E-mail Chain on Winner's Curse

• Jim to Ed-

Hi Ed,

I have been looking at the referee's comments more carefully. We have only a three month time horizon and a fair amount of work to do. We should start very soon if you can liberate an RA to work on the paper. Please see my comments below interspersed in the referee comments.

I hope, as I said before, that this is project to which you can legitimately assign one or even two RAs to work on. Most of the early work needed is very RA-intensive. If that is a problem, please let me know.

More generally, please let me know what you think of my comments below.

Thanks,

Jim

Reviewers' comments:

Report on MS 16-00014

The paper seeks to explain whether acquisitions of private targets are characterised by a winner's curse. To do so, the authors consider a simple of acquisitions between 1985 and 2011 and study the announcement returns around the acquisition. they find evidence that there is a winner curse that is higher with informational asymmetry proxies, and with overconfidence on the part of the acquirer.

The authors should put more emphasis on why it is important to prove there is a winner curse in private acquisitions? what are the implications for the existing literature? if WC characterises both public and private targets acquisitions, what do the readers get out of it? in one word, more motivation is needed here. The contribution of the study should also be highlighted.

I would be happy to rewrite this section after we do the additional analysis.

The authors should run robustness checks using for instance the malmendier and tate proxy for overconfidence.

It is a lot of work, but if you can liberate RA time we should download the basic data again. Given the amount of extra data we need we might as well re-do everything anyway. This is clearly something an RA would need to do. We should also download the Malmendier and Tate proxies for overconfidence. I am not sure what they are but I think we cite their paper in our paper and the RA could check out that paper and remind us what the proxies for overconfidence are and download them.

In the model, there are no controls for the corporate governance characteristics of the bidder: this is a potentially important variable in determining announcement returns.

I am not sure what corporate governance characteristics should be downloaded. Perhaps the RA could start by doing a literature review to see what corporate governance explanatory variables are used in finance papers as control for acquisitions and/or other decisions of interest.

The model should also alternatively control for the competitiveness of the sector, as a determinant of announcement returns.

We might need to create Herfindahl indexes as I think we have done before. But it might be worth getting the RA to do a literature search to see what competitiveness controls are normally used in Finance papers.

the authors place a strong emphasis on the weighted results, but do not justify why this is the metric one should focus on .

I can write up a response to this.

Despite the length of the study period, no analysis is run to determine the impact of the crisis or the IT bubble.

If we download a few more years of data (up to 2014 or 2015?) then this makes more sense. We can control for these events when we run regressions. One approach is to just use fixed effects. But we could also use some quantitative measure like changes in GDP.

In the model, the authors should also consider whether the acquisition was friendly or hostile, as this conditions the announcement returns.

I hope it is not too hard to categorize mergers into friendly and hostile. I am not even sure what that means for acquisitions of private companies. Maybe an RA could see if there is anything in the literature on this.

the measure of overconfidence captures frequent acquirers, but not whether the acquirers buy private or public firms.

I am not sure about the implications of this comment. I guess it means that the overconfidence measures used by Malmendier and Tate include frequency of acquisitions, and we should distinguish between frequency of private vs. public acquisitions.

Finally, the study would have gained insights should the authors have considered all acquisitions as in Faccio et al (2006) and then compared private targets and public targets. This would tackle the section biais issue.

This requires a big data download. But I guess we should do this comparison if it is feasible.

• Ed to Jim-

Hi Jim,

The biggest problem with the R&R is going to be getting the Malmendier and Tate proxy for overconfidence. Here's a link to their papers. https://scholar.google.com/scholar?hl=en&as_sdt=0,22&q=malmendier+tate+overconfidence

The source paper, and two papers that talk about the data, is attached. We might be able to do this with Execucomp or the Capital IQ data. This is available through wrds (https://wrds-web.wharton.upenn.edu/wrds/about/WRDS%20FAQs.cfm) but I'm not sure whether Rice has a subscription to this WRDS dataset. I'll check tomorrow. The second hardest thing is the measures of corporate governance of the acquirer. I'm still looking into this.

Essentially these two things could be deal-breakers. We might not be able to get the data. The rest is just time and effort. Whether the acquisitions are hostile or not is in SDC. Herfindahl or CR4 measures, or other measures of concentration, are pretty easy and we've done them before. And doing public acquisitions and extending the dataset isn't hard, it's just time consuming.

I can absolutely put several RAs on this. We will have a bit of a timing problem - my current set are about to end and I'll have a gap until I get others who will be on fewer hours - but the logistics aren't prohibitive. Ed

Addendum:

I have access to Execucomp! It goes back to 1992 and covers CEO options. https://wrds-web.wharton.upenn.edu/wrds/ds/comp/execcomp/stgrts/index.cfm?navId=72

I have also have access to the BoardEx Annual Option Compensation data from 1999 to present: https://wrds-web.wharton.upenn.edu/wrds/ds/boardex/na/options_compensation_drilldown/index.cfm?navId=23

The governance side still looks problematic. I have access to ISS (Risk Metrics) data on governance from 2007 forward: https://wrds-web.wharton.upenn.edu/wrds/ds/riskmetrics/rmgovernance/index.cfm?navId=246

But maybe we can find some governance measures elsewhere. I vaguely remember that the fraction of ownership by institutional investors was used as a popular governance measure...

Ed

*Jim to Ed-

Hi Ed,

Thanks very much for this message and for your previous email message. If I understand correctly all the data is now potentially covered except for corporate governance. I admit that I do not know what measures the referee has in mind.

I just checked on Google Scholar and found many recent finance papers that use measures of corporate governance. Here are three that seem very relevant:

1. Masulis, Ronald W., Cong Wang, and Fei Xie. "Corporate governance and acquirer returns." The Journal of Finance 62.4 (2007): 1851-1889.

2. Bebchuk, Lucian, Alma Cohen, and Allen Ferrell. "What matters in corporate governance?." Review of Financial studies 22.2 (2009): 783-827.

and

3. Wang, Cong, and Fei Xie. "Corporate governance transfer and synergistic gains from mergers and acquisitions." Review of Financial Studies 22.2 (2009): 829-858.

Also corporate governance is discussed in

4. Moeller, Sara B., Frederik P. Schlingemann, and René M. Stulz. "How do diversity of opinion and information asymmetry affect acquirer returns?." Review of Financial Studies 20.6 (2007): 2047-2078.

Here is another random example:

5. Bhagat, Sanjai, and Brian Bolton. "Corporate governance and firm performance." Journal of corporate finance 14.3 (2008): 257-273.

The following paper contains a measure that seems to be widely used. I am not sure how hard it is to obtain or construct.

6. Gompers, Paul, Joy Ishii, and Andrew Metrick. "Corporate Governance and Equity Prices." The Quarterly Journal of Economics 118.1 (2003): 107-155.

The first task for an RA is to look at the literature and identify the measures of corporate governance that are commonly used and where they can be obtained or constructed.

So, given that overconfidence is not a deal-breaker, the next step is to see what we can do on corporate governance. I suspect that there will be some feasible possibilities. But if you can liberate an RA for a few hours, I would suggest this task (i.e. check the finance literature for corporate governance measures used, particularly in the context of acquisitions, and determine where the data comes from).

Thanks,

Jim

• Ed to Jim & Jake-

Hi Jake,

Here's the email from Jim (cc'd below). Please:

1a) Get the BibTeX and papers for the refs below 1b) Maybe do your own dig, perhaps using these papers as a starting point, for any other key papers. But don't get carried away. 3) Work out what measures are commonly used and how they are built 4) Whether we can get the data to build these measures for publicly traded companies, and for what data ranges (and subject to what restrictions, etc.).

Speed is more important than prettiness or perfect comprehensiveness here. But please do build yourself a wiki page to log your process and findings.

Thanks very much for offering to do this. I really appreciate it.

Ed

cc. Jim - Jim meet Jake. Jake has offered to take a quick crack at this for us. He's been working with me over the summer on an LBO innovation project and will hopefully stay with me next term and work on this R&R too. Also, to get you set up on the new wiki asap could you please go to www.mcnaircenter.org and click "request an account" in the top right corner?

Jake makes Corporate Governance Measures Page

• Jim to Ed & Jake-

Thanks, Jake

This is very helpful. I think we have a good idea of what the major measures of corporate governance are. The next question is how we could obtain these measures.

Just to make sure you understand the big picture, what we want is a corporate governance variable with a value for each acquiring firm in our database. Possibly it might change over time if a given firm changes its corporate governance.

I am a little confused about the availability of measures. Answers to the following question would be helpful.

1. Are there any “off-the-shelf” measures maintained by credible third parties (such as TCL – The Corporate Library)? You indicate a TCL benchmark score that goes up to 2003. Is there anything more recent?

2. Do most of the papers construct their own indices? How difficult would it be for us to do something similar? Could we, for example, construct a GovScore variable fairly easily by downloading data from ISS?

Bottom line:

What would be the best and/or easiest way for us to obtain one or more corporate governance variables that we could use.

Thanks,

Jim

• Jake to Jim & Ed-

Hi Jim,

The vast majority of papers use the GIM index or some variation that is created as a subset of the original GIM index. Each index is constructed using the Investor Responsibility Research Center data from Corporate Takeover Defenses Rosenbaum 1990, 1993, 1995, 1998, 2000, 2002. I can only locate physical copies of these books, so I assume research assistants put together databases by hand using those physical copies.

In order to obtain this data, I would contact the authors of one of the more recent papers (Masulis et al 2011, for example) or the IRRC at http://irrcinstitute.org/contact/. It seems as though other authors contacted Gompers et al (2003) for the IRRC data, so that could also be a lead.

The TCL benchmark score and the BC Govscore only seem to be used in one of the major papers about corporate governance. If you want to use the TCL benchmark score, you need a subscription to the database through a company called MSCI, which you can contact at https://www.msci.com/contact-us. Rice does not have access to this database.

The BC Govscore comes from the Institutional Shareholder Services (ISS), and you can find the specific database at https://www.issgovernance.com/solutions/iss-analytics/quickscore/.

Papers on the subject, even recent ones, indicate that the corporate governance data doesn't go past 2003. It is possible that some of these databases could have more recent data, although I couldn't say if that is the case because I don't have access to the databases.

I hope this helps and that the email isn't too messy to digest.

Regards,

Jake

• Jim to Ed & Jake-

Thanks, Jake

Ed, I am not sure what is the best way to proceed. Corporate Governance is not our main issue so all we need to do is something reasonable. However, if we have data that only goes up to 2002 or 2003 I am not sure that would be enough. I assume that many of the acquirers in our data did not even exist in 2002, especially if we extend our data to 2014 or 2015.

I checked the IRCC website. I am not sure what they have but I am bit surprised they do not have something more recent than 2002. Jake, what is the relationship between Rosenbaum and IRCC?

Another possible source is the ESG database maintained by MSCI. I am not sure if that is related to the MSCI information mentioned by Jake. I have data from the MSCI ESG database up to about 2011 or so, so that might be useful.

So, here are some possible actions.

1. I would suggest that it would make sense for Jake to ask IRRC if they have more recent information of this type.

2. Failing that, we probably should ask for the GIM data previously used by others. That request might most reasonably come from Ed given his role at the McNair Center and as co-author on the paper. I am not sure which author we should ask.

3. Jake, how much work would it take to reconstruct the GIM index from the hard copies.

4. Jake, it might also make sense you to look into the MSCI ESG database. You can start with this website: https://www.msci.com/esg-ratings then click on the factsheet. In 2014 they had a study on corporate governance.

So the question is “what data do they have” and “how much would they charge to sell that data to an individual academic user”? This is not likely to be helpful, but it is worth finding out.

Ed, what do you think?

Thanks,

Jim

• Jake to Jim & Ed-

Hi Jim,

As it turns out, we have access to the MSCI ESG databases of Takeover Defenses from 2005-2014. We can use this database to construct the ATI index described in Cremers & Nair 2005.

I've updated the wiki page to include the relevant variables listed in each database.

Please let me know if you have any questions!

Regards,

Jake

• Jim to Ed & Jake-

Thanks, Jake

This is good news. Even if we have nothing else, this measure is probably enough to go forward.

But we should still look into the possibility of obtaining or reconstructing GIM data.

Ed, could you take the lead in discussing this with Jake. I am not sure what the best way to proceed is and we might conclude that it is not worth trying to do anything with GIM data. But we should at least check it out. Even if we only have GIM data from pre-2003 or so, we could simply assume that each acquiring company has whatever it most recent GIM score is and we could use an indicator variable if a company has no GIM variable.

Jim