An Analysis of Farmer Preferences Regarding Filter Strip Programs - PowerPoint Presentation

An Analysis of Farmer Preferences Regarding Filter Strip Programs Greg Howard Work in collaboration with Dr. Brian RoeDepartment of AED Economics Ohio State UniversityNovember19, 2012howard.761@osu.edu

Lake Erie: A Big Freaking Deal Drinking water for 11 million peopleOver 20 power plants300 marinas in Ohio alone40% of all Great Lakes charter boatsOne of top 10 sport fishing locations in the worldThe most valuable freshwater commercial fishery in the world (Walleye capital of the world)Coastal county tourism value is over $10 billion (7 coastal counties = over 25% of Ohio 88-county total) Issues with nutrient pollution Phosphorous and Nitrogen 2 Howard: An Analysis of Farmer Preferences Regarding BMPs

Nutrient PollutionHigh nutrient loads in lakes can cause harmful algal blooms (HABs) Why are large algal blooms harmful?Released toxinsLower water qualityHypoxic (dead) zones3 Howard: An Analysis of Farmer Preferences Regarding BMPs

Lake Erie HistoryIn ‘60s, huge nutrient problems Cuyahoga river burns in 1969Clean Water Act passes in 1972P levels stable from 1970-75Improving from 1975-95How did we do it? Point source reductionsMajority of loading in 1970 was point sourceNow agriculture accounts for 2/3 of loading1995-present: Getting worse 4Howard: An Analysis of Farmer Preferences Regarding BMPs

Microcystis in Lake Erie The Microcystis- Anabaena bloom of 2009 was the largest in recent years in our sampling region 2011 …until 2011 Source: Tom Bridgeman, UT and Jeffrey M. Reutter, Ohio Sea Grant 5 Howard: An Analysis of Farmer Preferences Regarding BMPs

Government ResponseRegulationMarket-based Solutions Nutrient taxesNutrient trading programs (Ohio River Basin)Payment for Ecosystem Services (PES) programsPay farmers for implementation of Best Management Practices (BMPs)6 Howard: An Analysis of Farmer Preferences Regarding BMPs

Best Management PracticesSoil testing and variable-rate application Avoiding fertilizer application before storm events or in winterWinter cover cropsFilter stripsRetention areasConservation tillage/No tillField retirement7 Howard: An Analysis of Farmer Preferences Regarding BMPs

Where is the Economic Problem?Question facing government: How to make these programs better? 1. More effective practices2. Greater adoption rates (more acres enrolled)3. Lower cost8Howard: An Analysis of Farmer Preferences Regarding BMPs

More Specifically…How do farmer perceptions of filter strip effectiveness influence filter strip program choice? Do farmers exhibit substantial preference heterogeneity for filter strip programs?9Howard: An Analysis of Farmer Preferences Regarding BMPs

Perceptions of Filter Strip Effectiveness Ma, Swinton, Lupi, and Jolejole-Foreman (2012)Consider a series of cropping systems, and control for farmer perceptions of ecosystem services from a cropping systemQualitative, and possibly endogenousThis study uses a quantitative measure and instruments for perceived efficacy using a two-stage estimation 10 Howard: An Analysis of Farmer Preferences Regarding BMPs

Preference HeterogeneityLatent Class Analysis (LCA) allows for preference heterogeneity Farmers belong to one of several latent (unobserved) groupsFor each group, variables of interest (predictors) can have different marginal effectsCan use other variables (covariates) to inform class membership11 Howard: An Analysis of Farmer Preferences Regarding BMPs

Latent Class Analysis (LCA)Example: Effect of LeBron James endorsementSome people are more likely to buy a product if James endorses itOther people (Ohioans and New Yorkers) may be less likely to buy if James endorsesAssuming preference homogeneityLittle or no effect of endorsementLCA can capture differences12 Howard: An Analysis of Farmer Preferences Regarding BMPs

FindingsEveryone likes more money and less paperwork Majority are more likely to choose program if perceived efficacy is higherNo status quo biasMinority for whom perceived efficacy little or no impactLarge status quo bias13Howard: An Analysis of Farmer Preferences Regarding BMPs

Rest of the TalkSurvey and data ModelResultsImplications and conclusion14Howard: An Analysis of Farmer Preferences Regarding BMPs

SurveySent to 2000 Ohio corn and soybean farmers in Maumee watershed December-February 2012Tailored Design Method (Dillman 2007)Completed surveys entered to win a pair of OSU football ticketsPilot tested with farmersResponse rate ≈ 40%15 Howard: An Analysis of Farmer Preferences Regarding BMPs

SurveyQuestions regardingDemographic information Field characteristics “Consider one of your fields where runoff is a potential problem and where no filter strip exists…”PES program enrollmentPreferences regarding hypothetical filter strip programs16 Howard: An Analysis of Farmer Preferences Regarding BMPs

Survey 17Howard: An Analysis of Farmer Preferences Regarding BMPs

Survey 18Howard: An Analysis of Farmer Preferences Regarding BMPs

Model: Conditional Logit Probability that farmer n will choose a series of t policy alternatives i , conditional on the farmer belonging to class s: X is a policy alternative-specific variable Probability that farmer n belongs to class s: Z is a farmer-specific variable 19 Howard: An Analysis of Farmer Preferences Regarding BMPs

Variables (Alternative-specific) VariableDescriptionRangeAvg. Std. Dev.Payment Dollars per acre {0, 125, 175, 200, 250} 126.4 96.1 FS Width Filter strip width in feet {0, 25, 75} 33.4 31.1 Paperwork Hours per year {0, 2, 5, 10} 3.8 3.8 Years Length of program{0, 5, 10} 5.14.1FS EfficacyDecrease in probability of runoff[-90, 100]11.919.0Status Quo=1 if current program{0, 1}0.30.520 Howard: An Analysis of Farmer Preferences Regarding BMPs

Variables (Farmer-specific) VariableDescriptionRangeAvg. Std. Dev.Risk Tolerant = 1 if risk tolerant in farm practices {0,1} 0.38 0.38 High School =1 if high school education or less {0,1} 0.42 0.42 First Gen =1 if first generation farmer {0,1} 0.14 0.35 Norm Till =1 if engages in conventional tillage {0,1} 0.250.43Lake Erie AlgaeLevel of awareness for Lake Erie algae issues{0,1,2}1.110.69 Models including age, income, environmental stewardship, and whether farmergrows organic yield same results. 21 Howard: An Analysis of Farmer Preferences Regarding BMPs

Model: First Stage (Endogenous Efficacy) OLS with FS Efficacy as dependent variable and field-specific variables as independent variablesLatent Class Analysis used in 1st stage as wellIndependent variables are exogenous and correlated with expected FS EfficacyPredicted values for FS Efficacy are used in the 2nd stage estimation 22Howard: An Analysis of Farmer Preferences Regarding BMPs

Variables (Field-specific) VariableDescriptionRangeAvg. Std. Dev.Drainage Field has working drainage tile {0, 1} 0.88 0.33 Width 25 =1 if 25 foot filter strip {0, 1} 0.34 0.47 Width 75 = 1 if 75 foot filter strip {0, 1} 0.33 0.47 Slope < 2=1 if slope of field is < 2 degrees{0, 1}0.510.50Slope > 5=1 if slope of field is > 5 degrees{0, 1} 0.100.30 D 25-75Indicator variables for distance to the nearest surface water in feet {0, 1} 0.19 0.39 D > 75 {0, 1} 0.25 0.43 23 Howard: An Analysis of Farmer Preferences Regarding BMPs

Results: 1st Stage 3 Classes (40%, 40%, 20%)Class 1 and Class 2: Wider filter strips and absence of drainage tile increase efficacyClass 2 believe filter strips are much more effective than Class 1 (21 vs. 6)Distance to water and slope not significantClass 3: Filter strips do nothing, regardless of field attributes24 Howard: An Analysis of Farmer Preferences Regarding BMPs

Results: 1st Stage Classes Class 1: Most profit-driven (marginally significant)Class 2: Better educated, already enrolled in PES programsClass 3: Older, more risk averse25 Howard: An Analysis of Farmer Preferences Regarding BMPs

Results: 2nd Stage Coefficients 26Independent Variable Traditional AnalysisLatent Class Analysis Class 1 (70%) Class 2 (30%) Payment Positive Positive Positive FS Width Negative Negative ------- Paperwork Negative Negative Negative Years ------ ------- ------- Status QuoPositive-------Positive (Large)FS EfficacyPositivePositive-------Howard: An Analysis of Farmer Preferences Regarding BMPs

Results: Marginal Effect on Probability that Program is “Best” VariableClass 1 (70%)Class 2 (30%)Payment 0.0025*** 0.0019** FS Width -0.0039*** -0.0027 Paperwork -0.0257*** -0.0235** Years -0.0124 0.0016 Status Quo -0.1288 0.5047** FS Efficacy 0.0105** 0.0060* Observations 526 R20.7920*, **, and *** denote statistical significance at the 90%, 95%, and 99% levels, respectively27Howard: An Analysis of Farmer Preferences Regarding BMPs

Results: Relative Importance of Independent Variables Independent VariableTraditional AnalysisLatent Class Analysis Class 1 (70%) Class 2 (30%) Payment 38% 32% 27% FS Width 19% 15% 12% Paperwork 16% 13% 14% Years 2% 7% 1% Status Quo 11%7%30%FS Efficacy14%26%16%Howard: An Analysis of Farmer Preferences Regarding BMPs28

Results: Relative Importance of Independent Variables Independent VariableTraditional AnalysisLatent Class Analysis Class 1 (70%) Class 2 (30%) Payment 38% 32% 27% FS Width 19% 15% 12% Paperwork 16% 13% 14% Years 2% 7% 1% Status Quo 11%7%30%FS Efficacy14%26%16%Howard: An Analysis of Farmer Preferences Regarding BMPs29

Results: 2nd Stage Profiles Class 1Class 2Risk Tolerant* 26%15% High School 37% 45% First Gen 10% 17% Norm Till*** 16% 35% Lake Erie Algae: Somewhat aware 51% 55% Lake Erie Algae: Very aware* 31% 20% *, **, and *** denote statistical significance at the 90%, 95%, and 99% levels, respectively 30Howard: An Analysis of Farmer Preferences Regarding BMPs

ImplicationsHow do we improve adoption rates? Increase payments, decrease paperworkTarget those most likely to belong to Class 1Educate farmers on value of FSsHow do we lower costs?Decrease paperworkFocus on educationEducation on the benefits of filter stripsEducation on the impacts of nutrient pollution (Lake Erie, Grand Lake St. Mary’s, etc.) Howard: An Analysis of Farmer Preferences Regarding BMPs 31

Thank You! 32Howard: An Analysis of Farmer Preferences Regarding BMPsSupport provided by NSF Coupled Human and Natural Systems Program (GRT00022685)