Transcript No Slide Title

Brent Ballard
ECON 700 Project
The Impact of Monetary Policy on
High Grade Corporate Yield Spreads
Fall 2009
Contents
I.
Introduction
II.
Method
III.
Data
IV.
Model
V.
Results
VI.
Conclusions
2
Introduction
Introduction


Research Goals
•
Explore the effect that changes in monetary policy have on the yield spread between high-grade corporate bonds and
treasury securities
•
Quantify that response by introducing a policy dummy into a model for the yield spread to measure the effect that it has
on the response variable
U.S. Monetary Policy
•
The U.S. Federal Reserve aims to promote maximum employment, stable prices, and moderate long-term interest rates
by influencing monetary and credit conditions in the economy
– Policy Objectives
– Transmission Mechanisms
– It is important that we understand how these policy transmission mechanisms operate, both to ensure that
monetary policy achieves its intended result, and so that we might predict any unintended financial consequences
of these policy actions

Yield Spread
•
Risk Premium
•
Idiosyncratic risk in the economy is manifested in the yield spread
4
Method
Method

The Policy Dummy
•
Public minutes of FOMC meetings between January, 1964 and December, 2003
•
Dummy = 0
– If a policy directive is issued instructing the committee to maintain the prevailing level of pressure on credit
conditions, dummy = 0 is assigned for that month
•
Dummy = 1
– If a directive is issued instructing the FOMC to either increase or decrease the level of pressure on credit conditions
by any magnitude, and that directive is maintained following four consecutive meetings or more, dummy = 1 is
assigned to each month during which that instruction persists
– Similarly, if a directive is issued ordering the committee to either increase or decrease the prevailing degree of
pressure on credit conditions significantly, dummy = 1 is assigned for that month

A Note Regarding FOMC Policy Directives
•
Whether the Fed pursues its policy objectives by targeting a specific federal funds rate, or by targeting a certain level for
the money supply, there is always some mention in the directive regarding the pressure to be exerted on credit conditions
in the economy.
•
Thus, the presence of a policy dummy is dependant on for how long and to what degree a change in the
pressure applied to these credit conditions persits
•
The dummy is consistent in the way that it distinguishes between different policy directions
6
Data
Data(1)
Dependent Variable

•
Difference between the Moody’s Aaa-rated corporate bond index yield and the 10 – Year yield
Independent Variables


•
U.S. Gross Domestic Product (GDP)
•
Standard & Poor’s 500 Equity Index (S&P)
•
Consumer Price Index (CPI)
•
Producer Price Index (PPI)
•
Spot Price of Oil (OIL)
Non-Stationary Variables
•
A graph of our outcome measure reveals that the yield spread exhibits a consistent upward trend over the course of the
sample period
– The presence of this trend demands that tests are performed to determine whether the series is non-stationary
(1) Data collected from the Federal Reserve Economic Database
8
Yield Spread
2.25%
2.25%
1.75%
1.75%
1.25%
1.25%
0.75%
0.75%
0.25%
0.25%
-0.25%
-0.25%
1976
1979
1982
1985
Yield Spread
•
The following tests for unit roots are applied in
order to determine whether the series is nonstationary
–
Augmented Dickey Fuller Test
–
Phillips-Perron Test
–
Correlogram
Trend-Stationarity
1.00
0.50
-0.50
•
Autocorrelations of yield spread
Non-Stationarity
1988
1.00
1973
0.50
1970
Autocorrelations of spread

1967
0.00
1964
1991
1994
1997
2000
2003
0
-0.50
0
0
5
5
Lag
10
10
15
15
Lag
Bartlett's formula for MA(q) 95% confidence bands
Bartlett’s formula for MA(q) 95% confidence bands
9
Yield Spread Series - Trend Removed
2.00%
2.00%
1.50%
1.50%
1.00%
1.00%
0.50%
0.50%
0.00%
0.00%
-0.50%
-0.50%
1964
1967
1970
1973
1976
1979
1982
1985
1988
1991
1994
1997
2000
2003
Correlogram confirms trend-stationarity
0.50
•
0.00
The modified series is stationary
Autocorrelations of yield spread
•
-0.50
Trend-Stationarity
Autocorrelations of dtspread

1.00
De-Trended Spread
1.00
0.50
0
-0.50
0
0
5
5
Lag
10
15
10
15
Lag 95% confidence bands
Bartlett’s formula for MA(q)
Bartlett's formula for MA(q) 95% confidence bands
10
The Model
Model

Model Specification
An ADL model is specified to predict the yield spread
•
–

Four consecutive single period lags of each variable are included in the model
The Model:
•
Yt = α0 + β0L4(Yt)+ β1S&Pt + β2L4(S&Pt) + β3GDPt + β4L4(GDPt) + β5CPIt + β6L4(CPIt) +
β7PPIt + β8L4(PPIt) + β9OILt + β10L4(OILt) + εt
•
(L4) indicates four consecutive single period lags of the preceding variable
2.75%
2.75%
2.25%
2.25%
1.75%
1.75%
1.25%
1.25%
0.75%
0.75%
0.25%
0.25%
-0.25%
-0.25%
1964
1967
1970
1973
1976
1979
1982
Observed Spread
1985
1988
1991
1994
1997
2000
2003
Predicted Spread
12
Independent Variables – Non-Stationarity
210
1200
175
1000
140
8000
105
6000
70
4000
35
2000
0
0
May-64
Jan-67
Sep-69
May-72
Jan-75
Sep-77
May-80
CPI

Non-Stationarity
•
Non-stationary independent variables
•
Stationary first differences
•
Possibility of cointegration
•
Stationary residuals
•
Model run with first differenced dependant and
independent variables predicts a similar result
•
We are confident that the results are robust
Jan-83
PPI
Sep-85
OIL
May-88
Jan-91
GDP
Sep-93
May-96
Jan-99
Sep-01
S&P
Augmented Dickey-Fuller Test For Unit Root
Interpolated Dickey-Fuller
Variable
Spread
Spread-Trend
GDP
GDP-Trend
S&P
S&P-Trend
CPI
CPI-Trend
PPI
PPI-Trend
OIL
OIL-Trend
Residuals
z(t)
z(t)
z(t)
z(t)
z(t)
z(t)
z(t)
z(t)
z(t)
z(t)
z(t)
z(t)
z(t)
Test
Statistic
1%
Critical
5%
Critical
10%
Critical
-3.021**
-4.178***
4.442
-0.284
0.491
-1.373
1.364
-3.182*
-0.708
-1.223
-1.588
-2.119
-9.425***
-3.442
-3.981
-3.442
-3.981
-3.442
-3.981
-3.442
-3.981
-3.442
-3.981
-3.442
-3.981
-3.442
-2.871
-3.421
-2.871
-3.421
-2.871
-3.421
-2.871
-3.421
-2.871
-3.421
-2.871
-3.421
-2.871
-2.57
-3.13
-2.57
-3.13
-2.57
-3.13
-2.57
-3.13
-2.57
-3.13
-2.57
-3.13
-2.57
13
Results
ADL Estimation Results
No Dummy Immediate
Constant
Time
Spread Lags
Lag1M
Lag2M
Lag3M
Lag4M
S&P Close
Lag1M
Lag2M
Lag3M
Lag4M
Spot
GDP
Lag1M
Lag2M
Lag3M
Lag4M
Spot
Oil Price
Independent Variable Coefficients
Spot
Lag1M
Lag2M
Lag3M
Lag4M
Dummy
R-Sq
F-Statistic
0.1290***
(2.93)
0.0018***
(3.45)
0.9680***
(16.25)
-0.2664***
(-3.01)
0.2342***
(3.20)
-0.0884*
(-1.72)
-0.0002
(-0.78)
-0.0009***
(-3.46)
0.0003
(1.27)
0.0004
(1.58)
0.0004
(1.52)
-0.0036***
(-3.71)
0.0057***
(2.68)
-0.0023
(-1.01)
0.0000
(0.01)
0.0001
(0.07)
-0.0072*
(-1.80)
0.0173***
(2.57)
-0.0115
(-1.63)
0.0092
(1.17)
-0.0040
(-0.74)
None
0.9354
214.73
0.1393***
(3.31)
0.0017***
(3.27)
0.9509***
(15.89)
-0.2528***
(-2.89)
0.2334***
(3.26)
-0.0777
(-1.49)
-0.0002
(-1.08)
-0.0009***
(-3.55)
0.0003
(1.13)
0.0005
(1.65)
0.0005**
(2.06)
-0.0036***
(-3.71)
0.0057***
(2.72)
-0.0023
(-1.04)
0.0001
(0.07)
0.0000
(-0.05)
-0.0072*
(-1.81)
0.0167**
(2.45)
-0.0107
(-1.52)
0.0082
(1.05)
-0.0037
(-0.69)
-0.0634***
(-3.43)
0.9370
212.86
Dependant Variables
Yield Spread
Policy Dummy
1M Lag
3M Lag
6M Lag
0.1426***
(3.77)
0.0018***
(3.66)
0.9507***
(15.77)
-0.0259***
(-2.94)
0.2390***
(3.12)
-0.0825
(-1.34)
-0.0002
(-0.86)
-0.0009***
(-3.72)
0.0003
(1.19)
0.0004
(1.58)
0.0005**
(2.06)
-0.0036***
(-3.86)
0.0056***
(2.75)
-0.0021
(-0.99)
0.0000
(-0.03)
0.0000
(0.12)
-0.0075*
(-1.89)
0.0178***
(2.64)
-0.0113
(-1.61)
0.0089
(1.14)
-0.0043
(-0.79)
-0.0443**
(-2.05)
0.9361
209.94
0.1625***
(3.77)
0.0020***
(3.66)
0.9460***
(15.77)
-0.2580***
(-2.94)
0.2223***
(3.12)
-0.0708
(-1.34)
-0.0002
(-0.86)
-0.0009***
(-3.72)
0.0003
(1.19)
0.0004
(1.58)
0.0005**
(2.06)
-0.0036***
(-3.86)
0.0057***
(2.75)
-0.0021
(-0.99)
0.0000
(-0.03)
0.0000
(0.12)
-0.0071*
(-1.84)
0.0167**
(2.56)
-0.0103
(-1.47)
0.0101
(1.30)
-0.0055
(-1.01)
-0.0639***
(-3.27)
0.9370
212.93
*** Denotes significance at the 1% level
** Denotes significance at the 5% level
* Denotes significance at the 10% level
0.1531***
(3.47)
0.0020***
(3.72)
0.9564***
(16.66)
-0.2657***
(-3.00)
0.2317***
(3.17)
-0.0850*
(-1.69)
-0.0002
(-0.93)
-0.0009***
(-3.46)
0.0003
(1.27)
0.0005*
(1.69)
0.0004*
(1.69)
-0.0037***
(-3.89)
0.0058***
(2.76)
-0.0023
(-1.01)
-0.0001
(-0.04)
0.0001
(0.18)
-0.0071*
(-1.79)
0.0172***
(2.58)
-0.0112
(-1.59)
0.0093
(1.18)
-0.0040
(-0.73)
-0.0319
(-1.34)
0.9353
206.27
1Y Lag
0.1239**
(2.45)
0.0019
(0.00)
0.9683***
(16.21)
-0.2660***
(-3.00)
0.2350***
(3.21)
-0.0879*
(-1.68)
-0.0002
(-0.77)
-0.0009***
(-3.46)
0.0003
(1.28)
0.0005
(1.58)
0.0004
(1.50)
-0.0036***
(-3.66)
0.0057***
(2.66)
-0.0023
(-1.01)
0.0000
(0.03)
0.0000
(0.04)
-0.0075*
(-1.78)
0.0174***
(2.58)
-0.0114
(-1.61)
0.0091
(1.16)
-0.0039
(-0.72)
0.0069
(0.34)
0.9336
197.79
10-Year
Aaa Index
Policy Dummy
Immediate Immediate
0.2356***
(2.65)
0.0007
(0.64)
1.2752***
(26.95)
-0.5747***
(-7.64)
0.3181***
(4.26)
-0.0788*
(-1.68)
-0.0001
(-0.23)
0.0009
(1.37)
0.0004
(0.63)
-0.0011
(-1.58)
-0.0001
(-0.19)
0.0086***
(4.73)
-0.0146***
(-4.02)
0.0048
(1.23)
0.0034
(0.90)
-0.0023
(-1.22)
0.0184*
(1.69)
-0.0395**
(-2.37)
0.0337*
(1.93)
-0.0138
(-0.79)
0.0099
(0.85)
0.1487***
(3.73)
0.9883
1292.56
0.2067***
(2.61)
0.0007
(0.80)
1.3293***
(28.13)
-0.6336***
(-8.17)
0.2864***
(3.70)
-0.0249
(-0.53)
-0.0003
(-0.78)
0.0001
(0.18)
0.0010
(1.94)
-0.0009*
(-1.66)
0.0002
(0.45)
0.0055***
(3.93)
-0.0088***
(-3.12)
0.0017
(0.56)
0.0043
(1.50)
-0.0029*
(-1.99)
0.0108
(1.27)
-0.0250*
(-1.93)
0.0206
(1.53)
-0.0015
(-0.11)
0.0027
(0.30)
0.0957***
(3.13)
0.9923
1971.36
Dummy Lags
Instantaneous: Yt = α0 + β0L4(Yt)+ … + βnDUMMYt
1M Lag:
Yt = α0 + β0L4(Yt)+ … + βnDUMMYt-1
3M Lag:
Yt = α0 + β0L4(Yt)+ … + βnDUMMYt-3
6M Lag:
Yt = α0 + β0L4(Yt)+ … + βnDUMMYt-6
1Y Lag:
Yt = α0 + β0L4(Yt)+ … + βnDUMMYt-12
15
Results
Effect of the policy dummy

•
The policy dummy is statistically significant (t=-3.43) and has improved the overall fit of our model. The instantaneous
effect of the policy dummy is a 6 basis point decline in the yield spread, while one-month and three-month lags of the
dummy decrease the spread by an additional 4 and 6 basis points respectively
– Because the six-month and one-year lags of the policy dummy are insignificant, we can be confident that the
measured effect of the dummy on the yield spread is not merely a coincidental result
•
Since the spread of the Aaa-rated corporate bond index averaged roughly 86 basis points between 1964 and 2003, the
effect of the policy dummy on the yield spread is equivalent to an immediate 7% decline relative to the average spread,
and a 20% decline over the three-month period following the initial policy action
Residuals

0.60
0.60
0.36
0.36
0.12
0.12
-0.12
-0.12
-0.36
-0.36
-0.60
-0.60
1964
1967
1970
1973
1976
1979
1982
Residuals: Original Model
1985
1988
1991
1994
1997
2000
2003
Residuals: Model With Dummy
16
Conclusions
Conclusions

Implications for market risk
•
While the empirical results of our research indicate that changes in monetary policy reduce the yield spread, they do not
necessarily suggest that policy interventions reduce the level of perceived risk in the economy
– Rather, the results reveal that treasury yields respond more acutely to monetary policy action than those of Aaa
corporates
– When the model is regressed employing the 10-year yield and the Aaa index yield as unique dependant variables,
we find that the policy dummy is positive and significant in both cases, but that it has a decidedly greater impact on
the treasury yield (15 bps) relative to the index yield (9 bps)

Explanations
•
Because investors in treasury securities are more attentive to the actions of the Fed, it may be that greater trading
volume in the treasury market following a policy intervention results in a more rapid adjustment in the prices and yields of
those assets
•
An alternative explanation for our result is that treasury yields overreact to changes in policy relative to the Aaa index
yield. Because high-grade corporate bonds are traded so infrequently, their tempered response to the policy dummy
could indicate that the long-term outlook regarding market risk is not so severely altered as changes in treasury yields
suggest
18
Conclusions

Final Thoughts
•
While it could not be concluded from this research that open market operations reduce the level of perceived risk in the
economy, one might reasonably propose that changes in short-term market outlook following a policy intervention are not
always indicative of the long-term trend in risk perception
•
In any case, further exploration of the mechanisms linking monetary policy to reactions in the financial markets is an
intriguing topic for future research
18.00%
18.00%
16.00%
16.00%
14.00%
14.00%
Aaa Index Yield Trend
12.00%
12.00%
10.00%
10.00%
8.00%
8.00%
6.00%
6.00%
4.00%
4.00%
10-Year Yield Trend
2.00%
2.00%
0.00%
0.00%
1964
1967
1970
1973
1976
1979
1982
10-Year Yield
1985
1988
1991
1994
1997
2000
2003
Aaa Index Yield
19
Thank You