03/28/2023

Publications

Objective 1: Evaluate Natural Resource Management Decisions and the Effects of Climate Change to Understand Associated Welfare Impacts

1. Alix-Garcia, J. M. and D. Millimet. 2023 “Remotely Incorrect? Accounting for Nonclassical Measurement Error in Satellite Data on Deforestation” Journal of the Association of Environmental and Resource Economists. https://doi.org/10.1086/723723


2. Bigelow, D.P., Lewis, D.J., and C. Mihiar, C. 2022. “A major shift in U.S. land development avoids significant losses in forest and agricultural land.” Environmental Research Letters, 17 024007. https://doi.org/10.1088/1748-9326/ac4537


3. Blumberg, J., Goemans, C. and Manning, D., 2022. Producer Beliefs and Conservation: The Impact of Perceived Water Scarcity on Irrigation Technology Adoption(No. w30080). National Bureau of Economic Research.


4. T. Bastian, A. Van Sandt, and R. H. Coupal. 2020-2021 Wyoming Comprehensive Snowmobile Recreation Report. University of Wyoming, Department of Agricultural and Applied Economics. Report prepared for the State of Wyoming, Department of Parks and Cultural Resources. May, 2022. pp. 138.


5. T. Bastian, A. Van Sandt, and R. H. Coupal. 2021 Wyoming Comprehensive Off-Road Vehicle Recreation Report. University of Wyoming, Department of Agricultural and Applied Economics. Report prepared for the State of Wyoming, Department of Parks and Cultural Resources. October, 2022. pp. 132.


6. Davis, E.C., Sohngen, B., and D.J. Lewis. 2022. “The effect of carbon fertilization on naturally regenerated and planted US forests.” Nature Communications, 13 (5490): https://doi.org/10.1038/s41467-022-33196-x.


7. Dundas, S. J. 2022. Economic and policy implications for coastal housing markets facing sea level rise and erosion. In: Conway-Cranos, L., J. D. Toft, D. J. Trimbach, H. Faulkner, J. Krienitz, D. Williams, and S. Des Roches (eds.) The 2021 Puget Sound Nearshore Restoration Summit Proceedings. Olympia, WA: Washington Department of Fish and Wildlife. p. 46-47. PDF


8. Dundas, S. J., S. Capalbo, and J. Sterns. 2023. The Economic Implications of Climate Change for Oregon. In: Fleishman, E. (ed.) Sixth Oregon Climate Assessment. Corvallis, OR: Oregon Climate Change Research Institute. pp. 134 – 157. doi: 5399/osu/1161


9. Filippelli, Steven, Matthew Sloggy, Jody Vogeler, Dale Manning, Christopher Goemans, Gabriel Senay (2022). Estimation of field-scale irrigation withdrawals in the Ogallala aquifer. Agricultural Water Management.


10. Gharib, Ahmed, Joey Blumberg, Dale T. Manning, Chris Goemans, and Mazdak Arabi (2023). Assessment of vulnerability to water shortage in semi-arid river basins: The value of demand reduction and storage capacity. Science of the Total Environment.


11. Ghimire, R., M. Suvedi, and M. Kaplowitz. 2022. Adoption of improved agricultural practices: Learning from off-season vegetable production in Nepal. Journal of International Agricultural and Extension Education, 29(3), 32-47.


12. Hashida, Y., and D.J. Lewis. 2022. “Estimating welfare impacts of climate change using discrete-choice models of land management: An application to western U.S. forestry.” Resource and Energy Economics, 68: 101295. https://doi.org/10.1016/j.reseneeco.2022.101295


13. Hrozencik, R.A., Manning, D.T., Suter, J.F. and Goemans, C., 2022. Impacts of Block‐Rate Energy Pricing on Groundwater Demand in Irrigated Agriculture. American Journal of Agricultural Economics, 104(1), pp.404-427.


14. Jensen, A. J., S. J. Dundas, and J. T. Peterson. 2022. Phenomenological and Mechanistic Modeling of Recreational Angling Behavior Using Creel Data. Fisheries Research 249: 106235. doi: 1016/j.fishres.2022.106235


15. Lee, M. C., Suter, J. F., & Bayham, J. (2022). Reductions in National Forest Campground Reservation Demand from Wildfire. Journal of Agricultural and Resource Economics.


16. Li, L. and A.W. Ando. 2022. “Bison and rural economies.” Agricultural and Resource Economics Review 51(3): 455-472. https://doi.org/10.1017/age.2022.13.


17. Li, L. and A.W. Ando. Forthcoming. “Early exposure to nature and willingness-to-pay for grassland restoration.” Land Economics.


18. Lupi, Frank, von Haefen, Roger H. and Li Cheng. “Distributional Effects of Entry Fees for Financing Public Beaches,” Land Economics, 98:509-519, 2022.


19. Lurbé, S., Burkhardt, J., Goemans, C., Manning, D. and Hans, L., (2022). Further evidence on social comparison and residential water use. Water Resources and Economics, p.100214.


20. Manning, D., Rad, M.R. and Ogle, S., 2022. Inferring the Supply of GHG Abatement from Agricultural Lands.


21. Montúfar, Rommel, Jake Gehrung, Neil Michael Ayala Ayala, and Shady S. Atallah. 2022. “Identifying the ecosystems services of the Ecuadorian ivory palm (Phytelephas aequatorialis Spruce): A qualitative study from the central coast of Ecuador.” Economic Botany. (DOI: 10.1007/s12231-022-09552-9)


22. Rice, W.L., Newman, P., Zipp, K.Y., Taff, B.D., Pipkin, A.R., Miller, Z.D. and Pan, B., 2022. Balancing quietness and freedom: Support for reducing road noise among park visitors. Journal of Outdoor Recreation and Tourism, 37, p.100474.


23. Tanner, S., F. Lupi and C. Garnache. 2022. Estimating visitor preferences for recreation sites in wildfire prone areas, International Journal of Wildland Fire. 31(9), 871-885.


24. Van Sandt, R. Coupal, and C. T. Bastian. 2022. The Economic Contributions of Wyoming’s Snowmobile Program. University of Wyoming, Department of Agricultural and Applied Economics. Report prepared for the State of Wyoming, Department of Parks and Cultural Resources. pp. 10.


25. von Haefen, Roger H. and Frank Lupi. “How Does Congestion Affect the Evaluation of Recreational Gate Fees? An Application to Gulf Coast Beaches,” Land Economics, 98:495-508, 2022.


26. Webster, M., K. Fisher-Vanden, V. Kumar, R. Lammers, J. Perla, “Integrated hydrological, power system and economic modeling of climate impacts on electricity demand and cost.”, Jan 2022. DOI: 10.1038/s41560-021-00958-8. https://www.nature.com/articles Nature Energy /s41560-021-00958- 8.


27. Wu, H., Miller, Z.D., Wang, R., Zipp, K.Y., Newman, P., Shr, Y.H., Dems, C.L., Taylor, A., Kaye, M.W. and Smithwick, E.A., 2022. Public and manager perceptions about prescribed fire in the Mid-Atlantic, United States. Journal of environmental management, 322, p.116100.

Objective 2: Advance Economic Valuation Methods and Uses to Enhance Natural Resource Management, Policy, and Decision-Making

1. Amy W. Ando, Titus O. Awokuse, Nathan W. Chan, Jimena González-Ramirez, Sumeet Gulati, Matthew G. Interis, Sarah Jacobson, Dale Manning and Samuel Stolper (accepted). Environmental and Natural Resource Economics and Systemic Racism. Review of Environmental Economics and Policy.


2. Ando, A., T, Awokuse, J. González-Ramirez, S. Jacobson, D. Manning, N. Chan, S. Stolper, and S. Gulati. Forthcoming. “Addressing systemic racism in environmental and resource economics.” Review of Environmental Economics and Policy.


3. Ando, A.W. 2022. “Equity and cost-effectiveness in valuation and action planning to preserve biodiversity.” Environmental and Resource Economics 83: 999 – 10145. https://doi.org/10.1007/s10640-022-00674-1.


4. Asbjornsen, Heidi, Yanhui Wang, David Ellison, Catherine M. Ashcraft, Shady S. Atallah, Kelly Jones, Alex Mayer, Monica Altamirano, and Pengtao Yu. 2022. “Multi-Targeted payments for the balanced management of hydrological and other forest ecosystem services.” Forest Ecology and Management522 (2022): 120482. (DOI:10.1016/j.foreco.2022.120482)


5. Brent, Daniel A., et al. "Reducing bias in preference elicitation for environmental public goods." Australian Journal of Agricultural and Resource Economics2 (2022): 280-308. https://doi.org/10.1111/1467-8489.12463


6. Chang, J.W., A.W. Ando, and M. Chen. 2023. “Valuing changes in the portfolio of service flows from climate-induced extremes on a linked food, energy, water system (C-FEWS)”. Frontiers in Environmental Science 11: https://doi.org/10.3389/fenvs.2023.1069483.


7. Lewis, D.J., Kling, D.M., Dundas, S.J., and D.K. Lew. 2022. “Estimating the value of threatened species abundance dynamics.” Journal of Environmental Economics and Management, 113: 102639. https://doi.org/10.1016/j.jeem.2022.102639


8. Manning, Dale and Amy Ando (2022). Ecosystem Services and Land Rental Markets: Producer Costs of Bat Population Crashes. Journal of the Association of Environmental and Resource Economists


9. Netusil, N., L. Rabe, S. Dissanayake, and A. Ando. 2022. “Valuing the public benefits of green roofs.” Landscape and Urban Planning 224: 104426. https://doi.org/10.1016/j.landurbplan.2022.104426


10. Netusil, N., S. Dissanayake, L. Lavelle, A. Ando, and K. Wells. Forthcoming. “Does presentation matter? An analysis of images and texts in a choice experiment of green roofs.” Q Open.


11. Nguyen, T., D. M. Kling, S. J. Dundas, S. D. Hacker, D. K. Lew, P. Ruggiero, and K. Roy. Quality over Quantity: Non-market Values of Restoring Coastal Dunes in the US Pacific Northwest. Land Economics 99(1): 63 – 79. doi: 10.3368/le.040721-0036R


12. Swedberg, K., Boyle, K. J., Stachelek, J., Ward, N. K., Weng, W., & Cobourn, K. M. (2022). Examining Implicit Price Variation for Lake Water Quality. Water Economics and Policy, 2240005.


13. Wang, Rui, Daniel Brent, and Hong Wu. "Willingness to pay for ecosystem benefits of green stormwater infrastructure in Chinese sponge cities." Journal of Cleaner Production 371 (2022): 133462. https://doi.org/10.1016/j.jclepro.2022.133462

Objective 3: Integrated Policy and Decision-Making

1. Albers, H.J., K. Kroetz, C. Sims, A.W. Ando, D. Finnoff, R.D. Horan, R. Liu, E. Nelson, and J. Merkle. 2023. “Using characteristics of migratory species to inform conservation policy questions.” Review of Environmental Economics and Policy 17(1) https://www.journals.uchicago.edu/doi/10.1086/724179.


2. Brent, Daniel A., Joseph H. Cook, and Allison Lassiter. "The effects of eligibility and voluntary participation on the distribution of benefits in environmental programs: an application to green stormwater infrastructure." Land Economics (2022): 102920-0166R. https://doi.org/10.3368/le.98.4.102920-0166R


3. Conte, Marc, Kristiana Hansen, Kyle Horton, Chian Jones Ritten, Leah H. Palm-Forster, Jason F. Shogren, Frank Wätzold, and Teal Wyckoff. (2023) “A Framework to Evaluate Mechanisms to Support Seasonal Migratory Species.” Review of Environmental Economics and Policy.


4. Jones Ritten, Chian, Amy Nagler, Kristiana Hansen, Drew Bennett, and Ben Rashford. (2022). “Incorporating Landowner Preferences into Successful Migratory Species Conservation Policy.” Western Economics Forum. 20(1): 83-94.


5. Kone, D., K. Biedenweg, A. Doerr, S. J. Dundas, P. Nelson, R. Niemiec, and A. Rogers. 2022. Establishing a roadmap for incorporating social science and human dimensions into potential sea otter reintroductions on the U.S. West Coast. Sacramento, CA: California Ocean Science Trust. 42 pp. PDF


6. Sims, C., Armsworth, P R., Blackwood, J., Fitzpatrick, B., Kling, D M., Lenhart, S., Neubert, M., Papeş, M., Sanchirico, J., Shea, K., & Springborn, M. (2023). Leveraging federalism for flexible and robust management of social-ecological systems. People and Nature, 00, 1– 9. https://doi.org/10.1002/pan3.10458


7. Ulrich-Schad, Jessica, Paul M. Jakus, Malieka Bordigioni, and Don Albrecht. 2022. “Preferences for Economic and Environmental Goals in Rural Community Development in the Western United States.” Rural Sociology, 87(2):605-641.


8. von Haefen, Roger H. et al. “Estimating the Benefits of Stream Water Quality Improvements in Urbanizing Watersheds: An Ecological Production Function Approach,” PNAS,


9. Zhang, Bo, Douglas H. Wrenn, Janak Joshi, and Edward C. Jaenicke (2022). "Housing Wealth, Food Spending, and Diet Quality: Evidence from Panel Data." Agricultural and Resource Economics Review. (2022): 1-25. https://doi.org/10.1017/age.2022.12

 

2023 Annual Meeting Abstracts

SESSION 1: Land Use Change & Protection

Title:               Individual environmental preferences and aggregate outcomes: an empirical socio-ecological model of forest landowner invasive species control

Authors:         Shadi S. Atallah and Ju-Chin Huang

W5133 Obj:   1

Presenter:      Shadi S. Atallah

Email:             satallah@illinois.edu

Abstract: Tree pests, diseases, and weeds threaten the ability of public and private forests to provide ecosystem service worldwide. While forest ecosystems span large spatial scales, forest disturbances are typically managed at multiple, smaller spatial scales due to jurisdiction. This mismatch in ecological and management spatial scales is most pronounced in the Eastern United States (U.S.), where most forests are owned by private parties with diverse ownership and management motivations. Given that these forest disturbances are mobile and renewable, they can lead to spatial-dynamic externalities if they are under-controlled from a landscape perspective.

            We combine the benefit of bioeconomic models in modeling the spatial dynamics of externalities and the benefit of choice experiments in characterizing preferences for control methods and ecosystem service outcomes. We develop a socio-ecological model of bio-invasion and control that is discrete in both time and space. The model includes computational private forest landowners that are ecologically connected beyond their property boundaries through the dispersal of the invasive shrub. The distribution of landowner characteristics and willingness to pay for invasive shrub control is based on a discrete choice experiment survey. According to the survey results with 939 landowners in Maine and New Hampshire, choices of invasive shrub control options are affected by control methods (chemical vs. mechanical), neighborhood control rate, and cost. Landowners prefer mechanical over chemical methods, on average, but mixed logit models suggest a significant individual heterogeneity regarding chemical use. The proportion of neighbors controlling the invasive shrub significantly affects own decision to control but only for landowners with a land size smaller than 24 acres. Larger landowners do not take into consideration whether neighbors control the invasive shrub.

We simulate the socio-ecological system over 50 years and collect data on the aggregate landowner welfare, total number of treatments, and the percentage of mature trees. We do so under no control, a cost-share payment, and a lumpsum subsidy. We find that under no subsidy, no control takes place. Under a lumpsum subsidy, and compared to a cost-share payment,  landowners choose mostly chemical rather than mechanical treatments, experience a lower net private welfare gain, achieve a better landscape level of bioinvasion control which allows more trees to reach the mature age with fewer treatments. The results highlight the counterintuitive role of individual pro-environmental preferences over invasive species control within one’s property on detrimental landscape-level environmental and economic outcomes. From a methodological perspective, the integrated model illustrates the promise of using choice experiments to generate willingness-to-pay functions for socio-ecological models and the role of the such models in providing insight over the dynamic outcomes of preferences identified in choice experiments.

 

Title:               Does Land Conservation Affect Local Employment? Evidence from the Conservation Reserve Program

Authors:         Liqing Li and Amy W. Ando

W5133 Obj:   1

Presenter:      Liqing Li

Email:             lli40@illinois.edu

Abstract:        Land conservation programs that improve environmental quality by retiring active croplands may entail a trade-off between environmental quality and local economic development. This paper examines the impact of private land conservation programs on local employment in the U.S. based on Conservation Reserve Program (CRP).

 

The CRP, established by the Food Security Act of 1985, is the largest federally funded private

land retirement program in the U.S. The program provides annual financial compensation to

landowners who voluntarily join the program by signing a ten to fifteen-year contract. While the

CRP provides several environmental benefits, land uses changing from agricultural production to

environmental conservation may have unintended negative impacts on the local economy (Beck

et al., 1999).

 

Previous research that focused on the economic impact of the CRP was mainly conducted in the

late 1980s to the early 1990s when the program was first implemented (Hyberg et al., 1991; Broomhall and Johnson, 1990; Martin et al., 1988; Mortensen et al., 1990; Siegel and Johnson,

1991, Sullivan et al., 2004). However, enrollment criteria and status have experienced significant

change since the late 1990s. Moreover, existing research mainly focused on the impacts in a

specific county, multi-county, or state. Sullivan et al. (2004) expanded the study region to 1481

counties located in the Great Plains and most likely to be affected by the CRP. However, the

effects of the CRP on local economic developments may vary by region. Research that can use

updated data and expand the study region to explore the impacts of CRP enrollment on local

employment for the entire U.S. and by region can bring new insights to this topic.

 

We utilize a panel fixed-effects model to study the impacts of CRP enrollment on local

employment from 1998 to 2019. As a robustness check, we utilize heteroskedasticity to construct

instruments (Lewbel, 2012) and examine the impacts of the CRP. The robustness check results

align with findings from a panel fixed effects model. We find CRP enrollment has a negative and

significant impact on employment for the agricultural sectors, though such impacts vary by

region. The CRP does not affect farm jobs in the Midwest and South, where CRP enrollment is

high but per acre farm labor requirement is low. The negative impacts of CRP enrollment on

farm jobs are mainly from the West and Northeast.

 

While the CRP decreases farm employment in certain regions in the U.S., improved natural amenities from CRP enrollment may promote rural development. We find that CRP enrollment increases the number of non-farm jobs in industries related to recreation, food, and lodging services. However, such non-farm jobs only account for a small portion of the total economy. The impact of the CRP on aggregated non-farm job opportunities is insignificant. Overall, we do not find evidence that the CRP harms total local employment. Future debates over the scale and geographic focus of the CRP and other agricultural conservation programs may draw on the results of this research to inform estimates of the economic effects of conservation activities

in different parts of the U.S.

 

Title:               The ecosystem service value of maintaining terrestrial protected areas in China

Authors:         Haojie Chen

W5133 Obj:   1

Presenter:      Haojie Chen

Email:            

Abstract:        Protected areas conserve and provide various ecosystem services, many of which are external to markets and invisible to commonly used development indicators (e.g., GDP), and provide indirect benefits to human well-being. The valuation of ecosystem services can assist in decision support associated with the management of protected lands, or with respect to deciding what to protect. Maintaining and expanding protected areas (PAs) can benefit humans and the rest of nature, but incurs management costs and opportunity costs . Using benefit transfer with data from existing studies, I estimated the annual value of the flow of the key ecosystem services for these lands. The value-flow of water retention, maintenance of soil fertility, sandstorm prevention, carbon sequestration, oxygen release, and recreation from China’s terrestrial protected areas is estimated be $2.64 trillion per year. The estimated benefit is over 15 times greater than the annual costs for preventing the degradation of the protected areas. These results indicate that, the benefits to social welfare for protecting these areas is substantially greater than the cost of protecting them, both in terms of management costs and opportunity costs. This study, like many others, assumes a constant unit value. That is, the value of a service equals the quantity of the service (e.g., tonnes of water) multiplies the unit value (e.g., value of water per tonne), which can be reflected by various indicators (e.g., market price, alternative cost, or avoided damage cost). Such an assumption ignores tipping points related to the provision of ES possible diminishing marginal value, and supply-demand relationships of the service. However, this assumption makes the assessment of values, especially at large special scales, practical and possible In order to better integrate these values into natural capital accounts, I am currently exploring ways to relax this assumption. I also look forward to exploring more sophisticated valuation approaches, including deliberative valuation is effective for soliciting ES values for protected areas.

 

Title:               Understanding tradeoffs residents make between fuel treatment and environmental benefits to reduce fire risk

Authors:         José J. Sánchez, Lorie Srivastava and Emily Schlickman

W5133 Obj:   1

Presenter:      José J. Sánchez

Email:            

Abstract:        Wildfires across the western United States are becoming larger, more frequent, and more severe. This shift has been largely attributed to a global change in climate along with local changes in the quantity of fuels in forested areas. Furthermore, increasing population and sprawl has expanded the wildland-urban interface (WUI), which has heightened fire-related risks for communities and people living in the WUI. In California, 3 out of every 10 homes are in the WUI – a number that is expected to increase in the coming years. 

 

This project investigates ways to reduce the risk of wildfire in diverse WUI communities across California by better understanding residents’ perceptions of fuel management and their associated environmental benefits. The study focuses on vulnerable communities, which was defined by having high fire risk, large percentage of minority population, low income, English as a second language, and unreliable internet service. A multimode survey approach (in-person, online, and mail) was used to obtain a representative sample.

 

Specifically, the research questions addressed by this study are: (1) what is the economic value that residents place on their preferred type of fuel reduction activity to help reduce fire risk? and (2) what benefits do residents derive from provisions of ecosystem services that are protected or improved by their preferred fuel reduction activity?  To address these research questions, we estimate the economic value of the trade-offs between wildfire and ecosystem benefits from fuel treatments. We used a choice experiment survey of residents in 16 communities throughout California. Each respondent was made to select between fuel treatment options in which the activity (e.g., prescribed fire, mechanical thinning, grazing, etc.) and benefits to their communities (e.g., improved air quality, water quality, and greater recreation opportunities) varied.

 

Data collection is complete for the in-person survey, and ongoing for the online and mail survey modes. The survey booklet is expected to be mailed out in mid-February. Preliminary results will be presented using the in-person and partial online survey data. The results of this study may be applicable and useful in several ways relating to both on-the-ground resource managers and in developing more research-based planning processes for the WUI. For example, the results will help stakeholders better understand their preferred fuel treatment activities and their potential ecosystem service benefits in WUI watersheds. Second, the results will help land and fire managers better target spatially-specifically fuel management strategies by better accounting for vulnerable communities’ preferences for their surrounding landscapes. Third, this project will provide critical information to improve policy uptake and decision-making such as the recent Climate 21 Memo’s opportunities for the USDA to address development in the WUI and the increased use of prescribed fire.

 

SESSION 2: Revealed Preferences

Title:               Travel cost method: traditional surveys versus mobile device data

Authors:         Alecia Evans, Jude Bayham, Charles Sims

W5133 Obj:   2

Presenter:      Alecia Evans

Email:            

Abstract:        Mobile device data are being increasingly leveraged in the valuation of recreational sites. This is in lieu of traditional survey-based methods such as the travel cost method (TCM). The use of mobile device data is promising for numerous reasons. Relative to traditional surveys, it is less costly  and it provides an opportunity to gather time-series data over longer time horizons (Jaung and Carrasco 2020). Despite its potential, there is a dearth of studies that employ mobile phone data in conducting TCM (Jaung and Carrasco 2020; Kubo et al. 2020). However, there is no quantitative evidence of how consumer surplus estimates derived from mobile device data compare to estimates using a traditional survey-based approach. In this paper, we address this information gap in an effort to highlight possible tradeoffs involved with using mobile device data for non-market valuation. We estimate, and compare, the recreational use value of the Knoxville Urban Wilderness (KUW) in Tennessee, using both the traditional survey-based method and aggregated mobile device data from SafeGraph.

Our estimation is threefold. First, we use online survey data collected throughout 2021 and 2022 to estimate both a hybrid individual – zonal TCM (IZTCM) and Zonal TCM (ZTCM). While the online survey collected individual-level socioeconomic data, the recall period for visits was only over the previous month and no data were collected on alternative sites. Therefore, we exploit the variation that exists in the individual level data rather than being limited to variations across zones by using zonal averages. Trips are aggregated according to the zip code of survey respondents. For the ZTCM, the individual level socio-economic data is aggregated at the zip code level.

Next, we use only the anonymized, aggregated mobile device data from SafeGraph to estimate the ZTCM. The data gives aggregated visits for mobile devices to KUW from each zip code in the same timeframe as the online survey. SafeGraph does not provide socio-economic data on users. As such, we complement this with aggregated socio-economic data, such as age and income, from sources such as the US Census Bureau and the Internal Revenue Service.

Finally, we supplement the traditional survey data with the mobile device data from SafeGraph and re-estimate a ZTCM. We use SafeGraph data to estimate travel cost from the zip codes individuals resided in from the online survey. This is complimented with the aggregated socio-economic data from the survey. In essence, mobile device data were used to refine the travel cost estimates (particularly distance travelled), whilst the traditional survey provided demographics on KUW users.

The traditional survey yields consumer surplus estimates of $115 per person per trip with 95% confidence interval of $31 to $1127. Analysis of the Safegraph data is ongoing. We expect SafeGraph data will lessen non-response bias and provide a more accurate estimate of distance travelled. However, it does not provide data on visitor spending potentially introducing new bias in the travel cost estimates. We expect improvements in the consumer surplus estimates when we supplement the survey with the Safegraph data relative to relying on the survey alone. 

Title:               Jointly Estimating Site-Choice and Trip Length for Non-Market Valuation

Authors:         Russel A. Dame, Daniel K. Lew, David M. Kling

W5133 Obj:   2

Presenter:      Russel A. Dame

Email:             damer@oregonstate.edu

Abstract:        In the study of recreation behavior, economists commonly use site-choice models to inform welfare effects from site-level closures or quality changes.  However, applications focused on site choices often make simplifying assumptions about the length of trips, frequently assuming they are the same length for all individuals in the sample. This oversimplification may be restrictive for samples with large variations in trip length leading to biased welfare estimates and limits the scope of the study to per trip welfare estimates. Researchers that consider site choice and trip length can calculate additional daily welfare effects and ecological endpoints that can be used in dynamic simulations, providing more information to policymakers.

We develop and estimate a joint site-choice and trip length model that links the site choice and trip length decisions via expected trip length. Our linked framework considers the total impact on the conditional indirect utility function from longer trips. We apply this model using data collected from non-resident anglers that participated in a recreational saltwater fishing trip in Alaska. Non-resident anglers are more likely than residents to participate in a single fishing trip to Alaska in a given year and participate in a multi-day fishing trip. Thus, fishing trip length is an important margin of choice. We estimate the trip length model as a one-inflated zero-truncated negative binomial distribution which accounts for the large proportion (~48%) of the sample participating in a single-day trip. We also estimate the site-choice model as a two-stage nested logit model that considers anglers’ fishing mode choice (charter v. non-charter). Both models are estimated simultaneously using a full-information maximum likelihood framework.  Matching our prior expectations, model estimates imply that an increase in expected daily harvest rates will increase the trip length for most key Alaskan species. However, as the historical harvest rate for Pacific halibut (Hippoglossus stenolepis) increases, the estimated coefficients suggest that trip length may decrease. Anglers may substitute fishing time with other non-fishing activities while in Alaska once they reach a satiation point of Pacific halibut harvest.

We conduct two simulations using the estimated model. The first considers a reduction in the statewide expected harvest rates of Pacific halibut due to proposed Blue Line management strategy plan. We find that reductions in the expected harvest of Pacific halibut led to an increase in fishing trip length by approximately 2 to 3 hours per trip. Our simulation suggests that expected recreational mortality will decrease by approximately 2 percentage points less than predicted in the Blue Line management strategy plan due to longer fishing trips. This implies that ignoring the on-site time component may bias oversimplified calculations of ecological endpoints and cause unintended policy consequences. The second simulation considers a closure to the harvest of silver salmon (Oncorhynchus kisutch) for a single site. We find a near zero effect on trip length as anglers substitute to sites with similar characteristics. Unlike the first simulation, the second simulation impacts a single site allowing for greater substitution effects than a statewide policy.

Title:               Regression and matching in hedonic analysis: Empirical guidance for estimator choice

Authors:         Klaus Moeltner, Roshan Puri, Robert J. Johnston

W5133 Obj:   2

Presenter:      Klaus Moeltner

Email:             moeltner@vt.edu

Abstract:        We illustrate how estimation results from hedonic regression, basic matching, and regression adjusted matching can be combined to provide guidance on model choice for housing market studies in the presence of unobserved spatial and temporal effects, when the primary goal of the analysis is to estimate an unbiased binary treatment effect. We show, conceptually, that the “golden rule” of single-control matching no longer holds when these effects are present, and exact matching in space and time is practically infeasible. A larger number of matched controls can, in fact, trigger a beneficial “cancellation effect,” a to date unexplored route towards unbiasedness. This, in turn, reveals an empirical prescription for model search by choosing the number of matched controls that drive the cancellation effect closest to zero. In our application to flood zone discounts for property values in Massachusetts, we illustrate that a close-to-zero cancellation effect is empirically feasible, and that the resulting preferred model produces reasonable estimates and has desirable efficiency properties.

 

Title:               Voting with their feet: Do Political Partisans Value Neighborhood Public Goods Differently?

Authors:         Corey Lang (clang@uri.edu) and Jarron VanCeylon (jvanceylon@uri.edu)

W5133 Obj:   2

Presenter:      Jarron VanCeylon

Email:             jvanceylon@uri.edu

Abstract:        Partisan divides are common among many civil issues in the United States. While understandable for issues like abortion and guns, conservation of land seems like something both political parties should get behind. Intuitively, the value of outdoor recreation and beautiful farm landscapes would seem universal. Hunters and anglers are predominantly aligned with the Republican Party, and these groups also push strongly for continued federal land protection and exhibit strong support for the Clean Water Act (National Wildlife Federation 2012). And yet, when it comes to the ballot box, a voter’s political affiliation is the single largest determinant of support for environmental conservation, with approval being 20 to 40 percentage points higher for Democrats than Republicans (Burkhardt and Chan 2017, JAERE; Lang and Pearson-Merkowtiz 2022, JEEM).

 

The purpose of this paper is to investigate if the partisan gap present at the ballot box is observed in the housing market, and hence if there is a true partisan gap in preferences for conservation, or if it is some phenomenon specific to voting. Following a Random Utility Maximization framework, we develop a residential sorting model to estimate household willingness to pay (WTP) for residential proximity to conserved lands, allowing for heterogeneity in WTP by partisan affiliation. This analysis requires the construction of a comprehensive dataset that comprises a cross-section of properties and proximate spatial amenities, as well as the occupant’s income, race, ethnicity, and political affiliation gathered from public mortgage application and voter registration data. Current data construction spans two major housing markets, greater Rhode Island, and greater Charlotte North Carolina. Our eventual goal is to add a third housing market from a Republican-leaning area.

 

Preliminary results suggest that average WTP for a residential location proximate to conserved land is about $3200 in purchase price. Importantly, we find no statistical differences in willingness to pay between partisan groups, indicating that open space is universally valued across the partisan divide. This paper illustrates a seeming paradox that partisan preferences are similar in the housing market but dramatically different at the ballot box. This has two implications. First, valuation studies using voting data may underestimate Republicans’ WTP for land conservation (and potentially many more public goods). Second, land conservation efforts may be hurt by relying on the ballot box. Currently, much of the FFOS preservation spending by states and municipalities in the United States is directly approved by voters – on average over $2 billion per year (The Trust for Public Land 2022). However, this reliance on voters may ironically lead to less FFOS conservation than is desired by the public, given the seeming change in preferences by Republicans moving across venues. For these reasons, more research is needed to understand mechanisms that drive this change in expressed preferences.

 

SESSION 3: Farm and Food

Title:               Who Values Local Food, Nature Conservation, and Farmland Preservation, and How Much?

Authors:         Frederick Nyanzu (UIUC), Bryan Parthum (Environmental Protection Agency), Corey Lang (University of Rhode Island)

W5133 Obj:   1

Presenter:      Amy Ando

Email:             amyando@illinois.edu

Abstract:        Peri-urban nature and farmland conservation activity is growing in the U.S. along with attention to expanding local food supplies. While many papers have quantified total or average public values of nature conservation, we know little about how the values people have for conservation vary among groups of people, and little research has actually quantified public willingness-to-pay for protecting farmland around cities and from increasing the supply of local food. There are also important questions to answer about who benefits most from conservation and local food. For example, other research shows that Black and Hispanic household are relatively unlikely to live near protected areas (Lang et al. 2023), but it isn’t clear whether that pattern reflects discriminatory barriers or preference heterogeneity. This paper uses a choice experiment survey of people in 10 cities in the East and Midwest to estimate the values of nature conservation, farmland preservation, and local food, and to explore how those values vary with features of people like race, ethnicity, income, and factors that influence mobility like disability and personal access to a car.

 

We craft a choice experiment survey to elicit the values people place on different kinds of conservation and local food. The survey explains that the respondents’ city and local non-profit groups could invest in efforts to establish protected natural areas and farmland around the edge of the city. Those protected areas could vary in several ways: number of acres of protected nature and protected farmland, amount of local food available to the respondent from the protected land in the growing season, type of recreational amenities (trails, picnic areas, or both), distance from the respondent’s home, and cost to the respondent in increased taxes. We field our survey online to about 800 respondents in 10 cities through Qualtrics.[1] Basic estimates of the sample-average marginal values of the attributes come from estimating fully correlated mixed multinomial-logit models in willingness-to-pay space with data from the responses to the choice questions. Then, to understand how preferences vary among groups of people we use two additional types of econometric models.

 

The results are still preliminary, but interesting findings have already emerged. Our sample displays a fairly strong generic preference for the status quo. However, the total annual value of the features of an actual conservation project is quite large. Who benefits the most from conservation and provision of local food? In fact, we find very little evidence of preference heterogeneity. There are no major differences in preferences between White, Black, and Hispanic people in our sample of respondents. Low income respondents do have relatively low MWTP for expanding local food that can be purchased from farm stands, but there are no major difference in preferences over local food between White and Black respondents.

 

Much research has already documented that people gain significant value from efforts to protect nature around cities. This research shows that residents of mid-sized cities in the U.S. would be willing to pay just as much to protect peri-urban farmland as they would nature. Our findings also lend evidence to inform current discussions about expanding local food in urban areas. We find that people of all races would value expanded farm stand food, though low-income people benefit less from that standard approach to food supply. To benefit low-income urban residents, policy makers and urban planners might think more about expanding areas where free foraging is available, since we find uniformly high WTP across all demographics for including access to fishing and plant-food foraging in protected natural areas.

 

Title:               Conservation and Food Security: Recreational Fishing, Foraging, and Willingness-To-Pay (WTP) for Local Food

Authors:         Frederick Nyanzu (UIUC), Bryan Parthum (Environmental Protection Agency), Corey Lang (University of Rhode Island)

W5133 Obj:   1

Presenter:      Frederick Nyanzu

Email:             fnyanzu2@illinois.edu

Abstract:        To combat land conversion around population centers, peri-urban nature conservation and farmland preservation efforts are growing in the United States. While studies have quantified total or average public values of nature conservation, we know little about the value different groups of people place on local food. Some scholars show that strategic local food systems can help mitigate food insecurity (O’Hara and Toussaint 2021), but others raise concerns that local food systems can fall prey to food gentrification (Rice 2015). Willingness-to-pay for local food can be influenced by the food insecurity status of the individual as well as historical or systemic factors, and failing to account for these factors can bias one’s understanding of the values that local food expansion will provide to marginalized people.

 

This paper fills the gap in our knowledge of the values people place on expanded access to local food. We use a choice experiment survey on nature and farmland conservation that draws 840 respondents from 10 cities in the East and Midwest States in the United States. We develop descriptive quantitative evidence about the roles that fishing and foraging play in food access, and how that varies among different socioeconomic groups. We estimate average willingness to pay (WTP) for expanding food available at local farm stands and in local natural areas managed to provide foraging opportunities. Finally, we explore how the values people have for those types of food access vary intersectionally with race, income, food insecurity status, and the experience they already have with fishing and foraging.

 

Preliminary descriptive data analysis shows that where people get food from do not vary by racial group and income level. However, foraging and fishing behavior vary by income level and racial groups; while high-income groups mainly fish for recreation, low-income and minority groups predominantly fish for food. Our data are consistent with previous findings (Odoms-Young 2018) that food insecurity is higher among minority or low-income groups; thus, our data show that food-insecure groups fish more to supplement food supply than for recreation. Early estimates of WTP for expanded local food from the choice experiment show that food insecurity has a negative (though statistically insignificant) relationship with willingness-to-pay for local food, even accounting for confounding factors that might bias the estimate. Early estimates of WTP for expanded local food from the choice experiment reveal mixed evidence on how those values vary with food insecurity. There certainly is no clear evidence that providing more food in local farm stands and foraging areas would convey large benefits to food insecure urban residents.

 

Our study contributes to the discussion on racial distribution of food insecurity and socio-cultural characteristics and values of subsistence fishing and foraging. Our findings will help policy makers and managers understand the roles that foraging and angling play in supplementing food access for low income and minority households, and inform discussions about whether expanding local food pathways will have value to the food insecure people we might hope could benefit the most.

 

Title:               Enrollment Restrictions and the Adoption of Conservation

Practices in the U.S. Corn Belt

Authors:         Xiaolan Wan, Gregory Howard, and Wendong Zhang

W5133 Obj:   1, 2, 3

Presenter:      Gregory Howard

Email:             glhoward@illinois.edu

Abstract:        Using a mixed-mode survey of 568 farmer respondents in the Boone and North Raccoon River watersheds in Iowa, we utilize a discrete choice experiment to examine farmer behavioral response to a new policy design of conservation programs—enrollment restrictions in cost-share programs. Using a random parameters logit model, we first show how farmers’ preferences for conservation practices change when enrollment restrictions are imposed in conservation programs. Our results suggest that eligible farmers are more likely to choose a conservation contract with enrollment restrictions. We consider three contracts with per-acre payments similar to those offered in Iowa’s EQIP program—incentivizing cover crops ($40), no-till ($10), and split nitrogen application ($9) for two years—and compare willingness-to-accept for these contracts with and without enrollment restrictions. Mean willingness-to-accept estimates for the cover crops, no-till, and split nitrogen application contracts decrease by 82%, 92%, and 93%, respectively, when enrollment restrictions are introduced. In addition, estimated participation supply curves demonstrate higher enrollment when introducing enrollment restrictions, though this is especially true for low compensation levels, and enrollment decreases as the proportion of farmers who are ineligible for the conservation contract rises. We show that this sharp reduction in WTA for enrollment restricted contracts (compared with contracts without enrollment restrictions) is not driven by sample selection, as we see similar results when we consider only subsamples of eligible farmers.

 

Title:               Moving to the Country: Understanding the Effects of Covid-19 on Property Values and Farmland Development Risk

Authors:         Kelsey Johnson, Lee Parton, Christoph Nolte, Matt Williamson, Theresa Nogeire-McRae, Jayash Paudel, Jodi Brandt

W5133 Obj:   1, 2, 3

Presenter:      Lee Parton

Email:             leeparton@boisestate.edu

Abstract:        As human populations grow, one strategy for meeting housing demand is through the development of agricultural land and other open space, which can generate negative externalities. This may be addressed at local, state, or federal levels with land-use planning, including farmland preservation policies (FPPA, 1981; Hunter 2022). Efficient land-use planning in the presence of competing land uses requires knowledge of development risk, housing preferences, and the full costs of farmland loss. We conduct a national scale hedonic analysis to investigate how COVID-19 has affected property prices in suburban and rural areas with high development risk and for the purpose of understanding the effects of COVID-19 driven shifts in housing preferences. This analysis is made possible through the use of data on U.S. housing transactions provided through the ZTRAX program (Zillow, 2020). Our analysis demonstrates that the pandemic caused differential price impacts across 21 states, suggesting heterogeneous shifts in preferences. Furthermore, our estimates suggest heterogeneous price effects driven by the characteristics of nearby urban areas, with prices appreciating faster on land at risk located near small urban areas. Our analysis finds that development pressure, as measured through transaction prices, changed in the aftermath of the COVID-19 pandemic.

 

Title:               Electricity Demand by the Irrigated Sector in Response to Climatic Shocks

Authors:         Dilek Uz

W5133 Obj:   3

Presenter:      Dilek Uz

Email:             dilekuz@unr.edu

Abstract:        Climate change is already impacting global agricultural productivity (Lobell et al., 2011; Burke et al., 2015) with crop losses due to extreme heat projected to rise substantially through the end of the century (Petersen, 2019). A rich economics literature characterizes how climate change impacts agricultural profits, land values, and yields (Mendelsohn et al., 1994; Schlenker et al., 2005; Burke and Emerick, 2016). In regions with access to water resources for irrigation the agricultural productivity losses associated with climate change are less severe as producers as can potentially adapt to reduced precipitation and higher temperatures by applying more water to their crops (Schlenker and Roberts, 2009). This paper addresses this form of climate change adaptation by characterizing how the irrigated agricultural sector responds to extreme heat and drought. Specifically, we use unique utility level data on electricity distributed to irrigated agricultural producers to empirically model how the sector’s energy demand responds to climatic shocks. Our results shed light on an important form of climate change adaptation for the agricultural sector while also informing projections of electricity demand by the irrigated sector under differing climate change scenarios.

 

Electricity is a vital input for the U.S. irrigated agricultural sector, powering nearly 75%

of all irrigation pumps which are generally used to extract water from underground aquifers

(NASS, 2018). Electricity use by the irrigated agricultural sector is also an important part of aggregate electricity demand, constituting approximately 1% of national demand annually (Sowby and Dicataldo, 2022). Despite irrigated agriculture’s importance for the electricity distribution and generation sector, relatively few studies have examined how climatic and economic factors influence irrigated agriculture’s demand for electricity. Notable exceptions include Maddigan et al. (1982) and Uri and Gill (1995) which both empirically model the price elasticity of demand for electricity by the irrigation sector. This paper builds on this work by

incorporating weather variables in empirical models of the irrigation sector’s demand for electricity. This extension to past modeling efforts is important as the sector’s demand for electricity exhibits substantial inter-annual variation based on growing season temperatures and precipitation as higher temperatures increase crop water needs and precipitation and pumped water are roughly substitutes. Projected increases in the intensity and severity of droughts and heat waves under global climate change scenarios underscores the importance of characterizing the irrigation sector’s responsiveness to such climatic shocks (Sch¨ar, 2016; Satoh et al., 2022).

 

SESSION 4: Recreation and Conservation

Title:               Measuring the Benefits and Costs of Conservation Program Design for Migratory Species

Authors:         Chian Jones Ritten, Roger Coupal, Kristi Hansen

W5133 Obj:   1, 3

Presenter:      Chian Jones Ritten

Email:             cjonesri@uwyo.edu

Abstract:        Conservation actions that support terrestrial migratory species, including elk and mule deer, have become a growing regional policy concern in the Pacific Northwest and the Intermountain West. Unlike sedentary species, these migratory species require habitat across expansive, usually connected areas. This unique need for spatially explicit conservation actions along pre-defined migration routes, or corridors, challenges the effectiveness of existing conservation programs (Conte et al., 2022). Further, because these species move throughout the landscape seasonally, effective conservation is species, time, and place specific, suggesting a ‘one-size-fits-all’ approach to conservation will be ineffective to meet the needs of migratory land species (Jones Ritten et al., 2022).

Previous literature has highlighted the need to measure ecological benefits when designing conservation policy for terrestrial migrating species (e.g., Conte et al., 2022; Jones Ritten et al., 2022). Yet, incorporating the full cost of conservation policy, specifically the value of any reduction in outdoor recreation, is also a critical factor that has yet to be addressed in the literature.

We use the case study of mule deer and elk migration corridors in Southwest Wyoming (Rudd, et al., 2018) to show the need for policy makers to incorporate the value of impacts to recreation and local public finance into conservation policy design. The choice of policy design, which will affect tourism and provision of public services in the county, will likely influence economic development including local government public finance and overall impacts to the region. The transition to taxation and public service regimes associated tourism and outdoor recreation increases can be constrained due to political and social considerations. Thus, misestimating the true cost of conservation design may lead to not only ineffective conservation policy, but broad regional impacts. 

Title:               Snow, skiing, and the impacts on the regional economy

Authors:         Joseph Snapp, Yvette Uwineza, Jude Bayham

W5133 Obj:   1

Presenter:      Jude Bayham

Email:             jude.bayham@colostate.edu

Abstract:        The ski industry is an important attraction of economic activity in many rural mountainous regions around the country. Climate change poses a threat to the ski industry and the economies that support ski tourism. Using high-resolution spatiotemporal mobile device data, we estimate the impact of snow depth on ski resort visitation and the subsequent impact of that visitation on supporting industries at different proximities from the resorts. We then quantify the direct impacts of visits to supporting industries on retail sales in the ski resort communities. We find that a 10% increase in snow depth leads to an increase of $1.7 million in daily retail sales for a representative community in Colorado. Our results highlight the economic vulnerability of resort communities to climate change.

 

Title:               Representing landscape type and wildfire burns in different sized recreation sites in a discrete choice framework

Authors:         Sonja Kolstoe, Trudy Cameron, Abby Kaminski

W5133 Obj:   2

Presenter:      Sonja Kolstoe

Email:             sonja.kolstoe@usda.gov

Abstract:        Recreational destinations in consumers’ consideration sets often come in all different shapes and sizes but are included in the choice set of a recreation demand model in the same fashion based on the presence and/or extent of their site attributes. Generally, when land cover is included in a recreation site choice model, it is included as a dummy variable for the dominant land cover type, as was done in Kolstoe, Cameron and Wilsey (2018). That study focused on the Pacific Northwest, where most of the narrowly defined sites of the type in question generally have an essentially homogenous land cover type, making land cover characterization for each site straightforward. However, there are two potential research contexts where this simple strategy may be problematic: (1) medium and large sites where only portions of the physical area relevant for a recreational day-trip (say) have been affected by wildfire (and potentially to differing degrees) and (2) large and very large sites (e.g., Yellowstone National Park) where the ecosystems themselves are vastly different but are technically within one “site” per the conventional definition. When wildfire has affected the “site,” then there may be parts of the site that are affected whereas others may remain untouched. In the literature, wildfire has been included in recreation demand models in a variety of ways (e.g., Englin et al. (2006)) and there have been concerns about how to account for individuals who visit multiple sites on a single trip (e.g., Parsons et al. (2021)). For example, Englin et al. (2006) valued Jasper National Park and its recovery following a forest fire. They used the Canadian National Parks Inventory data to break down three main types of vegetation (lodgepole pines, spruce-juniper and alpine meadows). They also account for the ages of tree stands, and time since the wildfire event. Data and computing power exist nowadays that make it feasible to account for site attributes (e.g., landcover type, wildfire burn extent, and burn severity) in a more granular fashion on a large spatial and temporal extent. However, the question then becomes how to uniformly include site attributes such as land cover and burn severity for medium and large sites where portions may be unaffected by wildfires? Medium and large sites may have to be treated differently, as collection of smaller subunits with essentially homogeneous land cover, if the researcher is to account for different portions of the site. This strategy is related to modeling visitation to multiple different sites during a single excursion (e.g., Parsons et al. (2021)). We will present a framework for different alternatives concerning how such attributes may be included in recreation demand and visitation models, including consideration of alternative strategies for incorporating heterogeneous medium and large destination “sites” into a discrete-choice model via different degrees of spatial aggregation.

 

Title:               Is the Wellbeing of Surviving Wildlife Welfare Relevant—And Should We Care? An Application to North Atlantic Marine Plastics

Authors:         Robert J. Johnston, Tobias Börger, Keila Meginnis, Nick Hanley, Tom Ndebele, Ghamz E. Ali Siyal, Nicola Beaumont, Frans de Vries

W5133 Obj:   1, 2

Presenter:      Robert J. Johnston

Email:             rjohnston@clarku.edu

Abstract:        People’s willingness to pay (WTP) for improvements to wildlife populations has been established by decades of research, largely applying stated preference (SP) methods. This work provides evidence of WTP for outcomes such as increased population sizes, reductions in mortality and biodiversity improvements. With few exceptions, this research focuses on measures that are related—directly or indirectly—to whether organisms survive. For example, measures such as population sizes, mortality, and biodiversity are fundamentally related to survival. In contrast, potential welfare effects linked to the quality of life of surviving organisms are often overlooked. Consider a pollutant such as marine plastics that causes mortality of marine species but can also cause non-fatal suffering for surviving animals. In such cases, pollution reductions may improve the survival and quality-of-life of affected species. If people care about animal wellbeing, SP welfare elicitation that overlooks these values—for example eliciting WTP for mortality reductions alone—may misrepresent benefits in possibly complex ways. The idea that people might value animal welfare improvements is not new. Research demonstrates that individuals are willing to pay for improvements in farm animal welfare, for example. Yet SP research evaluating WTP for environmental policies that benefit wildlife almost universally proceeds under an implied assumption that valid welfare measures may be obtained through WTP estimates linked to survival or existence alone.

 

To consider the implications of this common practice, this article develops a theoretical model and discrete choice experiment (DCE) to evaluate whether and how the omission of potentially relevant information on species wellbeing influences the validity of WTP estimation, drawing from a case study of marine plastic reductions in the North Atlantic. Empirical evaluation proceeds via a DCE that evaluated preferences of US households for programs to reduce plastics pollution in the North Atlantic. Two otherwise-identical DCE treatments were developed, both of which included an attribute that quantified reductions in anticipated mortality of marine species.

The first treatment included an attribute that quantified additional reductions in non-fatal harm to the same species. Scenarios in the second treatment omitted this attribute along with any mention of non-fatal harm. The questionnaire was implemented during 2022 over a sample of households in US Atlantic coast states, yielding a final sample of N=4,681 responses.

 

Random utility models were estimated via mixed logit in WTP space with structural comparisons across DCE treatments. Results support hypotheses from the theoretical model. Omitting information on changes to the wellbeing of surviving wildlife from SP scenarios that convey other environmental impacts leads to profound impacts on welfare estimates. For example, such omissions cause WTP estimates for mortality reductions to roughly double in magnitude—presumably because respondents (when not given information on wildlife harm) speculate that mortality reductions are accompanied by equivalent reductions in non-fatal wildlife suffering. These effects imply that welfare estimates linked to wildlife survival in the literature may be unknowingly affected in non-trivial and theoretically predicable ways by respondents’ speculations about concomitant impacts on species wellbeing that might also occur.

 

SESSION 5: Elicitation Improvements

Title:               Revealed Preferences from Voluntary Contributions

Authors:         Zhenshan Chen, Stephen K. Swallow

W5133 Obj:   2

Presenter:      Zhenshan Chen

Email:             zhenshanchen@vt.edu

Abstract:        We introduce an approach to modeling a binary-choice contribution solicitation for public goods. This approach recovers the compensating variation of the underlying public good by integrating nonparticipation, free riding, and warm glow factors in an expected utility framework. Monte Carlo simulation suggests the proposed approach reliably recovers the willingness to pay measures with only moderate bias, and the estimates are more accurate if the provision probabilities are elicited. We also provide an empirical example to show how the proposed approach can be applied to real donation datasets. Providing the basis of donation serving as a value elicitation tool, this study potentially opens a novel area for future research valuing public goods.

Being natural and familiar to the public, voluntary contributions have the potential to function as a value elicitation tool for certain public goods. A typical value elicitation tool for public goods is the stated preference approach, which was often criticized (e.g., Hausman, 2012) for involving hypothetical bias. Eliciting value measures from voluntary contributions (i.e., a revealed preference approach) could mitigate such concerns, serving as an alternative to the widely adopted approach to establish consequentiality in a stated preference survey (Vossler, Doyon, and Rondeau, 2012). Moreover, when raising tax might be unjustified or potentially widely protested for certain applications, voluntary contributions could serve as an effective payment vehicle. As voluntary contributions have been widely used to fund public goods and employed in “stated” preference studies (e.g., Champ and Bishop, 2006), they form a rich but underutilized data source on public good valuation. This study establishes an integrated framework showing how voluntary contributions can serve as a value-elicitation tool for public goods.

We establish an integrated framework showing how to estimate WTP from voluntary contribution data. We investigate a clearly defined public good, which unambiguously offers potential benefits to a certain group of people. Our model assumes that an individual doesn’t exactly know about others’ compensating values or the distribution of value within the population, and his decision is based on subjective beliefs regarding the probability of provision, which is then built on certain heuristics about the features specified in the solicitation. The likelihood function derived from the framework is presented below, where  denotes the probability of having a positive potential donation (under a random utility model framework) and  represents the probability of participation.

;                                       (1)

.       (2)

We present two applications with this framework: one is a Monte Carlo simulation, the other is an empirical analysis based on a real voluntary contribution solicitation (the Bobolink Project, see Swallow, Anderson, and Uchida, 2018).

Preliminary Results and Discussion

The Monte Carlo simulation shows that the utility parameter estimates from the proposed framework are generally centered around the true parameter value. As the sample size goes up to a certain level (e.g., pass 5000 total elicitations in our setting), the dispersion of the estimates considerably decreases, and the proposed estimation becomes reliable. The results from the Bobolink application show that the estimated WTP is considerably higher than WTD, although the relationship might be reversed when WTP is small and the warm glow effect dominates the donation behavior (Figure 1).

Figure 1. Predicted WTD vs. WTP

Note: Each point represents the predicted WTP and WTD for one individual.

 

Title:               Consequential Design for Valuing Private Goods

Authors:         Craig E. Landry, Twinkle Roy, Ben Campbell, Greg Colson, Eileen Schafer

W5133 Obj:   2

Presenter:      Craig E. Landry

Email:             clandry@uga.edu

Abstract:        Economic valuation can provide vital information on benefit flows associated with public goods, and stated preference methods have come a long way in providing more reliable and robust data on these values. In some cases, however, analysts wish to focus on the value of private goods that exhibit complementarity with public goods (e.g., recreation trips, organic food, green energy products). There has been considerably less research to bias and validity of stated preference methods in this domain. We propose an extension of the ex ante, consequential design for public goods to attenuate hypothetical bias in valuing private goods that complement environmental quality.

 

Our variant of consequentiality attempts to harness other-regarding preferences by identifying other parties that maintain a stake in results of the valuation exercise.  For example, indicating that one would take a trip to the Gulf of Mexico in the wake of the Deepwater Horizon oil leak may provide a warm glow to the respondent, but when the subject is told that survey results could be used to make costly stocking and hiring decisions by hospitality businesses in the Gulf region, the subject may undertake greater effort to evaluate a prospective trip, thus providing more realistic responses.

 

Using experimental methods, we compare valuation results for private goods (with complementary public goods) that highlight the potential for socio-economic consequences of others. In the first experiment, we assess the economic value of locally grown herbs, highlighting economic and environmental benefits that can accompany their trade. We split the sample and use an information treatment to convey potential negative impacts of hypothetical bias on local farmers. Relative to the control (no information on potential negative impacts on farmers), we find reduced WTP for locally grown herbs with consequential treatment, though the effect is only significant for those subjects that spent substantial time reading the treatment text (dwell time exceeds median for treatment).

 

In our second experiment, we assess the economic value of energy audits that rate respondents’ home energy efficiency and offer specific recommendations to improve efficiency in energy use. Again, we split the sample and use an information treatment that highlights the potential consequences of hypothetical bias. The information treatment indicates that “… national supplier is working with local military veterans groups that can supply experienced contractors to assist in the energy assessment partnership. Your answer to the following question could be used by these veterans groups to make investment decisions to provide for contractor services. If you provide inaccurate information, they may make poor decisions about costly investments.” In addition to testing for the influence of consequential treatment on WTP for energy audits, we explore other-regarding preferences as a mechanism. We include a standard instrument for empathy and explore whether these measures correlate with treatment effectiveness.

 

Title:               Correcting Voters’ Cost Misperception of Public Good Referendums

Authors:         Corey Lang, Shanna Pearson-Merkowitz

W5133 Obj:   2

Presenter:      Corey Lang

Email:             clang@uri.edu

Abstract:        Direct democracy is an important determinant of local public goods; US voters annually authorize billions of dollars in public spending through bond and tax referendums. An open question however is to what extent these mechanisms supply public goods that match preferences of constituents. A neoclassical model of voting behavior would stipulate that a referendum states a quantity of public good and a cost, and the voter performs a cost-benefit calculation to decide their choice. However, if voters misperceive costs, then they may vote in discordance with their preferences.

 

This paper examines referendum voting behavior and understanding of referendum costs. Specifically, we conduct an exit poll focused on a statewide bond referendum, and couple this survey with randomized information about true cost implications of the bond. The goals of the project are to understand cost perception, assess if information provision is effective at increasing cost understanding, and examine if correct cost information influences voter choice.

 

Data

Our research design couples randomized information provision and an exit poll survey. We focus on a 2022 Rhode Island statewide bond referendum, the $50 million Green Economy Bonds (GEB), which proposed spending on a number of environmental priorities (clean water, land conservation, etc.). We trained 95 undergraduate students and sent them to 38 polling locations on Election Day. Half of the locations were “treated”, students held signs and verbally communicated factual information about average household cost of a specific bond referendum on the ballot. Specifically, we communicated that the cost of GEB would be about $7/year for the average household. These students engaged voters entering the poll. At all locations, different students conducted an exit poll. The survey was anonymous and self-administered, which mitigates social desirability bias. The survey asked about votes for governor, referendum approval, demographic information, and importantly how much the bond would cost their household in additional taxes if it passed. We created three versions of the survey that differed only in the range of cost answers. Our reasoning was that if cost guesses are dependent on the survey version (ie, anchoring bias), then that is a strong indicator that people are ignorant about costs. We collected 2,057 completed surveys on Election Day.

 

Results

There are three sets of results. The first result is that voters are generally uninformed about personal costs of bond referendums. Examining voter respondents at control locations, those without the information treatment, the distribution of cost guesses was fairly uniform, with the highest probability choices being the lowest and highest costs regardless of survey version. Cost guesses were highly dependent on survey version – when indicator variables for survey version are added to a model of cost guess, R-squared increases from 0.05 to 0.45. Lastly, only 5.4% of respondents chose the correct cost answer, less than would be expected with random guessing.

 

The second result is that the information treatment worked, but was not universally effective, nor homogenously effective across groups. Survey takers at treated location were 13.2 percentage points more likely to choose the correct cost guess. While that is a 244% increase over baseline, clearly the majority of voters are not receiving the information, which is consistent with observed voter behavior of not wanting to engage with canvassers when entering a poll. Importantly, some voters were seemingly more receptive to information. Treated respondents who voted for the Democratic nominee for governor were 18.2 percentage points more likely to be accurate, while treated respondents who voted for the Republican nominee for governor were only 11.2 percentage points more likely to be accurate. Even though the information was presented in a non-partisan manner, because the information was about a referendum more favored by Democrats, it could have been construed in a partisan manner. This interpretation and Republican’s rejection of information is consistent with ideas of motivated reasoning (Kahan, 2013).

 

The last set of results investigates whether correct information affects voter choice. We estimate an instrumental variable model using treatment as an instrument for accuracy. Across a series of models, the effect of accurate cost beliefs is always a statistically insignificant determinant of voter choice. That being said, models that allow the effect of accurate information to be different across partisan groups suggest divergent effects. When knowing the correct cost information, Democrats are more likely to vote in favor of the Green Economy Bonds but Republicans are less likely to vote in favor of it. This result may again point to motivated reasoning or truly different interpretations of the information. For example, Democrats may interpret the information as a small cost and worth increasing their support. In contrast, the cost information may prime Republicans to think about taxes and decrease support.

 

Title:               Does excludability reduce free-riding in stated and real charitable donations?

Authors:         Jerrod Penn

W5133 Obj:   2

Presenter:      Jerrod Penn

Email:             jpenn@agcenter.lsu.edu

Abstract:        A known challenge of using donations as a payment mechanism to elicit willingness to pay for a public good is potential free-riding (Champ et al. 1997). The free-rider may strategically underbid knowing that others may donate towards the provision of the good. Yet studies continue to use a donation payment mechanism because of its plausibility relative to compulsory payment schemes such as taxes or fees. One suggested means of mitigating free-riding is to describe the good as quasi-public rather than a pure public good such that there is at least partial excludability. For example, Ready, Champ, and Lawton (2010) elicit donations for wildlife rehabilitation, that if a donation is not made, then an animal will be turned away. While the good is non-rival in that everyone benefits from an animal’s rehabilitation, it is excludable in that a person who positively values animal rehabilitation has an incentive to donate, and the misincentive to free-ride is removed.

As far as we know though, this assertion remains untested; we do not know whether such an excludability statement has the impact of increasing WTP.  The purpose of this study is to conduct an empirical test of whether divisible donations can increase WTP. We conduct this test using a split-sample experiment contained with a discrete choice experiment for donations to several charities. This choice experiment is also implemented as both a hypothetical and real elicitation to verify the impact of excludability in both hypothetical and real WTP.  This study can provide evidence of whether stating excludability can reduce free-riding in donation-based payment vehicles, increasing stated preference practitioners’ confidence in its use.

Data and Methods

The choice experiment is based on donations to one of three charities, Feeding America, the National Forest Foundation, and the Against Malaria Foundation. These three were selected since each provides information supporting a claim of excludability. Feeding America states $1 equals ten meals provided, the National Forest Foundation states $1 equals one tree planted, and the Against Malaria Foundation states that $2 equals one additional bed net (to protect against mosquitoes). We are nearly finished collecting 600 responses, with roughly half being provided this impact per dollar information (excludability treatment) and the remaining half without (traditional free-riding control). Our hypothesis is that WTP should increase in the excludability treatment since the incentive to free-ride is reduced. Each choice requested a donation of $1 to $5 to the charities and participants answered six choice sets. For the real treatment to occur, all respondents received an extra $5 in participation incentives. Real DCE respondents were informed that one of the six choice sets will be selected as real and their decision to donate or not donate would be carried out and potentially deducted from their extra $5. As is common in DCE, we will model the data using conditional, mixed, and generalized multinomial logit models.

 

SESSION 6: Marine and Coastal

 

Title:               Wetlands and Water Quality: Evidence from the Coastal US

Authors:         Mani Rouhi Rad, Yukiko Hashida

W5133 Obj:   1

Presenter:      Mani Rouhi Rad

Email:             Mani.RouhiRad@ag.tamu.edu

Abstract:        Wetlands are critical ecosystems that provide a range of ecosystem services, including protection from flooding during storms and improving water quality and wildlife habitat. Wetlands have been shown to reduce river nitrate concentration in intensively managed agricultural watersheds, for example (Hansen et al. 2021). The protection for wetlands under the Clean Water Act (CWA) has gone through changes over time. Specifically, the Clean Water Rule (CWR) protected wetlands that were not adjacent to streams. However, the ruling from the Supreme Court meant that wetlands that do not directly connect to streams protected under the CWA do not fall under the regulatory authority of the Federal government. In a recent study, Taylor and Druckenmiller (2021) estimated the benefits wetlands provide for flood mitigation and found that non-coastal wetlands provide significant flood mitigation benefits. However, the wetland likely has different spatial effects on water quality, another important ecosystem service benefit. This could lead to a different set of policy recommendations (e.g., where the wetland change has the largest negative impact; where to target the efforts to mitigate wetland conversions, etc.). To evaluate the impact of policy change, we need to understand the economic value of wetlands that account for all ecosystem benefits they provide. In this study, we estimate the effects of changes in wetland extent on water quality using land cover and water quality changes in the coastal United States between 1996 and 2016. Our empirical analysis exploits the spatial direction of river flows to estimate the causal effects of wetland area changes on water quality. The wetland extent changes as a result of conversion to other uses (e.g., farmland or development). One empirical challenge is spatially modeling the effect of wetlands on controlling water quality while accounting for spatial spillovers. We explicitly consider the geospatial relationship between the river/stream and wetland network in our estimation, including the direction of the river flow to account for the extent of the wetlands’ water purification service. We estimate the impact of change in wetland extent (change in area size) on downstream water quality by comparing changes in water quality over time, between downstream and upstream of the watersheds, and across watersheds with different wetland extent. In our preliminary empirical analysis, we focus on changes in nutrient pollution and find that an increase in wetland area results in decreases in phosphorus and nitrogen, two of the main sources of nutrient pollution in water bodies. Protecting water quality is a main purpose of the CWA. Our analysis provides insights into the importance of wetland protections under the CWR. Furthermore, the estimates of the effects of wetland area on water quality allow us to estimate the loss of wealth in regions where wetlands have been lost.

 

Title:               Valuation of Oyster Reef Restoration

Authors:         Dan Petrolia, Seong Yun, Freedom Enyetornye, Zhenshan Chen

W5133 Obj:   2

Presenter:      Dan Petrolia

Email:             d.petrolia@msstate.edu

Abstract:        A contingent valuation survey was designed and administered to households in the five Gulf Coast states to understand public opinion and estimate the value of oyster reef restoration efforts. Methodologically, the survey was an occasion to implement one thing and test another. First, we implemented videos in the survey, rather than written text, as recent research has argued that this makes it easier for respondents to understand and follow the information provided. Second, we tested whether uncertainty in the scenario outcome impacts responses. Specifically, we implemented a split-sample design where half of the respondents received a "certain" scenario whose outcome was depicted as a fixed value (for example, an expected oyster harvest of 2.6 million pounds per year over the next 10 years for the State of Florida without the project versus a 1-million pound increase with the project), whereas the other half received an "uncertain" scenario whose outcome was depicted as a range (for example, an expected oyster harvest of between 2.0 and 3.3 million pounds per year without the project versus an increase of between 2.7 and 4.9 million pounds with the project). Harvest was chosen as the preferred metric as a signal of reef health/abundance for its acceptance by respondents and for its data availability. The survey instrument was programmed and administered using the Qualtrics platform, and a convenience sample of respondents was obtained from Qualtrics. A total of 6,338 responses were collected during October 2022. The design featured five bids, proposed as a $25, 50, 100, 250, or 500 one-time tax per household collected on state tax returns for AL, LA, and MS and as a fee collected by local counties for FL and TX (who have no income tax). Two scales our restoration were tested, a "low" and "high", specific to each state's historical harvest levels. The uncertainty treatment was texted on FL and TX respondents only, given the larger number of respondents available for interviews. Initial results indicate widespread support for oyster restoration (70% willing to pay some amount of money), though this proportion drops significantly when adjusted for certainty of response. We find greater support among those that eat oysters or that go fishing. Votes follow the law of demand as expected, with the proportion of yes votes declining as bid increases. Scope effects appear to be minimal, which is of some concern.

 

Title:               The Nature of Discrimination in Recreation Decision Making

Authors:         Jesse D. Backstrom, Richard T. Woodward

W5133 Obj:   3

Presenter:      Jesse D. Backstrom

Email:             jbackstrom@txstate.edu

Abstract:        Introduction: In the economics literature, discrimination has been explored in numerous settings, including employment (e.g., Becker 1957), policing (e.g., Donohue and Levitt 2001), jury decisions (e.g., Anwar et al. 2012), housing markets (e.g., Bayer and McMillan 2006), and environmental outcomes (e.g., Banzhaf et al. 2019), among others. In this paper, we explore how racial and ethnic (RE) considerations influence a common task: recreation destination choice. We propose RE distance as a new driver and find economically significant aversion to diversity in our sample of marine recreational fishermen. Not only do our findings have implications for coastal communities of color, but we believe our approach lends a unique opportunity to enrichen our understanding of the ways in which subtle biases influence day-to-day behavior.

Methods:

Our focus in this paper is to explore how differences between the RE composition of a potential recreation site destination and that of a trip origin (RE distance) affects the site choice decisions made by marine recreational anglers. We make use of the standard site-choice econometric model with site-level fixed effects. As developed by McFadden (1974), our conditional-logit model assumes that the ith participant seeks to maximize her anticipated utility from choosing site j=1,…,J. We model participant i’s expected utility at site j, Vij, as a function of the cost to reach the site, cij, covariates that vary across participants and sites, xij, and a set of site-specific constants that measure other time-invariant site characteristics, dj.

Data:

Our travel cost models are estimated using site choice data from the Marine Recreational Information Program (MRIP), the primary program that counts and reports catch and effort data for public fishing sites along coastal U.S. waters. The complete MRIP dataset consists of thousands of point-intercept surveys conducted in six two-month waves throughout each year. Anglers are surveyed at the close of their fishing day to obtain information on their trip and fish catch. We narrow our focus to a set of about 50,000 trips taken by private boat anglers that took trips to sites in Mississippi, Alabama, and the Florida gulf coast over 2013-2016. Our RE data for all trip origins and fishing sites come from the US Census.

Results:

We find that anglers in our sample have a positive WTP to avoid sites with Black and Hispanic proportions that are more predominate than in their home ZIP Code. Our dataset also allows us to investigate the nature of the observed discrimination. First, following the predictions of statistical discrimination theory, we exploit data on reported fishing experience and find evidence of lower levels of aversion to diversity from more avid anglers (our proxy for experience). Second, after sorting anglers into two groups – those originating from ZIP Codes with an above (below) national-median White population proportion – we find evidence of taste-based discrimination as the site choice decisions of anglers coming from less diverse ZIP Codes are more sensitive to RE considerations than those of anglers originating from ZIP codes with a greater degree of diversity.

 

Title:               Impacts of the COVID-19 Pandemic on Willingness-To-Pay for Environmental Goods and Services: A Case Study of Coastal Wetlands in Tampa Bay

Authors:         Julian J. Hwang

W5133 Obj:   1,3

Presenter:      Julian J. Hwang

Email:             Julian.Hwang@mail.wvu.edu

Abstract:        The COVID-19 pandemic has impacted not only public health and economies around the world, but also various aspects of daily lives and the society.  For market goods, impacts of such shocks would be realized in the market.  For nonmarket goods such as environmental goods, however, such impacts cannot be realized.  This paper identifies two potential effects of COVID-19 on preferences for environmental goods: income effect and preference shift.  It empirically tests 1) if willingness-to-pay for an environmental good is impacted by COVID; and 2) if impacted, how much is attributed by income effect and/or preference shift, using a survey data that was collected in the midst of COVID-19 to elicit Floridians’ preferences for coastal wetlands in Tampa Bay.

 

SESSION 7: Stated Preference Methods

 

Title:               Unpacking Differences of Who is in the Sample Based on Survey Contact Mode

Authors:         Amila Hadziomerspahic, Sonja Kolstoe, Steve Dundas

W5133 Obj:   2

Presenter:      Amila Hadziomerspahic

Email:             Hadziomerspahic.Amila@epa.gov

Abstract:        In nonmarket valuation, surveys are designed to ask the who, what, when, where and why for a population of interest to understand preferences over environmental goods, services and policies. Sample representativeness remains an issue for survey-based research due to declining response rates with traditional contact modes (e.g., mail, phone) and the uptick in use of quota-based online samples. Here we contribute to the sample selection literature by asking the question - Are there systematic differences of who is in the sample, based on how the respondent was contacted? Quota-based panel sampling yield samples with marginal distributions matching the population of interest, but they may or may not be representative. Address-based sampling provides an opportunity to get a random sample of the population, but often low response rates make sample selection methods necessary.

We build upon prior work in this area (e.g., Cameron and DeShazo, JEEM, 2013; Kolstoe and Cameron, Ecol Econ, 2017; Johnston and Abdulrahman, JEEP, 2017; Cameron and Kolstoe, Land Econ, 2022) by unpacking the differences in sample composition relative to the population based on how respondents were contacted to take a survey: via letters/postcard at their home address or an email from a survey company (e.g., Qualtrics). The objective is to investigate sociodemographic and response differences between an address-based sample – a traditionally preferred but more costly probability sample – with an opt-in online panel sample from Qualtrics – a less costly non-probability sampling alternative that may be subject to potential selection biases. The survey instrument was designed to estimate Oregonians’ total value (WTP) for erosion management conditional on differences in coastal armoring policy for private landowners using a contingent valuation framework. The difference in contact mode also changes the order of assumptions about whether the respondent had access to internet and when they knew the topic of the survey. In the address-based sample, letters were sent out, thus respondents knew of the topic first and then opt in by taking the survey online. Whereas in the quota-based sample, respondents are invited to take surveys as part of the commitment of being on the survey panel, thus they already have internet and learn the topic of the survey when they read the consent page. For both samples, participant and non-participant ZIP-code information is paired with tract-level data from the American Community Survey to model selection into each sample.

Preliminary results suggest that the percent of renters, the percent of individuals with a college education, and the percent of individuals below the poverty level influenced whether individuals contacted in the address-based sample frame responded to the survey. We are currently investigating additional factors that could impact selection into the sample including broadband internet speed and the availability and uptake of gig economy jobs. We hypothesize that proximity to the coast, high broadband speeds and the lack of gig economy substitutes (to online survey taking) could be contributing to the overrepresentation of coastal county respondents in the Qualtrics panel sample frame. We also anticipate differences in WTP for erosion management between samples.

Title:               Using Observations of Marginal Willingness-to-Pay in Willingness-to-Pay Meta-Regressions for Benefits Transfer

Authors:         Matthew G. Interis, Seong Yun

W5133 Obj:   2

Presenter:      Seong Yun

Email:             seong.yun@msstate.edu

Abstract:        When using meta-regression for purposes of benefits transfer, the researcher must decide which observations from the literature to include in the meta-regression data set. There is a tradeoff between consistency of observations—in terms of what environmental change is measured and how and also in the theoretical value measure estimated—and the quantity of observations: the more consistent the observations are restricted to be, the fewer observations can be included in the regression. Some practitioners have adopted relatively strict consistency of observations, e.g. Hicksian measures of (non-marginal) willingness to pay (WTP) from stated preference methods, whereas others have combined less consistent measures together, e.g. Hicksian and Marshallian values from varying non-market valuation methods, using approaches varying from simple dummy controls to models with strong theoretical consistency but which are relatively challenging to implement.

We investigate a middle-ground approach in which Hicksian measures of both marginal willingness-to-pay (MWTP) and (non-marginal) willingness-to-pay (for water quality changes) are combined in a theoretically-consistent manner. Including MWTP observations allows us to increase our number of observations by about 48%, increasing our confidence in using frequentist estimation (rather than, say, Bayesian estimation which has been championed for use with small samples). We compare results across two main functional forms—a commonly-used linear form and a more structural form with desirable theoretical properties—and across several sub-variations of each regarding whether and how the two types of observations (WTP and MWTP) are combined: not at all, in a theoretically-consistent way, with simple dummy controls, or in a naïve manner ignoring the difference between the two entirely. We examine model performance both within- and out-of-sample.

We find that in all but a few cases, combining the two types of observations in a theoretically-consistent way performs best. The procedure is relatively easy to implement using non-linear least squares estimation.

Title:               Unintended Consequences of Removing Multiple Types of Unqualified Survey

Responses in Discrete Choice Welfare Analysis

Authors:         Qi Jiang, Jerrod Penn, Wuyang Hu

W5133 Obj:   2

Presenter:      Qi Jiang

Email:             jiang.1885@osu.edu

Abstract:        In stated preference valuation surveys, respondents might not give answers that would reflect their true preferences or behave strategically to influence the outcome of the survey, leading to an inaccurate estimate of the economic value of the good or service in question. Many factors may induce respondents to answer untruthfully. This paper focuses on three main reasons including protest, inconsequentiality, and inattention (Meyerhoff et al., 2014; Vossler and Evans, 2009; Borger, 2016). Protest describes a phenomenon where respondents may reject some components of the constructed valuation scenario by bidding zero regardless of their true positive demand. Inconsequentiality depicts a situation where respondents do not believe that their answers can potentially affect the outcome of the goods/services or the implementation of the policy in question. Inattention describes respondents who rush through the survey without fully understanding and considering the information or instructions. Instead of referring to these different types of behaviors separately, we refer to all such responses as unqualified responses.

Two important patterns of the literature emerge. First, each definition of unqualified responses following a specific behavioral reason is only identified by its own corresponding measurement/question. Removing unqualified responses from all possible sources may cause the removal of a substantial number of observations if the unqualified responses do not overlap. Additionally, many of these qualification questions have no other use than simply being a measure to check for qualification, taking up respondents’ survey time and effort. Therefore, our first research question is whether different types of unqualified responses are correlated. If so, we may consider one criterion instead of many to clean the data. Second, a common practice is to purge unqualified responses but previous literature suggests that the WTP estimated from the cleaned up sample can be significantly different from the one estimated from the uncleaned sample. Different combinations of definitions/standards to clean the sample might lead to significantly different welfare estimates (e.g., WTP), thus possibly causing inadvertent publication bias. Thus, the second goal of this paper is to compare the WTPs after cleaning the data following different combinations of definitions/standards of response removal criteria and shed light on possible publication bias.

Based on a contingent valuation survey with a split sample design including a hypothetical and a real treatment, we have three main findings. First, no large overlaps exist between protest, inconsequentiality, and inattention. Second, the WTPs estimated from the cleaned sample after excluding unqualified responses can be significantly different across different definitions/standards for qualification. Also, the WTPs could be dramatically inflated if we removed too many observations by using more stringent standards or excluding too many types of unqualified responses. This may generate opportunity for publication bias which might be insignificant for a single paper but otherwise severe collectively when considering all the literature. Third, hypothetical bias (HB), measured by the difference of WTPs between the hypothetical and real elicitation, remain stable if researchers apply similar exclusion criteria to both hypothetical and real data. However, the HB might be significantly inflated when only the hypothetical data are cleaned for unqualified responses.

Title:               Fat Tails and Willingness to Pay: Kristrom revisited

Authors:         Lynne Lewis, Leslie Richardson, John Whitehead

W5133 Obj:   2

Presenter:      Lynne Lewis

Email:             llewis@bates.edu

Abstract:        Best practices for estimating willingness to pay (WTP) from stated preference survey data typically recommend sensitivity analysis and the presentation of WTP values calculated using different methods (e.g. Bengochea-Morancha et al. (2005), Johnston et al. (2017). Both parametric and non-parametric methods are commonly used. Non-parametric methods are appealing since they are a less restrictive alternative to parametric models and do not rely on the analyst knowing or assuming the distribution of WTP. Commonly used approaches are the Turnbull (Haab and McConnel, 2002) and the Kristrӧm estimator (Kristrӧm; 1990; Boman, Bostedt, Kristrӧm, 1999).

 

Kristrӧm (1990) uses linear interpolation to characterize the distribution between bid amounts. This approach assumes that the distribution (i.e., survival) function is piece-wise linear between bid amounts (Haab & McConnell, 2002). This approach is appealing in that it allows for the estimation of the right tail of the WTP distribution. The challenge, however, is knowing what assumption to make regarding the choke price, or the point at which the probability of a “yes” response falls to zero. This choke price is somewhat arbitrary, typically calculated using either linear interpolation from the last two bid amounts, the approach taken in Kristrӧm 1990, or simply by truncating at the highest bid amount.

 

Both of these approaches can result in inaccuracies, especially with data sets for which a relatively large number of respondents say yes to the highest bid amount (fat tails). Parsons and Myers (2017) find that 60% of contingent valuation studies from 1995-2014 exhibit fat tails.

Richardson and Lewis (2022) present a combined approach for the calculation of the choke price. Using a large and very well-behaved data set, they utilize four different methods to calculate the choke price. First, they follow the Kristrӧm approach of simply calculating the slope using the last two data points. They then estimate a linear probability model of WTP to calculate a choke price, using the slope over the entire range of the bid curve (as suggested by Whitehead, 2017). This method results in a choke price that is lower than their highest bid. The third option is to simply truncate the Kristrӧm estimate at the highest bid amount. These methods result in choke prices of $2,300, $450, and $500, respectively. In addition to those approaches, they present a unique, slightly modified approach. They use the (steeper) constant slope from the linear probability model to calculate a choke price but extend it from highest bid amount. This results in a more reasonable choke price than simply calculating the slope using the last two data points, at which point the slope is very flat and results in a high choke price.

 

Richardson and Lewis’ results do not show a statistically significant difference between WTP calculated with a truncated Kristrӧm and the combined approach, likely because so few respondents said “yes” to the highest bid, but it begs the question of at what point might it make a difference? This paper explores that question.

 

First, we extend Parsons and Myers to determine how many stated preference data sets exhibit fat tails. Second, using both real and simulated data sets, we present a suite of Kristrӧm estimators and offer guidance for when each approach is most useful. Since it would be extremely rare, if not impossible, to capture the tail of the distribution exactly (the bid at which 0% of the respondents say yes), it seems important to explore whether or not an alternative approach is merited.

 

SESSION 8: Climate Change

 

Title:               Consumer Preferences For Battery Electric Vehicles Considering Energy Mix: A Choice Experiment Study

Authors:         Jamal Mamkhezri

W5133 Obj:   2

Presenter:      Jamal Mamkhezri

Email:             Jamalm@nmsu.edu

Abstract:        As the adoption rates of battery electric vehicles (BEVs) continue to grow in the United States, it is important to understand how consumers respond to these markets' features and the changes associated with using new technology. While many studies have investigated how consumers respond to vehicle attributes and related aspects of BEV ownership, there has not yet been a study that looks into the energy source as a BEV attribute or estimates consumers' willingness to pay (WTP) for clean electricity as a BEV fuel source in the U.S. This study aims to investigate the attitudes of U.S. drivers towards BEVs and the source of electricity that fuels them.

 

The study uses an online discrete choice experiment survey of 1,150 U.S. drivers to assess their WTP for clean energy as an attribute of BEVs. The survey also explores consumers' attitudes towards various policy incentives for BEVs, changes in the number of jobs, and ultimate changes in electricity costs. The data is analyzed using various conventional and advanced logit models to assess marginal WTP for each attribute and transportation plans. GIS is applied to calculate the spatial heterogeneity variables, the charging infrastructure exposure, and rural and urban areas within the U.S. This paper extends the literature by constructing marginal WTPs coupled with their confidence intervals at the individual level and in the WTP-space.

 

We find that there is considerable heterogeneity in preferences among decision makers, which cannot be explained by differences in sociodemographic variables. Results show that U.S. drivers have a positive WTP for increasing the share of BEVs in the transportation system and support the use of clean energy to power them. On average, respondents are willing to pay $1.24 for a 1% increase in the current BEV level. Respondents prefer renewable energy sources such as solar, wind, and hydro-power over nuclear energy as a clean energy source. Drivers are willing to pay $12 per month to replace the existing 15% nuclear power in the electric grid with renewable energy sources. They also support increasing the number of jobs associated with accelerated vehicle fleet decarbonization in the U.S., prefer tax credit incentives over free charging and free parking initiatives, and dislike the current transportation plan. Controlling for spatial and individual heterogeneity results in a divergence of values. Respondents who live in urban areas, are pro-environment, male, young, have high income and education level, and affiliate with the democratic political party are more supportive of BEV and clean energy policies. The study concludes that an efficient energy policy requires both technological efficiency and economic viability, as well as public acceptance.

 

Title:               Estimating the effect of climate change on outdoor recreation using short-run weather deviations and passively collected trip data for U.S. grasslands

Authors:         Kaylee Wells

W5133 Obj:   1, 2

Presenter:      Kaylee Wells

Email:             kkwells2@illinois.edu

Abstract:        In 2021, outdoor recreation activities accounted for about 2% of U.S. GDP (BEA 2022) in addition to providing other mostly unobservable non-market goods. The contribution of outdoor recreation to society’s output and welfare is likely to be impacted by climate change. Despite recent progress, the relationship between weather, outdoor recreation and climate change remains poorly understood. The major barrier to estimating the causal relationship between outdoor recreation and weather/climate is often data availability. The data must be temporally and spatially explicit and cover a range of sites and time periods; requirements that are generally not met with standard survey methods. Recent research on weather/climate and outdoor recreation has used multi-year, daily bicycling trip data from a bikeshare company (Chan and Wichman, ERE, 2020) and multi-year fishing data from a government agency (Dundas and von Haefen, JAERE, 2020). This research builds on Chan and Wichman (2020) and Dundas and von Haefen (2020) to estimate a causal model of the relationship between grassland recreation and weather using passively collected trip data from StreetLight Data. StreetLight data uses information from car GPS and cell phone location tracking with machine learning to predict the number of trips made to a site. I will use the predicted number of car trips to a subset of U.S. grasslands between January 2016 and April 2022 as the basis for this analysis. Grasslands offer an interesting case study because we know society values the recreational opportunities they provide (Wells, AAEA Poster, 2022) and those recreational activities are particularly susceptible to the impacts of climate change. Increasing heat is likely to be the source of most of these impacts as grasslands offer few options for visitors seeking to mitigate the heat (e.g., finding shade under a tree). This research is in early stages, and I plan to present my proposed model and summary statistics for the trip count data.

 

Title:               An econometric analysis of prescribed fire as a climate adaptation tool

Authors:         Yukiko Hashida, David Lewis

W5133 Obj:   1

Presenter:      Yukiko Hashida

Email:             yhashida@uga.edu

Abstract:        Prescribed fires can reduce future wildfires by removing excess fuels, improve habitat for some wildlife species, and recycle nutrients back into the soil. Since wildfires are expected to increase in frequency and intensity with climate change (Abatzoglou and Williams, PNAS 2016), fire mitigation will be a critical land management option for landowners to adapt to climate change. The purpose of this study is twofold: 1) empirically estimate the impact of long-term climate and short-term weather events on private forest landowners' decisions to conduct prescribed fires; and 2) estimate the impact of prescribed fires on wildfire occurrence. Much of the previous literature focuses on wildfire suppression issues, such as resource allocation and public health impacts (e.g., Bayham et al., ARRE 2022). However, an empirical economic analysis of prescribed burning decisions of landowners is important as the economic incentives that drive private landowner decisions on when and where to conduct prescribed burning are not well understood, and there is no empirical work linking private burning decisions to climate or weather outcomes. Our study fills this gap in the literature by linking private landowners' resource management actions, wildfire mitigation, and wildfire outcomes. Our model estimates prescribed burning as a climate adaptation tool that could potentially generate externalities by altering wildfire occurrence. Our methodology also allows us to study how implementing prescribed fires – and resulting wildfires – evolves with climate change.

We focus on the prime timber-growing region of the southeastern United States, where prescribed burns have been widely used. A key part of the analysis is prescribed fire permits data collected by the Non-Governmental Organization Tall Timbers, which serves as the empirical foundation to identify the location and magnitude of prescribed fire usage. Our data extends to ten states (AL, GA, FL, NC, MS, TN, SC, AK, LA, and VA) that have mandatory systems, creating 2.1 million permit records, which we subset to non-agricultural fires, resulting in about 1.6 million records over 11 years. We aggregate individual permit data into county acreages of total private forestland subjected to prescribed fires in each year between 2010 and 2021, resulting in a panel of about 10,000 county-year observations.

Our first model uses the inverse hyperbolic sign of prescribed fire acreage as the dependent variable, with independent variables consisting of annual realized weather variables (annual temperature, precipitation, and vapor pressure deficit (VPD)), forest types, soil productivity, stand age, stand volume, slope, elevation, and ownership type. We apply spatial and temporal fixed effects to control for unobservables. Our preliminary results provide strong evidence that prescribed fire acreage decisions are a function of annual climate realizations. In particular, results indicate that landowners respond to warmer and drier years by substantially increasing the application of prescribed fires, which is a robust result across multiple specifications. A one-degree C increase in temperature and a one-unit increase in VPD each result in about a 40 percent increase and a 4 percent increase in burned areas, respectively.

Next, we model wildfire events (burned acreages at the county-year level) as a function of prescribed burn. As the prescribed burns are likely endogenous, we explore a variety of instrumental variables such as forest sector wages, hunting permits, and extension support. Our preliminary IV results suggest that prescribed burns reduce wildfire occurrence, though the result is noisy. Our fully linked model provides evidence that landowner adaptation to climate change can raise prescribed burning, which can, in turn, provide positive externalities by lowering nearby wildfire rates.

Title:               Optimizing nonmarket ecosystem service flows from coastal landscapes under climate change

Authors:         Steven J. Dundas, Emma A. Gjerdseth, Sally Hacker, Peter Ruggiero, John S. Stepaneck, & Mohsen Taherkhani

W5133 Obj:   3

Presenter:      Steven J. Dundas

Email:             Steven.Dundas@oregonstate.edu

Abstract:        Integration of ecosystem services and their nonmarket values into spatial planning is a key hurdle to designing welfare-improving land-use policies. This is especially true in dynamic environments like beaches and dunes that are subject to changing conditions from natural hazards and the impacts of climate change. Using the Tillamook County coastline in Oregon, we combine primary nonmarket values with ecological and geomorphological production functions and underlying landscape conditions to estimate the overall flow and value of ecosystem services under the current spatial distribution. We then estimate the benefit flow changes across future scenarios, where we adjust the quantity of the services provided in response to climate change (e.g., sea-level rise, storm surges) and both public and private land-use adaptation strategies (e.g., shoreline armoring, habitat restoration areas). This latter exercise then searches for “coast-use” scenarios that maximize the social net benefits of alternative land management options, in the spirit of Polasky et al. (2008).

 

The key challenge for the proposed integrated model to inform natural resource management decisions is measurement of both prices and quantities related to coastal ecosystem services. For the nonmarket values, we use recent, Oregon-specific revealed and stated preference studies to estimate willingness to pay for coastal protection (Dundas & Lewis 2020, Gjerdseth & Dundas 2023), recreation and safe beach access (Hadziomerspahic 2022), dune restoration (Nguyen et al. 2023) and viewshed (Gjerdseth & Dundas 2023) along with recent federal government estimates of the social cost of carbon to value carbon storage in dunes. The quantities come from measurements derived from remote sensing data (e.g., beach width, dune height) and a probabilistic climate emulator based on total ocean level observations and predictions as well as direct in-field measurements of habitat quality and carbon storage by ecologists.

 

Ecosystem service prices and quantities are applied to individual land parcels. Each parcel is then grouped into decision units that face similar land-use choices and benefit flows are aggregated. Then using a combination of long-term coastal change, total water level and coastal dune recovery models, we examine the evolving probability of coastal change and the subsequent impact on ecosystem service flows on each decision unit at multiple future periods. Additional variation in future scenarios and ecosystem service provision are introduced with decision unit-specific policy options such as relaxing shoreline armoring restrictions, localized managed retreat, and expansion of habitat restoration areas. The outputs include future patterns of coastal land use optimized on the economic returns of nonmarket service flows through climate adaptation. We also provide insight on specific tradeoffs, like habitat restoration and carbon sequestration in dune management (i.e., more native habitat means less carbon storage), that may directly inform decision-making about climate adaptation on public shorelines.

 

[1] The cities are: New Haven, CT; Wilmington, DE; Augusta, GA; Louisville, KY; Columbia, SC; Grand Rapids, MI; Pittsburgh, PA; Charlotte, NC; Columbus, OH; and Nashville, TN.

04/01/2024