Activity-based hedonic pricing model of smart buildings

The second model presented in this book focuses on an essential dimension of smart space: its user-centricity. Smart buildings are totally focused on their occupants, thereby turning real estate space into “¡Space” (Lecomte, 2020). This is in sharp contrast to the previous model which does not incorporate user centricity in its formulation as interactions between occupants and buildings are purely captured in terms of smart space’s four strata.

Known as activity-based hedonic pricing model of smart real estate, this second model accommodates coded space’s user centricity by pricing individual interactions between space users and a building. In doing so, it echoes Liu and Gulliver’s

(2013) definition of a smart building as the sum of individual interactions with its occupants in smart space. Lecomte (2020) asserts:

In smart buildings, real estate merges with technology to form cognitive assemblages [...] User-centric interactions with real estate are primarily designed by code and mediated by non-conscious cognitive. Hence, smart space becomes real estate’s main productive component, whereas its physical characteristics move to the periphery of space users’ attention.

Whilst the previous pricing model is very much rooted in physical space, an activity-based model solidly positions a building into smart space. In order to implement this model, a taxonomy of all potential smart interactions has first to be drawn up (Lecomte, 2019a). Let Yt be the (In) price of the building (dependent variable) such that:

where wt = Swit with wit is the weight of activity i in the building’s use at date t. Coefficients wit are derived from actual smart use of the building (e.g. average use over a period). Zt is the matrix of independent quality variables representing the interactions in smart space related to all activities in the building. pt is the estimated coefficient vector for these interactions. st is the regression error term.

If the building encompasses N activities for K interactions2 then:

  • - wt is a IxN vector of N activities’ weights in the building’s actual smart use at date t with the sum of all weights w, (from 1 = 1 to N) being equal to 1;
  • - pt is a Kxl vector of K estimated coefficients;
  • - and Zt is a NxK matrix of N activities in the building’s taxonomy triggering K independent pre-defined interactions in smart space. Zt’s ith row consists of a vector of K interactions for activity i (z^ for k = 1 to K). Zt is such that:

zll

zNl

zlK

zNK

To implement the model, the following elements are required:

  • - a comprehensive taxonomy of activities and pre-defined interactions in smart space,
  • - a very dynamic formulation and calibration of these activities and interactions so as to take care of omitted variables and changes in the building’s usage patterns over time, which might outdate the current taxonomy of activities, and
  • - the ability to assess space users’ preferences at an extremely granular level and in terms of technology-driven experiences (linked to activities).

In this model, a smart building is akin to a set of on-demand services which are individually experienced by occupants in coded space. It is hypothesised that pre-defined interactions in matrix Zt can be experienced both successively and concomitantly with potentially synergistic effects while an activity takes place. By contrast, modelled activities are supposed to be exclusive (e.g. a building occupant is either working or shopping but not doing two activities at the same time).

Hence, one can expect that most interactions will occur as a succession of events rather than concomitant occurrences within an activity. Major exceptions to this rule are interactions pertaining to the building’s environment which truly remain in the background at all times. By definition, environmental interactions will be overwhelmingly concomitant (e.g. HVAC system, lighting), whereas activity-supporting interactions are conceptualised to occur individually and sequentially as activities unfold in digital-time. That is, smart space is inherently a specialised space. It might be an omni-use space (Lecomte, 2019a), albeit in a series of very specialised interactions matching with each modelled activity carried out in the building. Selection of quality variables in Zt should reflect the fact that smart space has both scope and depth.

Another dimension concurring to space specialisation is the presence of two generic types of interactions: (i) standard interactions shared by all tenants in a smart building and (ii) specific interactions which are idiosyncratic to a particular tenant. The latter might stem from Built-To-Order requirements, or result from the implementation of proprietary technologies? These assumptions about activities and interactions correspond to a new model of space user in smart real estate known as the cyber-dasein (Lecomte, 2019a). As pointed out by Dourish and Bell (2011), “new technologies inherently cause people to reencounter space”. The cyber-dasein purposefully engages in smart space in view of achieving tasks, one task at a time. His concernful absorption in carrying out his daily tasks in smart space is characteristic of pervasive computing’s phenomenology (Lecomte, 2020)?

Lecomte (2019a) describes a variant of the activity-based hedonic pricing model of smart buildings, by superimposing a behavioural map of space users’ patterns of activities in smart space. Such maps at the intersection of psychology and geography, called “preoccupation maps”, derive from psychogeography, a concept defined by French Situationist philosophers in the 1950s. Similar maps can be found in the pervasive computing literature where user-generated classifications of space are implemented (e.g. Calabrese et al., 2010). In the same line of thinking, Batty (2002) suggests to “conceive cities as being clusters of spatial events, events that take place in time and space, where the event is characterised by its duration, intensity, volatility and location”. The resulting model known as “behavioural spatial hedonic model of smart real estate” accounts for spatial effects in smart cities, especially spatial heterogeneity as smart buildings become more widespread in increasingly smart urban environments?

Defining utility in smart real estate: decision or experienced?

Hedonic pricing models are based on the fundamental assumption that a good’s observed market can be linked to observed utility-bearing characteristics

(Lancaster, 1966). Price differences among goods are explained by “the economic content of the relationship between observed prices and observed characteristics” (Rosen, 1974). Edmonds (1984) explains that a hedonic price model provides “a distinct, homogenous marketable tied bundle of characteristics [which are] objectively measurable, utility affecting attributes”.

In hedonic pricing models applied in real estate (see Chapter 1 of this book), utilities attached to a property’s characteristics are decision utilities. Their formation is underpinned by logical criteria assessed by a system of preferences, assuming that agents are rational. In smart environments where buildings are equivalent to a series of activities translated into interactions experienced in smart space, does a concept of utility disconnected from space users’ actual experience make sense for pricing smart buildings? Alternatively, should the concept of experienced utility introduced by Nobel Prize laureate Daniel Kahneman be applied instead?

Kahneman (1994) defines experienced utility as “the measure of the hedonic experience of [an] outcome” whereas “the decision utility of an outcome [...] is the weight assigned to that outcome in a decision”. Experience utility differs from decision utility insofar as it is “the rewards we realize once the choices are made” (Robson and Samuelson, 2011).

Lancaster (1966) who invented the hedonic pricing theory propounds “the objective nature of the goods-characteristics relationship” in a hedonic model, i.e. “the characteristics possessed by a good or a combination of goods are the same for all consumers, and [...] are in the same quantities”. Therefore, there is no need for involving the subjectivity of individual experiences in utility formation. Indeed, Robson and Samuelson (2011) explain:

Experienced utilities are of no interest to a fiercely neoclassical economist. Decision utilities suffice to describe behaviour. However, if we are to consider welfare questions, the difference may be important. If experienced utilities do not match decision utilities, should we persevere with the standard economists’ presumption that decision utilities are an appropriate guide to well-being?

In smart real estate where space is coded, utility should encompass smart space’s three realms: the space-time, money-time, and digital-time realms. When considering smart space as the good to be priced in a model of smart buildings, it is obvious that Lancaster’s assumption cannot hold. Each space user experiences real estate in a unique manner resulting in what Curry and Sheth (2018) call “a market-of-one”. To describe space users’ unique surrounding worlds in smart space, Lecomte (2019a, 2020) refers to “Umwelt”, a concept borrowed from Heidegger’s phenomenology (see Appendix 2.1). To paraphrase Situationist precursor Chtcheglow (1933-1998) in his Formulary of a New Urbanism, in a smart building “everyone will live in his own cathedral”.

Hence, due to its experiential dimensions and focus on building occupants’ well-being, smart space questions the use of decision utilities in hedonic pricing models of smart real estate. Furthermore, smart buildings’ occupants do not make every decision impacting their well-being in smart space since ubicomp coupled

Digitalisation of commercial real estate 89 with non-conscious cognitive instruments transparently and continuously act on their behalf as soon as they step into a smart environment. Defining the scope of decision utility is therefore problematic in a smart building.

Indeed, decision utility would struggle to gauge the full spectrum of subtle utilities involved in human-building interactions in smart environments. As mentioned before, decision utility also supposes that smart building occupants are capable of making rational choices on a range of utility-bearing characteristics which are technological in essence. This seems quite unrealistic in practice.

Despite its irrelevance for hedonic pricing models of smart buildings, decision utility can serve a useful purpose when analysed in comparison with experienced utility. The difference between the two utilities may even become an important indicator in smart real estate analysis at the micro-scale.

• The difference between decision utility and experienced utility in smart space as an indicator of affordances in smart buildings

The difference between decision utility and experienced utility, customarily known as “focusing illusion” in economics, can be of great interest to smart buildings’ owners provided they have a way to derive it. This is a granular indicator at the building level, which focuses on occupants rather than physical or digital structures. Let’s define the difference D by

n

Dt = ^U decision — U experienced

t=l

where n is the total number of building occupants at a given time t (instant utility) or over a period of time t (concept of extended utility mentioned in Kahneman, Wakker and Sarin, 1997).

Dt is the difference between decision utility and experienced utility for a specific interaction (or series of interactions attached to an activity) in a smart building.

Pervasive computing and non-conscious cognitive systems in smart real estate impose on space users’ pre-defined decision utilities inferred from revealed preferences in smart space (and derived from space users’ Umwelt). In that sense, in smart environments, decision utility is ubicomp’s modelled utility. It is an ex-ante utility whose technicalities might elude building occupants.

However, space users’ highly customised interactions with the built environment produce utilities which are first and foremost experiential, i.e. experienced utilities which quantify the actual amount of pleasure and pain evoked while interacting with the building. D, the difference between decision utility and experienced utility in smart buildings, measures the degree of‘non-predictability’ of building occupants’ experiences in smart space. It indicates how their own unique ways of experiencing space are not yet fully calibrated by the built environment’s embedded technological apparatus. As technology becomes smarter and space users’ personal surrounding worlds (Umwelt) take shape in smart space, one canenvision decision utility and experienced utility to converge (D = 0) to the point where space users’ experiences in space are perfectly forecasted by technology.6

Until this point is reached, the ability of embedded technologies to optimise space users’ utility will affect the price of utility-bearing characteristics in the two proposed models of smart buildings. A negative difference (D < 0) hints at affordances in smart space which are not properly modelled in the building’s technological apparatus. It is a good sign for any smart building owners inasmuch as more value can be accounted for than the building’s enterprise currently does (e.g. obsolete taxonomy of activities, accrued synergies among existing interactions).

Conversely, a positive D (i.e. experienced utility is less than decision utility) hints at misaffordances in smart space. These could stem from an array of causes, such as (i) dysfunction at one or several of smart space’s four layers, (ii) a disconnect between a building’s physical structure and its digital infrastructure resulting in failed synergies among pre-defined human-building interactions, or (iii) mismatches with the changing ways that space users perceive and/or consume smart space and its characteristics such as level of monitoring, respect for privacy, memory duration, and overall trust in context-aware environments. In any case, it is a warning signal for smart property owners.

Thus, it makes no doubt that unless D = 0, basing a hedonic pricing model of smart real estate on decision utility is inherently faulty. It would result in wrong estimates of a building’s price, with either underestimation if D < 0 or overestimation if D > 0. Experienced utility in hedonic pricing models of smart buildings is another break from aggregate thinking in real estate finance. Utilities in smart buildings are no longer deemed to be homogeneous. On the contrary, after 18th-century English economist Jeremy Bentham, experienced utility acknowledges the uniqueness of human interactions with the built environment in terms of their “intensity, duration, certainty and propinquity” (Stigler, 1950).

 
Source
< Prev   CONTENTS   Source   Next >