Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Regression Discontinuity Manipulation

    I am considering whether some reasonable processes can be applied to a regression discontinuity design.

    I focus on the consequence of a corporate regulation: The running variable is income, let say the threshold is 100 million dollars, the treated firms are required to conduct an audit.

    The problem is:
    (1) There is no jump at the cut-off point (Stata shows that mass points gathering around the threshold),
    (2) some treated firms did not conduct an audit, and some non-treated firms conducted a voluntary audit,
    (3) I used Fuzzy RDD and the situation did not improve.

    My process is like, I only keep firms if their incomes always stay below or above the threshold during the sample period.

    By doing this, there appears a sharp jump at the cut-off.

    My reasons for this process are:
    (1) Some companies still conduct an audit if sometimes their incomes drop below the threshold, thus firms with fluctuating incomes will be as noises,
    (2) Like Donut RDD, this action drop firms that may manipulate their incomes.

    I would appreciate any suggestions or information on this process or my setting.
    Last edited by Stanley Huang; 17 Mar 2024, 05:23.

  • #2
    Any comments are welcome!

    Comment


    • #3
      One add-on:
      The number of dropped firms and remained firms have similar distribution in the running variable, which means that this action is likely to be random drop (does not drop companies with specific characters)

      Comment

      Working...
      X