do ratings matter for re structured debt?...do ratings matter for re structured debt? rating para...
TRANSCRIPT
DO RATINGS MATTER FOR RE STRUCTURED DEBT?
RATING PARA CRIS: VALE A PENA?
Eduardo Fernandes [email protected](11) 9-9448-8771Co-autor: Richard Saito
Agradecimentos:
Purpose
โข Investigate the determinants of pricing in CRIs
โข Verify the impact of rating in the return of CRIs
2
Approach
โข Analyze the market and its key characteristics
โข Compare returns with debentures
โข Empirical test and comparison of models whichtake segments into account and other controls
Key Findings/Value
โข Although CRIs follow relative standardization,we find that they can represent different levelsof risk and underlying assets
โข There are different pricing levels for theirspecific risk characteristics. Few are rated(only 23% of all CRI issuances)
โข Versus debentures: average premium of 1.0 p.p.
โข The average spread in CRIs is 3.2 p.p.โ Higher for residential and allotment (+1.4p.p.)โ High level of guarantees lowers spreads
โข Rating effect:โ Residential CRIs show benefits of any ratingโ Commercial CRIs need high grade or be
assessed by international agencyโ Consequence: accumulation of guarantees
BriefSummary
3
The CRI Marketโข Brazil started to develop its secondary mortgage market in 1997 and regulation improved in 2004โข Strong development since 2005 when issued volumes surpassed R$1.0 billionโข Volumes issued reached R$ 14 billion in 2013 and 2014โข Deceleration in 2015 (R$ 10 billion) and in 2016 (R$ 7 billion until Sep)โข Eight securitizing entities concentrate 90% of all issuancesโข The main clearance where CRIs are traded is CETIP with 90% of the market
Relevancy of StudyThe analysis and implications of this study are important for real estate financing in general by:
(i) increasing access of companies to capital(ii) reducing the cost of capital(iii) increasing the allocation of more resources to the Brazilian real estate financing system
Context
4
All literatureโข Campbell (2013): three main categories of mortgage systemsโข Challenges of the securitized mortgage and covered bonds: liquidity and a low level of information asymmetry
(Campbell, 2013; Leece, 2008; Gyamfi-Yeboah et.al., 2010; Desa, 2005)โข Effects of securitization on the lending process: lowers renegotiation rates but shows less efficient
renegotiation outcomes in case of default (Agarwal et al., 2011) and bad practices such as poor screening standards (Keys et al., 2010)
โข Advantages of securitization mortgage systems provide increased supply of credit (Mian and Sufi, 2009) and lower the cost of capital (Desa, 2005)
โข The efficient integration of mortgage markets and capital markets has the consequences of a better environment with reduced interest rates for borrowers and reduction in mortgage credit rationing
โข Caselli and Gatti (2005), the caveats for the financial disintermediation include three matters: (1) transactional difficulty, (2) informational difficulty, and (3) perceived risk.
โข The experience of the USA is relatively successful as it has regularly revised its legal frameworks since the 1970s to great advances in the 1990s (Poindexter et al., 2002) and it presents an electronic market with liquidity and cheaper funding (Vickery and Wright, 2013). Gyamfi-Yeboah et.al. (2010) studied the South African experience and concludes that it supplied additional mortgage credit, however at no lower cost in that country.
Gap in the literature for Brazilโข This is the first study to measure and evaluate the factors behind the high returns of CRIs in the
Brazilian market.โข Other authors (Vink, D., and A. Thibeault, 2008; Pinto et al., 2016; Buscaino et al., 2012; Hu et al., 2006; Maris
and Segal, 2002) have developed important strategies to investigate structured finance instruments and have been used in this study whenever possible.
LiteratureReview
5
Main takeawaysโข 46% are commercial CRIs and 49% are residential CRIsโข the majority of CRIs (59%) are indexed at inflation (mostly IPCA or IGP-M) plus spread in percentage points โข Around 26% of all CRI volumes have been issued by banks that have used it as a tool for recycling (TR-linked)โข Average volume: R$ 29 million (excluded TR-linked CRIs)โข currently only 23% of the issuances are rated โข Risks of CRIs: default, liquidity, interest rate, prepayment โข Common credit enhancements: tranching, excess collateral, guaranteesโข 3 main types of guarantees: corporate, personal and propertyโข 82% have one or more of the 3 types of guaranteesโข Personal guarantees is present in 41% of issuances
Investors' point of view literatureโข 33%ofinvestorsareindividualsโข 56%ofinvestorsareinstitutional investorswhothemselves designandstructuretheCRIโข Min.TicketisR$300,000andusually R$1,000,000 atissuanceโข Averageremunerationwas11.4%p.a.Innominal terms,17.8%p.a.vs. 13.4%p.a.avg.Selicrate(2002-2014)โข Averagespreadwas3.2%p.a.(or321bps), higherforresidential andallotmentโข CVMandANBIMAhaveimprovedmarkettransparencyโข NoCRIhasdefaultedasfarastheresearchcould verify
MarketOverview
6
DescriptiveAnalysisGRAPH 1 - Nominal Remuneration
Average from 2004 to 2015: inflation plus 11.4%
2,0%
3,0%
4,0%
5,0%
6,0%
7,0%
8,0%
9,0%
10,0%
11,0%
12,0%
13,0%
14,0%
May-04
Aug-06
Jan-08
Jun-08
Jan-09
Apr-09
Aug-09
Dec-09
Mar-10
May-10
Aug-10
Oct-10
Dec-10
Dec-10
Feb-11
Mar-11
May-11
Jun-11
Aug-11
Sep-11
Oct-11
Nov-11
Dec-11
Feb-12
Apr-12
May-12
Jun-12
Aug-12
Oct-12
Nov-12
Dec-12
Jan-13
Feb-13
Mar-13
May-13
Jun-13
Jul-1
3Au
g-13
Sep-13
Oct-13
Nov-13
Nov-13
Dec-13
Feb-14
Mar-14
May-14
Jun-14
Jul-1
4Sep-14
Oct-14
Dec-14
Feb-15
GRAPH 2 โ Avg. Spread at issue Average spread from 2004 to 2015: 3.2%
-3,0%
-2,0%
-1,0%
0,0%
1,0%
2,0%
3,0%
4,0%
5,0%
6,0%
7,0%
8,0%
9,0%
10,0%
11,0%
May-04
Sep-07
May-08
Dec-08
Mar-09
Aug-09
Dec-09
Feb-10
May-10
Aug-10
Oct-10
Dec-10
Dec-10
Feb-11
Mar-11
May-11
Jun-11
Aug-11
Sep-11
Oct-11
Nov-11
Dec-11
Feb-12
Apr-12
May-12
Jun-12
Aug-12
Oct-12
Nov-12
Dec-12
Jan-13
Feb-13
Mar-13
May-13
Jun-13
Jul-1
3Au
g-13
Sep-13
Oct-13
Nov-13
Nov-13
Dec-13
Feb-14
Mar-14
May-14
Jun-14
Jul-1
4Sep-14
Nov-14
Dec-14
Mar-15
Remuneration (average) Full Sample Senior Junior/SubordinatedInflation (IGP, IPCA, IPC) plus | n=501 11.4% 8.8% 21.8%DI plus | n=106 4.5% 4.6% 3.6%INCC (Construction index) plus | n=3 9.0% 9.0% N/ATR (Bank mortgage contracts) plus | n=221 10.5% 10.2% 14.0%Fixed | n=22 18.0% 13.8% 27.0%Notes: (i) 'n' refers to the size of the sample; (ii) the percentages refer to remuneration of each category.Source: CETIP
Segments Number of Issuances
Average Spread
Commercial 181 2.3 Residential 183 4.0 Allotment 19 4.4 Total 383 3.2
โข Databases: ANBIMA, CETIP and CVMโข 853 issuances have been analyzed โข Two main subsamples:
Subsample 1: data from 2013 to 2015 with 41 observations
Subsample 2: data from 2004 to 2015, with 373 observations (senior inflation-indexed CRIs)
โข โSpreadCRIโ in the main regressions has been tested for normality through Kolmogorov โSmirnov one-sample test and due to failure to reject the null hypothesis, the sample approximates a Gaussian distribution (at critical value 1%). For all models, Whiteโs heteroscedasticity-consistent estimator was used.
7
DataDescription
HypothesesH1: Existence of difference in spreads between CRIs and debentures (given the same rating and duration).
Based on existing research, no significant difference would be verified based solely on rating grades.
H2: Determinants of spreads are different based on the segments of the CRIs (cross-section study).
Based on existing literature, the determinants should be highlighted for the two segments studied โ the residential and the commercial, and their guarantees.
8
Methodology
9
(1)
(2)
๐๐๐๐๐๐๐ถ๐ ๐ผ = ๐ผ + l n(๐๐๐๐๐๐๐ท๐๐) + ๐ท๐ข๐๐๐ฆ(๐ ๐๐ก๐๐๐๐ ) + ๐
๐๐๐๐๐๐๐ถ๐ ๐ผ= ๐ผ + ๐ฝ1. ๐๐๐ง๐๐๐+ ๐ฝ2.๐ ๐๐ ๐๐๐๐๐ก๐๐๐ + ๐ฝ3. ๐ถ๐๐๐๐๐๐๐๐๐ + ๐ฝ4. ๐น๐๐๐๐๐๐๐๐+ ๐ฝ5.๐ด๐๐๐๐ก๐๐๐๐ก + ๐ฝ6. ๐ฟ๐ก๐๐๐๐ + ๐ฝ7. ๐บ๐ก๐๐ฟ๐๐ฃ๐๐ + ๐ฝ8. ๐๐๐๐๐บ๐ก๐๐+ ๐ฝ9.๐๐๐๐ ๐บ๐ก๐๐ + ๐ฝ10. ๐ถ๐๐๐๐บ๐ก๐๐+ ๐ฝ11. ๐ถ๐๐๐ก๐๐๐๐ ๐ฝ๐ ๐๐๐ต+ ๐ฝ12.๐ถ๐๐๐๐๐๐๐๐ก๐๐ + ๐ฝ13.๐๐๐ก๐๐๐๐๐๐+ ๐ฝ14.๐ ๐๐ก๐๐ + ๐ฝ15. ๐ ๐๐ก๐ด๐ด๐ด+ ๐ฝ16.๐ ๐๐ก๐ด๐ด + ๐ฝ17. ๐ ๐๐ก๐ด + ๐ฝ18.๐ ๐๐ก๐๐๐๐ผ๐๐ก + ๐
Results(DescriptiveStatisticsI)
10
Summary statistics for the variables that present a comparable debenture information.
Variable Obs Mean Std.Dev. Min MaxT-student (null hyp.)
SpreadCRI 41 2.61 2.18 -0.69 8.78 7.67***SpreadDeb 41 1.64 0.43 0.8 2.28 24.42***โ Spread 41 0.97 2.05 -1.81 7.82 3.03***โ Spread | Rating AAA 5 -0.81 0.39 -1.5 -0.56 -4.64***โ Spread | Rating AA 7 1.38 2.99 -1.03 7.82 1.22โ Spread | Rating A 19 1.05 1.8 -1.69 4.9 2.54***โ Spread | Rating A โ wide* 29 1.18 1.85 -1.81 4.9 3.43***โ Spread | International R.Ag. 17 -0.56 0.81 -1.81 1.04 -2.85***โ Spread | National R.Ag. 24 2.05 1.97 -1.65 7.82 5.1***Note: * several rated issuances did not disclose grades and it was added to the A category.
Results(RegressionI)
11
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)a.1 a.2 a.3 a.4 a.5 a.6 a.7 a.8 a.9 a.10
Pooled OLS Rob
Pooled OLS Rob
Pooled OLS Rob
Pooled OLS Rob
Pooled OLS Rob
Pooled OLS Rob
Pooled OLS Rob
Pooled OLS Rob
Pooled OLS Rob
Pooled OLS Rob
Dependent Variable SpreadCRI SpreadCRI SpreadCRI SpreadCRI SpreadCRI SpreadCRI SpreadCRI SpreadCRI SpreadCRI SpreadCRI
VARIABLES Full sample RatingInt RatAAA RatAA RatA RatAamplo RatAAAint RatAAint RatAint RatAam.int
SpreadDebln 2.752** 1.002 0.322 2.860** 3.031* -1.477 0.322 2.800** 3.177*** 3.477***(2.57) (1.11) (0.19) (2.62) (2.02) (-0.47) (0.19) (2.50) (2.84) (3.29)
RatingInt -2.106***(-4.95)
RatAAA -2.598**(-2.08)
RatAA 0.179(0.25)
RatA -0.272(-0.37)
RatAamplo 2.747(1.61)
RatAAAint -2.598**(-2.08)
RatAAint 0.180(0.34)
RatAint -1.568***(-4.14)
RatAamploint -1.963***(-4.69)
Constant -0.115 2.198*** 1.146 -0.201 -0.078 -0.210 1.146 -0.170 0.449 0.490(-0.15) (3.24) (1.19) (-0.25) (-0.11) (-0.25) (1.19) (-0.21) (0.87) (0.98)
Observations 41 41 41 41 41 41 41 41 41 41Adjusted R-squared 0.3700 0.5429 0.4111 0.3516 0.3540 0.4029 0.4111 0.3514 0.4252 0.4897Source: Author
Results(RegressionII)
12
(1) (2) (3)b.1 b.2 b.3
Pooled OLS 4_D Rob Pooled OLS 4_D Rob Pooled OLS 4_D Rob
Sample Residential and Commercial
Residential Commercial
Dependent Variable SpreadCRI SpreadCRI SpreadCRI
VARIABLES Residential and Commercial
Residential Commercial
Sizeln -0.352*** -0.519*** -0.258*(-3.34) (-2.64) (-1.82)
Residencial 1.392***(5.63)
Loteamento 1.378***(3.78)
GteeLevel -0.429* -0.665* 0.070(-1.92) (-1.76) (0.21)
PersGtee 0.797** 0.821 0.442(2.34) (1.56) (0.85)
CorpGtee 0.521* 0.909* 0.261(1.70) (1.82) (0.54)
TotalTermln -1.102*** -0.822** -1.185***(-4.57) (-2.32) (-2.79)
Rated 0.234 -1.041** 1.305**(0.65) (-2.34) (2.08)
RatAAA -0.481 -1.229*(-1.00) (-1.74)
RatAA -0.545 0.903 -1.312*(-1.13) (0.98) (-1.85)
RatingInt -0.824** -0.929 -1.222**(-2.55) (-0.74) (-2.35)
Constant 10.775*** 13.754*** 8.282***(5.16) (4.13) (2.76)
Observations 373 197 176Adjusted R-squared 0.4181 0.2204 0.4009Source: Author
MainResultsH1:
โข Versus debentures: average premium of 1.0 p.p.
โข Lower grades and be assessed by a national rating agency provide less benefits than the highestrating grade or being assessed by an international rating agency.
H2:
โข The average spread in CRIs is 3.2 p.p.โข Higher for residential and allotment (+1.4 p.p.)โข High level of guarantees lowers spreads
โข Larger and longer-maturity CRIs have lower spread
โข Rating effect:โข Residential CRIs show benefits of any ratingโข Commercial CRIs need high grade or be assessed by international agency
13
Conclusions
โข Although CRIs follow relative standardization, we find that they can represent different levelsof risk and underlying assets
โข Versus debentures: average premium of 1.0 p.p.
โข The spread is relevant. The average spread in CRIs is 3.2 p.p.โข Higher for residential and allotment (+1.4 p.p.)โข High level of guarantees lowers spreads (-0.4 p.p.)
โข Rating effects:โข Residential CRIs show benefits of any ratingโข Credit rating is a key determinant but not equally seenโข Commercial CRIs need high grade or be assessed by international agency
In order to minimize risks or the perceived risks and compensate for the lowlevel of rated issuances, credit enhancements such as accumulation ofguarantees, tranching and excess spread seem to be the rule.
14
Thank you!Obrigado!
15