Content uploaded by Newsha K. Ajami
Author content
All content in this area was uploaded by Newsha K. Ajami on Aug 29, 2019
Content may be subject to copyright.
AnAlysis
https://doi.org/10.1038/s41893-019-0228-z
1Brown and Caldwell, Lakewood, CO, USA. 2ReNUWIt Engineering Research Center, Stanford University, Stanford, CA, USA. 3Woods Institute for the
Environment, Stanford University, Stanford, CA, USA. *e-mail: newsha@stanford.edu
The water sector today faces social and hydrological uncertain-
ties related to climate change, growing urban populations,
shifting water use patterns and ageing physical infrastruc-
ture1–4. Although there is no single solution to these challenges,
many communities are exploring implementation of alternative
water supplies, such as water recycling, storm-water capture and
reuse, and water conservation5–8. These adaptation strategies focus
on a transition towards a ‘One Water’ approach9—a concept that
is closely aligned with, and builds on, work on integrated water
resources management10,11. The One Water concept is centred on
the realization that all water resources have value and should be
managed in a sustainable and integrated way, both throughout the
full water cycle and across the traditionally siloed institutions that
manage water resources today. This transition towards One Water
systems is gradually disrupting the conventional centralized water
infrastructure model, by introducing hybrid infrastructure net-
works that combine centralized and decentralized sources at vari-
ous scales. Effectively managing these hybrid systems requires new
platforms that can address the needs of diverse stakeholders and
take advantage of possible funding and operational partnerships.
To help support this transition, this paper presents a goal-based
water-trading platform for supply portfolios. As with cap-and-trade
applications in other sectors, this approach relies on both regulatory
and market incentives, and allows various actors to engage in the
decision-making process while investing in local and economical
solutions, from water conservation to the expansion of water sup-
plies beyond traditional sources.
Water supply portfolio diversification can reduce the risks asso-
ciated with physical and institutional characteristics of each supply
source, such that a disruption or stressor that affects one compo-
nent of the system is unlikely to affect the other components. For
example, a portfolio that introduces a single centralized recycled
water plant may not be as robust as a portfolio that leverages mul-
tiple supply projects from various sources and locations, as each
project will be subject to different risks under different challenges12.
Unfortunately, diversification efforts are typically pursued either by
local agencies in isolation, with insufficient consideration of poten-
tial impacts and benefits to the larger system, or via a top-down
approach without enough information about the roles and charac-
teristics of individual system components. To address these limita-
tions, reinvention of the water sector and the transition towards a
true One Water approach will require not only technological solu-
tions but also institutional changes that address regulatory, financial
and governance components13.
In this transition, the water sector could learn from other sec-
tors that have effectively implemented change with a combination
of regulatory control and market incentives. Traditional water mar-
kets have seen varying success around the world, in part due to the
complex top-down nature of water rights, which often allocate more
water than is physically available, as well as a lack of institutional
capacity and information support, which lead to poor stakeholder
engagement14. In contrast, more flexible approaches, such as goal-
based trading systems used for emissions control, renewable energy
adoption and even water quality enhancement have proven effec-
tive across various contexts and scales15–17. Goal-based frameworks
couple regulatory requirements (such as an emissions cap, a target
energy portfolio or multiple requirements in tandem) with eco-
nomic incentives to allow actors to implement changes in the way
that best fits their needs. This can lead not only to efficient collective
action towards environmental goals, but also to more cost-effective
allocation of resources than traditional administrative systems18,
as well as enabling an environment for innovation17,19–21 while pro-
viding an avenue to manage and share the risks and costs of new
investments22,23. These characteristics make goal-based trading
approaches highly relevant in supporting water supply diversifica-
tion (WSD) efforts in the One Water movement.
To demonstrate the potential for a goal-based trading model in the
water sector, we propose a system of tradable WSD credits (Fig. 1).
Under this approach, a diversification goal would be set for a region
of water utilities interconnected through common supplies, such as a
river system or a groundwater basin. Utilities receive credits for imple-
menting new water supply projects locally to ease their reliance on the
Goal-based water trading expands and diversifies
supplies for enhanced resilience
P.Gonzales 1 and N.K.Ajami 2,3*
In response to rising water scarcity concerns around the world, water utilities are expanding and diversifying their water supply
portfolios, including attempting to value and reuse every drop throughout the water cycle. This transition is creating new hybrid
infrastructure systems that combine centralized and decentralized sources at various scales. To help strategize and manage
these emerging hybrid systems, we introduce a flexible goal-based water-trading framework using a combination of regulatory
and market incentives. We apply this framework in a simulation of the San Francisco Bay Area, and test the effects of policy
parameters, local preferences and collaboration between utilities. Results show that the option of engaging in an open trading
scheme can lead to a more strategic and holistic path to higher regional resiliency through increased diversity in water portfo-
lios, at lower costs than if utilities were tasked with pursuing goals independently. Highest benefits are observed when utilities
cooperate in the exchange of information, which highlights the importance of transparency and trust operating in conjunction
with regulatory and market forces.
NATURE SUSTAINABILITY | VOL 2 | FEBRUARY 2019 | 138–147 | www.nature.com/natsustain
138
AnAlysis
NATuRe SuSTAINABIlITy
common supply sources, which they can then trade in a regional mar-
ket. Utilities that can generate credits at or below market price would
have an incentive to do so with the expectation that they will be sub-
sidized for their investments, whereas utilities with higher marginal
costs would be likely to engage in buying credits instead. Through
the proposed goal-based water-trading scheme, we argue that exist-
ing networks (both physical and social) present an opportunity for
communities within a region to coordinate their efforts and leverage
locally beneficial diversification projects in the pursuit of common
regional goals. This paper presents a framework for developing such a
scheme, and demonstrates its implementation by exploring a range of
possible outcomes under various policy parameters, decision-making
dynamics and levels of information exchange, simulated in a case
study of the San Francisco Bay Area. Our combination of conceptual
framing and simulation modelling demonstrates the potential bene-
fits of collaboration, addresses important policy design questions, and
thus lowers entry barriers for more widespread adoption of innova-
tive management strategies for WSD.
Simulation study in the San Francisco Bay Area
The San Francisco Bay Area exemplifies many of the water chal-
lenges of the twenty-first century that make WSD a relevant issue.
With rapid growth, stressed water supplies and increasing vul-
nerability to climate risks, the region faces significant pressure to
enhance the resilience of its water systems24. Previous studies esti-
mate economic losses to businesses, residents and infrastructure
on the order of US$7–14 billion under scenarios of water supply
disruption, depending on the location and extent of water supply
deficiencies25. At the same time, the Bay Area, though collectively
known as a unit, is composed of numerous diverse communities
interconnected through complex networks of water supplies24. To
effectively capture the local characteristics of a diverse region,
we focus this study on a single water system served by Sonoma
County Water Agency (SCWA) (Supplementary Fig. 1). SCWA is
a wholesale agency in the northern region of the Bay Area that
provides water and services to nine cities and special districts in
Sonoma and Marin counties, which collectively serve 600,000
residents. SCWA is unique in that most of the water it delivers
to its retailers comes from local sources—primarily the Russian
River, Lake Sonoma and Lake Mendocino. All 9 water utilities that
purchase water from SCWA rely heavily on these common sup-
plies (61% of the regional supply portfolio), but also have access to
local surface water (28% regionally) and groundwater (5% region-
ally). Most utilities are also currently using recycled water as a
resource (5% regionally). The existence of a common pool supply,
heterogeneity in local water supply portfolios among utilities,
and the presence of SCWA as a coordinating governance layer are
all elements that make SCWA a good test bed for the proposed
goal-based water-trading scheme.
Figure 2 shows marginal cost curves for several alternative sup-
ply projects identified as feasible by each of eight water utilities
served by SCWA (the ninth utility was excluded due to limited
data availability) (Supplementary Information). Dashed lines rep-
resent the equivalent of a hypothetical target requiring that at least
25% of the total supply for each utility must come from alternative
sources (representing a total of ~21,000 acre-feet yr−1 for the region
as a whole). This shows that while not every utility would be able
to achieve such a goal independently, the region as a whole could
do so when aggregating all projects. For demonstration purposes,
only alternative supply projects (water recycling, rainwater and grey
water reuse, storm-water, conservation and efficiency, desalination
and emerging options) that have been identified by each utility in
publicly available documents are counted towards the diversifica-
tion goal in this study.
We used data from this case study basin to simulate the effective-
ness of our goal-based water-trading scheme under various policy
parameters and interactions between actors’ preferences. We com-
pared a combination of three deterministic conditions to provide
insights that can help guide future policy development. Details on
the policy framework and agent-based simulation approach are pro-
vided in the Methods. The following conditions were included in
simulations (Supplementary Fig. 2):
(1) Policy drivers. To provide a baseline, we simulated either a
purely regulatory policy (that is, utilities must comply with the
goal individually and no trading is allowed) or a combination
of regulatory and market incentives (that is, trading is allowed).
In both cases, the simulated goal is that at least 25% of the total
supply for each utility must come from alternative sources by
the year 2040.
(2) Decision strategies. We modelled utility decisions as either
based solely on the per-acre-foot cost of a given supply (that
is, cost-driven decision strategies) or skewed by the perceived
benets of multi-faceted projects (that is, multi-criteria de-
cision strategies). e multi-criteria strategies in this study
simulate a hypothetical preference for water recycling and
storm-water over other types of projects. is preference ex-
emplies additional local benets associated with drought
resilience, ood control and water quality enhancement that
result from implementing these two supply sources over tra-
ditional options.
WSD goal
Credits sold
$ subsidy
$ payment
Credits bought
WSD goal
Common pool
of water
resources
Water supply
diversification
(WSD) goals
Flexible implementation of alternative water
sources with incentives to work together
towards common goals
System of tradable
credits
+=
Fig. 1 | Goal-based trading scheme concept. Conceptual diagram of a goal-based trading scheme for water supply portfolio diversification.
NATURE SUSTAINABILITY | VOL 2 | FEBRUARY 2019 | 138–147 | www.nature.com/natsustain 139
AnAlysis NATuRe SuSTAINABIlITy
(3) Trading dynamics. We modelled the trading dynamics as either
independent or coordinated behaviours. Independent trading
strategies assume that utilities do not engage in collective plan-
ning, and instead make decisions based only on their own infor-
mation about market results, making strategic decisions to mini-
mize their own costs. Conversely, coordinated trading strategies
assume that utilities are acting collectively to achieve the diversi-
cation goal while minimizing the overall costs to the region.
Results
Evolution of water supply portfolios over time. We assessed the
evolution of water supply portfolios over time under each set of
simulated conditions. The resulting water supply portfolios show
that a regional diversification goal of 21,000 acre-feet yr−1 (corre-
sponding to 25% of the total supply portfolio) could be achieved
under various conditions. The specific portfolio composition var-
ies under different simulations of policy drivers, decision strategies
and trading dynamics (Fig. 3), with a mix of conservation and effi-
ciency (26–33%), water recycling (43–65%) and storm-water cap-
ture and reuse (8–10%) in each simulated portfolio. Desalination
also becomes a preferred option (up to 22% of the portfolio) under
cost-driven trading simulations, although it is displaced by recycled
water projects under multi-criteria decision strategies as utilities are
willing to implement those projects even if they are costly, due to
their non-monetary benefits.
Very different outcomes result from the two contrasting policy
drivers. An important finding is that the regional diversification
goal is never achieved under a no-trading policy as not all utilities
have cost-effective opportunities, or may instead choose to pay a
non-compliance fine. Even though the fine for this simulation study
was set at a level consistent with the regional marginal costs to
achieve the desired target (~US$3,000 acre-foot−1 , as derived from
Fig. 2), a central agency defining a regulatory penalty has only lim-
ited information about the internal cost structure of individual utili-
ties, which makes it difficult to provide an appropriate economic
incentive for the diverse actors involved. This issue reflects informa-
tion asymmetries—a problem widely documented in other markets
that in some cases has been addressed via additional policy specifi-
cations, such as adaptive penalties26.
In contrast with no trading, or a solely regulatory approach, we
find that the trading policy driver leads to a regional water sup-
ply portfolio that more closely follows the regional goal over time
(Fig. 3). While the two potential trading dynamics, independent
and coordinated, lead to similar portfolios, there are important
differences in the timing and composition of the two. First, the
independent trading dynamics lead to the development of surplus
diversification credits early on, resulting in considerable banking
of credits that are used to satisfy the more stringent diversification
goal in later years. Surplus credits and banking occur to a lesser
extent in the coordinated trading dynamics. Thus, while both trad-
ing dynamics lead to the desired goal in terms of diversification
credits at the end of the policy period, independent and coordi-
nated simulations result in a shortfall by volume on the order of
1,200–3,200 and 500–1,500 acre-feet yr−1, respectively. This is intui-
tively consistent with the expectation that independent decisions
0
10,000
20,000
30,000
0
2,000
4,000
0
2,000
4,000
0
2,000
4,000
01,000 2,000 3,000 4,000 5,000
0
2,000
4,000
0
2,000
4,000
0
2,000
4,000
0 5,000 10,000 15,000 01,000 2,000 3,000 4,000 5,000
0 1,000 2,000 3,000 4,000 5,000 01,000 2,000 3,000 4,000 5,000
01,000 2,000 3,000 4,000 5,000
0 1,000 2,000 3,000 4,000 5,000 01,000 2,000 3,000 4,000 5,000
0
2,000
4,000
25%
25%
25%
25%
25%
25%
25% 25%
0 5,000 10,000 15,000 20,000 25,000
0
10,000
20,000
30,000
25%
Marginal cost (US$ acre-foot
−1
)
Desalination
Stormwater
Emerging
options
Rainwater/
greywater
Conservation
and efficiency
Water recycling
and reuse
Cumulative supply capacity (acre-feet yr
−1
)
SCWA region (aggregate)
Utility C Utility D
Utility A Utility B
Utility E Utility F
Utility G Utility H
Fig. 2 | Marginal cost curves of alternative supply-diversification projects identified in eight water utilities served by the SCWA. Note that the results
for utility A are plotted on different x and y axis scales compared with the other utilities, owing to the much larger supply portfolio of this utility. Bar widths
represent project capacity (normal year yield), and bar heights represent annualized project cost. Red dashed lines intersect the volume equivalent to a
hypothetical goal of introducing 25% diverse water supplies (by volume) into each portfolio. Existing recycled water supplies are not shown, but were
included in the accounting. Emerging options include fog capture, cloud seeding and watershed management.
NATURE SUSTAINABILITY | VOL 2 | FEBRUARY 2019 | 138–147 | www.nature.com/natsustain
140
AnAlysis
NATuRe SuSTAINABIlITy
are subject to greater uncertainty and are thus more likely to result
in optimistic investments, whereas coordinated decisions are made
with the knowledge that there will be willing buyers and sellers in
the goal-based trading market for the diversification credits gener-
ated. The role of coordination in managing investment uncertain-
ties is also visible in the composition of the final portfolios, where
the coordinated trading dynamics lead to the staging and scaling
down of a desalination project. This result is consistent with better
information availability, which allows utilities to plan for the timely
implementation of capital-intensive projects and may lead to more
willingness to invest in new infrastructure.
Another important finding is that decision strategies, represent-
ing different internal motivations to pursue specific types of proj-
ects, can affect the resulting water supply portfolio of the region.
While the final portfolios are not considerably different between
cost-driven and multi-criteria preference simulations, multi-crite-
ria preference simulations lead to a 20–35% increase in contribu-
tion from recycled water and storm-water projects combined, as
well as displacing a desalination project, with specific portfolios
varying across simulations of policy drivers and trading dynam-
ics. While we explicitly defined this multi-criteria preference for
demonstration purposes in our model (Methods), it is qualitatively
consistent with One Water principles, including recent efforts in
the region to develop water recycling and storm-water as valu-
able resources that can help augment water supply portfolios.
Preferences for these projects over desalination are also consistent
with additional environmental costs typically associated with brine
disposal from desalination technologies, as well as higher energy
costs and carbon emissions.
Regional outcomes. In addition to the composition of specific water
supply portfolios resulting from various simulated conditions, we
looked at cost-effectiveness, market engagement and the final vol-
ume of alternative supplies as measures of policy success (Methods).
Under our simulations, a no-trading policy leads to an average cost
of alternative supplies of US$1,524 acre-foot−1, compared with
US$1,818 acre-foot−1 when trading based on independent decision-
making, and US$1,346 acre-foot−1 when trading efforts are coordi-
nated (all when decisions are cost-driven; Fig. 4a). The introduction
of multi-criteria preferences leads to slightly higher costs (6–14%
higher) for the no-trading simulation or trading with coordinated
decisions, as utilities are incentivized to pursue projects even if they
are costly because of their perceived multi-faceted benefits.
In contrast, multi-criteria preferences can lead to lowering
costs in a trading scenario where utilities are making decisions
independently. This discrepancy is the result of utilities’ ability to
bank excess diversification credits. As detailed in the Methods, the
independent trading dynamic assumes a lack of information about
other utilities’ investments, resulting in an imbalance between sup-
ply and demand for credits. As a result, when utilities are incentiv-
ized to start implementing storm-water and recycled water projects,
which have a shorter implementation timeline than larger projects
such as desalination, the independent trading dynamic leads to a
greater excess of diversification credits early on, which are banked
and applied to comply with targets in later periods. Thus, the lower
costs in these scenarios compared with coordinated trading should
be considered in conjunction with the observation of a larger short-
fall from the regional goal.
The imbalance of diversification credits under independent
trading dynamics is also observed in the low volume of trading
(Fig. 4b). These results suggest that a lack of coordination can
lead to the region not taking full advantage of the market incen-
tives, and can in fact create disincentives for utilities to meet their
diversification targets. Nevertheless, while the option of banking
credits can result in the region not meeting its end goal, it pro-
vides an important avenue for utilities to manage the risk of new
investments. Additional policy rules and limitations to the bank-
ing of credits could help mitigate this uncertainty while promoting
compliance with the regional goal, as is often done in markets for
ecosystem services27.
As these results represent a limited set of simulated conditions,
we also performed a sensitivity analysis on policy parameters and
modelling assumptions (Supplementary Figs. 4–6). We found that
the level of penalty for non-compliance and the WSD goal play an
important role in final outcomes, whereas the effect of behavioural
assumptions and the level of multi-criteria preferences is not as
salient. This is an important consideration for policy-makers who
would need to define regulatory conditions a priori. Most impor-
tantly, we find that the general benefits of coordinated trading
over independent trading dynamics or a no-trading policy hold
true across a wide range of simulated conditions, highlighting the
importance of promoting collaboration and transparency as drivers
of policy success.
Diversity as a measure of resilience. The concept of diversification
as a strategy to mitigate risk has been widely adopted in portfolio
0
10,000
20,000
30,000
0
10,000
20,000
30,000
0
10,000
20,000
30,000
0
10,000
20,000
30,000
0
10,000
20,000
30,000
2020 2025 2030
Year Year Year
Year Year Year
2035 2040 2020 2025 2030 2035 2040 2020 2025 2030 2035
2040
2020 2025 2030 2035 2040 2020 2025 2030 2035 2040 2020 2025 2030 2035
2040
0
10,000
20,000
30,000
Acre-feet yr−1
Acre-feet yr−1
No trading Independent trading Coordinated trading
WSD goal
Conservation
and efficiency
Water recycling
and reuse
Desalination Stormwater Emerging
options
Rainwater/
greywater
WSD goal
WSD goal
WSD goal
WSD goal
WSD goal
Acre-feet yr−1
Acre-feet yr−1
Acre-feet yr−1
Acre-feet yr−1
Multi-criteria Cost driven
Fig. 3 | Alternative supplies added to the regional portfolio over time under different simulated conditions of policy drivers, decision strategies and
trading dynamics. Dashed lines show the phased implementation of a WSD goal.
NATURE SUSTAINABILITY | VOL 2 | FEBRUARY 2019 | 138–147 | www.nature.com/natsustain 141
AnAlysis NATuRe SuSTAINABIlITy
theory in finance, ecology and social sciences28,29 because diverse
systems are better able to handle uncertainty. Diversification
holds similar promise as a quantitative measure of resilience for
water resource systems30. To better account for the possible resil-
ience benefits of a diverse water supply portfolio, we looked not
only at the volume of alternative supplies, but also the breadth and
depth of these expanded portfolios. To measure this, we used the
Gini–Simpson diversity index—a common indicator of diversity in
other sectors28 that has recently been adapted to assess water supply
portfolios30,31. This diversity index considers the proportion of each
supply (s) as a fraction of the total portfolio (n sources) and com-
bines them into a 0–1 scale, where a higher index represents higher
diversity:
−∑=s1i
ni
12
. This metric is not intended to be used as a goal
in itself, but rather as a way to benchmark the current state of the
system against possible scenarios that can lead to higher resilience.
Ideally, each region implementing WSD policies should set regional
goals that explicitly target improving resilience and minimizing
risks. Nevertheless, a diversity index provides a useful heuristic
metric for analysis given a lack of empirical data on the relative risks
of alternative supply sources.
The current regional water supply portfolio without the simu-
lated policy (based on 2015 data) has a diversity index of 0.54. As
shown in Fig. 5, even a modest diversification goal of 10% (equiv-
alent to ~8,400 acre-feet of new alternative supplies) can lead to
clear enhancements in the breadth and depth of the regional port-
folio, but a higher degree of diversification correlates with higher
costs. This is consistent with the expectation that diversification
requires additional investments on a wide variety of projects, as
opposed to relying on a single solution that may be associated
with lower costs, such as investing solely in conservation. Fig. 5
also shows the variation between supply diversity and costs when
different diversification goals and multi-criteria preferences are
simulated. These results show that higher portfolio percentage
goals can effectively lead to a higher diversity index. The relation-
ship between multi-criteria preferences and the diversity index
is weaker, as demonstrated by multiple scenarios overlapping or
closely following the same results for a given diversification goal.
Thus, we find that setting a supply diversification goal in terms
of a portfolio percentage composition of alternative supplies,
though an imperfect metric of resilience, is an effective way to
increase both breadth and depth in supply portfolios regardless
of uncertainties associated with utilities’ preferences or imperfect
information. Furthermore, this correlation holds across policy
drivers and trading dynamics.
Discussion
Policy design. Our simulation study demonstrates the potential for
regulatory drivers and market incentives to complement each other
in the pursuit of regional WSD goals through a goal-based water-
trading system. Examples of regulatory goal setting in line with the
One Water movement are already prevalent in the water sector at
various scales, particularly in regions facing chronic water secu-
rity concerns around the world, such as South Africa, the Western
United States and Australia. In California, for instance, recent
regulations are calling for state-wide water conservation standards
(Assembly Bill 1668) and better management of groundwater
resources (Sustainable Groundwater Management Act), and regions
in densely populated southern California have set long-term goals
to decrease dependence on imported supplies32. In each of these
adaptation measures, California regulators are also experiment-
ing with new ways to provide flexible incentives and bottom-up
accountability measures such as providing financing mechanisms
for collective projects and requiring regular reporting to the state.
However, measures to bridge the gap between top-down and bot-
tom-up approaches are still far from perfect. Our work demon-
strates that goal-based water-trading schemes can bridge this gap
by providing a platform for compliance with emerging regulations
while prompting collective action and coordination, thus ensuring
funds are deployed to the most cost-effective and beneficial water
projects. These kinds of regional efforts will become increasingly
relevant as local and alternative water supplies become more cost-
effective compared with the cost of building large centralized infra-
structure projects or doing nothing24,33.
To aid in policy development, the goal-based water-trading frame-
work presented here is flexible and could include targets, timelines,
market trading mechanisms and eligible resources that are suitable
to a given location, whether that location is defined as a neigh-
bourhood, city, utility service area or larger region (see Fig. 6 and
Methods). Furthermore, our work suggests that this flexibility should
expand beyond the initial policy design to also allow for adaptation
over time. For example, our model results demonstrate the important
role that the policy goal and the non-compliance penalty can play
on shaping water supply portfolios, despite uncertainties in defin-
ing these parameters. Although information gaps and asymmetries
can create a barrier to selecting appropriate policy drivers, previous
studies suggest that negative impacts can be reduced through adap-
tive management approaches11,34. Adaptive management is generally
defined as a systematic strategy to continually improve policies and
practices based on learning from the outcomes of past management
No trading
Independent
Coordinated
No trading
Independent
Coordinated
No trading
Independent
Coordinated
0
500
1,000
1,500
2,000
abc
0
500
1,000
1,500
0
10
20
30
40
50
Cost-driven Multi-criteria
Average cost of
alternative supplies
(US$ acre-foot
–1
yr
–1
)
Average volume traded
(acre-feet yr
–1
)
Alternative supplies in
the portfolio (%)
1,524
1,614 1,818
1,411
1,346
1,537
00 32 83
1,423 1,348
18
21 23 21 24 23
Fig. 4 | Regional outcomes under various conditions at the end of the simulation period. a–c, Measured outcomes include the cost of alternative supplies
(a), volume traded (b) and percentage of alternative supplies in the portfolio (c). In c, the red dashed line represents the WSD goal of 25%; ‘alternative
supplies’ account only for the volume of actual water supplies (and do not reflect additional diversification credits that may be acquired through the market
and banked to help to meet the 25% end goal).
NATURE SUSTAINABILITY | VOL 2 | FEBRUARY 2019 | 138–147 | www.nature.com/natsustain
142
AnAlysis
NATuRe SuSTAINABIlITy
actions11. For example, previous studies suggest the use of adaptive
penalties to ensure compliance26. Allowing social learning as part of
the policy development process could help ensure a balance between
regulatory and market drivers, thus maximizing the benefits of a
goal-based water-trading scheme.
The role of regional coordination. Our results highlight the impor-
tance of incentivizing close collaboration and transparency among
stakeholders. This is exemplified in the fact that coordinated trad-
ing dynamics consistently led to cost savings and greater diversifica-
tion, and ultimately higher resiliency, both in terms of the volume
of alternative supplies and the diversity index, compared with inde-
pendent trading dynamics and no-trading conditions. It is worth
noting that our coordinated and independent trading dynamics
demonstrate a wide range of possible outcomes, whereas real-world
situations are likely to fall somewhere in between depending on
collective behaviours and incentives. Nevertheless, these findings
are important considerations because they indicate that individual
behaviours can affect the results of a given policy and cause them to
diverge from the expectations of static top-down economic models
typically used in policy analysis35.
In addition, although our simulation results agree with previ-
ous literature on the importance of collective action in achieving
regional goals36, the act of coordinating is a process in itself, subject
to uncertainties and politics, and requiring effective interaction
with existing management systems. The presence of bridging orga-
nizations, such as the role played by SCWA as a coordinating agency
in this study basin, can greatly decrease the barriers to effective col-
laboration and help create tools to reduce transaction costs37. In this
sense, our work suggests that placing the priorities and governance
of water diversification policies in the hands of regional actors can
better incorporate local realities, promote transparency and increase
efficiency by leveraging existing interconnected resources. Most
importantly, for coordination to be effective, such regional scope
should be aligned with the institutional and governance context of
a given location, whether that is abounded by a neighbourhood, a
city, a watershed, and so on.
Institutional needs and limitations. An important consideration
for designing a goal-based water-trading scheme as an effective
incentive for collective action is how such a scheme would fit in
with existing governance and institutional systems. Policy develop-
ment requires that regulatory and market incentives be properly
aligned, which in many cases could require significant reform to the
complex structure of current approaches for water supply, wastewa-
ter and storm-water management. A literature review of the institu-
tional challenges to a One Water approach highlights five key areas:
legislation and regulations, economics and finance, planning and
Average cost of alternative supplies
(US$ acre-foot−1 yr−1)
Multi-criteria tradeoff parameter
500
1,000
1,500
2,000
abc
0.60 0.65 0.70 0.75 0.80
500
1,000
1,500
2,000
500
1,000
1,500
2,000
Supply diversity index (0–1)
0.60 0.65 0.70 0.75 0.80 0.60 0.65 0.70 0.75 0.80 10
15
20
25
30
35
40
WSD goal (%)
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Fig. 5 | Average cost of alternative supplies versus supply diversity (Gini–Simpson index). a–c, Three scenarios are shown: no trading (a); independent
trading (b); and coordinated training (c). Each bubble represents the result of a simulation under a given WSD goal (colour scale) and trade-off parameter
(size, where a larger bubble puts more emphasis on multi-criteria decision-making), all other factors being constant. Thick bubble edges reflect overlapping
results where multiple concentric circles yield the same x and y axis values from different simulations. A higher value of the diversity index indicates better
performance. A higher multi-criteria trade-off parameter indicates higher preference to implement recycled water and storm-water projects.
Drivers
Regional
goal
Market
rules
Compliance
mechanisms
Considerations
Long-term needs
Geographical scope
Policy timeline
Source 25% of the water
supply portfolio from
alternative sources by 2040
Centralized uniform price
auction
Forward market
(every 5 years)
Current-year market (yearly)
Phased implementation
Banking of credits is allowed
Fixed volumetric fine
(US$ acre-foot
−1
)
Auction type and format
Auction timeline
Selection criteria
Compliance criteria
Allowances
Penalties
Example
(this study)
Fig. 6 | Policy design considerations. Policy design considerations for a goal-based trading scheme for water portfolio diversification.
NATURE SUSTAINABILITY | VOL 2 | FEBRUARY 2019 | 138–147 | www.nature.com/natsustain 143
AnAlysis NATuRe SuSTAINABIlITy
collaboration, culture and capacity, and citizen engagement10. This
study addresses two of these areas by providing a new economic
framework and a platform for collaboration that can add flexibility
to existing systems. However, fully achieving a One Water system
can only be possible if the other three areas are addressed as well.
While solutions will need to be location specific, some of these bar-
riers are common to the traditional water management paradigm
and are prevalent around the world. For example, adaptation strat-
egies will generally require overcoming the dominant paradigm
of centralized and siloed systems, re-defining cost structures and
financing strategies, clarifying policy positions regarding new tech-
nologies, and shifting public perceptions about non-traditional
water sources. These changes are already being incentivized in
response to emerging water challenges.
Both new research and new technologies could help address
some of the existing limitations. While this study presents a start-
ing point for new policy development, it only addresses a subset
of important questions. Future research should additionally test
the proposed policy framework under different contexts, such as
incorporating the effects of technological advancement and climate
uncertainties, conducting comparative case studies across diverse
regions facing different local characteristics and challenges, and
explicitly modelling distribution systems to incorporate storage
requirements and system-level robustness under trading scenarios.
Additionally, this study does not explicitly address the practical pro-
cess of coordinating decision-making. This could be envisioned as
an informal process arising from existing networks between utility
decision-makers or through formal venues such as board meetings,
but additional benefits could be derived from information technol-
ogy applications. For example, previous studies have recommended
leveraging interactive data management platforms to help enable
markets and promote coordination38. This can help reduce infor-
mation, exchange and monitoring costs, reducing transaction costs
overall similar to tools that are already in place in goal-based trading
schemes for water quality, renewable energy and carbon emissions.
Conclusion
This study demonstrates the untapped potential to develop goal-
based trading schemes for water portfolio diversification as a way
of furthering the reliability and resilience of urban water systems.
In line with the One Water movement, this policy instrument
would enable more diverse water sources to be valued and incorpo-
rated into supply portfolios, while allowing diverse stakeholders to
achieve multiple benefits, encouraging more inclusive partnerships,
and promoting a system-level mindset in the management of water
resources. While meaningful reform will not happen overnight,
proof-of-concept analyses and guiding frameworks as presented in
this work can help catalyse new policy adoption and implementa-
tion. In practice, enhancing the resilience of our water systems will
require both new knowledge and synergistic policy development.
These efforts should include exploring regulations, incentives and
new tools that can incorporate the valuation of all available water
resources and the different perspectives of their diverse users.
Methods
Data. Data on existing water supply portfolios and potential diversication
projects were obtained from urban water management plans and other
publicly available documents. A complete list of data sources and assumptions
is presented in the Supplementary Information. For each potential supply
diversication project, we extracted data on the expected supply yield (acre-
feet yr−1), annualized costs (US$ acre-foot−1) and implementation timeframe
(years) as estimated by each utility.
Goal-based water-trading policy framework. We designed a set of conditions,
rules and mechanisms to demonstrate the operation of a goal-based trading system
for WSD. These policy design considerations include elements generally applicable
to any market process, as well as elements specific to the water sector (Fig. 6).
Regional goal. Defining a regional diversification goal, though not a trivial task,
is an elementary step that acts as the catalyser for action. This goal may be either
intrinsically motivated or driven by regulations. For example, in this study, we
adopted a hypothetical goal where the overall regional policy dictates that at least
25% of total supplies must come from alternative sources such as recycled water
and conservation, by the year 2040. This goal is comparable in scope to efforts
already pursued in many water management regions. Specific targets could be
further refined to fit metrics that are reflective of system resilience in a given
community, such as an agreed-on supply portfolio composition (for example,
doubling the amount of recycled water use), a reduced dependence on imported
supplies, compliance with in-stream flow requirements, or a given level of water
conservation to address projected shortfalls. Equally as important is defining an
appropriate geographical and institutional scope of the region undertaking the
goal. As suggested in previous studies examining water management strategies in
the presence of many diverse decision-making agents, we recommend a scope that
overlaps with common supply pools38–40, where all stakeholders affected by a given
supply source are subject to the trading scheme. The common supply could be a
groundwater basin, river system or common dependence on imported supplies.
The ‘region’ could also exist on even smaller scales, such as a neighbourhood,
city or utility service area interconnected through common water sources. Such a
scope has the dual benefit of defining common grounds for problem definition and
collective action, as well as facilitating interactions between interconnected agents
and reducing the costs for monitoring and enforcement36.
Market rules. Similar to other environmental goal-based trading applications, we
propose a system where water utilities receive credits for implementing alternative
water projects. Each year, utilities have an opportunity to buy or sell diversification
credits in a centralized auction process. Because water infrastructure projects are
subject to planning, permitting and widely variable implementation timeframes,
in this study, we propose the availability of two coexisting markets: a forward
market happening every five years and a current-year market happening every
year. The purpose of the forward market is to allow utilities to bid and trade future
diversification credits for large projects before they are implemented, thereby
receiving encouragement to invest in long-term projects, whereas the current-year
market allows utilities to trade diversification credits from projects already in place,
thus balancing existing credits to meet that year’s target. To ensure transparency
and equal access to the market, all bids are cleared through a centralized uniform
price auction managed by a regional coordinating agency.
Compliance mechanisms. To ensure that each utility makes continual progress
towards its goal, we demonstrate a phased policy implementation of the
diversification goal over a 20-year period, starting with a milestone of 5% in
2020 and following a linear progression every year until the 25% goal is achieved
in 2040. Utilities that are unable to secure sufficient diversification credits in
a given year are subject to a fine for non-compliance. This penalty should be
consistent with the marginal cost of achieving the region-wide diversification goal.
Conversely, utilities that surpass their goal in a given year are allowed to bank
diversification credits that can be used in future trading periods.
Even though we selected these parameters to demonstrate a goal-based
water-trading scheme for water portfolio diversification, additional policy
design considerations could be tailored to fit local realities. Examples of design
considerations that are commonly addressed in other environmental applications
include defining prequalification criteria to ensure that the projects being proposed
are feasible, offering technical support to lower the entry barriers for utilities with
limited institutional capacity, and addressing potential interactions with existing
local rules41. Overall, goal-based trading schemes are flexible in that they allow
a variety of different target goals and deadlines, market trading mechanisms,
social incentives and water supply types to be used to comply with the policy. This
flexibility makes this policy tool especially relevant for the diverse needs of the
water sector in different locations and at different scales.
Agent-based model. We simulated the proposed goal-based water-trading policy
using an agent-based modelling (ABM) framework. ABM is a dominant paradigm
in the study of collective behaviour and adaptation42, and has gained prominence
in the study of water resources management43,44. ABM is particularly suited to
simulating how the heterogeneous behaviour of agents with different internal
motivations, decision rules and information affect the performance and macro
behaviour of the overall system45,46. In this study, we developed a stylized ABM
informed by theory and recent literature in decision-making, water management
and markets. Because our simulation study deals with hypothetical scenarios
subject to future uncertainties, the focus was not on specific predictions but rather
on exploring dynamics and demonstrating the potential range of outcomes of a
regional virtual water-trading market, thus providing insights to support policy
development42. For this purpose, we used a deterministic model that explores a
specific range of conditions, as described below.
Overview. The purpose of this model is to simulate the selection of water supply
portfolio investments over time under a goal-based trading policy for water
diversification. Each water utility is modelled as a decision-making agent with
NATURE SUSTAINABILITY | VOL 2 | FEBRUARY 2019 | 138–147 | www.nature.com/natsustain
144
AnAlysis
NATuRe SuSTAINABIlITy
a portfolio of available water supplies to be implemented, a target WSD goal, a
projected water demand to be met, and associated costs and preferences.
On a yearly timescale, utilities make long-, medium- and short-run decisions
about which new supply projects to implement, based on the projected availability
of water supplies, to meet their yearly diversification goals. Each decision period
determines: (1) how much of each local supply and alternative supply to implement
in a given year (subject to capacity constraints); (2) whether to participate in
trading based on whether they can meet or exceed their individual diversification
goal, and if so; (3) the offer price and quantity of diversification credits to be
bought or sold. The bids are then aggregated and cleared in a market simulation,
following a uniform price auction protocol. The use of multiple planning horizons
is comparable to portfolio selection problems that have been more widely modelled
in the context of renewable portfolio standards47. In the long run (looking ten
years ahead), utilities make capital investments that are not easily altered from year
to year. Projects with a long implementation timeframe are thus negotiated in a
forward market. Medium-term (five-year moving window) and short-term (one-
year moving window) decisions are then made to adjust for interannual variability
in supply availability, and balanced in a current-year market.
Design concepts. This model assumes that water utilities make decisions based on
simplified objectives as well as intrinsic behavioural characteristics, as described below.
Objectives and constraints. Utilities’ decisions in any given year are driven by
the incentive to minimize the total costs (C) of their local water portfolio while
meeting the needs of their customers and satisfying policy requirements. To
accomplish this, utilities consider the marginal costs of each supply source, existing
or alternative (c and calt, respectively), as well as the projected market price for
trading diversification credits (p), and select a portfolio of existing supplies
(q, out of the set of existing supply sources M), new supplies (qalt, out of the set of
alternative supply sources N) and bids for buying or selling credits (bid) (equation (1)).
These decisions are subject to constraints related to meeting the projected water
demand (D) of each service area (equation (2)), meeting the WSD goal set by
the policy (equation (3)), and capacity constraints (Q and Qalt) for each supply
source (equation (4)). There is no constraint on the number of credits traded, as
diversification credits are treated as a virtual exchange, rather than representing
a physical transfer of supplies. Instead, we assume utilities could leverage the
common pool supply to balance local supply and demand, such that a new supply
source in one location represents an equivalent decrease in withdrawals from the
common pool, which is then available for other utilities to use. The capacity of the
common pool is addressed in equation (4), where each utility has access to a given
withdrawal amount limited by contractual allocations and conveyance capacity:
∑∑
=×+×+×
==
CcqcqpMinimize bid(1
)
m
M
mm
n
N
nn
11
∑∑
+≥
==
qqD(2
)
m
M
m
n
N
n
11
alt,
∑+≥
=
qbidWSD (3
)
n
N
n
1
alt,
≤≤qQqQ,(4)
m
m
n
n
alt,
alt,
Portfolio selection criteria. While we model cost as the main driver of portfolio
selection, consistent with previous studies and practical decision-making in the
water sector48, we acknowledge that utilities may also have the internal motivation
to pursue certain projects due to their perceived multi-criteria benefits. Multi-
criteria preferences could come about from the value of local benefits such as
drought resilience, water quality and flood control, from additional motivations
such as reduced energy consumption and carbon emissions, or from external
funding availability. Preferences for specific technologies have also been observed
in electricity markets for renewable portfolio standards, and have been recognized
as having important impacts on policy49. In this model, we incorporate multi-
criteria preferences with a trade-off parameter (TF) that adjusts the perceived cost
of alternative projects, where the value of TF can vary between 0 (no trade-off;
utilities require subsidization for the full cost of the project) and 1 (complete
trade-off; utilities would implement the project even if it were unsubsidized)
(equation (5)). Values for TF are assigned intrinsically to each utility for each project.
=− ×cc(1 TF)(5)
perceived
Market clearance. Every year, once utilities select their water supply portfolios and
trading strategies, all bids are aggregated and cleared through a uniform price
auction. Bids for future projects with long implementation timelines are cleared
in the forward market through a reverse auction process, where bids for selling
diversification credits are aggregated in increasing order of price and the lowest-
cost sellers are successful until the demand for future credits is satisfied. The
market price in the forward market is determined by the last successful seller, and
all successful sellers receive the same price per credit. In contrast, bids for projects
already in place are cleared in the current-year market through a double auction
process, where bids for selling diversification credits are aggregated in increasing
order of price and bids for buying credits are aggregated in decreasing order
of price. The market price is then determined by the intersection or mid-point
between buyers and sellers, such that all seller bids below market price and all
buyer bids above market price are successful.
Learning. Utilities can adapt their bidding strategy and expectations based on
the market outcomes from year to year. To represent this learning process, we
followed a Zero-Intelligence Plus learning algorithm50,51, which updates market
price expectations based on the previous trading price, whether or not a utility was
successful in its trading strategy, and an assumed learning rate, as further described
in the Supplementary Information.
Details. Each simulation run is initialized by specifying a diversification goal.
The model then draws inputs from a data matrix defining each utility’s portfolio
of available water supplies, potential alternative supplies, associated costs and
trade-off parameters, as well as learning characteristics. An example of specific
model variables as used in this study is further described in the Supplementary
Information.
Simulation experiments. We used the general ABM framework above to represent
various scenarios of policy drivers, decision-making strategies and trading
dynamics in our case study of the San Francisco Bay Area.
Policy drivers. To baseline the performance of the proposed goal-based trading
policy for water portfolio diversification, we compared simulation results between
two scenarios: one where utilities follow the trading framework and one where
utilities are given the same diversification target of 25% but are not able to trade
with each other. The goal-based water-trading policy simulation follows the agent-
based model with all parameters as described above. In contrast, the no-trading
policy was modelled as individual optimization problems without bidding or
learning parameters, and thus no market clearing process. In both simulations,
utilities that do not comply with their diversification target in a given year are
subject to a fine of US$3,000 acre-foot−1. Both models follow a yearly simulation
timescale and all other parameters and assumptions are held constant.
Decision strategies. We represented portfolio selection criteria by varying
the trade-off parameter TF in the agent-based model. We simulated two
deterministic scenarios: one where utilities’ decisions are cost driven (TF = 0)
and one where utilities consider multi-criteria preferences (TF > 0). For the
multi-criteria scenario, TF was set as 0.6 for storm-water and recycled water
projects and 0 for all others, exemplifying a preference that would bring
perceived costs down by 60% for these projects. This deterministic simulation is
only meant to show the potential effect of multi-criteria drivers in the decision-
making process. Multi-criteria decision support tools have been extensively
studied elsewhere, and utilities may choose to quantify benefits in a variety
of different ways52. Although potentially significant, these simulations do not
incorporate transaction costs associated with setting up and running a market.
Assumptions about transaction costs are subject to significant uncertainty based
on the context of information, exchange and monitoring costs, and are beyond
the scope of this work.
Trading dynamics. We simulated two different trading dynamics: one where
utilities coordinate their efforts (coordinated trading) and one where they select
supply portfolios independently (independent trading). The coordinated dynamic
assumes that all utilities (U) share information and engage in collectively finding
the optimum regional portfolios that achieve the desired diversification target
at a minimum overall cost. Thus, the problem of choosing an optimum supply
portfolio for a given planning horizon t under coordinated trading dynamics
is represented by equation (6) (adapted from equation (1) in the general agent-
based model), where utilities’ decisions are simulated as a single objective
function with all utilities’ existing supplies and potential projects on the table, and
willingness to trade is defined each year based on the perceived marginal costs of
each utility independent of previous market outcomes. This global optimization
model is consistent with market representations under full information-sharing
assumptions53. In contrast, the independent trading dynamic assumes that utilities
find their optimum portfolios based on their own information and information
about past market outcomes only, without knowing what the other utilities are
planning for the upcoming period, and they adjust their expectations every year
through a learning process. The independent model was thus simulated as distinct
objective functions for each utility i, where utilities gauge the marginal cost of their
local projects in relation to their expected market price p and decide independently
NATURE SUSTAINABILITY | VOL 2 | FEBRUARY 2019 | 138–147 | www.nature.com/natsustain 145
AnAlysis NATuRe SuSTAINABIlITy
on their bid quantity (bid) (equation (7), adapted from equation (1) in the general
agent-based model):
∑∑ ∑
=×+−××
== =
CcqcqMinimize (1 TF )(6
)
i
U
m
M
mi mit
n
N
ni ni nit
11
,,,
1
,alt,, alt, ,,
∑∑
=×+−×× +× ∀
==
Ccqcqp i
Minimize
(1 TF
)b
id (7
)
m
M
mi mit
n
N
ni ni nititit
1
,,,
1
,alt,, alt, ,, ,,
Policy analysis. The criteria we used to compare simulation results are: cost-
effectiveness, level of engagement and diversification. First, to be cost-effective,
a set of policy conditions must achieve the same policy goal at a lower cost
(US$ acre-foot−1) than other possible policies. Second, engagement was measured
by the level of diversification achieved through trading or the volume of credits
(average acre-feet yr−1 traded) that result from interaction between utilities through
the market. This is intended to reflect the level of regional collaboration motivated
by the goal-based water-trading policy. Finally, diversification in this context was
used as a proxy for resilience benefits; thus, we measured both the proportion of
alternative supplies being added to the regional portfolio (%) and the
Gini–Simpson diversity index—a metric that accounts for both the breadth and
depth achieved by the expansion of water supply portfolios.
Data availability
The data that support the findings of this study are available within the paper and
its Supplementary Information files.
Received: 2 August 2018; Accepted: 15 January 2019;
Published online: 11 February 2019
References
1. Dienbaugh, N. S., Swain, D. L. & Touma, D. Anthropogenic warming has
increased drought risk in California. Proc. Natl Acad. Sci. USA 112,
3931–3936 (2015).
2. Quesnel, K. J. & Ajami, N. K. Changes in water consumption linked to heavy
news media coverage of extreme climatic events. Sci. Adv. 3, e1700784 (2017).
3. Gonzales, P. & Ajami, N. Social and structural patterns of drought-related
water conservation and rebound. Water Resour. Res. 53, 10619–10634 (2017).
4. Viviroli, D. et al. Climate change and mountain water resources: overview
and recommendations for research, management and policy. Hydrol. Earth
Syst. Sci. 15, 471–504 (2011).
5. Hering, J. G., Waite, T. D., Luthy, R. G., Drewes, J. E. & Sedlak, D. L. A
changing framework for urban water systems. Environ. Sci. Technol. 47,
10721–10726 (2013).
6. Beh, E. H. Y., Maier, H. R. & Dandy, G. C. Adaptive, multiobjective optimal
sequencing approach for urban water supply augmentation under deep
uncertainty. Water Resour. Res. 51, 1529–1551 (2015).
7. Porse, E. et al. Systems analysis and optimization of local water supplies in
Los Angeles. J. Water Resour. Plan. Manage. 143, 04017049 (2017).
8. Low, K. et al. Fighting drought with innovation: Melbourne’s response
to the Millennium Drought in Southeast Australia. WIREs Water 2,
315–328 (2015).
9. US Water Alliance One Water Roadmap: e Sustainable Management of Life’s
Most Essential Resource (2016); http://uswateralliance.org/sites/
uswateralliance.org/les/publications/Roadmap%20FINAL.pdf
10. Mukheibir, P., Howe, C. & Gallet, D. What’s getting in the way of a One
Water approach to water services planning and management? Wate r 41,
67–73 (2014).
11. Pahl-Wostl, C. et al. Managing change toward adaptive water management
through social learning. Ecol. Soc. 12, 30 (2007).
12. Aerts, J. C. J. H., Botzen, W., van der Veen, A., Krywkow, J. & Werners, S.
Dealing with uncertainty in ood management through diversication. Ecol.
Soc. 13, 41 (2008).
13. Kiparsky, M., Sedlak, D. L., ompson, B. H. & Truer, B. e innovation
decit in urban water: the need for an integrated perspective on institutions,
organizations, and technology. Environ. Eng. Sci. 30, 395–408 (2013).
14. Marston, L. & Cai, X. An overview of water reallocation and the barriers to
its implementation. WIREs Water 3, 658–677 (2016).
15. Mayr, D., Schmidt, J. & Schmid, E. e potentials of a reverse auction in
allocating subsidies for cost-eective roof-top photovoltaic system
deployment. Energy Policy 69, 555–565 (2014).
16. urston, H. W., Taylor, M. A., Shuster, W. D., Roy, A. H. & Morrison, M. A.
Using a reverse auction to promote household level stormwater control.
Environ. Sci. Policy 13, 405–414 (2010).
17. Taylor, M. R. Innovation under cap-and-trade programs. Proc. Natl Acad. Sci.
USA 109, 4804–4809 (2012).
18. Rahimova, N., Hesseln, H. & Silven, P. Factors aecting water market
acceptance: a case study of stakeholders in southern Alberta, Canada.
Can. Geogr. 60, 245–258 (2016).
19. Izadpanah, H., Youse, G. R. & Makhdoumi Kaviri, S. Asset and risk
management for a wind farm in a competitive electricity market. In 21st
Iranian Conference on Electrical Engineering (IEEE, 2013).
20. Rogge, K. S., Schneider, M. & Homann, V. H. e innovation impact of the
EU Emission Trading System—ndings of company case studies in the
German power sector. Ecol. Econ. 70, 513–523 (2011).
21. Stern, J. Introducing competition into England and Wales water industry—
lessons from UK and EU energy market liberalisation. Util. Policy 18,
120–128 (2010).
22. Espey, S. Renewables portfolio standard: a means for trade with electricity
from renewable energy sources? Energy Policy 29, 557–566 (2001).
23. Morthorst, P. E. e development of a green certicate market. Energy Policy
28, 1085–1094 (2000).
24. Gonzales, P. & Ajami, N. K. e changing water cycle: impacts of an evolving
supply and demand landscape on urban water reliability in the Bay Area.
WIREs Water 4, e1240 (2017).
25. Brozovic, N., Sunding, D. L. & Zilberman, D. Estimating business and
residential water supply interruption losses from catastrophic events. Water
Resour. Res. 43, W08423 (2007).
26. Kwon, T.-h Rent and rent-seeking in renewable energy support policies:
feed-in tari vs. renewable portfolio standard. Renew. Sustain. Energy Rev. 44,
676–681 (2015).
27. BenDor, T. K., Riggsbee, J. A. & Doyle, M. Risk and markets for ecosystem
services. Environ. Sci. Technol. 45, 10322–10330 (2011).
28. Morris, E. K. et al. Choosing and using diversity indices: insights for
ecological applications from the German Biodiversity Exploratories. Ecol.
Evol. 4, 3514–3524 (2014).
29. Van Beers, C. & Zand, F. R&D cooperation, partner diversity, and
innovation performance: an empirical analysis. J. Prod. Innovat. Manag. 31,
292–312 (2014).
30. Gonzales, P. & Ajami, N. K. An integrative regional resilience framework for
the changing urban water paradigm. Sustain. Cities Soc. 30, 128–138 (2017).
31. Palazzo, J. et al. Urban responses to restrictive conservation policy during
drought. Water Resour. Res. 53, 4459–4475 (2017).
32. Mika, K. et al. LA Sustainable Water Project: Los Angeles City-Wide Overview
(UCLA Institute of the Environment and Sustainability, UCLA Sustainable
LA Grand Challenges & Colorado School of Mines, 2018).
33. Porse, E. et al. e economic value of local water supplies in Los Angeles.
Nat. Sustain. 1, 289–297 (2018).
34. National Research Council Adaptive Management for Water Resources Project
Planning (e National Academies Press, Washington DC, 2004).
35. Aerts, J. C. J. H. et al. Integrating human behaviour dynamics into ood
disaster risk assessment. Nat. Clim. Change 8, 193–199 (2018).
36. Ostrom, E. Polycentric systems for coping with collective action and global
environmental change. Glob. Environ. Change 20, 550–557 (2010).
37. Hughes, S. & Pincetl, S. Evaluating collaborative institutions in context: the
case of regional water management in southern California. Environ. Plan. C
32, 20–38 (2014).
38. Gonzales, P., Ajami, N. & Sun, Y. Coordinating water conservation eorts
through tradable credits: a proof of concept for drought response in the San
Francisco Bay area. Water Resour. Res. 53, 7662–7677 (2017).
39. Madani, K. & Dinar, A. Cooperative institutions for sustainable common pool
resource management: application to groundwater. Water Resour. Res. 48,
W09553 (2012).
40. Doyle, M. W., Patterson, L. A., Chen, Y., Schnier, K. E. & Yates, A. J.
Optimizing the scale of markets for water quality trading. Water Resour. Res.
50, 7231–7244 (2014).
41. Mora, D. et al. Auctions for Renewable Energy Support—Taming the Beast of
Competitive Bidding (AURES, 2017); http://auresproject.eu/publications/
auctions-renewable-energy-support-taming-the-beast-of-competitive-bidding
42. Nay, J. J., Abkowitz, M., Chu, E., Gallagher, D. & Wright, H. A review of
decision-support models for adaptation to climate change in the context of
development. Clim. Dev. 6, 357–367 (2014).
43. Haer, T., Botzen, W. J. W., Moel, H. & Aerts, J. C. J. H. Integrating household
risk mitigation behavior in ood risk analysis: an agent-based model
approach. Risk. Anal. 37, 1977–1992 (2016).
44. Ali, A. M., Shaee, M. E. & Berglund, E. Z. Agent-based modeling to
simulate the dynamics of urban water supply: climate, population growth,
and water shortages. Sustain. Cities Soc. 28, 420–434 (2017).
45. Bonabeau, E. Agent-based modeling: methods and techniques for simulating
human systems. Proc. Natl Acad. Sci. USA 99, 7280–7287 (2002).
46. Lempert, R. Agent-based modeling as organizational and public policy
simulators. Proc. Natl Acad. Sci. USA 99, 7195–7196 (2002).
47. Ritzenhofen, I., Birge, J. R. & Spinler, S. e structural impact of renewable
portfolio standards and feed-in taris on electricity markets. Eur. J. Oper. Res.
255, 224–242 (2016).
NATURE SUSTAINABILITY | VOL 2 | FEBRUARY 2019 | 138–147 | www.nature.com/natsustain
146
AnAlysis
NATuRe SuSTAINABIlITy
48. Wu, W. Y. et al. Including stakeholder input in formulating and solving
real-world optimisation problems: generic framework and case study. Environ.
Model. Sow. 79, 197–213 (2016).
49. Novacheck, J. & Johnson, J. X. e environmental and cost implications of
solar energy preferences in Renewable Portfolio Standards. Energy Policy 86,
250–261 (2015).
50. Cli, D. & Bruten, J. Less than human: simple adaptive trading agents for
CDA markets. IFAC Proc. Vol. 31, 117–122 (1998).
51. Du, E., Cai, X., Brozović, N. & Minsker, B. Evaluating the impacts of farmers’
behaviors on a hypothetical agricultural water market based on double
auction. Water Resour. Res. 53, 4053–4072 (2017).
52. Hajkowicz, S. & Collins, K. A review of multiple criteria analysis for water
resource planning and management. Water Resour. Manag. 21, 1553–1566 (2007).
53. Giuliani, M. & Castelletti, A. Assessing the value of cooperation and
information exchange in large water resources systems by agent-based
optimization. Water Resour. Res. 49, 3912–3926 (2013).
Acknowledgements
We appreciate the partnership with Sonoma County Water Agency on this project,
especially C. Pollard and J. Jasperse for their continuous support and constructive
feedback. We also thank K. Quesnel, P. Bolorinos and P. Womble for providing
useful suggestions and technical guidance that improved this paper. This work was
financially supported by Sonoma County Water Agency and the National Science
Foundation Engineering Research Center for Re-Inventing the Nation’s Urban
Water Infrastructure (award number EEC-1028968). Any opinions, findings and
conclusions or recommendations expressed in this material are those of the authors
and do not necessarily reflect the views of Sonoma County Water Agency or the
National Science Foundation. This work was completed at Stanford University as
part of P.G.’s PhD dissertation.
Author contributions
P.G. and N.K.A. designed the study. P.G. performed the modelling work. P.G. and N.K.A.
analysed the results and wrote the manuscript.
Competing interests
The authors declare no competing interests.
Additional information
Supplementary information is available for this paper at https://doi.org/10.1038/
s41893-019-0228-z.
Reprints and permissions information is available at www.nature.com/reprints.
Correspondence and requests for materials should be addressed to N.K.A.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.
© The Author(s), under exclusive licence to Springer Nature Limited 2019
NATURE SUSTAINABILITY | VOL 2 | FEBRUARY 2019 | 138–147 | www.nature.com/natsustain 147
A preview of this full-text is provided by Springer Nature.
Content available from Nature Sustainability
This content is subject to copyright. Terms and conditions apply.
