Replication folder ECMODE-D-24-00441R3

Description

Title: Firm-Level Analysis of Bubble Formation in Chinese Real Estate Equities Economic Modelling This study investigates evidence of bubble preponderance in China’s real estate sector and seeks to identify the main determinants of exuberance in the equity prices of listed developers, relative to their dividend-based fundamentals. In contrast to the focus on property prices and rents that characterizes prior research, we emphasize real estate equity prices and firm-specific metrics. This shift in perspective, and the corresponding use of a dividend-based proxy, separates speculative-driven bubbles from those linked to fundamentals and thus enables us to better interpret the nature of exuberance as well as assess the alignment—or misalignment—between prices and fundamentals. Our empirical examination, based on the equity prices of 25 publicly listed developers included in the BICHODVP Chinese benchmark real estate index, detects bubbles in developer equity prices as well as the presence of common bubble dynamics among BICHODVP index components. Additionally, by incorporating firm-specific characteristics and macroeconomic variables, we provide a more granular understanding of how company characteristics—especially corporate valuation multiples and leverage—interact with broader market and policy conditions to generate equity price bubbles in the real estate sector.

Steps to reproduce

Readme file - Always run lines 1 to 19 before performing any analysis to load required R packages Bubble detection - Data loading -> run lines 23 to 50; remember to change folder to the one where you have saved data - Bubble analysis -> run lines 54 to 68; we recommend saving the environment after running simulations to save time in later analysis (line 71); you can reload the environment by running line 72 - Results -> can be obtained with different methods (montecarlo, wb and panel) running lines 74-86 - Saving results -> Lines 89 to 128 will save the results as xlsx files: 1) companies in table 3a in the paper will be named res_wb_1.xlsx and parameters for the MC specification can be found in res_mc_1.xlsx, 2) companies in table 3b in the paper will be named res_wb_2.xlsx and parameters for the MC specification can be found in res_mc_2.xlsx - Datestamping -> Lines 131-146 will produce datestamping data; 1) Tables 3c-3d may be reproduced by running lines 140 to 146), 2) Fig. 7a to 7h will be reproduced running lines 351 to 389 Logit - Data loading -> run lines 151 to 165; remember to change folder to the one where you have saved data - Correlation -> Table 4 can be obtained by running line 168; after analysing the correlation we disregard some of the variables (line 171) - Logit for price -> Run lines 174 to 206. Data for table 5a can be obtained by running lines 209 to 215. Please be aware that Step function in R stops after reaching an optimal AIC.To obtain the rest of the models run regressions iteratively with remaining variables (this is not included in the code). The authors run these additional regressions and used Excel files to compile the data. - Logit for dividends-> Run lines 221 to 260. Data for table 5b can be obtained by running lines 262 to 269. Please be aware that Step function in R stops after reaching an optimal AIC.To obtain the rest of the models run regressions iteratively with remaining variables (this is not included in the code). The authors run these additional regressions and used Excel files to compile the data. Rest of the charts - Figure 2 -> Run script in lines 277-294 - Figure 3 -> Run script in lines 297-317 - Figure 4 -> Run script in lines 320-328 - Figure 5 -> It is not included in the code; the authors used Excel - Figure 6 -> Run script in lines 331-337

Institutions

Categories

Funding

Licence