The growing popularity of using forward contracts to pre-sell properties is attributed to the benefits generated to both sellers/developers and buyers. For developers, it can improve not only the cashflow of the development, but can also help developers hedge against any possible financial loss on the unsold properties when a price decline is expected by the time the construction is completed. As such, selective hedging can be attained through the use of futures markets to increase the efficiency of the price-forming process in the forward property market. On the other hand, if the economic sentiment favors the property market, developers can use the earnings collected from the presales to reinvest in other construction projects to yield further returns.

On the other side, prospective presale property buyers would like to deploy anticipatory hedging through a purchase of an uncompleted property. The hedge is to take advantage of the current price against any anticipated price appreciation of properties in the future; in particular, when a boom market is anticipated by investors. Furthermore, presale properties can offer more choices to both home-seekers and investors to look for their ideal dwellings/investments in terms of the location choice and the attributes available, which are short of supply in the spot market.

Forward Risks on Presale Properties

Apart from the respective benefits yielded, there are also risks embedded that are specific to presale properties. With the use of forward contracts, developers can pass part of their project risks to the buyers through the transfer of the equity interests of the uncompleted properties (Leung and Hui, 2005). As shown in Exhibit 1, apart from the transfer of market risk during the construction time-lag, presale property buyers have to bear the additional cost of capital invested in the uncompleted properties. Unlike financial contracts, no dividend in the form of rent will be generated to the buyers on the uncompleted properties to cover the additional cost of capital incurred during the forward contract period. Therefore, it is a common practice of developers to impose a discount on the listed prices of the properties if full payments are settled by the buyers at the time of the presales (Apple Daily, 2006a).

Furthermore, there are also asymmetric information issues in a presale property. Once a forward contract is executed, the buyer has to rely on the developer to finish the construction work in accordance with the terms stipulated in the contract. However, the buyer cannot be sure whether his/her best interests are served by the developer of up-keeping the quality work after the developer has collected the money, which may result in overpricing of the final product. According to the study conducted by Yang (2001), the problems of presale properties in China derived not only from default of the projects in the middle of the construction and of low quality, but also features mismatched with what had been promised in the presale arrangement. Similar issues are also found in other countries like Singapore and Malaysia (Ong, 1997; and Esha, 2003). Surveys conducted in Hong Kong also revealed that prospective buyers were often given inaccurate, insufficient or even misleading information in the presale brochures and showflats, and developers used grey areas for not adhering to the requirements stipulated in the presale guidelines.1 For example, no price list was attached in the presale brochures to allow buyers to compare prices of different flats, and some developers had selectively announced part of the details of the transactions concerned in order to ramp up prices higher than the equilibrium prices by creating an impression that the market was in dire demand for properties (HKSAR 1997, 2005; SCMP, 2005; and Lai, 2006).

Pricing Framework of Presale Properties

According to the "return equivalence theorem" of Hendershott (1996), the expected risk-adjusted returns are equal across different investments that have the same risk level. Therefore, if presale properties possess the same amount of risks as those sold in the spot market, their returns should, in principle, be the same. However, if there are asymmetric information issues in the forward property market, developers who have access to the information would enjoy a comparative advantage in the pricing of presale properties. According to the conceptual framework developed by Monroe (2003), the equilibrium prices of spot properties is a function of a list of internal and external factors as shown in Exhibit 2. They include the costs of producing the housing attributes, the age of the property, and the market return required from the investment. Regarding presale properties, the pricing takes into consideration not only the internal and external factors, but also the additional "forward factors," which are specific to the forward markets as shown on the right-hand side of Exhibit 2. These forward factors include the risks arising from market uncertainty during the construction time-lag, the missing rents to compensate for the additional capital financed within the forward contract period, and the hidden factors arising from the asymmetric information issues.

It has been suggested in the property markets that the hidden forward risks that are specific to presale properties have imposed an additional cost to the buyers compared to buying a spot property. Developers might have charged a premium on the housing attributes higher than the equilibrium market price by the time they offered the presales, thereby taking advantage of the asymmetric information issues (SCMP, 2005; Apple Daily, 2006b; and Lai, 2006); in particular, when the demand for properties was high in the market. On the other side, buyers accepted the extra premium imposed on the presales by developers in order to get developments with the attributes they desired (Yang, 2001). As Duffie (1989) cited, it is widely reckoned that forward markets are used by those who are risk-averse and are willing to securing their needs by use of a forward contract.

Despite the importance of understanding the pricing mechanism of presale properties, there are few researchers studying how the forward factors, in particular how the hidden forward risks affect the pricing of presale properties. In this regard, this study sets up a pricing model that is capable of incorporating the forward risks listed in Exhibit 2 into the pricing mechanism of presale properties.

Literature Review

There is rich literature about the pricing of spot properties, but only limited studies have been conducted on the pricing of presale properties. Among the very few, Chau, Wong, and Yiu (2003) set up a price discovery function by use of repeat sales method for constructing a Forward Contract Repeat Sales (FCRS) method for constructing a forward property price index and for investigating the equilibrium relationship between the spot and forward property prices in Hong Kong. The study has identified the returns generated from the presale properties to cover the market uncertainty during the construction time-lag. It also confirms that presales of uncompleted properties are traded at a discount to the expected spot properties to compensate for the missing rents during the forward contract period. However, the FCRS Index is constructed using registered2 pair sales of only uncompleted properties transacted in the forward property market in previous years. Therefore, the price variations shown in the Index cannot reflect the up-to-date market information. Furthermore, the price discovery function has not accounted for the hidden forward risks contained in the forward property market.

Centaline Property Agency (2006), one of the leading property agencies in Hong Kong, set up the Centa-City Leading (CCL) Index aimed at providing an up-to-date trend for gauging the pricesetting in the first-hand property market in which presales are active. The CCL Index is a weekly index based on the preliminary contract prices of spot properties conducted through Centaline to monitor the up-to-date property price variations reflecting the more recent market sentiment. Although the market generally considers that the prices of first-hand properties leads the pricing of spot properties sold in the second-hand market, due to the bias caused on the property prices because of the limited coverage of sales in terms of location and types of properties conducted in the first-hand market within a particular time-period, practitioners still have to rely on the price trend of spot properties, like the CCL Index, as a reference for price-setting of new developments using the comparison method (HKSAR, 2005; and Apple Daily, 2006C;). This practice is in line with the findings of Yiu, Hui, and Wong (2005) that suggest that "during periods of low-volume ratios (i.e., the forward market is relatively less active than the spot market), the spot return Granger causes the returns of forward contracts; and during periods of higher-volume ratios, there are feedback relationships between the two markets." However, the preliminary ASPs contained in the CCL Index are usually not registered in the Land Registry and are thus not publicly available for further validation of the model. Furthermore, the CCL Index is actually an index constructed using data collected from the second-hand market, which contains little information specific to the presales of uncompleted properties. Nevertheless, it gives important insights for pricing of presale properties through the use of comparison method.

The market comparison method to property appraisal is used to predict the market value of a particular real estate asset through the analysis of recent sales of similar properties. In a competitive market, economic assets that provide equivalent services or prospects for future benefits must have the same market prices no matter they are sold in the spot or forward market (Isaac, 2000). This principle also applies to property markets and serves as the foundation for the market comparison method. At any point in time, properties with similar characteristics must have the same market values and the value of a property, upon adjustment on the quality, must be equal to the price of recently sold, similar properties. The advantages of the market comparison method are that not only it is quick and easy to use, it is also the backbone incorporated in other appraisal methods (Baum, 1991). To apply the comparison method for pricing presale properties, the Single Index Model developed by Sharpe (1985) could help. Sharpe advocated that the returns on any security could be determined solely by random factors and the relationship with some common indices. Given this assumption, the returns on a security in a period relative to an index by use of an index-tracking strategy is incorporated into the Single Index Model (Brown and Matysiak, 2000; and Byrne and Lee, 2000). This comparison model has been widely adapted by researchers and industry practitioners is based on the very popular Capital Asset Pricing Model (CAPM).

Methodology

Following the framework outlined in Exhibit 2, it is possible to develop a model by encompassing the Hedonic Pricing Model (HPM) (Rosen, 1974), the Repeat Sales Price Model (RSPM) (Shiller, 1993), and the Single Index Model (SIM) (Sharpe, 1985), taking the forward factors into consideration, for pricing presale properties in the forward market in which asymmetric information is contained.

Hedonic Pricing Model and Repeat Sale Pricing Model

Index tracking on the price changes of presale properties. The FSIT Model expressed in Equation 2 is in line with the comparison method used in the industry for property appraisals. The beta value, ^bgr;^sub 2^, measures the elasticity of the two streams of return rates. The implications carried in β^sub 2^ are the following: (1) the higher the β, the more volatile of the forward market compared to the spot market; (2) the lower the β, the less volatile of the forward market compared to the spot market; and (3) if β [asymptotically =] 1, the relative forward-spot price changes track along the rates of return of the spot properties in exact synchronization

Discounts required to compensate for the additional cost of capital. The discount required by the buyer is regarded as the compensation for the missing rent to cover the additional cost of capital for holding the uncompleted property during the construction time-lag. The coefficient β^sub 3^ is expected to be positive, i.e., the higher the interest rate and the longer the construction time-lag, the more discount required in the presales and thus the larger the relative price changes between the pair sales (Chau, Wong, and Yiu, 2003).

Depreciation due to aging of the properties. In reality, there are seldom repeat sales of the same property conducted in the forward market at t^sub -1^, and then the second sale conducted again when it is just completed at t^sub 0^ (i.e., with zero-age). If the second sale is conducted some time after the completion at t^sub 1^, then adjustment has to be made into the model to control the aging impact of the spot properties transacted in the second sales in order to keep the price level free from a change in quality over time. A negative sign on β^sub 4^ is expected to show the depreciation of the property during the holding period.

Risk premium arising from the hidden forward risks and the three repeat sales method. As mentioned in previous sections, the hidden forward risks representing a group of uncertainties arising from asymmetric information issues create an opportunity for developers to charge an extra premium in pricing presale properties. Since there is hardly any proxy available in the market for identifying this group of risks in one measure, a dummy variable, h, is built into the model to investigate the impact of the hidden forward risks imposed on the pricing of the presale properties.

For the dummy variable, h, if the coefficient β^sub 5^ significantly lowers the intercept (in negative sign), the differential intercept suggests that a risk premium relative to the price changes has been embedded in pricing the presale properties. The larger the negative differential intercept, the higher the presale properties were priced at t^sub -1^ compared to their equilibrium spot prices at t^sub 1^.

Data Source

Quarterly data from years 1993 to 2005 are extracted from various sources for the validation.

The housing estates chosen. A total of 2,748 pair sales, including both the forward-spot pairs and the spot-spot pairs, are extracted from the twelve large housing estates located in different districts of Hong Kong with years of presales given in Exhibit 3. The units in the sample are all drawn from high-rise buildings in self-contained housing estates that have similar structural characteristics, neighborhoods, and amenities (Tse and Love, 2000). As required by the FSIT Model, only properties with three repeat sales are selected so that a forward-spot pair and a spot-spot pair can be formed for the same individual property. The property sales contained in the period from years 1993 to 2005 are extracted from the Economic and Property Research Centre (EPRC) and are spread evenly over the study period so that both boom and bust periods were covered in the tests. In order to narrow the impact arising from aging, the ages of the properties selected for the study were no more than ten years old when the sales were conducted.

Benchmark spot property price index. Different price indices on spot properties are published by the Rating and Valuation Department of the Hong Kong Government (RVD, 1995-2005) to measure the property price movements. Among the indices, the one for the Selected Popular Residential Developments (SPRD) is considered appropriate for use in this study. The housing estates included in the SPRD are all large self-contained estates that share similar characteristics with the twelve housing estates chosen for the study in terms of period of construction, building style, facilities, and property attributes. Also, the data are available on a quarterly basis, which provides more up-to-date information on recent market sentiment and has been the primary source that practitioners use. The SPRD price index is, therefore, used as the benchmark proxy for measuring returns generated from the spot property market.

Discount to compensate for the additional cost of capital. In Hong Kong, most property sales involve mortgages. The mortgage interest rates can be considered as the discount rates required by buyers to cover the additional cost of capital incurred during the construction time-lag. The mortgage rates offered generally fluctuate with the best lending rates, which can be extracted from the database of the Hong Kong Monetary Authority (HKMA).3 Exhibit 4 contains the summary statistics of the sample set from the twelve housing estates.

Estimation Results

OLS Estimates of the FSIT Model

The Ordinary Least Squares (OLS) method is used for statistical validation of the FSIT Model outlined in Equation 3. The results in Exhibit 5 shows that the signs of all the estimated coefficients turned out as expected except the dummy variable, which represents the hidden forward risks. The coefficient attached to the spot market return, β^sub 2^, at 0.7905 is not only significant but also high, which demonstrates that the relative price changes of both the forward property market and the spot market are not only in the same direction but also with a very close degree of synchronization. The coefficient attached to the aging effect, β^sub 4^, is negative, which confirms depreciation due to the aging of properties. Regarding the coefficient attached to the hidden forward risks, β^sub 5^, it is interesting to find that the coefficient is the opposite sign of that expected (a positive sign) and is also statistically insignificant. Before drawing a conclusion on this issue, regression on the individual estates is conducted to examine whether there is an abnormality in any particular estate.

OLS Estimates of the FSIT Model on the Twelve Individual Estates

Exhibit 6 shows the OLS estimates of the twelve individual estates generated from the FSIT Model. Although the coefficients attached to their respective explanatory variables are not all significant, some insights can be drawn from the results. It is noted that the coefficients attached to the hidden forward risks of nearly all the estates are negative except for three estates: La Vista Villa (LV94), Kingswood Villa (KV95), and East Point City (EPC97). Upon further examination of the characteristics of these three estates, they have something in common-they were located in newly developed districts in which transportation and the necessary amenities for the community were limited when the presales were offered.

La Vista is located on Lantau Island and the only means of transportation was by ferries when it was constructed in 1994. Similar to La Vista, Kingswood Villa was located in an isolated area in Tin Shui Wai and East Point City was located in Tseung Kwun O in which the transportation means and amenities available were also limited when they were constructed in 1995 and 1997 respectively. By now, there are additional transportation modes to link the three districts to urban areas including highways and Mass Transportation Rails. The districts have also been developed into self-contained communities with all the necessary amenities available. As such, these three housing estates have violated the assumption that "no quality change" take place during the repeat sales period in terms of the amenities available and their accessibility and should, therefore, be excluded from the sample set.

OLS Estimates of the FSIT Model on the Revised Data Set

After taking out the three housing estates that violated the assumption of "no quality change" from the sample set, the nine remaining estates in the revised sample set are tested. The results in Exhibit 7 show that the signs of all coefficients attached to the explanatory variables are generated as expected including that of the hidden forward risks. The coefficient β^sub 2^ measuring the elasticity of the percentage change in the price difference of the forward-spot properties for a percentage change of the price difference of the benchmark spot properties is at 0.7869. It is not only significant but also high, which indicates that the relative price changes of both the forward property market and the spot market are not only in the same direction but also with a very close degree of synchronization. It should also be noted that the coefficient is less than one, which suggests that although the presale prices are tracking along the spot prices, the forward property market may be less volatile than that of the spot market. This agrees with the proposition made by Chau, Wong and Yiu (2003) that relaxation of presales help smooth out the price fluctuations in the spot market and thus help dampen the volatility in the property market as a whole.

The discount factor shows a positive coefficient of 1.298, which is evidence that a discount was embedded in the presale prices to compensate for the missing rent to cover the additional cost of capital invested during the construction time-lag. The aging factor shows a negative coefficient of -0.0266, which is approximated at a property value depreciation rate of 2.62% per annum.4 The negative coefficient of -0.054 attached to the hidden forward risks suggests that an extra premium is found only in the price changes of forward-spot pair sales, not in the spot-spot pair sales of the same properties. It supports the proposition that presale property buyers had paid an extra premium of approximately 5.25% on the presale property prices higher than the equilibrium prices of the same set of properties sold in the spot market.

OLS Estimates on Presale Properties Only

The effect of the hidden risks that are specific to the forward market is, therefore, absorbed in the intercept, γ^sub 1^, through the regressing process. Based on what the FSIT Model proposed-that an extra risk premium is imposed by the developers on the presale prices taking advantage of the asymmetric information issues-then a lower value should be obtained from the intercept generated from Equation 4, γ^sub 1^, compared to that of the FSIT Model, β^sub 1^, outlined in Equation 3 by the amount of the extra premium indicated.

The results of the OLS estimates on the forward-spot pair sales only (Equation 4) compared to that of the FSIT Model (Equation 3) on the same set of properties are presented in Exhibit 8. Exhibit 8 shows that the coefficients attached to the explanatory variables of the spot market return, aging impact and discount required between the two tests are very similar. However, the intercept generated from the forward-spot pair sales is -0.0243, which is very much lower than that of the FSIT Model at 0.0351, of which the impact of the hidden forward risks has been considered under a separate variable. The range of the differences is -0.059 [from 0.0351 (β^sub 1^ in E3) to -0.0243 (γ^sub 1^ in E4)], which is very close to β^sub 5^ of -0.054 in Equation 3. The result once again confirms the suggestion proposed by the FSIT Model that presale property buyers paid an extra premium of about 5% in the forward property market, which is higher than the equilibrium prices required in the spot property market, ceteris paribus.

The FSIT Model is considered efficient in pricing presale properties with a high adjusted explanatory power at 84% (Exhibit 7). The regression model itself is also significant. The multicollinearity issue among the explanatory variables is minimal as detected by the tolerance (TOL) tests and variance inflation factors5 (VIF). As shown in Exhibit 9, the TOLs of each of the explanatory variables with respect of the remaining explanatory variables are not close to zero and their VIFs are far from ten, which evidence that the degree of collinearity among the explanatory variables is low.

Conclusion

Understanding the pricing mechanism of presale properties is important for practitioners and investors conducting investment appraisals; however, there are limited studies of how the forward risks, in particular, the hidden forward risks, affect the pricing of presale properties. This study sets up a Forward-Spot Tracking Index-Tracking (FSIT) model for pricing presale properties, which can capture not only the risks arising from market uncertainty during the construction time-lag and the discount required to compensate for the additional cost of capital during the forward contract period, but also the possible premium imposed by developers on presale properties, which take advantage of the asymmetric information issues that are specific to forward property markets.

First, the findings indicate that the relative price changes in the forward property market of Hong Kong during the study period from 1993 to 2005 tracked closely to that of the spot property market. second, the findings reveal that presale properties were traded at a discount, which is positively correlated to the interest rate at the time the presale was conducted and the length of the construction time-lag to cover the additional cost of capital. Third, a risk premium averaged at about 5% is found in the presales of the uncompleted properties, which supports the proposition that developers could have charged a premium on the presales higher than the equilibrium market price, taking advantage of the asymmetric information problems in the forward property market. This is a valuable finding to property practitioners and demonstrates the extent of the risk premium that buyers might accept to pay on the presales in order to get the properties they desire from the forward property market.