Office price index lagging in Singapore and Hong Kong.

ABSTRACT. Price discovery of real estate investment has been getting lots of attentions from researchers and it is generally believed that lagging errors exist in appraisal-based returns of commercial real estate investments, in comparison to other investment instruments traded in the stock market. Due to fewer transactions in the commercial real estate market, it is reasonable to notice a difference in the handling of current market information. By introducing two study approaches along with a test case using Singapore's data, this paper explores the extent of lagging in Hong Kong's commercial (office) real estate values, in a State Space Model with Kalman filter. The findings first suggest that whether appraisal-based indices overstate or understate true values lies in the economy condition at the time. Then, commercial real estate values in Hong Kong are about three months behind the stock market property indices. Also, as indicated by the findings, data collection/selection bias may render a de-lagged index not as efficient as it is--supposed to be. This paper provides a different perspective on price discovery and the process of de-lagging property values.

KEYWORDS: Price discovery; State Space Model; Kalman filter; Commercial Real Estate (Office) values

SANTRAUKA

BIURU KAINU INDEKSO ATSILIKIMAS SINGAPURE IR HONKONGE

Eddie Chi Man HUI, Ka Hung YU, David Kim Hin HO

Investiciju i nekilnojamaji turta kainos mokslininkus itin domina. Daznai manoma, kad, palyginti su kitais akciju birzoje siulomais investiciniais instrumentais, investiciju i komercini nekilnojamaji turta graza vertinama klaidingai del atsilikimo. Kadangi komercinio nekilnojamojo turto rinkoje sandoriu sudaroma maziau, verta pabrezti, kaip skirtingai tvarkoma aktuali rinkos informacija. Pristatant du tyrimo budus kartu su atvejo tyrimu pagal Singapuro duomenis, siame darbe, remiantis busenu erdves modeliu ir naudojant Kalmano filtra, nagrinejamas Honkongo komercinio nekilnojamojo turto (biuru) verciu atsilikimas. Isvados pirmiausia rodo, kad tai, ar vertinimu pagristi indeksai padidina ar sumazina realias vertes, priklauso nuo esamu ekonominiu salygu. Be to, komercinio nekilnojamojo turto vertes Honkonge nuo akciju rinkoje naudojamu nuosavybes indeksu atsilieka apie tris menesius. Isvados rodo ir tai, kad del salisko duomenu rinkimo (atrankos), neatsiliekantis indeksas gali buti ne toks efektyvus, kaip turetu buti. Siame darbe pateikiamas kitas kainu nustatymo budas ir aprasomas nuosavybes vertes atsilikimo panaikinimo procesas.

1. INTRODUCTION

Real Estate is an investment option not traded on the stock market. It has been illustrated in previous studies that the risk-adjusted return of real estate is higher than that of other investment goods. In other words, risk factors are priced differently on the stock market and the private commercial real estate market (Fu, 2002). Therefore, investment would be less risky when real estate is included in an investment portfolio. Also, real estate investment is considered a good hedging tool against inflation (Fama and Schwert, 1977; Webb and Sirmans, 1980; Miles and McCue, 1982 & 1984; Ibbotson and Siegel, 1984; and Brueggeman, et al., 1986).

However, there is a problem regarding the evaluation of real estate values and its return rates. Unlike the immense amount of information available on investment instruments traded on the stock market, information circulating on real estate is much less. According to Quan and Quigley (1991), costly search would be incurred by potential buyers, given the heterogeneity and fixity of real estate. Moreover, trading activities in the real estate market are decentralized and market prices are the result of negotiations between buyers and sellers. Those involved in the real estate market usually rely on very limited and partial knowledge to make decisions. This could probably lead to a gap in the bargaining power between the buyer and the seller, affecting the final transaction price in the process. Moreover, insufficient market information on real estate provides difficulties for investment managers to evaluate its returns accurately, in order to judge whether real estate is a good enough instrument to be included in an investment portfolio.

Singapore and Hong Kong are selected for this study, due to two reasons. First, both of such have been perhaps two of the most prosperous cities in the world, since the 1980s. They were one-half of the famous four Little Dragons in Asia. The economic systems of Hong Kong and Singapore are similar, and both are favored by foreigners to do business with. Entrepot trade has played a crucial role in both economies. In Singapore, under intensive urban planning and rapid diversification of the Singapore economy, specifically the rapid growth of the financial and business services sectors, demand for modern office space has been robust and sustainable. It is very similar to the situation in Hong Kong, as service-oriented industries have taken over Hong Kong's economy in the 1990s, once the secondary industries shifted their bases to Mainland China. This provides the backbone for office space demand in both cities. Secondly, both Singapore and Hong Kong were seriously wounded by the Asian Financial Crisis. As a result, the office price movements in both places were similar.

This paper will be divided into five sections. The first part is a literature review on appraisal-based returns and the problems caused by this method. Then, a literature review of the attempts to eliminate the problem of lagging in appraisal-based commercial real estate returns will be presented. The third section first introduces a single-index approach which is then applied to a test Singapore case study. Then, the model is trial-tested on a Hong Kong study, as presented in the fourth section. An alternate approach is proposed in case the first approach is not applicable within the context of Hong Kong. Further, a comparison is made with respect to how data selection bias plays a role in office price discovery and lagging. The last section concludes and summarizes the findings in this study.

2. LITERATURE REVIEW

2.1. Appraisal-based returns and its problems

It is reasonable to say that professional appraisals are generally referred to as a representation of the market value of real estate investment. The use of such appraisals aims to reduce search and information cost. Researchers, for instance, form indices on real estate values based on appraisal value, instead of transaction prices (Clayton, Geltner, and Hamilton, 2001). It goes so far as such appraisals are used to even evaluate the performance of investment managers, though such practice is generally viewed as an improper way to utilize the appraisals (Giliberto, 1988; Fisher and Geltner, 2000). Due to the impact that appraisals may have caused, one has to wonder how accurate these appraisal-based returns are, compared to its "true" return values.

Numerous studies were conducted previously to investigate the plausibility of the appraisal-based returns for real estate investments. The results are mostly negative, due to the problem of lagging (or smoothing). The reliance on previous prices (Ibbotson and Siegel, 1984), or previous appraisals (Ross and Zisler, 1987; Geltner, 1989) by appraisers in the estimation of the current market value, are the reasons contributing to the "smoothed" results. Valuations on the stock market is more updated than that on the real estate market, as real estate prices reflect the changes in market conditions slower. Fu and Ng (2001) find that about one half of the effect of the current market news is captured by the contemporaneous adjustment in real estate transaction prices. Therefore, it is possible that some of the impact of the outdated information is still present in the appraisal values (Fisher and Geltner, 2000). Corgel and deRoos (1999) state that the extraordinary risk/return relationship in appraisal-based real estate can be deciphered as a compensation for illiquidity and inefficient information. On the other hand, analyses would be erroneous without correcting the effect of smoothing (Geltner, 1991). Examples can be found on studies by Brueggeman, Chen, and Thibodeau (1984); Miles and McCue (1984); Hartzell, Hekman, and Miles (1986, 1987); Hartzell, Shulman, and Wurtzebach (1987).