An Overview of Public Information and the Agriculture and Food System

1
An Overview of Public Information and the
Agriculture and Food System

• Richard E. Just
• C-FARE Fall Symposium on
• "Public Information and the
• Agricultural and Food System
• Washington DC
• November 6, 2002

1
2
Unprecedented Change

• Industrialization of Agriculture
• Role of Off-farm Income
• Biotechnology GM Traceability
• Environmental Concerns Sensitivity
• Alternative Agriculture Niche Markets
• Nonmarket Activity (Contracting Internet)
• Information Technology the Internet
• Consolidation of Agribusiness (Supply, Mktg)

2
3
Shortcomings of Data

• Heterogeneity

• 1. Spatial Allocation (of Inputs to Crops)
• 2. Temporal Allocation (Planting, Growing,
  Harvest)
• 3. Statistical Distribution (Variation not only
  Average)
• 4. Correlation of Multifunctional Attributes
• 5. Capital Stock Long-Term Behavior
• 6. Financial Structure Off-farm Activity

• Identification of Structure vs Reduced Form
• Widely Accessible Panel Data is Needed
• Nontraditional Markets
• Consolidation Issues
• Anticipatory Policy Support

3
4
Todays Farm Sector is Truly DiverseOffutt (2002
AAEA Presidential Address AJAE Dec 2002)

• Size
• Choice of farming enterprise(s)
• Business organization
• Environmental performance

Analysts are obligated to investigate
differential response and impact.
Relying on aggregate data has obscured
distributional facts about effects of ag policy.
4
5
Unique Features of Ag ProductionJust Pope
(2002 synthesis of ag production in the Handbook)

• Temporal allocation with biological production
• Flexible output mix by spatial allocation
• Fragmented technology adoption (role of capital)
• Uncertainty (weather pests) biological
  production
• Heterogeneity
• Land/Soil Quality Climate/Weather/Pests
• Water Availability Environmental Sensitivity

5
6
The Problem of HeterogeneityJust Pope (1999
ASSA Meetings AJAE Aug 1999)

• Standard theory fails at the aggregate level if
  heterogeneity is not considered
• Explains why many models dont predict
• Implied policy/welfare impacts are false
• Distributional data is required

6
7
1. Spatial AllocationJust (2000 NEARA Meetings
ARER Oct 2000)

• Data do not include allocations of fertilizer,
  pesticides, labor to crops
• Models can study only aggregate production
  possibilities
• Models allow increased fertilizer application on
  wheat land to increase corn yields
• Necessary aggregation conditions are dubious
  prevent meaningful policy analysis

7
8
Adequacy of Aggregate ModelsJust Pope (1999
ASSA Meetings AJAE Aug 1999)

• Aggregates assumed to obey Adam Smiths invisible
  hand (equating marginal conditions)
• Ignores realities of farming
• SR fixities constraints (e.g., financial
  structure, physical capital, land quality, family
  labor)
• Price/weather/pest variation
• Ex post adjustment (responses to states of
  nature)
• Entry/exit/failure (bankruptcy)

• If policy responses are affected by these,
  aggregate models that ignore heterogeneity are
  inappropriate

8
9
2. Temporal AllocationJust (2000 SERA-IEG-31 Ag
Systems 2002)

• Data do not include timing of input applications
• Many risk-reducing inputs are stage-dependent
• Pesticide applications
• Pre-emergent - Preventative
• Post-emergent - Prescriptive
• Models cannot discern motivations
• Risk aversion vs
• Simple profit max (or loss minimization)

To understand behavior, we must study decisions
given information available to the farmer at the
time
9
10
The Crop Insurance ExampleJust/Calvin/Quiggin
(AJAE Nov 1999)

• Risk-based justification (missing risk market)
• Nondistortionary correction of market failure
• Research shows farmers are motivated by
  subsidies
• The risk benefit is only .65/acre
• Federal Costs
• 1.4-1.7 billion per year throughout the 1990s
• Efforts to address moral hazard/adverse selection
• Multi-Peril Crop Insurance (MPCI) Catastrophic
  Risk Protection (CAT)
• Crop Revenue Coverage (CRC) Revenue Assurance
  (RA)
• Income Protection (IP) Group Risk Protection
  (GRP)

Why are large subsidies required if SR risk
matters?
10
11
The Identification ProblemJust (2000
SERA-IEG-31 Ag Systems 2002)

• Is crop diversification due to risk aversion?
• Or labor constraints, scheduling of fixed inputs
  crop rotation?
• Is heavy use of pesticides due to risk aversion?
• Or expected profit benefits?
• Is irrigation used to reduce risk?
• Or increase profits?

Most risk effects are subject to identification
problems.Without allocations, discernment is
possible only artificially(by imposing
assumptions which in effect determine results).
11
12
Cannot Truly Test New Advanceswith Aggregate Data

• Methodological advances are illustrated in
  academic journals with token aggregate data
• Risk response in supply
• Risk effects of inputs
• Structure of farmers' risk preferences
• Exceptions primarily in less developed ag
• Aggregation tends to eliminate and alter risk
• Farm-level yield variation 2-10 times greater
  than aggregate data (Just Weninger AJAE May
  1999)

12
13
3. Statistical DistributionJust Pope (1999
ASSA Meetings AJAE Aug 1999)

• NASS Focus Averages (prices) and Totals
  (production capital)
• Economists cant generate the benefit of data
  that has been collected because all of its
  characteristics are not available for research
• Failure of aggregate models can be mitigated by
  data on variation as well as averages

• The alternative Assume identical farms
  circumstances

13
14
4. Correlation of AttributesJust Antle (1990
ASSA Meetings AER May 1990)

• Local correlations of multifunctional
  characteristics are critical for policy impacts
• Productivity
• Erodability
• Environmental sensitivity
• Value in preservation
• Data collection has tended to be independent
• ERS (ARMS) NASS CENSUS NRCS EPA GIS
• Public data do not allow linking observations by
  location

14
15
An Environmental Use Restriction
Environmental Sensitivity Pollution- Output
Ratio (z/y)
Input Intensity Input-Output Ratio (x/y)
16
16
An Input Intensity Restriction
Environmental Sensitivity Pollution- Output
Ratio (z/y)
Input Intensity Input-Output Ratio (x/y)
15
17
Effect of a Target Price
Pollution- Output Ratio (z/y)
Input-Output Ratio (x/y)
Py
PyT
Px
Px
17
18
Social Optimality
Pollution- Output Ratio (z/y)
Py Marginal value of output Px Marginal cost
of input Pz Marginal cost of pollution
Slope -Px/Pz
Py/Pz
PyY PxX PzZ gt 0
PyY PxX PzZ 0
Input-Output Ratio (x/y)
18
19
Social Optimality Environmental Use Restriction
Works Well
Pollution- Output Ratio (z/y)
Py Marginal value of output Px Marginal cost
of input Pz Marginal cost of pollution
Slope -Px/Pz
Py/Pz
PyY PxX PzZ 0
PyY PxX PzZ gt 0
Input-Output Ratio (x/y)
19
20
Social Optimality Input Intensity Restriction
Works Well
Pollution- Output Ratio (z/y)
Py Marginal value of output Px Marginal cost
of input Pz Marginal cost of pollution
Slope -Px/Pz
PyY PxX PzZ 0
PyY PxX PzZ gt 0
Input-Output Ratio (x/y)
20
21
5. Capital Stock LR BehaviorJust Pope (2002
synthesis of ag production in the Handbook)

• The ag risk that really matters is risk of farm
  failure (the long swings)
• Lesson of 1970s boom, 1980s debt crisis
• Capital investment/replacement is the key
• No data on capital vintages, retirement, salvage
• Crude, inaccessible data on debt/equity/wealth
• Hardly any study of LR preferences/behavior
• Farmers willingness to tradeoff annual
  variability for serial correlation of profits is
  not understood

21
22
The Move to Service Flow DataJust Pope (2002
synthesis of ag production in the Handbook)

• Lacking better capital data, service flow data
  has been used for estimating production
• Construction relies on marginal assumptions
  necessary for using aggregate data
• Ignores
• SR fixities constraints
• Ex post adjustment (response to state of nature)
• Price/weather/pest variation
• Entry/exit

22
23
6. Financial Structure Off-farm ActivityOffutt
(2002 AAEA Presidential Address AJAE Dec 2002)

• The family farm as a household
• Most farmers major occupation is not farming
• Hobby farming ¾ have sales lt 50K ½ lt 10K
• Production behavior is affected by financial
  constraints
• Production behavior may be motivated by
  consumption preferences (smoothing of risk)
• Hobby farming may be a consumption activity

23
24
Tendency Toward Reduced Form EstimationJust
Pope (2002 synthesis of ag production in the
Handbook)

• Thinking Appropriately restricted reduced forms
  relieve data requirements (Offutt)
• Estimation of outcomes w/o underlying how
• Cannot learn basic properties of technology or
  preferences with reduced form models
• Composition of technologies is key
• Estimated parameters of reduced form or aggregate
  technologies embody policies

24
25
Structure vs Reduced FormJust Pope (2002
synthesis of ag production in the Handbook)

• Lucas Critique Models estimated under one
  policy cannot be used for another because the
  estimated parameters embody policies under which
  the data were generated

• Solution Capture deep structure
• Production Structure (allocation to technologies)
• Behavior given Technology Financial Structure
• Structure of Institutions Markets
• Model Change in Internal Structure vs Exogenous
  Factors
• Policy- Behavior-Relevant Aggregation

25
26
Need Widely-Accessible Panel Data

• Few data sets reflect individual farms
• ARMS KSU Farm Mgmt Survey ICRISAT
• Limited accessibility (access analysis)
• Debate scientific progress is choked
• Labor economics Current Population Survey, Panel
  Study of Income Dynamics
• Survey exposure vs matching existing data
• Confidentiality broadening the circle

26
27
My Experience on Crop Insurance

• Needed data
• FCRS (now ARMS) data on crop production
• FCIC data on participation and yield histories
• Piggy-back survey to get farmer perceptions
• My grant paid for NASS to conduct the survey
• NASS matching problems
• Delays discarding of data
• Limited access to data
• Delays in research revisions for publication
• Three-quarters of research abandoned after 10
  years

27
28
Nontraditional MarketsPotential Declining
Inclusiveness of Public Data

• Decline in central cash markets (Ag Statistics)
• Industrial Agriculture
• Poultry 90 contracted since 1950s but vertical
  integration doubled 1975-94 only 1 cash market
• Pork Contracted share 2 to 56 1970-99
• Beef Non-cash marketing 19 to 42 1994-2000
• Internet marketing (disintermediation)
• Niche markets direct marketing
• Information markets

28
29
Consolidation in Ag Supply MarketingInability
to Research Concentrated Industries

• The 80 largest pesticide companies have merged
  into 10 huge conglomerates
• Noncompetitive pricing premiums are typically
  20-50 and profit margins are higher
• Strategic competitive practices maintain
  monopolies up to 10 yrs beyond patents
• Public data give no way to estimate the
  associated welfare losses

29
30
Anticipatory Policy Support

• Actions are preceded by perceptions
• Perceptions depend on information
• Models are weakest w.r.t. expectations
• Models are weakest when needed the most
• The formative stages of policy making
• Examples GM seeds (Starlink Corn), Risk,
• Rapidity of adoption in the age of information
• Understanding policy impacts may depend on
  understanding information markets (Internet)

32
31
Shortcomings of Data

• Heterogeneity

• 1. Spatial Allocation (of Inputs to Crops)
• 2. Temporal Allocation (Planting, Growing,
  Harvest)
• 3. Statistical Distribution (Variation not only
  Average)
• 4. Correlation of Multifunctional Attributes
• 5. Capital Stock Long-Term Behavior
• 6. Financial Structure Off-farm Activity

• Identification of Structure vs Reduced Form
• The Need for Widely Accessible Panel Data
• Nontraditional Markets
• Consolidation Issues
• Anticipatory Policy Support

33