Kimberly

Comments#

Ook heb ik tevergeefs geprobeerd data van WRDS af te trekken, want ik kreeg alleen maar errors. Heb jij wellicht tips hoe ik dit het beste aan kan pakken zo?

  • Er was (is) een mankement met WRDS en de EUR - voor zover ik weet zijn ze daarmee bezig, de errors zijn dus niet jouw schuld!

Extend paper to all economic downturns instead of only Covid –> Differences between crises?

Hier zou ik geen tegenstander van zijn, maar het is ook geen vereiste.

  • Main research question: wees precies, specificeer de announcement returns van welke firm, en de CEO’s van welke firm! :)

But negative acquirer CARs are not unambiguous proof of overbidding. Acquisition announcements deliver information not only about the transaction itself but also about the acquirer’s current condition and strategy (the revelation effect). Akdogu (2011), for example, emphasizes that acquisitions can be undertaken in response to competitive pressures of which the market is unaware prior to the bid. In such a circumstance, negative acquirer CARs are compatible with value creating transactions because, in the absence of the transaction, acquirers would have been even worse off.

I agree, but there might be a way to proxy for those effects, by e.g. taking into account tangibility of a firm’s assets, as well as other control variables.

It states 8 weeks bid premium but states t-42? In other literature it refers to ‘one month’ which is not 42 days? Could I perhaps use this 8 weeks as a robustness check?

Yes, you can use different intervals, and it is good to use 2 or more for robustness. Do you use the closing price as the price at $t-42$ or $t-30$?

  • You did not chose the best example for a regression table. You can use stargazer in R, or outreg in Stata to generate nice regression tables. We can go over some examples if you want.

  • I think you can run the regression(s) with different intervals for the bid premium, and then overconfidence + covid + interaction effect in subsequent regressions.

  • Then, you can do the same thing for regressions with different CAR-intervals.

  • Why use only control variables in the CAR analysis and not in the BP analyses?

Cumulative abnormal returns berekenen#

2 Stappen: estimation period and event period

  • Estimation period, per bedrijf twee variables: $R_i$ & $R_m $

  • Evaluation period: ook per bedrijf, $AR_it = R_it - \mathbb{E}[R_it] = R_it - \beta_i \cdot R_{mt}$

  • (met het CAPM)

  • Daarna, cumulative abnormal returns berekenen:

  • In R:


df %>%
    group_by(company_name) %>%
    arrange(evttime) %>%
    mutate(car = cumsum(abret))