Upload
independent
View
1
Download
0
Embed Size (px)
Citation preview
ORIGINAL PAPER
Assessing sustainability when data availability limits real-timeestimates: using near-time indicators to extend sustainabilitymetrics
Matthew T. Heberling • Matthew E. Hopton
Received: 7 April 2011 / Accepted: 29 October 2013 / Published online: 19 November 2013
� Springer-Verlag Berlin Heidelberg (outside the USA) 2013
Abstract The goal of this paper is to highlight the
problem of time lags in data releases that are necessary for
calculating sustainability metrics and its effect on making
informed management decisions. We produced a method-
ology to assess whether a regional system is on a sustain-
able path and tested it in south-central Colorado. We
identified key components of the system and selected four
sustainability metrics that measure those components.
Metrics included: (1) ecological footprint (i.e., environ-
mental burden), (2) green net regional product (GNRP)
(i.e., economic well-being), (3) emergy (i.e., energy flows),
and (4) Fisher information (i.e., dynamic order). Having
calculated these metrics, we identified future research
recommendations and limitations. One limitation was the
delay between when an event occurred and when data on
the event were released. Given, the recent push in gov-
ernment agencies for calculating sustainability metrics,
finding solutions for the time lag will be important. To
address this limitation, we explore the potential of using
both sustainability metrics and indicators that are available
near-time to provide decision makers with better decision
support. For the pilot study in Colorado, the metric cal-
culations were 3 years behind present. Using near-time
indicators that are publicly available before the metrics can
be calculated might help to predict the path of the metric.
As an example, we examine if specific near-time indicators
are correlated with ecological balance (a component of
ecological footprint) and GNRP. We use Spearman rank
correlations and scatter plots to identify the relationship of
the metrics and near-time indicators in an exploratory
analysis. We offer research recommendations to consider.
Keywords Sustainability � Ecological footprint �Green net regional product � Data availability � Data
release � Time lag � Ecological balance � Green
accounting
Introduction
The United States Environmental Protection Agency
(USEPA) is interested in developing integrated, system-
based decision tools that focus on sustainability. We
wanted a general methodology applicable at many spatial
scales (e.g., city, country, state, region, etc.) with appro-
priate modification as data needs and availability change
for the specific area under study. To test our general
methodology, we initiated a pilot study in south-central
Colorado, a region we refer to as the San Luis Basin (SLB).
The research objectives of the pilot study were to: (1)
determine the applicability of using existing datasets to
estimate four metrics1 of sustainability at a regional scale;
(2) calculate metrics through time from 1980–2005; and (3)
M. T. Heberling (&) � M. E. Hopton
National Risk Management Research Laboratory, Sustainable
Technology Division, Sustainable Environments Branch, US
EPA/ORD/NRMRL (MS 443), US EPA, 26 W MLK Dr.,
Cincinnati, OH 45268, USA
e-mail: [email protected]
M. E. Hopton
e-mail: [email protected]
1 For clarification, we distinguish between ‘‘metric’’ or ‘‘index,’’ and
‘‘indicator.’’ Mayer (2008) defines an indicator as measuring one
characteristic or variable (e.g., CO2 emissions), but an index
combines many indicators or variables through aggregation (e.g.,
ecological footprint). Therefore, indices can provide a multidimen-
sional view of sustainability and are capable of quantifying the
condition of an entire system. For this paper, metric and index are
synonymous.
123
Clean Techn Environ Policy (2014) 16:739–748
DOI 10.1007/s10098-013-0683-6
compare and contrast the results to determine if the system
was moving toward or away from sustainability (i.e., the
path of sustainability).
To identify the trends in sustainability for the SLB, the
interdisciplinary research team, which included ecologists,
engineers, economists, geographers, physical scientists,
and outreach specialists, first determined the major com-
ponents of the system. Next, the team identified and tested
the four sustainability metrics thought to capture most of
these basic properties of a system. Specifically, (1) envi-
ronmental burden characterized by ecological footprint
analysis (EFA), (2) economic well-being ascertained from
green net regional product (GNRP), (3) flow and conser-
vation of energy through the system as computed from an
emergy analysis (EmA) and specific emergy indices such
as the ratio of renewable emergy used to total emergy used,
and (4) dynamic order (i.e., a well-functioning system has
an inherent order and departure from such a state can lead
to decreased function) as estimated from Fisher informa-
tion (FI). We successfully collected data appropriate to the
region and we were able to calculate the metrics through
time. A detailed description of the project, methodology,
and results can be found in San Luis Basin Sustainability
Metrics Project: A Methodology for Evaluating Regional
Sustainability (USEPA 2010) and a special collection in
Journal of Environmental Management (Heberling and
Hopton 2012).
Based on results from the pilot study, a potential limi-
tation of the methodology was how close to present time
the metrics could be calculated (USEPA 2010). The last
year we included in the calculations depended on when
collected data were publicly available (USEPA 2010). The
time lag would not allow us to calculate any metric more
recently than 3 years before present (i.e., 2005 data were
available in 2008). In other words, the calculations were
always 3 years behind the present year, limiting the use-
fulness for decision support.2 For instance, the effect of
management decisions made in 1 year (e.g., 2008) would
be based on metrics that were 3 years in the past (i.e.,
2005) and the results of the management action would not
be observable in the data or metrics for at least 3 years (i.e.,
2011). Although the delay of available data was 3 years for
this project, it could be more or less depending upon the
spatial scale under investigation and the types of data
needed. Knowledge gained from the pilot project led to a
number of avenues for future research, including ways to
mitigate the delay in available data (USEPA 2010). There
are potentially many approaches for dealing with this issue
including developing future scenarios to estimate how the
metrics might change or modeling the important systems
and varying important variables. However, these have their
own set of limitations including linkages that are difficult
to predict.
The goal of this paper is to highlight the problem of time
lags in data releases and availability of those data necessary
for calculating sustainability metrics. Both impact support-
ing decisions related to sustainability. In addition, we
explore the possibility of combining both the calculated
sustainability metrics and near-time indicators to provide
better information on sustainability. We define a near-time
indicator, as opposed to the lagging metric, as a variable
whose data are released to the public and are available close
to present time. Such variables that are available in a timely-
manner and show a strong correlation with a sustainability
metric could be used as proxies for the metric until all of the
data necessary to calculate the metric are available. These
proxies might provide insight to how the metric is moving
during those recent years for which all data necessary to
calculate the metric are not yet available. In effect, we are
proposing to extend the sustainability metrics to more recent
years using relevant near-time indicators. We explore fur-
ther below why this issue is a problem.
In this exploratory analysis, and to begin the conversa-
tion on time lags and sustainability metrics, we identify
potential near-time indicators by examining correlations
with the sustainability metrics. We are not suggesting
stopping the estimation of the sustainability metrics; rather,
we only are proposing using near-time indicators to predict
the trends until data are available for the full calculation.
By continuing the metric calculations, we also provide an
annual check on the relevant near-time indicators (i.e.,
examining whether the indicators still provide some pre-
dictability of trends). We did not find any existing sus-
tainability literature addressing data availability.
Methodology
For this paper, we only focus on two of the four metrics in
the SLB study (EFA and GNRP) because we were able to
calculate values for all 26 years for these metrics. These
metrics, relatively speaking, are easier to disaggregate into
the variables and indicators used for calculations. We have
described in detail the data sources and approaches used to
calculate each of these metrics in USEPA (2010), Hopton
and White (2012), Heberling et al. (2012), but briefly
describe each approach below.
EFA is an effort to quantify human burden on the
environment. Rees (1992) and Wackernagel and Rees
(1996) applied and inverted the concept of carrying
capacity (i.e., the number of individuals a system of a given
2 We state there was a three-year lag for the original study, which
was based primarily on CO2 emissions data. For this particular
variable, a new approach for these emissions data was under
consideration (pers. comm., P. Lindstrom 2008). As Table 1 points
out, CO2 emissions now have a 2-year delay.
740 M. T. Heberling, M. E. Hopton
123
size or area can support) to humans by estimating the
amount of biologically productive (bioproductive) land that
is required to support a population of a given size at a given
standard of living and comparing this demand to the bio-
productive land that is available (i.e., biocapacity or sup-
ply) to support the population. The area of land necessary
for the production and maintenance of goods and services
consumed by the population was termed the ecological
footprint (the metric is named after this demand component
of the overall analysis). Sustainability is estimated by
relating the supply to demand of environmental resources.
We conducted conventional EFA accounting using the
compound approach as introduced by Wackernagel and
Rees (1996) and expanded by Chambers et al. (2000). This
approach is more inclusive and robust compared to the
component-based approach (Chambers et al. 2000). In
general, an EFA calculation requires: (1) the size of the
population in the study area, (2) the amount of consum-
ables used per individual, (3) the amount of energy con-
sumed per individual, and (4) the amount of biologically
productive land, as defined by EFA, available in the study
region. For this paper, we use ecological balance, supply
minus demand, for the SLB as the metric of interest (for the
remaining paper, we use EFA and ecological balance
interchangeably when we refer to the analyses in this
paper; see also Hopton and White 2012). Ecological bal-
ance could be a surplus if available biocapacity exceeds the
ecological footprint, or it could be a deficit if the ecological
footprint exceeds the available biocapacity. Ultimately, the
trend of ecological balance is important to reveal if the
system is moving toward sustainability.
Green net national product captures the economic well-
being of a nation by adjusting the standard measurements
of a nation’s economy (e.g., gross national product),
through explicitly incorporating the depreciation of all
capital stocks (e.g., manufactured, human, and natural
capital), consumption of ecosystem services, and damage
from pollution flows. Economists define sustainability as
non-declining utility over generations (e.g., Pezzey 2004;
Pezzey et al. 2006), which provides the decision rule for
examining a nation’s sustainability. Because we are look-
ing at a region, we call this metric GNRP (Heberling et al.
2012). Few economic studies have examined sustainability
at the regional scale as opposed to countries (e.g., Gram-
bsch et al. 1993; Gundimeda et al. 2007; Gren and Isacs
2009; Heberling et al. 2012).
Calculating net regional product (NRP), which is gross
regional product minus the depreciation of man-made capital,
moves us closer to our measure of interest. It does not include
the depreciation of natural capital. This component still has to
be estimated to calculate GNRP. The depreciation of natural
capital included primary components of natural capital in this
system (i.e., water quantity from groundwater, soil erosion
from wind, and CO2 emissions; USEPA 2010). We converted
these variables into dollars using standard economic approa-
ches. We also estimated the value of time from technological
progress (e.g., see Pezzey et al. 2006). Starting from NRP, we
deduct the costs of emissions, costs of soil erosion, and the
value of rent from groundwater stock. We then add the value
of time to estimate GNRP for the SLB (Heberling et al. 2012).
To compare the existing metric calculations with near-
time indicators, we need an approach for identifying a
potential set of indicators. There are a number of sources
for presenting criteria related to choosing indicators and
Ochola et al. (2003), for example, summarize several
studies that describe approaches for choosing and evalu-
ating indicators. For illustrative purposes, we present two
different sets of criteria used for choosing indicators.
Zarnowitz (1992) lists six criteria The National Bureau of
Economic Research uses for choosing business cycle
indicators. The following questions were used to identify
the appropriate measures (Zarnowitz 1992: 317–318):
1. How well understood and how important is the role in
business cycles of the variables represented by the
data? (the judgment on this is quantified in the score
for economic significance).
2. How well does the given series measure the economic
variable or process in question (statistical adequacy)?
3. How consistently has the series led (or coincided or
lagged) at business cycle peaks and troughs (timing at
recessions and revivals)?
4. How regularly have the movements in the specific indicator
reflected the expansions and contractions in the economy at
large (conformity to historical business cycles)?
5. How promptly can a cyclical turn in the series be
distinguished from directional change associated with
shorter, irregular movements (smoothness, which is
inversely related to the degree of statistical noise)?
6. How promptly available are the statistics and how
frequently are they reported (currency or timeliness)?
Jackson et al. (2000) provide guidelines for selecting
ecological indicators as part of the Office of Research and
Development’s Environmental Monitoring and Assessment
Program. There are 15 guidelines for evaluating indicators
organized into four evaluation phases. The evaluation
process can be summarized into four related questions
(Jackson et al. 2000: 1–5):
Phase 1. Conceptual relevance: Is the indicator relevant to
the assessment question (management concern)
and to the ecological resource or function at risk?
Assessing sustainability when data availability limits real-time estimates 741
123
Phase 2. Feasibility of implementation: Are the methods
for sampling and measuring the environmental
variables technically feasible, appropriate, and
efficient for use in a monitoring program?
Phase 3. Response variability: Are human errors of mea-
surement and natural variability over time and
space sufficiently understood and documented?
Phase 4. Interpretation and utility: Will the indicator
convey information on ecological condition
that is meaningful to environmental decision-
making?
Neither example is completely appropriate for identi-
fying near-time indicators for sustainability metrics
because their purposes differ from our objective. As the
number of sustainability studies increase, developing spe-
cific criteria for choosing near-time indicators should be a
research priority. We can glean relevant information from
the above criteria, however. As Jackson et al. (2000)
emphasize, the choice of indicators depends on the user’s
needs and objectives. Requests from stakeholders and
reviewers of the San Luis Basin Sustainability Metrics
Project focused on being able to predict future values of
the metrics. However, we were unable to calculate the
metrics for recent years, yet alone predict future values.
Zarnowitz (1992) highlights important criteria that cover
some aspects for this research. These include timely data
that cover the period between the final year of metric
calculation and present.
We specifically need indicators that are capable of
representing the sustainability metrics for the missing
years. One place to look for such near-time indicators that
may relate to the metrics is within data used to calculate the
sustainability metrics. By examining the variables included
in each of the metrics as an initial investigation, we can
identify data that are reported more frequently and more
quickly and provide an indication how the metric is
responding during the missing years. If additional indica-
tors are necessary, finding variables related to the main
components of the metrics makes sense. Once selected, we
determine if the indicators are significantly correlated with
the metrics and could serve as a proxy for the metric.
For the SLB project, GNRP can be broken down into
NRP, natural capital depreciation, and value of time. As
stated earlier, soil erosion, groundwater storage, and CO2
emissions were used for natural capital depreciation. For
EFA, the main components were population, agricultural
production and consumption, and energy consumption.
Given our methodologies for calculating GNRP and EFA,
Table 1 shows when data necessary to estimate the metrics
for year 2010 were available. Our goal was to identify
variables that could serve as near-time indicators, could
represent each metric, and were publicly available prior to
Table 1 Major variables needed to calculate sustainability metric
and potential near-time indicators for 2010 and year when data would
become available
Variables Month and year data
for 2010 were released
GNRP NRPa April 2012
Soil erosion October 2011
Groundwater storageb 2010
CO2 emissions October 2012
Near-time indicator Population estimatesc April 2011
Potatoes plantedd October 2011
Precipitatione April 2011
Total wagef March 2010 available
October 2010
Total employeesf March 2010 available
October 2010
Groundwater storageb 2010
EFA Food productiong February 2011
Food consumptionh February 2012
Energy consumptioni June 2012
Population estimatesc April 2011
Bioproductive landj 2011
Near-time indicator Population estimatesc April 2011
Potato productiong February 2011
Cereal productiong February 2011
Meat productiong February 2011
Precipitatione April 2011
a To estimate NRP, we require gross domestic product and consumption
of fixed capital for the US, gross domestic product and personal income
for Colorado and New Mexico, and personal income for the SLB
counties (USEPA 2010). The reason NRP only can be calculated in
April 2012 is that personal income for the SLB counties is released on
that date (all other variables are released in 2011)b Colorado’s Decision Support System can be used to examine histor-
ical water uses and future scenarios, so we state that water quantity
could be estimated for the current year (CWCB 2000)c Typically, estimates through July are released in December of the
current year and annual estimates are released in April of the following
year (http://www.census.gov/popest/topics/schedule.html and pers.
comm.—Katie Wengert, US Census Bureau, Population division)d http://www.nass.usda.gov/Statistics_by_State/Colorado/index.aspe Pers. comm.—Jan Curtis, PRISM Project Manager, http://www.prism.
oregonstate.edu/products/matrix.phtmlf Total wage and employees are available from the US Bureau of Labor
Statistics. Accessed at: http://www.bls.gov/cew/releasecalendar.htmg http://www.nass.usda.gov/Data_and_Statistics/County_Data_Files/
Release_Schedule/index.asph Pers. comm.—Jean Buzby, USDA, Economic Research Servicei Pers. comm.—Barbara T. Fichman, Integrated Energy Statistics, Sur-
vey Development and Statistical Integration, Office of Energy Statisticsj Bioproductive land is calculated from National Agricultural Statistics
Service, USDA-Forest Service data, US Census Bureau data, and
National Land Condition Data and is dependent on release dates for
those datasets (USEPA 2010). National Agricultural Statistics Service
2007 was the last report (NASS 2009); 2012 will be released in February
2014. However, missing years are interpolated annually
742 M. T. Heberling, M. E. Hopton
123
when those metric components were available. Because of
the exploratory nature of this analysis, the near-time indi-
cators were our initial attempt and we selected a subset of
potentially relevant variables (see Table 1 for data sources).
For GNRP, we chose to test population, potatoes plan-
ted, precipitation, groundwater storage, total wage, and
total employees because of their timeliness of release. We
propose population, total wage, and total employees could
be correlated with GNRP because they are related to NRP.
Because of our approach for estimating wind erosion
(where potatoes planted is one of the major variables), we
thought that potatoes planted (as opposed to potato pro-
duction or potatoes harvested) is related to GNRP through
the soil erosion component (USEPA 2010). Finally, we
wanted to test whether precipitation and groundwater
storage were correlated with GNRP through agricultural
production and water quantity; groundwater storage was
our actual measure of water quantity in the calculation of
GNRP. Unfortunately, data for total wage and total
employees were not available for all 26 years; for this
paper, they were limited to 2001–2005.
For EFA, we selected meat, cereal, and potato produc-
tion (kg) as well as population and precipitation to test. We
thought agricultural production could be correlated with
ecological balance because it quantifies bioproductive land
and is available relatively soon after data are collected. We
test whether precipitation is correlated with ecological
balance because it can affect agricultural production (e.g.,
Rosenzweig et al. 2002). Finally, we thought population
might be correlated with EFA because human burden and
population are intricately linked (i.e., an increasing popu-
lation with level per capita consumption will lead to
increased burden).
We used Spearman rank correlation coefficients to
examine whether any of the near-time indicators correlate
closely with GNRP and EFA. Spearman rank correlation
coefficient is a non-parametric procedure that allows sig-
nificance to be assessed (Zar 1999). Often it is used as an
alternative to Pearson product-moment correlation coeffi-
cient because it is not as ‘‘sensitive to the shape of the
distribution or the presence of outliers’’ (Siegel and Mor-
gan 1996: 564). Most of our data failed the normality
assumption. The coefficient ranges between ?1 and -1,
which indicates a perfect positive correlation and a perfect
negative correlation, respectively. If the correlation is
strong, either negative or positive, we suggest the variable
could be used as a proxy or alternative for the actual cal-
culation of the metric (until the complete metric data set is
available) to cover the missing years. Rather than sepa-
rating into individual correlation tables, we decided to run
all the variables and metrics at one time. We did this for
two reasons. First, we wanted to see if any of the near-time
indicators were related and second, we wanted to see if
GNRP and EFA were correlated. We also created scatter
plots of the near-time indicators with the individual metrics
to provide additional insight into the relationship of the
metrics and near-time indicators. We analyzed these data
using STATISTICA, version 10.0.10113 and significance
was tested at P B 0.05.
Results
Table 2 presents the Spearman rank correlation coefficients
for both GNRP and EFA. First, the analysis shows GNRP
and EFA are significantly negatively correlated (rs =
-0.75, P \ 0.0001) for these 26 years of data. Because of
this result, it is not surprising that near-time indicators
important for GNRP are important to ecological balance
and vice versa, albeit in an opposite manner. For example,
meat and cereal production (proposed near-time indicators
for ecological balance) were positively correlated with
EFA and were negatively related to GNRP whereas the
opposite was true for potato production (Table 2).
We are aware that calculating Spearman rank correla-
tion coefficients with very small samples could cause
interpretation problems. Therefore, we did not estimate the
correlation coefficients for GNRP, total wages, and total
employees. GNRP appeared strongly positively correlated
with population (rs = 0.84, P \ 0.0001). It exhibited a
positive correlation with potatoes planted, but the coeffi-
cient was smaller than the coefficient with population.
GNRP was not correlated with either precipitation or
groundwater storage for these 26 years.
Ecological balance showed a significant correlation with
all of the variables, excluding groundwater storage. Specif-
ically, ecological balance was negatively correlated with
population (rs = -0.92, P \ 0.0001), potatoes planted,
(rs = -0.51, P = 0.0084), potatoes produced (rs = -0.45,
P = 0.0221), and positively correlated with precipitation
(rs = 0.49, P = 0.0106), cereal production (rs = 0.46,
P = 0.0191), and meat production (rs = 0.45, P = 0.0197).
Figure 1 presents the scatter plot matrix for GNRP and
potential near-time indicators. Figure 2 displays those
potential near-time indicators for EFA.
Discussion
We proposed and tested using near-time indicators to cover a
time lag in data availability that prevents calculating metrics
and to help identify the trends in two sustainability metrics.
These near-time indicators may serve as proxies for the
3 StatSoft, Inc. (2011). STATISTICA (data analysis software
system), version 10. www.statsoft.com.
Assessing sustainability when data availability limits real-time estimates 743
123
metrics during intervening years while waiting on data to be
released. If, for example, we focus on one or two variables,
both GNRP and EFA appear to have potential near-time
indicators based on our correlation analyses. However,
given the original proposed set of near-time indicators (i.e.,
Table 1), there were no strong negative correlations that
were statistically significant (i.e., |rs| C 0.7, P B 0.05) with
GNRP and no strong positive correlations that were statis-
tically significant with ecological balance.
For GNRP, population and potatoes planted are potential
near-time indicators. To be useful, near-time indicators
should correlate with the metrics and provide a warning for
movement away from sustainability, which would not be
found in this study unless population or potatoes planted
falls. Interestingly, notice that as GNRP increased, potatoes
planted peaked around $1.2 billion and then began to fall
(Fig. 1). This type of nonlinear relationship highlights the
limitation of only calculating Spearman rank correlation
coefficients and the importance of identifying threshold
effects for decision makers. For this region, GNRP was
driven by the market side (i.e., NRP) rather than the
depreciation of natural capital (USEPA 2010). For future
sustainability studies that have a large contribution from
the depreciation of natural capital, near-time indicators
might reveal information that would provide an early
warning for movement away from sustainability. For eco-
logical balance, population had a strong negative correla-
tion. Population makes sense as a near-time indicator
because EFA attempts to quantify the amount of resources
a population is consuming. As the population increases, all
things being equal (i.e., consumption per capita does not
change) the demand placed on the resources will increase
and fewer resources available per capita. Moreover, pop-
ulation is a good near-time indicator because estimated
data are released relatively quickly (i.e., April 2011 for
20104; Table 1).
It is interesting to note there is an inverse relationship
between GNRP and EFA (USEPA 2010), potentially con-
sistent with Nourry (2008), at least graphically, who
examined the sustainability of France. However, other
studies have not identified such a relationship between
GNRP and EFA (e.g., Hanley et al. 1999). With that said,
when examining the indicators proposed for EFA, GNRP
did have a negative correlation with ecological balance,
meat production, and cereal production. Because ecologi-
cal balance had a positive and significant relationship with
meat and cereal production, it is not surprising that GNRP
had a negative and significant correlation with the pro-
duction variables. For this region, if we calculate one
metric with less lag time, it could predict the other metric.
A final note about including all of the metrics and
indicators in the same correlation table revealed that
potato production (i.e., tuber production) and potatoes
Table 2 Spearman correlation coefficients for sustainability metrics and potential near-time indicators (1980–2005)
GNRP Population Potatoes
planted
Precipitation Ground-
water
EFA Potato
production
Meat
production
Cereal
production
GNRP 1.000
Population 0.84205
(\0.0001)
1.000
Potatoes
planted
0.55711
(0.0031)
0.66347
(0.0002)
1.000
Precipitation -0.33402
(0.0954)
-0.55214
(0.0034)
-0.22298
(0.2735)
1.000
Ground-
water
-0.14467
(0.4746)
-0.33607
(0.0932)
-0.18263
(0.3719)
0.68342
(0.0001)
1.000
EFA -0.75385
(\0.0001)
-0.91863
(\0.0001)
-0.50581
(0.0084)
0.49265
(0.0106)
0.29641
(0.1415)
1.000
Potato
production
0.47419
(0.0144)
0.60615
(0.0010)
0.91416
(\0.0001)
-0.20889
(0.3058)
-0.09538
(0.6430)
-0.44684
(0.0221)
1.000
Meat
production
-0.70188
(\0.0001)
-0.56034
(0.0029)
-0.16518
(0.4200)
0.44342
(0.0233)
0.31350
(0.1189)
0.45436
(0.0197)
-0.07829
(0.7038)
1.000
Cereal
production
-0.46393
(0.0170)
-0.63077
(0.0006)
-0.35978
(0.0710)
0.53504
(0.0049)
0.26906
(0.1838)
0.45641
(0.0191)
-0.36821
(0.0642)
0.36821
(0.0642)
1.000
Boldnames are the sustainability metrics of interest
P-values are in parentheses
4 Typically, population estimates for counties through July of a given
year are available in December of that year and annual estimates are
released in April of the following year. Because of the decadal census,
county estimates for 2010 were released March 2011.
744 M. T. Heberling, M. E. Hopton
123
planted showed a positive correlation, as one might
expect. Notice data for potatoes planted were released in
October 2011 for a 2010 calculation (Table 1). However,
data for potato production actually were released
8 months earlier. By using potato production for this
examination, a researcher would have near-time indica-
tors available the April after the year of interest. This
would be an improvement over the two or three year lag
we faced for the actual calculations of the GNRP and
EFA. For example, in April 2011, given the years for
which data necessary to compute the metrics were
available, we could only calculate up to year 2008
because of limitations (i.e., data availability) from CO2
emissions and energy consumption. However, using the
near-time indicators, we might be able to predict the
trends of the metrics for 2009 and 2010, thereby cov-
ering some of the time lag and better enabling decisions
to be made.
For the SLB project, the last year we could calculate
GNRP and EFA was 2005 because of data availability
issues. In 2005, the population in the SLB was 47,530.
Population declined in 2006 and 2007 to 47,125 and
46,888, respectively. From this information along with the
correlation results, a decision maker may guess that EFA
would improve in 2006 and 2007, whereas GNRP would
decline.
Fig. 1 Scatter plot for green net regional product (GNRP) and potential near-time indicators (1980–2005)
Assessing sustainability when data availability limits real-time estimates 745
123
Conclusion
Having calculated sustainability metrics to support deci-
sions related to land use, we revealed a potential limitation
of our methodology. The limitation depends on when data
are made available and the earliest possible calculation of
GNRP or EFA. GNRP is an economic metric for examin-
ing the trends in sustainability and ecological balance is a
component of EFA. For the pilot study in the SLB, the lag
was 3 years (meaning that in 2008 we could only calculate
upto 2005).
As the USEPA and other Federal, state, or regional
agencies pursue calculating sustainability metrics for dif-
ferent purposes, we advocate acknowledging the impact of
potential lag times based on data availability. By doing
this, agencies can begin to think about how to supplement
these metrics with other indicators or other analyses to
make the calculations closer to real time and better support
management decisions.
Our exploratory technique to reduce the time lag was to
examine scatter plots and use Spearman rank correlation
coefficients to determine if near-time indicators could
Fig. 2 Scatter plot for ecological balance (EFA) and potential near-time indicators (1980–2005)
746 M. T. Heberling, M. E. Hopton
123
represent changes in the sustainability metrics. The initial
attempt found potential near-time indicators for both
GNRP and EFA for the SLB. These included population,
precipitation, and a number of agricultural production
indicators. To continue with this line of research, we rec-
ommend reassessing the correlations annually as new data
may result in changes to the correlations. This will also
help to examine how consistent the indicators represent the
sustainability metrics as the time series grows.
We realize that simple correlation calculations and
paired plots are restricted in their ability to predict actual
changes in the different sustainability metrics and, although
informative, other approaches for such predictions may be
more appropriate. Our purpose for this paper simply was to
emphasize the potential problem for practical use of sus-
tainability metrics and provide a potential avenue to address
the problem. Time lags will clearly hamper the usefulness
of sustainability metrics for a number of reasons. First, the
lags limit decision support for understanding changes from
management actions in the present year because the impact
of a management decision would not be seen until data are
available. The result is that decisions are based on old data
and may not improve or address the current state of the
system. Second, decision makers require real-time data to
understand how close their system is to critical thresholds.
If a system is nearing a threshold or undergoing a regime
shift, providing the ability to respond to changes earlier will
give managers and decision makers the best opportunity to
make sound decisions to prevent the shift or better prepare
for resulting changes. Sustainability metrics that are always
3 years behind will not help to identify how close the sys-
tem is to the threshold.
We have limited the analysis to correlation analyses even
though some of the scatter plots identify potential nonlinear
relationships. One future research recommendation is
applying econometric analyses, such as time series analyses
or autogressive models (AR), to predict future estimates of
GNRP and EFA. Time series analysis can describe the path
of a sustainability metric in terms of lagged or same-year
variables, previous years of the sustainability metric, and
disturbances (Greene 1993). This econometric technique can
be used for forecasting the sustainability metrics if appro-
priate data are available (including some of these near-time
indicators). These approaches also can identify the statistical
significance of covariate relationships and explore con-
founding variables. We emphasize that the paper presents an
exploratory methodology, so future research is warranted to
explore other methodologies and indicators to eliminate the
time lag between metric calculations and current year.
Having closer to real time data would enable decision makers
to manage better large complex systems such as the SLB
because there would no longer be a 3-year lag in the sus-
tainability metric calculations.
Acknowledgments The views expressed herein are strictly the opin-
ions of the authors and in no manner represent or reflect current or
planned policy by the federal agencies. Mention of trade names or
commercial products does not constitute endorsement or recommenda-
tion for use. The authors acknowledge the San Luis Basin Sustainability
Metrics team and collaborators for their contribution on the pilot study.
We thank the two anonymous reviewers for their insightful comments.
References
Chambers N, Simmons C, Wackernagel M (2000) Sharing nature’s
interest: ecological footprints as an indicator of sustainability.
Earthscan Publications Ltd., London
Colorado Water Conservation Board (CWCB) (2000) Rio Grande
Decision Support System: An Overview. Section 4.0 Water
Administration Issues. CWCB, State of Colorado. Available
online at http://cdss.state.co.us/DNN/RioGrande/tabid/57/Default.
aspx. Accessed January 24, 2011
Grambsch A, Michaels R, Peskin H (1993) Taking stock of nature:
environmental accounting for Chesapeake Bay. In: Lutz E (ed)
Toward improved accounting for the environment. A UN Stat-
World Bank Symposium, IBRD-World Bank, New York, NY,
pp 184–197
Greene W (1993) Econometric analysis. Prentice Hall, NJ
Gren I, Isacs L (2009) Ecosystem services and regional development:
an application to Sweden. Ecol Econ 68:2549–2559
Gundimeda H, Sukhdev P, Sinha R, Sanyal S (2007) Natural resource
accounting for Indian states—illustrating the case of forest
resources. Ecol Econ 61:635–649
Hanley N, Moffatt I, Faichney R, Wilson M (1999) Measuring
sustainability: a time series of alternative indicators for Scotland.
Ecol Econ 28:55–73
Heberling MT, Hopton ME (2012) San Luis Basin sustainability
metrics project: a methodology for evaluating regional sustain-
abilitySpecial Collection. J Environ Manage 111:272–278
Heberling MT, Templeton JT, Wu S (2012) Green net regional
product for the San Luis Basin, Colorado: an economic measure
of regional sustainability. J Environ Manage 111:287–297.
doi:10.1016/j.jenvman.2012.02.035
Hopton ME, White D (2012) A simplified ecological footprint at a
regional scale. J Environ Manage 111:279–286. doi:10.1016/j.
jenvman.2011.07.005
Jackson LE, Kurtz JC, Fisher WS (eds) (2000) Evaluation guidelines
for ecological indicators. EPA/620/R-99/005. U.S. Environmen-
tal Protection Agency, Office of Research and Development.
Kluwer Academic Publishers, Research Triangle Park, NC
Mayer A (2008) Strengths and weaknesses of common sustainability
indices for multidimensional systems. Environ Int 34:277–291
National Agricultural Statistics Service, United States Department of
Agriculture (NASS) (2009) United States Census of Agriculture:
2007. Available online at http://www.nass.usda.gov/Statistics_
by_State/Colorado/index.asp. Last accessed June 13, 2012
Nourry M (2008) Measuring sustainable development: Some empir-
ical evidence for France from eight alternative indicators. Ecol
Econ 67:441–456
Ochola W, Kerkides P, Argyrokastritis I (2003) Water quality: an
indicator for sustainable land management? European Water
3(4):17–24
Pezzey J (2004) One-sided sustainability tests with amenities, and
changes in technology, trade, and population. J Environ Econ
Manag 48:613–631
Pezzey J, Hanley N, Turner K, Tinch D (2006) Comparing augmented
sustainability measures for Scotland: is there a mismatch? Ecol
Econ 57:60–74
Assessing sustainability when data availability limits real-time estimates 747
123
Rees WE (1992) Ecological footprints and appropriated carrying
capacity: what urban economics leaves out. Environ Urban
4:120–130
Rosenzweig C, Tubiello FN, Goldberg R, Mills E, Bloomfield J
(2002) Increased crop damage in the US from excess precipi-
tation under climate change. Global Environ Chang 12:197–202
Siegel AF, Morgan CJ (1996) Statistics and data analysis: an
introduction. Wiley, New York
U.S. Environmental Protection Agency (USEPA) (2010) San Luis
Basin Sustainability Metrics Project: A Methodology for Eval-
uating Regional Sustainability. EPA/600/R-10/182 USEPA,
Cincinnati
Wackernagel M, Rees W (1996) Our ecological footprint: reducing
human impact on the earth. New Society Publishers, Gabriola
Island
Zar JH (1999) Biostatistical analysis. Prentice Hall Inc., Upper Saddle
River
Zarnowitz V (1992) Business cycles: theory, history, indicators, and
forecasting. University of Chicago Press, Chicago
Author Biographies
Matthew T. Heberling is an
economist at the U.S. Environ-
mental Protection Agency’s
(USEPA’s) National Risk Man-
agement Research Laboratory in
Cincinnati, Ohio. He holds a
Ph.D. in agricultural economics
from The Pennsylvania State
University, where he specialized
in environmental and natural
resource economics. Matt joined
the USEPA in 2001 to work in a
program of research integrating
ecological risk assessment and
economic analyses. He is now
studying the effects of ancillary benefits on market mechanisms and
conducting research on sustainability metrics. His research experience
also includes using economic valuation methods to examine indi-
viduals’ preferences for recreational fishing and to prioritize stream
restoration.
Matthew E. Hopton is an ecol-
ogist at theUSEPA in the National
Risk Management Research Lab-
oratory in Cincinnati, OH. He has
a Ph.D. in biology from Univer-
sity of Cincinnati, where he
examined the relationship
between environmental heteroge-
neity and biological diversity.
Matt started at the USEPA as a
NRC post-doctoral research
associate in 2006 in the Sustain-
able Environments Branch. He is
co-leader of the Regional Envi-
ronmental Management research
group, a multidisciplinary group that consists of ecologists, economists,
engineers, geographers, and physical scientists. His current research
interests include two foci; the ecological function and service provided by
green infrastructure in urban systems and measuring and managing sus-
tainability in regional systems.
748 M. T. Heberling, M. E. Hopton
123