Expected Utility - normative model of decision making

 

utility - usefulness in achieving a goal

 

units of utility don=t have to be dollars, but often are--

- from economic models

- common demoninator

 

Determining utility with uncertainty:

1) Expected utility = Average expected utility

Utility of beach trip = Average of past 10 trips

fuu's = fun utility units

(8 * 25) fuu=s + (2 * -3) fuu=s = 194 fuu=s

194/10 = 19.4 fuu=s = expected utility

 

(8 * 25) fuu=s + (1 * 4) fuu's + (1 * -3) fuu's = 201 fuu=s

201/10 = 20.1 fuu=s = expected utility

Compare with trip to mall -- always same (10 fuu=s)

 

2) Expected utility = probability * utility

20% chance of rain, 80% chance of sun:

(.20 * -3 fuu=s) + (.80 * 25 fuu=s) = 19.4 fuu=s

 

Picking highest expected utility option brings us greatest utility in the long run--

thus, normative or Abest@ choice

 

Marginal utility -- utility associated with each additional unit

Marginal utility tends to taper off

(each subsequent unit increases total utility less)

Money keeps its marginal utility better than most commodities

 

Prospect Theory (Kahneman & Tversky, 1979) --

Redraws expected utility curve around neutral reference point

Loss function is steeper than gain function

e.g.: Losing beach trip moves you further from 0 than gaining beach trip moves you in opposite direction from 0

(compare to standard Aaccounting@)

 

Reference point can change

Suppose you weight the following three vacation spot criteria as follows:

outdoor beauty     nightlife     chances for meeting true love
.30                .30           .40
Now, you have to decide between Bend and L.A. for a vacation spot. Their scores on these criteria are:
         Beauty     Nightlife      Love

Bend:     25         10             20

(.30 * 25) + (.30 * 10) + (.40 * 20) = 18.5

L.A.: 2 25 20 (.30 * 2) + (.30 * 25) + (.40 *20) = 16.1

High score on one can compensate for low score on other--e.g., Bend=s a lot lower on nightlife, but still wins because it=s so high on outdoor beauty.


Too much work to look at all the dimensions?

Lexicographic decision: Pick option highest on most important dimension

E.g.: Nightlife - pick L.A.

(despite fact it isn=t as good overall)

 

Satisfice: Agood enough vacation@

e.g., any city over 15, like Ashland -

(.30* 30) + (.30 * 5) + (.40 * 10) = 15.4