Market Mechanisms for Financing Green Real Estate
Investments
Dwight Ja?ee and Nancy Wallace
Draft: November 30, 2009
1 Introduction
U.S. buildings consume almost 39% of the total U.S. total energy consumption, making real
estate the largest consuming sector by a considerable margin; see Table 1.1. In comparison,
the industrial consumption share is about 33% of the total and the transportation share is
about 28% of the total. Within the real estate sector, the share of total energy used is almost
equally split between residential (20.9%) and commercial (18.0%) buildings.
Table 1.1: Buildings Share of U.S. Primary Energy Consumption (Percent), 2006
Residential
Buildings
Commercial
Buildings
Total
Buildings
Industry Transportation Total Total Con-
sumption
(quads)
20.9% 18.0% 38.9% 32.7% 28.4% 100% 99.5
Source: U.S. Department of Energy (2009)
The energy consumption of U.S. real estate appears, furthermore, to be substantially
less e?cient than comparable European buildings, even after controlling for such factors
such as climate, GDP, and population.
1
There is no mystery as to the reason: U.S. costs
for electricity and heating oil range from 50 to 75 percent of the levels in most European
countries; see IEA (2009). This indicates there must be feasible technologies that would allow
the energy consumption of U.S. buildings to be reduced signi?cantly. Such investments could
arise as the result of building codes and comparable requirements, or as a voluntary response
to high and volatile energy prices.
Most economists would agree that the “?rst-best” solution to reducing U.S. energy con-
sumption, across the three major uses of real estate, industry, and transportation, is to raise
1
IEA (2008) and (2004) provide comparisons of residential energy use in the U.S. and Europe corrected for
climate and measured per unit of GDP or per capital. McKinsey (2007) also shows substantially higher U.S.
energy consumption compared to Europe after controlling for GDP and population. Rand (2009) provides
a discussion comparing energy use in the U.S., Australia and the European Union.
1
the U.S. price of energy through appropriate ?scal instruments. While this may occur in
the future, the timing of this change in policy remains highly uncertain. So it is sensible,
even critical, to look for alternative, accelerated, mechanisms to reduce U.S. energy con-
sumption. Indeed, if and when U.S. energy prices do rise, the economic adjustment will be
easier and faster if the transformation toward a more energy e?cient technology is already
well along. Facing inadequate energy-e?cient investments in real estate, governments in
both the U.S. and Europe have intervened to provide additional incentives. To date, three
approaches have dominated theses interventions: expanded building codes, energy-e?ciency
certi?cations, and direct ?scal subsidies.
Expanded building codes have been the primary mechanism to ensure energy-e?cient
structures in Europe and are likely to expand over time in the U.S. as well. Building code
requirements, however, have their primary impact on new construction, and given the long
durability of most buildings, new construction annually represents a very small percentage
of the existing building inventory. Most building codes, furthermore, are prescriptive rather
than performance-based, which limits the incentive they provide for new innovative solutions.
Disclosure certi?cates at the time of building sales have also become an important mecha-
nism in Europe. The 2002 EU Energy Performance in Buildings Directive requires an energy
performance certi?cate based on either the building’s design or usage characteristics. While
such disclosures might well have an impact on the sales prices, it is unclear to what extent
such disclosures motivate either the seller or buyer to initiate energy e?cient investments. In
the U.S., certi?cations such as Leeds and Energy Star are available, and such certi?cations
are sought by builders and developers who are planning to create energy e?cient buildings.
But these certi?cations primarily apply to new buildings, and it is unclear to what extent
the certi?cations are the factor actually motivating energy-e?ciency in these buildings.
2
Direct subsidies provided by either government agencies and public utilities may provide
economic stimulus to energy e?cient investments. In these times of harsh ?scal budgets,
however, direct subsidies are unlikely to be a major driver of new energy-e?cient investments.
Given the limitations of building code requirements, energy-e?ciency certi?cations, and
direct subsidies, it is critical to consider whether there are other solutions to the private mar-
ket failures that are contributing to the underinvestment in U.S. building energy e?ciency.
It is possible, of course, that the available technology is just not pro?table in net present
value (NPV) terms. However, it appears that NPV positive energy-e?cient investment op-
portunities already exist, and that further innovations and cost-savings from scale economies
are literally in process. In this paper, we focus on loan market frictions as a market fail-
ure that raises the cost and limits the availability of mortgage funding for energy-e?cient
investments. Given the large ratio of the existing stock to new investment, our proposed so-
lutions emphasize market mechanisms to eliminate the frictions that inhibit energy retro?ts
in existing structures. These same solutions will also expand the incentives to embed the
best technology in new construction. In either case, we can identify three steps that must
2
Eichholtz, Kok, and Quigley (2009) show that buildings with certi?cations obtain higher sales prices,
which provides a motivation for obtaining such certi?cations. However, it is unclear whether the existence
of the certi?cations motivates energy e?cient investments that otherwise would not have been carried out.
The existence of higher sales prices on certi?ed buildings also does not indicate whether the initial energy-
e?ciency investments were NPV positive; to show that, the investments costs must be compared with the
sales prices.
2
occur if building owners are to invest voluntarily in energy saving equipment:
1. Identi?cation of worthwhile investments, Building owners must be able to identify those
energy-e?cient investments that meet at least a minimal standard of pro?tability. As a
simple example, for any investment to add value, it must at least provide a non-negative
NPV when evaluated with a zero interest rate. In other words, the investment must
pass an internal rate of return (IRR) hurdle of zero, which is to say that the sum of
the actual cash ?ow bene?ts must exceed the investment costs.
2. Computation of the true NPV. Investment decisions, of course, will be based on a
NPV > 0 criterion, where the NPV is evaluated at the owner’s true opportunity cost
of capital. For many real estate investments, this opportunity cost will equal the
interest rate that is applied in ?nancing the investment. The NPV criterion also allows
the property owner to prioritize alternative investments, giving the highest priority to
the highest NPV projects. This is particularly relevant if there is a cap on the total
amount of available funding.
3. Identi?cation and removal of ?nancial frictions. Frictions in the lending market raise
the discount rate and lower the NPV computations, thus limiting the set of feasible
investments and acting as a deterrent to those investments with the longest payo?
structures. Or to put it in a positive vein, the removal of such frictions can stimulate
the aggregate amount of energy e?cient investments, and provide particular bene?ts
to those investments with the longest-term payo?s. The identi?cation of these frictions
and proposals to remove them is the direct focus of this paper.
The agenda of this paper is as follows. In Section 2, we describe the loan market fric-
tions that currently inhibit energy-e?cient investments in the commercial real estate sector.
In Section 3, we describe our proposals to help eliminate these frictions. Our proposals in
Section 3 require a mechanism through which ?rms can compute the NPV of their proposed
energy investments and convince their lenders of the bene?ts of funding these investments.
Thus, in Section 4, we provide a feasible method to compute the net present value of alterna-
tive energy-e?cient investments. The key innovation here is to use Monte Carlo simulations
to incorporate stochastic energy prices into the valuation process. In Section 5, we provide
a summary and conclusions.
2 Commercial Mortgage Market Frictions that Inhibit
Energy-E?cienty Investments
Commercial property mortgage lending in the United States traditionally focuses on two key
risk measures for underwriting mortgages: the loan-to-value ratio (LTVR, the ratio of the
mortgage balance to the value of the building), and the debt-service-coverage ratio (DSCR,
the ratio of the propertys net operating income to the principal and interest payments on the
mortgage debt). These ratios are also monitored by bank regulators, including the Federal
Reserve, the Comptroller of the Currency, and the O?ce of Thrift Supervision, because they
3
are important indicators of the quality of commercial bank underwriting and the mortgage-
related default-risk exposure. Sound underwriting standards typically require a LTVR no
more than 65% and a DSCR no less than 1.25, although, of course, other factors such as
the quality and timing of tenant leases will have an impact on these ratios. These ratios
are important because mortgage performance data have shown that mortgage borrowers are
more likely to default as the LTVR rises toward 1.0 or as the DSCR falls toward 1.0.
The key role of the DSCR and the LTVR for underwriting commercial mortgages has
become a critical impediment for the recognition of volatile energy costs as an additional and
major source of mortgage default risk for commercial buildings. Net operating income (NOI,
the net rental income associated with a commercial structure) is the key input in computing
both the DSCR and the LTVR. NOI appears directly in the numerator of the DSCR. The
present value of NOI is also the fundamental input in determining the property valuation
that appears in the denominator of the LTRV. In practice, the forecasted net operating
income of a commercial building is constructed by:
1. Aggregate the forecasted contractual rental income from the tenants’ leases;
2. Subtract the forecasted building operating expenses, including the costs of energy; and
3. Adding back the forecasted energy-use reimbursements from the tenant to the landlord,
as required on triple net leases that are the most common commercial lease contracting
structure.
The triple-net lease, and the software package, ARGUS, which is widely used to organize
and display the information, thus “nets” out the energy risk exposure of buildings, other
than those energy costs borne solely by the landlord due to vacancies, joint costs associated
with common areas such as lobbies, and incomplete energy reimbursements from the tenants
to the building owner. The result is that mortgage lenders typically do not account for the
risks created by a commercial buildings energy costs. In particular, commercial mortgage
underwriters currently have no validated system to score the default risk created by energy-
cost volatility in a manner similar to their use of the DSCR and the LTVR to score the
overall default risk.
Of course, with triple-net leases, tenants in more energy-e?cient buildings should bene?t
from lower energy bills, and thus should be willing to pay higher rents for space in such
buildings. Three conditions have to be realized, however, if the result is to be more energy-
e?cient investments:
1. Tenants must validate the lower expected energy costs before they accept the higher
rents;
2. Landlords must con?rm the higher rents before undertaking the energy-e?cient invest-
ments;
3. The actual investments must be NPV positive in order to ensure that a true net bene?t
exists.
Imperfect information regarding future energy costs and the e?cacy of energy-e?cient
investments makes each of these steps problematic. The result is that the existing systems
4
for commercial mortgage lending and property leasing inhibit energy-e?cient investments in
this sector.
The positive news is that our proposed tools for evaluating and disclosing energy e?cient
investments, to which we next turn, can play an important role in mitigating the frictions that
currently constrain both owner-lender mortgage contracts and owner-tenant lease contracts.
3 Mitigating Financing Frictions for Commercial Real
Estate
The unsatisfactory situation for the ?nancing of energy-e?cient investments in commercial
real estate can be potentially mitigated by energy e?ciency metrics and disclosures for
mortgage underwriting and tenant leasing. These metrics allow the comparison of energy-
related costs and risks across building types, building qualities, and tenanting structures.
Just as LTVR and DSCR are informative about the potential default risk of newly originated
mortgages, so too our proposed Energy E?ciency Score (EES) and Energy Volatility Score
(EVS) provide default risk metrics based on the impact of energy-cost savings on a building’s
net operating income. The bene?ts include both a higher net operating income as a result
of lower energy costs and a more stable net operating income in the face of volatile energy
costs. The goal is that the two new energy e?ciency metrics will be used by commercial loan
underwriters to provide more favorable loan terms to loan applicants with energy-e?cient
buildings.. The two scores will also provide property owners with a means to determine the
most cost-e?ective energy-e?cient investments, and tenants with a means to estimate their
savings in energy costs as a result of renting space in an energy-e?cient building.
Figure 3.1: Asset Valuation
5
3.1 Energy-E?ciency Metrics to Facilitate Commercial Mortgage
Lending
Figure 3.1 provides an overview of how our proposed energy e?ciency (EE) mortgage un-
derwriting rating system would work. It uses both engineering-based simulation methods
focusing on an integrated analysis of whole building systems to benchmark the relative en-
ergy consumption of lighting, HVAC, and plug-load functions in commercial buildings of
di?ering types and qualities; and performance-based empirical benchmarking using statisti-
cal estimation methods. The output from our coordinated energy e?ciency benchmarking
analyses will be technical metrics that are summarized in the proposed scoring metrics for
mortgage underwriting: the EES and EVS. These two scores are related to the level and
volatility of the net operating income of commercial buildings. The scores can also be incor-
porated as inputs into mortgage valuation models that are based on the dynamics of interest
rates and asset prices.
As shown on the right-hand side of the ?ow diagram, the mortgage valuation technology
can be used prescriptively to determine the appropriate interest rate to be charged, and/or
the maximum loan size to be granted for buildings with di?ering EES and EVS. This pre-
scriptive exercise takes the energy e?ciency scores as given and solves for the actuarially fair
mortgage contract terms (e.g. by setting lower mortgage rates, higher LTVRs, and lower
DSCRs as compensation for a higher energy e?ciency score or a lower energy volatility
score). The scores may also be used to induce renters to pay higher rents in exchange for
the bene?t of lower energy costs.
By measuring the EES and EVS that result from alternative energy-e?cient investments,
we can also derive a third measure, the Energy E?ciency Action Ratio (EEAR). As shown in
the ?ow diagram, this third ratio allows a decision maker to rank order the cost e?ectiveness
of alternative investments to change the energy e?ciency of buildings. This analysis is
a particularly important if mortgage lenders are going to recognize that energy-e?cient
investments in commercial buildings can directly reduce their mortgage default risk by raising
the property’s NOI and by reducing the sensitivity of the NOI to future energy cost volatility.
In Section 4 below, we take the ?rst steps to implement this program by carrying out
dynamic Monte Carlo valuation of the possible future energy costs of a representative building
based on alternative stochastic paths for energy costs. Future work will then expand this tool
to compute the energy e?ciency and energy volatility scores and the energy e?ciency action
ratio for buildings with any actual or simulated physical and environmental characteristics.
3.2 Alternative Methods to Facilitate Commercial Mortgage Lend-
ing
Figure 3.2 introduces two alternative green ?nancing instruments–Managed Energy Con-
tracts and Property Tax Assessed Energy Retro?t Financing and compares them with a
straight commercial mortgages. Managed Energy Contracts (MEC) are privately provided
leasing contracts in which an energy service provider commits to provide all energy related
services to a property owner, typically a Real Estate Investment Trust or large real estate
portfolio operator, in exchange for the payment of the property operators historical energy
costs. The energy service provider pro?ts if it achieves energy cost savings in excess of the
6
Figure 3.2: Location in the “Cash Flow Stack” of Commercial Real Estate Mortgages
(CREM), Managed Energy Contracts (MEC), and Property Tax Assessed Energy Retro?t
Financing (PTA-ERF) for Commercial Real Estate Assets
present value of any investments it makes. At the end of the lease term, typically ten years,
the energy e?ciency improvements remain with the building, thus enhancing the value of
the building owner’s asset. Since the MEC lease payments are equivalent to the historical
utility fee structure of the property, the full utility fee structure remains fully recoverable
from tenant reimbursements. While energy e?ciency gains are achieved through MEC, the
full cost savings of these investments are captured by the tenants or the building owner only
at the lease maturity date.
Property Tax Assessed Energy Retro?t Financing (PTEF) is a very recent innovation in
green energy ?nancing pioneered by the city of Palm Desert, California through the Palm
Desert Energy Partnership, and subsequently was codi?ed into California State Law in July,
2008 with the passage of Assembly Bill 811. The goal of AB811 is to allow California cities
to ?nance permanently ?xed renewable energy generation improvements in private property
including commercial real estate assets. Although the ?nancing strategies may di?er by
city, the bill allows for the creation of special tax assessment districts in which individual
property owners can opt for energy retro?t project loans with long maturities (e.g. 20 years
for the CityFIRST program in Berkeley, California). These loans have a contractual payment
structure that is similar to loans, except that they are paid back using a fully amortizing
installment payment that is added to the property taxes of the asset. In addition, the PTEF
loan liability is a tax lien with priority over mortgage liens and it is attached to the property
in contrast to a mortgage loan which is the liability of the borrower and is collateralized by
7
the property.
As shown in ?gure 3.2, the three types of commercial real estate green ?nancing are
attached at di?erent positions in the cash ?ow stack of commercial real estate assets. All
three options, usually involve a commercial real estate mortgage (CREM) and therefore a
substantial fraction of the net operating income, the primary mortgage underwriting cash
?ow, is allocated to the debt service on the loan subject to the limitations of the debt service
coverage ratio. The relative energy risk exposure of the CREM would, however, di?er across
the three structures represented in ?gure 3.2. The straight commercial mortgage presents the
lender with the maximal energy risk exposure. In contrast, the PTEF increases the property
tax liability of the property and thus reduces both the size and risk of the commercial
mortgage loan through the cost savings achieved by an energy-e?cient retro?t investment.
Even further, the use of a MEC largely eliminates the energy risk of the commercial real
estate loan because the property’s energy costs are ?xed at the historical expense level.
Although MEC and PTEF have di?erent payment priorities, the commitment of the
building owner is similar across the two contracts (technically MEC are lease ?nancings).
Both of these commitments represent forms of ?xed income securities in which an initial
investment by the lender, or lessor, is paid back with a set of installment payments by
the borrower, lessee. For this reason, they both can be underwritten for risk and priced
using the same form of ?nancial modeling techniques that we develop in Section 4 below.
Alternatively, the market prices of these two contract types can be directly related to our
two energy e?ciency scores, and the preferred investments will be identi?ed our the energy
e?ciency corrective action ratios.
Table 3.1 provides a further comparison across these three green ?nancing instruments.
For one thing, currently there are no standardized energy e?ciency metrics or valuation
tools to underwrite or price any of these ?nancing alternatives. Without an organized capital
market for MEC and PTEF instruments, valuation occurs on a case by case basis. Both
CREM and PTEF involve loan contracts, or promissory notes, whereas the MEC involves a
lease contract which is simply another form of ?xed income security. The seniority of these
instruments, as previously discussed, also di?ers with PTEF instruments having the most
secure interest.
The U.S. commercial real estate mortgage market, based on its size and long history, does
have well developed and standardized methods for valuing the risks that arise from macro-
fundamentals such as interest rates and asset price dynamics. In contrast, the valuation
standards applied in MEC and PTEF markets are currently entirely private and proprietary.
The CREM market also currently services all sizes and types of commercial building, whereas
MEC usually require larger scale to achieve energy e?ciency gains and thus they are used
primarily for Real Estate Investment Trusts or large real estate operating companies.
Two further limitations arise with regard to the MEC and PTEF mechanisms. With
regard to the MECs, there is a potential principal-agent problem in that the energy-saving
innovations introduced by the service provider could have a negative impact on the build-
ing’s serviceability. In fact, this may be an important factor that has limited the use of
these contracts. With regard to PTEFs, most commercial mortgages have a covenant that
constrains the property owner from creating new liens senior to the mortgage lien. The
mortgage lender’s permission is thus required to use the PTEF mechanism. In principal, the
mortgage lender would bene?t and thereby grant permission– if the energy-e?cient invest-
8
ment were NPV positive, since the increase in property value (i.e., the loan collateral) would
then exceed the value of the new, senior, property lien. The method we develop in Section
4 below would provide the lender with the necessary information to make the decision.
Table 3.1: Alternate Financing Methods Comparison
Commercial Real Estate
Mortgage (CREM)
Managed Energy Con-
tracts (MEC)
Property-Tax Assessed
Energy Retro?t Financing
(PTA-ERF)
Standardized Energy E?-
ciency Metrics
No No No
Energy E?ciency Valua-
tion Tools
None Net present value of
agent’s energy e?ciency
investments and pass-
through of existing utility
payments
Net present value of en-
ergy retro?t investment
and amortized loan pay-
ments
Security of Interest Mortgage lien on Property Commitment on utility
fees
Tax Lien on Property
Contract Type Promissory Note Utility provision lease. Promissory Note
Valuation Standards Actuarial and regulated Private Private
Installment Payment
Structure
Long amortization (15 to
20 years) due in 7 or 10
years
10 year renewable con-
tracts
Add-on to Property Taxes
with 20 year payback
Ownership of Energy Re-
lated Investment
Building Owner Building Owner Building Owner
Capital Market Funding
Sources
Commercial Banks, Com-
mercial Mortgage Backed
Securities, Insurance Cos.,
Pension Funds
Private Companies (e.g.
Transcend Equity De-
velopment Corp.) - if
scale su?cient potential
for lease securitization
capital markets
Private Companies (e.g.
Renewable Funding Fi-
nancial Partners) if scale
su?cient Municipal Bond
Market
Commercial Real Estate
Applications
All commercial property
types and building sizes
Real Estate Investment
Trusts, Large Real Estate
Operating Companies
Residential only, autho-
rization exists for com-
mercial real estate
Market Penetration $2.37 Trillion Capital
Market
Small Private Operating
Companies (e.g. Tran-
scend - MESA)
Small Private Operating
Companies (e.g. Re-
newable Funding Finan-
cial Partners)
Performance Risk Default and Prepayment
Risk of Borrowers
Agent’s failure to render
energy services, property
owner’s failure to pass-
through utility fees
Property owners non-
payment of property
taxes
Energy Retro?t Expenses
Recoverable (Escalatable)
from Tenants
Yes, if operating cost,
repairs/maintenance, or
cost reducing. No, if
capital improvements.
Lease payment is an esca-
latable operating expense.
Accounting opinions exist.
Tax payment is escalat-
able, no accounting rul-
ings.
Finally, CREM, MEC, and PTEF share comparable default risks due to the failure to
make payments per the contractual agreements, by the borrower, lessee, or taxpayer respec-
tively. Finally, the ability of the building owner to pass energy e?ciency related expenses
9
through to the tenants depends on the language of the tenant leases and whether the invest-
ments are related to ?xed capital improvements or to maintenance and repairs. MEC and
PTEF may blur this distinction, however, since there are no de?nitive FASB rulings. What
is clear from the summary presented in Table 3.1 is that all three green ?nancing strategies
could bene?t from the development of suitable energy e?ciency metrics, the topic to which
we now turn.
4 Monte Carlo Simulation to Compute NPV for En-
ergy Investments
As previously introduced, summary statistics, or scores, for the energy e?ciency and energy
volatility of commercial real estate assets are critical if we are to achieve private market
solutions to the mortgage market frictions that currently inhibit energy-e?cient investments
in commercial buildings. The key is to quantify the impact of possible future energy costs
on a building’s NOI. NOI is, in turn, a critical component in computing the property’s
DSCR and LTVR, the ?nancial ratios used to determine the mortgage contract terms and
amount. Thus NOI is the critical statistic through which energy cost variability is linked to
the mortgage lending decision.
We next provide a prototype model that allows us to compute the impact of energy
price volatility on the NOI (net of energy costs), for a representative building. We use
a Monte Carlo simulation methodology in which alternative future paths of energy prices
are generated as the output of a stochastic process. We then compute the impact of the
alternative energy paths on the property’s NOI.
4.1 Stochastic Modeling of Energy Spot Dynamics
The classical stochastic process for the spot dynamics of commodity prices was developed
by Schwartz (1997) who speci?ed the process as the exponential of an Orenstein Uhlenbeck
(OU) process. Mean reversion has been identi?ed as an important empirical regularity of the
price dynamics of natural gas and electricity. There is also a growing literature considering
improvements on the Schwartz model such as incorporating jumps to handle the empirical
spikes in many commodity prices, such as electricity, modeling the seasonality of commodity
prices over heavy and lighter consumption months, and ?tting the correlation structure of
commodity prices and exogenous factors such as temperature dynamics.
3
For our modeling of energy price dynamics, we will also use an Schwartz-type OU process.
We implement our simulation model as ?exibly as possible to allow for inclusion of important
empirical characteristics of electricity such as mean-reversion, seasonalities in price levels
3
Recent work in modeling energy dynamic processes include: Benth, F. E., Cartea, A., and Kiesel, R.
(2006). Benth, F. E., Koekebakker, S., and Ollmar, F. (2007), Bessembinder, H., and Lemmon, M. (2002),
Cartea, A. and Figuerosa, M. G. (2005), Deng, S. (2000), Eberlein, E. and Stahl, G. (2003), Geman, H.
and Roncoroni, A. (2006), Karlis, D. and Lillestl, J. (2004), Weron, R. (2000), Weron, R. (2004), Weron, R.
(2005), and Weron, R., Kozlowska, B., and Nowicka-Zagrajeck, J. (2001). For excellent surveys of modern
energy modeling techniques see Weron, R. (2006) and Benth, F.E. and Benth J.S., and Koekebakker, S.
(2008).
10
and volatility, heteroscedasticity, co-variation with fuel prices, and high frequency (daily
or intraday) jumps. Natural gas spot prices also have mean reversion, seasonalities and
heteroscedasticity. Thus, the price dynamics for the natural log of electricity and natural
gas prices, dp
et
and dp
gt
,
4
can be represented by generic equations of the form:
5
dp
et
= ?
e
[?
e
(t) ?p
et
]dt + v
e
(p
et
, t, . . .)dX
e
t
dp
gt
= ?
g
[?
g
(t) ?p
gt
]dt + v
g
(p
gt
, t, . . .)dX
g
t
where the local volatility functions, v
e
(p
et
, t, . . .) and v
g
(p
gt
, t, . . .), can be constant, deter-
ministically dependent on time t, or deterministically dependent on any other relevant state
variables. The Brownian motion increments, dX
e
t
and dX
g
t
, have historically been highly
correlated for electric power and gas.
Most commodity option pricing models assume that the risk neutral dynamics have the
same qualitative characteristics as the objective dynamics. The parameters of the risk neutral
dynamics are then determined using a mixture of statistical calibration (using time series
data from the spot market for natural gas and electricity, such as the Henry Hub natural gas
spot prices shown in Figure 4.1) and implied calibration of the volatility parameters from the
natural gas and electricity futures and options markets in the monthly block market. The
table in Appendix I gives the data used for the electric power and gas derivatives markets
and for the term structure of U.S. Treasury securities in April 2008.
In this paper, we model the actual resource costs of natural gas and electricity, not the
utility power company regulated pass-through of these resource costs. We focus on two
distinct markets in which energy is traded. These markets di?er in the “time granularity”
of the underlying variables and types of optionality traded and in their market participants.
1. The ?rst market is the relatively liquid over-the-counter (OTC) market for futures and
options on monthly blocks (i.e., portfolios) of daily ?ows of electric power and gas (e.g.,
1 MW of electricity each hour for each day over the delivery month or 1 mmBTU of gas
each day over the delivery month). Prices are quoted on a 1 megawatt hour (MWH)
or 1 mmBTU basis. Some of this trading occurs on exchanges (e.g., NYMX) and some
is OTC. We will call this the monthly block market. Since a capability for physical
delivery is not necessary in this market, parties without the capacity to handle energy
physically can simply close out their positions via o?setting trades to avoid delivery.
The future and options markets have the broadest investor participation.
2. The second market is also an OTC market in which same-day blocks of spot electric
power (or spot gas in daily blocks) are bought and sold in real-time. In this spot
market, trades occur through a decentralized search/negotiation process. Nevertheless,
a daily spot price index is publicly reported based on surveys of transactions by electric
power broker/dealers. Agents that trade in this market must be able to deliver/receive
electrons or gas physically in real time.
6
4
Where p
et
= lnP
et
and p
gt
= lnP
gt
.
5
For simplicity, in this version of the paper we will ignore Poisson price jumps in both the gas and
electricity modeling. In future versions of our simulations we intend to model these Poisson processes as a
function of temperature dynamics.
6
For simplicity in this version of our simulations we are making some signi?cant simpli?cations about
11
Figure 4.1: Henry Hub Spot Price
Source: Bloomberg, Henry Hub spot prices
We calibrate our OU processes to the futures and options prices reported in Appendix
I for April 2008 and to electricity and Henry Hub spot price dynamics.
7
We use a simple
deterministic volatility speci?cations v
e
(t) and v
g
(t) and ?t an empirical value for ?
e
of 5.2
for electricity and an ?
p
of 4.4 for natural gas and an electricity/gas correlation of 0.9 for
their respective Brownian motions. We ?t the ?
e
(t), ?
g
(t), v
e
(t), and v
g
(t) to the futures
and options data using least squares. We are thus ?tting the OU dynamics to the observed
term structure of forward prices for electricity and natural gas as of April 1, 2008. We then
simulate ?ve years of monthly spot price dynamics using 1000 Monte Carlo simulations.
Representative price paths using a ?xed correlation structure of 0.9 for our ?tted OU energy
dynamic processes are reported in Figures 4.2 and 4.3.
energy markets. First, the spot market actually trades hourly rather than daily blocks of power in real time.
Thus, the spot price energy is literally the price of energy over the shortest operational decision intervals:
one hour for power and one day for gas. Second, there is one more important market which is ignored in our
current simpli?ed simulation strategy. This is an OTC market for one-day ahead forward contracts on one
day’s worth of power (or gas). For example, on Tuesday an investor can lock in the price for a day-long ?ow
of a 1 MW of electricity (or delivery of 1 mmBTU) on Wednesday. Because physical delivery is required,
this market is also limited to parties with contractual or operational control over actual power or gas. There
is no term structure information for daily deliveries since there are no daily forwards for maturities longer
than one-day ahead.
7
For this version of the paper our electricity spot data are actually from the Electricity Reliability Council
of the Mid-Atlantic (ERCMA) because our CALPX data only goes through 2001. We hope to obtain CALPX
data to re?t the OU for electricity to California derivatives markets.
12
Figure 4.2: Representative paths for Monthly Electricity Spot Prices (2008-2016)
(Dollars/MWhs/Month) Authors’ Simulations
Figure 4.3: Representative paths for Monthly Henry Hub Natural Gas Spot Prices (2008-
2016)
(Dollars/mmBTUs/Month) Authors’ Simulations
13
4.2 Simulating the Impact of Stochastic Energy Prices on a Prop-
erty’s NOI
The goal of our simulation exercise is to consider the importance of energy dynamics on the
valuation of commercial real estate and the derivative contracts, such as mortgages or bonds,
that are collateralized by building cash ?ows (i.e. the lease portfolio rents). Toward this end,
we exploit a variety of information sources to characterize a “representative” central business
district commercial real estate o?ce building in California. The information we require
includes: transaction prices, capitalization rates, square footage, location, net operating
income, market rents, primary tenants, and information concerning the energy usage and
cost of energy for similar buildings in California.
Our real estate market data comes from two sources. The transaction data, including
tenant and building characteristics, come from Real Analytics. We select a building type and
size that corresponds to available information on energy usage and cost using ENERGYIQ,
based upon the Commercial Energy Use Survey (CEUS), developed by the Lawrence Berke-
ley National Laboratory. Our representative building is located in San Diego, it recently
transacted, and it has 225, 000 rentable square feet and tenants that work in ?nancial ser-
vices, insurance, administration, and real estate industries (FIRE). The building was built
in 1999 again giving it appropriate correspondence with the CEUS performance summaries.
Table 4.1: Representative Building Example
Location San Diego, CA
Square Footage 225, 000
Cap rate at sale 6.97%
O?ce-tenant FIRE
Sale price $127, 100, 000.00
NOI at sale $8, 858, 870.00
We obtained information on class A asking rents for San Diego from Grubb & Ellis,
O?ce Market Trends for quarter 1, 2006 through Quarter 1, 2009. This market information
is presented in Table 4.4.
Table 4.2: Class A Asking Rents
Annual $/Square Feet (full Servicegross)
2006 $35.4
2007 $37.10
2008 $39.3
2009 $35.70
Grubb & Ellis, O?ce Market Trends
14
We calculated the energy performance comparables for California commercial o?ce build-
ings of 150, 000 square feet or larger with FIRE tenants, using ENERYIQ. These data sum-
marize the energy performance using information from utility bills for a sample drawn in
2002. The energy consumption breakdown for such buildings is presented in Table 4.3. As
shown, the average electricity consumptions per Kilo Watt Hours (KWh) per square foot per
year for a building of this size and type is 16.24 for electricity and 16.63 thousand British
Thermal Units (KBTU) per square foot per year. Most of the electricity cost is heating and
ventilation, lighting, and the operation of o?ce equipment. Most of the natural gas cost is
heating.
Table 4.3: 2002 Energy Consumption - EnergyIQ- Commercial Energy Use Survey (CEUS )
Electricity Consumption $/ Natural Gas
(kWh/sf/yr) (kBTU/sf-yr)
Year Built Vintage 1991-present 1991-present
Heating 0.36 13.49
Cooling 4.06
Vent 2.3
Lighting 4.86
Drinking & Hot Water 0.13 2.93
O?ce Equip. 3.40
Refrigeration 0.19
Cooking 0.03 0.20
Motors 0.67
Air Compressors 0.07
Misc. 0.16 0.01
Total 16.24 16.63
All Buildings with tenancy as Financial, Insurance, and Real Estate activities, as aggregated by EnergyIQ for
commercial real estate buildings with 150, 000 square feet or more of rentable space. Consumption measured
from 2002 energy bills.
We report the results of our Monte Carlo simulations of monthly energy costs for elec-
tricity and natural gas for our representative building, using our calibrated OU processes
and our assumptions concerning the characteristics of this building in Figures 4.4, 4.5, and
4.6. We make one added assumption that the property was underwritten at the CEUS esti-
mates for the average cost per square foot of electricity use ($1.62) and the average cost per
square foot of natural gas use ($.166) for buildings with rentable square footage of greater
than 155, 000 that were built post 1991. This assumption implies that the observed Net
Operating Income is computed using the CEUS estimates of the energy component of op-
erating costs. Interestingly, the annual average cost of energy for the San Diego building is
$402, 851 using the CEUS/EnergyIQ values for such o?ce buildings and our simulated long
run annual average is $384, 341 using our market calibrated, correlated OU processes. Thus,
our assumption is a conservative representation of the unobserved actual underwriting on
15
the building.
As shown in Figures 4.4, 4.5 and 4.6, there is considerable seasonal variability in both the
electricity and the natural gas components of energy consumption in our building. Figure
4.5 presents the overall simulated total energy consumption of the San Diego building from
April of 2008 through April of 2013. Given current forward contract pricing (See Appendix
I), there is an important summer and winter seasonal to these costs, the summer seasonal
exhibits very signi?cant volatility, and the electricity costs dominate.
Figure 4.4: Simulated Monthly Total Energy Cost (Sum electricity and natural gas) San
Diego FIRE O?ce.
Aggregate monthly costs in dollars for the 225,000 Square Foot Representative San Diego Building.
In Figures 4.5 and 4.6, we report the results of simulating the electricity and natural
gas components of the energy costs and compare these with the CEUS benchmark. In Fig-
ure 4.5, we present the mean and the plus and minus one standard deviation simulation
results for total costs of electricity per month for the representative San Diego building. As
shown, about 17% of the months exhibit energy costs that signi?cantly exceed the assumed
CEUS underwriting cost level and all of the one standard deviation values exceed the CEUS
electricity cost levels. Clearly, the volatility of the summer electricity cost represents an
important source of risk both to building owners and to lenders who ignore energy consump-
tion volatility in their underwriting. It also represents a potentially important opportunity
to consider o?setting positions that would hedge this risk such as installing photovoltaic
technology that would enable the building owner to sell energy to the grid at high energy
16
demand periods and thus o?set their own higher electricity cost periods or buying various
types of energy options.
Figure 4.5: Simulated monthly electric power costs ($ electricity price × MWhs used × total
square footage per Month), the monthly cost for the 225,000 square foot representative San
Diego building.
The CEUS underwriting value is the total monthly cost of electricity assuming that the realized cost is
$1.62 per square foot per year for a 225,000 square foot building.
Figure 4.6 presents the simulation results for the natural gas component of the energy
consumption for the representative San Diego building. Again, there is an evident seasonal,
although the peak to trough dynamics of natural gas is signi?cantly less than that of the of
the electricity total cost dynamics for the building. Again, the natural gas cost component
exceeds the assumed CEUS natural gas energy cost in nearly all periods, so this assumption
appears less conservative for natural gas. A one standard deviation shock to natural gas cost
signi?cantly exceeds the CEUS natural gas cost levels.
To better understand, the underwriting implications of the obvious volatility that charac-
terizes these costs, we carry out a simulation where we assume that the San Diego building
was underwritten to a Debt Service Coverage ratio (DSCR) of 1.25, based on the CEUS
energy cost projections, and the observed net operating income (NOI) of the building at
the beginning of 2008. We assume that the lender underwrites the debt service payments
based on the 1.25 DSCR and that the building owner pays the energy costs over and above
energy costs of the CEUS underwritten values and experiences an augmentation to NOI if
energy costs fall below the CEUS underwritten cost of energy.
8
. We then consider the e?ects
8
We are thus assuming a gross rent structure in which the leases are structure such that rents to the lessee
are set accounting for average energy cost based on CEUS and the lessor/owner pays the realized energy
cost
17
Figure 4.6: Simulated monthly natural gas costs ($ gas price× MBTUs used × total square
footage per Month), the monthly cost for the 225,000 square foot representative San Diego
building.
The CEUS Underwriting value is the total monthly cost of gas assuming that the realized cost is $.166 per
square foot per year for a 225,000 square foot building.
on the mean and standard deviation of the computed DSCR for a one standard deviation
shock and a two standard deviation shock to total energy costs. We consider these e?ects
for the monthly underwriting NOI, of $738, 239. We then reconsider the same experiment,
for the following year, 2009, when rents have fallen to $35.70 per year, or a monthly NOI of
$669, 375.
As shown, the 2008 DSCR just rounds to 1.25, but with a one standard deviation positive
shock the DSCR falls to 1.24 and falls to 1.21 with a two standard deviation shock in
energy prices. Thus, possible shocks lead to signi?cantly lower DSCR for the lender that,
when combined with lower property values due to the increased costs of energy, increase
the likelihood of default on the part of the borrower. The e?ects of this volatility are
exacerbated in 2009 when market rents fall and the NOI falls. The new average DSCR is
1.14 rounded but the one and two standard deviation shocks cause these values to fall to
1.12 and 1.11 respectively (accounting for the standard deviation around the e?ects of these
shocks, the DSCR falls to 1.00.) These negative outcomes are again associated with decreases
in the capitalized value of the NOI, which implies a lower market price (particularly because
capitalization rates rose to 8.35% in San Diego implying a new lower market price). Both
of these outcomes would increase the possibility of default and the need for the lender to
account for this risk in higher initial underwriting standards. Again, the obvious bene?ts of
undertaking energy retro?ts that could mitigate the high volatility of the seasonal in these
costs would clearly be advantageous if fairly priced ?nancing instruments could be designed
to ?nance such investments.
18
Table 4.4: Simulation results for the Average Debt Service Coverage Ratio for the Mean
total energy cost, for a one standard deviation positive shock to total energy cost, and a two
standard deviation shock to total energy costs at Q1 2008 and Q1 2009 Class A rents in San
Diego
Mean One Standard Two Standard
Deviation Deviation
Q1 2008 Rents 1.25 1.24 1.21
Std. Dev. (.007) (.008) (.011)
Q1 2009 Rents 1.14 1.12 1.11
Std. Dev. (.007) (.008) (.011)
Grubb & Ellis, O?ce Market Trends
We also simulate the price of our representative 225,000 square foot San Diego o?ce
building over a ?ve year investment horizon. This simulation is meant to be illustrative of
the importance of adding the stochastic dimension of energy costs into a valuation analysis.
For our analysis we again assume a gross leasing structure in which the level of gross rent
su?cient to achieve the underwritten net operating income of $8,858,870.00 at the CEUS
average energy cost for this type of building given its square footage and use by tenants in
?nancial, insurance, and real estate economic activities. We then simulate the distribution
of the realized net operating income of the representative San Diego property assuming that
NOI varies by the net realized energy cost relative to the CEUS underwritten energy level.
9
We discount using the term structure reported in Appendix I and we capitalize the month
sixty NOI using the observed end of year 2008 cap rate in San Diego of 6.7%. As shown in
4.7, we ?nd under our assumptions that the price of the San Diego building fall below its
underwritten price of $127,100,000.00 more than 10% of the simulated draws. This suggests
that there is a more than 10% chance of default on the mortgage loan on the building
assuming that the borrower would default on the loan when the loan-to-value ratio exceeds
one from below. Obviously, more realism could be added to this simulation, however, given
the relative accuracy of our match to the CEUS initial operating cost of the San Diego
building, we believe that, as structured, our simulation is indicative of the potential risk to
lender’s mortgage holdings that may arise from energy cost risk that is not underwritten
into the mortgage contract structure.
5 Summary and Conclusion
U.S. buildings consume almost 39% of the total U.S. energy usage, and appear ine?cient
compared to the available technology and the e?ciency of European structures. Govern-
ment building codes, disclosure requirements, and ?scal subsidies have been the primary
9
More realism could be added here using stochastic rents that are perhaps correlated with the energy
costs
19
Figure 4.7: Simulated Value of the Representative San Diego O?ce Building on April 2008
Assuming a Five Year Investment Horizon, CEUS Average Energy Cost for this building to
realized Stochastic Electricity and Natural Gas Costs to Compute the Realized Net Operating
Income Over the Holding Period.
0
50
100
150
200
250
300
instruments used to stimulate the adoption of energy-e?cient technology in both the U.S.
and Europe. While these instruments have achieved important successes, they also have
signi?cant limitations.
In this paper, we instead look at market mechanisms to stimulate voluntary private sector
energy-e?cient investments. We focus on a particular market failure, namely the absence of
energy e?ciency as an input to the underwriting decision for mortgage loans on commercial
properties. In principal, rising and volatile energy prices can be as important a source of
mortgage defaults as the standard inputs used in commercial mortgage underwriting.
We argue that the failure of commercial mortgage lenders to take energy-e?ciency into
account is primarily the result of insu?cient information to connect high and volatile energy
prices to mortgage default. Current commercial mortgage underwriting uses the loan to value
ratio (LTVR) and debt service coverage ratio (DSCR) to account for the macroeconomic and
housing market sources of market default. We propose a new energy e?cient score (EES)
and an energy volatility score (EVS) to account for the impact of energy prices on mortgage
default.
We carry out the ?rst step in the development of EES and EVS measures by creating
a dynamic model to compute the impact of stochastic energy prices on the net operating
income (NOI) of a representative building. We use a Monte Carlo methodology to provide
quantitative measures –both expected values and the range of variability–of the resulting
changes in NOI. Our preliminary results indicate that energy price volatility is a previously
unrecognized and signi?cant source of potential commercial mortgage default.
20
A
p
p
e
n
d
i
x
I
:
A
s
s
u
m
e
d
M
a
r
k
e
t
C
o
n
d
i
t
i
o
n
s
A
p
r
i
l
2
0
0
8
M
o
n
t
h
l
y
P
o
w
e
r
&
N
a
t
u
r
a
l
G
a
s
F
u
t
u
r
e
s
a
n
d
E
u
r
o
p
e
a
n
c
a
l
l
s
o
n
f
u
t
u
r
e
s
p
r
i
c
e
s
T
h
e
z
e
r
o
-
p
e
r
i
o
d
f
u
t
u
r
e
s
p
r
i
c
e
s
a
r
e
m
o
n
t
h
l
y
b
l
o
c
k
p
r
i
c
e
s
f
o
r
d
e
l
i
v
e
r
y
i
n
t
h
e
c
u
r
r
e
n
t
m
o
n
t
h
.
T
h
e
u
n
d
e
r
l
y
i
n
g
f
o
r
t
h
e
f
u
t
u
r
e
s
a
r
e
m
o
n
t
h
l
y
?
o
w
s
o
f
p
o
w
e
r
a
n
d
o
f
g
a
s
o
v
e
r
t
h
e
d
e
l
i
v
e
r
y
m
o
n
t
h
.
T
h
e
u
n
d
e
r
l
y
i
n
g
f
o
r
p
o
w
e
r
a
n
d
g
a
s
c
a
l
l
o
p
t
i
o
n
s
i
s
t
h
e
f
u
t
u
r
e
s
p
r
i
c
e
w
h
e
r
e
o
p
t
i
o
n
e
x
p
i
r
y
–
f
u
t
u
r
e
d
e
l
i
v
e
r
y
.
A
l
l
o
f
t
h
e
c
a
l
l
s
a
r
e
s
t
r
u
c
k
a
t
-
t
h
e
-
m
o
n
e
y
-
f
o
r
w
a
r
d
.
(
S
o
u
r
c
e
:
2
0
0
8
F
i
n
a
n
c
i
a
l
E
n
g
i
n
e
e
r
i
n
g
C
a
s
e
C
o
m
p
e
t
i
t
i
o
n
.
F
i
l
l
m
o
r
e
I
n
v
e
s
t
o
r
s
L
L
P
,
a
u
t
h
o
r
e
d
b
y
D
u
a
n
e
S
e
p
p
i
,
C
a
r
n
e
g
i
e
M
e
l
l
o
n
U
n
i
v
e
r
s
i
t
y
.
D
a
t
a
s
o
u
r
c
e
s
a
r
e
t
h
e
E
l
e
c
t
r
i
c
i
t
y
R
e
l
i
a
b
i
l
i
t
y
C
o
u
n
c
i
l
o
f
t
h
e
M
i
d
-
A
t
l
a
n
t
i
c
(
E
R
M
C
M
A
)
a
n
d
B
l
o
o
m
b
e
r
g
,
H
e
n
r
y
H
u
b
s
p
o
t
a
n
d
f
u
t
u
r
e
s
g
a
s
p
r
i
c
e
s
.
A
p
r
.
M
a
y
J
u
n
e
J
u
l
y
A
u
g
.
S
e
p
t
.
O
c
t
.
N
o
v
.
D
e
c
.
J
a
n
.
F
e
b
.
M
a
r
.
A
p
r
i
l
2
0
0
8
2
0
0
8
2
0
0
8
2
0
0
8
2
0
0
8
2
0
0
8
2
0
0
8
2
0
0
8
2
0
0
8
2
0
0
9
2
0
0
9
2
0
0
9
2
0
0
9
U
.
S
.
T
r
e
a
s
u
r
y
T
e
r
m
S
t
r
u
c
t
u
r
e
z
e
r
o
y
i
e
l
d
s
0
.
0
3
0
0
0
.
0
3
2
0
0
.
0
3
2
0
0
.
0
3
1
5
0
.
0
3
2
0
0
.
0
3
2
0
0
.
0
3
2
0
0
.
0
3
3
0
0
.
0
3
3
0
0
.
0
3
3
0
0
.
0
3
3
0
0
.
0
3
3
0
z
e
r
o
p
r
i
c
e
s
$
0
.
9
9
7
5
$
0
.
9
9
4
7
$
0
.
9
9
2
0
$
0
.
9
8
9
6
$
0
.
9
8
6
8
$
0
.
9
8
4
1
$
0
.
9
8
1
5
$
0
.
9
7
8
2
$
0
.
9
7
5
6
$
0
.
9
7
2
9
$
0
.
9
7
0
2
$
0
.
9
6
7
5
E
l
e
c
t
r
i
c
i
t
y
P
r
i
c
e
s
(
p
e
r
M
W
h
)
f
u
t
u
r
e
s
p
r
i
c
e
s
$
8
2
.
0
0
$
8
8
.
1
5
$
9
8
.
3
5
$
1
1
6
.
0
0
$
1
2
4
.
4
0
$
9
0
.
5
0
$
8
6
.
2
5
$
8
0
.
6
5
$
8
6
.
0
5
$
9
6
.
0
0
$
9
7
.
2
0
$
9
1
.
7
5
$
8
0
.
8
0
d
o
l
l
a
r
p
r
i
c
e
o
f
E
u
r
o
p
e
a
n
c
a
l
l
s
$
3
.
5
4
$
5
.
8
9
$
9
.
6
2
$
1
1
.
8
8
$
8
.
9
5
$
1
0
.
0
2
$
9
.
8
5
$
1
0
.
6
5
$
1
2
.
5
6
$
1
3
.
3
6
$
1
3
.
1
8
$
9
.
3
2
N
a
t
u
r
a
l
G
a
s
P
r
i
c
e
s
(
p
e
r
m
m
B
T
U
)
f
u
t
u
r
e
s
p
r
i
c
e
s
$
9
.
6
0
$
9
.
7
9
$
9
.
8
8
$
9
.
9
7
$
1
0
.
0
3
$
1
0
.
0
5
$
1
0
.
1
2
$
1
0
.
4
0
$
1
0
.
7
5
$
1
0
.
9
7
$
1
0
.
9
5
$
1
0
.
7
1
$
9
.
7
5
d
o
l
l
a
r
p
r
i
c
e
o
f
E
u
r
o
p
e
a
n
c
a
l
l
s
$
0
.
3
9
$
0
.
5
8
$
0
.
7
3
$
0
.
8
9
$
1
.
0
7
$
1
.
2
6
$
1
.
3
3
$
1
.
4
3
$
1
.
5
8
$
1
.
7
3
$
1
.
7
7
$
1
.
4
2
21
References
[1] Benth, F.E. and Benth J.S., and Koekebakker, S. (2008), Stochastic Modelling of Elec-
tricity and Related Markets, London, World Scienti?c, Advanced Series on Statistical
Science & Applied Probability, Volume II.
[2] Benth, F. E., Cartea, A., and Kiesel, R. (2006). Pricing forward contracts in power
markets by the certainty equivalence principle: explaining the sign of the market risk
price premium, Forthcoming Journal of Banking Finance.
[3] Benth, F. E., Koekebakker, S., and Ollmar, F. (2007), Extracting and applying smooth
forward curves from average-based commodity contracts with seasonal variation, Journal
of Derivatives, Fall 15(1), pp. 52-66.
[4] Bessembinder, H., and Lemmon, M. (2002), Equilibrium pricing and optimal hedging in
electricity forward markets, Journal of Finance, 57, pp. 1347-1382.
[5] Cartea, A. and Figuerosa, M. G. (2005), Pricing in electricity markets: a mean reversion
jump di?usion model with seasonality, Applied Mathematical Finance, 12 (4), pp. 313-
335.
[6] Deng, S. (2000), Stochastic Models of Energy commodity prices and their applications:
mean-reversion with jumps and spikes, Working paper, PWP-073 of the Program on
Workable Energy Regulation (POWER), University of California Energy Institute.
[7] Eberlein, E. and Stahl, G. (2003), Both sides of a fence: A statistical and regulatory
view of electricity risk, Energy & Power Risk Management, 8, pp. 34-38.
[8] Eichholtz, Piet, Nils Kok, and John Quigley (2009), “Doing Well by Doing Good: Green
O?ce Buildings,” Working Paper No W08-001, Fisher Center for Real Estate and Urban
Economics, University of California, Berkeley.
[9] Geman, H. and Roncoroni, A. (2006), Understanding the ?ne structure of electricity
prices, Journal of Business, 79 (3), pp. 1225-1261.
[10] International Energy Agency (IEA) (2009), Energy Prices and Taxes.
[11] International Energy Agency (IEA) (2008), Energy E?ciency Requirements in Building
Code, Energy E?ciency Policies for New Buildings.
[12] International Energy Agency (IEA) (2004), 30 Years of Energy Use in IEA Countries.
Karlis, D. and Lillestl, J. (2004), Bayesian estimation of NIG models via Markov chain
Monte Carlo methods, Applied Stochastic Models in Business and Industry, 20 (4),
323-338.
[13] McKinsey Global Institute (2007), Curbing Global Energy Demand Growth: The Energy
Productivity Opportunity.
[14] Rand (2009), Improving the Energy Performance of Buildings.
22
[15] Schwartz, E. S. (1997), The stochastic behavior of commodity prices: Implications for
valuation and hedging, Journal of Finance, LII, 3, pp. 923-973.
[16] U.S. Department of Energy (2009), Buildings Energy Data Book.
[17] Weron, R. (2000), Energy Risk Management, Physica A 285, pp. 127-134.
[18] Weron, R. (2004), Computationally intensive Value at Risk calculations, In Handbook
of Computational Statistics, J.E. Gentle, W. Hrdle, and Y. Mori (eds). Springer, Berlin,
pp. 911-950.
[19] Weron, R. (2005), Heavy tails and electricity prices, The Deustsche Bundesbanks 2005
Annual Fall Conference, Eltville.
[20] Weron, R. (2006), Modeling and forecasting electricity loads and prices: A statistical
approach, Chichester, Wiley Finance, John Wiley & Sons.
[21] Weron, R., Kozlowska, B., and Nowicka-Zagrajeck, J. (2001), Modeling electricity loads
in California: A continuous time approach, Physica A 299, pp. 344-350.
23
doc_850285196.pdf
Investments
Dwight Ja?ee and Nancy Wallace
Draft: November 30, 2009
1 Introduction
U.S. buildings consume almost 39% of the total U.S. total energy consumption, making real
estate the largest consuming sector by a considerable margin; see Table 1.1. In comparison,
the industrial consumption share is about 33% of the total and the transportation share is
about 28% of the total. Within the real estate sector, the share of total energy used is almost
equally split between residential (20.9%) and commercial (18.0%) buildings.
Table 1.1: Buildings Share of U.S. Primary Energy Consumption (Percent), 2006
Residential
Buildings
Commercial
Buildings
Total
Buildings
Industry Transportation Total Total Con-
sumption
(quads)
20.9% 18.0% 38.9% 32.7% 28.4% 100% 99.5
Source: U.S. Department of Energy (2009)
The energy consumption of U.S. real estate appears, furthermore, to be substantially
less e?cient than comparable European buildings, even after controlling for such factors
such as climate, GDP, and population.
1
There is no mystery as to the reason: U.S. costs
for electricity and heating oil range from 50 to 75 percent of the levels in most European
countries; see IEA (2009). This indicates there must be feasible technologies that would allow
the energy consumption of U.S. buildings to be reduced signi?cantly. Such investments could
arise as the result of building codes and comparable requirements, or as a voluntary response
to high and volatile energy prices.
Most economists would agree that the “?rst-best” solution to reducing U.S. energy con-
sumption, across the three major uses of real estate, industry, and transportation, is to raise
1
IEA (2008) and (2004) provide comparisons of residential energy use in the U.S. and Europe corrected for
climate and measured per unit of GDP or per capital. McKinsey (2007) also shows substantially higher U.S.
energy consumption compared to Europe after controlling for GDP and population. Rand (2009) provides
a discussion comparing energy use in the U.S., Australia and the European Union.
1
the U.S. price of energy through appropriate ?scal instruments. While this may occur in
the future, the timing of this change in policy remains highly uncertain. So it is sensible,
even critical, to look for alternative, accelerated, mechanisms to reduce U.S. energy con-
sumption. Indeed, if and when U.S. energy prices do rise, the economic adjustment will be
easier and faster if the transformation toward a more energy e?cient technology is already
well along. Facing inadequate energy-e?cient investments in real estate, governments in
both the U.S. and Europe have intervened to provide additional incentives. To date, three
approaches have dominated theses interventions: expanded building codes, energy-e?ciency
certi?cations, and direct ?scal subsidies.
Expanded building codes have been the primary mechanism to ensure energy-e?cient
structures in Europe and are likely to expand over time in the U.S. as well. Building code
requirements, however, have their primary impact on new construction, and given the long
durability of most buildings, new construction annually represents a very small percentage
of the existing building inventory. Most building codes, furthermore, are prescriptive rather
than performance-based, which limits the incentive they provide for new innovative solutions.
Disclosure certi?cates at the time of building sales have also become an important mecha-
nism in Europe. The 2002 EU Energy Performance in Buildings Directive requires an energy
performance certi?cate based on either the building’s design or usage characteristics. While
such disclosures might well have an impact on the sales prices, it is unclear to what extent
such disclosures motivate either the seller or buyer to initiate energy e?cient investments. In
the U.S., certi?cations such as Leeds and Energy Star are available, and such certi?cations
are sought by builders and developers who are planning to create energy e?cient buildings.
But these certi?cations primarily apply to new buildings, and it is unclear to what extent
the certi?cations are the factor actually motivating energy-e?ciency in these buildings.
2
Direct subsidies provided by either government agencies and public utilities may provide
economic stimulus to energy e?cient investments. In these times of harsh ?scal budgets,
however, direct subsidies are unlikely to be a major driver of new energy-e?cient investments.
Given the limitations of building code requirements, energy-e?ciency certi?cations, and
direct subsidies, it is critical to consider whether there are other solutions to the private mar-
ket failures that are contributing to the underinvestment in U.S. building energy e?ciency.
It is possible, of course, that the available technology is just not pro?table in net present
value (NPV) terms. However, it appears that NPV positive energy-e?cient investment op-
portunities already exist, and that further innovations and cost-savings from scale economies
are literally in process. In this paper, we focus on loan market frictions as a market fail-
ure that raises the cost and limits the availability of mortgage funding for energy-e?cient
investments. Given the large ratio of the existing stock to new investment, our proposed so-
lutions emphasize market mechanisms to eliminate the frictions that inhibit energy retro?ts
in existing structures. These same solutions will also expand the incentives to embed the
best technology in new construction. In either case, we can identify three steps that must
2
Eichholtz, Kok, and Quigley (2009) show that buildings with certi?cations obtain higher sales prices,
which provides a motivation for obtaining such certi?cations. However, it is unclear whether the existence
of the certi?cations motivates energy e?cient investments that otherwise would not have been carried out.
The existence of higher sales prices on certi?ed buildings also does not indicate whether the initial energy-
e?ciency investments were NPV positive; to show that, the investments costs must be compared with the
sales prices.
2
occur if building owners are to invest voluntarily in energy saving equipment:
1. Identi?cation of worthwhile investments, Building owners must be able to identify those
energy-e?cient investments that meet at least a minimal standard of pro?tability. As a
simple example, for any investment to add value, it must at least provide a non-negative
NPV when evaluated with a zero interest rate. In other words, the investment must
pass an internal rate of return (IRR) hurdle of zero, which is to say that the sum of
the actual cash ?ow bene?ts must exceed the investment costs.
2. Computation of the true NPV. Investment decisions, of course, will be based on a
NPV > 0 criterion, where the NPV is evaluated at the owner’s true opportunity cost
of capital. For many real estate investments, this opportunity cost will equal the
interest rate that is applied in ?nancing the investment. The NPV criterion also allows
the property owner to prioritize alternative investments, giving the highest priority to
the highest NPV projects. This is particularly relevant if there is a cap on the total
amount of available funding.
3. Identi?cation and removal of ?nancial frictions. Frictions in the lending market raise
the discount rate and lower the NPV computations, thus limiting the set of feasible
investments and acting as a deterrent to those investments with the longest payo?
structures. Or to put it in a positive vein, the removal of such frictions can stimulate
the aggregate amount of energy e?cient investments, and provide particular bene?ts
to those investments with the longest-term payo?s. The identi?cation of these frictions
and proposals to remove them is the direct focus of this paper.
The agenda of this paper is as follows. In Section 2, we describe the loan market fric-
tions that currently inhibit energy-e?cient investments in the commercial real estate sector.
In Section 3, we describe our proposals to help eliminate these frictions. Our proposals in
Section 3 require a mechanism through which ?rms can compute the NPV of their proposed
energy investments and convince their lenders of the bene?ts of funding these investments.
Thus, in Section 4, we provide a feasible method to compute the net present value of alterna-
tive energy-e?cient investments. The key innovation here is to use Monte Carlo simulations
to incorporate stochastic energy prices into the valuation process. In Section 5, we provide
a summary and conclusions.
2 Commercial Mortgage Market Frictions that Inhibit
Energy-E?cienty Investments
Commercial property mortgage lending in the United States traditionally focuses on two key
risk measures for underwriting mortgages: the loan-to-value ratio (LTVR, the ratio of the
mortgage balance to the value of the building), and the debt-service-coverage ratio (DSCR,
the ratio of the propertys net operating income to the principal and interest payments on the
mortgage debt). These ratios are also monitored by bank regulators, including the Federal
Reserve, the Comptroller of the Currency, and the O?ce of Thrift Supervision, because they
3
are important indicators of the quality of commercial bank underwriting and the mortgage-
related default-risk exposure. Sound underwriting standards typically require a LTVR no
more than 65% and a DSCR no less than 1.25, although, of course, other factors such as
the quality and timing of tenant leases will have an impact on these ratios. These ratios
are important because mortgage performance data have shown that mortgage borrowers are
more likely to default as the LTVR rises toward 1.0 or as the DSCR falls toward 1.0.
The key role of the DSCR and the LTVR for underwriting commercial mortgages has
become a critical impediment for the recognition of volatile energy costs as an additional and
major source of mortgage default risk for commercial buildings. Net operating income (NOI,
the net rental income associated with a commercial structure) is the key input in computing
both the DSCR and the LTVR. NOI appears directly in the numerator of the DSCR. The
present value of NOI is also the fundamental input in determining the property valuation
that appears in the denominator of the LTRV. In practice, the forecasted net operating
income of a commercial building is constructed by:
1. Aggregate the forecasted contractual rental income from the tenants’ leases;
2. Subtract the forecasted building operating expenses, including the costs of energy; and
3. Adding back the forecasted energy-use reimbursements from the tenant to the landlord,
as required on triple net leases that are the most common commercial lease contracting
structure.
The triple-net lease, and the software package, ARGUS, which is widely used to organize
and display the information, thus “nets” out the energy risk exposure of buildings, other
than those energy costs borne solely by the landlord due to vacancies, joint costs associated
with common areas such as lobbies, and incomplete energy reimbursements from the tenants
to the building owner. The result is that mortgage lenders typically do not account for the
risks created by a commercial buildings energy costs. In particular, commercial mortgage
underwriters currently have no validated system to score the default risk created by energy-
cost volatility in a manner similar to their use of the DSCR and the LTVR to score the
overall default risk.
Of course, with triple-net leases, tenants in more energy-e?cient buildings should bene?t
from lower energy bills, and thus should be willing to pay higher rents for space in such
buildings. Three conditions have to be realized, however, if the result is to be more energy-
e?cient investments:
1. Tenants must validate the lower expected energy costs before they accept the higher
rents;
2. Landlords must con?rm the higher rents before undertaking the energy-e?cient invest-
ments;
3. The actual investments must be NPV positive in order to ensure that a true net bene?t
exists.
Imperfect information regarding future energy costs and the e?cacy of energy-e?cient
investments makes each of these steps problematic. The result is that the existing systems
4
for commercial mortgage lending and property leasing inhibit energy-e?cient investments in
this sector.
The positive news is that our proposed tools for evaluating and disclosing energy e?cient
investments, to which we next turn, can play an important role in mitigating the frictions that
currently constrain both owner-lender mortgage contracts and owner-tenant lease contracts.
3 Mitigating Financing Frictions for Commercial Real
Estate
The unsatisfactory situation for the ?nancing of energy-e?cient investments in commercial
real estate can be potentially mitigated by energy e?ciency metrics and disclosures for
mortgage underwriting and tenant leasing. These metrics allow the comparison of energy-
related costs and risks across building types, building qualities, and tenanting structures.
Just as LTVR and DSCR are informative about the potential default risk of newly originated
mortgages, so too our proposed Energy E?ciency Score (EES) and Energy Volatility Score
(EVS) provide default risk metrics based on the impact of energy-cost savings on a building’s
net operating income. The bene?ts include both a higher net operating income as a result
of lower energy costs and a more stable net operating income in the face of volatile energy
costs. The goal is that the two new energy e?ciency metrics will be used by commercial loan
underwriters to provide more favorable loan terms to loan applicants with energy-e?cient
buildings.. The two scores will also provide property owners with a means to determine the
most cost-e?ective energy-e?cient investments, and tenants with a means to estimate their
savings in energy costs as a result of renting space in an energy-e?cient building.
Figure 3.1: Asset Valuation
5
3.1 Energy-E?ciency Metrics to Facilitate Commercial Mortgage
Lending
Figure 3.1 provides an overview of how our proposed energy e?ciency (EE) mortgage un-
derwriting rating system would work. It uses both engineering-based simulation methods
focusing on an integrated analysis of whole building systems to benchmark the relative en-
ergy consumption of lighting, HVAC, and plug-load functions in commercial buildings of
di?ering types and qualities; and performance-based empirical benchmarking using statisti-
cal estimation methods. The output from our coordinated energy e?ciency benchmarking
analyses will be technical metrics that are summarized in the proposed scoring metrics for
mortgage underwriting: the EES and EVS. These two scores are related to the level and
volatility of the net operating income of commercial buildings. The scores can also be incor-
porated as inputs into mortgage valuation models that are based on the dynamics of interest
rates and asset prices.
As shown on the right-hand side of the ?ow diagram, the mortgage valuation technology
can be used prescriptively to determine the appropriate interest rate to be charged, and/or
the maximum loan size to be granted for buildings with di?ering EES and EVS. This pre-
scriptive exercise takes the energy e?ciency scores as given and solves for the actuarially fair
mortgage contract terms (e.g. by setting lower mortgage rates, higher LTVRs, and lower
DSCRs as compensation for a higher energy e?ciency score or a lower energy volatility
score). The scores may also be used to induce renters to pay higher rents in exchange for
the bene?t of lower energy costs.
By measuring the EES and EVS that result from alternative energy-e?cient investments,
we can also derive a third measure, the Energy E?ciency Action Ratio (EEAR). As shown in
the ?ow diagram, this third ratio allows a decision maker to rank order the cost e?ectiveness
of alternative investments to change the energy e?ciency of buildings. This analysis is
a particularly important if mortgage lenders are going to recognize that energy-e?cient
investments in commercial buildings can directly reduce their mortgage default risk by raising
the property’s NOI and by reducing the sensitivity of the NOI to future energy cost volatility.
In Section 4 below, we take the ?rst steps to implement this program by carrying out
dynamic Monte Carlo valuation of the possible future energy costs of a representative building
based on alternative stochastic paths for energy costs. Future work will then expand this tool
to compute the energy e?ciency and energy volatility scores and the energy e?ciency action
ratio for buildings with any actual or simulated physical and environmental characteristics.
3.2 Alternative Methods to Facilitate Commercial Mortgage Lend-
ing
Figure 3.2 introduces two alternative green ?nancing instruments–Managed Energy Con-
tracts and Property Tax Assessed Energy Retro?t Financing and compares them with a
straight commercial mortgages. Managed Energy Contracts (MEC) are privately provided
leasing contracts in which an energy service provider commits to provide all energy related
services to a property owner, typically a Real Estate Investment Trust or large real estate
portfolio operator, in exchange for the payment of the property operators historical energy
costs. The energy service provider pro?ts if it achieves energy cost savings in excess of the
6
Figure 3.2: Location in the “Cash Flow Stack” of Commercial Real Estate Mortgages
(CREM), Managed Energy Contracts (MEC), and Property Tax Assessed Energy Retro?t
Financing (PTA-ERF) for Commercial Real Estate Assets
present value of any investments it makes. At the end of the lease term, typically ten years,
the energy e?ciency improvements remain with the building, thus enhancing the value of
the building owner’s asset. Since the MEC lease payments are equivalent to the historical
utility fee structure of the property, the full utility fee structure remains fully recoverable
from tenant reimbursements. While energy e?ciency gains are achieved through MEC, the
full cost savings of these investments are captured by the tenants or the building owner only
at the lease maturity date.
Property Tax Assessed Energy Retro?t Financing (PTEF) is a very recent innovation in
green energy ?nancing pioneered by the city of Palm Desert, California through the Palm
Desert Energy Partnership, and subsequently was codi?ed into California State Law in July,
2008 with the passage of Assembly Bill 811. The goal of AB811 is to allow California cities
to ?nance permanently ?xed renewable energy generation improvements in private property
including commercial real estate assets. Although the ?nancing strategies may di?er by
city, the bill allows for the creation of special tax assessment districts in which individual
property owners can opt for energy retro?t project loans with long maturities (e.g. 20 years
for the CityFIRST program in Berkeley, California). These loans have a contractual payment
structure that is similar to loans, except that they are paid back using a fully amortizing
installment payment that is added to the property taxes of the asset. In addition, the PTEF
loan liability is a tax lien with priority over mortgage liens and it is attached to the property
in contrast to a mortgage loan which is the liability of the borrower and is collateralized by
7
the property.
As shown in ?gure 3.2, the three types of commercial real estate green ?nancing are
attached at di?erent positions in the cash ?ow stack of commercial real estate assets. All
three options, usually involve a commercial real estate mortgage (CREM) and therefore a
substantial fraction of the net operating income, the primary mortgage underwriting cash
?ow, is allocated to the debt service on the loan subject to the limitations of the debt service
coverage ratio. The relative energy risk exposure of the CREM would, however, di?er across
the three structures represented in ?gure 3.2. The straight commercial mortgage presents the
lender with the maximal energy risk exposure. In contrast, the PTEF increases the property
tax liability of the property and thus reduces both the size and risk of the commercial
mortgage loan through the cost savings achieved by an energy-e?cient retro?t investment.
Even further, the use of a MEC largely eliminates the energy risk of the commercial real
estate loan because the property’s energy costs are ?xed at the historical expense level.
Although MEC and PTEF have di?erent payment priorities, the commitment of the
building owner is similar across the two contracts (technically MEC are lease ?nancings).
Both of these commitments represent forms of ?xed income securities in which an initial
investment by the lender, or lessor, is paid back with a set of installment payments by
the borrower, lessee. For this reason, they both can be underwritten for risk and priced
using the same form of ?nancial modeling techniques that we develop in Section 4 below.
Alternatively, the market prices of these two contract types can be directly related to our
two energy e?ciency scores, and the preferred investments will be identi?ed our the energy
e?ciency corrective action ratios.
Table 3.1 provides a further comparison across these three green ?nancing instruments.
For one thing, currently there are no standardized energy e?ciency metrics or valuation
tools to underwrite or price any of these ?nancing alternatives. Without an organized capital
market for MEC and PTEF instruments, valuation occurs on a case by case basis. Both
CREM and PTEF involve loan contracts, or promissory notes, whereas the MEC involves a
lease contract which is simply another form of ?xed income security. The seniority of these
instruments, as previously discussed, also di?ers with PTEF instruments having the most
secure interest.
The U.S. commercial real estate mortgage market, based on its size and long history, does
have well developed and standardized methods for valuing the risks that arise from macro-
fundamentals such as interest rates and asset price dynamics. In contrast, the valuation
standards applied in MEC and PTEF markets are currently entirely private and proprietary.
The CREM market also currently services all sizes and types of commercial building, whereas
MEC usually require larger scale to achieve energy e?ciency gains and thus they are used
primarily for Real Estate Investment Trusts or large real estate operating companies.
Two further limitations arise with regard to the MEC and PTEF mechanisms. With
regard to the MECs, there is a potential principal-agent problem in that the energy-saving
innovations introduced by the service provider could have a negative impact on the build-
ing’s serviceability. In fact, this may be an important factor that has limited the use of
these contracts. With regard to PTEFs, most commercial mortgages have a covenant that
constrains the property owner from creating new liens senior to the mortgage lien. The
mortgage lender’s permission is thus required to use the PTEF mechanism. In principal, the
mortgage lender would bene?t and thereby grant permission– if the energy-e?cient invest-
8
ment were NPV positive, since the increase in property value (i.e., the loan collateral) would
then exceed the value of the new, senior, property lien. The method we develop in Section
4 below would provide the lender with the necessary information to make the decision.
Table 3.1: Alternate Financing Methods Comparison
Commercial Real Estate
Mortgage (CREM)
Managed Energy Con-
tracts (MEC)
Property-Tax Assessed
Energy Retro?t Financing
(PTA-ERF)
Standardized Energy E?-
ciency Metrics
No No No
Energy E?ciency Valua-
tion Tools
None Net present value of
agent’s energy e?ciency
investments and pass-
through of existing utility
payments
Net present value of en-
ergy retro?t investment
and amortized loan pay-
ments
Security of Interest Mortgage lien on Property Commitment on utility
fees
Tax Lien on Property
Contract Type Promissory Note Utility provision lease. Promissory Note
Valuation Standards Actuarial and regulated Private Private
Installment Payment
Structure
Long amortization (15 to
20 years) due in 7 or 10
years
10 year renewable con-
tracts
Add-on to Property Taxes
with 20 year payback
Ownership of Energy Re-
lated Investment
Building Owner Building Owner Building Owner
Capital Market Funding
Sources
Commercial Banks, Com-
mercial Mortgage Backed
Securities, Insurance Cos.,
Pension Funds
Private Companies (e.g.
Transcend Equity De-
velopment Corp.) - if
scale su?cient potential
for lease securitization
capital markets
Private Companies (e.g.
Renewable Funding Fi-
nancial Partners) if scale
su?cient Municipal Bond
Market
Commercial Real Estate
Applications
All commercial property
types and building sizes
Real Estate Investment
Trusts, Large Real Estate
Operating Companies
Residential only, autho-
rization exists for com-
mercial real estate
Market Penetration $2.37 Trillion Capital
Market
Small Private Operating
Companies (e.g. Tran-
scend - MESA)
Small Private Operating
Companies (e.g. Re-
newable Funding Finan-
cial Partners)
Performance Risk Default and Prepayment
Risk of Borrowers
Agent’s failure to render
energy services, property
owner’s failure to pass-
through utility fees
Property owners non-
payment of property
taxes
Energy Retro?t Expenses
Recoverable (Escalatable)
from Tenants
Yes, if operating cost,
repairs/maintenance, or
cost reducing. No, if
capital improvements.
Lease payment is an esca-
latable operating expense.
Accounting opinions exist.
Tax payment is escalat-
able, no accounting rul-
ings.
Finally, CREM, MEC, and PTEF share comparable default risks due to the failure to
make payments per the contractual agreements, by the borrower, lessee, or taxpayer respec-
tively. Finally, the ability of the building owner to pass energy e?ciency related expenses
9
through to the tenants depends on the language of the tenant leases and whether the invest-
ments are related to ?xed capital improvements or to maintenance and repairs. MEC and
PTEF may blur this distinction, however, since there are no de?nitive FASB rulings. What
is clear from the summary presented in Table 3.1 is that all three green ?nancing strategies
could bene?t from the development of suitable energy e?ciency metrics, the topic to which
we now turn.
4 Monte Carlo Simulation to Compute NPV for En-
ergy Investments
As previously introduced, summary statistics, or scores, for the energy e?ciency and energy
volatility of commercial real estate assets are critical if we are to achieve private market
solutions to the mortgage market frictions that currently inhibit energy-e?cient investments
in commercial buildings. The key is to quantify the impact of possible future energy costs
on a building’s NOI. NOI is, in turn, a critical component in computing the property’s
DSCR and LTVR, the ?nancial ratios used to determine the mortgage contract terms and
amount. Thus NOI is the critical statistic through which energy cost variability is linked to
the mortgage lending decision.
We next provide a prototype model that allows us to compute the impact of energy
price volatility on the NOI (net of energy costs), for a representative building. We use
a Monte Carlo simulation methodology in which alternative future paths of energy prices
are generated as the output of a stochastic process. We then compute the impact of the
alternative energy paths on the property’s NOI.
4.1 Stochastic Modeling of Energy Spot Dynamics
The classical stochastic process for the spot dynamics of commodity prices was developed
by Schwartz (1997) who speci?ed the process as the exponential of an Orenstein Uhlenbeck
(OU) process. Mean reversion has been identi?ed as an important empirical regularity of the
price dynamics of natural gas and electricity. There is also a growing literature considering
improvements on the Schwartz model such as incorporating jumps to handle the empirical
spikes in many commodity prices, such as electricity, modeling the seasonality of commodity
prices over heavy and lighter consumption months, and ?tting the correlation structure of
commodity prices and exogenous factors such as temperature dynamics.
3
For our modeling of energy price dynamics, we will also use an Schwartz-type OU process.
We implement our simulation model as ?exibly as possible to allow for inclusion of important
empirical characteristics of electricity such as mean-reversion, seasonalities in price levels
3
Recent work in modeling energy dynamic processes include: Benth, F. E., Cartea, A., and Kiesel, R.
(2006). Benth, F. E., Koekebakker, S., and Ollmar, F. (2007), Bessembinder, H., and Lemmon, M. (2002),
Cartea, A. and Figuerosa, M. G. (2005), Deng, S. (2000), Eberlein, E. and Stahl, G. (2003), Geman, H.
and Roncoroni, A. (2006), Karlis, D. and Lillestl, J. (2004), Weron, R. (2000), Weron, R. (2004), Weron, R.
(2005), and Weron, R., Kozlowska, B., and Nowicka-Zagrajeck, J. (2001). For excellent surveys of modern
energy modeling techniques see Weron, R. (2006) and Benth, F.E. and Benth J.S., and Koekebakker, S.
(2008).
10
and volatility, heteroscedasticity, co-variation with fuel prices, and high frequency (daily
or intraday) jumps. Natural gas spot prices also have mean reversion, seasonalities and
heteroscedasticity. Thus, the price dynamics for the natural log of electricity and natural
gas prices, dp
et
and dp
gt
,
4
can be represented by generic equations of the form:
5
dp
et
= ?
e
[?
e
(t) ?p
et
]dt + v
e
(p
et
, t, . . .)dX
e
t
dp
gt
= ?
g
[?
g
(t) ?p
gt
]dt + v
g
(p
gt
, t, . . .)dX
g
t
where the local volatility functions, v
e
(p
et
, t, . . .) and v
g
(p
gt
, t, . . .), can be constant, deter-
ministically dependent on time t, or deterministically dependent on any other relevant state
variables. The Brownian motion increments, dX
e
t
and dX
g
t
, have historically been highly
correlated for electric power and gas.
Most commodity option pricing models assume that the risk neutral dynamics have the
same qualitative characteristics as the objective dynamics. The parameters of the risk neutral
dynamics are then determined using a mixture of statistical calibration (using time series
data from the spot market for natural gas and electricity, such as the Henry Hub natural gas
spot prices shown in Figure 4.1) and implied calibration of the volatility parameters from the
natural gas and electricity futures and options markets in the monthly block market. The
table in Appendix I gives the data used for the electric power and gas derivatives markets
and for the term structure of U.S. Treasury securities in April 2008.
In this paper, we model the actual resource costs of natural gas and electricity, not the
utility power company regulated pass-through of these resource costs. We focus on two
distinct markets in which energy is traded. These markets di?er in the “time granularity”
of the underlying variables and types of optionality traded and in their market participants.
1. The ?rst market is the relatively liquid over-the-counter (OTC) market for futures and
options on monthly blocks (i.e., portfolios) of daily ?ows of electric power and gas (e.g.,
1 MW of electricity each hour for each day over the delivery month or 1 mmBTU of gas
each day over the delivery month). Prices are quoted on a 1 megawatt hour (MWH)
or 1 mmBTU basis. Some of this trading occurs on exchanges (e.g., NYMX) and some
is OTC. We will call this the monthly block market. Since a capability for physical
delivery is not necessary in this market, parties without the capacity to handle energy
physically can simply close out their positions via o?setting trades to avoid delivery.
The future and options markets have the broadest investor participation.
2. The second market is also an OTC market in which same-day blocks of spot electric
power (or spot gas in daily blocks) are bought and sold in real-time. In this spot
market, trades occur through a decentralized search/negotiation process. Nevertheless,
a daily spot price index is publicly reported based on surveys of transactions by electric
power broker/dealers. Agents that trade in this market must be able to deliver/receive
electrons or gas physically in real time.
6
4
Where p
et
= lnP
et
and p
gt
= lnP
gt
.
5
For simplicity, in this version of the paper we will ignore Poisson price jumps in both the gas and
electricity modeling. In future versions of our simulations we intend to model these Poisson processes as a
function of temperature dynamics.
6
For simplicity in this version of our simulations we are making some signi?cant simpli?cations about
11
Figure 4.1: Henry Hub Spot Price
Source: Bloomberg, Henry Hub spot prices
We calibrate our OU processes to the futures and options prices reported in Appendix
I for April 2008 and to electricity and Henry Hub spot price dynamics.
7
We use a simple
deterministic volatility speci?cations v
e
(t) and v
g
(t) and ?t an empirical value for ?
e
of 5.2
for electricity and an ?
p
of 4.4 for natural gas and an electricity/gas correlation of 0.9 for
their respective Brownian motions. We ?t the ?
e
(t), ?
g
(t), v
e
(t), and v
g
(t) to the futures
and options data using least squares. We are thus ?tting the OU dynamics to the observed
term structure of forward prices for electricity and natural gas as of April 1, 2008. We then
simulate ?ve years of monthly spot price dynamics using 1000 Monte Carlo simulations.
Representative price paths using a ?xed correlation structure of 0.9 for our ?tted OU energy
dynamic processes are reported in Figures 4.2 and 4.3.
energy markets. First, the spot market actually trades hourly rather than daily blocks of power in real time.
Thus, the spot price energy is literally the price of energy over the shortest operational decision intervals:
one hour for power and one day for gas. Second, there is one more important market which is ignored in our
current simpli?ed simulation strategy. This is an OTC market for one-day ahead forward contracts on one
day’s worth of power (or gas). For example, on Tuesday an investor can lock in the price for a day-long ?ow
of a 1 MW of electricity (or delivery of 1 mmBTU) on Wednesday. Because physical delivery is required,
this market is also limited to parties with contractual or operational control over actual power or gas. There
is no term structure information for daily deliveries since there are no daily forwards for maturities longer
than one-day ahead.
7
For this version of the paper our electricity spot data are actually from the Electricity Reliability Council
of the Mid-Atlantic (ERCMA) because our CALPX data only goes through 2001. We hope to obtain CALPX
data to re?t the OU for electricity to California derivatives markets.
12
Figure 4.2: Representative paths for Monthly Electricity Spot Prices (2008-2016)
(Dollars/MWhs/Month) Authors’ Simulations
Figure 4.3: Representative paths for Monthly Henry Hub Natural Gas Spot Prices (2008-
2016)
(Dollars/mmBTUs/Month) Authors’ Simulations
13
4.2 Simulating the Impact of Stochastic Energy Prices on a Prop-
erty’s NOI
The goal of our simulation exercise is to consider the importance of energy dynamics on the
valuation of commercial real estate and the derivative contracts, such as mortgages or bonds,
that are collateralized by building cash ?ows (i.e. the lease portfolio rents). Toward this end,
we exploit a variety of information sources to characterize a “representative” central business
district commercial real estate o?ce building in California. The information we require
includes: transaction prices, capitalization rates, square footage, location, net operating
income, market rents, primary tenants, and information concerning the energy usage and
cost of energy for similar buildings in California.
Our real estate market data comes from two sources. The transaction data, including
tenant and building characteristics, come from Real Analytics. We select a building type and
size that corresponds to available information on energy usage and cost using ENERGYIQ,
based upon the Commercial Energy Use Survey (CEUS), developed by the Lawrence Berke-
ley National Laboratory. Our representative building is located in San Diego, it recently
transacted, and it has 225, 000 rentable square feet and tenants that work in ?nancial ser-
vices, insurance, administration, and real estate industries (FIRE). The building was built
in 1999 again giving it appropriate correspondence with the CEUS performance summaries.
Table 4.1: Representative Building Example
Location San Diego, CA
Square Footage 225, 000
Cap rate at sale 6.97%
O?ce-tenant FIRE
Sale price $127, 100, 000.00
NOI at sale $8, 858, 870.00
We obtained information on class A asking rents for San Diego from Grubb & Ellis,
O?ce Market Trends for quarter 1, 2006 through Quarter 1, 2009. This market information
is presented in Table 4.4.
Table 4.2: Class A Asking Rents
Annual $/Square Feet (full Servicegross)
2006 $35.4
2007 $37.10
2008 $39.3
2009 $35.70
Grubb & Ellis, O?ce Market Trends
14
We calculated the energy performance comparables for California commercial o?ce build-
ings of 150, 000 square feet or larger with FIRE tenants, using ENERYIQ. These data sum-
marize the energy performance using information from utility bills for a sample drawn in
2002. The energy consumption breakdown for such buildings is presented in Table 4.3. As
shown, the average electricity consumptions per Kilo Watt Hours (KWh) per square foot per
year for a building of this size and type is 16.24 for electricity and 16.63 thousand British
Thermal Units (KBTU) per square foot per year. Most of the electricity cost is heating and
ventilation, lighting, and the operation of o?ce equipment. Most of the natural gas cost is
heating.
Table 4.3: 2002 Energy Consumption - EnergyIQ- Commercial Energy Use Survey (CEUS )
Electricity Consumption $/ Natural Gas
(kWh/sf/yr) (kBTU/sf-yr)
Year Built Vintage 1991-present 1991-present
Heating 0.36 13.49
Cooling 4.06
Vent 2.3
Lighting 4.86
Drinking & Hot Water 0.13 2.93
O?ce Equip. 3.40
Refrigeration 0.19
Cooking 0.03 0.20
Motors 0.67
Air Compressors 0.07
Misc. 0.16 0.01
Total 16.24 16.63
All Buildings with tenancy as Financial, Insurance, and Real Estate activities, as aggregated by EnergyIQ for
commercial real estate buildings with 150, 000 square feet or more of rentable space. Consumption measured
from 2002 energy bills.
We report the results of our Monte Carlo simulations of monthly energy costs for elec-
tricity and natural gas for our representative building, using our calibrated OU processes
and our assumptions concerning the characteristics of this building in Figures 4.4, 4.5, and
4.6. We make one added assumption that the property was underwritten at the CEUS esti-
mates for the average cost per square foot of electricity use ($1.62) and the average cost per
square foot of natural gas use ($.166) for buildings with rentable square footage of greater
than 155, 000 that were built post 1991. This assumption implies that the observed Net
Operating Income is computed using the CEUS estimates of the energy component of op-
erating costs. Interestingly, the annual average cost of energy for the San Diego building is
$402, 851 using the CEUS/EnergyIQ values for such o?ce buildings and our simulated long
run annual average is $384, 341 using our market calibrated, correlated OU processes. Thus,
our assumption is a conservative representation of the unobserved actual underwriting on
15
the building.
As shown in Figures 4.4, 4.5 and 4.6, there is considerable seasonal variability in both the
electricity and the natural gas components of energy consumption in our building. Figure
4.5 presents the overall simulated total energy consumption of the San Diego building from
April of 2008 through April of 2013. Given current forward contract pricing (See Appendix
I), there is an important summer and winter seasonal to these costs, the summer seasonal
exhibits very signi?cant volatility, and the electricity costs dominate.
Figure 4.4: Simulated Monthly Total Energy Cost (Sum electricity and natural gas) San
Diego FIRE O?ce.
Aggregate monthly costs in dollars for the 225,000 Square Foot Representative San Diego Building.
In Figures 4.5 and 4.6, we report the results of simulating the electricity and natural
gas components of the energy costs and compare these with the CEUS benchmark. In Fig-
ure 4.5, we present the mean and the plus and minus one standard deviation simulation
results for total costs of electricity per month for the representative San Diego building. As
shown, about 17% of the months exhibit energy costs that signi?cantly exceed the assumed
CEUS underwriting cost level and all of the one standard deviation values exceed the CEUS
electricity cost levels. Clearly, the volatility of the summer electricity cost represents an
important source of risk both to building owners and to lenders who ignore energy consump-
tion volatility in their underwriting. It also represents a potentially important opportunity
to consider o?setting positions that would hedge this risk such as installing photovoltaic
technology that would enable the building owner to sell energy to the grid at high energy
16
demand periods and thus o?set their own higher electricity cost periods or buying various
types of energy options.
Figure 4.5: Simulated monthly electric power costs ($ electricity price × MWhs used × total
square footage per Month), the monthly cost for the 225,000 square foot representative San
Diego building.
The CEUS underwriting value is the total monthly cost of electricity assuming that the realized cost is
$1.62 per square foot per year for a 225,000 square foot building.
Figure 4.6 presents the simulation results for the natural gas component of the energy
consumption for the representative San Diego building. Again, there is an evident seasonal,
although the peak to trough dynamics of natural gas is signi?cantly less than that of the of
the electricity total cost dynamics for the building. Again, the natural gas cost component
exceeds the assumed CEUS natural gas energy cost in nearly all periods, so this assumption
appears less conservative for natural gas. A one standard deviation shock to natural gas cost
signi?cantly exceeds the CEUS natural gas cost levels.
To better understand, the underwriting implications of the obvious volatility that charac-
terizes these costs, we carry out a simulation where we assume that the San Diego building
was underwritten to a Debt Service Coverage ratio (DSCR) of 1.25, based on the CEUS
energy cost projections, and the observed net operating income (NOI) of the building at
the beginning of 2008. We assume that the lender underwrites the debt service payments
based on the 1.25 DSCR and that the building owner pays the energy costs over and above
energy costs of the CEUS underwritten values and experiences an augmentation to NOI if
energy costs fall below the CEUS underwritten cost of energy.
8
. We then consider the e?ects
8
We are thus assuming a gross rent structure in which the leases are structure such that rents to the lessee
are set accounting for average energy cost based on CEUS and the lessor/owner pays the realized energy
cost
17
Figure 4.6: Simulated monthly natural gas costs ($ gas price× MBTUs used × total square
footage per Month), the monthly cost for the 225,000 square foot representative San Diego
building.
The CEUS Underwriting value is the total monthly cost of gas assuming that the realized cost is $.166 per
square foot per year for a 225,000 square foot building.
on the mean and standard deviation of the computed DSCR for a one standard deviation
shock and a two standard deviation shock to total energy costs. We consider these e?ects
for the monthly underwriting NOI, of $738, 239. We then reconsider the same experiment,
for the following year, 2009, when rents have fallen to $35.70 per year, or a monthly NOI of
$669, 375.
As shown, the 2008 DSCR just rounds to 1.25, but with a one standard deviation positive
shock the DSCR falls to 1.24 and falls to 1.21 with a two standard deviation shock in
energy prices. Thus, possible shocks lead to signi?cantly lower DSCR for the lender that,
when combined with lower property values due to the increased costs of energy, increase
the likelihood of default on the part of the borrower. The e?ects of this volatility are
exacerbated in 2009 when market rents fall and the NOI falls. The new average DSCR is
1.14 rounded but the one and two standard deviation shocks cause these values to fall to
1.12 and 1.11 respectively (accounting for the standard deviation around the e?ects of these
shocks, the DSCR falls to 1.00.) These negative outcomes are again associated with decreases
in the capitalized value of the NOI, which implies a lower market price (particularly because
capitalization rates rose to 8.35% in San Diego implying a new lower market price). Both
of these outcomes would increase the possibility of default and the need for the lender to
account for this risk in higher initial underwriting standards. Again, the obvious bene?ts of
undertaking energy retro?ts that could mitigate the high volatility of the seasonal in these
costs would clearly be advantageous if fairly priced ?nancing instruments could be designed
to ?nance such investments.
18
Table 4.4: Simulation results for the Average Debt Service Coverage Ratio for the Mean
total energy cost, for a one standard deviation positive shock to total energy cost, and a two
standard deviation shock to total energy costs at Q1 2008 and Q1 2009 Class A rents in San
Diego
Mean One Standard Two Standard
Deviation Deviation
Q1 2008 Rents 1.25 1.24 1.21
Std. Dev. (.007) (.008) (.011)
Q1 2009 Rents 1.14 1.12 1.11
Std. Dev. (.007) (.008) (.011)
Grubb & Ellis, O?ce Market Trends
We also simulate the price of our representative 225,000 square foot San Diego o?ce
building over a ?ve year investment horizon. This simulation is meant to be illustrative of
the importance of adding the stochastic dimension of energy costs into a valuation analysis.
For our analysis we again assume a gross leasing structure in which the level of gross rent
su?cient to achieve the underwritten net operating income of $8,858,870.00 at the CEUS
average energy cost for this type of building given its square footage and use by tenants in
?nancial, insurance, and real estate economic activities. We then simulate the distribution
of the realized net operating income of the representative San Diego property assuming that
NOI varies by the net realized energy cost relative to the CEUS underwritten energy level.
9
We discount using the term structure reported in Appendix I and we capitalize the month
sixty NOI using the observed end of year 2008 cap rate in San Diego of 6.7%. As shown in
4.7, we ?nd under our assumptions that the price of the San Diego building fall below its
underwritten price of $127,100,000.00 more than 10% of the simulated draws. This suggests
that there is a more than 10% chance of default on the mortgage loan on the building
assuming that the borrower would default on the loan when the loan-to-value ratio exceeds
one from below. Obviously, more realism could be added to this simulation, however, given
the relative accuracy of our match to the CEUS initial operating cost of the San Diego
building, we believe that, as structured, our simulation is indicative of the potential risk to
lender’s mortgage holdings that may arise from energy cost risk that is not underwritten
into the mortgage contract structure.
5 Summary and Conclusion
U.S. buildings consume almost 39% of the total U.S. energy usage, and appear ine?cient
compared to the available technology and the e?ciency of European structures. Govern-
ment building codes, disclosure requirements, and ?scal subsidies have been the primary
9
More realism could be added here using stochastic rents that are perhaps correlated with the energy
costs
19
Figure 4.7: Simulated Value of the Representative San Diego O?ce Building on April 2008
Assuming a Five Year Investment Horizon, CEUS Average Energy Cost for this building to
realized Stochastic Electricity and Natural Gas Costs to Compute the Realized Net Operating
Income Over the Holding Period.
0
50
100
150
200
250
300
instruments used to stimulate the adoption of energy-e?cient technology in both the U.S.
and Europe. While these instruments have achieved important successes, they also have
signi?cant limitations.
In this paper, we instead look at market mechanisms to stimulate voluntary private sector
energy-e?cient investments. We focus on a particular market failure, namely the absence of
energy e?ciency as an input to the underwriting decision for mortgage loans on commercial
properties. In principal, rising and volatile energy prices can be as important a source of
mortgage defaults as the standard inputs used in commercial mortgage underwriting.
We argue that the failure of commercial mortgage lenders to take energy-e?ciency into
account is primarily the result of insu?cient information to connect high and volatile energy
prices to mortgage default. Current commercial mortgage underwriting uses the loan to value
ratio (LTVR) and debt service coverage ratio (DSCR) to account for the macroeconomic and
housing market sources of market default. We propose a new energy e?cient score (EES)
and an energy volatility score (EVS) to account for the impact of energy prices on mortgage
default.
We carry out the ?rst step in the development of EES and EVS measures by creating
a dynamic model to compute the impact of stochastic energy prices on the net operating
income (NOI) of a representative building. We use a Monte Carlo methodology to provide
quantitative measures –both expected values and the range of variability–of the resulting
changes in NOI. Our preliminary results indicate that energy price volatility is a previously
unrecognized and signi?cant source of potential commercial mortgage default.
20
A
p
p
e
n
d
i
x
I
:
A
s
s
u
m
e
d
M
a
r
k
e
t
C
o
n
d
i
t
i
o
n
s
A
p
r
i
l
2
0
0
8
M
o
n
t
h
l
y
P
o
w
e
r
&
N
a
t
u
r
a
l
G
a
s
F
u
t
u
r
e
s
a
n
d
E
u
r
o
p
e
a
n
c
a
l
l
s
o
n
f
u
t
u
r
e
s
p
r
i
c
e
s
T
h
e
z
e
r
o
-
p
e
r
i
o
d
f
u
t
u
r
e
s
p
r
i
c
e
s
a
r
e
m
o
n
t
h
l
y
b
l
o
c
k
p
r
i
c
e
s
f
o
r
d
e
l
i
v
e
r
y
i
n
t
h
e
c
u
r
r
e
n
t
m
o
n
t
h
.
T
h
e
u
n
d
e
r
l
y
i
n
g
f
o
r
t
h
e
f
u
t
u
r
e
s
a
r
e
m
o
n
t
h
l
y
?
o
w
s
o
f
p
o
w
e
r
a
n
d
o
f
g
a
s
o
v
e
r
t
h
e
d
e
l
i
v
e
r
y
m
o
n
t
h
.
T
h
e
u
n
d
e
r
l
y
i
n
g
f
o
r
p
o
w
e
r
a
n
d
g
a
s
c
a
l
l
o
p
t
i
o
n
s
i
s
t
h
e
f
u
t
u
r
e
s
p
r
i
c
e
w
h
e
r
e
o
p
t
i
o
n
e
x
p
i
r
y
–
f
u
t
u
r
e
d
e
l
i
v
e
r
y
.
A
l
l
o
f
t
h
e
c
a
l
l
s
a
r
e
s
t
r
u
c
k
a
t
-
t
h
e
-
m
o
n
e
y
-
f
o
r
w
a
r
d
.
(
S
o
u
r
c
e
:
2
0
0
8
F
i
n
a
n
c
i
a
l
E
n
g
i
n
e
e
r
i
n
g
C
a
s
e
C
o
m
p
e
t
i
t
i
o
n
.
F
i
l
l
m
o
r
e
I
n
v
e
s
t
o
r
s
L
L
P
,
a
u
t
h
o
r
e
d
b
y
D
u
a
n
e
S
e
p
p
i
,
C
a
r
n
e
g
i
e
M
e
l
l
o
n
U
n
i
v
e
r
s
i
t
y
.
D
a
t
a
s
o
u
r
c
e
s
a
r
e
t
h
e
E
l
e
c
t
r
i
c
i
t
y
R
e
l
i
a
b
i
l
i
t
y
C
o
u
n
c
i
l
o
f
t
h
e
M
i
d
-
A
t
l
a
n
t
i
c
(
E
R
M
C
M
A
)
a
n
d
B
l
o
o
m
b
e
r
g
,
H
e
n
r
y
H
u
b
s
p
o
t
a
n
d
f
u
t
u
r
e
s
g
a
s
p
r
i
c
e
s
.
A
p
r
.
M
a
y
J
u
n
e
J
u
l
y
A
u
g
.
S
e
p
t
.
O
c
t
.
N
o
v
.
D
e
c
.
J
a
n
.
F
e
b
.
M
a
r
.
A
p
r
i
l
2
0
0
8
2
0
0
8
2
0
0
8
2
0
0
8
2
0
0
8
2
0
0
8
2
0
0
8
2
0
0
8
2
0
0
8
2
0
0
9
2
0
0
9
2
0
0
9
2
0
0
9
U
.
S
.
T
r
e
a
s
u
r
y
T
e
r
m
S
t
r
u
c
t
u
r
e
z
e
r
o
y
i
e
l
d
s
0
.
0
3
0
0
0
.
0
3
2
0
0
.
0
3
2
0
0
.
0
3
1
5
0
.
0
3
2
0
0
.
0
3
2
0
0
.
0
3
2
0
0
.
0
3
3
0
0
.
0
3
3
0
0
.
0
3
3
0
0
.
0
3
3
0
0
.
0
3
3
0
z
e
r
o
p
r
i
c
e
s
$
0
.
9
9
7
5
$
0
.
9
9
4
7
$
0
.
9
9
2
0
$
0
.
9
8
9
6
$
0
.
9
8
6
8
$
0
.
9
8
4
1
$
0
.
9
8
1
5
$
0
.
9
7
8
2
$
0
.
9
7
5
6
$
0
.
9
7
2
9
$
0
.
9
7
0
2
$
0
.
9
6
7
5
E
l
e
c
t
r
i
c
i
t
y
P
r
i
c
e
s
(
p
e
r
M
W
h
)
f
u
t
u
r
e
s
p
r
i
c
e
s
$
8
2
.
0
0
$
8
8
.
1
5
$
9
8
.
3
5
$
1
1
6
.
0
0
$
1
2
4
.
4
0
$
9
0
.
5
0
$
8
6
.
2
5
$
8
0
.
6
5
$
8
6
.
0
5
$
9
6
.
0
0
$
9
7
.
2
0
$
9
1
.
7
5
$
8
0
.
8
0
d
o
l
l
a
r
p
r
i
c
e
o
f
E
u
r
o
p
e
a
n
c
a
l
l
s
$
3
.
5
4
$
5
.
8
9
$
9
.
6
2
$
1
1
.
8
8
$
8
.
9
5
$
1
0
.
0
2
$
9
.
8
5
$
1
0
.
6
5
$
1
2
.
5
6
$
1
3
.
3
6
$
1
3
.
1
8
$
9
.
3
2
N
a
t
u
r
a
l
G
a
s
P
r
i
c
e
s
(
p
e
r
m
m
B
T
U
)
f
u
t
u
r
e
s
p
r
i
c
e
s
$
9
.
6
0
$
9
.
7
9
$
9
.
8
8
$
9
.
9
7
$
1
0
.
0
3
$
1
0
.
0
5
$
1
0
.
1
2
$
1
0
.
4
0
$
1
0
.
7
5
$
1
0
.
9
7
$
1
0
.
9
5
$
1
0
.
7
1
$
9
.
7
5
d
o
l
l
a
r
p
r
i
c
e
o
f
E
u
r
o
p
e
a
n
c
a
l
l
s
$
0
.
3
9
$
0
.
5
8
$
0
.
7
3
$
0
.
8
9
$
1
.
0
7
$
1
.
2
6
$
1
.
3
3
$
1
.
4
3
$
1
.
5
8
$
1
.
7
3
$
1
.
7
7
$
1
.
4
2
21
References
[1] Benth, F.E. and Benth J.S., and Koekebakker, S. (2008), Stochastic Modelling of Elec-
tricity and Related Markets, London, World Scienti?c, Advanced Series on Statistical
Science & Applied Probability, Volume II.
[2] Benth, F. E., Cartea, A., and Kiesel, R. (2006). Pricing forward contracts in power
markets by the certainty equivalence principle: explaining the sign of the market risk
price premium, Forthcoming Journal of Banking Finance.
[3] Benth, F. E., Koekebakker, S., and Ollmar, F. (2007), Extracting and applying smooth
forward curves from average-based commodity contracts with seasonal variation, Journal
of Derivatives, Fall 15(1), pp. 52-66.
[4] Bessembinder, H., and Lemmon, M. (2002), Equilibrium pricing and optimal hedging in
electricity forward markets, Journal of Finance, 57, pp. 1347-1382.
[5] Cartea, A. and Figuerosa, M. G. (2005), Pricing in electricity markets: a mean reversion
jump di?usion model with seasonality, Applied Mathematical Finance, 12 (4), pp. 313-
335.
[6] Deng, S. (2000), Stochastic Models of Energy commodity prices and their applications:
mean-reversion with jumps and spikes, Working paper, PWP-073 of the Program on
Workable Energy Regulation (POWER), University of California Energy Institute.
[7] Eberlein, E. and Stahl, G. (2003), Both sides of a fence: A statistical and regulatory
view of electricity risk, Energy & Power Risk Management, 8, pp. 34-38.
[8] Eichholtz, Piet, Nils Kok, and John Quigley (2009), “Doing Well by Doing Good: Green
O?ce Buildings,” Working Paper No W08-001, Fisher Center for Real Estate and Urban
Economics, University of California, Berkeley.
[9] Geman, H. and Roncoroni, A. (2006), Understanding the ?ne structure of electricity
prices, Journal of Business, 79 (3), pp. 1225-1261.
[10] International Energy Agency (IEA) (2009), Energy Prices and Taxes.
[11] International Energy Agency (IEA) (2008), Energy E?ciency Requirements in Building
Code, Energy E?ciency Policies for New Buildings.
[12] International Energy Agency (IEA) (2004), 30 Years of Energy Use in IEA Countries.
Karlis, D. and Lillestl, J. (2004), Bayesian estimation of NIG models via Markov chain
Monte Carlo methods, Applied Stochastic Models in Business and Industry, 20 (4),
323-338.
[13] McKinsey Global Institute (2007), Curbing Global Energy Demand Growth: The Energy
Productivity Opportunity.
[14] Rand (2009), Improving the Energy Performance of Buildings.
22
[15] Schwartz, E. S. (1997), The stochastic behavior of commodity prices: Implications for
valuation and hedging, Journal of Finance, LII, 3, pp. 923-973.
[16] U.S. Department of Energy (2009), Buildings Energy Data Book.
[17] Weron, R. (2000), Energy Risk Management, Physica A 285, pp. 127-134.
[18] Weron, R. (2004), Computationally intensive Value at Risk calculations, In Handbook
of Computational Statistics, J.E. Gentle, W. Hrdle, and Y. Mori (eds). Springer, Berlin,
pp. 911-950.
[19] Weron, R. (2005), Heavy tails and electricity prices, The Deustsche Bundesbanks 2005
Annual Fall Conference, Eltville.
[20] Weron, R. (2006), Modeling and forecasting electricity loads and prices: A statistical
approach, Chichester, Wiley Finance, John Wiley & Sons.
[21] Weron, R., Kozlowska, B., and Nowicka-Zagrajeck, J. (2001), Modeling electricity loads
in California: A continuous time approach, Physica A 299, pp. 344-350.
23
doc_850285196.pdf