Closed JNing0 closed 2 years ago
Excellent, Jie! Very clear picture of effects.
How did you define, bi-,tri-, and quad-? If t is the current quarter, are they over the following, preceding, or one before and some after quarters?
@vob2 It is defined such as the sum of delays in the current quarter and the preceding n-1 quarters. If you look at the x-axis of the figures, the quarterly plot starts from 1st quarter of 2010, the bi-quarterly plot starts from 2nd quarter of 2010 as the first point includes the delays in the 1st and 2nd quarters of 2010, etc.
Thanks, Jie. Could you please clarify how you define "non-zero" delays in these plots? What I mean is the following. Suppose a project is 3 quarters long, and had a delay of 10 days in the first quarter but no delay in the next two quarters. Would this project be included in the plots for bi- and tri- quarterly delay?
@vibhuti6 Yes, it will be included. The delay over N quarters is computed using the original quarterly data with zero delays. After aggregation, I take out the nonzero N-quarter delays and plot the figures.
Great, thanks -- just wanted to confirm!
Could you please also plot the figures that include zero delays? Just wondering if the results look different in the two cases!
I did some preliminary figures about the nonzero delays over N quarters (N=1,2,3,4). The figures suggest no "corrective behavior" from the firms over time. The effect of QP seems persistent and may even accumulate over time. The figures also provide some ideas about running linear regressions.
Please review and comment. Thanks!