Cuñado, J., & De Gracia, F. P. (2012). Environment and happiness: New evidence for Spain. Social Indicators Research, 112(3), 549-567.; Goetzke & Rave (2015) entry above.

### Valuation methods, environmental economics, and hedonic approaches
- Brown, T. C., & Gregory, R. (1999). Why the WTA–WTP disparity matters. Ecological Economics, 28(3), 323-335. 
- Callan,  S.  J.,  &  Thomas,  J.  M.  (2013). Environmental  Economics  and  Management:  Theory,  Policy, and Applications. Cengage learning. 
- Cunningham entries on welfare/valuation theory: Hicks, J. R. (1941). The rehabilitation of consumers' surplus. The Review of Economic Studies, 8(2), 108-116.; Freeman,  A.M.,  III.  (1993),  The  Measurement  of  Environmental  and  Resource  Values,  Washington,  DC: Resources for the Future. 
- Egan, K.J., Corrigan, J.R. & Dwyer, D.F. (2015) Three reasons to use annual payments in contingent valuation surveys: Convergent validity, discount rates, and mental accounting. Journal of Environmental Economics and Management, 72, 123–136.  
- Kuminoff, N.V., Parmeter, C.F. & Pope, J.C. (2010) Which hedonic models can we trust to recover the marginal willingness to pay for environmental amenities? Journal of Environmental Economics and Management, 60, 145–160.  
- Luechinger, S. (2010). Life satisfaction and transboundary air pollution. Economics Letters, 107(1), 4-6. 
- Marrouch, W., & Sinclair-Desgagné, B. (2012). Emission taxes when pollution depends on location. Environment and Development Economics, 17(4), 433-443. 
- Kuminoff et al. and Hicks (above) on theoretical foundations. 

### Methods, data sources, and institutional/legal references
- Baetschmann, G., Staub, K.E. & Winkelmann, R. (2011) Consistent Estimation of the Fixed Effects Ordered Logit Model. IZA Discussion Paper No. 5443 
- Chin/Survey design and data: Lynn, P. (2009). Sample design for understanding society. Underst. Soc. Work. Pap. Ser, 2009. 
- Nicholson,  W.,  &  Snyder,  C.  M.  (2012). Microeconomic  theory:  Basic  principles  and  extensions.  Cengage Learning. 
- ONS. (2017). Population Weighted Centroids Guidance. 
- ONS. (2020). Labour Market Overview.  
- DEFRA. (2015). Valuing the Impacts of Air Quality on Productivity. 
- Department for Environment, Food, and Rural Affairs (DEFRA). (2021). September 2021: Air Quality Factsheet (Part4). 
- Environmental Protection Agency. (2016).  National Ambient Air Quality Standards Table. 
- United States Environmental Protection Agency. (2016).  National Ambient Air Quality Standards Table. 
- Environmental Protection UK. (2021). Air pollution laws. Environmental Protection UK. https://www.environmental-protection.org.uk/policy-areas/air-quality/air-pollution-law-and-policy/air-pollution-laws/. 
- European Environmental Agency. (2009). La calidad del aire en Europa, http://www.eea.europa.eu/es/.  
- The Environment Bill. (2021). Parliament: House of Commons. Bill no. 220. 
- UNICEF UK. (2018). One in three UK children live in toxic air zones. 
- WHO REVIHAAP (2013) entry above. 

### Cross-cutting empirical and regional studies
- Dibeh, G., Fakih, A., Marrouch, W., & Matar, G. (2021). Who Cares About Environmental Quality in the MENA Region?. Social Indicators Research, 1-27. 
- Wang,  B.  Z.,  &  Cheng,  Z.  (2017).  Environmental  perceptions,  happiness  and  pro-environmental  actions  in China. Social Indicators Research, 132(1), 357-375.
- Klein studies referenced above linking noise, particulate pollution, hypertension and stroke: Babisch et al. (2014); Shah et al. (2015); Sørensen et al. (2014). 

*Source: wpiea2023035-print-pdf - References*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023035-print-pdf.pdf_