DEVELOPMENT OF SENTIMENT INDEX AND ITS EFFECTS ON GOVERNMENT LAND SALES IN SINGAPORE

Abstract

This study aims to develop a real estate sentiment index by tapping on an innovative source of data. Through sentiment analysis of land-related news articles, the tone of news articles are examined at the sentence-level to generate the sentiment scores. The newly constructed sentiment index is found to correlate well with macroeconomic events in Singapore. A test of the robustness of the sentiment index has proven that it is able to account for variations in GDP growth, hence it serves as a robust proxy of market sentiment. Following which, this study seeks to investigate the effects of sentiment on the Government Land Sales (GLS) programme using the sentiment index. The regression results have shown that GLS transaction volume is positively correlated with market sentiment, which is consistent with the profit-maximising behaviour of the government and developers. The sentiment index is also proven to be a useful prediction tool of GLS transaction volume in the subsequent quarter. The price premium of tender bids is found to be positively correlated with market sentiment, which indicates more aggressive bidding behaviour during positive market sentiment. However, no significant results are observed on the number of bidders and Coefficient of Variation (COV). A possible explanation may be that the prospect of a development site is of greater importance for developers in deciding to bid for a site as the market sentiment is subjected to frequent changes. As for the COV, the “bottom-fishing” behaviour of developers may have resulted in skewed mean and standard deviation.