Experimental Design
This field experiment studies how dividend-related information and dividend tax information affect investors’ participation in dividend events. The experiment is organized around three key information nodes in the dividend process: (1) the announcement of a dividend proposal, (2) the announcement of the dividend implementation plan, and (3) the reminder related to the record date.
The experimental sample is determined before the intervention begins. At a fixed baseline date, we identify investors who hold at least one dividend-related stock. These stocks form a fixed stock pool for the experiment and are not updated during the intervention period. Investors with missing gender, age, or risk-tolerance information are excluded. Based on this sampling procedure, approximately 100,000 investors are selected through stratified randomization.
After the eligible sample is constructed, each investor is assigned to one prompted stock from the fixed stock pool and then randomly assigned to one of seven groups: three treatment groups, three placebo groups, and one control group. The treatment groups receive messages containing both dividend information and dividend tax information. The placebo groups receive messages containing only dividend information. The control group receives no message. Once the sample, group assignment, stock pool, and prompted stock are fixed, they are not adjusted according to subsequent changes in investors’ holdings.
The information interventions are implemented at the three pre-specified dividend information nodes. The first treatment-placebo pair receives information about the dividend proposal announcement at a fixed intervention date. The second treatment-placebo pair receives information when the prompted stock announces its dividend implementation plan. The third treatment-placebo pair receives information related to the record date when the implementation plan, including the record date, is disclosed.