The factorisation of commercial real estate: factor-based real estate derivatives

Factopbased real estate dérivatives

Introduction: a brief history of real estate derivatives

As far back as recorded history goes, there is no evidence of the use of property derivatives until very recently. In fact, property derivatives did not exist. Mesopotamians could hedge the price of grain, barley, or red garlic, but they had no way to hedge the price of their houses (Swan, 2000).

In modern times, property derivatives were first introduced on the London Futures and Options Exchange (FOX) in London in May 1991 (Patel, 1994). Trading in the four index-based contracts only lasted for a few months until the contracts plagued with illiquidity were withdrawn amid allegations of false trading.

A few years later, Barclays Capital started to issue Property Income Certificates (PICs) akin to structured notes linked to Investment Property Databank (IPD) indices (all property income return and all property capital return). Almost GBP 800 million of PICs were issued in the 1990s. McNamara (2010) reports that from 1996 to 1998, Barclays Capital also issued Property Index Forwards (PIFs) designed after contracts for differences with returns linked to the IPD UK All Property Index.

In the early 2000s, after the Financial Services Authority and the Inland Revenue in the UK clarified their stances on property derivatives, interest for property derivatives was rekindled. By 2004, 21 investment banks acquired licences to commercialise IPD index-based property derivatives (Baum, 2015). IPD Total Return Swaps grew rapidly, reaching 265 contracts (GBP 3.5 billion notional) written in QI 2008 when they fell victim to the subprime crisis (Torous, 2017).

Whilst the UK has been at the centre of property derivatives innovation (due to the serendipity of a well-organised property sector and innovative financial markets), the USA swiftly jumped on the bandwagon with the first NCREIF Property Index (NPI)-based swap agreement organised by Credit Suisse First Boston in January 2006 for US$10 million over two years (Fisher, 2005; Syz, 2008).

By October 2007, the Chicago Mercantile Exchange (CME) launched trading in futures and options on the S&P/Global Real Analytics Commercial Real Estate indices (Syz, 2008). Jud and Winkler (2009) note that the plan was to trade cash-settled commercial real estate futures in office, warehouse, apartment, and retail property sectors for all US regions with electronic trading out 20 quarters. This attempt to list direct real estate derivatives on standardised markets abruptly ended in December 2008 when the underlying index was no longer produced (Torous, 2017).

Meanwhile, in continental Europe, the largest European futures and options market, EUREX, listed IPD index-based futures. These cash-settled annual contracts were based on the total returns of MSCI-1PD UK quarterly indices. Starting with the UK All Property Index listed in February 2009, Eurex enlarged its listing in 2013 and 2015, by offering quarterly index futures on granular IPD UK indices (All property type x Region). Nine standardised futures contracts were open for trade until June 2020 when the market was shut down (Eurex circular 042/2020).

Despite the ebullient 30 years since the FOX Futures contracts were introduced, Tunaru and Fabozzi (2017) sound quite disillusioned with the actual success of property derivatives. They write:

Commercial real estate is directly linked to the real economy; by total size, it represents a significant spot market. However, it is still quite difficult for investors to hedge the risk exposure arising from investing in this important asset class [...]. Almost 25 years [after Shiller’s Marco Markets (1993)], we are still waiting for standard derivatives such as such futures and options to be established as main contract with healthy liquidity.

As a matter of fact, in spite of the Global Financial Crisis, the real estate derivatives market is still in its infancy and has failed to develop beyond “an embryonic stage” (Tunaru, 2017). Lecomte (2007) argues that the main shortcoming of past attempts to launch property derivatives lies in these instruments’ poor hedging effectiveness for direct real estate assets. Tunaru (2017) notes that “overall the pace of financial innovation is very slow in this area [of real estate derivatives]”. Past research have focused on underlyings by looking at ways to improve the reliability of private real estate indices (e.g. Geltner, 1989). The problem with property derivatives stems not only from their reliance on direct commercial real estate indices, which do not capture the full range of idiosyncrasies in real estate risk at the granular level, but also and perhaps more importantly from their use of an index-based format.

Indeed, irrespective of the forms they have materialised into (e.g. Total Return Swaps, contracts for differences, listed futures contracts and options, structured notes), real estate derivatives have systematically been structured as index-based instruments using private commercial real estate indices (such as IPD indices in the UK or NCREIF Property Index in the USA) as underlyings. That is, they imply that private property indices are the optimal choice of underlying for a property derivative. The application of the index-based format of derivatives to commercial real estate supposes that whatever parameters led to the creation of index-based derivatives for equity portfolios also apply to commercial real estate. But, is it so? This chapter addresses this question and proposes two alternatives to

Factorisation of commercial real estate 3 the index-based format of derivatives in real estate finance: combinative derivatives and factor-based derivatives.

These two models of real estate derivatives epitomise what Shiller (1993) calls “a different approach to identifying potential new markets”. As this chapter explains, combinative derivatives have already been applied to other asset classes. Furthermore, factor-based real estate derivatives have been mentioned by Shiller (1993) in his analysis of macro markets applied to real estate:

One could imagine some factor analytic modelling to discover factors underlying variation in prices of claims on incomes, to enable contracts to cash settle on the basis of these factors.

This chapter reviews the notion of factors in classical finance and real estate finance. It describes the two new factor-based models of real estate derivatives proposed by Lecomte (2007) in line with Shiller’s early intuition. It then proposes solutions to overcome the difficulties inherent to the use of factor models. It concludes by reviewing the concept of stochastic process in real estate finance and devising a novel way to model commercial real estate’s price dynamics.

Factors versus indices

Models of derivatives and real estate risk

The perfect world of CAPM

The year 1982 was crucial in the history of modern finance with the introduction of index-based derivatives on US markets. The negotiation leading to this outcome was long and arduous with the first attempt to launch Dow Jones Futures dating back to the late 1960s (Brine and Poovey, 2O17).1 Against all odds, Leo Melamed, chairman of the Chicago Mercantile Exchange, managed to convince regulators that index-based futures contracts were not akin to gambling, but instead that they met a real need from investors and hedgers.

Index-based derivatives’ innovativeness results from two unique features: cash settlement and the use of a broad financial index as underlying. Both conceptually and historically, index-based derivatives are linked to the Capital Asset Pricing Model (Neiderhoffer and Zeckhauser, 1980). Bernstein (1995) notes that Capital Asset Pricing Model (CAPM)’s dominance in the financial industry following the 1973 oil price crisis provided some badly needed relief to investors unsettled by market turbulences.

Indeed, index-based derivatives were designed for the perfect world of CAPM’s normalities. They take care of systematic risk while portfolio diversification is supposed to reduce idiosyncratic risk to a negligible entity. CAPM’s reductionist approach to risk imposes on index-based derivatives a binary framework in terms of systematic risk/specific risk, which does not sit well with individual properties’ heterogeneity. Young and Graff (1996) sum up the dilemma facing real estatefinance when applying Modern Portfolio Theory and its antecedent the Efficient Market Hypothesis:

MPT and EMH seem to have been introduced into real estate to justify the use of particular statistical techniques and portfolio strategies rather than as a consequence of empirical analysis of investment return and risk characteristics. In science, the situation is generally reversed: theories are developed to explain observations.

Real estate risk is different

As repeatedly mentioned in the academic literature (e.g. Weimer, 1966; Miles and Graaskamp, 1984), real estate has a number of idiosyncrasies which set it apart as an asset class from equities and fixed income: asymmetry of information, high transaction cost, illiquidity, importance of physical characteristics. Numerous evidence show that real estate fundamentally differs from CAPM’s perfect world (Clapp, Goldberg and Myers, 1994)-

Since the early 1980s, literature on portfolio diversification has abundantly explored real estate risk. These studies focus on diversified portfolio’s risk, in particular as it relates to risk breakdown between systematic risk and specific risk. For instance, Miles and McCue (1984), Hartzell, Hekman and Miles (1986) indicate that systematic risk as defined in CAPM rarely exceeds 20% of total risk for a portfolio of buildings in the USA. Likewise, Brown and Matysiak (2000) who research the UK commercial real estate market identify that market risk accounts for less than 10% of the average fluctuation in a building’s total returns, versus 30% on average for a listed stock.

Bruggemann, Chen and Thibodeau (1984), Titman and Warga (1986) confirm these findings, and conclude that CAPM is not adapted to capture the risk/re-turn relationship of direct real estate assets. As a result, researchers have explored other models more suited to explain real estate returns (e.g. Hoag, 1980). Some of these models are based on hedonic price regression described by Clapp and Myers as one of real estate’s fundamental paradigms. In terms of risk structure, the hedonic model is comparable to a multifactor model insofar as hedonic indices are made up of a range of hedonic variables whose individual contribution to the asset’s total utility makes up the asset price.

Whilst CAPM supposes that there is only one source of non-diversifiable risk captured in the market portfolio, factorial models acknowledge that risk might stem from multiple sources. The CAPM explicitly validates a mono-causal approach to real estate risk, whereas multifactorial models take into account various causes and their varying degrees of influence. The consequence of the index-based derivatives model applied to heterogeneous assets whose risk structure is not compatible with the CAPM is basis risk (Figlewski, 1984). In contracts for difference, a model which has been used for real estate derivatives, cross-hedge basis risk stemming from intrinsic differences between the asset returns to be hedged and the index price changes can become an insurmountable problem for hedgers and an obstacle to the smooth working of any standardised derivative market.

One cannot help wondering why real estate derivatives are, or have been, designed after a CAPM framework (i.e. using a composite index as underlying). A possible explanation is that there are no derivatives market capable of dealing with the myriad of factors that an instrument designed to address multiple sources of risk would entail.

The two realms of commercial real estate

Given the conceptual deadlock that composite index-based derivatives represent for real estate assets, an alternative model of derivatives using a hedonic index as underlying has been mentioned as part of Shiller’s macro markets (1993). Shiller’s model puts the spotlight on the true nature of commercial real estate. In a nutshell, when designing a derivative, shall a building be considered as a physical entity or as a financial asset?

Grissom and Liu’s analysis of the different paradigms in real estate sheds some light on this question (1993). Two models can be applied to commercial real estate:

  • - A space-time model best defined by James A. Graaskamp (1976) and urban land economists before him, such as Richard U. Ratcliff (1949), and
  • - A money-time model.

Any real estate investment aims to achieve the monetary cycle leading to the conversion of space-time into money-time, or more prosaically square feet into money. A real estate derivative instrument designed after commodity derivatives focuses on real estate’s physical characteristics. It positions real estate risk in the space-time realm by emphasising real estate’s physical characteristics over its financial dimension. Conversely, a real estate derivative designed according to equity index-based derivatives assumes that real estate assets are akin to financial assets. It overlooks real estate’s physical dimension and reduces risk to the moneytime realm. Grissom and Liu (1993) make clear that it is essential to capture both dimensions:

The money-time component represented in the classical literature is only a portion of the total return equation. The space-time dimension must be fully comprehended to understand the nature of real estate products, markets and problems. In a financial context, the failure to grasp space-time component will result in inappropriate analysis of the risk dimension of real estate.

A hedonic index seems like an astute way to combine commercial real estate’s two realms: a hedonic underlying encapsulates the spatial dimension through the selection of utility contributing variables while the derivative’s index-based structure accounts for the financial dimension. Notwithstanding its great theoretical interest, is such model operational? As noted by Shiller (1993), hedonic index-based derivatives instruments would be very challenging to implement in practice. Widespread arguments over the choice of underlyings (i.e. factor selection and pricing) have traditionally been a major hurdle in the implementation of hedonic demand theory beyond the confines of academia.

As a matter of fact, there is no agreement on what a hedonic underlying should be for commercial real estate assets (Shiller, 1993). Hedonic indices are difficult to construct, in particular for buildings with limited transaction flows and/or significant alterations in-between transactions. In Macro Markets, Shiller proposes an alternative which would rely on a rental index to create a market for “perpetual futures”. Noticeably, these perpetual futures are based on indices of revenue rather than price. Such model which would overlook the full space-time dimension of commercial real estate has not managed to take hold among promoters of property futures markets since it was introduced as a concept in the 1990s.

Narrow price index-based derivatives and other possible models of real estate derivatives

Fisher (2005) proposes a simpler alternative to Shiller’s perpetual futures: derivatives on narrow price indices of direct properties. In any market for physical assets, the more granular the index, the narrower the definition of the market. A real estate index is more or less narrow according to the choice of property type(s) and location(s). A narrow index should theoretically allow for better hedging than a large index, by making it possible to closely replicate the returns to be hedged.

However, the more precisely defined the property type and sub-type as well as the location in the underlying, the larger the risk of manipulation of the real estate derivatives market using such narrow price index as underlying. Market authorities have traditionally been very suspicious of derivatives instruments based on narrow indices. There are good reasons for such distrust. Shiller (1993) underlines the risk of market manipulation as a major concern for commercial real estate derivatives, contrary to what the situation would be for residential real estate. Manipulation risk of derivatives based on narrow residential price indices seems fairly remote due to the intrinsically atomised nature of residential property markets. Whilst investments in commercial real estate involve a much smaller number of investors, residential real estate imply large numbers of owners, buyers, and sellers in most submarkets.

That’s one of the paradoxes of real estate derivatives. Real estate derivatives ought to reflect as much as possible commercial real estate’s space-time dimension. However, the more they do so, the less likely they are to make it into the world of finance. Finding the right balance between buildings’ physical characteristics and the granularity of the index used as underlying is a challenge.

Factor hedges

The existing index-based model of derivatives is not well suited for commercial real estate. Its binary framework does not capture real estate risk, which is overwhelmingly unique. Lecomte (2007) proposes an alternative called “factor hedges”. Factor hedges apply a biomedical analogy to define derivatives as a combination of a format (e.g.

Factorisation of commercial real estate 7 index-based) and underlying (e.g. narrow index). Based on this analysis, two alternative models of derivatives are proposed: combination hedges and pure factor hedges.

Combination hedges for commercial real estate

With combination hedges, the idea is to deal with risk from different concomitant angles. The main shortcoming of this approach stems from the potential interactions between instruments used in the ad-hoc combination. Lecomte (2007) suggests that a combination hedge materialise as an index-based futures contracts with some granularity (e.g. property type x location) combined with add-on instruments, resulting in a hybrid customisable risk management tool. Figure 1.1 illustrates this format.

Model of combinative derivatives

Figure 1.1 Model of combinative derivatives.

The combination hedge model implies a hierarchy of causes with the principal source of risk being encapsulated in a futures contract whilst additional causes serve as underlying to, for instance, options. Such a causal analysis of real estate risk is supported by past research’s findings which show that property type is usually the most important criteria in terms of portfolio diversification (e.g. Miles and McCue, 1982 for the USA; Hamelink et al., 2000 for the UK), whereas geographic criteria are secondary.

In its simplest form, a combinative instrument could be made up of a futures contract tied to a property type sub-index and options linked to selected economic indicators. Lecomte (2007) suggests the design of options based on economic bases (after Mueller and Ziering, 1992; Mueller, 1993). Components of this aggregate hedge might even be individually tradable. Appendix 1.6 presents a model of combinative derivatives prioritising property type over the other factors in the combinative.

The idea of combining several standard derivatives instruments within one hedge is not new. It has been used in the context of yield curve risk management. Appendix 1.2 presents a detailed analysis of alternative hedging methods in interest rate risk management. By comparison, an index-based derivative follows a dual model (Figure 1.2). Risk analysis is minimal, which results in potentially significant basis. Basis risk is all the more important as the hedged asset does not comply with CAPM’s hypotheses. The derivatives instrument hedges systematic risk whilst basis is left unhedged.

Model of index-based derivatives

Figure 1.2 Model of index-based derivatives.

Pure factor hedges for commercial real estate

Real estate derivatives can also be designed as pure factor hedges, i.e. a combination of derivatives instruments using a single ‘pure’ risk factor as underlying. The atomisation of the hedged asset’s risk structure is essential in this approach, as each factor should be theoretically exchangeable as stand-alone. Pure factor hedges are highly customised instruments. Besides, they have the ability to adapt to changes in the hedged asset’s risk structure, thereby enabling very dynamic hedging strategies and reducing basis risk to a minimum.

Lecomte (2007) mentions that factors in real estate derivatives may come from two sources:

  • • Macro-factors resulting from the interaction of buildings and their environments. Akin to economic indicators, they would easily be exchangeable. The search for macro-factors explaining commercial real estate returns have been extensively covered in the academic literature (e.g. Kling and McCue, 1987; Chen, Hsieh and Jordan, 1997; Ling and Naranjo, 1997).
  • • Micro-factors stemming from a building’s intrinsic characteristics. They are specific to each building, its property type, its location. Defining them, valuing them, and ultimately hedging them would be a challenge. Contrary to macro-factors which are standardisable and applicable to a wide range of assets classes other than commercial real estate, micro-factors are asset class specific.

This distinction between macro-factors or micro-factors is consistent with Hoag’s analysis (1980) which is presented later in this chapter. Lecomte (2007) explains that the factor-based structure would be comparatively easier to implement than the process of identifying and pricing underlying micro-factors. Central to this issue are the number of factors to include in a factor hedge (Breitung and Eick-meier, 2005) as well as individual factors’ contribution to a building’s total risk and factor collinearity. These questions have traditionally plagued the implementation of factor models in finance (Shiller, 1993). Furthermore, they have also prevented the widespread use of multifactor models in real estate (Draper and Findlay, 1982).

Model of factor-based derivatives

Figure 1.3 Model of factor-based derivatives.

illustrates a derivatives instrument with atomised underlyings

Figure 1.3 illustrates a derivatives instrument with atomised underlyings. The model relies on six factors, which is within the usual range of factor models in real estate (e.g. Chen, Hsieh and Jordan, 1997 who select five factors). Each model of derivatives implies a specific way of conceptualising risk. From a binary structure in the index-based model, real estate risk becomes multifactorial in the factor-based model. The link between factors and risk components is achieved through customisation, whereby each factor impacts a specific part of total risk. One factor can affect one risk or several risk component(s), either alone or concomitantly with other factors.

By combining macro-factors with micro-factors specific to commercial real estate, the model produces sophisticated risk management tools for dynamic hedging strategies. Basis risk is reduced to a minimum. Nonetheless, the model comes with its fair share of problems. First, as previously mentioned, a factor can impact several risk components in the total risk structure. Collinearity issues are instrumental to the model implementation. This is all the more important as factors might be selected out of a clearly defined risk model, which explicitly sets the network of causal relations between factors and risk.

Pending issues with factor hedges

Based on the analysis presented in this chapter, it should make no doubt that the way one thinks about commercial real estate risk as well as tools available to deal with it are directly dependent on models applied to analyse that risk. In theory, derivatives should adapt to models which are themselves representative of empirical observations. With index-based derivatives, a financial economics model not applicable to commercial real estate is driving the choice of derivative format and underlying. Unless the application of CAPM ends up modifying empirical observations pertaining to commercial real estate risk and return, there is no way that index-based real estate derivatives can efficiently deliver what they are supposed to. MacKenzie (2006) notes that “financial models shape markets” inasmuch they are engines designed to power markets, and not cameras meant to perfectly reflect reality.

Hence, one can legitimately wonder whether efficient risk management tools can ever be designed for commercial real estate without a widely accepted risk model. Irrespective of the asset class, any derivative instrument designed without a solid underlying risk model, or at least the belief that it can provide a relevant framework to analyse risk is likely to be simplistic and reductionist in its approach. A poor understanding of risks involved in assets will sooner or later jeopardise such derivatives’ hedging effectiveness, and overall ability to fulfil its role.

Factorisation is the key to more gradated approaches to commercial real estate risk management effectively addressing real estate’s granular dimension. When it comes to derivatives, real estate should set free from the theoretical orthodoxy of financial economics in modern finance. Many exciting ideas can be explored beyond the restrained world of CAPM. For instance, Lecomte and McIntosh (2005) mention the concept of “standardised customisation” to describe a derivative model able to optimally capture commercial real estate’s highly idiosyncratic risk while setting up liquid standardised markets for these real estate derivatives of a new kind. ‘Standardised customisation’ of derivatives instruments is made possible owing to factorisation.

Notwithstanding their attractiveness on paper, factor hedges would come with their fair share of unresolved practical issues if they were to be implemented:

  • - What factors should be selected and how?
  • - What risk model could underpin factor selection in factor hedging?
  • - Can a standardised market be set up to trade such instruments?
  • - What market structure would enable liquid and efficient trading of real estate factor-based derivatives and their components?

The rest of the chapter addresses these questions.

 
Source
< Prev   CONTENTS   Source   Next >