$850 million
USD
annually if Landsat was used to maximize
crop yield in northeast Iowa. A survey conducted in 2006
by the American Society for Photogrammetry and Remote
Sensing estimated a loss of $936 million
USD
to users who
completed the survey: if Landsat imagery was no longer avail-
able (Green, 2008). Each of these analyses pertains only to the
application and/or user group in which they were conducted,
which makes generalizing the results to other areas difficult.
As the above literature shows, monetizing the economic
value of this publically provided good beyond a specific
group or application can be challenging, partly because
there is no market price to reflect the value of the imagery to
society. Similar to many other sources of data and informa-
tion, Landsat imagery has characteristics of a public good: (a)
once provided to one user, there is little or no cost to provide
it to additional users; and (b) one person’s use of a Landsat
image does not consume or “use up” the image, so the same
image is available to others. As is well known in the econom-
ics literature, markets often fail in providing goods that have
characteristics of a public good. Markets tend to provide less
than the economically efficient or the socially optimal level
of provision of public goods, whether national defense or air
quality (Krugman and Wells, 2013, p. 479; Nicholson, 1992, p
758). As a result, public goods are more frequently provided
by group action through government or associations of users.
Even when the
USGS
had priced Landsat images, this price
did not accurately reflect economic benefits because the price
was administratively set. To capture the economic benefits
provided, user’s net willingness-to-pay or consumer surplus
is needed. This is the standard measure of benefits in benefit-
cost analysis (Sassone and Schaffer, 1978) and is recommend-
ed by the Office of Management and Budget to capture the
societal benefits from a government program or project (1992).
An Economic Method to Measure the Value of Landsat Images
A variety of methods can be used to value earth science
information (Macauley, 2005 and 2006). Economists apply
a specific set of methods to monetize the economic benefits
provided by goods and services that are not traded in markets
(see Freeman, 2003 for a review of these methods). One of the
most commonly applied methods and one most suitable for
valuing Landsat images is a stated preference or intended be-
havior technique known as the Contingent Valuation Method
(
CVM
).
CVM
is a survey-based approach used to estimate the
economic benefits individuals receive from a non-market
good or service. This method is recommended for use by
Federal agencies (US Environmental Protection Agency, 2000;
US Water Resources Council, 1983). While the method has
been used worldwide with more than a thousand studies,
nearly all of these involve environmental public goods or
outdoor recreation (see Carson, 2012). To our knowledge, only
a handful of studies have applied
CVM
to value publically pro-
vided sources of information, such as weather warnings and
forecasts (for instance, Lazo and Chestnut, 2002; Cavlovic
et
al
., undated). The work herein is the first application of
CVM
to satellite imagery and specifically, Landsat imagery.
CVM
is a survey-based approach in which individuals
are asked what they would pay to have access to a particu-
lar public good. There are several design features of such
surveys. The first is determining how the willingness-to-pay
(
WTP
) question is to be asked. Since the 1990s most such
questions do not involve directly asking an open-ended “fill
in the blank”
WTP
question format. Instead, users are asked if
they would “buy or not buy” the public good at a particular
price that varies across the sample of potential respondents.
This dichotomous choice format has several advantages.
First, it essentially simulates a market, where the individual
is confronted with a price and asked if they will pay that
amount or not. Second, there is less opportunity for “strate-
gic” behavior (e.g., inflating or understating
WTP
). As argued
by Carson and Groves (2007), this question format potentially
provides the appropriate incentive structure to get respon-
dents to reveal whether they value the good more or less than
the price they are asked to pay. When using this dichotomous
choice survey format, an important issue becomes having a
large enough sample to vary the bid amounts within different
potential groups of respondents that themselves may have
different valuations. Finally, whether the sample responses
exhibit economically rational behavior can be tested by
determining whether respondents facing a higher price have
a lower probability of “buying” (responding
Yes
), than those
facing lower prices. For these reasons, as well as the recom-
mendation of the US National Oceanic and Atmospheric Ad-
ministration’s (
NOAA
) blue ribbon panel (Arrow
et al
., 1993),
we use the dichotomous choice
CVM
approach.
We also implemented the
NOAA
panel’s suggestion to
remind users of their budget constraint when answering the
WTP
question. In this study the budget constraint was not the
usual household budget constraint but rather the user’s proj-
ect and/or agency budget constraint. Since we were sampling
from the entire spectrum of imagery users (which include
for-profit businesses; academics; Federal, State, and local gov-
ernments; and non-profit organizations), we emphasized that
the money to pay for the Landsat images would have to come
out of their current project or agency budgets.
Nonetheless, a longstanding issue in
CVM
has been the con-
cern over hypothetical bias, e.g., will respondents overstate
their
WTP
since they do not actually have to pay. While the
literature is voluminous on this topic (see Loomis, 2014 for a
summary), recent evidence suggests that use of the dichoto-
mous choice
WTP
question format can help minimize this
tendency. The overstatement is also less for users of a public
good than non-users (see Carson
et al
., 1996). When users are
valuing a well-defined product for which they have had past
“consumption” experience with, hypothetical bias is also
minimized (Cummings
et al
., 1986; List and Lucking-Reiley,
2000). The advantages and disadvantages of
CVM
have been
recently summarized by Kling
et al
. (2012), and will not be
dealt with here other than to say best practices were imple-
mented wherever possible.
2009 Pilot Survey
The survey wording and range of
WTP
bid dollar amounts
were originally pretested on a large sample of users in 2009
(see Miller
et al
., 2011, for more details). In particular, the
sample was drawn from a web search of self-identified Land-
sat users. Convenience samples such as this are often used to
pretest surveys.
The pilot survey was administered online, as was the 2012
survey; the results of which are reported here. Based on the
pilot, the dollar amounts asked in the dichotomous choice
WTP
question at the upper end were increased for the 2012
survey. In the dichotomous choice framework, high bids need
to be high enough so that almost all respondents answer that
they are not willing to pay, and similarly, low bids need to
be low enough so that typical respondents are willing to pay.
Due to the fact that a large portion of the respondents were
willing to pay the highest bid amount in the 2009 survey, high
bids in the final survey were increased. Further, the magni-
tude of the step up in bids for initial
Yes
responses and step
down in bids to initial
No
responses were adjusted from the
traditional doubling and halving to less obvious increments
and decrements.
648
August 2015
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
