A National Survey of Willingness to Pay for Restoration of Louisiana’S Coastal Wetlands

America’s Wetland? A National Survey of Willingness to Pay for Restoration of Louisiana’s Coastal Wetlands

Final Project Report Department of Agricultural Economics Mississippi State University

Revised April 1, 2013

Daniel R. Petrolia * Matthew G. Interis Joonghyun Hwang Mississippi State University

Michael K. Hidrue University of Delaware

Ross G. Moore USDA Farm Service Agency, Selmer, TN

GwanSeon Kim University of Georgia

*Corresponding Author: [email protected]

The authors wish to express their sincere gratitude to Kerry St. Pe and all the staff of the Barataria-Terrebonne National Estuary Program for providing valuable information and their assistance in improving the quality of the survey instrument.

This research was conducted under award NA06OAR4320264 06111039 to the Northern Gulf Institute by the NOAA Office of Ocean and Atmospheric Research, U.S. Department of Commerce; and supported by the USDA Cooperative State Research, Education & Extension Service, Multistate Project W-2133 “Benefits and costs of Natural Resources Policies Affecting Public and Private Lands” (Hatch # MIS-033120).

EXECUTIVE SUMMARY

A nationwide survey was conducted in the summer of 2011 via Knowledge Networks to estimate

the willingness to pay (WTP) for a large-scale restoration project in the Barataria-Terrebonne

National Estuary in coastal Louisiana. A split-sample approach was used to administer both a binary-choice (contingent valuation) and multinomial-choice (choice experiment) version of the survey, with the latter used to estimate willingness to pay for increments in three specific wetland ecosystem services: wildlife habitat, storm surge protection, and fisheries productivity.

A total of 3,464 respondents completed the valuation exercise, of which 3,228 (93%) had neither visited nor live/lived in the study region. (Note also that only 32 respondents, < 1%, were

Louisiana residents). Of the 1,397 respondents who completed the binary-choice version of the survey, 601 (43%) were willing to pay some positive amount of money for the proposed restoration project (costs ranged between $25 and $2,825 per household). Of the 2,067 respondents who completed the multinomial-choice version of the survey, 1,250 (60%) were willing to pay for some version of the restoration. Results indicate that confidence in federal and state government agencies, political leanings, and “green” lifestyle choices were significant explanatory factors regarding support. All three wetland ecosystem services significantly affected project support, with increased fisheries productivity having the largest marginal effect, followed by improved storm surge protection, and increased wildlife habitat. Mean WTP for an intermediate-scale restoration program ranges between $909 and $1,751 per household, with a total value between $105 billion and $201 billion, which exceeds a recent (and by far the largest)

$100 billion restoration cost estimate.

TABLE OF CONTENTS

I. Introduction…………………………………………………………….…..4

II. Study Area: Barataria-Terrebonne National Estuary………………….….13

III. Survey Design……………………………………………………………..17

IV. Survey Administration…………………………………………………….26

V. Data ……………………………………………………………………….29

VI. Overview of Economic Theory, Non-Market Valuation,

Stated Preference Methods, and Consequentiality ……………………….43

VII. Random Utility and Econometric Models………………………………...51

VIII. Results…………………………………………………………………….56

IX. Summary and Conclusions………………………………………………..69

X. References………………………………………………………………...72

XI. Appendix: Survey Instrument……………………………………………78

CHAPTER I

INTRODUCTION

The wetlands of coastal Louisiana make up one of the most important, productive ecosystems in

the United States. They support the largest commercial fishery in the lower 48 states, and

comprise the seventh-largest delta on Earth, containing 37 percent of the estuarine herbaceous

marshes in the continental United States (Couvillion et al. 2011).1 Over 2 million residents - more than 47% of the state’s population- live in Louisiana’s coastal parishes (Wang 2012).

These wetlands provide habitat for mammals, amphibians, fishes, and migratory birds and act as nurseries for shellfish and fish. The wetlands also provide valuable nutrients to surrounding habitats and help improve Louisiana’s water availability and quality. A primary function is that they act as natural water collection areas. As water moves through wetlands, many processes occur that benefit nature and society. These processes include cleaning pollution out of the water, absorption of excess nitrogen and phosphorous by wetland plants, denitrification, and destruction of intestinal bacteria in wastewater (Coreil and Barrett-O’Leary

2004).

Wetlands also play a vital role in reducing damage from storms along Louisiana’s Gulf

Coast. The wetlands provide a natural barrier for the inland by reducing storm surge and decreasing wave energy. This can be seen through a comparison of the damage to Florida’s

Atlantic Coast and Louisiana’s Gulf Coast from Hurricane Andrew in 1992. Florida’s coast does not have the wetland barriers that Louisiana’s coast has, and Florida received much more

1 The coastal zone covers approximately 14,913 square miles, of which 6,737 square miles is

water and 8,176 square miles is land (Louisiana Oil Spill Coordinator's Office 2005).

damage from the Hurricane Andrew than did Louisiana (Coreil and Barrett-O’Leary 2004). After the devastation of Hurricanes Katrina and Rita, models were used to examine the role that wetlands play in reducing storm damage. One such model examined the hypothetical situation where the wetlands east of the Mississippi River Gulf Outlet, the Gulf Intracoastal Waterway, and Lake Borgne turned into open water eight feet deep. That model found the if this open water scenario had existed the storm surge from Hurricane Katrina would have been three to six feet higher in St. Bernard Parish and New Orleans East (Working Group for Post-Hurricane Planning for the Louisiana Coast 2006). Wetlands also capture storm water runoff that can cause flooding.

Wetlands that make up 15 percent of the acreage in a watershed have the ability to reduce flood peaks by up to 60 percent (Coreil and Barrett-O’Leary 2004).

The wetlands of Louisiana also play a vital role for the state economy. The wetlands provide habitat for harvestable animals and timber. Twenty six percent (by weight) of commercial fish landings in the lower 48 states is provided by Louisiana’s wetlands. Also, this area is the nation’s largest shrimp, blue crab, and oyster producing areas (National Marine

Fisheries Service 2005). Almost 30,000 of Louisiana’s citizens have a job that is provided through this commercial fishing industry (Louisiana Department of Wildlife and Fisheries 2005).

The wetland timber industry contributes about $2 billion to the state, along with many jobs for the people of Louisiana (Coreil and Barrett-O’Leary 2004).

Louisiana’s wetlands also play an important role in protecting the state’s and the nation’s energy infrastructure from storm damage. Nearly 9,300 miles of oil and gas pipelines cross the wetlands of coastal Louisiana (United States Army Corps of Engineers 2004). Coastal Louisiana is also the home of the pricing point for natural gas throughout North America (Henry Hub), and

Port Fourchon is a port and supply point for hundreds of offshore drilling operations in the Gulf

5 of Mexico. A third of the nation’s oil and gas supply and 50 percent of the nation’s oil refining capacity is produced or transported in or near Louisiana’s wetlands (Louisiana Department of

Natural Resources 2006).

International commerce infrastructure is also protected by the wetlands. There are ten major navigation routes that are located in southern Louisiana. The ports in this area are some of the largest ports in the United States. These ports handle approximately 469 million tons of cargo each year, which represents 19% of the annual waterborne commerce in the United States

(United States Army Corps of Engineers 2003).

Louisiana’s wetlands also provide enjoyment, employment, and revenue through many recreational opportunities. In 2001, hunters spent $446 million, anglers spent $670 million, and wildlife watchers spent $165 million in Louisiana. Much of this can be attributed to the state’s coastal wetlands (Coreil and Barrett-O’Leary 2004).

Wetland Loss

Louisiana has been one of the states most affected by wetland loss in the United States.

Couvillion et al. (2011) estimate that coastal Louisiana has undergone a net change in land area of approximately 1,883 square miles from 1932 to 2010, representing a 25 percent decrease, which is about the size of the state of Delaware. These losses in Louisiana account for about 90 percent of the total wetland loss in the lower 48 states (Couvillion et al. 2011). Estimated lost rates vary, with Coreil and Barrett-O’Leary (2004) reporting about 2.3 square miles per year since the 1930s, but in excess of 40 square miles per year during the last 50 years, and between

25 and 35 square miles per year during the 1990s. The most recent estimate is a loss rate of

16.57 square miles per year for the period 1985-2010 (Couvillion et al. 2011). Recently,

Hurricanes Katrina and Rita destroyed approximately 200 square miles of marshlands in a single

hurricane season (Coastal Protection and Restoration Authority of Louisiana (CPRA) 2007).

Overall, during the past 100 years, these losses represent an acceleration of 10 times the natural

rate (Coastal Protection and Restoration Authority of Louisiana 2012b).

Losses are partly due to natural phenomena, such as sea level rise, subsidence, erosion, saltwater intrusion, and tropical storm impacts, as well as human activities such as dredged

canals, levees, upstream dams and soil conservation practices which have modified the

movement of freshwater and suspended sediment, and other development (Barras et al., 2003;

(Caffey, Savoie, and Shirley 2003; Dunbar et al., 1992; Coastal Protection and Restoration

Authority of Louisiana 2007). If no action is taken to curb this trend, future losses are forecasted

to be between 700 and 1,756 square miles by the year 2060 (United States Army Corps of

Engineers 2004; Desmond 2005; Coastal Protection and Restoration Authority of Louisiana

2012). The dollar value of these losses is estimated at $37 billion by 2050 (Louisiana Coastal

Wetlands Conservation and Restoration Task Force and the Wetlands Conservation and

Restoration Authority 1998); CPRA (2012) predicts annual damage from flooding in 50 years

could increase from an average of $2.4 billion to $23.4 billion. Still, this dollar amount does not

encompass the environmental, social, and cultural losses that would be incurred.

Restoration Efforts

Actions have been taken by the state and federal government to prevent wetland loss and

attempt to restore the wetlands of Louisiana. Some of the key initiatives that have been

undertaken are: the Federal Coastal Zone Management Act (1972), Louisiana Coastal Wetlands

Conservation Restoration and Management Act (1989), Louisiana Act 6 (1989), Barataria-

Terrebonne National Estuary Program (1990), The Gulf of Mexico Program (1991), and Sections

2004, 206, and 1135 of the Water Resources Development Act (of 1986, 1992, and 1996)

(United States Army Corps of Engineers 2004).

The most prominent piece of legislation that has addressed Louisiana’s coastal wetland loss is the Coastal Wetlands Planning, Protection and Restoration Act (CWPPRA). CWPPRA, also known as the Breaux Act, became law in November 1990. CWPPRA authorized the

Federal government to use funds to address wetland loss across the nation. Louisiana was a major focus of the Breaux Act. CWPPRA is a cost sharing program between the federal government and the states to fund projects that restore, maintain, and prevent loss of wetlands in the United States. In Louisiana, projects are sponsored by five federal agencies (U.S. Army

Corps of Engineers, Natural Resources Conservation Service, National Marines Fisheries

Service, U.S. Fish and Wildlife Service, U.S. Environmental Protection Agency) and the State of

Louisiana. The total cost over the life of the program is estimated to reach $2 billion (United

States Army Corps of Engineers 2005). From 1990 to 2006, CWPPRA funding averaged approximately $60 million for each coastal restoration project. As of 2006, there are 78 projects costing $624.5 million that have been constructed, are in the process of being constructed, or have been approved for construction. In addition, 47 projects that cost approximately $913.4 million are in the engineering and design phase. The combined benefits of all of the projects are estimated to be 103,281 acres re-established or protected and 515,213 acres restored. Funding was reauthorized by Congress for CWPPRA through 2019 (Louisiana Coastal Wetlands

Conservation and Restoration Task Force 2006).

In 1998 a new plan for the Louisiana coast was published. It was entitled Coast 2050:

Toward a Sustainable Coastal Louisiana (Louisiana Coastal Wetlands Conservation and

Restoration Task Force and the Wetlands Conservation and Restoration Authority 1998). The overarching goal of the plan was, “to sustain a coastal ecosystem that supports and protects the environment, economy and culture of southern Louisiana, and that contributes greatly to the economy and well-being of the nation” (pg. 2). The Coast 2050 Plan provides a set of restoration strategies for restoring southern Louisiana’s wetlands to a sustainable level

(Louisiana Coastal Wetlands Conservation and Restoration Task Force 2006). The plan emphasized that CWPPRA was indeed making a positive contribution, but not to the desired level. The Coast 2050 Plan called for $14 billion to be spent over the next 30 years starting in

1998. However, funding was not granted to cover all of the Coast 2050 plans.

The Coast 2050 Plan laid the ground work for the Louisiana Coastal Area study (United

States Army Corps of Engineers 2004). The study produced the Louisiana Coastal Area

Comprehensive Coastwide Ecosystem Restoration study report and Draft Programmatic

Environmental Impact Study (DPEIS). Another report on Louisiana’s restoration efforts was the

Louisiana Coastal Area (LCA) Near-Term Ecosystem Restoration Plan: Evolution of Coastal

Restoration in Louisiana (United States Army Corps of Engineers 2004). The report identified plans to rehabilitate Louisiana’s coast. The LCA plan would cost approximately $1.9 billion to implement (Louisiana Coastal Wetlands Conservation and Restoration Task Force 2006). The

LCA plan was not funded.

The Energy Policy Act of 2005 provided Louisiana with $450 million over the following four years for restoration efforts and to mitigate some impacts of Outer Continental Shelf oil and gas production. The money comes through the Coastal Impact Assistance Program (CIAP) in the

2005 energy bill (Louisiana Coastal Wetlands Conservation and Restoration Task Force 2006).

Then in 2005, Hurricane Katrina and Rita devastated coastal Louisiana. Reports

estimated that approximately 128,000 acres of wetlands were converted into open water. The

hurricanes also had devastating effects on human life, infrastructure, and property. It is

estimated that the losses of physical capital from Hurricanes Katrina and Rita totaled between

$70 and $130 billion (Coastal Protection and Restoration Authority of Louisiana 2007). This

prompted the State of Louisiana to restructure the Wetland Conservation and Restoration

Authority to form the Coastal Protection and Restoration Authority. Prior to these hurricanes,

planning for coastal restoration and hurricane protection were separated. After these hurricanes,

the CPRA began considering “hurricane protection and the protection, conservation, restoration

and enhancement of coastal wetlands and barrier shorelines or reefs” jointly (Coastal Protection

and Restoration Authority of Louisiana 2010).

Over the next few years CPRA began working on a Master Plan that would outline how to achieve a sustainable coast. In April 2007, the state released the Integrated Ecosystem

Restoration and Hurricane Protection: Louisiana's Comprehensive Master Plan for a Sustainable

Coast (Coastal Protection and Restoration Authority of Louisiana 2007). The hurricanes of 2005 drastically changed the approach to maintaining and restoring Louisiana’s coastal wetlands. Pre-

Katrina and pre-Rita protection and restoration efforts were primarily focused on environmental and ecological benefits of the wetlands. Post-Katrina and post-Rita efforts have become increasingly more concerned with the protection that wetlands can provide against hurricane and flood damage. Most recently, the Master Plan has been updated, and a draft version was released in January 2012 (CPRA 2012a).

Project Objectives

Federal and State governments have already begun to address the wetland loss problem in

Louisiana, but a positive perception of these efforts by the public is important for the continued

progress of these efforts. However, evidence supporting the claim that public support actually exists is limited. Several studies have estimated the value of benefits of wetlands, although most of them really do not directly address this question of public support. Kazmierczak provides summaries of the literature on wetland value estimates specific to habitat/species protection

(2001a), hunting and fishing (2001b), and water quality (2001c). Brouwer et al. (1999), Brander,

Florax, and Vermaat (2006), and Woodward and Wui (2001) provide summaries of the literature as well, and also provide meta-analyses of the literature. Costanza et al. (2008) estimated the value of coastal wetlands for storm surge attenuation using the replacement cost method, and

Farber (1996) estimated the values of wetlands for fisheries production in coastal Louisiana

using the benefit transfer method.

Bergstrom et al. (1990) was the first to address the question of public support directly by

utilizing stated-preference methods (they also utilized the travel-cost method) to estimate benefit

values of wetlands in Louisiana. However, they focused only on resource users and only on

recreational values. Petrolia and Kim (2011) and Petrolia, Moore, and Kim (2011) provide more

recent estimates, and surveyed residents from across the entire state of Louisiana, thus including

both resource users and non-users, and did not limit values to one particular benefit. Combined,

these studies provide strong evidence that support does in fact exist, at least within the state.

Given the large share of total U.S. wetlands that are situated in Louisiana, and given the benefits

provided by Louisiana’s wetlands that reach beyond the state’s borders, some have begun to

promote the state’s wetlands as “America’s Wetland” (America’s Wetland Foundation 2012).

Thus, although the work cited above takes a step in the right direction, it does not address the

11 question of whether Louisiana really is “America’s Wetland.” For this purpose a survey of national scope is required. Whether the American public perceives it as such is a testable hypothesis that forms the focus of this research.

The overall objective of this research is to answer the following two questions:

1. How much money are U.S. households willing to pay to restore Louisiana’s coastal wetlands?

2. What specific ecosystem services provided by Louisiana’s coastal wetlands are the key drivers of U.S. households’ willingness to pay, and what are U.S. households willing to pay for specific increments of these services?

CHAPTER II

STUDY AREA: BARATARIA-TERREBONNE NATIONAL ESTUARY

The Barataria-Terrebonne estuarine complex encompasses the 4.2 million acres of wetlands, ridges, forests, farmlands, and communities between the Mississippi and Atchafalaya

River Basins in southeast Louisiana. Bayou Lafourche separates this complex into two basins,

Barataria Basin to the east, and Terrebonne Basin to the west (see Figure 1). The Lower

Barataria-Terrebonne Estuary covers about 2.7 million acres, an area roughly three-quarters of the size of the state of Connecticut. More than 80% of the land area is wetlands (swamps and marshes) and barrier islands. The remainder of the land contains homes, businesses, and farms.

The estuary is home to over 500,000 people, and provides storm protection for over 1 million people, including the city of New Orleans.

The estuary provides habitat for 735 species of birds, finfish, shellfish, reptiles, amphibians, and mammals. At least 28 species are either endangered or threatened. The estuary supports almost 20% of the estuarine-dependent fisheries of the U.S. and is a major location for fishing, hunting, and bird watching. The estuary is rich in minerals including crude oil, natural gas, and salts, and is home to some of the busiest shipping routes in the U.S.

Couvillion et al. (2011) estimate a 398 and 358 square mile net loss of land area in the

Terrebonne and Barataria basins, respectively, since 1956. Figure 2 shows these changes for the period 1956-2000. Of Louisiana’s nine major basins, these two have the highest and second highest land loss rates, respectively. Furthermore, these two basins alone account for 49 percent of the total persistent losses across all basins (Couvillion et al. 2011).

Figure 1. Barataria-Terrebonne Estuarine System (courtesy of Barataria-Terrebonne National Estuary Program)

Barataria-Terrebonne National Estuary Program

The Barataria-Terrebonne National Estuary Program (BTNEP) is one of 28 National

Estuary Programs throughout the United States and its territories. The National Estuary Program was established by Congress through section 320 of the Clean Water Act of 1987. The

Barataria-Terrebonne estuarine complex became a National Estuary in 1990.

Upon inclusion in the National Estuary Program, a diverse group of stakeholders was assembled including government, business, scientists, conservation organizations, agricultural interest, and individuals. This group is known as the Management Conference and it ensures the voice of all that live, work, and play in the Barataria-Terrebonne estuarine complex is heard.

The Barataria-Terrebonne National Estuary Program operates under the Comprehensive

Conservation and Management Plan (CCMP). The CCMP was developed by consensus of the

Management Conference members following a five year collaborative effort. A formal management plan with specific action plans was developed to promote and preserve the

Barataria-Terrebonne estuarine complex.

The severity of loss in these basins, as well as the presence of the Barataria-Terrebonne

National Estuary Program (BTNEP) led to the selection of this area as the focus of the study.

The staff of BTNEP, located on the campus of Nicholls State University, in Thibodaux,

Louisiana, proved an invaluable resource for improvement in the quality of the survey instrument used in this study.

Figure 2. Habitat changes in the Lower Barataria-Terrebonne Estuary (1956-2000) (Courtesy of the Barataria-Terrebonne National Estuary Program)

CHAPTER III

SURVEY DESIGN

A survey was designed to estimate non-market values (i.e., welfare measures) for changes in ecosystem services associated with coastal wetland and barrier island restoration in

Louisiana’s Barataria-Terrebonne National Estuary. The survey proposed to respondents one or more wetland and barrier island restoration programs and asked respondents if they would hypothetically be willing to pay a specified amount to implement one of the proposed restoration programs.

The survey explained to respondents that wetlands and barrier islands in the estuary were being lost due to “natural erosion, sea-level rise, sinking of land, winds, tides, currents, and major storms”, as well as human development such as the construction of river channels and levees. Respondents were asked to consider, evaluate, and indicate their preference for a set of proposed projects that would restore roughly 50% of land lost since 1956. The year 1956 was chosen because this was the year when diligent measurement of land loss began, according to experts at the Barataria-Terrebonne National Estuary Program center in Thibodaux, Louisiana.

The projects under consideration were large-scale land restoration projects which included “wetland building, barrier island restoration, freshwater and sediment diversions, and the movement of large amounts of soil on barges and via pipelines.” The survey focused on three main benefits of restoration: improved wildlife habitat, measured as the percentage of created land generally suitable for wildlife habitat; storm surge protection, measured as the percentage of residents in the area that would have improved storm surge protection; and improved commercial fish harvest, measured as the percentage improvement in harvest levels of

major commercial (Gulf of Mexico) fish such as oysters and shrimp. The specific levels of

changes to these ecosystem services depended on the version of the survey each respondent

received, as detailed in the next paragraphs.

Two versions of the survey were constructed. In the first version, respondents were

presented with a single restoration program and were asked whether they were willing to pay a

stated amount to implement the program, or to not implement any project, incur no cost, and

allow land loss to continue at its current rate. This version is referred to as the binary-choice version since respondents are choosing between two options, “yes” and “no”. The project in the binary-choice version proposed to restore 50% of land lost since 1956, 50% of which would be suitable for wildlife habitat, which would increase storm surge protection for 30% of residents in the estuary, and increase fish harvest levels by 15%. (Note that these levels correspond to the

“intermediate” levels used in the multinomial-choice version discussed below. See Table 1.)

The price to the respondent for the project took on one of nine randomly-assigned dollar values

{$25, 90, 155, 285, 545, 925, 1305, 2065, 2825}. Figure 3 shows an example choice question for the binary-choice version. In the second version, respondents were asked to choose between two different restoration programs, each available at a specified price, which differ according to how much habitat is restored, how many people receive increased storm protection, and by how much fish harvest levels increase (see Table 1 for the levels used). Alternatively, people could vote to implement neither of these programs, incur no cost, and allow land loss to continue at its current rate. This version is referred to as the multinomial-choice version because respondents are choosing between three options (either of the two programs or neither). The specific attribute levels actually shown to each respondent depended on the choice set to which the respondent

18 was randomly assigned. 2 The twelve possible choice sets are shown in Table 2. In all cases, the third alternative was the no-cost, no-action (status-quo) alternative.3 Figure 2 shows an example choice question for the multinomial-choice version. The reason to use both formats is that each has its own advantages and disadvantages, and welfare estimates have been shown repeatedly to differ across formats. Thus, we split the sample across both in order to obtain a robust set of results. Note that all respondents under both versions were also given the option to not vote, i.e., to opt out of responding to the vote question entirely.

2 Given six possible levels for the price and three possible levels for each of the other three attributes, there are a total of 162 possible choice sets (33 * 61). However, the more choice sets actually used in the survey, the fewer observations the researcher obtains on each choice set for a fixed number of respondents. Furthermore, little information is gained from certain choice sets, for example those in which one of the three options is obviously better than the other two since everyone will tend to pick that superior option. Therefore, it is common practice to restrict the number of choice sets to a smaller subset from which it is still possible to estimate welfare measures. With a linear specification for price and a dummy specification for the other attribute levels, the minimum required number of choice sets is 8, but to allow for some flexibility in estimation, we developed a design with 12 choice sets. See Kuhfeld (2010) for more information.

3 Although most multinomial-choice surveys utilize “repeated choice”, where each individual respondent evaluates multiple choice sets, we wished to avoid any of the confounding effects associated with this approach (see Carson and Groves, 2007) and presented each respondent with exactly one choice set to evaluate.

We used 100 responses to two pilot tests of the multinomial-choice survey to estimate the parameters of a multinomial logit model (see Greene 2012) which are used in SAS's choice

macro function (Kuhfeld 2010) to generate an efficient survey design. This macro searches for a

design with the highest D-efficiency and our design had a relative D-efficiency of 4.71. See

Kuhfeld (2010) for more details on multinomial-choice survey design.

Table 1. Attribute levels and descriptions. All non-price attributes set to the medium level for the contingent valuation version. No action alternative (SQ): Land loss Action Alternatives: expected to continue 50% of lost land restored at 4,500 to 7,100 acres per year Low Medium High Wildlife Habitat: No additional habitat x% of restored land and current habitat suitable as habitat 25% 50% 75% expected to decline Storm surge protection: No improvement and improved protection current habitat for x% of residents 5% 30% 50% expected to decline Commercial Maintains fisheries harvest: current No improvement and x% higher harvest harvest current harvest levels levels levels 15% 30% expected to decline Price: $x one-time $25, $90, $155, $285, $545, $925, tax $1305*, $2065*, $2825* $0 * These prices were used in the contingent-valuation version only

Table 2. Attribute levels for multinomial-choice survey choice sets. Commercial Wildlife Habitat: Storm surge protection: fisheries harvest: x% of restored land Choice improved protection for x% x% higher harvest Price: $x one- suitable as habitat Set Alternative of residents levels time tax 1 A 25% 30% Maintains current $155 B 50% 50% 15% $285 2 A 50% 50% 15% $545 B 25% 5% 30% $155 3 A 75% 5% Maintains current $285 B 50% 30% 30% $90 4 A 50% 30% Maintains current $545 B 25% 5% 15% $90 5 A 25% 30% 15% $25 B 75% 5% Maintains current $90 6 A 75% 30% 15% $545 B 25% 50% 30% $925 7 A 25% 5% 30% $90 B 75% 30% 15% $155 8 A 75% 30% 30% $925 B 50% 5% 15% $285 9 A 25% 50% Maintains current $155 B 75% 5% 30% $285 10 A 75% 50% Maintains current $155 B 50% 30% 30% $25 11 A 50% 5% Maintains current $25 B 75% 30% 30% $285 12 A 50% 5% Maintains current $25 B 25% 30% 15% $90

The survey was subject to a fairly extensive vetting process. First, the researchers met with staff at the Barataria-Terrebonne National Estuary Program center in Thibodaux, Louisiana, to discuss the feasibility and believability of projects like the one proposed in the survey, the relevant project attributes that people would most likely care about, etc. Then in early 2011, two focus groups were held, using staff from various departments at Mississippi State University, the first of which was used only to narrow down the appropriate attributes for the survey, and the second of which focused on a more complete version of the survey to check for clarity, bias, etc.

These participants were deliberately chosen not to be “experts” in anything related to the study since the target population was the general U.S. population. The multinomial-choice version of the survey instrument was then pre-tested through Knowledge Networks, who administered the survey to approximately 30 respondents, the main focus being on honing the prices used in the survey. A second pilot study was administered to roughly 100 respondents, three quarters of whom received the multinomial-choice version of the survey and the remainder receiving the binary-choice version.

Figure 3. Example binary-choice valuation question.

With Project: Without Project (No Action): 50% of lost land restored Land loss expected to continue at 4,500 to 7,100 acres per year

50% of restored land suitable No additional habitat and current Wildlife habitat as habitat habitat expected to decline

Storm surge Improved protection for 30% No improvement and current of residents protection protection expected to decline

Commercial fish No improvement and current 15% higher harvest levels harvest harvest levels expected to decline

Share of total cost to your household $925 $0 (one-time tax)

I prefer: ⁬ ⁬

I prefer not to vote: ⁬

Figure 4. Example multinomial choice set (corresponds to Choice Set 1 from Table 2)

Project A: Project B: No Action: 50% of lost land restored 50% of lost land restored Land loss expected to continue at 4,500 to 7,100 acres per year

25% of restored land suitable 50% of restored land suitable No additional habitat and current Wildlife habitat as habitat as habitat habitat expected to decline

Storm surge Improved protection for 5% of Improved protection for 30% No improvement and current protection residents of residents protection expected to decline

Commercial fish Maintains current harvest No improvement and current 15% higher harvest levels harvest levels harvest levels expected to decline Share of total cost to your household $155 $285 $0 (one-time payment) I prefer: ⁬ ⁬ ⁬

⁬ I prefer not to vote.

Questions were included to collect other relevant data. Respondents were asked how

closely (0 = not at all, 1 = somewhat closely, 2 = very closely) they followed news about the

Deepwater Horizon oil spill in the Gulf of Mexico in April, 2010; whether they had ever lived in

or visited the Barataria-Terrebonne Estuary (=1 if yes, 0 otherwise), whether they had made

changes in their lifestyle within the past five years to help protect the environment (= -1 if no

changes, 0 if minor changes, 1 if major changes); their perceived consequentiality of the survey

(i.e., the degree to which they perceived their own vote would influence the outcome of the survey and the degree to which the outcome of the survey would influence actual policy); their confidence in federal agencies and Louisiana state government to implement the project proposed in the survey (“a lot of confidence”, “some confidence”, “little confidence”, “no confidence” and “I don’t know); and whether the survey provided enough information for them to “make an informed choice”. Demographic data, including a measure of respondent political ideology (rated on a seven-point scale (1 = extremely liberal, 7 = extremely conservative)), were pre-collected by Knowledge Networks.

CHAPTER IV

SURVEY ADMINISTRATION

Knowledge Networks was contracted on behalf of the principal investigators (Mississippi State

University) to administer the survey. The target population consisted of the following: non-

institutionalized adults age 18 and over, residing in the United States. Knowledge Networks sampled households from its KnowledgePanel, a probability-based web panel designed to be representative of the United States. The data collection field periods were as follows:

Stage Start Date End Date

Pretest 4/28/2011 5/2/2011

Pilot 5/7/2011 5/23/2011

Main 6/10/2011 7/5/2011

Participants completed the main survey in 10 minutes (median). Out of 5,185 people sampled,

3,464 (66.8%) responded. Of the 3,464 respondents, 1,397 took the binary-choice version and

2,067 took the multinomial-choice version. Besides the standard measures taken by KN to enhance survey cooperation, the following steps were also taken: email reminders to non- responders were sent on day three of the field period; additional email reminders to non- responders were sent on June 24, June 27, and June 30, 2011; and participants were eligible to win an in-kind prize through a monthly KN sweepstakes.

Knowledge Networks Methodology

Knowledge Networks (KN) has recruited the first online research panel that is representative of the entire U.S. population. Panel members are randomly recruited through probability-based sampling, and households are provided with access to the Internet and hardware if needed.

Knowledge Networks selects households by using address-based sampling methods; formerly, KN relied on random-digit dialing (RDD). Once households are recruited for the panel, they are contacted by e-mail for survey taking or panelists visit their online member page for survey taking (instead of being contacted by phone or postal mail). This allows surveys to be fielded very quickly and economically. In addition, this approach reduces the burden placed on respondents, since e-mail notification is less intrusive than telephone calls, and most respondents find answering Web questionnaires more interesting and engaging than being questioned by a telephone interviewer. Furthermore, respondents have the freedom to choose what time of day to participate in research.

Unlike Internet convenience panels, also known as “opt-in” panels, that include only individuals with Internet access who volunteer themselves for research, KnowledgePanel recruitment uses dual sampling frames that includes both listed and unlisted telephone numbers, telephone and non-telephone households, and cell-phone-only households, as well as households with and without Internet access. Only persons sampled through these probability-based techniques are eligible to participate on KnowledgePanel. Unless invited to do so as part of these national samples, no one can participate in the panel.

Documentation regarding KnowledgePanel sampling, data collection procedures, weighting, and IRB-bearing issues are available at the online resources below:

• http://www.knowledgenetworks.com/ganp/reviewer-info.html

• http://www.knowledgenetworks.com/knpanel/index.html

• http://www.knowledgenetworks.com/ganp/irbsupport/

CHAPTER V

DATA

This chapter reports the raw responses to the questions asked in the survey. Questions as they appeared in the survey are shown in the table title (where appropriate), with results following.

Table 3. Sample Demographics (N = 3464) % (Mean) % (Mean) Age (48.9) Marital status Education Married 0.58 Less than high school 0.09 Widowed 0.04 High school 0.29 Divorced 0.11 Some college 0.29 Separated 0.01 bachelor's degree or 0.33 Never married 0.19 higher Ethnicity Living with partner 0.08 White, non-Hispanic 0.75 Region Black, non-Hispanic 0.09 Northeast 0.19 Other, non-Hispanic 0.03 Midwest 0.23 Hispanic 0.10 South 0.35 2+races, non-Hispanic 0.03 West 0.23 Male 0.49 Employment status Household Income Paid employee 0.49 Less than $20,000 0.13 Self-employed 0.07 $20,000-$50,000 0.28 Temporarily layoff 0.01 $50,000-$100,000 0.33 Looking for work 0.08 Over $100,000 0.26 Retired 0.20 Internet access Not working-disabled 0.07 No 0.22 Not working-other 0.08 Yes 0.78

Table 4. Comparison of the Sample and Population (U.S.) Sample Population (N=3,464) (U.S.) % % Age (years)* (48.9) (36.8) Education: High school or higher* 0.91 0.85 bachelor's degree or higher* 0.33 0.28 Ethnicity: White, non-Hispanic* 0.75 0.64 Black, non-Hispanic* 0.09 0.13 Other, non-Hispanic* 0.03 0.05 Hispanic* 0.10 0.16 2+races, non-Hispanic 0.03 0.03 Gender: Male 0.49 0.49 Household Income (median)* (67,500) (51,914) * indicates that the sample and the population are statistically different based on t-test (for age and income) and proportions test (for all other categories).

Table 5. Survey Question: How familiar are you with the wetland and barrier island loss issue in coastal Louisiana? Frequency Percent Very familiar 95 0.03 Somewhat familiar 1053 0.30 Not at all familiar 2287 0.66 Refused 29 0.01 Total 3464

Table 6. Survey Question: Have you ever visited New Orleans or another part of coastal Louisiana?

Frequency Percent Yes 1113 0.32 No 2311 0.67 Refused 40 0.01 Total 3464

Table 7. Survey Question: How familiar are you with the Barataria-Terrebonne Estuary? Frequency Percent I have visited or lived in the 212 0.06 area. I have never visited or lived in 664 0.19 the area, but have heard of it. I have never heard of it. 2564 0.74 Refused 24 0.01 Total 3464

Table 8. Survey Question: Overall, how concerned are you about these changes in the Lower Barataria- Terrebonne Estuary? Frequency Percent Very concerned 481 0.14 Concerned 1187 0.34 Mildly concerned 1191 0.34 Not at all concerned 575 0.17 Refused 30 0.01 Total 3464

Table 9. Survey Question: How likely do you think it is that the results of this survey will shape the direction of future policy in the Lower Barataria- Terrebonne Estuary? Frequency Percent Very likely 197 0.06 Somewhat likely 1346 0.39 Unlikely 1088 0.31 I don’t know 796 0.23 Refused 37 0.01 Total 3464

Table 10. Survey Question: Have you heard of the Mississippi Flyway, a migratory bird route? Frequency Percent Yes 739 0.21 No 2223 0.64 Not sure 475 0.14 Refused 27 0.01 Total 3464

Table 11. Survey Question: Which, if any, of the following outdoor activities do you engage in? Please check all that apply. Frequency Percent of Sample engaged in activity Freshwater fishing 884 0.26 Saltwater fishing 321 0.09 Boating/Canoeing 754 0.22 Hunting 324 0.09 Bird watching 566 0.16 Hiking/nature walking 1482 0.43 Other 411 0.12

Table 12. Survey Question: How closely did you follow the BP Oil Spill and Deepwater Horizon Accident in the Gulf of Mexico last summer? Frequency Percent Very closely 705 0.20 Somewhat closely 2176 0.63 Not at all 486 0.14 I was not aware of the oil spill and accident in the Gulf 67 0.02 of Mexico last summer. Refused 30 0.01 Total 3464

Table 13. Survey Question: Thinking about your own shopping and living habits over the last five years, would you say you have made major changes, minor changes, or no changes to help protect the environment? Frequency Percent Major changes 525 0.15 Minor changes 2163 0.63 No changes 740 0.21 Refused 36 0.01 Total 3464

Table 14. Survey Question: The projects discussed in this survey would likely involve the cooperation of federal agencies, Louisiana state and local governments, as well as private companies. How much confidence do you have in each of these to carry out these projects? Federal Louisiana state Louisiana local Private Agencies government government companies Freq % Freq % Freq % Freq % A lot of 163 0.05 307 0.09 358 0.10 210 0.06 confidence Some confidence 1066 0.31 1212 0.35 1168 0.34 1089 0.31 Little confidence 1127 0.33 901 0.26 836 0.24 1012 0.29 No confidence 676 0.20 346 0.10 394 0.11 600 0.17 I don’t know 374 0.11 638 0.18 646 0.19 493 0.14 Refused 58 0.02 60 0.02 62 0.02 60 0.02 Total 3464 3464 3464 3464

Table 15. Survey Question: Did you file a Federal income tax return this year? Frequency Percent Yes 2954 0.85 No 477 0.14 Refused 33 0.01 Total 3464

Table 16. Survey Question: Thinking about your overall experience in this survey, indicate how strongly you agree with each of the following statements. The survey provided Information in the Information in the enough information for survey was easy to survey was presented me to make an understand. in an unbiased way. informed choice. Frequency Percent Frequency Percent Frequency Percent Strongly disagree 122 0.04 107 0.03 102 0.03 Disagree 228 0.07 135 0.04 217 0.06 No strong opinion 1131 0.33 659 0.19 1112 0.32 (neutral) Agree 1555 0.45 1985 0.57 1570 0.45 Strongly agree 382 0.11 522 0.15 410 0.12 Refused 46 0.01 56 0.02 53 0.02 Total 3464 3464 3464

Hypothetical Referendum Questions

Binary Choice Format

In the binary choice format, respondents evaluate one fixed version of the project (but with a randomly-assigned cost) and vote either for it or against it (i.e., no action). Forty percent of the total respondents (1,397 out of 3,464) received this format. Below is the referendum question as presented in the survey:

Once again, here are the expected outcomes and project cost. The project would be completed in 5 years and the benefits are expected to last for 50 years. The No Action option means that the restoration project would not be implemented. For this advisory vote, assume that the choice receiving the most votes would be adopted. Please indicate your choice at the bottom of the table below.

Table 17 shows how many people voted for the project, voted against the project, and refused to vote by randomly-assigned cost. For example, 117 people voted for the project, 25 people voted against the project, and 36 people refused to vote when project cost was stated to be $25 per household. Generally speaking, the proportion of votes for the project decreases as cost increases. Also, at relatively low cost levels, a majority of respondents vote for the project.

Table 17. Binary-choice vote by offered price Frequency (%) Vote for Vote against Refused Total $25 117 (0.66) 25 (0.14) 36 (0.20) 178 $90 102 (0.52) 45 (0.23) 50 (0.25) 197 $155 75 (0.42) 58 (0.33) 44 (0.25) 177 $285 94 (0.51) 46 (0.25) 44 (0.24) 184 $545 65 (0.39) 59 (0.35) 45 (0.27) 169 $925 45 (0.30) 53 (0.35) 52 (0.35) 150 $1,305 38 (0.28) 53 (0.39) 45 (0.33) 136 $2,065 44 (0.31) 57 (0.40) 43 (0.30) 144 $2,825 21 (0.34) 22 (0.36) 19 (0.31) 62 Total 601 (0.43) 418 (0.30) 378 (0.27) 1397

Multinomial Choice Format

In the multinomial choice format, respondents evaluate two versions of the project (each with randomly-assigned attribute levels and cost) and vote for one of the two versions, or neither (i.e., no action). Sixty percent of the total respondents (2,067 out of 3,464) received this format. We re-present the set of blocks indicating the specific attribute and cost levels presented to respondents below:

Table 18. Attribute levels for multinomial choice experiment survey blocks. Commercial Wildlife Habitat: Storm surge protection: fisheries harvest: x% of restored land improved protection for x% x% higher harvest Price: $x one- suitable as habitat Block Alternative of residents levels time tax 1 A 25% 30% Maintains current $155 B 50% 50% 15% $285 2 A 50% 50% 15% $545 B 25% 5% 30% $155 3 A 75% 5% Maintains current $285 B 50% 30% 30% $90 4 A 50% 30% Maintains current $545 B 25% 5% 15% $90 5 A 25% 30% 15% $25 B 75% 5% Maintains current $90 6 A 75% 30% 15% $545 B 25% 50% 30% $925 7 A 25% 5% 30% $90 B 75% 30% 15% $155 8 A 75% 30% 30% $925 B 50% 5% 15% $285 9 A 25% 50% Maintains current $155 B 75% 5% 30% $285 10 A 75% 50% Maintains current $155 B 50% 30% 30% $25 11 A 50% 5% Maintains current $25 B 75% 30% 30% $285 12 A 50% 5% Maintains current $25 B 25% 30% 15% $90

Below is the referendum question as presented in the survey:

Once again, here are the available options. Both Project A and Project B would be completed in 5 years and the benefits are expected to last for 50 years. The No Action option means that neither restoration project would be implemented. For this advisory vote, assume that the choice receiving the most votes would be adopted. Please indicate your choice at the bottom of the table below.

Table 19 shows how many respondents voted for Alternative A, Alternative B, no action, and refused to vote by randomly-assigned choice set. For example, of the respondents assigned to

Choice Set 1, 38 voted for Alternative A, 49 votes for Alternative B, 39 voted for no action, and

42 refused to vote. It is difficult to ascertain any general trends directly from the table given the multiple attributes being varied across the choice sets. Trends and relative importance of attributes are more readily identified in the multiple regression analysis presented in the next chapter.

Table 19. Frequency of respondents’ votes to proposed alternatives (%) Choice Alternative Alternative No Action Refused Total Set A B 1 38 (0.23) 49 (0.29) 39 (0.23) 42 (0.25) 168 2 28 (0.17) 55 (0.34) 35 (0.22) 44 (0.27) 162 3 19 (0.11) 98 (0.56) 29 (0.17) 29 (0.17) 175 4 25 (0.14) 84 (0.46) 27 (0.15) 47 (0.26) 183 5 92 (0.51) 28 (0.16) 26 (0.15) 33 (0.18) 179 6 77 (0.42) 21 (0.12) 36 (0.20) 48 (0.26) 182 7 36 (0.21) 70 (0.41) 31 (0.18) 33 (0.19) 170 8 38 (0.21) 64 (0.36) 31 (0.17) 46 (0.26) 179 9 49 (0.28) 41 (0.23) 43 (0.24) 44 (0.25) 177 10 44 (0.26) 83 (0.49) 18 (0.11) 23 (0.14) 168 11 61 (0.37) 42 (0.26) 19 (0.12) 42 (0.26) 164 12 57 (0.36) 51 (0.32) 15 (0.09) 37 (0.23) 160 Total 564 686 349 468 2067

These referendum questions were followed by debriefing questions regarding respondent perceptions of the survey and how others were likely to vote.

Table 20. Survey Question: When voting, what expectations, if any, did you have about how others might vote? Binary-choice Frequency Percent I thought most people would vote 208 0.15 for the project. I thought most people would vote 358 0.26 against the project. I thought the votes would be roughly even across the two 325 0.23 options. I didn’t really think about it. 498 0.36 Refused 8 0.01 Total 1397 1.0 Multinomial-choice Frequency Percent I thought most people would vote 313 0.15 for “Project A”. I thought most people would vote 351 0.17 for “Project B”. I thought most people would vote 399 0.19 for “No Action”. I thought the votes would be roughly even across the three 210 0.10 options. I didn’t really think about it. 769 0.37 Refused 25 0.01 Total 2067 1.0

Table 21. Survey Question: When voting, how important did you think your vote would be in determining which option received the most votes? Frequency Percent Very important 397 0.12 Somewhat important 1355 0.39 Not important 896 0.26 I didn’t really think 775 0.22 about it Refused 41 0.01 Total 3464 1.0

The following question was asked of respondents who chose “No Action” in question 5 or 5-1.

Table 22. Survey Question: You chose the “No Action” option. Would you mind telling us why? Frequency Percent I don’t really have a specific reason why. 23 0.03 I’m interested, but I can’t afford it. 172 0.22 I don’t think the expected benefits are worth 55 0.07 it. Society has more important problems than 134 0.17 restoring wetlands and barrier islands. I do not support any kind of tax increases. 130 0.17 I do not live in the area – only people who 116 0.15 live in the area should pay for the project. Other 132 0.17 refused 5 0.01 Total 767

The following question 9A is asked to 764 respondents who chose “I prefer not to vote”.

Table 23. Survey Question: You chose not to vote. Would you mind telling us why? Frequency Percent I don’t really have a specific reason why. 136 0.16 I’m not interested. 142 0.19 I don’t feel that my opinion should 200 0.24 influence policy in the area. The options seemed equally desirable so I 84 0.11 could not decide which I preferred. The survey did not give me enough 98 0.12 information to make a proper choice. Other 145 0.17 refused 14 0.02 Total 819

CHAPTER VI

OVERVIEW OF ECONOMIC THEORY, NON-MARKET VALUATION, STATED PREFERENCE METHODS, AND CONSEQUENTIALITY

Environmental goods and services provide benefits to humans in much the same way as

market goods. However, because of the public goods nature of environmental goods, many do

not have markets, and thus prices do not exist. Thereby, the value of environmental goods is

difficult to measure. Theoretical concepts of economic welfare used to evaluate the value of

environmental goods include consumer surplus, compensating and equivalent variation, and

compensating and equivalent surplus (see Kolstad 2011). Compensating surplus, which is a

special case of compensating variation, is more appropriate for stated preference methods, as

used in the present study, because the response elicited in the present survey is one in which

utility is assumed held constant given a change in income, and the good being valued is a case of

“restricted demand” (see Kolstad 2011).

Compensating surplus can be defined as the monetary compensation (positive or negative) needed in order to return and individual to his original level of wellbeing (or, “utility” in economics jargon) after the quantity change occurs. Figure 5 shows compensating surplus graphically where Y on the vertical axis represents income, or equivalently, consumption of all other market goods, and q on the horizontal axis represents quantity of the environmental good.

Suppose a person has income Y0 and the current quantity of the environmental good is q0. This

person is then at point A and has a utility of U0. Then there is an increase in the quantity of the environmental good q from q0 to q1. This change moves consumption from point A to point B and raises the person’s utility to U1. An indifference curve shows the locus of points that give a

43 person the same level of well-being or utility. IU0 is an indifference curve with initial utility of

U0, and IU1 is a new indifference curve with new, higher utility of U1. In figure 3, compensating surplus is the difference between income levels Y0 and Y1 because, if, from consumption point B, you take away this amount of income, the person will once again be at his initial utility level, U0, at consumption point C. In this case, the compensating surplus represents the individual’s maximum willingness to pay (WTP) to obtain the quantity change from q0 to q1.

Income (Y)

A B Y0

CS (=WTP) IU1 C Y1 IU0

q0 q1 Quantity (q)

Figure 5. Compensating surplus

Contingent valuation

The binary-choice version of our valuation question is commonly referred to as a

“contingent valuation” question. Contingent valuation (CV) is a stated-preference survey-based

economic technique to value non-market goods. It is a stated-preference technique because it

presents respondents with a proposed program and asks them to directly state their preferences,

in particular, which program, if any, they prefer be implemented. In contrast, revealed-preference studies use choices people have already made which indirectly reveal their preferences (see

Haab and McConnell 2002). Contingent valuation can be used to estimate individuals’ willingness to pay for changes in environmental quantity or quality (Haab and McConnell 2002).

Ciriacy-Wantrup (1947) first proposed the contingent valuation method to estimate the values of non-market goods (Carson, Meade, and Smith 1993). Later, Davis (1963) implemented the contingent valuation method to measure the benefits of a recreational area in

Maine. In the 1970s, many researchers and economists used the CV method to measure the value of different recreational areas. One of the most well-known early CV studies was conducted by Randall, Ives, and Eastman (1974) to measure the value of improved air quality, and this study was published in the first volume of the Journal of Environmental Economics and

Management. Thereafter, the CV method was developed and improved and was popularly used in various environmental economic studies.

However, the CV method was subject to debate by economists and other researchers regarding whether it was a reliable method to estimate the value of environmental goods. The

Exxon Valdez oil spill in 1989 was a challenge and developmental opportunity for the CV method in order to assess its validity. By way of the Exxon Valdez oil spill, there grew a controversy between economists on the CV method. Passive use values are values people place

on an environmental good even though they do not directly consume the good. Examples might

include the value of experiencing the pristine natural environment of Prince William Sound, or

the value of lost wildlife to people who never plan to visit the sound. Diamond and Hausman

(1994) and Desvousges et al. (1993) argued that CV is not an appropriate method for measuring

passive-use values. On the other hand, Randall (1993), Hanemann (1994), and Portney (1994) argued that CV is a valid and useful approach. The National Oceanic and Atmospheric

Administration (NOAA) appointed a panel of six expert economists, Kenneth Arrow Robert

Solow, Edward Leamer, Paul Portney, Roy Randner, and Howard Schuman, in order to refine

previous CV studies (Arrow et al. 1993). The NOAA panel concluded that “CV studies can

produce estimates reliable enough to be the starting point for a judicial or administrative

determination of natural resource damage including lost passive-use value” (p. 4610). Moreover, the NOAA panel provided guidelines for the proper use of the method that are still followed today.

Contingent Valuation is a special case of a choice experiment (CE). Note here that the multinomial-choice version of our valuation question is commonly referred to as a “choice experiment” question. In a CE, respondents are presented with two or more competing programs which have varying “attributes” – for example, the amount of wetlands restored, or the time to completion, or the cost to the respondent. There is often a “no action” or “implement none” option, and respondents are asked to indicate their most preferred option. Furthermore, they may

make more than one such choice, where each “choice set” presents the respondent with different

options from which to choose. For example, a respondent may choose his most preferred of

three proposed programs under eight different choice sets. Choice experiments have a number of advantages over CV studies, in particular that they glean more information about the

46 respondents’ preferences, which often increases the statistical efficiency of value estimates.

Also, they can be used to value not only the implementation of a particular project, but also small changes in the different attributes of the project (Adamowicz et al. 1998).

Consequentiality

The usefulness of responses to surveys on hypothetical referenda, in particular, standard contingent valuation (CV) surveys, has long been debated. Carson and Groves (2007, 2011) argue that as long as the survey question is consequential we can predict how agents should respond, given their incentive structure. Their work has largely shifted the debate on stated- preference methods from the issue of “real versus hypothetical” to the issue of “consequential versus inconsequential”. A survey question is consequential if the agent believes his response will affect some outcome that he cares about. From such questions we can expect “useful information” (Carson and Groves 2007, p.183). The usual desirable property of a standard CV question, however, is the more restrictive incentive compatible property. An agent’s dominant strategy in responding to an incentive-compatible question is to tell the truth. Carson and Groves

(2007) point out that a standard, dichotomous choice, advisory survey CV question is incentive compatible if it is consequential and the respondent can be compelled to pay if the public good is provided.4

4 An additional condition is that only one service or public good is provided and that there are no other ways to provide the good. For example, if respondents believe they will be given a different set of choices if they give a “no” response, they may vote “no” despite actually preferring project implementation. Or, if respondents believe the good will be provided using existing tax revenue, incentive compatibility is lost. However, in this paper, we consider only

Several experiments have supported the idea that a referendum with payment need only

have a positive probability of being binding to yield a similar willingness to pay estimate to that

of a fully-binding referendum. In a controlled field experiment, Landry and List (2007) find that

sports memorabilia collectors’ responses to a dichotomous choice willingness-to-pay (WTP)

question in which the probability of the votes being binding is 50% are statistically

indistinguishable from responses to the same question in which votes are binding with 100%

certainty. Vossler and Evans (2009) find similar results when comparing real and advisory

referenda in an experimental setting: when participants regard their vote as consequential, no

elicitation bias is observed, but when viewed as inconsequential, elicitation bias is present.

Related to these findings, Mitani and Flores (2010) provide predictions as to the direction of

these effects: they suggest that the probability of provision has a positive effect, whereas the

probability of payment has a negative effect, on contributions in a lab setting with open bidding.

Landry and List (2007) emphasize that it is critical to “[ensure] that survey respondents

view the instrument as consequential” (p.427), yet they are surprised to find that there are no

contingent valuation studies of which they are aware which explicitly elicit agent perceptions of

the consequences of the survey.5 Since then, there have been two such studies. Herriges et al.

single issue referenda, and assume respondents do not substitute alternative policies for the one presented in the survey.

5 Landry and List (2007) do not explicitly define consequentiality although one can infer that

they mean something like “one’s response will affect the probability that the good is provided

and he has to pay.” It can easily be shown that such a definition has equivalent implications to

the two conditions for incentive compatibility proposed by Carson and Groves (2007) if it is

assumed that one cares whether or not he has to pay.

(2010) directly ask respondents about their perceptions of the consequences of their survey.

They find similar willingness to pay distributions among respondents who find the survey to be

at least minimally consequential, whereas respondents who believed the survey to be

inconsequential have statistically lower willingness to pay distributions. Unfortunately, the

Herriges et al. (2010) study did not have any data on the second necessary condition for

incentive compatibility: whether or not respondents believed they would have to pay for the good

if it were provided.6 Vossler, Doyon, and Rondeau (2012) conduct a field experiment and find that, as long as participants believed there would be more than a “weak” impact on policy, stated and real willingness-to-pay functions were statistically identical. Furthermore, they find that

willingness to pay does not necessarily depend on whether an incentive-compatible decision rule

is specified.

Further work on the effects of consequentiality were conducted by the present authors in

Interis and Petrolia (2012), which focused on the binary-choice setting, and Petrolia and Interis

(2012), which focused on the multinomial-choice setting. Petrolia and Interis (2012) concluded that a respondent who believes neither that his response will affect the outcome of the survey, nor that the survey will affect future policy is the least likely type of respondent to vote in favor of the proposed restoration program. They find that those who believe in both are the most

6 Herriges et al. (2010) refer to this condition as payment consequentiality. Furthermore, what

Herriges et al. refer to as policy consequentiality is what Carson and Groves (2007) refer to simply as consequentiality. They evidently do so to distinguish this from what they call payment consequentiality.

likely, and those who believe in one or the other are in between. They conclude that proper

handling of these respondents is critical for value estimation.

Petrolia and Interis (2012) find that that behavior of respondents for whom the valuation

survey is fully consequential is consistent with theory, but not so for respondents for whom the

valuation survey is less-than-fully consequential. In particular, although price effects are

significant and negative across consequentiality types, non-price choice attributes are either not significant, but with expected sign, as in the case of partially-consequential types, or in some cases, significant with the wrong sign among not-consequential types. Welfare estimates are found to be statistically equivalent between fully- and partially-consequential types, but statistically different for not-consequential types.

Based on the findings in the literature that the behavior of non-consequential respondents is not necessarily consistent with economic theory and empirically not consistent with consequential respondents, we segment results into two sets: our preferred results, based on the sample that excludes such respondents, and, for comparison and completeness, the sample that includes them.

CHAPTER VII

RANDOM UTILITY AND ECONOMETRIC MODELS

The Random Utility Model

The random utility model is a well-known model to analyze discrete stated preference responses. Hanemann (1984) was the first to develop a basic model for the random utility, and

McFadden (1974) utilized a framework for the random utility. For simplicity, we discuss the random utility model in a binary (two) choice setting in which the respondent indicates whether he is for or against implementation of the proposed program. The multinomial-choice setting is a generalization of this form. This discussion follows Haab and McConnell (2002). In the binary

case there are two choices or alternatives, either “for” or “against” the proposed program. Let

indirect utility for respondent j be written

uij= uy i(,,,) j zxjiεij

th where i =1 if the program is implemented, and i = 0 is for the status quo. y j is the j

respondent’s discretionary income, and z j is an m-dimensional vector of respondent

characteristics, xi is a vector of choice-specific attributes, andεij is a component of preference

known by the individual respondent but not observed by the researcher. Based on this model,

respondent j will answer yes to a program with required payment of t j if utility with the program, minus the payment, exceeds utility of the status quo:

10 uy1(jj−> t ,,,zxjjεε10 j ) uy (,,, jzxjj 0 j )

Because a random part of preferences is unknown, only a probability statement about yes or no

can be made. The probability of a yes response is the probability that the respondent believes

that he is better off if the proposed program is implemented and he makes the required payment,

so that uu10> . For respondent j, this probability is

10 Pr(yesj )= Pr( u1 ( yjj − t ,zxji , ,εε10 j ) >−u ( y jj t ,zxji , ,0 j ))

Two modeling decisions are needed to estimate the model. First, the functional form of utility

must be specified. Second, the distribution of εij must be specified. All approaches clearly

identify that the indirect utility function be additively separable in deterministic (v) and random parts:

uyi(,,,) j zxjiεεij= vy i (,,) j zxji+ ij

Using the additive specification of the equation, the probability of respondent j becomes

10 Pr(yesj )= Pr( v1 ( yjj − t ,zxji , ) +>εε10j )v ( y j ,zxji , ) +0j ))

This also can be written as

= −10 − >−εε Pr(yesj ) Pr v1 ( yjj t ,zxji , ) v0 ( y j , zxji , )01j j )

However, the differences in the random components between the status quo and the proposed

program cannot be identified. Therefore, the random term can be written as εεεij≡−10 j j , a

single random term. Then let Faε () be the probability that the random variable εij is less than a.

Therefore the probability of a yes is

=−−−10 − Pr(yesj ) 1 Fε  ( v10 ( yjj t ,zxji , ) v ( y j , zxji , ))

At this point, a more specific indirect utility function is needed for estimation. For example, the

linear indirect utility function, which is the simplest and most commonly estimated function, is

specified as follows. The linear indirect utility function results when the deterministic part of the

preference function is linear in income and covariates

vyij(,,) j zxj i =+++δβi iy j αzij γx ii

where δi is a coefficient on a constant utility term, αi is an m-dimensional vector of parameters,

m so that = α . The deterministic utility for the proposed program is then αzij ∑ k =1 ikz jk

1 1 vyt1jjj(− ,,)zxj i =+δβ11 (ytjj −+ )αz1j + γx 1i

th where t j is the price offered to the j respondent. The status quo utility is

00 vy0jj(,,)zxj i =+++δβ00yjαz0j γx 0i

The change in deterministic utility is

10 vv1jj− 0 =−+δδ 10() ββ 1 − 0yj − β 0 t j + ()α1 − α 0 z j + γx 1i − γx 0i

If one assumes that the marginal utility of income and the marginal utility of the environmental

good attributes are constant between the two states, then ββ10= , γγγ10= = , and the utility

difference becomes

10 vv10jj−=−+δβ t jαzj + γ() x ii − x

where αα=10 − αand δδδ=10 − . With the deterministic part of preferences specified, the

probability of responding yes becomes

10 Pr(yes jj )= Pr(δβ −t +αzj + γ( x ii − x ) +>εj 0)

where εεεjjj≡−10 as defined already.

Econometric (Multiple Regression) Analysis

Binary-Choice Model

Recall from above that the probability of a yes vote can be expressed as

10 Pr(yes jj )= Pr(δβ −t +αzj + γ( x ii − x ) +>εj 0) .

We know that 10 10 Pr(δβ−++ttjαzj γ( x ii −+>=−−++ x )ε jj 0) Pr( ( δβ αzj γ( x ii − x )) <ε j ) 10 =−−− 1 Pr( (δβt jj +αzj + γ( x ii − x )) >ε ) = Pr(δβ −++t αz γ()) x10 −> x ε jjj ii

The last equality exploits the symmetry of the distribution. For symmetric distributions 2 2 Fx()=−− 1 F ( x ). Suppose that εσj ~N (0, ) . If we convert εσj ~N (0, ) to a standard normal (N(0,1)) variable, and let θ= εσ/ , then θ ~N (0,1) and

δβ−++t αz γ() x10 − x δβ−++10 −>=ε j j ii>θ Pr( t jjαzj γ( x ii x ) ) Prj σ δβ−++t αz γ() x10 − x = Φ j j ii  σ

Where Φ()x is the cumulative standard normal, i.e., the probability that a unit normal variate is less than or equal to x. This is the probit model. The probit model was used to estimate the

binary-choice model.

Multinomial-Choice Model

For the multinomial-choice case, let the probability of choosing alternative j be

** exp(xit γ ) Pr(yji = ) = J ** ∑exp(xit γ ) i=1

* * where xit and β are defined as follows. Let the deterministic component of the random-utility

model be expressed as:

vii=x β + ( zAi )'

=xiJβz +⊗ (i I )vec( A ') β = (,xziJi ⊗ I ) vec(A ') ** = xi β

Where β is a p x 1 vector of alternative-specific coefficients and A = (αα1 ...J ) is a q x J matrix of

individual-specific coefficients. It is necessary to fix one of the α j to the constant vector to

normalize the location. Here, I J is the J x J identity matrix, vec( ) is the vector function that

creates a vector from a matrix by placing each column of the matrix on top of the other, and ⊗

is the Kronecker product. This is McFadden’s Alternative-Specific-Constants model in Stata, which was used to estimate all multinomial-choice models (Statacorp 2011). All models were estimated using Stata 12 (StataCorp 2011).

CHAPTER VIII

RESULTS

Tables 24 and 25 display descriptions of the variables included in the analysis. The

reader will notice that the low levels of the project attributes are not displayed in the Table 25.

In order for the multinomial choice models to be estimable, one level of each attribute must serve

as the “base”, and be omitted from the regression equation. Thus the estimates on the non-

omitted variable levels are interpreted as effects relative to the omitted base level. The choice of

which level of each attribute to omit is arbitrary and we chose to omit the lowest level of each

attribute.

Table 26 displays the means and standard errors of the individual-specific variables for both the binary choice and multinomial choice models. Note that the number of observations is lower than that reported in the earlier chapter, due to item non-response. For each model, the means and standard deviations for both the complete sample and consequential respondents only are shown. The means of several of the variables are similar across the full sub-sample for each model and the consequential respondents only – BTNE visitor / resident, non-taxpayer, income, head of household, age, minority, male, and politically conservative. Others indicate that certain respondents might be more likely to believe the survey to be consequential, including those who have greater confidence in federal and state government to implement the projects, those who had followed the Deepwater Horizon Oil spill more closely, and those who had made more major changes in their lifestyle for environmental reasons are all more likely to believe the survey to be consequential. If respondents believed the survey did not provide enough information to make an informed choice, they were less likely to perceive the survey as consequential.

Table 27 reports the parameter estimates for the binary-choice version of the survey. The significant variables are largely the same across the two sub-samples, and so we focus on the results of the consequential-respondents-only model. Importantly, the coefficient on bid is negative and significant. This indicates that the higher the bid price the respondent must pay to implement the project, the less likely he is to vote for its implementation. Resource users, defined as those who have visited or live in the BTNE, are significantly more likely to vote in favor of the proposed restoration, and this accounts for the largest single effect of one variable on the probability of a yes vote: resource users are 16 percent more likely to vote yes. Respondents with greater confidence in federal and state governments are also more likely to be in favor of the project at 8 and 10 percent, respectively. People who rate themselves relatively more conservative are less likely to be in favor of the project (7 percent for a one-unit change in

political rating). People who have made greater changes to their lifestyle for environmental

reasons are more likely to be in favor of the project (8 percent for a one-unit change). Regarding

demographic indicators, age and head-of-household are also significant.

Table 28 shows the parameter estimates for the multinomial-choice version of the survey.

The significant individual-specific variables are similar to those in the binary-choice model, with two key exceptions: the first is that the BTNE visitor / resident variable is not significant, indicating no statistically-significant difference in the voting behavior of this group relative to resource non-users; and second, minorities and males are statistically less likely to vote for a program in the multinomial-choice setting.

The coefficients on the alternative-specific variables are all highly significant for the consequential respondents. These coefficients indicate whether the respondent is more likely to vote for an option with the specified attribute level than an option with the lowest level attribute,

all else equal. Furthermore, we should expect the coefficients on the highest-level attributes to

be greater in magnitude than those on the intermediate level attributes because, all else equal, we

would expect a high level of the attribute to increase the probability that the respondent chooses

that option more than an intermediate level of the attribute. This is true in the consequential- respondents-only results with the exception of the storm surge protection attribute. For storm surge protection, the coefficient on the high level is equal to that of intermediate level. This indicates that an intermediate or high level of this attribute increases the likelihood of the respondent voting for the program over the lowest level of these attributes, but that respondents do not derive additional utility beyond the intermediate level.

Increased fisheries productivity has the largest marginal effect on project choice (13 and

14 percent for intermediate and high levels, respectively), followed by increased storm surge protection (10 percent), and increased wildlife habitat (7 and 9 percent, respectively). Among individual-specific characteristics, confidence in Louisiana state government has the largest effect at 6 percent.

Comparing the results of the consequential-respondents-only and all-respondents, we find that the significant individual-specific variables are largely the same across the two sub-samples.

However, more substantial differences are found for the choice-specific attributes. Although the coefficient signs, significance, and relative magnitudes are as expected for the consequential- respondents only model, the high level of wildlife habitat is only marginally significant and the high level of storm protection is not significant for the all-respondents model. This result highlights the fact that including respondents who do not find their responses to be consequential can lead to some counterintuitive results; it would be odd if people truly were more likely to vote for a program if it had the intermediate level of the attribute provided but not if it provided the

highest level of the attribute. Thus, we have less confidence in the validity of these results given

that we have less confidence in the responses of individuals that do not perceive the survey as

consequential (i.e., as meaningful).

Table 29 reports the estimated WTP for the proposed program based on the binary-choice

results.7 The numbers in brackets show the 95% confidence intervals. Recall that the binary- choice version proposed a program fixed at the intermediate levels of the attributes. We report the resource user and non-user specific WTP estimates along with the sample weighted mean

WTP. Given that the sample is comprised of approximately 92 percent non-users, the weighted mean is closely aligned with the resource non-user estimates. Resource user WTP for the consequential-respondents-only subsample is estimated at $3,125 per household (with a confidence interval of $2029-$4825), approximately twice that of resource non-users (mean of

$1,637 with confidence interval of $1271-$2242). Overall, the sample-weighted mean WTP is estimated at $1,751 per household, with a confidence interval of $1382-$2396. For comparison, we also report an alternative estimate based on the Turnbull Lower Bound method (see Turnbull

1976; Cosslett 1982; Ayer et al. 1995).8 The Turnbull estimates are substantially lower,

hovering closer to $1000 per household.

Table 30 reports the estimated WTP for the proposed program based on the multinomial-

choice results. For the multinomial-choice survey, it is possible to derive value estimates for a

7 The mean WTP estimate for a given model is calculated by multiplying each coefficient

(except that of price) by its sample mean (setting the “mean” of the intercept term to one), then

dividing each by the coefficient on price. These are then summed to yield the sample mean

WTP.

program at various attribute levels. We report the value estimates for a program with all of the

attributes at the lowest level, all of the attributes at the intermediate level, and all of the attributes

at the highest level. We report estimates for both resource users and non-users, but keep in mind

that the BTNE visitor / resident variable was not significant in this model, and so these

difference should not be viewed as statistically different. Estimated WTP for resource users is

$524, $971, and $1,018 per household for the proposed project at the low, intermediate, and high

attribute levels, respectively, just slightly above those of resource non-users. The estimates based on the all-respondents model are slightly lower.

The estimates for the binary-choice model can be directly compared to those of the multinomial-choice model for the intermediate scale program. Although the regression-based

binary-choice model estimates are substantially higher, the Turnbull estimates are fairly consistent with the multinomial-choice model based estimates.

As mentioned earlier, one of the advantages of a multinomial-choice survey is that it is possible to derive value estimates for incremental changes in the attribute levels. This allows the analyst to identify the specific contribution to overall WTP of a particular attribute, and to identify the relative importance of the various attributes. The bottom half of Table 30 shows

these value estimates. The willingness to pay values indicate how much a household is willing

to pay for the specified level of the attribute relative to the lowest level of the attribute.

Comparing across attributes, results indicate that increases in fisheries productivity make the

largest contribution to overall WTP, followed by improvements in storm protection, followed by

increases in wildlife habitat. Thus, we estimate that respondents are willing to pay an average of

$189 per household for an increase in fisheries productivity from the low level to the

intermediate level, and $204 per household for an increase from the low level to the high level,

60 all else equal. These results also imply the WTP for an increase from the intermediate to the high level of fisheries productivity: $204 - $189 = $15. Similarly, WTP for an increase in storm surge protection to the intermediate level is estimated at $149, but WTP for a further increase is just an additional $2. Finally, WTP for an increase in wildlife habitat to the intermediate level is

$109 per household, and an additional $30 for a further increase to the high level. Thus, results indicate that although respondents are willing to pay additional dollars for improvements in wildlife habitat beyond the intermediate level, they do not appear to be willing to pay much, if anything, for improvements in either storm protection or fisheries productivity above and beyond the intermediate level.

Table 24. Multiple regression model: dependent and individual-specific independent variable names and descriptions Variable Name Type Description Dependent Variable Vote Binary = 1 if vote for alternative, = 0 otherwise Individual-specific Variables BTNE Visitor / Resident Binary = 1 if visited or resides in BTNE, = 0 otherwise Non-taxpayer Binary = 1 if did not file 2010 federal tax return Income Ordered Cat. Household income; 19 categories, ranging from = 1 (Less than $5,000) to 19 ($175,000 or more) Head of Household Binary = 1 if respondent is head of household, = 0 otherwise Age Continuous respondent age in years Minority Binary = 1 if minority race, = 0 otherwise Male Binary = 1 if male, = 0 otherwise Confidence in Fed Gov. Binay = 1 if has at least some confidence in federal agencies to carry out project, = 0 otherwise Confidence in LA Gov. Binary = 1 if has at least some confidence in Louisiana state agencies to carry out project, = 0 otherwise Politically Conservative Ordered Cat. political preference, ranging from 1 (very liberal) to 7 (very conservative) Oilspill Binary = 1 if followed DWH oil spill at least somewhat closely, = 0 otherwise Green Ordered Cat. = -1 if has made no changes in behavior for environmental reasons, = 0 if minor changes, = 1 if major changes

Table 25. Multiple regression model: alternative-specific independent variable names and descriptions Variable Name Type Description Bid Continuous offered project cost, in dollars Wildlife Habitat- Intermediate* Binary = 1 if wildlife habitat attribute level specified as "50% of restored land suitable as habitat", = 0 otherwise Wildlife Habitat - High* Binary = 1 if wildlife habitat attribute level specified as "75% of restored land suitable as habitat", = 0 otherwise Storm Protection - Intermediate* Binary = 1 if storm protection attribute level specified as "Improved protection for 30% of residents", = 0 otherwise Storm Protection - High* Binary = 1 if storm protection attribute level specified as "Improved protection for 50% of residents", = 0 otherwise Fisheries Productivity - Intermediate* Binary = 1 if fisheries productivity attribute level specified as "15% higher harvest levels", = 0 otherwise Fisheries Productivity - High* Binary = 1 if fisheries productivity attribute level specified as "30% higher harvest levels", = 0 otherwise * Appears in multinomial-choice model only

Table 26. Mean and standard errors of regression model individual-specific variables. Binary-choice Sample Multinomial-choice Sample Consequential Consequential Respondents Respondents Only All Respondents Only All Respondents N = 652 N = 959 N =1048 N = 1518 Std. Std. Std. Std. Variable Name Mean Dev. Mean Dev. Mean Dev. Mean Dev. Vote (dep. variable) 0.67 0.470 0.59 0.49 * * * *

Bid 657.69 763.86 673.04 770.84 * * * *

BTNE Visitor / Resident 0.08 0.27 0.06 0.24 0.08 0.27 0.08 0.26

Non-taxpayer 0.10 0.30 0.11 0.31 0.12 0.33 0.11 0.32

Income 12.30 4.21 12.34 4.24 12.28 0.33 12.52 4.32

Head of Household 0.83 0.37 0.82 0.38 0.81 0.39 0.82 0.38

Age 48.76 16.92 48.49 16.65 48.95 17.13 48.88 16.65

Minority 0.23 0.42 0.22 0.42 0.22 0.42 0.21 0.40

Male 0.50 0.50 0.51 0.50 0.48 0.50 0.50 0.50

Confidence in Fed Gov. 0.41 0.49 0.36 0.48 0.48 0.50 0.42 0.49

Confidence in LA Gov. 0.56 0.50 0.49 0.50 0.58 0.49 0.49 0.50

Politically Conservative 4.15 1.49 4.21 1.50 4.13 1.50 4.20 1.52

Oilspill 0.91 0.28 0.89 0.32 0.90 0.30 0.89 0.31 Green 0.06 0.56 -0.04 0.58 0.06 0.58 -0.004 0.59 * Details given in Table 19.

Table 27. Multiple Regression Probit Model Results for Binary-choice Valuation Data Consequential Respondents Only All Respondents N = 652 N = 959 Std. Marg. Std. Marg. Coef. Err. Effect Coef. Err. Effect Bid -0.0004 *** 0.0007 -0.0001 -0.0004 *** 0.0006 -0.0001 BTNE Visitor / Resident 0.61 *** 0.23 0.16 0.54 *** 0.19 0.16 Non-taxpayer 0.22 0.21 0.06 0.05 0.15 0.02 Income -0.01 0.01 -0.003 -0.01 -0.01 -0.004 Head of Household -0.36 ** 0.18 -0.10 -0.23 * 0.13 -0.07 Age 0.01 *** 0.003 0.003 0.008 *** 0.003 0.003 Minority 0.15 0.14 0.04 0.19 * 0.11 0.06 Male -0.01 0.11 -0.002 -0.04 0.09 -0.01 Confidence in Fed Gov. 0.26 ** 0.13 0.08 0.43 *** 0.10 0.14 Confidence in LA Gov. 0.37 *** 0.12 0.11 0.38 *** 0.09 0.13 Politically Conservative -0.22 *** 0.04 -0.07 -0.19 *** 0.03 -0.06 Oilspill 0.30 0.19 0.09 0.18 0.14 0.06 Green 0.28 *** 0.10 0.08 0.26 *** 0.08 0.09 Constant 0.89 *** 0.34 0.72 *** 0.27 Log-likelihood Value -343.91 -550.84 Likelihood Ratio Chi-sq(12) 138.93*** 199.24*** McFadden's Pseudo R-sq 0.17 0.15 ***, **, * indicates statistical significance at the p = 0.99, 0.95, and 0.90 levels, respectively.

Table 28. Multiple Regression Conditional Logit Model Results for Multinomial-choice Valuation Data Consequential Respondents Only All Respondents N = 1048 N = 1518 Coef. Std. Err. Marg. Eff. Coef. Std. Err. Marg. Eff. Alternative-specific Variables Bid -0.003 *** 0.0004 -0.001 -0.002 *** 0.0003 -0.001 Wildlife Habitat: Intermediate 0.30 *** 0.11 0.07 0.27 *** 0.09 0.07 Wildlife Habitat: High 0.38 *** 0.14 0.09 0.21 * 0.11 0.05 Storm Protection: Intermediate 0.41 *** 0.10 0.10 0.37 *** 0.08 0.09 Storm Protection: High 0.41 ** 0.16 0.10 0.15 0.14 0.04 Fisheries Productivity: Intermediate 0.52 *** 0.13 0.13 0.53 *** 0.11 0.13 Fisheries Productivity: High 0.56 *** 0.15 0.14 0.47 *** 0.13 0.12 Individual-specific Variables BTNE Visitor / Resident 0.18 0.30 0.01 0.13 0.27 0.01 Non-taxpayer -0.05 0.20 -0.002 0.001 0.24 0.0001 Income 0.03 0.01 0.001 0.02 0.02 0.001 Head of Household -0.49 * 0.23 -0.02 -0.23 0.20 -0.02 Age 0.02 *** 0.19 -0.001 0.01 ** 0.005 0.001 Minority -0.58 ** 0.22 -0.03 -0.32 * 0.18 -0.02 Male -0.59 *** 0.20 -0.03 -0.41 *** 0.14 -0.03 Confidence in Fed Gov. 0.57 ** 0.07 0.03 0.52 *** 0.16 0.04 Confidence in LA Gov. 1.11 *** 0.29 0.06 1.03 *** 0.15 0.07 Politically Conservative -0.04 *** 0.17 -0.02 -0.40 *** 0.05 -0.03 Oilspill 0.30 0.33 0.02 0.28 0.21 0.02 Green 0.39 ** 0.02 0.52 *** 0.12 0.04 Constant (Alt A) 1.13 * 0.51 1.10 ** 0.43 Constant (Alt B) 0.99 * 0.52 1.01 ** 0.44 Log-likelihood Value -923.23 -1409.56 Wald Chi-sq(18) 206.70*** 316.91*** ***, **, * indicates statistical significance at the p = 0.99, 0.95, and 0.90 levels, respectively. 66

Table 29. Estimated Means and Confidence Intervals (in brackets) of Willingness to Pay (WTP)based on Binary-Choice Results Consequential Respondents Only All Respondents $3,125 $2,710 Resource Users* [2029, 4825] [1618, 4181] $1,637 $1,184 Resource Non-Users [1271, 2242] [894, 1592] $1,751 $1,281 Weighted Mean [1382, 2396] [989, 1708] $1026 $973 Non-parametric Turnbull** [955, 1096] [916, 1031] * BTNE Visitors and Residents ** Provided for comparison; not based on regression results

Table 30. Estimated Means and Confidence Intervals (in brackets) of Willingness to Pay (WTP) based on Multinomial-Choice Results Overall WTP Consequential Respondents Only All Respondents

Low Intermediate High Low Intermediate High $524 $971 $1,018 $388 $911 $757 Resource Users* [241, 877] [673, 1376] [719, 1393] [148, 679] [643, 1282] [470, 1038] $457 $904 $951 $331 $854 $700 Resource Non-Users [319, 662] [724, 1181] [795, 1150] [216, 476] [692, 1086] [534, 855] $463 $909 $956 $335 $858 $704 Weighted Mean [321, 664] [732, 1185] [800, 1156] [220, 479] [696, 1093] [542, 860] WTP for Attribute Increments (Relative to Low Levels) Intermediate High Intermediate High $109 $139 $121 $92 Wildlife Habitat [37, 184] [47, 212] [46, 203] [-7, 172] $149 $151 $165 $68 Storm Protection [83, 225] [44, 246] [99, 245] [-61, 165] $189 $204 $237 $210 Fisheries Productivity [97, 309] [106, 310] [141, 359] [111, 315] * BTNE Visitors and Residents

CHAPTER IX

SUMMARY AND CONCLUSIONS

Wetland loss is occurring at a rapid rate all around the United States, but one of the

fastest rates of loss is occurring in the Barataria-Terrebonne National Estuary, south of New

Orleans. Coastal wetlands provide many benefits to humans and the natural world including protection from storms, breeding and spawning grounds for many important commercial and recreational fish species, habitat for birds, small mammals, and other wildlife, and a location for fishing, birding, and other recreational activities. Several wetland restoration and preservation programs have been proposed and some have been implemented in coastal Louisiana over the years; however, they have generally been of a much smaller scale than the programs proposed in our survey. Our proposed programs involved restoring roughly fifty percent of the land lost in the area since 1956, an area of roughly 375 square miles.

The programs we proposed differed by the percentage of restored land that would be suitable for wildlife habitat, the percentage of residents in the area who would receive improved protection from storm surge, and the percentage increase in harvest of key Gulf of Mexico commercial fish species. For an intermediate-scale program in which 50% of restored land was suitable for wildlife, 30% of respondents received improved storm protection, and harvest levels increased by 15%, we estimated that the average U.S. household is willing to pay between $909

(the weighted mean taken from the consequential-only multinomial-choice results) and $1,751

(the weighted mean taken from the consequential-only binary-choice results), with resource- using households willing to pay substantially more.

Our study involved the use of a multinomial choice survey, one of the advantages of

which is that it can be used to estimate the values of changes in various attributes of the program.

We found that the largest share of total willingness to pay for the program came from the

desirability of increases in fisheries productivity (valued at between roughly $100 and $360 for

the intermediate-scale program), followed by the desirability of protection from storms ($80 to

$245), followed by the desirability of wildlife habitat ($35 to $210).

In addition to being a resource user, several other factors increase the probability that a

respondent is willing to pay to implement the proposed program. Older respondents,

respondents who considered themselves more politically liberal, respondents who had made

more lifestyle changes in the past for environmental reasons, and respondents who had greater

confidence in federal and Louisiana State governments to implement the programs were more

likely to vote in their favor.

Before closing, we wish to provide some simple comparisons to estimates in the existing

literature as well as to estimated costs of restoration. First, we find that our value estimates are in

line with those in the existing literature, keeping in mind that value estimates vary widely (see

Woodward and Wui 2001). Our estimates correspond to a value of roughly $0.01 per person per

acre of wetland. Putting estimates into January 2012 dollars, Bergstrom et al. (1990) and

Petrolia, Moore, and Kim (2011) find estimates lower than ours ($0.0002 and $0.001,

respectively) whereas Udziela and Bennett (1997) and Bauer, Cyr, and Swallow (2004) find

estimates higher than ours ($1.25 and $0.60, respectively).

Given that our estimates are consistent with those found by other researchers in

independent studies, we wish next to provide estimates of total value of the proposed restoration.

Using the U.S. Census Bureau (2011) one-year estimate of the number of households in the U.S.

in 2011 (114,991,725), this represents a total value of the intermediate-scale program of between

$105 billion and $201 billion.9 The lower of these values still exceeds the most recent, and by far the largest, estimate of restoration cost of $100 billion (Graves 2009). Furthermore, our

estimate, strictly speaking, applies only to our study area, the Barataria-Terrebonne National

Estuary, which comprises only a fraction of the total Louisiana coast on which the above cost estimate is based. Thus, even if one discounts our estimates further for a variety of reasons, it is difficult to argue that the benefits do not justify the costs of restoration. Thus, our results certainly give credence to the claim that Louisiana is “America’s Wetland”.

9 Alternatively, one could use the number of U.S. tax returns filed in 2010, but this number is

slightly larger, at 142,823,000 (U.S. Census Bureau 2012), so the estimated total WTP would be

even larger.

REFERENCES

Adamowicz, W., Boxall, P., Williams,M., and Louviere, J. 1998. “Stated Preference Approaches for Measuring Passive Use Values: Choice Experiments and Contingent Valuation.” American Journal of Agricultural Economics 80(February): 64-75.

America’s Wetland Foundation. 2012. Website: http://www.americaswetland.com/. Last accessed March 20, 2012.

Arrow, K., R. Solow, P. Portney, E. Leamer, R. Radner, and H. Schumar. 1993. “Report of NOAA Panel on Contingent Valuation.” Federal Register 58:4601-4614.

Ayer, M., H.D. Brunk, G.W. Ewing, W.T. Reid, and E. Silverman. 1955. “An Empirical Distribution Function for Sampling with Incomplete Information.” Annals of Mathematical Statistics 26: 641-647.

Barras, J.A., S. Beville, D. Britsch, S. Hartley, S. Hawes, J. Johnston, P. Kemp, Q. Kinler, A. Martucci, J. Porthouse, D. Reed, K. Roy, S. Sapkota, and J. Suhayda. 2003. “Historical and projected coastal Louisiana land changes: 1978-2050.” USGS Open File Report 03-334.

Bauer, D. N. Cyr, and S. Swallow. 2004. “Public Preferences for Compensatory Mitigation of Salt Marsh Losses: A Contingent Choice of Alternatives.” Conservation Biology 18 (no.2): 401- 11.

Bergstrom, J., J. Stoll, J. Titre, and V. Wright. 1990. “Economic Value of Wetlands-Based Recreation.” Ecological Economics 2: 129-147.

Brander, L.M., R.J.G.M. Florax, and J.E. Vermaat. 2006. “The Empirics of Wetland Valuation: A Comprehensive Summary and a Meta-Analysis of the Literature.” Environmental & Resource Economics 33: 223-50.

Brouwer, R., I.H. Langford, I.J. Bateman, and R.K. Turner. 1999. “A meta-analysis of wetland contingent valuation studies.” Regional Environmental Change 1: 47-57.

Caffey, R.H., K. Savoie, and M. Shirley. 2003. “Stewardship Incentives for Louisiana’s Coastal Landowners.” Interpretive Topic Series on Coastal Wetland Restoration in Louisiana, Coastal Wetlands Planning, Preservation and Restoration Act. National Sea Grant Library.

Carson, R.T., and T. Groves. 2007. “Incentive and informational properties of preference questions.” Environmental & Resource Economics 37: 181-210.

Carson, R.T. and T. Groves. 2011. “Incentive and Information Properties of Preference Questions: Commentary and Extensions,” in Jeff Bennett, ed., International Handbook of Non- Market Environmental Valuation (Northampton, MA: Edward Elgar, 2011).

Carson, R.T., N.F. Meade, V.K. Smith. 1993. “A Contingent Valuation and passive-use values.” Choices 2nd Qtr: 5-8.

Ciriacy-Wantrup, S.V. 1947. “Capital returns from soil-conservation practices.” Journal of Farm Economics 29: 1181-96.

Coastal Protection and Restoration Authority of Louisiana. 2007. Integrated Ecosystem Restoration and Hurricane Protection: Louisiana’s Comprehensive Master Plan for a Sustainable Coast. Baton Rouge, LA.

Coastal Protection and Restoration Authority of Louisiana. 2010. “About Us.” http://www.lacpra.org/index.cfm?md=pagebuilder&tmp=home&nid=4&pnid=0&pid=2&fmid=0 &catid=0&elid=0. Last accessed March 20, 2012.

Coastal Protection and Restoration Authority of Louisiana. 2012a. Louisiana’s Comprehensive Master Plan for a Sustainable Coast, Draft January 2012. Baton Rouge, LA.

Coastal Protection and Restoration Authority. 2012b. “What’s causing the crisis?” http://coastal.louisiana.gov/index.cfm?md=pagebuilder&tmp=home&pid=118. Last accessed on April 4, 2012.

Coreil, P. and M. Barrett-O’Leary. 2004. “Functions and Values of Wetlands in Louisiana.” Louisiana Agricultural Center, Publication #2519.

Cosslet, S.R. 1982. “Distribution-Free Maximum Likelihood Estimator of the Binary Choice Model.” Econometrica 51: 765-782.

Costanza, R., O. Pérez-Maqueo, M.L. Martinez, P. Sutton, S.J. Anderson, and K. Mulder. 2008. “The Value of Coastal Wetlands for Hurricane Protection.” AMBIO: A Journal of the Human Environment 37: 241-8.

Couvillion, B.R., J.A. Barras, G.D. Steyer, W. Sleavin, M. Fischer, H. Beck, N. Trahan, B. Griffin, and D. Heckman. 2011. “Land area change in coastal Louisiana from 1932 to 2010.” U.S. Geological Survey Scientific Investigations Map 3164, scale 1:265,000, 12 p. pamphlet.

Davis, R. 1963. The Value of Outdoor Recreation: An Economic Study of the Maine Woods. Doctoral Dissertation in Economics, Harvard University.

Desmond, J.M. 2005. “The Dying Louisiana Wetlands.” Janus Head 8(2):485-492.

Desvousges, W.H., A.R. Gable, R.W. Dunford, and S.P. Hudson. 1993. “Contingent valuation: the wrong tool to measure passive-use losses.” Choices, 2nd Qtr: 9-11.

Diamond, P.A., and J.A Hausman. 1994. “Contingent Valuation: Is Some Number Better than No Number?” Journal of Economic Perspectives 8:45-64.

Dunbar, J.B., L. Britsch, and E. Kemp. 1992. “Land loss rates: Louisiana Coastal Plain.” Technical Report GL-92-3, U.S. Army Engineer Waterways Experiment Station, Vicksburg, MS.

Farber, S. 1996. “Welfare Loss of Wetlands disintegration: A Louisiana Study.” Contemporary Economic Policy 14: 92-106.

Graves, G. 2009. “Navigating the Environment - Managing Risks and Sustaining Benefits, Coastal Systems: Navigation, Flood, Ecosystems and protecting Louisiana’s coast.” PIANC Conference, October 28, 2009. Retrieved from www.pianc.iwr.usace.army.mil/docs/NTE2009Conf/16-Graves_PIANC_10-28-09.v2.pdf (Accessed on Oct. 14, 2009).

Greene, W.H. 2012. Econometric Analysis, 7th ed. Boston: Prentice Hall.

Haab, T.C. and K.E. McConnell. 2002. Valuing Environmental and Natural Resources: The Econometrics of Non-Market Valuation. Northampton, MA: Edward Elgar.

Hanemann, M.W. 1984. “Welfare Evaluations in Contingent Valuation Experiments with Discrete Responses.” American Journal of Economics 66:332-341.

Hanemann, M.W. 1994. “Valuing the Environment Through Contingent Valuation.” Journal of Economic Perspectives 8:19-43.

Herriges, J., C. Kling, C.-C. Liu, and J. Tobias. 2010. “What are the consequences of consequentiality?” Journal of Environmental Economics & Management 59: 67-81.

Interis, M.G. and D.R. Petrolia. “Face the Consequentiality: Useless, Useful, and Truthful Responses to Hypothetical Surveys.” Working paper, Department of Agricultural Economics, Mississippi State University. April 2012.

Kazmierczak, R.F. 2001a. “Economic Linkages Between Coastal Wetlands and Habitat/Species Protection: A Review of Value Estimates Reported in the Published Literature.” Staff Paper 2001-04, Natural Resource and Environment Committee, Department of Agricultural Economics & Agribusiness, Louisiana State University.

Kazmierczak, R.F. 2001b. “Economic Linkages Between Coastal Wetlands and Hunting and Fishing: A Review of Value Estimates Reported in the Published Literature.” Staff Paper 2001- 03, Natural Resource and Environment Committee, Department of Agricultural Economics & Agribusiness, Louisiana State University.

Kazmierczak, R.F. 2001c. “Economic Linkages Between Coastal Wetlands and Water Quality: A Review of Value Estimates Reported in the Published Literature.” Staff Paper 2001-02, Natural Resource and Environment Committee, Department of Agricultural Economics & Agribusiness, Louisiana State University.

Kolstad, C.D. 2011. Environmental Economics, 2nd ed. New York: Oxford University Press.

Kuhfeld, W.F. 2010. “Experimental Design, Efficiency, Coding, and Choice Designs.” SAS Technical Papers: Marketing Research. http://support.sas.com/techsup/technote/mr2010c.pdf. Last accessed March 20, 2012.

Landry, C.E. and J.A. List. 2007. “Using Ex Ante Approaches to Obtain Credible Signals for Value in Contingent Markets: Evidence from the Field.” American Journal of Agricultural Economics 89: 420-429.

Louisiana Coastal Wetlands Conservation and Restoration Task Force. 2006. “Coastal Wetlands Planning, Protection and Restoration Act (CWPPRA): A Response to Louisiana’s Land Loss.”

Louisiana Coastal Wetlands Conservation and Restoration Task Force and the Wetlands Conservation and Restoration Authority. 1998. Coast 2050: Toward a Sustainable Coastal Louisiana. Louisiana Department of Natural Resources. Baton Rouge, LA.

Louisiana Department of Natural Resources. 2006. “America’s Energy Corridor: Louisiana Serving the Nation’s Energy Needs.” Technology Assessment Division. Baton Rouge, Louisiana.

Louisiana Department of Wildlife and Fisheries. 2005. “The Economic Benefits of Fisheries, Wildlife and Boating Resources in the State of Louisiana.” Southwick Associates, 34.

Louisiana Oil Spill Coordinator's Office. 2005. “Land and Water Interface for Louisiana from 2002 Landsat Thematic Mapper Satellite Imagery.” Geographic NAD83.

McFadden, D. 1974. “Conditional Logit Analysis of Qualitative Choice Behavior.” in P. Zarmebka, ed., Frontiers in Econometrics. New York: Academic Press 105-142.

Mitani, Y. and N.E. Flores. 2010. “Hypothetical Bias Reconsidered: Payment and Provision Uncertainties in a Threshold Provision Mechanism.” Paper presented at the World Congress on Environmental and Resource Economics, Montreal Canada. http://www.webmeets.com/files/papers/WCERE/2010/1592/Mitani_Flores_HB.pdf.

National Marine Fisheries Service. 2005. “Fisheries of the United States.” United States Department of Commerce, National Oceanic and Atmospheric Administration, Washington, DC.

Petrolia, D.R. and M.G. Interis. “Consequentiality Effects in a Multinomial Choice Experiment Survey.” Working paper, Department of Agricultural Economics, Mississippi State University. February 2012.

Petrolia, D.R. and T. Kim. 2011. “Preventing land loss in coastal Louisiana: Estimates of WTP and WTA.” Journal of Environmental Management 92: 859-65.

Petrolia, D.R., R.G. Moore, and T. Kim. 2011. “Preferences for Timing of Wetland Loss Prevention in Louisiana.” Wetlands 31: 295-307.

Portney, P.A. 1994. “The Contingent Valuation Debate: Why Economists Should Care.” Journal of Economic Perspectives 8:3-17.

Randall, A. 1993. “Passive-use values and contingent valuation-valid for damage assessment.” Choices, 2nd Qtr: 12-15.

Randall, A., B. Ives, and C. Eastman. 1974 “Bidding games for valuation of aesthetic environmental improvements.” Journal of Environmental Economics and Management 2:132- 149.

StataCorp. 2011. “Stata Base Reference Manual.” Volume 3, Release 12. College Station: Stata Press.

Turnbull, B. 1976. “The Empirical Distribution Function with Arbitrarily Grouped, Censored and Truncated Data.” Journal of the Royal Statistical Society 38B: 290-95.

Udziela, M. and L. Bennett. 1997. “Contingent Valuation of an Urban Salt Marsh.” in Restoration of an Urban Salt Marsh: and Interdisciplinary Approach. D. Casagrande, ed., Bulletin Series 100. School of Forestry and Environmental Studies, Yale University, New Haven, CT.

United States Army Corps of Engineers. 2003. “Water-borne Commerce of the United States, Calendar Year 2003.” Part 5 – National Summaries. Fort Belvoir, VA: Water Resources Support Center, US Army Corps of Engineers.

United States Army Corps of Engineers. 2004. Louisiana Coastal Area (LCA), Louisiana. Ecosystem Restoration Study. New Orleans, Louisiana. November.

United States Army Corps of Engineers. 2005. Breaux Act (Coastal Wetlands Planning, Protection, Restoration Act).

United States Census Bureau. 2011. “Selected Social Characteristics in the United States”. American Community Survey, http://factfinder2.census.gov/faces/nav/jsf/pages/searchresults.xhtml?refresh=t. Accessed April 1, 2013.

United States Census Bureau. 2012. The 2012 Statistical Abstract: The National Data Book. “481 - Individual Income Tax Returns Filed--Examination Coverage.” http://www.census.gov/compendia/statab/cats/federal_govt_finances_employment/federal_indivi dual_income_tax_returns.html. Last accessed April 16, 2012.

Vossler, C.A. and M.F. Evans. 2009. “Bridging the gap between the field and the lab: Environmental goods, policy maker input, and consequentiality.” Journal of Environmental Economics & Management 58: 338-45.

Vossler, C.A., M. Doyon and D. Rondeau. 2012. “Truth in Consequentiality: Theory and Field Evidence on Discrete Choice Experiments. Forthcoming, American Economic Journal: Microeconomics.

Wang, Hua. 2012. Economic Assessment of Rapid Land-Building Technologies for Coastal Restoration. M.S. thesis, Louisiana State University.

Woodward, R.T. and Y.-S. Wui. 2001. “The economic value of wetland services: a meta- analysis.” Ecological Economics 37: 257-70.

Working Group for Post-Hurricane Planning for the Louisiana Coast. 2006. A New Framework for Planning the Future of Coastal Louisiana after the Hurricanes of 2005. University of Maryland Center for Environmental Science. Cambridge, Maryland.

APPENDIX: SURVEY INSTRUMENT

Consent Form to be used by Knowledge Networks

o This study is being conducted for research at Mississippi State University. o Your participation is absolutely voluntary and you may quit at any time. o The survey will take approximately 15 minutes of your time to complete. o You will not be individually identified and your responses will be used for statistical purposes only. o If you have questions about your rights as a participant in this survey, or are dissatisfied at any time with any aspect of the survey, you may contact Knowledge Networks at 800-782- 6899.

[DISPLAY]

This survey is about the wetlands and barrier islands of coastal Louisiana.

Wetlands

In the U.S., more than half of commercially-harvested fish (including shrimp, blue crabs, and oysters) depend on coastal wetlands at some stage in their lives. Coastal habitats provide spawning grounds, nurseries, shelter, and food for finfish, shellfish, birds, and other wildlife. Nearly 45% of the nation’s endangered and threatened species are dependent on coastal habitats.

Barrier Islands

Barrier islands are islands located just offshore of the mainland that also provide habitat for many wildlife species. They also provide protection to fragile inland wetlands by absorbing the impact of waves, slowing erosion of inland shorelines, and protecting homes and other infrastructure from storm surge.

[RADIO]

Land Loss

Wetlands and barrier islands throughout the U.S. are losing land. These losses are due to natural erosion, sea-level rise, sinking of land, winds, tides, currents, and major storms. Losses also occur from construction of river channels, levees, and other land development.

Land loss is most severe along the Gulf of Mexico, especially in Louisiana. Louisiana’s 3 million acres of wetlands represent 40% of all coastal wetlands in the U.S., but Louisiana’s lost wetland acres account for 80% of all the wetland losses in the U.S.

If nothing is done, more land will be lost.

Q1) How familiar are you with the wetland and barrier island loss issue in coastal Louisiana? a) Very familiar b) Somewhat familiar c) Not at all familiar

[DISPLAY]

The Barataria-Terrebonne Estuary

We’d like to focus your attention on the Lower Barataria-Terrebonne Estuary in coastal Louisiana, located just south and west of New Orleans. The small map below shows Louisiana in relation to the rest of the U.S., and the large map shows the estuary (in the red box) in relation to the rest of Louisiana.

[DISPLAY]

An estuary is a coastal area where salt water from the ocean mixes with fresh waters from rivers, rainfall, and runoff. An estuary is made up of many types of habitat. The map below shows a close-up of the estuary.

The Lower Barataria-Terrebonne Estuary covers about 2.7 million acres, an area roughly three- quarters of the size of the state of Connecticut. More than 80% of the land area shown is wetlands (swamps and marshes) and barrier islands. The remainder of the land contains homes, businesses, and farms.

The estuary is home to over 500,000 people, and provides storm protection for over 1 million people, including the city of New Orleans.

[RADIO] [Q2 SHOULD BE ON THE SAME DISPLAY AS THE PREVIOUS]

Q2) Have you ever visited New Orleans or another part of coastal Louisiana? a) Yes b) No

[RADIO]

The estuary also provides habitat for 735 species of birds, finfish, shellfish, reptiles, amphibians, and mammals. At least 28 species are either endangered or threatened. The estuary supports almost 20% of the estuarine-dependent fisheries of the U.S. and is a major location for fishing, hunting, and bird watching. The estuary is rich in minerals including crude oil, natural gas, and salts, and is home to some of the busiest shipping routes in the U.S.

Q3) How familiar are you with the Barataria-Terrebonne Estuary? a) I have visited or lived in the area. b) I have never visited or lived in the area, but have heard of it. c) I have never heard of it.

[DISPLAY]

Changes in the Estuary

Between 1956 and 2000, the estuary lost about 444,000 acres, or 30% of its land area, that have converted to open water. The map below shows these changes. Wetland and barrier-island loss rates range from 4,500 to 7,100 acres each year. These estimates don’t account for major hurricane events, like Hurricanes Katrina and Rita, which eliminated an additional 24,000 acres in 2005.

Map of the Lower Barataria-Terrebonne Estuary – Habitat Change 1956-2000

[RADIO]

As barrier islands and wetlands erode, the water in the estuary is less protected from saltwater intrusion, leading to the conversion of many areas from fresh marsh to salt marsh or open water. This compromises the habitat of many animal species that rely on fresh marsh and wetlands, and affects commercial fishing.

Further, as wetlands and barrier islands disappear, the storm surge and flood protection they provide to buildings and infrastructure is reduced.

Q4) Overall, how concerned are you about these changes in the Lower Barataria-Terrebonne Estuary? a) Very concerned b) Concerned c) Mildly concerned d) Not at all concerned

[CE AND CV VERSIONS DIFFER FROM THIS POINT FORWARD.] [RANDOMLY ASSIGN RESPONDENTS TO TWO GROUPS] [GROUP=1: CE (70% OF RESPONDENTS) GROUP=2: CV (30% OF RESPONDENTS)]

[DISPLAY]

What’s Your Opinion?

[if group=1] Federal, state, and local governments are considering ways to slow further land loss and recover part of what is already lost. These projects would use a variety of methods including wetland building, barrier island restoration, freshwater and sediment diversions, and the movement of large amounts of soil on barges and via pipelines. Small-scale projects using these methods have already been tested and implemented across coastal Louisiana, but now larger- scale projects such as those presented in this survey are being considered.

[if group=2] Federal, state, and local governments are considering ways to slow further land loss and recover part of what is already lost. These projects would use a variety of methods including wetland building, barrier island restoration, freshwater and sediment diversions, and the movement of large amounts of soil on barges and via pipelines. Small-scale projects using these methods have already been tested and implemented across coastal Louisiana, but now larger- scale projects such as the one presented in this survey are being considered.

Proposed projects can restore some of the land already lost and partially prevent future losses. However, since part of the land loss is due to natural causes and these natural causes will still exist after these projects are completed, these efforts should not be thought of as ending land loss entirely, but as slowing it down. It will fall to future generations to decide whether to continue such projects or abandon them.

[DISPLAY]

[if group=1] Here, we’d like to ask your opinion on some of these proposed projects which, if adopted, would be financed with federal tax dollars. We will ask you to compare two alternative projects being considered and to tell us which one you would most likely support, if either.

[if group=2] Here, we’d like to ask your opinion on one of these proposed projects which, if adopted, would be financed with federal tax dollars. We will ask you to evaluate the project being considered and to tell us whether you would support it.

The results of this survey are advisory. In other words, they will be used to inform policymakers on the opinions and preferences of taxpayers to help them decide if and what projects should be funded in Louisiana’s Lower Barataria-Terrebonne Estuary.

[DISPLAY] [if group=1] The projects proposed here involve restoring roughly 50% of the wetlands and barrier islands lost in the Lower Barataria-Terrebonne Estuary since 1956.

The projects would take approximately 5 years to complete, and the benefits, as discussed on the next page, are not expected to diminish significantly for about 50 years.

The projects are not expected to affect oil and gas exploration and production in any significant way.

[if group=2] The project proposed here involves restoring roughly 50% of the wetlands and barrier islands lost in the Lower Barataria-Terrebonne Estuary since 1956.

The project would take approximately 5 years to complete, and the benefits, as discussed on the next page, are not expected to diminish significantly for about 50 years.

The project is not expected to affect oil and gas exploration and production in any significant way.

[DISPLAY]

[if group=1:] The two projects would result in the same amount of land being created, but are designed to produce different benefits and can differ in cost. Specifically, the projects can differ in the following 4 ways:

[if group=2:] The project is designed to produce 3 key benefits but will come at a cost. We describe the benefits and the cost in detail below:

• Wildlife habitat: the percentage of land created that is generally suitable for wildlife habitat.

• Storm surge protection: the percentage of residents in the estuary that will have improved storm surge protection.

• Change in commercial fish harvest: the expected change in commercial fish harvest, such as oysters and shrimp, relative to the current harvest level. This may affect jobs in commercial fishing as well as market prices for these seafood products.

• Cost to your household: the share of the total cost of the project that your household would incur as a one-time payment on your 2011 Federal income tax return, if the project were actually carried out. All revenue collected would be placed into a fund set up exclusively for the completion of this program. By law, no additional payments would be required.

[DISPLAY] [IF GROUP=1] [SEE EXCEL DESIGN FILE. RANDOMLY ASSIGN RESPONDENTS TO ONE OF THE 12 BLOCKS. RECORD BLOCK]

The table below shows the expected outcomes for the project options, labeled “Project A” and “Project B”, as well as the expected outcomes of not taking any action (No Action). Both Project A and Project B would be completed in 5 years and the benefits are expected to last for 50 years. Please compare these options carefully. In a few moments, you will be asked to choose which course of action you prefer.

Project A: Project B: No Action: 50% of lost land restored 50% of lost land restored Land loss expected to continue at 4,500 to 7,100 acres per year

No additional habitat and current Wildlife habitat habitat expected to decline

Storm surge No improvement and current

protection protection expected to decline

Commercial fish No improvement and current

harvest harvest levels expected to decline

Share of total cost to your household $0 (one-time tax)

[DISPLAY] [IF GROUP=2] [SEE EXCEL DESIGN FILE. RANDOMLY ASSIGN RESPONDENTS TO ONE OF THE 6 BLOCKS. RECORD BLOCK]

The table below shows the expected outcomes of the project and the expected outcomes of not taking any action (No Action). The project would be completed in 5 years and the benefits are expected to last for 50 years. Please compare the two options carefully. In a few moments, we will ask you to choose which course of action you prefer.

With Project: Without Project (No Action): 50% of lost land restored Land loss expected to continue at 4,500 to 7,100 acres per year

50% of restored land suitable No additional habitat and current Wildlife habitat as habitat habitat expected to decline

Storm surge Improved protection for 30% No improvement and current of residents protection protection expected to decline

Commercial fish No improvement and current 15% higher harvest levels harvest harvest levels expected to decline

Share of total cost to your household $0 (one-time tax)

[DISPLAY] On the next page, you will be asked to make your choice, but before doing so, please keep in mind the following:

• At this point, this is only an advisory vote, so you will not have to make any actual tax payments as a direct response to answering this survey. However, your honest answer is important because we will share the opinions we collect with policy makers, and future policy in the area and any taxes you might actually pay will be influenced by your responses in this survey today.

• [if group=1] Please think about your budget and keep in mind other things you might spend your money on instead of the project. Honestly assess the tradeoffs of supporting a proposed project and not supporting it.

[if group=2] Please think about your budget and keep in mind other things you might spend your money on instead of the project. Honestly assess the tradeoffs of supporting the proposed project and not supporting it.

• [if group=1] There is no right or wrong answer. We have found some people would support these kinds of projects and others would not support them. Both kinds of voters have good reasons for why they would vote one way or the other.

• [if group=2] There is no right or wrong answer. We have found some people would support this kind of project and others would not support it. Both kinds of voters have good reasons for why they would vote one way or the other.

[RADIO] [PROMPT IF SKIP] [ADD CHECK BOX FOR “PREFER NOT TO VOTE] [CHECK BOX AND RADIO EXCLUSIVE OF EACH OTHER] [IF GROUP=1] Q5. Once again, here are the available options. Both Project A and Project B would be completed in 5 years and the benefits are expected to last for 50 years. The No Action option means that neither restoration project would be implemented. For this advisory vote, assume that the choice receiving the most votes would be adopted. Please indicate your choice at the bottom of the table below.

Project A: Project B: No Action: 50% of lost land restored 50% of lost land restored Land loss expected to continue at 4,500 to 7,100 acres per year

No additional habitat and current Wildlife habitat habitat expected to decline

Storm surge No improvement and current

protection protection expected to decline

Commercial fish No improvement and current

harvest harvest levels expected to decline

Share of total cost to your household $0 (one-time payment) I prefer: ⁬ ⁬ ⁬ I prefer not to vote:

[RADIO] [PROMPT IF SKIP] [ADD CHECK BOX FOR “PREFER NOT TO VOTE] [CHECK BOX AND RADIO EXCLUSIVE OF EACH OTHER] [IF GROUP=2]

Q5_1. Once again, here are the expected outcomes and project cost. The project would be completed in 5 years and the benefits are expected to last for 50 years. The No Action option means that the restoration project would not be implemented. For this advisory vote, assume that the choice receiving the most votes would be adopted. Please indicate your choice at the bottom of the table below.

With Project: Without Project (No Action): 50% of lost land restored Land loss expected to continue at 4,500 to 7,100 acres per year

50% of restored land suitable No additional habitat and current Wildlife habitat as habitat habitat expected to decline

Storm surge Improved protection for 30% No improvement and current of residents protection protection expected to decline

Commercial fish No improvement and current 15% higher harvest levels harvest harvest levels expected to decline

Share of total cost to your household $0 (one-time tax)

I prefer: ⁬ ⁬

I prefer not to vote: ⁬

[RADIO] [IF GROUP=1] Q6) When voting, what expectations, if any, did you have about how others might vote? a) I thought most people would vote for “Project A”. b) I thought most people would vote for “Project B”. c) I thought most people would vote for “No Action”. d) I thought the votes would be roughly even across the three options. e) I didn’t really think about it.

[RADIO] [IF GROUP=2] Q6) When voting, what expectations, if any, did you have about how others might vote? a) I thought most people would vote for the project. b) I thought most people would vote against the project. c) I thought the votes would be roughly even across the two options. d) I didn’t really think about it.

[RADIO] Q7) When voting, how important did you think your vote would be in determining which option received the most votes? a) Very important b) Somewhat important c) Not important

e) I didn’t really think about it.

[RADIO] Q8) How likely do you think it is that the results of this survey will shape the direction of future policy in the Lower Barataria-Terrebonne Estuary? a) Very likely b) Somewhat likely c) Unlikely d) I don’t know.

[RADIO] [IF Q5=NO ACTION] Q9) [Ask only if answer to Choice Question is “No Action”] You chose the “No action” option. Would you mind telling us why?

a) I don’t really have a specific reason why. b) I’m interested, but I can’t afford it. c) I don’t think the expected benefits are worth it. d) Society has more important problems than restoring wetlands and barrier islands. e) I do not support any kind of tax increases. f) I do not live in the area – only people who live in the area should pay for the project. g) Other: ______

RADIO] [IF Q5=NO VOTE] Q9A) [Ask only if answer to Choice Question is “I prefer not to vote”] You chose not to vote. Would you mind telling us why?

h) I don’t really have a specific reason why. i) I’m not interested. j) I don’t feel that my opinion should influence policy in the area. k) The options seemed equally desirable so I could not decide which I preferred. l) The survey did not give me enough information to make a proper choice. m) Other: ______

Closing Questions

[RADIO] Q10) Have you heard of the Mississippi Flyway, a migratory bird route? a) Yes b) No c) Not sure

[CHECK BOX] Q11) Which, if any, of the following outdoor activities do you engage in? Please check all that apply. a) Freshwater fishing b) Saltwater fishing c) Boating/Canoeing d) Hunting e) Bird watching f) Hiking/nature walking g) Other______

h) I don’t engage in any outdoor activities

[RADIO] Q12) How closely did you follow the BP Oil Spill and Deepwater Horizon Accident in the Gulf of Mexico last summer? a) Very closely b) Somewhat closely c) Not at all d) I was not aware of the oil spill and accident in the Gulf of Mexico last summer.

[RADIO] Q13) Thinking about your own shopping and living habits over the last five years, would you say you have made major changes, minor changes, or no changes to help protect the environment? a) Major changes b) Minor changes c) No changes

[GRID] Q14) The projects discussed in this survey would likely involve the cooperation of federal agencies, Louisiana state and local governments, as well as private companies. How much confidence do you have in each of these to carry out these projects? A lot of Some Little No I don’t confidence confidence confidence confidence know Federal agencies Louisiana state government Louisiana local government Private companies

[RADIO] Q15) Did you file a Federal income tax return this year? a. Yes b. No

[GRID] Q16) Thinking about your overall experience in this survey, indicate how strongly you agree with each of the following statements. Strongly Disagree No strong Agree Strongly disagree opinion agree (neutral) The survey provided enough information for me to make an informed choice. Information in the survey was easy to understand. Information in the survey was presented in an unbiased way.

[RADIO] [IF RESPONSE TO Q5 WAS EITHER A OR B OR Q5_1= “WITH PROJECT”]

Q17-A) THIS IS THE LAST QUESTION ON THE SURVEY:

Earlier, you indicated that you are willing to pay $X [if group=1: COST from A or B in Q5; if group=2; COST from “With Project” in Q5_1] to fund a coastal restoration project proposed in this survey.

All things considered, are you sure about your answer?

a. Yes, I am willing to pay the tax to fund the project. b. No, I’d like to see the question again and reconsider my answer [if group=1: go to Q5A; if group=2: go to Q5A_1].

[RADIO] [IF RESPONSE TO Q5 WAS “NO ACTION” OR Q5_1= “NO ACTION”]

Q17-B) THIS IS THE LAST QUESTION ON THE SURVEY:

Earlier, you indicated that you are NOT willing to fund [if group=1: either; if group=2: the] coastal restoration project proposed in this survey.

All things considered, are you sure about your answer?

a. Yes, I am NOT willing to pay to fund [if group=1: either / if group=2: the] project. b. No, I’d like to see the question again and reconsider my answer [if group=1: go to Q5A; if group=2: go to Q5A_1].

[RADIO] [IF Q5 = “NO VOTE” OR Q5_1= “NO VOTE”]

Q17-C) THIS IS THE LAST QUESTION ON THE SURVEY:

Earlier, you indicated that you prefer not to vote on implementing a project.

All things considered, are you sure about your answer?

a. Yes, I prefer NOT to vote. b. No, I’d like to see the question again and reconsider my answer [if group=1: go to Q5A; if group=2: go to Q5A_1].

[RADIO] [PROMPT IF SKIP] [ADD CHECK BOX FOR “PREFER NOT TO VOTE] [CHECK BOX AND RADIO EXCLUSIVE OF EACH OTHER] [IF GROUP=1] [IF Q17A=NO OR Q17B=NO OR Q17C=NO] Q5A. Once again, here are the available options. Both Project A and Project B would be completed in 5 years and the benefits are expected to last for 50 years. The No Action option means that neither restoration project would be implemented. For this advisory vote, assume that the choice receiving the most votes would be adopted. Please indicate your choice at the bottom of the table below.

Project A: Project B: No Action: 50% of lost land restored 50% of lost land restored Land loss expected to continue at 4,500 to 7,100 acres per year

No additional habitat and current Wildlife habitat habitat expected to decline

Storm surge No improvement and current

protection protection expected to decline

Commercial fish No improvement and current

harvest harvest levels expected to decline

Share of total cost to your household $0 (one-time payment) I prefer: ⁬ ⁬ ⁬ I prefer not to vote: ⁬

[RADIO] [PROMPT IF SKIP] [ADD CHECK BOX FOR “PREFER NOT TO VOTE] [CHECK BOX AND RADIO EXCLUSIVE OF EACH OTHER] [IF GROUP=2] [IF Q17A=NO OR Q17B=NO OR Q17C=NO]

Q5A_1. Once again, here are the expected outcomes and project cost. The project would be completed in 5 years and the benefits are expected to last for 50 years. The No Action option means that the restoration project would not be implemented. For this advisory vote, assume that the choice receiving the most votes would be adopted. Please indicate your choice at the bottom of the table below.

With Project: Without Project (No Action): 50% of lost land restored Land loss expected to continue at 4,500 to 7,100 acres per year

50% of restored land suitable No additional habitat and current Wildlife habitat as habitat habitat expected to decline

Storm surge Improved protection for 30% No improvement and current of residents protection protection expected to decline

Commercial fish No improvement and current 15% higher harvest levels harvest harvest levels expected to decline

Share of total cost to your household $0 (one-time tax)

I prefer: ⁬ ⁬

I prefer not to vote: ⁬

[RADIO] [IF Q5A OR Q5A_1=NO ACTION] Q9_1) [Ask only if answer to Choice Question is “No Action”] You chose the “No action” option. Would you mind telling us why?

n) I don’t really have a specific reason why. o) I’m interested, but I can’t afford it. p) I don’t think the expected benefits are worth it. q) Society has more important problems than restoring wetlands and barrier islands. r) I do not support any kind of tax increases. s) I do not live in the area – only people who live in the area should pay for the project. t) Other: ______

[RADIO] [IF Q5A OR Q5A_1=NO VOTE] Q9A_1) [Ask only if answer to Choice Question is “I prefer not to vote”] You chose not to vote. Would you mind telling us why?

u) I don’t really have a specific reason why. v) I’m not interested. w) I don’t feel that my opinion should influence policy in the area. x) The options seemed equally desirable so I could not decide which I preferred. y) The survey did not give me enough information to make a proper choice. z) Other: ______

[TEXT BOX]

QF1. Please share with us any comments you would like to leave: