Economics of Precision Farming

Marvin BatteVanBuren Professor Emeritus, OSU

Research Professor, University of Kentucky

Florian DiekmannAssistant Professor

University Libraries, OSU

What is Precision Farming?

A suite of technologies that allow:• Capture of data at an appropriate scale and time• Interpretation and analysis of that data• Implementation of a management response at an

appropriate scale and time.

Source: National Research Council

Precision farming technology components include:

• Global Positioning System technologies• Yield Monitoring devices• Grid or zone soil sampling• Georeferenced scouting for weeds/pests/disease• Variable rate fertilizer application• Variable rate herbicide application• Variable rate planting• Auto guidance or assisted guidance

Organization of my remarks:

1. Summarize results for the 2010 precision farming survey regarding adoption

2. Discuss an econometric model of precision farming adoption and evaluation by adopters

3. Speculate a bit on the future of this emerging technology.

Results of the 2010 Precision Farming Survey

•Statewide survey, $50,000 + gross sales•Mailed Questionnaire•3,000 contacts•40.4% Effective response rate• Larger farms were over-sampled to provide more observations of larger farms.

•All analyses are weighted using Census of Agriculture numbers to return estimates that are representative of the Ohio farm business population.

•These analyses are restricted to the corn belt region of Ohio (42 counties).

Sample descriptive statistics by Adopter and Non-adoper farmers

Item All farms Adopters Non-adopters

Percent of sample 100 45.3 54.7Farm size (cropped acres) 830 1202 523

Gross farm sales (%) $50,000-$99,000 28.3 12.5 41.4 $100,000-$499,999 52.1 52.9 51.5 $500,000-$999,999 10.4 16.6 5.2 $1,000,000 + 9.2 18.0 1.9Livestock Sales as Percent of Gross Income

10.3 6.1 13.9

Work off-farm (%) 34.6 33.4 35.6

Age of Respondent 58.6 55.0 61.5

College degree or more 26.7 30.8 23.4

Percent of Ohio farmers who had adopted various precision farming components, February 2010.

ComponentPercent adopting

Satellite GPS receiver 36.1Precision Guidance 32.7Yield Monitor 31.0Georeferenced grid soil sampling 29.1Variable Rate Application of Lime 24.7Variable Rate Application of Potassium 24.1Variable Rate Application of Phosphorus 23.5Variable Rate Application of Nitrogen 7.1Variable Rate Application of Manure 2.3Percent who have adopted one or more of above 45.3

Percent of Ohio farmers who had adopted various precision farming components, 2007and 2010.

Component2010 2007

Satellite GPS receiver 36.1 21.1Precision Guidance 32.7 25.2Yield Monitor 31.0 25.7Georeferenced grid soil sampling 29.1 23.5Variable Rate Application of Lime 24.7 18.4Variable Rate Application of Potassium 24.1 17.9Variable Rate Application of Phosphorus 23.5 17.8Variable Rate Application of Nitrogen 7.1 10.8Variable Rate Application of Manure 2.3 --Percent who have adopted one or more of above 45.3 45.3

Percent adopting

Adoption of various precision farming components by farm sales class.

Measure$50,000 -

99,999100,000 - 499,999

500,000 - 999,999

Over $1,000,000

Satellite GPS receiver 11.3 36.2 62.6 82.6Precision Guidance 10.0 31.0 60.7 81.1Yield Monitor 10.0 26.5 64.0 84.7Georeferenced grid soil sampling 11.3 29.0 49.1 62.1Variable Rate Application of Lime 10.1 23.4 41.7 57.4Variable Rate Application of Potassium 8.9 23.4 41.2 55.3Variable Rate Application of Phosphorus 10.1 22.6 39.3 52.1Variable Rate Application of Nitrogen 2.5 6.9 10.9 17.9Variable Rate Application of Manure 1.3 2.0 3.3 5.8Percent who have adopted one / more of above 20.0 46.0 72.6 88.5

Adoption Percent

Net Benefits of the Total PF System

For your farm situation, are the total benefits of the overall precision farming system greater than the total costs of this system? (Circle ONE answer only)

1. Yes, Benefits are significantly greater than Costs2. Yes, Benefits are slightly greater than Costs3. Benefits and Costs are about equal4. No, Costs are slightly greater than Benefits5. No, Costs are significantly greater than Benefits

Net Benefits of the Total PF System, 2010

Net Benefits of the Total PF System, 2007 vs. 2010

Net Benefits of the Precision Farming System by Survey Year

0

10

20

30

40

50

60

70

80

Costs Exceed Benefits Costs, Benefits Equal Benefits Exceed Costs

1999200320072010

Net Benefits of the Total PF System, by Farm Gross Sales Class, 2010

Net Benefits of the Total PF System by level of operator education

Summary Observations• “Simple”, turn-key technologies are adopted more

quickly.• Large farmers are adopting more quickly

– Spreading of fixed costs of investment, knowledge development

• Small farmers are adopting VRT methods, but often contract out the entire process.

• Small farmers typically are pleased with the benefit / costs of the PF system.

• Over time, farmers have indicated higher Benefit/Cost. Is this learning or technology improvement?

Part II – Econometric model of PF adoption and evaluation

• A double-hurdle model– Phase 1 – what explains PF adoption– Phase 2 – For those who have adopted, what

explains satisfaction with the PF system• A variant of the Heckman sample selection model was

tested.– Phase 1 – binomial probit model– Phase 2 - An ordered probit model

• Correlation of error terms for the two models is not significantly different than zero.

Variable Mean Std.Dev. Min Max NUSE_PF 0.45 0.50 0 1 845PF_EVAL 2.02 0.87 0 3 465Operator Age 58.57 11.75 19 90 839Post High School Education (=1) 0.43 0.50 0 1 845Operator works part-time away from the farm (=1) 0.11 0.31 0 1 837Operator works full-time away from the farm (=1) 0.24 0.43 0 1 837Cropped acres (1000 acres) 0.83 0.85 0.07 8.4 835LIVE_PCT 10.32 22.96 0 100 801COM_USER (=1) 0.63 0.48 0 1 755USE_GPS (=1) 0.82 0.39 0 1 465USE_YM (=1) 0.69 0.46 0 1 465YM*GPS (=1) 0.51 0.50 0 1 465GRID SAMPLING (=1) 0.53 0.50 0 1 451ZONE SAMPLING (=1) 0.14 0.35 0 1 451VRT_PK (=1) 0.55 0.50 0 1 461LIGHTBAR (=1) 0.48 0.50 0 1 465AUTOSTEER (=1) 0.40 0.49 0 1 465

Sample statistics for dependent and independent variables

VariableEstimated

Coefficient P[|Z|>z]Marginal Effects P[|Z|>z]

Constant 0.6731 0.075 * 0.263 0.068 *Operator Age -0.0286 0.000 *** -0.011 0.000 ***Post High School Education (=1) 0.1873 0.096 * 0.070 0.094 *Operator works part-time away from the farm (=1) 0.1567 0.376 0.057 0.362Operator works full-time away from the farm (=1) -0.3912 0.006 *** -0.151 0.007 ***Cropped acres (1000 acres) 0.6264 0.000 *** 0.233 0.000 ***Percentage of gross revenue from livestock -0.0087 0.001 *** -0.003 0.001 ***Computer user (=1) 0.7565 0.000 *** 0.289 0.000 ***

Log-likelihood function -357.4Restricted log-likelihood -482.4Chi-squared 250.1 ***Prediction success:Percent of adopters correcly predicted 81.3Percent of non-adopters correcly predicted 71.4Overall correct prediction success 77.4

* One, two and three asterisks indicate significance at the 0.10, 0.05, and 0.01 probability levels, respectively.

A binomial probit model of the probability of adoption of precision farming technologies

Net Benefits of the Total PF System, 2010

0

1

2

3

VariableEstimated coefficient P[|Z|>z

Costs exceed benefits

Costs equal benefits

Benefits somewhat exceed costs

Benefits substantially exceed costs

Constant 0.978 0.000 ***Cropped acres (1000 acres) 0.149 0.015 ** -0.018 -0.022 -0.014 0.053USE_GPS (=1) 0.117 0.460 -0.015 -0.017 -0.009 0.041USE_YM (=1) -0.481 0.002 *** 0.046 0.065 0.070 -0.181YM*GPS (=1) 0.446 0.004 *** -0.058 -0.066 -0.031 0.154GRID SAMPLING (=1) 0.352 0.011 ** -0.042 -0.051 -0.032 0.125ZONE SAMPLING (=1) 0.065 0.707 -0.007 -0.009 -0.007 0.023VRT_PK (=1) 0.330 0.014 ** -0.041 -0.048 -0.027 0.116LIGHTBAR (=1) -0.025 0.834 0.003 0.004 0.002 -0.009AUTOSTEER (=1) -0.051 0.717 0.006 0.008 0.005 -0.018Threshold parameters for indexMu(1) 0.653 0.000 ***Mu(2) 2.029 0.000 ***Log-likelihood function -499.1Restricted log-likelihood -511.7Chi-squared 25.1 0.003 ***

Marginal effects

* One, two and three asterisks indicate significance at the 0.10, 0.05, and 0.01 probability levels,

Ordered Probit model of net benefits of the overall PF system

Part III: What does the future hold for Precision Farming developments and adoption?

What are the motivations for precision farming technology adoption?1. To more precisely regulate agricultural inputs to

maximize farm profitability from production or to facilitate meeting environmental compliance goals / regulations

2. To further allow substitution of capital for relatively expensive labor and management

3. To allow expansion of farm business size or to facilitate new production opportunities

•VRT applications will continue to expand over time• fertilizers, lime• seed planting rate and variety

• VRT is data intensive and requires specialized analytical skills• Currently much of this is contracted to special service providers.• On-the-go soil sampling sensors will facilitate adoption of VRT

• Auto guidance• Has been adopted very rapidly by large farmers• This technology is an important tool of precision farming•But also is a technological change that continues trends of substitution of capital for labor and expansion of farm business sizes.

• allows operation at higher speeds and longer hours with less operator fatigue.• allows operators with less experience / skill.• allows operation in low light conditions, effectively extending the day.•Two machines can be controlled by one operator to work together in the field.

• New methods of global positioning will facilitate.• Computer vision, radar, and other sensing technologies may be important.

• Supervised autonomous (robotic) machines• Further enables substitution of capital for labor.

• This technology may reverse many trends of past century• Previously, capital was substituted for labor using increasing larger machines. But, robotic tractors may be smaller, without operator amenities -- May be lower cost to own and operate.• Farmer ownership/operation of machines may change.•Issues of liability will need to be addressed.

• This technology may allow for significant changes in production practices.

• Strip intercropping has important agronomic benefits, but is difficult to implement using large equipment. • Controlled traffic systems / reduced soil compaction more feasible.• Intelligent spraying• Intelligent mechanical weed control

Spirit robotic tractor demonstration - Automation Research Group in Minneapolis, Minnesota

The tractor of the future???

Last word from farmers – what emerging technologies do they think will be most Important?

Technology N Meana

Precision placement of seeds 485 3.22On-the-to sensors for machinery (VRT of nutrients) 479 3.02Integrated software for managing PF data 474 2.88PF technology designed for small farms 476 2.88Precision guidance for implements 463 2.69Ultra-precise tillage technology 471 2.54Nano-sensor soil implants 473 2.39Tracking technology for grazing livestock 449 1.87Robotic machinery 472 1.74a Responses ranged from 1=Not Important to 4=Important