1. Preventing Land Loss in Coastal Louisiana: Estimates of WTP and WTA Dan Petrolia & Ross Moore Mississippi State University Tae-goun Kim Korea Maritime University 2010 CNREP Conference, New Orleans, May 26-28

2. Abstract  A dichotomous-choice contingent-valuation survey was conducted on a sample of Louisiana households to estimate compensating surplus (CS) and equivalent surplus (ES) welfare measures for the prevention of future coastal wetland losses in Louisiana  Valuations were elicited using both WTP (tax) and WTA compensation (tax refund) payment vehicles  PV of welfare estimates were very sensitive to discount rates, but were estimated in the neighborhood of $9,000 per LA household for CS (WTP) and $21,000 for ES (WTA)  The results of a probit model using a Box-Cox specification on income indicate that the major factors influencing support for land-loss prevention were: perceived hurricane protection benefits (positive) environmental and recreation protection (positive) distrust of government (negative) income age (positive) race (positive for whites)

3. Motivation  Louisiana’s 3 million acres of wetlands represent about 40% of coastal wetlands in U.S. but account for about 80% of losses (USGS 1995).  Coastal LA lost 1,900 miles 2 between 1932 and 2000 (USGS 2003)  Katrina and Rita eliminated an additional 217 miles 2 in 2005 (Barras 2006)  Additional 700 miles 2 expected to be lost over next 50 years (USGS 2003)  Several reports have been published over the last decade to heighten awareness of these losses and to identify solutions  Ongoing restoration projects: CWPPRA & state As of 2006, an estimated 32,345 acres of coastal land had been re-established under CWPPRA, at cost of $624.5 million  Most projects are small and independent; do not appear to be part of a comprehensive coast-wide restoration strategy called for in reports. Cost estimates for a comprehensive strategy:  from as low as $1.9 billion over 10 years for a scaled-down version of the Coast 2050 plan (National Research Council 2006),  to $14 billion over 30 years for the full-blown Coast 2050 plan (LCWCR Task Force 1998),  to as much as $100 billion (Winkler-Schmit 2009).

4. Prior work on Wetland Valuation  Brander, Florax, and Vermaat (ERE 2006)  Woodward and Wui (Ecol. Econ. 2001)  Kazmierczak (LSU Staff Paper 2001)  Research specific to Louisiana is now 14-23 years old Farber (Cont. Econ. Pol. 1996) Bergstrom et al. (Ecol. Econ. 1990) Costanza, Farber, and Maxwell (Ecol. Econ. 1989) Farber (JEEM 1987) Farber and Costanza (JEM 1987)

5. Our Approach  We address wetland valuation from the perspective of future land-loss prevention Similar to how one approaches damage assessment  Our referendum asks respondents to evaluate a policy that will prevent expected future losses rather than a policy that will restore land already lost. The former is more readily-feasible than the latter The scenario on which our results are based are coast- wide.  our estimates are intended to reflect preferences for a unified large-scale approach to land-loss prevention that spans the entire Louisiana coast

7. The CV Scenario The questionnaire opened with this introduction:  Coastal Louisiana has lost an average of 34 square miles of land, primarily marsh, per year for the last 50 years. From 1932 to 2000, coastal Louisiana lost 1,900 square miles of land, roughly an area the size of the state of Delaware.  PLEASE SEE MAP INCLUDED WITH YOUR SURVEY.  Additionally, Hurricanes Katrina and Rita eroded an additional 217 square miles in 2005 alone. (Not shown on map.)  If no action is taken, Louisiana could potentially lose an additional 700 square miles of land, about equal to the size of the greater Washington D.C. – Baltimore area, by the year 2050.

8. Other scenario details  Program timing: To make the scenario appear realistic, we specified that it would take 5 years for the program to be fully implemented.  To make the scenario consistent with the map included with the questionnaire, we specified that the program, if implemented, would maintain land area at current levels through the year 2050.  In other words, the prevention program would, at minimum, shift projected losses as shown 35 years into the future.

9. WTP vs. WTA  One can conceive of at least two sets of respondents in the LA restoration case: WTA: perceives coastal land to be their birthright, where WTA would be the appropriate welfare measure  This set may, perhaps, perceive any losses as the result of human activity such as oil and gas exploration or the building of levees WTP: perceives coastal land loss as a natural phenomenon  likely perceives future losses as a given, with any losses prevented being something gained, rather than something saved, relative to no action Unable to identify each respondent ex ante  we split the sample in two, half receiving a WTP-style referendum question, and the other a WTA-style question.

10. WTP Question: A tax Suppose the State of Louisiana proposed a coast-wide project that would prevent the projected future losses (as shown in yellow on the map) from occurring. It is expected that it would take approximately 5 years for the project to be fully implemented, and projected land area would be maintained at current levels until the year 2050. If the project received a majority vote of support, it would be implemented, and each tax-paying household in Louisiana would be obligated to pay an additional tax of $X per year for 10 years. The tax payments would be collected on annual state income tax returns. How would you vote? 1- I would vote FOR the project: In other words, PREVENT future land losses and PAY AN ADDITIONAL ANNUAL TAX OF $X FOR 10 YEARS. 2- I would vote AGAINST the project: in other words, DO NOT PREVENT future land losses and pay NO ADDITIONAL TAX.

11. WTA Question: A tax refund If the project did not receive a majority vote of support, it would not be implemented, and the State would redistribute the funds such that each tax-paying household in Louisiana would receive an additional tax refund of $X per year for 10 years. The refunds would be distributed on annual state income tax returns. How would you vote? 1- I would vote FOR the project: In other words, PREVENT future land losses and receive NO ADDITIONAL TAX REFUND. 2- I would vote AGAINST the project: in other words, DO NOT PREVENT future land losses and RECEIVE AN ADDITIONAL TAX REFUND OF $X FOR 10 YEARS.

12. Data Collection  Questionnaire mailed to a stratified (by parish) random sample of 3,000 Louisiana households, obtained from Survey Sampling International, Inc.  1 st mailing sent during 3rd week of May 2009 included pre-paid $1 cash incentive, shown to increase response rates relative to either no incentive or post- paid incentives (Dillman 2007, Petrolia and Bhattacharjee 2009).  Replacement questionnaires sent during 3rd week of June 2009.  A total of 680 questionnaires were returned (22.7% response rate).

13. Referendum Responses by Bid ($/year)

14. Empirical Model  Standard RUM approach Utility of individual i is defined as U i = U(y i, z i ; q )  where y is household income; z is a vector of individual- specific characteristics, and q is coastal land quantity In WTP case, Individual votes Yes if U i (y i – t i, z i ; q 0 ) > U i (y i,z i ; q 1 ) In WTA case, Individual votes Yes if U i (y i + t i, z i ; q 1 ) > U i (y i,z i ; q 0 )  where t is the bid  q 1 is the state of nature where the program is implemented  q 0 is where it is not (q 0 > q 1 )  Weighted likelihood function to mitigate sample bias (pseudo- likelihood) Ratio of population income category proportion over sample income category proportion  Estimated (weighted) pseudo-likelihood probit model using Stata 11

15. Modeling Income and Bid  Because bids were relatively large, ranging from $50 to $1,189, we did not wish to impose the commonly-made assumption of constant marginal utility of income.  We adopted the Box-Cox Transformation, which specifies a composite bid-income term of where λ is Box-Cox Transformation parameter with K possible values When λ = 0, the Box-Cox transformation converges to the log- linear specification Following Greene (2000), estimated model for λ = {-2, 2} in increments of 0.1. The survey gathered household income by income categories. To construct our composite variable, income was interpolated as the mid-point in the category.  Our search resulted in the adoption of λ = 0.7.

16. Variable Descriptions

17. Regression Results

18. Nominal (Annual) Welfare Estimates

19. PV(WTP)

20. PV(WTA)

21. Summary of Results  The WTA dummy variable, included to capture any residual treatment differences was not found to be significantly different from zero Although it does affect welfare estimates  The Box-Cox income-bid term was significant and positive  Age was significant among WTA respondents only (at the p = 0.1 level), increasing the probability of a Yes vote by 15% for a 10-year increase in age  Whites were 19% more likely to vote Yes  Respondents with no confidence in government to enact restoration efforts in a timely manner were 13% less likely to vote Yes  Those citing storm protection were 45% more likely to vote Yes relative to all others over half of respondents indicated storm protection as their leading concern while voting  respondents citing some other concern (including environmental protection, protection of recreational opportunities, and protection against sea-level rise) were 28% more likely to vote Yes  Depending on the discount rate, our results indicate: $1,000 < WTP < $18,000 $5,000 < WTA < $45,000

22. Questions and Comments  Contact: Dan Petrolia  petrolia@agecon.msstate.edu Sponsoring Agency (NOAA):