Public Information Creates Value
A case study of the USDA
Soybean Rust Coordinated Framework finds that the
value of the information provided by the framework
exceeds its cost.
Michael
J. Roberts; David
Schimmelpfennig
|
|
|
|
With
accurate information, individuals can
make sound decisions that allow them
to adjust their actions to the situation
at hand. Information comes from
many sources, but the value of publicly
provided information is often underestimated.
|
|
|
For
farmers who are trying to react to a
potential pest infection, such as soybean
rust, information about the likelihood
of infection can help them to make better
decisions about the amount and timing
of fungicide applications, which will
ultimately increase their profits. |
|
| This
article is drawn from . . . |
| The
Value of Plant Disease Early-Warning Systems:
A Case Study of USDA’s Soybean Rust
Coordinated Framework, by Michael
J. Roberts, David Schimmelpfennig, Elizabeth
Ashley, and Michael Livingston, with contributions
by Mark Ash and Utpal Vasavada, ERR-18, USDA,
Economic Research Service, March 2006 |
| You
may also be interested in . . . |
| Economics
of Food Labeling, by Elise Golan,
Fred Kuchler, Lorraine Mitchell, Cathy Greene,
and Amber Jessup, AER-793, USDA, Economic
Research Service, January 2001
ERS Briefing Room on Invasive
Species Management
ERS Briefing Room on Traceability
in the U.S. Food Supply |
Soybean rust (SBR), which is believed
to have been transmitted to the United States on
the winds of hurricanes during the summer of 2004,
is a new pest threatening the U.S. soybean crop.
In response to this threat, USDA leads an initiative
to monitor outbreaks of SBR and provide real-time
forecasts of its spread to help farmers efficiently
apply preventative and curative fungicides. In 2005,
SBR turned out to be less damaging than expected,
which led some to question whether or not development
of the initiative was worthwhile. But ERS research
finds that the public information about SBR was
still quite valuable because it helped farmers make
better decisions in managing their operations. In
general, the more information influences decisions,
the greater its value. Although the precise value
of the SBR information is unclear, with estimates
ranging broadly from $11 million to $299 million
in 2005, even the lowest estimated value is several
times the costs of providing the information to
farmers.
Information is an unusual kind
of economic good. It is not bought and sold in stores
like apples, cars, or DVD players, mainly because
people can easily share or replicate information.
As a result, markets do not always create and disseminate
information as efficiently as other kinds of goods
and services because it is hard for businesses to
control access and charge all users. Sometimes the
government can step in and provide information,
like hurricane or crop forecasts, that private markets
do not provide.
USDA and other agencies also implement
regulations that create incentives for individuals
and businesses to provide information they otherwise
may not. For example, the Food and Drug Administration
(FDA) requires “Nutrition Facts” labels
on food products, which helps consumers make better
dietary choices. These examples are a few of the
many ways government influences the creation and
dispersal of information.
Information is not normally traded
in competitive markets like apples. Thus, quantifying
its value is difficult because it involves determining
the decisions farmers would have made without the
information and what the consequences of those decisions
would have been. ERS estimated the value of public
information from USDA’s SBR initiative by
comparing farmers’ expected profits, as viewed
from the beginning of the season, with and without
the information. The value reflects the degree to
which information allows farmers to adjust their
decisions to suit the particular situation at hand.
Estimating the value involves quantifying how large
a threat farmers would have perceived SBR to have
been without the real-time forecasts. It also involves
evaluating what farmers’ decisions and profit
outcomes would have been without the framework.
The Soybean Rust Coordinated
Framework
In 2005, USDA initiated the Soybean
Rust Coordinated Framework to track and forecast
the incidence and spread of a new pest threatening
the U.S. soybean crop: Phakopsora pachyrhizi, a
fungus with the common name soybean rust. SBR has
been a recurrent problem for soybean producers in
much of the southern hemisphere. In recent years,
SBR has reduced yields and raised production costs
for soybeans in every major production region of
the world except the United States. Although SBR
has the potential to cause significant yield losses,
these can be almost entirely mitigated with application
of fungicides. The fungicides, however, are expensive,
so application reduces farmers’ profits if
SBR does not occur.
Almost 60 percent of the U.S.
soybean crop is produced in areas where climatic
conditions are expected to support establishment
of SBR in at least 5 of 10 years. SBR was first
detected in the Southern U.S. in fall 2004, late
enough in the season that it posed no threat to
that year’s soybean crop. After overwintering
in the South, SBR posed a new, uncertain, and potentially
large threat at the beginning of the 2005 U.S. soybean
season. Fields infected with SBR were anticipated
to see markedly reduced soybean yields if not treated
with fungicides.
With sufficient notice of an SBR
threat, farmers could treat their fields in advance
with preventative fungicides. Another approach to
the threat was to carefully monitor fields and immediately
treat with curative fungicides once the disease
was detected. Because curative fungicides must be
applied immediately after first infection, this
approach also benefits from timely information on
the spread of SBR by allowing farmers to limit scouting
to times when infection risks are highest. Fungicides
are costly, and the efficacy of both preventative
and curative fungicides is sensitive to the timing
of application, which means that better information
about the likelihood of infection helps farmers
improve management decisions and increase profits.
Information collected and analyzed
by the framework is communicated to the public via
the website, www.sbrusa.net. The public website
includes a regularly updated map showing where field
and test-plot monitoring has found and not found
evidence of SBR; national and local commentary discussing
the incidence and likely spread of SBR; and management
strategies, often delineated by county. The framework
also uses a web-based system to facilitate communication
between the many experts, comprised from government
and nongovernment agencies and universities, who
monitor for SBR in soybean fields and sentinel plots
strategically located throughout the country. USDA
built and tested the new information infrastructure
before SBR had caused any significant U.S. crop
losses.
The website, which was updated
almost daily during the growing season, was viewed
about 4.9 million times in 2005. Approximately 4,500
users of USDA’s SBR Internet website signed
up to be alerted via email when new information,
such as new incidence of SBR in the U.S., was posted.
This was the broadest USDA delivery over the Internet
of an information system to provide pest forecasts
to farmers and other stakeholders.
Estimating the Value of Soybean
Rust Forecasts
How valuable is information provided
by the framework? This question has become particularly
salient in light of the modest outbreak of SBR during
the 2005 season. Given the expense of developing
the website and its underlying infrastructure, some
have questioned whether the infrastructure was a
worthwhile endeavor. After all, if some farmers
had simply managed their crops as if there were
no SBR threat, it is possible that they would have
fared as well as or better than they actually did
in 2005.
This view overlooks the widespread
perception that SBR posed a threat (of unknown magnitude)
at the beginning of the season, and it is not clear
how farmers might have prepared for that threat
in the absence of the framework, which provided
real-time information about local, more imminent
threats. It could not have been known in advance
that optimal conditions for infection ultimately
would not arise in most areas. Indeed, without the
framework, individual farmers in some areas may
have incurred even greater expenses in monitoring
their own fields and perhaps spraying fungicides
for a threat that did not exist. Without the framework,
some farmers may have forgone planting soybeans
entirely and planted a less profitable alternative
crop.
ERS assessed the framework’s
value by estimating farmers’ expected profits,
as viewed from the beginning of the growing season,
with and without the information from the framework.
Making this calculation involved quantifying farmers’
expectations about the likelihood of SBR at the
beginning of the season—that is, how likely
they perceived the SBR threat to be. It also involved
evaluating what farmers’ decisions and profit
outcomes would have been without the framework.
Farmers’ decisions are fundamentally
different with and without information about the
SBR threat. With no information (the left decision
tree), farmers have to decide whether to spray or
not without knowing if their fields will become
infected with SBR. In this instance, farmers will
sometimes spray when not needed and sometimes not
spray when needed. Information about natural events
will seldom be perfect, but for illustrative purposes
a decision tree is shown that displays the outcome
with a perfect forecast. With perfect information,
farmers can always make the correct decision and
have higher profits.
The ERS analysis of the value of
SBR information addressed several intermediate scenarios
and used decision trees that allowed for less than
perfect information and a wait-and-see (monitor-and-cure)
treatment option. This richer analysis allowed for
sensitivity tests of the results to changes in assumptions
about risk aversion, heterogeneous beliefs of farmers,
and market price feedbacks from soybean yield changes
(see “Different Scenarios Affect
Estimated Information Values, But Not by Much”).
We examined how values might have
varied over different soybean-producing regions
and what the information values would have been
if farmers had different expectations about the
likelihood of SBR at the beginning of the season.
Although pinning down a precise value is impossible,
the analysis provides some perspective on the likely
benefits from the publicly provided information.
Across all scenarios and forecast
accuracies considered, we found the value of information
from the framework to range between $11 million
and $299 million in 2005, or about $0.16 to $4.12
per acre. This value is made up of a combination
of reduced expected costs and higher expected yields,
as viewed from the beginning of the season. The
range of possible information values is small relative
to total U.S. soybean sales (about $16.1 billion,
or $214 per acre), but quite large relative to the
cost of establishing the framework. Although we
did not conduct a comprehensive cost analysis, including
amortization of any fixed one-time costs, the framework’s
total development cost in 2005 was $2.6 million,
which suggests that the benefits of the framework
exceeded its costs.
Note that forecast quality pertains
to forecast accuracy, not the incidence of SBR.
A poor (imprecise) forecast is one that resolves
20 percent of farmers’ uncertainty about whether
or not they will be infected; medium and good (accurate)
forecasts resolve 50 and 80 percent of their uncertainty.
As one would expect, accurate
forecasts have much higher value than do imprecise
forecasts. More surprising, perhaps, is that risk
aversion (how much soybean farmers prefer steady
profits over variable profits), anticipated price
shocks (i.e., price feedback) from large rust outbreaks,
and widely varying farmer expectations (i.e., heterogeneous
beliefs) about the risk of infestation have relatively
little influence on the value of information, when
keeping the accuracy of the forecast fixed.
Public information has been particularly
valuable for SBR management, mainly because the
forecasts aided farmers in their decisions about
whether or not to apply fungicides. Because of the
high cost of monitoring and applying fungicides,
farmers would have wanted to apply these management
strategies only if an SBR threat were likely. Without
a forecast, they would have been more likely to
spray when it was unnecessary and not spray when
it was necessary. If preventative measures had not
been available and the only management options were
to lose crops to infection, if infection were certain
to occur, the forecasts would have had little or
no value. Thus, in evaluating the cost effectiveness
of developing public monitoring and forecasting
services for pests other than SBR, a key feature
to consider is whether or not preventative management
strategies might take advantage of any information
provided. The lesson learned is that the more information
influences decisions, the greater its value. This
is true regardless of whether information takes
the form of hurricane forecasts, food nutrition
labels, crop production forecasts, Internet searches,
or SBR forecasts.
|
Different Scenarios Affect Estimated Information
Values, But Not by Much |
| The
Base Case: The value of information
is determined by estimating the increase in
expected profit per acre of soybeans planted
and assuming soybean prices were fixed at
May 2, 2005, futures prices.
Risk Aversion:
Like the base case, except farmers are assumed
to be strongly risk averse, meaning they strongly
prefer a steady flow of profits over one that
is variable, holding expected profits the
same. Risk-averse farmers are more prone to
apply preventative fungicides in the absence
of information. They may also derive less
value from a lot of information because, somewhat
counterintuitively, fine-tuning their decisions
in response to information may cause their
profits to be more variable, even though profits
are higher on average. For example, without
an accurate forecast a farmer may always apply
the preventative fungicide, resulting in a
steady but low level of profits. With an accurate
forecast, farmers apply fungicides only when
needed, leading to higher average profits
but somewhat more variability. Because risk-averse
farmers dislike profit variability, the information
is therefore less valuable than it is to a
farmer who cares only about average profits.
The difference, however, is small.
Price Feedback:
Like the base case, except in
the event of an SBR outbreak, soybean prices
adjust to the reduced supply. The price response
is estimated using historical price response
to local yield shocks. Because price increases
offset farmers’ profit losses but hurt
those who purchase soybeans, the analysis
considers the total effect of information
on both soybean farmers and soybean purchasers.
Because prices tend to move in the opposite
direction as yield shocks, these effects tend
to offset each other, and thus have little
overall effect on the value of information.
Heterogeneous
Beliefs: Like the base case,
except farmers within each region are assumed
to have held widely varying expectations about
the likelihood of an SBR outbreak. In this
scenario, some farmers value information far
more than others do, but on average, the value
is close to the base case. |
|
