4 C LASSIFICATION AND T YPES OF C OVENANTS
6.5 Determinants of Financial Covenant Restrictiveness 163
6.5.2 Multiple Regression Analysis 165
6.5.2.1 Target related 169
Model 1 shows that increased profitability, measured as EBITDA/sales339, at transac- tion (t = 0) leads to significantly (at 10% level) less restrictive covenants, i.e. higher headroom. This finding suggests that lenders consider achieved profitability as a signal that future behavior of management will be positive and therefore lenders grant more flexibility to management in terms of less restrictive financial covenants. An increase in the EBITDA margin by one percentage point or one standard deviation (9.7 per- centage points), increases the leverage headroom by about 19 basis points or 187 basis points, respectively. The results for profitability are robust across all models.
Next, the analysis provides evidence that financial covenants are less restrictive for high growth firms, i.e. they receive higher D/EBITDA headroom. This result is signif- icant across all specifications, mainly at the 5% level. Accordingly, increasing the ex- pected EBITDA growth forecast over the first three years after the buyout by one per- centage point or one standard deviation increases the D/EBITDA headroom by 30 ba- sis points or 173 basis points, respectively. Substituting EBITDA growth by expected sales growth yields robust results.340 Whereas theory predicts different directions, the author finds direct evidence that shareholders and management of high-growth firms value flexibility more than shareholders of comparable low-growth firms. At first, this finding appears to contradict the results of many scholars, e.g. BILLETT/KING/ MAUER (2007)341, who suggest that higher growth firms receive more restrictive cov- enants. But most past studies evaluate covenant protection by the mere counting of action-restricting covenants in the context of agency theory.
339
In the view of the author EBITDA/sales represents a better proxy for profitability than EBITDA/ assets for LBOs because assets represent the purchase price at t = 0. Therefore, EBITDA/assets would be the reciprocal of the EBITDA multiple of the enterprise value and highly negatively cor- related with leverage, leading to multicollinearity problems. However, including EBITDA/assets does not change the results and profitability remains significant in all models.
340
The market-to-book ratio is difficult to determine for non-public private equity transactions because (1) the book value before acquisition is not known and (2) even if it were known, the corporate structure changes significantly in a buyout disqualifying the historical book value of equity.
341
Also, Billett/King/Mauer (2007) find a significant relationship for the growth proxy market-to- book, although they did not find any significant relationship for their sales growth proxies.
This research design allows to draw direct conclusions regarding the potential shift of control rights, which are not in conflict with action-restricting covenants, but are com- plementary and might adhere to a different economic logic. Previous literature shows that high growth firms receive more action-restricting covenants; however, the analy- sis shows that they negotiate greater flexibility for their financial type covenants. Nev- ertheless, the significant influence of growth expectations on headroom could also have a mere technical reason in leveraged loan contract design, which can be ex- plained by way of an example: a lead arranger analyzes two companies A and B, which are identical in each aspect except for their growth forecast, with A having higher growth prospects. The lead arranger calculates worst case scenario analyzes for both firms, which are consequently identical. Setting the covenant according to this analysis, both firms receive the same threshold. However, since the headroom express- es the distance between forecast and threshold, firm A automatically receives more headroom.
Interestingly, model 2 shows that the size of the target firm does not have a significant effect on financial covenant restrictiveness, while the base specification remains ro- bust. One explanation for this result might be that lenders are more concerned with the reputation of the private equity sponsor than with the size of the target firm, implicitly transferring the sponsor related reputation to the borrowing firm. In line with this ar- gument, removing the reputation variables from the equation does not alter the coeffi- cients and t-values of the size variables. Anecdotal evidence supports this argument. Practitioners state that if KKR were to buy a significantly smaller than usual company they would still enjoy the favorable credit terms. The author could not test for this ef- fect because there are only very few deals where a small company was acquired by a high reputation PE group and vice versa.
Model 3 tests whether the share of fixed to total assets has an impact on the decision of financial covenant restrictiveness. Interestingly, the analysis at best shows a negative effect of fixed to total assets on the D/EBITDA headroom. This result makes sense when taking into account that LBO loans constitute cash flow based lending, i.e. the
shares are pledged, in contrast to collateral based lending, where the lenders have di- rect access to the collateral in case of default.
Model 4 includes business risk in the D/EBITDA headroom equation. There is no em- pirical relation between business risk and D/EBITDA headroom. One explanation for this finding might be that PE groups but also banks select targets which are expected to have a low volatility in cash flows and no further attention is devoted to this aspect in negotiations. A business which generated highly volatile operating cash flows is simply not a candidate for an LBO. This piece of evidence is supported when looking at the process of covenant construction which does not take into account any historical cash flow volatility. Implicitly, cash flow volatility might be taken into account when determining leverage levels. Nevertheless, the finding still seems puzzling as higher business risk should incentivize lenders to demand a shift in control early in order to reduce the potential for risk shifting.