Existing literature proposes several proxies to measure private equity group reputa- tion.332 This study constructs all relevant measures, which are described in the follow- ing. A common measure for PE group reputation is the size of the PE group measured by assets under management (PE group size).333 This study constructs the size variable by aggregating the capital committed to the funds of the PE group over the last five
332
For an overview see Demiroglu/James (2010b).
333
Cf. among others see Demiroglu/James (2010b).
Deal year Number of
deals Supermajorities Yank the bank Snooze and lose
Equity cure right (injection used to
pay back debt)
Equity cure right (injection used to increase EBITDA)
Creditor rights dilution index
(frequ.) (frequ.) (frequ.) (frequ.) (frequ.) (mean)
2000 1 0 0 0 0 0 0 2001 2 0 0 0 0 0 0 2002 6 1 1 0 0 0 0.33 2003 7 0 1 1 0 0 0.29 2004 20 3 9 4 3 2 1.05 2005 28 2 10 6 8 0 0.96 2006 30 9 9 5 13 2 1.23 2007 29 11 14 10 21 7 2.03 2008 7 2 7 3 7 0 2.38 Total 130 28 51 29 52 11 1.29
years before the transaction. This procedure mirrors the one applied by Private Equity International. The fund data is obtained from Thomson Venture Economics (TVE). Table 25 Panel A reports the distribution for the PE group size variable in the LBO dataset. The average PE group had €3 billion under management at the date of the deal (median €1.3 billion). Due to the inclusion of very large PE groups like e.g. KKR, Bain Capital, Goldman Sachs Capital Partners, Carlyle, among others, the standard deviation is €5 billion.
Table 25: Descriptive statistics for sponsor related variables
Source: LBO dataset
Another way to measure PE group reputation is proposed by GOMPERS/LERNER (1999). They argue that investors learn about the characteristics and behavior of pri- vate equity organizations over time. Therefore this study also uses the lifetime of a PE group as another proxy for reduced information asymmetry costs (PE group lifetime).
Panel A: Descriptive statistics of LBO sample
Obs. Unit 25% Median Mean 75% SD
Sponsor related
PE group size 130 € millions 91.3 1,316.4 3,033.5 3,936.4 4,961.1
PE group lifetime 130 no. 6.0 13.0 15.6 24.0 12.3
PE group number of deals 130 no. 5.0 11.0 17.4 28.0 17.3
BR volume 130 % 0% 16.3 23.2 34.7 26.5
BR number of deals 130 % 0% 13.9 18.8 25.0 23.1
BR number of deals (abs) 130 no. 0.0 1.5 3.3 4.0 4.7
Panel B: Correlation matrix for reputation measures
Information asymmetry measure
Information asymmetry measure (1) (2) (3) Description
(1) PE group size 1 LN(1 + PE group size)
(2) PE group lifetime 0.69 *** 1 LN(1 + PE group lifetime)
The age of the private equity organization at transaction date is aggregated by using the fund database of TVE. The year of the first fund is then compared to available pub- lic information coming directly from the PE group. If the data from TVE and the pub- lic information do not match, the study uses the information provided directly from the PE group. Panel A reports the distribution for the PE group lifetime variable in the LBO dataset. The average PE group in the sample has a lifetime of 15.6 years (median 13.0).
The frequency of interactions with the syndicated loan market serves as a third meas- ure for PE group reputation (PE group number of deals). The number of transactions a PE group was able to complete before the respective deal in the sample approximates the frequency of interactions. The variable counts all deals recorded in TVE and Mer- gerMarket during three years prior to the transaction. Table 25 Panel A reports the dis- tribution for the PE group number of deals variable in the LBO dataset. The average deal in the sample was sponsored by a PE group which completed 17.4 deals before the respective transaction (median 11.0).
Table 25 Panel B presents the correlation matrix for the measures of information asymmetry costs. The individual measures have been logged because the economic characteristics suggest a diminishing marginal effect of the variables. For example, a completed deal is more important when the PE group has only two completed deals as compared to when it has completed 22 deals in the last years.
The analysis shows that the three measures for private equity group reputation are highly and significantly correlated (between 0.62 and 0.77 all being significant at the 1% level). Although there is a high correlation, the regression analysis tests the results for robustness for each of the reputation proxies. As reputation is a latent construct, the authors also constructed a latent reputation variable that included the information of all three proxies. However, since results remained robust and the economic interpretation of the reputation proxies is easier, this study does not report the results for the latent variable.
Similar to IVASHINA/KOVNER (2010) and BHARATH et al. (2007) this study measures bank relationship by the Euro value of loans in the previous five years un- derwritten by the bank functioning as lead arranger divided by the total Euro value of all loans sponsored by the respective private equity group as reported in Dealscan (BR volume). In case more than one bank functioned as lead arranger, the highest relation- ship value is selected. Table 25 Panel A reports the distribution for the BR volume var- iable in the LBO dataset. The average deal is sponsored by a PE group that has a BR volume of 23.2% with the bank functioning as the lead arranger (median of 16.3%). This means, the average deal is sponsored by a PE group that financed 23.2% of its previous loans with the same lead arranging bank of the deal.
For reasons of robustness, the author also computes an analogous bank relationship variable that results from dividing the number of loans of the lead arranging bank sponsored by the respective PE group in the five years prior the transaction divided by the total number of loans sponsored by this private equity group as reported in Deals- can, Thomson Venture Economics and Mergermarket (BR number of deals). Table 25 Panel A reports the distribution for the BR number of deals variable in the LBO da- taset. On average BR number of deals equals 18.8% (median of 13.9%).
Additionally, a simple count variable is aggregated (which is logged in the regression), summing the number of loans underwritten in the previous five years by the respective lead arranging bank (BR number of deals (abs)). As the reduction of information asymmetry is a learning process, it may likely result from the absolute frequency of interactions between bank and the PE group. However, as noted by IVASHI- NA/KOVNER (2010) this measure for bank relationship conflates with measures of PE group reputation. Table 25 Panel A reports the distribution for the BR number of deals (abs) variable in the LBO dataset. The average PE group in the sample complet- ed 3.3 transactions in the previous five years before the respective deal with the re- spective lead arranging bank (median 1.5).
One obvious point of critique is that the variables measuring banking relationship merely capture the view of the PE group. More specifically, the variables measure how the PE group evaluates the relationship to the bank but not how the bank sees the PE
group what would be more important in the eyes of the author. These variables meas- ure how important the bank is for the PE group but not how important the PE group is for the bank. However, due to the incomplete data on historical loans of the banks, this analysis is not possible for this dataset.