Which is a better investment choice in the Hong Kong residential property market: a big or small property?

Abstract

The relationship between property size and property investment yield is an interesting issue in the real estate market. Previous studies usually use the mean–variance criterion to compare the return-risk profiles of the yields of different property sizes in the United States. However, this criterion has a few shortcomings. This article provides the first attempt to use a stochastic dominance approach to analyse this issue. We adopt two powerful stochastic dominance tests to compare the yields of five property size classes in the Hong Kong residential property market. In our study, we analyse two possible investment outcomes: (1) investors could not rent out their properties, and thus they would gain/lose from the appreciation/depreciation of residential property prices; (2) investors could also gain from rental incomes. Our empirical results provide strong evidence to show that the yields of smaller property classes stochastically dominate the yields of bigger property classes, suggesting that buying smaller properties is a better investment choice in the Hong Kong residential property market.

Keywords:

• mean–variance criterion
• stochastic dominance
• property size
• Hong Kong residential property market

JEL Classification:

• C12
• C14

Acknowledgement

The authors are grateful to the Editor and two anonymous referees for helpful comments. This research is supported by University of Macau and Hong Kong Baptist University.

Notes

¹ Property prices of Hong Kong are also among the most volatile in the world (Xiao and Liu, 2010).

² The Hong Kong residential property market is also known for its low vacancy rate. Even after the Asian financial crisis it peaked at just 6.8%. In recent years, the vacancy rate has been trending down and touched a decade low level of 4.1% at the end of 2013, which is below the estimated natural vacancy rate (4.35%) in Hong Kong (Tse and MacGregor, 1999).

³ Hong Kong Island and Kowloon are the very well-developed commercial centre of Hong Kong, while the New Territories is a more recently developed district.

⁴ The Jarque-Bera statistics reported in Table 1 indicate that many of the property yields we examined are not normally distributed, suggesting that the conclusion drawn from the MV results may be misleading.

⁵ In our sample, the mean of the yield of the biggest size property (Class E) is higher than that of the smallest size property (Class A), while the standard deviation of the yield of Class E is also higher than that of Class A. Therefore, the MV criterion could not indicate which property size is a better choice for investors.

⁶ A possible disadvantage of the SD approach is that the SD test result may not be reliable if there are too few observations.

⁷ ‘X almost stochastically dominates Y’ means that, except for a very small probability, or very small portions of the distribution, say, 1%, the distribution of X lies below that of Y.

⁸ To save space, we do not report the DD test results for all 100 grid points as introduced in Section IV. We report only the results for grid point 11 to grid point 60, which provide enough information to make relevant inferences. The results for other grid points are available upon request.

⁹ To save space, we do not report the DD test results for all 100 grid points as introduced in Section IV. We report only the results for grid point 11 to grid point 60, which provide enough information to make relevant inferences. The results for other grid points are available upon request.

¹⁰ The findings for investment outcome 2 are actually the same as those for investment outcome 1.

¹¹ Following Linton et al. (2005), Cho et al. (2007) and Fong (2010), we draw our statistical inference based on the median value of simulated p-values.

¹² The LMW test results for investment outcome 2 are similar to those for investment outcome 1. The only difference is that, for Hong Kong Island, Class A dominates Class B at the third order for investment outcome 1, while Class A dominates Class B at the first, second, and third orders for investment outcome 2.

¹³ Investments in the residential property market are subject to investment costs. When we calculate the investment yields in Section II, we do not deduct various costs because we do not have complete and detailed data about the cost part. These costs include stamp duties, property tax, facility costs, management fees and homeowner association fees. This could be a limitation of this study. But we think this will not change our findings qualitatively for two reasons. First, all of these costs, except property tax, favor the owners of smaller properties. Second, as for property tax, it is uniformly applied to the rental incomes of all property size classes (e.g. the current property tax is 15% of rental income, regardless of whether it is a small or a big property). Therefore, the empirical results using the yields with these costs deducted (subject to data availability) will strengthen our conclusion that investing in smaller properties is a better choice.