Alternative Econometric Speci…cations

To summarize at this point, we …nd a highly signi…cant downward-sloping term structure of transparency spreads. Furthermore, the e¤ect of accounting transparency on the term structure of CDS spreads is largest for the most risky

…rms. We now show that the term structure of transparency spreads remains downward-sloping under alternative econometric speci…cations. Furthermore, while highly signi…cant in the short end, it is often insigni…cant at maturities exceeding 5 years. Hence, the …ndings are strongly supportive of hypotheses H2 and H3 from Du¢ e & Lando (2001). The …ndings only weakly support the overall level e¤ect due to discretionary disclosure in hypothesis H1

Regression (4) to (6) extend these speci…cations and control for a constant time e¤ect. We do that by addressing the latter parametrically using monthly dummies. Clustering by month while including monthly dummies allows one to separate the time e¤ect into a constant and non-constant part. A non-constant time e¤ect is present, if a shock in a given month has a di¤erent e¤ect on di¤erent

…rms.

The cross-sectional Fama & MacBeth (1973) regression from Table 1.5 is re-peated in regression (7). This regression also accounts for a cross-correlation in residuals stemming from a time e¤ect, and it assumes that the monthly coe¢ cient estimates are independent of each other. However, when estimating the standard error of their mean the annual accounting transparency measure may imply a serial correlation in the monthly coe¢ cient estimates. We adopt the method in Abarbanell & Bernard (2000) and present the adjusted standard errors in regres-sion (8). This adjustment is designed to correct for a …rm e¤ect arising from persistent …rm characteristics.^{20 21}

The conclusion from Table 1.7 Panel A is that the transparency spread is very robust in the short end and estimated around 20 bps at the 1-year maturity. At longer maturities the transparency spread narrows and is estimated around 14, 8, 7 and 5 bps at the 3, 5, 7 and 10-year maturity, respectively. While highly signi…cant in the short end across all speci…cations, the transparency spread is most often insigni…cant after the 7-year maturity. The same conclusion results from Panel B, where the di¤erent econometric speci…cations are applied on full curves²².

20To be conservative, the adjustment is not applied when the estimated serial correlation is less than zero.

21We do not report standard errors after clustering at the …rm level or introducing …rm dummies for a number of reasons. First, the short time-series implies that we only have 1 year of data for a signi…cant number of …rms (as noted in Table 1.1 the data consists of 368 …rms and 890 …rm-years in 2002 to 2004). This makes an identi…cation of a …rm e¤ect separate from accounting transparency impossible. Second, as shown in Petersen (2007) the bias from a …rm e¤ect is increasing in the number of periods. Third, the inclusion of …rm …xed e¤ects would force an identi…cation of the transparency spread from time-series changes in accounting transparency, which is unreasonable.

22Other unreported speci…cations such as purely sectional regressions and annual cross-sectional regressions based on the time-series average CDS spreads and control variables support these …ndings.

Table1.7:TheTermStructureofTransparencySpreadsUnderVariousEconometricSpeci…cations ThistableestimatesthegapbetweenthehighandlowtransparencyCDScurvesundervariouseconometricspeci…cations.(1) isapooledOLSregressionwithWhiteerrors,while(2)and(3)controlforresidualdependencebyestimatingcluster-robust errorsbycurvesandtime,respectively.Regression(4)to(6)extend(1)to(3)byincludingmonthlydummies.TheFama& MacBethestimatesarereportedin(7),andFama&MacBethstandarderrorsadjustedforadependencebetweenthecross-sections byAbarbanell&Bernard(2000)arereportedin(8).T-statisticsarereportedinparantheses.PanelAdisplaystheresultsfor unrestrictedcurves,whilePanelBdisplaysresultsforfullcurveswithanobservationatamaturityof1,3,5,7and10years.The regressionsareSpreaditT=1m1it+2m3it+3m5it+4m7it+5m10it+6dm1it+7dm3it+8dm5it+9dm7it+10dm10it+ 11Volit+12Levit+13QdispitT+"itT:*,**and***denotesigni…canceat10,5and1percent,respectively. PanelA.Unrestrictedcurves (1)(2)(3)(4)(5)(6)(7)(8) WhiteClusterClusterWhiteClusterClusterF-MAdj.F-M m1-m10supp.supp.supp.supp.supp.supp.supp.supp. dm1-20.08*** -20.08*** -20.08*** -17.12*** -17.12*** -17.12*** -22.66*** -22.66*** (-3.43)(-3.50)(-3.29)(-3.03)(-3.11)(-2.84)(-4.22)(-2.89) dm3-14.44*** -14.44*** -14.44*** -13.93*** -13.93*** -13.93*** -20.04*** -20.04*** (-3.68)(-3.89)(-3.57)(-3.68)(-3.92)(-3.53)(-6.58)(-3.66) dm5-9.05*** -9.05*** -9.05*** -7.66** -7.66*** -7.66** -13.15*** -13.15** (-2.83)(-2.95)(-2.85)(-2.49)(-2.64)(-2.47)(-5.56)(-2.14) dm7-8.49** -8.49*** -8.49*** -7.47** -7.47** -7.47** -12.88*** -12.88*** (-2.49)(-2.59)(-2.94)(-2.29)(-2.39)(-2.65)(-5.98)(-4.37) dm10-6.34* -6.34* -6.34** -3.53-3.53-3.53-10.94*** -10.94*** (-1.90)(-1.91)(-2.21)(-1.10)(-1.15)(-1.29)(-5.26)(-3.81) Volatility686.82*** 686.82*** 686.82*** 767.05*** 767.05*** 767.05*** 805.44*** 805.44*** (34.28)(20.99)(10.49)(28.69)(18.58)(10.52)(16.50)(5.24) Leverage355.97*** 355.97*** 355.97*** 349.27*** 349.27*** 349.27*** 317.20*** 317.20*** (54.58)(29.19)(12.65)(53.41)(29.20)(12.98)(12.98)(3.37) Qdisp-63.04*** -63.04** -63.0412.1112.1112.11-33.37-33.37 (-2.76)(-2.01)(-1.31)(0.50)(0.36)(0.25)(-1.07)(-0.60) Cluster-CurveMonth-CurveMonth-- Dummy---MonthMonthMonth--

Dummy---MonthMonthMonth--Table1.7:TheTermStructureofTransparencySpreadsUnderVariousEconometricSpeci…cations(cont.) ThistableestimatesthegapbetweenthehighandlowtransparencyCDScurvesundervariouseconometricspeci…cations.(1) isapooledOLSregressionwithWhiteerrors,while(2)and(3)controlforresidualdependencebyestimatingcluster-robust errorsbycurvesandtime,respectively.Regression(4)to(6)extend(1)to(3)byincludingmonthlydummies.TheFama& MacBethestimatesarereportedin(7),andFama&MacBethstandarderrorsadjustedforadependencebetweenthecross-sections byAbarbanell&Bernard(2000)arereportedin(8).T-statisticsarereportedinparantheses.PanelAdisplaystheresultsfor unrestrictedcurves,whilePanelBdisplaysresultsforfullcurveswithanobservationatamaturityof1,3,5,7and10years.The regressionsareSpreaditT=1m1it+2m3it+3m5it+4m7it+5m10it+6dm1it+7dm3it+8dm5it+9dm7it+10dm10it+ 11Volit+12Levit+13QdispitT+"itT:*,**and***denotesigni…canceat10,5and1percent,respectively. PanelB.Fullcurves (1)(2)(3)(4)(5)(6)(7)(8) WhiteClusterClusterWhiteClusterClusterF-MAdj.F-M m1-m10supp.supp.supp.supp.supp.supp.supp.supp. dm1-21.35*** -21.35*** -21.35*** -19.33*** -19.33*** -19.33*** -23.56*** -23.56*** (-3.43)(-3.53)(-4.13)(-3.27)(-3.39)(-3.86)(-3.91)(-2.91) dm3-16.54*** -16.54*** -16.54*** -14.48*** -14.48*** -14.48*** -20.52*** -20.52** (-3.01)(-3.12)(-3.33)(-2.79(-2.92)(-2.98)(-3.57)(-2.59) dm5-13.75*** -13.75*** -13.75*** -11.71** -11.71*** -11.71*** -17.61*** -17.61** (-2.75)(-2.86)(-3.24)(-2.50)(-2.62)(-2.84)(-3.11)(-2.47) dm7-10.99** -10.99** -10.99** -9.02** -9.02** -9.02** -14.67** -14.67** (-2.30)(-2.39)(-2.75)(-2.00)(-2.10)(-2.33)(-2.71)(-2.24) dm10-9.07** -9.07** -9.07** -5.01* -5.01* -5.01* -13.08** -13.08** (-2.03)(-2.09)(-2.26)(-1.66)(-1.73)(-1.78)(-2.47)(-2.06) Volatility712.37*** 712.37*** 712.37*** 793.03*** 793.03*** 793.03*** 873.06*** 873.06*** (38.52)(18.72)(8.83)(37.96)(18.68)(9.40)(12.92)(5.94) Leverage368.84*** 368.84*** 368.84*** 359.81*** 359.81*** 359.81*** 315.71*** 315.71*** (47.12)(21.59)(12.65)(47.46)(21.77)(12.80)(11.80)(5.71) Qdisp-93.68*** -93.68* -93.68-29.26-29.26-29.26-122.68** -122.68* (-2.68)(-1.81)(-1.17)(-0.85)(-0.59)(-0.45)(-2.37)(-1.80) Cluster-CurveMonth-CurveMonth-- Dummy---MonthMonthMonth--

Dummy---MonthMonthMonth--Table 1.8 repeats the speci…cations in Dummy---MonthMonthMonth--Table 1.7, but includes the senior un-secured credit rating from Standard & Poor’s as an additional control variable in equation (1.5). As noted in Sengupta (1998) and Yu (2005), credit agencies claim to have incorporated the quality of information disclosure in the credit rat-ings. The results show that credit ratings do not absorb the e¤ect of accounting transparency on the term structure of credit spreads. After accounting for the information content in credit ratings, the transparency spread continues to be highly signi…cant at the 1-year maturity and downward-sloping. However, now the gap between the high and low transparency credit curves is insigni…cant after the 5-year maturity. As expected, the credit rating is highly signi…cant and a one notch increase in rating lowers the CDS spread by approximately 50 bps.

Unreported results based on full curves support these …ndings.

Consistent with empirical …ndings in Du¤ee (1998), structural models such as Longsta¤ & Schwartz (1995) predict an inverse relationship between the risk-free rate and credit spreads. An increase in the risk-free rate increases the risk-neutral drift of the asset value process and reduces the risk-neutral default probability. If an increase in the slope of the risk-free yield curve increases the expected future short rate, then by the same argument as above it implies a decrease in credit spreads. From a di¤erent perspective, as noted in Collin-Dufresne et al. (2001), a decrease in the slope of the risk-free yield curve may imply a weakening economy with decreasing expected recovery rates and higher default rates. Once again, a negative relationship between the slope of the risk-free yield curve and credit spreads is expected. The risk-free term structure variables are constant across all …rms in a given month. Hence, they cannot be included in the empirical speci…cations from Table 1.7 based on Fama & MacBeth (1973) or when including monthly dummies. Table 1.9 presents the results from including the slope of the yield curve in addition to credit ratings in equation (1.5). The slope is de…ned as the di¤erence between the 10 and 1-year constant maturity treasury yields.²³ The slope of the risk-free yield curve is highly signi…cant and estimated with a negative coe¢ cient. However, the transparency spread continues to be highly signi…cant in the short end, downward-sloping and insigni…cant after the 5-year maturity.

23The level of the risk-free yield curve is discussed in section 1.7.2, where individual maturity classes are studied.

Table1.8:TheTermStructureofTransparencySpreadsandCreditRatings ThistableestimatesthegapbetweenthehighandlowtransparencyCDScurvesundervariouseconometricspeci…cations.(1)is apooledOLSregressionwithWhiteerrors,while(2)and(3)controlforresidualdependencebyestimatingcluster-robusterrors bycurvesandtime,respectively.Regression(4)to(6)extend(1)to(3)byincludingmonthlydummies.TheFama&MacBeth estimatesarereportedin(7),andFama&MacBethstandarderrorsadjustedforadependencebetweenthecross-sectionsby Abarbanell&Bernard(2000)arereportedin(8).T-statisticsarereportedinparantheses.Theseniorunsecuredcreditratingsfrom S&Paretransformedtoanumericalscale,where…rmsratedAAAareassignedascoreof10,AAascoreof9andsoforth.The regressionsareSpreaditT=1m1it+2m3it+3m5it+4m7it+5m10it+6dm1it+7dm3it+8dm5it+9dm7it+10dm10it+ 11Volit+12Levit+13QdispitT+14Ratingit+"itT:*,**and***denotesigni…canceat10,5and1percent,respectively. (1)(2)(3)(4)(5)(6)(7)(8) WhiteClusterClusterWhiteClusterClusterF-MAdj.F-M m1-m10supp.supp.supp.supp.supp.supp.supp.supp. dm1-15.93*** -15.93*** -15.93*** -14.28*** -14.28*** -14.28** -18.90*** -18.90*** (-2.76)(-2.81)(-2.89)(-2.66)(-2.62)(-2.57)(-3.52)(-2.68) dm3-11.76*** -11.76*** -11.76*** -11.83*** -11.83*** -11.83*** -17.23*** -17.23*** (-3.07)(-3.25)(-3.24)(-3.19)(-3.41)(-3.33)(-6.18)(-3.52) dm5-6.43** -6.43** -6.43** -5.69* -5.69** -5.69** -10.16*** -10.16** (-2.09)(-2.17)(-2.45)(-1.92)(-2.02)(-2.16)(-5.17)(-2.26) dm7-4.77-4.77-4.77** -4.48-4.48-4.48* -8.87*** -8.87*** (-1.45)(-1.50)(-2.10)(-1.42)(-1.48)(-1.99)(-4.88)(-4.20) dm10-3.02-3.02-3.02-1.51-1.51-1.51-7.45*** -7.45*** (-0.94)(-0.98)(-1.30)(-0.49)(-0.51)(-0.65)(-3.80)(-2.81) Volatility639.80*** 639.80*** 639.80*** 694.29*** 694.29*** 694.29*** 702.68*** 702.68*** (32.33)(19.44)(9.38)(26.06)(16.55)(9.14)(13.65)(4.05) Leverage263.48*** 263.48*** 263.48*** 262.34*** 262.34*** 262.34*** 228.60*** 228.60*** (43.04)(21.85)(10.26)(44.96)(22.70)(11.21)(11.67)(3.47) Qdisp74.71*** 74.71** 74.71* 126.28*** 126.28*** 126.28*** 72.30** 72.30 (3.30)(2.45)(1.83)(5.38)(4.01)(3.04)(2.73)(1.55) Rating-49.70*** -49.70*** -49.70*** -46.62*** -46.62*** -46.62*** -48.73*** -48.73*** (-37.41)(-20.70)(-15.14)(-27.60)(-16.66)(-12.73)(-17.54)(-5.96) Cluster-CurveMonth-CurveMonth-- Dummy---MonthMonthMonth--

Dummy---MonthMonthMonth--Table 1.9: The Term Structure of Transparency Spreads and the Yield Curve

This table estimates the gap between the high and low transparency CDS curves under various econometric speci…cations. (1) is a pooled OLS regression with White errors, while (2) and (3) control for residual dependence by estimating cluster-robust errors by curves and time, respectively. T-statistics are reported in parantheses. The senior unsecured credit ratings from Standard & Poor’s are transformed to a numerical scale, where …rms rated AAA are assigned a score of 10, AA a score of 9 and so forth. The slope of the yield curve is the di¤erence between the 10 and 1-year constant maturity treasury rates. Panel A displays the results for unrestricted curves, while Panel B displays results for full curves with an observation at a maturity of 1, 3, 5, 7 and 10

years. The regressions areSpread_itT = ₁m_1it+ ₂m_3it+ ₃m_5it+ ₄m_7it+ ₅m_10it+

6dm_1it+ ₇dm_3it+ ₈dm_5it+ ₉dm_7it+ ₁₀dm_10it+ ₁₁V ol_it+ ₁₂Lev_it+ ₁₃Qdisp_itT+

14Ratingit+ ₁₅Slopet+"itT:*, ** and *** denote signi…cance at 10, 5 and 1 percent,

respectively.

Panel A. Unrestricted curves Panel B. Full curves

(1) (2) (3) (1) (2) (3)

White Cluster Cluster White Cluster Cluster

m₁-m₁₀ supp. supp. supp. supp. supp. supp.

dm₁ -15.48^{* * *} -15.48^{* * *} -15.48^{* * *} -13.93^{* *} -13.93^{* *} -13.93^{* * *}

(-2.70) (-2.75) (-2.77) (-2.30) (-2.35) (-3.16)

dm₃ -11.60^{* * *} -11.60^{* * *} -11.60^{* * *} -9.02^* -9.02^* -9.02^{* *}

(-3.05) (-3.23) (-3.20) (-1.72) (-1.78) (-2.13)

dm₅ -6.27^{* *} -6.27^{* *} -6.27^{* *} -6.22 -6.22 -6.22^*

(-2.05) (-2.14) (-2.37) (-1.33) (-1.38) (-1.72)

dm₇ -4.68 -4.68 -4.68^{* *} -3.76 -3.76 -3.76

(-1.43) (-1.49) (-2.05) (-0.84) (-0.87) (-1.10)

dm₁₀ -2.89 -2.89 -2.89 -1.50 -1.50 -1.50

(-0.91) (-0.95) (-1.23) (-0.36) (-0.37) (-0.42)

Volatility 649.90^{* * *} 649.90^{* * *} 649.90^{* * *} 682.60^{* * *} 682.60^{* * *} 682.60^{* * *}

(32.52) (19.57) (9.56) (35.31) (17.12) (7.99)

Leverage 263.51^{* * *} 263.51^{* * *} 263.51^{* * *} 270.80^{* * *} 270.80^{* * *} 270.80^{* * *}

(43.31) (22.04) (10.83) (34.71) (15.91) (10.58)

Qdisp 94.90^{* * *} 94.90^{* * *} 94.90^{* *} 75.96^{* *} 75.96 75.96

(4.18) (3.09) (2.29) (2.25) (1.54) (1.13)

Rating -49.38^{* * *} -49.38^{* * *} -49.38^{* * *} -57.97^{* * *} -57.97^{* * *} -57.97^{* * *}

(-37.37) (-20.74) (-13.66) (-35.79) (-16.72) (-13.58)

Slope -0.51^{* * *} -0.51^{* * *} -0.51^{* * *} -0.65^{* * *} -0.65^{* * *} -0.65^{* * *}

(-22.26) (-11.09) (-3.19) (-18.99) (-8.89) (-2.82)

Cluster - Curve Month - Curve Month

Dummy - - -