Do Credit Agencies Add Value? Evidence from the Sovereign ...
Upcoming SlideShare
Loading in...5
×
 

Do Credit Agencies Add Value? Evidence from the Sovereign ...

on

  • 545 views

 

Statistics

Views

Total Views
545
Views on SlideShare
545
Embed Views
0

Actions

Likes
0
Downloads
7
Comments
0

0 Embeds 0

No embeds

Accessibility

Upload Details

Uploaded via as Microsoft PowerPoint

Usage Rights

© All Rights Reserved

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Processing…
Post Comment
Edit your comment

    Do Credit Agencies Add Value? Evidence from the Sovereign ... Do Credit Agencies Add Value? Evidence from the Sovereign ... Presentation Transcript

    • Eduardo Cavallo, IADB Andrew Powell, IADB Roberto Rigobon, MIT
    • Motivation
      • Do credit agencies add informational value to an already well functioning financial market?
        • Rating changes are usually anticipated. Hence, they should have been incorporated in interest rates and other financial variables.
        • In sovereign debt, does the rating adds information beyond the information already in the interest rate?
      • Very difficult to disentangle informational content of credit ratings
    • What do we do?
      • Evaluate informational content using methodology robust to several misspecification errors
      • Evaluate impact of rating changes on stock markets, future spreads, and exchange rates – after controlling for current interest rates and VIX
    • What we find?
      • Ratings provide information in additional to interest rates
      • Rating upgrades
        • Reduce future interest rate spreads
        • Increase stock markets
        • Appreciate exchange rates
      • Results are quite robustness
    • Agenda
      • Methodology
      • Data
      • Results
      • Conclusions
    • Methodology
      • Technically we are asking if the interest rate is a sufficient statistic for the credit rating.
      • We have to allow for misspecification.
      • To test this hypothesis we assume that there is an underlying fundamental for the economy, and interest rates and credit ratings are imperfect measures of it.
      • We evaluate the “sufficient statistic” property of the interest rate trying to explain other financial variables
        • Future spread
        • Stock market
        • Exchange rate
    • Methodology X(t) I(t)
    • Methodology X(t) R(t)
    • Methodology X(t) I(t) R(t)
    • Methodology X(t) I(t) R(t) S(t)
    • Methodology
      • Idea
        • If the true model is then we can estimate by OLS or using ratings as IV.
      • Test
        • Under the null hypothesis the OLS estimate and the IV estimate are identical.
        • Under the alternative hypothesis, the OLS and IV are different. The OLS is biased because of EIV, but IV is consistent.
    • Methodology
      • After we have found that the rating has informational content, we run a horse race between interest rates and ratings.
        • We estimate in a window surrounding credit rating changes. (+/- 10 days)
        • Fixed effect per event
        • Cumulative returns – to deal with endogeneity and anticipation.
    • Methodology
      • Typical event
    • Agenda
      • Methodology
      • Data
      • Results
      • Conclusions
    • Data
      • Source: Bloomberg
      • Daily information
      • 32 emerging market economies
      • January 1 st 1998 and April 25 th 2007
      • Macro variables: stock market, interest rate spread, dollar exchange rate, VIX
      • Ratings: Moody, S&P, Fitch – transformed to a numerical scale.
      • Unbalanced panel with ~80k observations
    • Data
    • Data 21 12 5 15
      • Concurrence of credit rating changes (21 days)
    • Agenda
      • Methodology
      • Data
      • Results
      • Conclusions
    • Results
      • Pooled all credit rating events.
      • Fixed effects for each event.
      • Analyze window of 21 days surrounding credit rating change.
      • Use cumulative returns.
      • We are not concerned with interpretation of coefficient. No attempt to disentangle channel of propagation.
    • Results
      • Table 4: OLS versus IV
    • Results
      • Table 5: summary
        Spreadt+1 Stock Market Exchange Rate Standard & Poor's (downgrades + upgrades) 0.001 0.018 0.848 Standard & Poor's (downgrades) 0.010 0.800 0.436 Standard & Poor's (upgrades) 0.001 0.140 0.001 Fitch (downgrades + upgrades) 0.430 0.600 0.001 Fitch (downgrades) 0.960 0.001 0.001 Fitch (upgrades) 0.190 0.001 0.031 Moodys (downgrades + upgrades) 0.066 0.061 0.082 Moodys (downgrades) 0.355 0.053 0.001 Moodys (upgrades) 0.078 0.009 0.001 Standard & Poor's - 5 day window (all) 0.001 0.078 0.771 Standard & Poor's - 5 day window (downgrades) 0.001 0.770 0.018 Standard & Poor's - 5 day window (upgrades) 0.100 0.017 0.001 Standard & Poor's - 20 day window (all) 0.001 0.660 0.850 Standard & Poor's - 20 day window (downgrades) 0.001 0.001 0.670 Standard & Poor's - 20 day window (upgrades) 0.001 0.068 0.001 Standard & Poor's - Without contemporanous change in rating 0.001 0.100 0.250 Rejection rate 2 75% 63% 63%
    • Lessons
      • Informational content
        • Around credit rating changes, ratings provide information beyond interest rates
          • EIV interpretation allows for a robust methodology
          • Robust to specification changes
          • Even though they are anticipated
    • Results
      • Macro variables and S&P
        S&P upgrades & downgrades   Spread Rating VIX Spreadt+1 0.884*** -0.006*** 0.006   [0.011] [0.0014] [0.015] Stock Market -0.205*** 0.004*** -0.104***   [0.011] [0.014] [0.001] Exchange Rate 0.098*** -0.0005 0.045***   [0.008] [0.0009] [0.010] Δ Spread -0.117*** -0.006*** 0.029*   [0.011] [0.001] [0.015]
    • Results
      • Macro variables, Fitch and Moody
        Fitch upgrades and downgrades   Moodys upgrades & downgrades   Spread Rating VIX   Spread Rating VIX Spreadt+1 0.863*** -0.002 0.036***   0.855*** -0.004** 0.040***   [0.010] [0.001] [0.011]   [0.013] [0.002] [0.015] Stock Market -0.404*** 0.002 -0.132***   -0.297*** 0.005** -0.140***   [0.016] [0.002] [0.017]   [0.014] [0.002] [0.016] Exchange Rate 0.225*** -0.009*** 0.033**   0.190*** -0.003** 0.046***   [0.013] [0.002] [0.014]   [0.010] [0.0014] [0.012] Δ Spread -0.139*** -0.001 0.064***   -0.147*** -0.004** 0.070***   [0.010] [0.001] [0.012]   [0.013] [0.002] [0.015]
    • Results
      • S&P upgrades and downgrades
        S&P downgrades   S&P upgrades   Spread Rating VIX   Spread Rating VIX Spreadt+1 0.894*** -0.006*** 0.013   0.876*** -0.007*** -0.003   [0.014] [0.001] [0.017]   [0.017] [0.001] [0.022] Stock Market -0.484*** -0.002 -0.067***   -0.018** 0.002** -0.085***   [0.020] [0.002] [0.025]   [0.008] [0.001] [0.012] Exchange Rate 0.196*** -0.003 0.089***   0.007** 0.002*** -0.006   [0.018] [0.002] [0.022]   [0.003] [0.0003] [0.005] Δ Spread -0.109*** -0.006*** 0.030*   -0.124*** -0.006*** 0.027   [0.014] [0.001] [0.018]   [0.017] [0.0019] [0.024]
    • Results
      • Typical event
    • Lessons
      • Informational content
        • Around credit rating changes, ratings provide information beyond interest rates
          • EIV interpretation allows for a robust methodology
          • Robust to specification changes
          • Even though they are anticipated
      • Rating changes
        • Upgrades
          • Decrease future spreads (0.7% per notch)
          • Increase stock market (0.2% per notch)
          • Appreciate real exchange rate (0.2% per notch)
        • Downgrades
          • Decrease future spreads (0.6% per notch)
          • No impact on stock markets
          • No impact on exchange rates
    • Results
      • Does changes in asset class have larger impact?
        • We find that changing the asset class has no additional effect for the rating variable.
      • What about outlook changes?
        • Replicate the results for outlook.
        • Estimate degree of anticipation using the outlook change prior to the rating change.
    • Results
      • Using outlook in the specification
        S&P upgrades & downgrades   Spread Outlook VIX Spreadt+1 0.856*** -0.0005 0.022**   [0.009] [0.002] [0.009] Stock Market -0.363*** 0.007*** -0.009   [0.011] [0.002] [0.010] Exchange Rate 0.083*** -0.005*** 0.015***   [0.006] [0.0008] [0.005] Δ Spread -0.148*** -0.0009 0.0536***   [0.009] [0.002] [0.0097] S&P downgrades   S&P upgrades Spread Outlook VIX   Spread Outlook VIX 0.876*** 0.002 0.02   0.808*** -0.003* 0.024* [0.013] [0.002] [0.014]   [0.016] [0.001] [0.012] -0.400*** -0.001 -0.017   -0.283*** [0.002] 0.0159*** [0.013] [0.002] [0.013]   [0.020] 0.015*** [0.0023] 0.090*** -0.007*** 0.020**   0.054*** -0.002*** 0.009** [0.009] [0.002] [0.009]   [0.004] [0.001] [0.003] -0.128*** 0.002 0.059***   -0.195*** -0.004** 0.045*** [0.013] [0.002] [0.014]   [0.016] [0.0018] [0.013]
    • Results
      • Outlook: days between outlook and change.
    • Results
      • Degree of anticipation
    • Conclusions
      • Ratings provide information in additional to interest rates
        • Different agencies provide different information
      • Rating upgrades
        • Reduce future interest rate spreads
        • Increase stock markets
        • Appreciate exchange rates
        • All even after controlling for, fixed effects, interest rate and VIX.
      • Robustness
        • Anticipation affects the quantitative results but not the qualitative message
        • Outlooks provide same conclusions