Experiments on Subjective Preference: Toward a New Methodology and Applications of Subjective Income Jeffrey E. Frank M.S. Candidate, Community Development and Applied Economics Certificate of Ecological Economics Advisor: Jane Kolodinsky, Ph.D. March 24, 2011
Outline Chapter 1: Continuum Happiness Surveys
Introduction: Why Happiness?
Burlington Happiness Survey
Surveying for Elasticity
Loss Aversion vs. Gain Adoration
Chapter 2: On Subjectivity: Burlington, VT and the Latin American Context Chapter 3:
Introduction: Middle Class
Methods and Data
Do Middle Class Attitudes Differ?
“BTV Ranked Happiest Small-City In The Country: Gallup Finds Burlington Among The Happiest, Healthiest” - March 17, 2011 (Terminology: Happiness = Well-Being, Subjective Well-Being)
Because the traditional assumption that people are income-based profit-seeking agents in the economy has proven itself incomplete.
Because profit and income are not innate human directives, but the pursuit of our happiness is.
(Note: Happiness data can be used both to inform the ways respondents accumulate/qualify their happiness, but also can inform how to best relieve misery.)
Diminishing Marginal Returns on Happiness
Environmental Recovery is Good for GDP
Prisons are good for GDP
Even a burning building is good for GDP
Happiness Going Becoming Mainstream “What we measure affects what we do.”
Self-Assessment Surveys with Ordinal Categories
“Taken all together, how would you say things are these days – would you say that you are very happy, pretty happy, or not too happy?”
General Social Survey
Burlington Happiness Survey Objectives: To test a 101-point interval response category for measuring well-being. To calculate elasticity values for variables associated with well-being. To calculate Gini Coefficient and Lorenz Curve for Happiness
Survey Construction 1) Objectives of the Survey 2) Population to be Sampled 3) Degree of Precision Desired 4) Data to be Collected 5) Methods of Measurement 6) Selection of the Sample 7) Organization of the Field Work 8) Summary and Analysis of Data
1) Objectives of Survey: To test a 101-point interval response category for measuring well-being. To provide data able to calculate elasticity values for variables associated with well-being. To provide data able to calculate Gini Coefficient and Lorenz Curve for well-being.
Static level of happiness for well-being variables
Hypothetical low happiness level for well-being variables
Hypothetical high happiness level for well-being variables
Static level of well-being variables
Survey Design Guiding Concept: By collecting happiness data that behaves more like continuous income data we can begin to conduct similar analyses and utilize happiness data in a variety of ways. Allows happiness data able to complement income data in new ways.
5) Methods of Measurement
5) Methods of Measurement
6) Selection of Sample Skip Interval = N/n = 1919/252 = 7.6 = 8 Skip interval of every 8th house on right
7) Organization of the Field Work - Pretested 7 versions among classes and volunteers - First surveyed alone (roughly 25). - Reviewed results - Then recruited RM 250 Lab students. - Used 2 class periods for training and survey methodology familiarization.
8) Summary and Analysis of Data
Coding of all survey questions both with 101-point gradient
Continuous to Interval scale
What Hypothetical Data Does: - Utilizes the brain’s ability to simulate experience in order to capture the directional forces of one’s pursuit of happiness. Prefrontal cortex simulations
Elasticity How to do this? - By measuring the estimated impact of increases or decreases in a well-being variable – family time, education, income, etc. Def: Elasticity - The measure of the percentage change in one variable brought about by a percent change in another.
Calculating Arc Elasticity
Calculating Arc Elasticity =
Experiments: What can this data do? Loss Aversion = The tendency for people to strongly prefer avoiding losses over acquiring gains. And its opposite: “Gain Adoration” = The measure of one’s preference for acquiring a gain.
Must account for static levels
For example, a low gain adoration score could mean the respondent already has a high level of the variable, or that the variable is less desired, or both.
Static Level = Respondent’s definition of Happiness w/ well-being variable is based on internal high and low anchors.
Measuring elasticities above and below static rather than the distance from static allows for directional forces to be compared across well-being variables.
LA and GA are not necessarily inversely related because each are measured against the preference for much more/less of the variable, not a finite sum of H.
Loss Aversion & Gain Adoration
Experiments: What can this data do? Gini Coefficients for Happiness: - A measures of statistical dispersion, typically used to measure the inequality of income among a population. Generates a value between 0-1 where 0=Max equality and 1=Max inequality. - Noted by UK Group as a frontier for Happiness research. - Illustrated using a Lorenz Curve:
Burlington Tract 9 Gini = .116
Conclusions and Future Research
It is possible:
To collect happiness data that can behave as though it is a continuous data, such as income.
To calculate elasticity values for variables contributing to well-being.
To calculate Gini coefficients for happiness.
Without repetitive studies, we can only consider these results tentative.
Burlington H. Survey offers experimental methods of capturing subjective assessments of well-being.
Subjectivity has flaws
We lie – intentionally or unintentionally.
We gravitate toward expected norms
- Policy implications of subjective well-being research is still undetermined and highly contentious. - To get closer to applications we study subjective income.
Revealed vs. Subjective Preferences
Revealed Preferences: Those displayed by purchasing decisions.
Subjective Preferences: Those displayed directly through surveys.
Much used RFK quote, “GDP tells us everything except that which makes life worthwhile.”
The wealthy get to have more preference.
Subjectivity can offer
Include more about complementary nature
In order to make the best complements between subjective and objective income, we can measure income subjectively and assess those issues untouched by objective measures.
An Exploration of Subjective Income in theLatin American Context Jeff Frank MS CDAE Candidate, University of Vermont
Carol Graham and Julie Markowitz Brookings Institution
Jane Kolodinsky, Jon Erickson and Qingbin Wang University of Vermont
What is the Middle Class? - Highly difficult to identify so many income cut-offs are arbitrary.
Methods Subjective definition of Income. Economic Ladder Scale: Imagine a staircase with 10 steps, in which on the first step are located the poorest and on the 10th step , the richest. Where would you put yourself on this staircase? Source: Latinobarometro Survey
Source: All figures are in percentages. Source for income groups is Cardenas and Henao (2011); source for other figures is Latinobaremetro data in Frank, Markowitz, and Graham (2011)
Data Latinobarometro Survey - Annual survey of 19,000 respondents from 19 Latin American countries representing over 450 Million inhabitants. - Dataset analyzed is comprised of data from 1997-2010, N= - Dataset coded and compiled by (Frank, Markowitz and Graham, 2011)
Strength of Correlation against Economic Ladder Scale
High Middle Low
Do Middle Class Attitudes Differ?
Multinomial logistic model:
Virtues – Allows for a step-by-step analysis of the Economic Ladder Scale by predicting different possible outcomes of a categorically distributed dependent variable, given a set of independent variables which are real-valued, binary-valued, or categorical-valued.
Separate regressions for each step of the ladder. Offers the ability to view the variation in how independent variables contribute to log probability of a respondent chosen a given step of the ladder based on their responses to attitudinal variables.
Some Findings: - Education increases slightly as subjective income increases. - Those at the higher end of the subjective income scale have a lower level of support for democracy. - Fear of unemployment has a strong influence with feeling poor and relieving this fear of unemployment has a strong influence on feeling wealthier. - People on the lower end of subjective income have lower expectations for increasing on that scale in the future and visa versa with those on the higher end.