1
Looks Like RainThe Value of Improved Weather
Forecasts

• David J. Chapman and Megan Harrod
• Stratus Consulting Inc., Boulder. CO
• Jeff Lazo, Barbara Brown, Rebecca Morss
• National Center for Atmospheric Research
• January 19, 2006

2
Outline

• Motivation of the Study
• Approach to Valuation
• Pre-test Survey
• Full Survey Development
• Next Steps

3
Motivation

• Understand the publics preferences for weather
  forecast attributes
• Estimate benefits to households for improvements
  in day-to-day weather forecasting products
• Provide evidence of value of forecast products
  and improvements to those products

4
Use of Results

• Help inform policy makers
• Provide information for program review
• Inform prioritization of research agendas

5
Focus of Study

• Focus on general population users (108 M
  households)
• Not focused on specialized users at this time
  (e.g., farmers, shippers)
• Day to day weather forecasts as opposed to
  severe weather
• Limited number of forecast attributes
• temperature, precipitation

6
Value of Weather Information

• Weather forecasts are a public good
• Cannot exclude some from using
• Cost to provide does not rise with the number of
  users
• General forecast are not exchanged in a market
• Estimation of the value of a public good
  necessitates use of non-market valuation methods
• Stated preference methods
• Individuals state their preference rather than us
  observing some specific behavior
• contingent valuation
• choice based methods

7
(No Transcript)
8
Study Development

• Atmospheric Science Advisors (ASA)
• attributes of weather forecasts
• current and potential level of attributes
• Pre-test survey development and testing
• Focus groups, one-on-one interviews, peer review,
  Pre-test
• Survey Instrument Revision
• Full Survey Implementation
• National representative sample of 1,400 subjects

9
Overview of Survey Components

• Introduction
• Perceptions, sources and uses of forecasts
• Description of forecast attributes
• Valuation of improved weather forecasts
• Stated choice - attributes of forecasts
• Socio-demographic information
• Value for Current Forecasts
• Communication of Uncertainty

10
Pre-Test PerceptionsImportance of Weather
Forecast Characteristic
11
Pre-Test Stated ChoiceQuestion
12
Pre-Test Stated Choice -Attributes and
Attribute Levels

• Dollars per year per household of 3, 8, 15,
  24
• Budget constraint reminder
• 20 versions of survey
• 9 Stated Choice and 1 Stated Value question

13
Pre-Test Results Perceptions - Importance of
Weather Forecast Characteristics (Q2)
14
Pre-Test Results Perceptions - Adequacy of
Current Levels of Forecast Attributes
15
Pre-Test Results Perceptions - Adequacy of
Current Levels of Forecast Attributes
16
Pre-Test Results

• Provided good information on peoples
  perceptions, sources and uses
• Provided evidence that there is significant value
  for current forecasts 110 / HH
• Provided evidence that values exist for improved
  forecasts 16 / HH (13.91 - 19.54)
• Indicated that description of alternative
  forecast attributes is necessary
• Cannot be generalized to general public

17
Main Survey Revisions

• Re-defining forecast characteristics
• Temperature 0-1 days 2-6 days 7-14 days
• Precipitation 0-1 days 2-6 days 7-14
  days
• Geographic and temporal specificity
• Probabilistic forecast information
• Sampling
• National representative sample ( 1400
  completes)
• Internet based implementation
• Modeling and analysis
• Incorporation of sample heterogeneity
• Regional estimates (?)
• Module on presentation and understanding of
  forecast uncertainty

18
Final Survey PerceptionQuestionMay help
develop an individuals attribute set
19
Communicating Uncertainty in Forecasts

• What characteristics of uncertainty does the
  public most care about ?
• Variability around a point estimate (e.g., max
  temp)
• Timing of events ( rain this afternoon v. rain at
  3 pm)
• What characteristics of uncertainty can best be
  understood
• Absolute, relative changes
• What are the most effective ways to communicate
  uncertainty?
• Graphical / Visual / Textual
• Examples
• Local Daily Forecast
• What does it mean, what is the variability?
• National Isotherms

20
Evidence from Economics Literature

• Individuals can understand trade-off in relative
  risk, changes in probabilities (health, value of
  statistical life)
• Changes in probability of saving endangered
  species
• Expected outcomes of lottery games
• Individuals can understand and incorporate
  uncertainty into decision-making
• Experimental economics

21
(No Transcript)
22
(No Transcript)
23
Presentation of Uncertainty in Isotherms
Nagy and Oleary http//www.mathcs.emory.edu/nag
y/Twinkle/
24
Presentation of Uncertainty in Isotherms
Nagy and Oleary http//www.mathcs.emory.edu/nag
y/Twinkle/
25
Next Steps

• Pre-test indicated values exist but need to
  refine instrument
• We have completed expert focus groups to develop
  attribute set to revise instrument
• Revising survey instrument
• Revisions/testing/revision/testing
• Experimental Design
• Regional variability
• Attribute levels (number, range)
• Developing uncertainty module
• Programming instrument for Internet
  administration
• Administration, analysis and reporting