• Ingen resultater fundet

CHAPTER VII - RESULTS OF STATISTICAL DATA ANALYSIS

7.1 Results of full data analysis

73

74 Table 14: Growth test statistics for the full sample

The results say a lot about the relative impact of PE ownership using different performance measures of growth.

The measure of operating performance in terms of FTEs suggests that portfolio companies at a 95% significance level over perform compared to the industry averages. More specific, growth in FTEs is on average 16.54% higher for portfolio companies relative to the industry averages.

It also appears from table 14 that portfolio companies over perform with regard to the performance measurement of revenue and underperform with regard to the performance measurement of assets, however, none of these measurements imply statistical significance.

Hence, we cannot conclude that the relative positive development in revenue and relative negative development in assets is due to PE ownership.

These findings are surprising due to several elements.

With regard to the result of no significance in growth of revenue, this was not expected due to the level of add-on investments in portfolio companies which was documented in chapter 2. One explanation that these add-on investments did not lead to a significant growth in revenue of portfolio companies relative to industry peers might be that mergers and acquisitions may not be limited to the PE market but is an increasing tendency in other markets

Growth test statistics [entry -1; exit +1]

Revenue

Observations 43

Median change (industry adjusted) 7.09%

Direction Increase

P-value (significance) 0.3280

Assets

Observations 43

Median change (industry adjusted) -1.99%

Direction Increase

P-value (significance) 0.8848

Employees (FTEs)

Observations 43

Median change (industry adjusted) 16.54%

Direction Increase

P-value (significance) 0.0490**

Note: Test statistics from Wilcoxon Signed Rank test. Median

changes are measured as percentage changes.

***, **, and * imply statistical significance at confidence levels 99%, 95%, and 90%, respectively.

75

as well (Appendix 7), thus add-on investments i.e. inorganic growth of the industry peers could possibly explain the lack of significance. In addition, the median change for growth in revenue could possibly be negatively affected by divestments in portfolio companies, which for instance was the case in F. Junckers Industrier A/S (Appendix 2 – Axcel Interview), and DAKO Danmark A/S (Appendix 1 – EQT Interview).

The finding of no significance in growth of revenue further contradict with the general growth focus of PE firms (Appendix 1-3) as well as the findings of previous studies e.g. Holm (2013), Jääskeläinen (2011), and Viitala (2012).

Asset growth is lower by 1.99% relatively to the industry averages, and this contradicts with our expectation cf. hypothesis #1. Vinten (2008) found that Danish portfolio companies are larger ex ante than the industry averages. Based on this finding one explanation to the lower asset growth of 1.99% of portfolio companies could be that since portfolio companies are larger, the lower relative asset growth may be explained by the fact that the industry peers are smaller and therefore possibly in a better position to grow.

From the tests statistics we can further see that the assets of portfolio companies increases (direction=increase), however not as much as for the industry averages. This increase could possibly be explained by the recognition of goodwill i.e. asset boosting post the entry (PwC, 2017). Asset boosting is a common feature in acquisitions and especially in PE-transactions where post-buyout goodwill is often written up. In our data, however, it does not seem to be the case since asset growth in portfolio companies is lower than for the industry averages cf.

table 14. It could be that asset boosting takes place at the holding company and not at the portfolio company level. Since there is little support for the asset boosting argument in our data, we do not find it relevant to adjust for goodwill effects.

As mentioned, the growth in employees is significant at a 95% significance level and supports the finding of Økonomi- og Erhvervsministeriet (2006) and Holm (2013), but violate the finding of Davis et al. (2014), however, the analysis of Davis et al. (2014) is not based on Danish data but US data, which should be taken into consideration.

This finding with regard to FTE growth is very interesting with respect to an often-highlighted critique of PE-firms that they violate the employment. Whether this significant finding can be attributed to gross vs. net increase in employment from a macroeconomic perspective will be further analyzed in chapter 10.

With the above growth test statistics in mind the results seems not to support Hypothesis #1 and we can hereby overall conclude that the null hypothesis 𝐻01 is accepted as not all of the

76

P-values of the three growth performance measures < significance levels. If instead hypothesis 𝐻01 only measured growth in terms of employees, we could reject the null hypothesis 𝐻01 on a 95% significance level as the P-value < 0.05 and conclude that the result is significantly different from zero at a 95% significance level and hence claim that portfolio companies grow significantly more than the average company of their industry over the entire holding period in terms of employment. However, as growth in employees do not weigh as much as growth in revenue when measuring operational value creation (Appendix 1 – EQT Interview), we cannot make such conclusion and thus the conclusion is that we cannot reject the null hypothesis 𝐻01.

7.1.2 Profitability metrics

Below table 15 presents the results from the statistical test of hypothesis #2 that a change to PE ownership leads to abnormal performance in profitability (measured in terms of EBITDA, EBITDA margin, EBITDA/Assets, and ROE) in portfolio companies, controlling for industry effects.

Table 15: Profitability test statistics for the full sample

Profitability test statistics [entry -1; exit +1]

EBITDA

Observations 43

Median change (industry adjusted) 28.57%

Direction Increase

P-value (significance) 0.3525

EBITDA margin

Observations 43

Median change (industry adjusted) 0.21%

Direction Decrease

P-value (significance) 0.8092

EBITDA/Assets

Observations 41

Median change (industry adjusted) -1.50%

Direction Decrease

P-value (significance) 0.8206

ROE

Observations 41

Median change (industry adjusted) -5.34%

Direction Decrease

P-value (significance) 0.5212

Note: Test statistics from Wilcoxon Signed Rank test Median changes are measured as percentage changes, except for

ratios, which are measured as percentage point changes.

***, **, and ** imply statistical significance at confidence levels 99%, 95%, and 90%, respectively.

77

Table 15 shows that portfolio companies realize a positive impact on profitability for 2/4 measurements i.e. EBITDA and EBITDA margin. However, neither of the median changes of any of the profitability performance measurements are statistically significant. Even though the impact is insignificant, the findings are notable and interesting since some of the previous empirical results show the opposite e.g. Vinten (2008).

Against our expectation, the statistical tests show that the development of the EBITDA margin and EBIDTA/Assets of portfolio companies points towards a decreasing direction in the event window. The decreasing direction of EBITDA/Assets and the negative median change of 1.50 percentage points may be explained by the absolute increase of the assets of portfolio companies cf. table 14. This finding is in line with the finding of Vinten (2008) who documents a negative development in profitability controlled for industry effect. With regard to the EBITDA margin, we initially assumed that a change to PE ownership would result in a significant increase controlling for industry effects, however, this cannot be evidenced through the Wilcoxon test. One reason for this finding might be the fact that PE-firms are willing to sacrifice profitability in the holding period if they are able to create a larger value creation in the long run:

“One of the strength of the private equity model is that we do not have to care as much of the performance in the holding period as listed corporations have to as our objective is rather to create the basis for future growth possibilities and a long run strategic position”

(Nicholas Hooge, EQT)

This way of thinking with regard to the profitability in terms of the EBITDA margin can theoretically be explained by the so-called J-curve where the EBITDA margin in the first years of the holding period is decreasing and later turning into a positive development (Grabenwarter & Weidig, 2005). However, the evidence of a J-curve in our data should be interpreted with caution as according to the theory, data should, all else equal, show an increase in EBITDA margin when using the event window of entry-1 to exit+1.

The measurement of growth in EBITDA is assumed to be a key element and focus for the PE firms when developing portfolio companies (Chapter 2; Appendix 2 – Axcel Interview). The median change in EBITDA in the event window is highly positive (28.57%) which is related to the result of the positive median growth in revenue of 7.09% cf. table 14. As the test statistics for the EBITDA growth is more positive than the test statistics for the revenue growth, this is an indication that PE-firms are relatively more efficient at cost cutting than creating sales growth, which is supported by the interviewed PE firms cf. Appendix 1-3, even

78

though none of the differences are significant and even though the interviewed PE-firms also states they are topline focused.

The return on equity is a function of ROIC, financial leverage, and cost of capital. The relative negative median change of 5.34% is a surprising result as the financial leverage in portfolio companies is in most cases higher than in the industry averages (refer to chapter 2). Due to a higher leverage in portfolio companies, it is easier to present a higher ROE, which contradicts our findings.

Note that the number of observations with regard to the testing of EBITDA/Assets and ROE is lower than for the other performance measures. This is due to the fact that for those measures no industry numbers for the year 2000 have been possible to conduct impacting 2 portfolio companies of our sample which have been excluded.

Concluding on the development of profitability of portfolio companies, we put most emphasis on the EBITDA margin due to the potentially asset boosting bias with regard to EBITDA/Assets. The result indicates a small improvement in profitability of portfolio companies in the event window, however, due to the insignificance we are not able to reject the null hypothesis 𝐻02. Thus we can conclude that the profitability of the portfolio companies, against our expectation, has not improved significantly different, controlling for industry effects.

This finding contradicts with the finding of Bergström et al. (2007), but supports the findings of Vinten (2008) and Holm (2013).

7.1.3 Productivity metrics

Table 16 presents the results from the statistical test of hypothesis #3 i.e. that the productivity of portfolio companies increases more than the average company of their industry over the entire holding period. The results put light on the relative impact of PE-ownership on employment effects and the utilization of assets.

79

Table 16: Productivity test statistics for the full sample

As explained in chapter 4 we have chosen to use the revenue-to-employees ratio as a proxy for labor productivity. According to hypothesis #3, we expected to see an increase in labor productivity relative to the industry peers. The data do show that the ratio is on average 7.09 percentage points higher for portfolio companies relative to the industry averages, however, this result is not statistically significant.

Even though a relative positive median of this ratio, the conclusion may be that we cannot reject the null hypothesis 𝐻03 due to insignificance. Hence, the improvement of the labor productivity cannot be credited to the change in ownership. This finding partly supports the study of Holm (2013), Lichtenberg & Siegel (1990), but is contrary to the finding of Davis et al. (2014).

Asset turnover is often interpreted as a proxy for managerial efficiency, i.e. the more sales the management generates from firm investments (assets) the better.

With regard to this ratio, we can document a positive median change (15.09 percentage points) of utilization of assets for portfolio companies over the entire holding period compared to the industry peers, but without significance. However, even though the relative positive median change is very close being statistically significant at a 90% significance level, we cannot reject the null hypothesis 𝐻03, and hence our data cannot evidence that PE-ownership lead to a higher degree of managerial efficiency in portfolio companies than in industry averages.

Productivity test statistics [entry -1; exit +1]

Revenue/FTEs

Observations 43

Median change (industry adjusted) 7.09%

Direction Decrease

P-value (significance) 0.3280

Asset Turnover

Observations 41

Median change (industry adjusted) 15.09%

Direction Increase

P-value (significance) 0.1067

Note: Test statistics from Wilcoxon Signed Rank test. Median changes are measured as percentage point changes.

***, **, and * imply statistical significance at confidence levels 99%, 95%, and 90%, respectively.

80