Jeff Lazo Societal Impacts Program WAS*IS Workshop Boulder, CO July 2006

1. Some Economics to Make a Wiser WAS*ISer Jeff Lazo Societal Impacts Program WAS*IS Workshop Boulder, CO July 2006

2. Economics ec·o·nom·ics (ĕk'ə-nŏm'ĭks, ē'kə-) n. • The social science that deals with the production, distribution, and consumption of goods and services and with the theory and management of economies or economic systems. http://www.answers.com/topic/economics • Lionel Robbins 1932: "the science which studies human behavior as a relation between scarce means having alternative uses." http://en.wikipedia.org/wiki/Economics • study of the allocation of scarce resources in light of unlimited wants

3. Economics Objectives • exposed to basic concepts of economics • exposed to basic methods of economics • discuss some applications Reality check • can’t teach economics in one hour • what economics do you need to know? • some people think economists think differently than other people – that is NOT true • other people think differently than economists

4. Why Value Forecasts? • program justification • benefit-cost analysis • program evaluation • guidance for research investment • any cases of true comparative analysis? • inform users of forecast benefits • developing end-to-end-to-end forecast and warning system

5. What Should Be Valued? • Weather impacts Dutton - $3T US • Forecasts • Improved forecasts • Research to improve forecasts • How forecasts are used

6. What Should Be Valued? Forecasts  Value Integrate forecasting and valuation - Meteorology - Economics

7. What Should Be Valued? Weather  Observation  Forecast  Communication  Perception  Use  Value

8. Some Basic Economics: Value Theory and Topics in ValuationEcon 101

9. Econ 101 TOPICS • value theory • consumers and producers • supply and demand • markets and prices • consumer surplus • producer surplus • net societal welfare • market failures

10. What is Value? Nelson and Winter QJE • Why ask “What is Value?” • Ensure that “economic value” is valid economics • Look at broader approach to economic valuation

11. What is Value?Market Failures • Public goods • Market power • Externalities • Information

12. What is Value? Topics: Public Goods • What is the price (i.e., value) of weather forecasts? • Weather forecast characteristics • Non-rival • Non-exclusive • Problems of public goods • No observable price information • No provision by private markets • Weather forecasts as “quasi-public goods”?

13. What is Value? Topics: Time Discounting 5% Rate of Time Preference Net Benefit 20.00 Net Benefit -16.15

14. What is Value?Topics: VSL • Value of Statistical Life (VSL) • 1,000,000 people each willing to pay $50 a year for a program to reduce the chance of death by 1 in 100,000 per year (say from 20 in 100,000 to 19 in 100,000 each year) • Means that the group is WTP $50,000,000 to prevent 10 deaths • VSL = $50,000,000/10 deaths = $5,000,000

15. What is Value?Topics: Benefits Transfer Application of results from one study for a different analysis context. Same commodity being valued? same baseline? same outcome? Adjusting for: date of study – changes in prices (inflation) changes in preference income differences availability of substitutes and complements other significant determinants of value

16. Evaluation of the Sensitivity of U.S. Economic Sectors to Weather (OUSSSA)Jeffrey K. Lazo – NCARPete Larsen – NCAR / Cornell UniversityMegan Harrod – Stratus ConsultingDonald Waldman – University of Colorado Purpose: Assess sensitivity of US economic sectors to weather variability

17. Outline • Motivation • Concept • What is Economic Sensitivity? • Data and Modeling • Results • Conclusions

18. Dutton – BAMS – September 2002 “. . . one-third of the private industry activities, representing annual revenues of some $3 trillion, have some degree of weather and climate risk. This represents a large market for atmospheric information . . . “

19. Conceptual Approach Model Building: using historical economic and weather data, we model the relationship between economic output in 11 sectors, economic inputs, and weather and weather variability ? Gross State Product Capital Labor Energy Temperature Precipitation

20. Conceptual Approach Gross State Product Capital Labor Energy Temperature  Precipitation  Sensitivity Analysis: Using these models, we then hold the economic inputs constant, and use 70 years of weather data to see how economic output varies as a result of variation in weather

21. Define “Sensitive” • No single correct definition • Characteristics of a meaningful approach • consistent with economic theory • amenable to empirical examination • provide meaningful information about economic impacts of Wx

22. What is Weather Sensitivity? S(K0, L0, E0;W0) P$ S(K0, L0, E0; W1) P1 P* D(W1) Change in GSP GSP D(W0) Q Q* Q1

23. Issues? • Weather or climate? • Sensitivity or something else?

24. Super Sectors

25. Economic Modeling Translog Function

26. Weather “Sensitivity”

27. Economic Data Economic Data - state x year x sector Gross State Product (dependent variable) Production Inputs • Capital (K) - dollars • Labor (L) - hours • Energy (E) – BTUs Weather Data - state x year Temperature Variability • CDD : Cooling Degree Days: (T - 65) on a given day • HDD : Heating Degree Days: (65 - T) on a given day Precipitation • P_Tot: Precipitation Total (per square mile) • P_Std: Precipitation Standard Deviation i = state 48 j = sector 11 t = year 1977-2000 = 24 years 48 x 11 x 24 = 12,672 “observations”

28. Temperature Weather Inputs • CDD: Defined as (T - 65) = daily CDD, where T is daily Average Temperature (F). If T is less than 65 degrees F, CDD=0. • HDD: Defined as (65 - T) = daily HDD, where T is daily Average Temperature (F). If T is greater than 65 degrees F, HDD=0. • Average (Mean) Temperature of the day : (High Temperature + Low Temperature) / 2 ; High and Low Temperature are whole integer values. http://www.weather2000.com/dd_glossary.html

29. Econometric Methods • Level data versus per capita • Panel data – time series – AR(1) • Heteroskedasticity • Fixed Effects • Covariance calculations for marginal effects

30. Econometric Results Sector: Agriculture ns = not significant at 10% * 10%, ** 5%, *** 1%

31. Parameter Estimates from Full Model Regressions

32. Marginal Responses

33. Marginal Responses

34. Wx Sensitivity Analysis 11 Sector Models: Q = f (K, L, E, W; Year, State) • Average K, L, E 1996-2000 • Set Year to 2000 • Historical weather data 1931-2000 • Fitted GSP values by sector by state by year • 11 sectors • 48 states • 70 “years” fit to 2000 “economic structure”

35. State Sensitivity(Billions $2000)

36. State Sensitivity(Billions $2000)

37. Sector Sensitivity(Billions $2000)

38. Sector Sensitivity(Billions $2000)

39. Sector Sensitivity(Billions $2000)

40. Sector Sensitivity(Billions $2000)

41. National Sensitivity(Billions $2000)

42. Future Research (1) • extend data past 2000 • better capital and energy data • include “storms” data • include forecast skill measure • value of weather forecasts? • split supply and demand • model uncertainty

43. Future Research (2) • finer spatial scales • county level data for a state • finer temporal scales • quarterly / monthly economic data • finer sectoral scales • 2, 3, or 4 digit sector study • other regions / countries

44. Conclusions • Economically valid analysis • Significant impact of weather • significant regression coefficients • significant marginal effects • Interpretation of weather sensitivity • upper-bound weather risk measure? • upper-bound measure of value of weather information? • 3.4% of annual US economic variability • $260B US economic variability related to weather variability

45. Benefits of Investing in Weather Forecasting ResearchJeff Lazo, Jennie Rice, Marca HagenstadSuperComp Purpose: Assess benefits of buying a new supercomputer for weather forecast research

46. SuperComp • TOPICS • Value of investments in research • Assess value chain • Benefit-cost analysis • Benefits transfer • Value of statistical life • Discounting • Sensitivity analysis

47. SuperComp Study Methods • Determine potential impact of supercomputer on forecast quality • Identify potential sectors/users and of improved forecast • Identify existing benefit studies for sectors/users • Quantify probabilities and timing of impacts • Develop benefits model for aggregating over time • Conduct sensitivity analysis

48. Example: “SuperComp” Marine Resource Mgt. Benefits Agriculture Benefits Private Sector Benefits (e.g., highways) Marine Transportation Benefits International Benefits Improved Environmental Modeling New Supercomputer Total Benefits Household Benefits Improved Operational Forecasts (NWS Benefits) Retail Benefits DOE Benefits (wind) Aviation Benefits Air Force Benefits Energy Benefits (temps, wind) Army Benefits

49. Example: “SuperComp” Household Benefits of Short Term Weather Forecasts Stratus Consulting (2002) – stated preference study

50. Example: “SuperComp”