Valuing the Reliability of the Electrical Power Infrastructure: A Two-stage Hedonic Approach

Abstract

The reliability of electrical power supply is amongst the conditions that inform house purchase decisions in all urban areas. Reliability depends in part on the conditions of the power generation and distribution infrastructures involved, and in part on environmental conditions. Its value to homeowners may be capitalised into the value of the house. In this paper, a hedonic pricing approach is used to estimate the capitalised value of the reliability offered by distribution infrastructures and the environmental conditions with which they interact in Phoenix, Arizona. A first stage estimates the impact of infrastructure and environmental conditions on reliability. In a second stage, the capitalised value of reliability from the marginal willingness to pay for reliability revealed by house purchase decisions is estimated and used to infer the value of both infrastructural characteristics and environmental conditions.

1. Introduction

The reliability of power infrastructures depends on three sets of conditions. One is the level of demand for the services provided by those infrastructures. A second is the engineered characteristics of the infrastructure that supplies those services. A third is the environmental conditions in which the infrastructure operates. Reliability can change with a change in either the capacity of an infrastructure to meet demand over existing environmental conditions, or with a change in environmental conditions. In this paper, we estimate the value of the reliability of urban residential electric power supply, and the infrastructural and environmental conditions that determine that reliability in Phoenix, Arizona.

Reliability is measured by the number and duration of unscheduled power outages experienced by consumers. In principle, reliability events are any deviations from a pure 60-cycle per second alternating current supply at 120 volts for residential customers or 480 volts for many commercial and industrial customers. In practice, however, reliability events are experienced as interruptions (incidents where voltage falls to zero) captured in any of the main reliability indices: The System Average Interruption Duration Index (SAIDI), the System Average Interruption Frequency Index (SAIFI), or the Momentary Average Interruption Frequency Index (MAIFI) (IEEE, 1995). The average annual count of unscheduled power outages over the period 2002–05 was just over 11 in Phoenix, Arizona. The reliability of residential power supply in Phoenix and other cities in the western US is slightly lower than in cities in Europe (Carlsson and Martinsson, 2007). The mean MAIFI in 11 western and northern European countries is 2.3 (CA Technologies, 2011). In contrast, the comparable measure for Pacificorp, a power utility in the western US, was 10–11 in California and 5–6 in Oregon (Oregon Public Utility Commission, 2010; California Public Utilities Commission, 2011). Developing countries experience significantly worse electrical reliability with SAIFIs up to 40 in some areas (Central Electricity Authority of India, 2007), with consumers having to carry the cost of reserve generating capacity or alternative energy sources.

Power interruptions impose both direct and indirect costs on householders. A first approximation of the direct cost of a power interruption can be obtained by multiplying the average load per unit of time by the average market price of that load per unit of time by the duration of that interruption. This is generally referred to as the expected cost per unserved kWh. An alternative approach is to estimate damage costs, often through surveys that elicit statements of expected costs associated with different types of interruption (Lawton et al., 2003; Sullivan et al., 2009). A study of expected damage costs in the wake of the major blackout in the north-eastern US and Canada in 2003, for example, identified costs to three categories of consumers: residential customers, commercial and industrial customers, and what the authors termed ‘wider infrastructure’—the capacity of municipal, state and federal authorities to maintain essential public services (LaCommare and Eto, 2004). The approach tends to minimise the cost of unreliability to residential consumers, this particular study concluding that residential users accounted for only 2 per cent of the total annual cost of outages.

The expected damage approach implies that, if an asset has the capacity to reduce the likelihood and severity of economic damage, the value of that capacity may be measured by the associated reduction in expected damage. It has, for example, been used in connection with human health (Cameron and Trivedi, 1998), airline safety (Rose, 1990), highway safety (Michener and Tighe, 1992), drug safety (Olson, 2004) and environmental hazard protection (Barbier, 2008). It does, however, focus on the direct costs of outage events. Indirect costs include loss of public services such as transport, communications, emergency response and security (Chang et al., 2007). Many of these costs are neighbourhood costs, affecting all those in an area suffering an outage. To capture the value of these requires a different approach.

Since interruptions are harmful, householders may be expected to be willing to incur some cost to improve the reliability of service. Where reliability depends on neighbourhood characteristics such as the type of infrastructure, environmental conditions or proximity to sites that either increase or decrease the risk of outages, we would expect the value of reliability to be capitalised into the value of the property. In this paper, we estimate the capitalised value of the reliability of residential electric power supplies in Phoenix, Arizona, using a hedonic house price model. Our aim is to demonstrate the effect of a particular approach to the estimation of the value of the reliability of power infrastructures. While Phoenix has some characteristics that might make it different from other cities, we are less interested in these characteristics than in the effectiveness of the approach relative to the survey-based damage cost approach.

Hedonic modelling decomposes a marketable item into a number of attributes over which purchasers have preferences. By estimating a hedonic price function it is possible to infer purchasers’ marginal willingness to pay (MWTP) for each attribute. For instance, house prices can be used to infer the value of a public service by estimating the MWTP for that service, controlling for salient housing characteristics, neighbourhood characteristics and other environmental characteristics. We estimate the impact of infrastructure reliability on the value of the asset served by that infrastructure using a hedonic property value approach and treating the reliability of power distribution infrastructures as an attribute of the properties serviced by those infrastructures.

The hedonic pricing approach has been used to value a wide range of asset attributes (Heal et al., 2005; Mendelsohn and Olmstead, 2009). Many such attributes are direct properties of the asset, such as the size of a house, the materials it is constructed from, its internal configuration and so on. However, others include neighbourhood or ambient conditions. Amongst these are infrastructural attributes such as electricity, water, road, health, recreation and other infrastructures. The hedonic pricing approach has been also been used to value environmental phenomena such as air quality (Ridker and Henning, 1967; McPherson, 1992; Smith and Huang, 1995; Zabel and Kiel, 2000; Anselin and Lozano-Gracia, 2008), parks (More et al., 1988; Lockwood and Tracy, 1995; Lindsey and Knaap, 1999; Morancho, 2003; Salazar and Menendez, 2007; Troy and Grove, 2008), forests (Tyrväinen, 2001; Popoola and Ajewole, 2002; Kwak et al., 2003; Mansfield et al., 2005), noise (Clark, 2006), wetlands and water quality (Doss and Taff, 1996; Leggett and Bockstael, 2000; Mahan et al., 2000; Lupi et al., 2002; Boyer and Polasky, 2004; Tong et al., 2007) and hazardous waste (Deaton and Hoehn, 2004; Greenstone and Gallagher, 2008). Researchers have also applied the method to value pest control (Jetter and Paine, 2004), recreation (Jim and Chen, 2006) and aesthetics (Kulshrestha and Gillies, 1993).

While the hedonic method has been applied to estimate the effect of high-voltage overhead (Sims and Dent, 2005) and underground power lines (McNair and Abelson, 2010) on property values, the majority of studies on the WTP for electrical reliability use survey methods such as contingent valuation and choice experiments (Beenstock et al., 1998; Goett et al., 2000; Carlsson and Martinsson, 2007, 2008; Abdullah and Mariel, 2010). To our knowledge, the hedonic approach has not previously been applied to estimate the MWTP for electrical power reliability.

We do not expect house buyers to know the reliability of electric power supplies directly. However, we do expect them to be able to observe neighbourhood characteristics that affect reliability. They can, for example, observe the age of feeder lines, whether they are trussed overhead and whether they are vulnerable to trees or birds. Since these are observable by the house buyer, we can estimate their impact on the value of houses. The following section details the data and methods involved in the analysis. A third section provides our results and a final section offers our conclusions.

2. Data and Methods

We aim to estimate the impact of the reliability of residential electric power supplies on the value of residential properties. To calculate the MWTP for reliability through house purchase decisions, we specify and estimate a hedonic price function of the form:

p_{i} = f (h_{i}, a_{i}, s_{i}, β, µ, λ) + ε_{i}

(1)

where, $p_{i}$ is the price of the ith property sold during the reference period; $h_{i}$ is a vector of house characteristics; $a_{i}$ is a vector of environmental or ambient conditions; $s_{i}$ is a vector of power infrastructural conditions; and $ε_{i}$ is an error term. The remaining terms are vectors of estimated coefficients on house characteristics, ambient conditions and infrastructure conditions. This implies an underlying household utility function of the form:

U_{j} = U (x_{j}, h_{j}, a_{j}, s_{j})

(2)

where, $x_{j}$ is a vector of all other commodities consumed by the jth household.

Household preferences are assumed to be weakly separable in the set of housing and housing-related services and all other commodities. The assumption allows us to estimate demand for housing services independent of the prices of other commodities—since demand for housing services depends only on service ‘prices’ and total expenditure.

The study area comprises a central transect within the municipal boundaries of the City of Phoenix, Arizona, stretching from the Sky Harbor International Airport in the south to the Carefree Highway (SR 74) in the north. Phoenix accounts for about 35 per cent of the metropolitan area’s population and is the sixth-largest city in the US.¹ Phoenix represents a good case study for three reasons. First, there is significant variation in the reliability of electrical power supplies within the metropolitan area. Secondly, Phoenix has a unique environment. Phoenix is situated in a desert and yet has a diverse urban-biophysical environment as well as a varied electrical distribution system. Roughly half of Phoenix is served by overhead lines and half by underground lines. Finally, over the past 40 years, Phoenix has been one of the most rapidly growing cities in the US and this is reflected in the frequency of house sales.

We use data on 2005 detached single-family housing sales in Phoenix, Arizona, obtained from the Maricopa County Assessor’s Office (MCAO). The data include a number of house and parcel characteristics such as price (dollars), house size (square feet), lot size (square feet), roof type (asphalt or tile), the presence/absence of a garage, the presence/absence of a pool and the age of the house. For the ambient socioeconomic conditions, data on median household income and median age of residents were collected from the 2000 US Census Bureau. Since tracts were the smallest spatial demographic unit provided, each housing parcel sale within each tract was assigned attribute information according to its corresponding tract’s demographic characteristics.

Amongst the direct environmental determinants of house prices, we considered two proximity variables. Distance to the city centre (feet) measures location with respect to Sky Harbor Airport and downtown Phoenix. Distance to the nearest natural desert (feet) is a measure of accessibility to the Sonoran desert or desert remnants, considered to be one of the main attractions of property in central Arizona. The spatial separations between the centroid (geometric centre) of sold parcels and the centroid of features of interest were calculated through ArcGIS using Euclidean (straight-line) distances in feet. We also considered one ambient environmental variable, vegetation abundance measured by the Soil Adjusted Vegetation Index (SAVI) from a Landsat Thematic Mapper (ETM) image obtained via the Central Arizona–Phoenix Long Term Ecological Research (CAP-LTER) project.

We expect vegetation to be a desirable environmental amenity through the positive effects it has on, for example, ambient temperature. As with proximity to natural desert, however, we test the hypothesis that vegetation will also have negative indirect effects through its interaction with the electrical power distribution system (described later). Vegetation is therefore a component of two separate variables: vegetation abundance and an interaction term between vegetation and overhead lines. We expected abundance to have a positive effect on house prices (for example, aesthetic, cooling). We expected the interaction term to have no effect in areas served by underground cables, but a negative effect in areas served by overhead lines.

Our chosen measure of the reliability of the power distribution infrastructure is accidental outages due to a failure of the distribution infrastructure (the system that delivers power from sub-stations to end users) as opposed to the power generation infrastructure or high-voltage bulk transmission lines. Failures of the distribution system explain most of the interruptions experienced by electricity consumers (Pahwa, 2004). Failures of the distribution system are also most closely linked to environmental conditions. The Phoenix metropolitan area’s electrical power is supplied by two major sources, Arizona Public Service (APS) and Salt River Project (SRP). Infrastructure data (power line location and type—overhead or underground) were obtained from APS. The outage data were also provided by APS; hence the areas within the municipal boundaries of the City of Phoenix serviced by SRP were excluded. The outage data included the number of events by feeder ID, their proximate cause, duration and the number of customers affected for the period 2002–05. We grouped the causes of outages into the following categories: scheduled outages, accidental outages and environmental outages (a subset of accidental outages). Outages can occur when there is a failure in any part of the power infrastructure including the generation, transmission, sub-transmission and distribution systems. We consider all unscheduled outages including both momentary interruptions that persist no longer than a few seconds and blackout incidents that persist longer than several minutes.

Since we are interested in the number of outages per house by feeder, each housing sale was assigned to its nearest feeder line to account for the number of outages at each sale location. The total number of unscheduled outages that are clearly identified with failures of the distribution system is our proxy for the reliability of the system. The data on outages between 2002 and 2005 corresponding to each housing sale in 2005 are mapped in Figure 1.

Figure 1.

Unscheduled power outages in Phoenix, Arizona, 2002–05.

The average count of unscheduled power outages per 2005 house sale over the period 2002–05 was just over 11 outages per year. The distribution around the mean was also quite sizeable with observed outages ranging from 1 to 130 between 2002 and 2005. This reflects the considerable geographical variation in both the conditions of the electrical infrastructure and the urban biophysical environment. Since most of the feeder lines north of Dunlap Avenue and Doubletree Ranch Road tend to be buried below ground, they are not affected by wind, rain or dust. Furthermore, vegetation abundance and the age of infrastructure generally increase with increasing proximity to downtown Phoenix.

An issue in estimating MWTP for electrical reliability is that house buyers do not necessarily have perfect knowledge about the electrical reliability of a house up for sale. This is because electrical reliability is not directly observable and is not amongst the data disclosed to buyers. However, since electrical reliability depends on both the infrastructure and the environment with which it interacts, and since these factors are directly observable by the house buyer (for example, whether the power distribution infrastructure is above or below ground, or overhead lines are exposed to trees), we are able to infer MWTP for reliability. Specifically, we treat unscheduled outages as an endogenous variable and interactions between the observable infrastructural and environmental conditions as instruments. We hypothesise that interactions between infrastructures and environmental conditions do not directly affect house prices, but do affect the reliability of those infrastructures. We further hypothesise that the reliability of infrastructures is an ambient or neighbourhood characteristic that may be capitalised into the value of residential properties.

Unscheduled environmental outages generally depend on environmental events, environmental conditions, infrastructural conditions and interactions between each of these (Brown, 2002). The pertinent interactions for Phoenix are illustrated in Figure 2.

Figure 2.

Outage diagram of the interactions between environmental events, environmental conditions and infrastructural conditions.

Unscheduled outages are hypothesised to depend on type of infrastructure (location above or below ground), electricity demand, vegetation abundance, proximity to desert (a proxy for wind-blown sand/dust) and the interaction between vegetation, desert and overhead lines. The effect of vegetation on power distribution reliability ranges from brief contact (if a branch bridges two conductors) to tree fall that can severely impair electrical equipment (Kuntz et al., 2002; Guikema et al., 2006). For example, growing branches can intrude upon conductors; animals may move branches into conductors; and dead trees may fall, interfering with equipment (Radmer et al., 2002). Tree-to-line contact is, however, more likely to occur if combined with a severe weather event. Tornados, hurricanes and major storms are accompanied by high wind speeds that cause branches to sway and, in the worst case, cause trees to topple onto overhead lines (Zhou et al., 2006). It has been estimated that 20–50 per cent of unscheduled outages are caused by vegetation in the Northeast US (Simpson and van Bossuyt, 1996). Additionally, the natural desert environment generates wind-borne dust and sand that interfere with insulators, reducing effective distribution and potentially resulting in flashover outages (Hamza et al., 2002; Sima et al., 2010). The desert environment effect ranges from dust deposition to the worst case of sand storms, when strong winds lead to sand particle saltation at high speeds occasionally leading to suspension (Hamza et al., 2002).

House sales greater than $749 999 and less than $60 001 were eliminated from the analysis to avoid undue outlier influence on the result, leaving 6061 observations. The variables and descriptive statistics are summarised in Table 1 and Figure 3.

Table 1.

Names, descriptions and statistics of variables (N = 6061)

Name	Description	Mean	S.D.	Minimum	Maximum
SALE	House price	255 113	126 889.9	61 319	745 000
ASPH	Asphalt roof	0.52	0.50	0	1
TILE	Tile roof	0.02	0.14	0	1
PORT	Garage	0.91	0.28	0	1
POOL	Pool	0.35	0.48	0	1
CAGE	Construction age	27.8	18.84	1	105
AREA	Lot size (square feet)	8 425	4 995.30	2 016	82 823
SQFT	House size (square feet)	1682	581.33	445	5360
SQFTTEMP*	House size (square feet) and August minimum temperature interaction	35 310	12 182.9	9 790	112 600
VEG	Vegetation abundance	39 297	4 765.17	22 013	64 659
VEGOH*	Vegetation abundance and overhead line interaction	10 680	18 120.69	0	59 950
BIRD	Bird abundance	130	18.31	78	168
PHX	Distance to central Phoenix (feet)	66 014	26 324.91	1 906	125 204
DES	Distance to nearest desert area (feet)	5 727.76	4 374.16	34.73	20 672.30
DESOH*	Distance to nearest desert area (feet) and overhead line interaction	2 203	4 591.66	0	20 672
AGE	Median resident age	34.11	4.63	22.5	48.7
INC	Median household income	53 055	16 994.54	10 607	110 924
BIRDVEGOH	Bird abundance, vegetation abundance and overhead line interaction	1570 263	2672 269	0	9 315 093
ART	Distance from major arterial (feet)	1 031.42	775.93	0.58	6 924.04
OH	Percentage of house sales in tract supplied by overhead lines	0.26	0.36	0	1
OUT	Number of unscheduled outages	44.94	26.82	1	130
LAKEOH*	Distance to nearest lake (feet) and overhead line interaction	1 129	2 162.22	0	15 400

Figure 3.

House prices, power infrastructure type and infrastructure age in Phoenix, Arizona, 2005.

The reliability model employed to identify candidate instrumental variables used in the two stage least squares hedonic model took the following form

y_{i} = f (w_{i}, c_{i}, z_{i}) + ε

(3)

where, $y_{i}$ is the number of outages at location i; $w_{i}$ is a proxy for energy demand at location i; $c_{i}$ is a vector of infrastructural conditions (such as feeder type) at location i; and $z_{i}$ is a vector of environmental conditions (such as vegetation abundance, desert proximity) at location i (see Table 2).

Table 2.

Outage determinants and estimated coefficients (N = 6061)

	Coefficient	T-statistic
(Constant)	−0.285	−0.313
SQFTTEMP*	4.863E-5	7.742
OH	1.168E+1	28.727
ART	−3.225E-4	−3.510
BIRD	1.362E-2	2.564
VEG	3.704E-6	0.240
BIRDVEGOH	1.252E-6	16.688
DESOH*	−3.835E-4	−14.145
LAKEOH*	−7.707E-4	−12.002
Adjusted R²	0.431

Energy demand was approximated by multiplying housing square footage by ambient temperature. Temperature data represented by August minimum degrees in Celsius were obtained through CAP-LTER. These data were derived from spatial interpolation of daily temperature data from 55 meteorological sensors from different sources including the Flood Control District of Maricopa County (ALERT), the National Weather Service (NWS), the Arizona Meteorological Network (AZMET) and the Phoenix Real-time Instrumentation for Surface Meteorological Studies (PRISMS) Network. Daily measurements were aggregated to bi-weekly periods. The variation in climate is certainly not as great as landscape, vegetation and ruggedness, but it does vary. Spatial heterogeneity of climate is a topic that is heavily researched in the Phoenix metropolitan area, as highlighted by several articles on microclimate differences within the city (Harlan et al., 2008; Buyantuyev and Wu, 2010; Grossman-Clark et al., 2010; Chow et al., 2012). We then tested each of these variables as instruments, using the test statistics reported in Table 2 to determine which instruments explained reliability, but did not explain house prices. Since some of the variables that explained reliability did explain house prices, we excluded those as instrumental variables. We ended up using two instruments in the first stage—namely, the interaction between vegetation and overhead lines and the interaction between desert conditions (sand/dust) and overhead lines—and the 13 variables described in Table 3 in the second stage.

Table 3.

Estimated coefficients and MTWP for a 1per cent change in attributes (N = 6061)

	Coefficient	T-statistic	MWTP for 1per cent change
(Constant)	10.732	69.65	—
ASPH	−1.342E-1	−13.25	−$160
TILE	7.711E-2	2.77	$3
PORT	1.113E-1	7.70	$232
POOL	4.861E-2	5.89	$38
AREA	1.240E-5	11.78	$238
SQFT	3.918E-4	46.78	$1504
AGE	2.557E-2	16.60	$1991
INC	3.79E-6	8.69	$459
lnPHX	−3.078E-2	−2.60	−$770
lnDES	2.714E-2	5.34	$516
CAGEOH*	1.875E-3	4.94	$55
VEG	2.90E-6	3.74	$260
OUT	−6.862E-3	−7.80	−$704
Adjusted R²	0.655	—	—
Durbin	68.132	p < 0.0001	—
Wu-Hausman	68.736	p < 0.0001	—
Sargan	2.529	p = 0.112	—
Basmann	2.524	p = 0.112	—
Cragg–Donald F	83.646	—	—

3. Results

Coefficients for the hedonic property model are reported in Table 3 along with corresponding MWTP estimates and t-statistics. The resulting model fit is good (adjusted R² = 0.655). Our selection of an instrumental variables approach to the estimation of impact of power supply reliability on house prices is validated. Both Durbin and Wu–Hausman tests (Wu, 1973; Hausman, 1978) rejected the null hypothesis that power supply reliability is exogenous (p < 0.0001) and the Cragg and Donald (1993) and Stock and Yogo (2005) tests confirmed that the instrumental variables are not weak. The Sargan (1958) and Basmann (1960) tests for overidentification were not significant at the 10 per cent level, indicating the instruments were not strongly correlated with the residuals. Specifically, the interaction between vegetation and overhead lines and the interaction between proximity to desert and overhead lines explain electrical reliability in Phoenix, but do not directly explain house prices.

The signs of the coefficients on house characteristics were all as expected, with house size being dominant. MWTP increased with tile roofs, lot size, garages and pools. By contrast, an asphalt roof tended to decrease the price people were willing to pay. The signs of the coefficients on socioeconomic household conditions and on ambient neighbourhood and environmental conditions were, for the most part, also as expected. An increase in the average age of residents and the median household income increased house prices. Houses farther away from downtown Phoenix tended to be less expensive. Houses with greater amounts of vegetation tended to be more expensive because vegetation in itself is a desirable amenity, despite the fact that vegetation can negatively interact with overhead feeder lines. The coefficient for distance from the nearest natural desert area was positive, meaning that house prices tended to increase as their distance to desert areas increased. This was not expected, but houses close to natural desert also tend to be in newer developments built during the mid 2000s housing boom (the houses being relatively inexpensive and generally perceived to be of poorer quality than more established houses).

We found that the relative reliability of the power distribution system positively affected house prices. Controlling for other factors, the less reliable the power supply in a neighbourhood, the less people were willing to pay for houses in that neighbourhood. Specifically, we found that a 1 per cent reduction in the number of unscheduled power outages due to different environmental and infrastructural factors would increase residential property values by an average of $704. This implies that the capitalised value of a 10 per cent increase in the reliability of the power infrastructure in the city of Phoenix was potentially as high as $6.1 billion considering all housing units across Phoenix (based on the 2000 census).

Using the same estimate of MWTP for the reliability of the power distribution system and the reliability model (equation (3)), we then inferred the MWTP for the contribution of each of the factors affecting reliability. The main factor determining the reliability of the power infrastructure is whether the feeder lines are buried below or trussed above ground. In Phoenix, electrical service through underground power lines reduces the number of unscheduled power outages by nearly 26 per cent. Overhead lines are vulnerable to hazardous environmental events, weather events, vegetation and sand saltation. Since we tested the percentage of overhead feeder lines in each census tract, the MWTP for a 10 per cent increase in underground lines per census tract, calculated in terms of their effect on the reliability of power supply, is $1830 per house.

We found that interactions between overhead lines, vegetation abundance and proximity to natural desert affect the number of unscheduled outages. Interaction between overhead power lines and proximity to the desert was both strongly negative and highly significant. Vegetation abundance was strongly and positively related to outages, but the interaction between vegetation and overhead power lines, although positive, was not significant at the 5 per cent level.

4. Conclusions

While our findings build on existing studies of WTP for electrical reliability and underground lines, we adopted a different approach from that commonly used in the literature. We considered the impact of interactions between power infrastructures and environmental conditions for the reliability of residential electric power supplies. The vulnerability of overhead power lines increases in the presence of certain environmental conditions. For example, the presence of abundant vegetation is a significant source of overhead distribution interference. Additionally, houses fed by overhead lines that are closer to natural desert areas are more exposed to adverse natural desert conditions such as excessive sand/dust that can interfere with distribution lines through either deposition or suspension. Overall, the more exposed infrastructure lines are to hazardous conditions, the less reliable the infrastructure is, and the more people are willing to pay to avoid this exposure.

Although we have focused on the residential sector, a similar analysis could be done for commercial and industrial properties. The central point is that, just as many environmental characteristics are capitalised into property values, so are many public services. By estimating the impact of changes in services on asset values, it is possible to infer the value those services have to different classes of consumer. The evidence from Phoenix, at least, indicates that consumers attach significant weight to the capacity of the power infrastructure to operate reliably over a range of environmental conditions.

The magnitude of the capitalised value of the reliability of different infrastructures at least raises a question about the findings of LaCommare and Eto (2004), who argued that damage costs of outages to residential consumers were very small relative to commercial and industrial consumers. Comparison with the findings of Carlsson and Martinsson (2007), who use a contingent valuation approach to estimate how much people are willing to pay to avoid a power outage in Sweden, suggests that willingness to pay for a marginal improvement in reliability may be expected to diminish as reliability increases. However, at current levels of reliability in our study area, it is still substantial. The significant capitalised value of the infrastructure reliability indicates that homeowners are currently able to capture many of the benefits of improving the quality of electrical service. These findings are potentially important for electrical reliability planning. From a management perspective, this suggests that there may be alternative options for financing infrastructural improvements through private electricity providers and not the state. For example, because undergrounding electricity lines increases property values, utilities may provide packages to consumers in exchange for increases in electrical reliability. Since the value of housing assets reflects the reliability of power infrastructures, there may be scope for power utilities to enter into novel contractual arrangements with consumers over the cost of reliability-enhancing investments.

Footnotes

Notes

Funding Statement

This research was supported by the National Science Foundation project EFRI-RESIN: Sustainable Infrastructures for Energy and Water Supply (SINEWS) award number 0836046.

References

Abdullah

Mariel

(2010) Choice experiment study on the willingness to pay to improve electricity services, Energy Policy, 38, pp. 4570–4581.

Anselin

Lozano-Gracia

(2008) Errors in variables and spatial effects in hedonic house price models of ambient air quality, Empirical Economics, 34, pp. 5–34.

Barbier

E. B.

(2008) In the wake of tsunami: lessons learned from the household decision to replant mangroves in Thailand, Resource and Energy Economics, 30, pp. 229–249.

Basmann

(1960) On finite sample distributions of generalized classical linear identifiability test statistics, Journal of the American Statistical Association, 55, pp. 650–659.

Beenstock

Goldin

Haitovsky

(1998) Response bias in a conjoint analysis of power outages, Energy Economics, 20, pp. 135–156.

Boyer

Polasky

(2004) Valuing urban wetlands: a review of non-market valuation studies, Wetlands, 24, pp. 744–755.

Brown

R. E.

(2002) Electric Power Distribution Reliability. New York: Marcel Dekker.

Buyantuyev

(2010) Urban heat islands and landscape heterogeneity: linking spatiotemporal variations in surface temperatures to land-cover and socioeconomic patterns, Landscape Ecology, 25, pp. 17–33.

California Public Utilities Commission (2011)Reliability annual reports: MAIFI 1998–2008( http://www.cpuc.ca.gov/PUC/energy/ElectricSR/Reliability/annualreports/maifi_1998to2008.htm ; accessed 27 March 2011).

10.

Cameron

C. A.

Trivedi

(1998) Regression Analysis of Count Data. Cambridge: Cambridge University Press.

11.

Carlsson

Martinsson

(2007) Willingness to pay among Swedish households to avoid power outages: a random parameter Tobit model approach, Energy Journal, 28, pp. 75–89.

12.

Carlsson

Martinsson

(2008) Does it matter when a power outage occurs? A choice experiment study on the willingness to pay to avoid power outages, Energy Economics, 30, pp. 1232–1245.

13.

CA Technologies (2011) The avoidable cost of downtime: the impact of IT downtime on employee productivity (http://www.ca.com/~/media/Files/SupportingPieces/acd_report_110110.ashx; accessed 27 March 2011).

14.

Central Electricity Authority of India (2007) Reliability of power supply. Ministry of Power, Government of India.

15.

Chang

S. E.

McDaniels

T. L.

Mikawoz

Peterson

(2007) Infrastructure failure interdependencies in extreme events: power outage consequences in the 1998 ice storm, Natural Hazards, 41, pp. 337–358.

16.

Chow

W. T. L.

Chuang

W.-C.

Gober

(2012) Vulnerability to extreme heat in metropolitan Phoenix: spatial, temporal and demographic dimensions, The Professional Geographer, DOI:10.1080/00330124.2011.60022510.1080/00330124.2011.600225.

17.

Clark

D. E.

(2006) Externality effects on residential property values: the example of noise disamenities, Growth and Change, 37, pp. 460–488.

18.

Cragg

J. G.

Donald

S. G.

(1993) Testing identifiability and specification in instrumental variable models, Econometric Theory, 9, pp. 222–240.

19.

Deaton

B. J.

Hoehn

J. P.

(2004) Hedonic analysis of hazardous waste sites in the presence of other urban disamenities, Environmental Science & Policy, 7, pp. 499–508.

20.

Doss

C. R.

Taff

S. J.

(1996) The influence of wetland type and wetland proximity on residential property values, Journal of Agricultural and Resource Economics, 21, pp. 120–129.

21.

Goett

A. A.

Hudson

Train

K. E.

(2000) Customers’ choice among retail energy suppliers: the willingness-to-pay for service attributes, Energy Journal, 21, pp. 1–28.

22.

Greenstone

Gallagher

(2008) Does hazardous waste matter? Evidence from the housing market and the superfund program, Quarterly Journal of Economics, 123, pp. 951–1003.

23.

Grossman-Clarke

Zehnder

J. A.

Loridan

T. A.

Grimmond

S. C.

(2010) Contribution of land use changes to near surface air temperatures during recent summer extreme heat events in the Phoenix metropolitan area, Journal of Applied Meteorology and Climatology, 49, pp. 1649–1664.

24.

Guikema

S. D.

Davidson

R. A.

Liu

(2006) Statistical models of the effects of tree trimming on power system outages, IEEE Transactions on Power Delivery, 21, pp. 1549–1557.

25.

Hamza

A. S. H. A.

Abdelgawad

N. M. K.

Arafa

B. A.

(2002) Effect of desert environmental conditions on the flashover voltage of insulators, Energy Conversion and Management, 43, pp. 2437–2442.

26.

Harlan

S. L.

Brazel

Jenerette

G. D.

Jones

N. S.

. (2008) In the shade of affluence: the inequitable distribution of the urban heat island, Research in Social Problems and Public Policy, 15, pp. 173–202.

27.

Hausman

(1978) Specification tests in econometrics, Econometrica, 46, pp. 1251–1271.

28.

Heal

G. M.

Barbier

E. B.

Boyle

K. J.

Covich

A. P.

. (2005) Valuing Ecosystem Services: Toward Better Environmental Decision-making. Washington, DC: The National Academies Press.

29.

IEEE (Institute of Electrical and Electronics Engineers) (1995) IEEE Standard 1159-1995: recommended practice for monitoring power quality. IEEE, New York.

30.

Jetter

Paine

T. D.

(2004) Consumer preferences and willingness to pay for biological control in the urban landscape, Biological Control, 30, pp. 312–322.

31.

Jim

C. Y.

Chen

W. Y.

(2006) Recreation-amenity use and contingent valuation of urban greenspaces in Guangzhou, China, Landscape and Urban Planning, 75, pp. 81–96.

32.

Kulshrestha

S. N.

Gillies

J. A.

(1993) Economic evaluation of aesthetic amenities: a case study of river view, Journal of the American Water Resources Association, 29, pp. 257–266.

33.

Kuntz

P. A.

Christie

R. D.

Venkata

S. S.

(2002) Optimal vegetation maintenance scheduling of overhead electric power distribution systems, IEEE Transaction on Power Delivery, 17, pp. 1164–1169.

34.

Kwak

S. J.

Yoo

S. H.

Han

S. Y.

(2003) Estimating the public’s value for urban forest in the Seoul metropolitan area of Korea: a contingent valuation study, Urban Studies, 40, pp. 2207–2221.

35.

LaCommare

K. H.

Eto

J. H.

(2004) Understanding the cost of power interruptions to U.S. electricity consumers. Ernest Orlando Lawrence Berkeley National Laboratory, Berkeley, CA.

36.

LaCommare

K. H.

Eto

J. H.

(2006) Cost of power interruptions to electricity customers in the United States (US), Energy, 31, pp. 1845–1855.

37.

Lawton

Eto

J. H.

Katz

Sullivan

(2003) Characteristics and trends in a national study of consumer outage costs. Population Research Systems, LLC, San Francisco, CA.

38.

Leggett

C. G.

Bockstael

N. E.

(2000) Evidence of the effects of water quality on residential land prices, Journal of Environmental Economics and Management, 39, pp. 121–144.

39.

Lindsey

Knaap

(1999) Willingness to pay for urban greenway projects, Journal of the American Planning Association, 65, pp. 297–313.

40.

Lockwood

Tracy

(1995) Nonmarket economic valuation of an urban recreation park, Journal of Leisure Research, 27, pp. 155–167.

41.

Lupi

Kaplowitz

M. D.

Hoehn

J. P.

(2002) The economic equivalency of drained and restored wetlands in Michigan, American Journal of Agricultural Economics, 84, pp. 1355–1361.

42.

Mahan

B. L.

Polasky

Adams

R. M.

(2000) Valuing urban wetlands: a property price approach, Land Economics, 76, pp. 100–113.

43.

Mansfield

Pattanayak

S. K.

McDow

McDonald

Halpin

(2005) Shades of green: measuring the value of urban forests in the housing market, Journal of Forest Economics, 11, pp. 177–199.

44.

McNair

Abelson

(2010) Estimating the value of undergrounding electricity and telecommunications networks, The Australian Economic Review, 43, pp. 376–388.

45.

McPherson

E. G.

(1992) Accounting for benefits and costs of urban greenspace, Landscape and Urban Planning, 22, pp. 41–51.

46.

Mendelsohn

Olmstead

(2009) The economic valuation of environmental amenities and disamenities: methods and applications, Annual Review of Environment and Resources, 34, pp. 325–347.

47.

Michener

Tighe

(1992) A Poisson regression model of highway fatalities, American Economic Review, 82, pp. 452–456.

48.

Morancho

A. B.

(2003) A hedonic valuation of urban green areas, Landscape and Urban Planning, 66, pp. 35–41.

49.

T. A.

Stevens

Allen

P. G.

(1988) Valuation of urban parks, Landscape and Urban Planning, 15, pp. 139–152.

50.

Olson

M. K.

(2004) Are novel drugs more risky for patients than less novel drugs?, Journal of Health Economics, 23, pp. 1135–1158.

51.

Oregon Public Utility Commission (2010) Seven Year Electric Service Reliability Statistics Summary 2003–2009 ( http://www.oregon.gov/PUC/safety/10reliab.pdf27 ; accessed March 2011).

52.

Pahwa

(2004) Effect of environmental factors on failure rate of overhead distribution feeders, in: IEEE Power Engineering Society General Meeting, Vol. 1, pp. 691–692.

53.

Popoola

Ajewole

(2002) Willingness to pay for rehabilitation of Ibadan urban environment through reforestation projects, International Journal of Sustainable Development and World Ecology, 9, pp. 256–268.

54.

Radmer

D. T.

Kuntz

P. A.

Christie

R. D.

Venkata

S. S.

Fletcher

R. H.

(2002) Predicting vegetation-related failure rates for overhead distribution feeders, IEEE Transactions on Power Delivery, 17, pp. 1170–1175.

55.

Ridker

R. G.

Henning

J. A.

(1967) The determinants of residential property values with special reference to air pollution, The Review of Economics and Statistics, 49, pp. 246–257.

56.

Rose

N. L.

(1990) Profitability and product quality: economic determinants of airline safety performance, Journal of Political Economy, 98, pp. 944–964.

57.

Salazar

Menendez

L. G.

(2007) Estimating the non-market benefits of an urban park: does proximity matter?, Land Use Policy, 24, pp. 296–305.

58.

Sargan

(1958) The estimation of economic relationships using instrumental variables, Econometrica, 26, pp. 393–415.

59.

Sima

Yang

G. Q.

Jiang

C. L.

. (2010) Experiments and analysis of sand dust flashover of the flat plate model, IEEE Transactions on Dielectrics and Electrical Insulation, 17, pp. 572–581.

60.

Simpson

Bossuyt

R. van

(1996) Tree-caused electric outages, Journal of Arboriculture, 22, pp. 117–121.

61.

Sims

Dent

(2005) High-voltage overhead power lines and property values: a residential study in the UK, Urban Studies, 42, pp. 665–694.

62.

Smith

V. K.

Huang

J. C.

(1995) Can markets value air-quality? A meta-analysis of hedonic property value models, Journal of Political Economy, 103, pp. 209–227.

63.

Stock

J. H.

Yogo

(2005) Testing for weak instruments in linear IV regression, in: Stock

J. H.

Andrews

D. W. K.

(Eds) Identification and Inference for Econometric Models: Essays in Honor of Thomas J. Rothenberg, pp. 80–108. Cambridge: Cambridge University Press.

64.

Sullivan

M. J.

Mercurio

Schellenberg

Freeman

M. A.

(2009) Estimated value of service reliability for electric utility customers in the United States. Ernest Orlando Lawrence Berkeley National Laboratory, Berkeley, CA.

65.

Tong

Feagin

R. A.

Zhanq

. (2007) Ecosystem service values and restoration in the urban Sanyang wetland of Wenzhou, China, Ecological Engineering, 29, pp. 249–258.

66.

Troy

Grove

J. M.

(2008) Property values, parks and crime: a hedonic analysis in Baltimore, MD, Landscape and Urban Planning, 87, pp. 233–245.

67.

Tyrväinen

(2001) Economic valuation of urban forest benefits in Finland, Journal of Environmental Management, 62, pp. 75–92.

68.

D. M.

(1973) Alternative tests of independence between stochastic regressors and disturbances, Econometrica, 41, pp. 733–750.

69.

Zabel

J. E.

Kiel

K. A.

(2000) Estimating the demand for air quality in four US cities, Land Economics, 76, pp. 174–194.

70.

Zhou

Pahwa

Yang

(2006) Modeling weather-related failures of overhead distribution lines, IEEE Transactions on Power Systems, 21, pp. 1683–1690.