Abstract
This paper studies the seller’s agency choice and the financial consequences of this decision. Specifically, at sale, sellers can choose to use either the same agency they originally purchased the house from, or a different agency. Since the same agency sold the property before, it has an informational advantage by knowing the property and principal. Using housing market transaction data in Sydney, the largest capital city in Australia, we show that sellers make their agency choice based on past purchase experience, non-salient features of the property, and the current market shares of the agency. Furthermore, our analysis of pair sale transactions reveals that using the same agency results in a 1.1–1.4% return discount compared to using a different agency. The findings support agency theory, which suggests that using the same agency may not be financially beneficial for the principals due to its informational advantage.
Acknowledgements
We acknowledge housing transaction data provided by the Rozetta Institute (formerly Securities Industry Research Centre of Asia-Pacific (SIRCA)) on behalf of CoreLogic. We also would like to thank Claire Shi for excellent research assistance in this paper. We are responsible for any remaining errors. © 2021 Song Shi and Junji Xiao. All Rights Reserved.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1 We are not studying the dual agency problem in this paper.
2 As agency firms can operate in multiple suburbs, the number of agency firms at the suburb level is higher than at the city level. The drop in the agency number in 2017 was mainly because we don’t have a complete sale data in 2017. Our data ended in September/October 2017.
3 CoreLogic is a leading property data provider in Australia, New Zealand, and the United States. It provides detailed residential property transaction data that is widely used by lending institutions, mortgage brokers, and property professionals such as real estate agents and valuers (appraisers) in real estate transactions. This dataset has been widely recognised and trusted in the industry for its accuracy and reliability. For more information about CoreLogic and its data services, please visit its website at https://www.corelogic.com.
4 If multiple transactions are observed with a particular property, any two consecutive transactions are defined as pair sales. For example, if one property is sold three times over the studied period, there will be two pair sales generated from this property.
5 The notation t − indicates the purchase time. As the duration of ownership varies across sellers, we leave the duration open.
6 We excluded the purchase agency value, type of purchase method, and HHI from the return regression as they affect the principal’s agency choice but not the current sale price.
7 We combined houses and units to increase the sample size in the pair sale analysis and used a dummy variable of property type to capture the behavioural difference in agency choice between the house and unit sellers.
8 0.011/0.033 = 0.333
9 The diverse nature of the Sydney housing market implies that various submarkets could develop their own price trends over time (e.g., see Pawson & Martin, Citation2021). To account for this dynamic heterogeneity across different regions, we have further grouped the Sydney metropolitan area into five main regions based on the Greater Sydney Region Plan. We then included interaction terms between year dummies and each main region in the pair sale analysis. Adding sale methods or regional time trends to Table 4 does not change our results. These results are available upon request. We further examined how neighbourhood characteristics can influence the principal’s agency choices and subsequent sale prices, using census suburb demographics as a proxy in Section 5.5.
10 The duration is retrieved from https://renovationjunkies.com.au/how-long-does-it-take-to-renovate-a-house-2/.
11 We first carry out the above hedonic price analysis without agency to obtain residuals and then we regress the residuals obtained in the first step on a constant and agency fixed effect. The value of R squared is about 10% for both house and unit regressions.