95 In column (1) we find that the odds ratio of Tranche Count is positive significant (Ɀ-stat=3.72), indicating that a one standard deviation change in Tranche Count increases the odds of experiencing two credit ratings by 4%. We test whether the results in column (1) are sensitive to modifications, by rotationally removing complexity indicators in columns (2) to (4). We observe, overall, that the caonmd ps il ge xniitfyi cianndti caat tao r1s%f o lsliogwn i af i csai mn ciel a rl epvaetl tienr na:l lt hceaisrecsoeexfcf iecpi et nftosr r e m a i n p o s i t ive Tranche Count, wanhde(r5e)t.hTehceoceoffeifcfiiecnietnsthoanve the correct sign, but are insignificant in columns (3) Log Tranche Size remains significant and stable across all specifications, with odds ratios ranging from 0.24 to 0.35 throughout. The odds ratios of our third complexity factor, Capital Allocation, are significantly above two in all specifications. In the first column of Table 3.2 we find that the magnitude of Capital Allocation’s coefficient increases to an odds ratio of 2.38 (Ɀ-stat=9.08), indicating that a one standard deviation increase in Capital Allocation increases the odds of a dual-rated tranche by 238%. This finding suggests that the security dl i ke se il gi hnowo dh eor fe da i Cs cLlOo si isnggi vtewnoa rhaitghheerr tchaapni t aolnael l corceadt ii ot nr ai st i ansgs oa ct i ai stseudawn ci teh. aWhei gahl seor find that, on average, larger issuers are more likely to sell CLO tranches with o- 0n. e2 6c)r, es du igt g er as tt ii nn gg tdhiastc ol ovseerda ,l l wi shs ue ne r csoi zme pma ar ey da l tsoo ps lma ya lal pi sasrut ei nr sd e( ot edrdms i nr ai nt igot hoef likelihood that issuers engage in rating shopping behavior. Columns (5) to (8) are constructed to analyze whether there has been a change in the observed pattern since the financial crisis. We look at the complexity characteristics in the period after 2007 using an interaction variable Post for each of the complexity factors. Overall, we find the same relationship still holds – the one exception being Log Tranche Size in column (6) where we observe a negative significant relationship at the 5% level (Ɀ-stat=-2.43), what means that opneraiovderaafgteerltahregecrritsrisa.nches are less likely to disclose two credit ratings in the Chapter 3 - Security Design and Credit Rating Risk
RkJQdWJsaXNoZXIy MTk4NDMw