arch model.ppt
TRANSCRIPT
-
7/22/2019 ARCH Model.ppt
1/26
Time Series Econometrics:
ARCH/GARCH Models
Measuring volatility: Conditional
heteroscedastic ModelsK.R. Shanmugam, MSE
-
7/22/2019 ARCH Model.ppt
2/26
Features of Financial Data (e.g
Stock returns)
1. Volatility Clustering: Some periods are highlyvolatile while others are less. Big shocks(residuals) tend to follow big shocks in eitherdirection and small shocks follow small. Thisimplies strong AC. In addition, constant varianceassumption is inappropriate
(If unconditional (or long-run) variance isconstant but there are periods in which thevariance is relatively high. Such series are calledconditionally heteroscedastic).
-
7/22/2019 ARCH Model.ppt
3/26
Features.
2. Leverage Effect: Volatility is higher in a
falling market than it is in a rising market
(there is a tendency for volatility to rise
more following a large price fall than
following a price rise of the same
magnitude).
-
7/22/2019 ARCH Model.ppt
4/26
Features
3. Leptokurtosis: They have distributionswhich exhibit flatter tails and excess peakedness
(due to a large number of excessive values).
(a normal distribution is skewed (3rd moment)and has a coefficient of kurtosis of 3 (it is
symmetric and said to be meso-kurtic)
Leptokurtic is one which has flatter tails and ismore peaked at mean than normal with same
mean and variances.
-
7/22/2019 ARCH Model.ppt
5/26
NSE (2/1/1996 to 20/12/2002)
600
800
1000
1200
1400
1600
1800
1996 1997 1998 1999 2000 2001 2002
CLOSE
-.12
-.08
-.04
.00
.04
.08
.12
1996 1997 1998 1999 2000 2001 2002
RETURNS
-
7/22/2019 ARCH Model.ppt
6/26
Impact of special characteristics
They lead to violations ofhomoscedasticity as well as auto-correlation assumptions of OLS
(consequences of heteroscedasticity:estimates of parameters are unbiased, butSEs are large, CIs are very narrow, andprecision is affected)
Linear Models are unable to explain thesespecial features
-
7/22/2019 ARCH Model.ppt
7/26
Tasks of the Asset Holder
He may be interested in forecasting the rate ofreturn and the variance of the stock asset overthe holding period.
ARMA models are useful to forecast meanreturns. But it ignores the risk factor (variance orSD is a measure of risk or volatility)
Some series are subject to fat tails, volatility
clustering and leptokurtic, the task is to specifyand forecast both mean and variance of theseries conditional on past information.
-
7/22/2019 ARCH Model.ppt
8/26
Historical Volatility
Calculating the variance of the series in theusual way over some historical period
(e.g. In option pricing model, historical average
variance is used as volatility measure) Rolling standard deviation (RSD): Studies such
as Tauchen and Pitts (1987) used this. Theycalculate standard deviation using fixed number
of observations. That is, first calculate it usingmost recent (say) 22 days data. Then drop firstday and add 23rd day data. This is also calledOfficiers approach.
-
7/22/2019 ARCH Model.ppt
9/26
Problems with RSD
It uses equal weight for all cases (more
recent observations should be more
relevant and be given higher weight)
Use of overlapping observations lead to
correlation issues
Zero weight for other observations are
unattractive
-
7/22/2019 ARCH Model.ppt
10/26
Summary of Problems
unconditional forecast has a greater
variance than the conditional forecast, plus
overlapping problem and usage of equal
weighting.
So conditional forecast (since they take
into account the known current and past
realization of series) are preferable.
-
7/22/2019 ARCH Model.ppt
11/26
Engle (1982) ARCH Model
It says that the variance of the error term at time
t depends on the squared error terms from
previous periods:
Rt =mt + et and et = N(0, t2) .(1)(or et = vtt ; vt ~ N(0, 1)) where
2t =0 + 1 et-12 + 2 et-2
2++ p et-p2..(2)
(here the AC in volatility is modeled) this is an ARCH (p) model
ARCH(1) Model: 2t =0 + 1 et-12
-
7/22/2019 ARCH Model.ppt
12/26
ARCH TESTARCH (joint) TEST:To choose number of lagged terms.
If we start with one lag, this test will tell us whether weneed to add any additional lag term.
Steps:
(i) Run mean regression Rt =mt + et and save residuals et
(ii) Squared the residuals and regress it on its own laggedterm to test for ARCH (e2t =0 + 1 et-12 + 2 et-2
2++p et-p
2
(iii) Define TR2 ~ 2 (p), where p is number of lagged termon the right has side of second equation
(iv) Null Hypothesis: H0: 1=0; 2=0. p =0Alternate Hypothesis: H1 : 10; 2 0. p 0
(v) If test value is greater than critical value, reject the null
-
7/22/2019 ARCH Model.ppt
13/26
Properties of ARCH (1) process
et = vtt and 2t =0 + 1 et-1
2
t = (0 + 1 et-12)1/2
Since E vt =0; E et = 0
Since E vt vt-i = 0, E et et-i =0
Unconditional variance:
E et2 = E [vt
2 (0 + 1 et-12)]
Since E vt2 =1, E et
2 = 0 + 1Eet-12
2 = 0 + 12 = 0 /(1- 1)
-
7/22/2019 ARCH Model.ppt
14/26
Conditional Mean and Variance
E [et | et-1, et-2..] =0
E [et2 | et-1, et-2..] = E [vt
2 (0 + 1et-12)]
2
t = 1. (0 + 1et-12
) Thus, the conditional variance depends on
realized value of et-12.
If et-12
is larger, the conditional variance in twill be larger as well.
-
7/22/2019 ARCH Model.ppt
15/26
ARCH (1) Model
Unconditional variance:
E et2 = 0 /(1- 1) ; we restrict that 0>0
and |1|
-
7/22/2019 ARCH Model.ppt
16/26
Conditional Forecast Vs
Unconditional Forecast
Engle (1982): conditional is better thanunconditional
Example: Consider an AR(1) Model:
yt = a0 + a1 yt-1 + et Conditional Forecast of yt+1:
E (yt+1|t) = a0 + a1 yt
Forecast Error Variance is:
E [(yt+1- a0 - a1 yt)2] = E et+1
2 =2
-
7/22/2019 ARCH Model.ppt
17/26
Unconditional Forecast of y
yt= a0 + a1 yt-1 + e1
yt (1-a1L)= a0 + et
yt = a0/( 1-a1L) + et/(1-a1L)
yt+1 = a0 /(1-a1L) + et+1 / (1-a1L)
E yt+1 = a0/ (1-a1L) = a0 / (1-a1): Long run mean
Unconditional Forecast of Error Variance
E [yt+1- Eyt+1]]2 = E (et+1/(1-a1)) =
2/ (1-a1)
-
7/22/2019 ARCH Model.ppt
18/26
Estimation
ML Estimation Method
Log-Likelihood function using a normality assumption forthe disturbance is:
ln L = -T/2 ln (2)-(1/2) ln t2 (1/2)[(yt
a0-a1 yt-1)2 (1/t2)] It is an updating formula: Observed variance of the
residuals is taken for 1st observation; then it calculatesvariance for the second and so on for any given set ofparameter values; thus the entire time series of varianceforecast is constructed; then the likelihood functionprovides a systematic way to adjust the parameter togive the best fit.
-
7/22/2019 ARCH Model.ppt
19/26
Generalized ARCH
Developed independently by Bollerslev (1986) andTaylor (1986)
Bollerslaves GARCH Specification:
Rt =mt + et, where et ~ N(0, t2)
2
t =0 + 1 et-12
+ 2 et-22
++ 1t-12
+ It allows conditional variance to be dependent on
previous own lags
It is a weighted average of long term average (variance),information about volatility during previous years and
fitted variance in previous years. Such an updating formula is a simple description of
adaptive or learning behaviour.
Estimation is same as ARCH
However, we use parsimonious model; GARCH (1,1) is
sufficient
-
7/22/2019 ARCH Model.ppt
20/26
GARCH (1,1) Vs. ARCH
Proof: Let 2t =0 + 1 et-12 + t-1
2......(1)
2t-1 =0 + 1 et-22 + t-2
2.................(2)
2t-2 =0 + 1 et-32 + t-3
2(3)
Substitute (2) in (1) and then (3) in 2t =0 (1+ +
2) + 1 et-12 (1+ L+ 2L2) + 3t-3
2
An infinite number of successive substitutions leads to
2t =0 (1+ + 2+.) + 1 et-1
2 (1+ L+ 2L2+.) + 02
The last term is almost zero
The rest is similar to ARCH infinite specification
Although GARCH (1,1) has 3 parameters in conditional varianceequation (parsimonious), it allows an infinite number of past squarederrors
-
7/22/2019 ARCH Model.ppt
21/26
(1) GARCH (1,1)
2
2 2 2
0 1 1 2 1
t t t
t t t
r
where t-1
and t-1
are ARCH and GARCH
terms respectively.
-
7/22/2019 ARCH Model.ppt
22/26
The (G)ARCH-M Model
2 2
2 2 2
0 1 1 2 1
t t t t
t t t
r
In this application, the dependent variable in the mean
equation is determined by its conditional variance.
For instance, the expected return on an asset is related
to expected asset risk.
-
7/22/2019 ARCH Model.ppt
23/26
The Asymmetric ARCH Models
Often, we notice that the downward movement in the
market are highly volatile than upward movement of the
same magnitude. In such case, symmetric ARCH model
undermine the true variance process. Engle and Ng
(1993) provide a news impact curve with asymmetricresponse to good and bad news.
Good newsBad News
Volatility
-
7/22/2019 ARCH Model.ppt
24/26
two types of asymmetric ARCH
models: TARCH and EGARCH
2
2 2 2 2
0 1 1 2 1 1 1
t t
t t
1 1
where d 1if < 0 and 0 otherwise
0 good news, 0 bad news
They have differential effects on conditional var.Good (bad) news has an effect of ( )
If 0,
t t t
t t t t t
r
d
there is leverage effect (bad news increases volatility).
if 0, the news impact is asymmetric.
The TARCH or Threshold ARCH due to Zakoian
(1990) can be defined as:
-
7/22/2019 ARCH Model.ppt
25/26
The EGARCH Model (by Nelson
1991)
2 2 1 10 1 1 3
1 1
log log t tt t
t t
If 0, there is leverage effect.
if 0, the impact is asymmetric
Log of conditional variance equation. This means that leverage effect is
exponential than quadratic. This guarantees that forecasts of conditional
variances are positive;
-
7/22/2019 ARCH Model.ppt
26/26
Power GARCH (PARCH)
Taylor (1986) and Schwert (1989) developedS.D GARCH model. The conditional variance isnot a linear function in lagged squared residuals
where >0.
For symmetric model =0; if0, asymmetric
effects are present If=2 and =1 for all i, then the model is
standard GARCH
2 20 1( ) ( ) ( )t i t i t i j t