ordinality and cardinality in economic utility theory
TRANSCRIPT
Applications of
to Economic Utility Theory
Benjamin DanielsMath 101Pomona College
Ordinality and Cardinality
How Happy Are You?
Okay, this is a tough
question to start with.
Could you be happier? What would make you happier? How much happier would it
make you?
Ordering Preferences
Basic Rules:
Ordered
Transitive
Positive
Convex
Guns
Butter
A
B
C
Indifference Curves
Equivalence classes in
utility space.
Making Decisions
Constrained
Optimization:
Some Lagrangian stuff you’ll be thrilled
to be reminded
of.
How Happy Are “We”?
Would something that makes any of us happier
make all of us
happier?
A Social Utility Function
Familiar Rules:
Ordered
Transitive
Positive
Convex
Bread
Circuses
A
B
C
Distributional Equity
Familiar Rules:
Ordered
Transitive
Positive
Convex
Bread
Circuses
A
B
C
Me
You
Cardinalizing Utility Space
For each person, we
can arbitrarily
assign useful
cardinal units,
based on their own subjective perception
s.1
2
3
4
Efficient Social Decisions
Me You
Are These Decisions Possible?Are utilities comparable between
people, even if we
assume that utility space is
cardinal for each
person?
If so, is that fair?
Measuring Happiness
To make the best
decisions, we need to measure
how happy people are.
Time Period Surveys Sum of Momentary Happiness Intensity Weighting
Estimating Social Utility
Most practical theories
avoid estimating individual
utility directly.
Econometric Analyses Assigning dollar values to
outcomes. Proportional Tradeoffs
“Doubling my happiness is equivalent to doubling yours.”
Transformations from Ordinal Values
First-Order Approaches
Remember, maximization problems are solved using first-
order conditions.
If we can cardinalize
the derivative,
we can avoid the problems of measuring
value directly.
In Practice: Cost-Benefit Analysis
Cost-benefit analysis tries to
estimate cardinal
values by observing
what people are willing to pay for various things.
Difference-in-difference analysis
“Marginal willingness to pay”
CBA as Cardinal Values
Can we estimate the value
of a human life?
In Practice: The QALY System
“How many
years of life in state
X are equivalent to one year of healthy
life?”
Participants answer this question, and their answers are normalized such that one year of perfect health has value 1.
The answers are averaged to create a social valuation of health states.
Each procedure is empirically observed to improve health by some value and extend longevity by some duration.
QALYs as Cardinal Values
The QALY conversion gives each procedure a cardinal value so the most efficient choices can be made.
QALYs as Derivatives
Once we have the marginal value of a procedure,
cost-efficiency is trivial.
Value = Quality ∙ Longevity
The marginal value of a procedure is given by its
expected impact on value:
∂Value = (Quality ∙ ∂Longevity) + (∂Quality ∙ Longevity)
In Practice: The U-Index
The U-Index is
specifically designed
to output a cardinal measure
from ordinal inputs.
At each time period, participants rank various emotions from 0 to 5.
Whenever any negative emotion scores higher than all positive emotions, a person is considered “unhappy.”
The U-Index is then reported as the proportion of “unhappy” outcomes.
The U-Index as Cardinal Values
The main benefit of
U-Indexation is the ease
of transforming answers
to useful values.
But it is very
difficult to account for differences in intensity.
The Space of Social Utility Functions
Rawls Benthamite
Formalizing Fairness
The theories of Bentham and Rawls are upper and lower bounds for this notion of fairness. Rawls: α -∞
Benthamite: α 1
In Conclusion…
Social choice is
really hard!
Kenneth Arrow proved that efficient social choice is impossible.
But policymakers need to make decisions somehow.
Even given very weak assumptions about fairness and human happiness, we can construct useful cardinal estimations of benefits.