econometrics i 25
TRANSCRIPT
-
7/25/2019 Econometrics I 25
1/25
Part 25: Time Series5-1/25
Econometrics I
Professor William Greene
Stern School of Business
Department of Economics
-
7/25/2019 Econometrics I 25
2/25
Part 25: Time Series5-2/25
Econometrics I
Part 25 Time Series
-
7/25/2019 Econometrics I 25
3/25
Part 25: Time Series5-3/25
Modeling an Economic Time Series
!ser"ed #$% #&% '% #t%'
What is the (sample) *andom sampling+
The (o!ser"ation ,indo,)
-
7/25/2019 Econometrics I 25
4/25
Part 25: Time Series5-4/25
Estimators
-unctions of sums of o!ser"ations
.a, of large num!ers+ /onindependent o!ser"ations
What does (increasing sample si0e) mean+
1s#mptotic properties+ There are no finite
sample properties34
-
7/25/2019 Econometrics I 25
5/25
Part 25: Time Series5-5/25
Interpreting a Time Series
Time domain: 1 (process) #t4 a6t4 7 !#t8&4 7 '
*egression li9e approachinterpretation
-re;uenc# domain: 1 sum of terms #t4
-
7/25/2019 Econometrics I 25
6/25
Part 25: Time Series5-6/25
-or e6ample%'
-
7/25/2019 Econometrics I 25
7/25Part 25: Time Series5-7/25
In parts'
-
7/25/2019 Econometrics I 25
8/25Part 25: Time Series5-8/25
Stud#ing the -re;uenc# Domain
-
7/25/2019 Econometrics I 25
9/25Part 25: Time Series5-9/25
1utocorrelation in *egression Yt = bxt + t
Cov(t, t-1) 0
Ex. RealConst= a + bRealIncome + t U.. !ata, "#a$te$l%,
1&'0-000
-
7/25/2019 Econometrics I 25
10/25Part 25: Time Series5-10/25
1utocorrelation
=o, does it arise+ What does it mean+ Modeling approaches
direct: corrective Estimation that accounts for autocorrelation Inference in the presence of autocorrelation
structural
Model the source Incorporate the time series aspect in the model
-
7/25/2019 Econometrics I 25
11/25Part 25: Time Series5-11/25
Stationar# Time Series
0t !t8&7 !2#t827 ' 7 !P#t8P7 et 1utoco"ariance: @9
-
7/25/2019 Econometrics I 25
12/25Part 25: Time Series5-12/25
Stationar# "s3 /onstationar# Series
T
-5.73
-1.15
3.43
8.01
12.59
-10.31
20 40 60 80 1000
YT ET
Va
ria
b
le
-
7/25/2019 Econometrics I 25
13/25Part 25: Time Series5-13/25
The .ag perator
.6t 6t8&
.26t 6t82
.P6t7 .H6t 6t8P7 6t8H
Pol#nomials in .: #t B.4#t7 et
1.4 #t et
In"erti!ilit#: #t A1.48&et
-
7/25/2019 Econometrics I 25
14/25Part 25: Time Series5-14/25
In"erting a Stationar# Series
#t #t8&7 et&8 .4#t et
#t A&8 .8&et et7 et8&7
2et827 '
Stationar# series can !e in"erted
1utoregressi"e "s3 mo"ing a"erage form of series
2 31 1 ( ) ( ) ( ) ...1
L L LL
= + + + +
-
7/25/2019 Econometrics I 25
15/25Part 25: Time Series5-15/25
*egression ,ith 1utocorrelation
#t 6t! 7 et% et et8&7 ut
&8 .4et utet &8 .48&ut
EAet EA &8 .48&ut &8 .4
8&EAut $
JarAet &8 .482JarAut &7
2u27 ' u
2&8 24
-
7/25/2019 Econometrics I 25
16/25Part 25: Time Series5-16/25
.S "s3 G.S
.S ?n!iased+
-
7/25/2019 Econometrics I 25
17/25Part 25: Time Series5-17/25
+----------------------------------------------------+
| Ordinary least squares regression |
| LHS=REALCONS Mean = 2999!"# || Auto$orrel %ur&in-'atson Stat = (92(!)( |
| R*o = $ore,e-./0 = 91"9#( |
+----------------------------------------------------+
+---------+--------------+----------------+--------+---------+----------+
|3aria&le | Coe44i$ient | Standard Error |t-ratio |5|6|7t0 | Mean o4 8|
+---------+--------------+----------------+--------+---------+----------+
Constant -)("1!!)) .!"(1)1.1 -1#. ((((
REAL%5 92.#)1# ((").1 2")(1! (((( ""!.!19)
| Ro&ust 3C Ne:ey-'est, 5eriods = .( |
Constant -)("1!!)) !.2"92.! -.92# (111
REAL%5 92.#)1# (.1("1.# #."(2 (((( ""!.!19)+---------------------------------------------+
| AR./ Model; et/ = r*o < et-./ + ut/ |
| inal >alue o4 R*o = 99))2 |
| ter= #, SS= ..)"#((, Log-L=-9!.".9.! |
| %ur&in-'atson; et/ = ((2!"# |
| Std %e>iation; et/ = !9(1#9.( |
| Std %e>iation; ut/ = 2!2(#92# |
| %ur&in-'atson; ut/ = .99!91 |
| Auto$orrelation; ut/ = ((212. |
| N(,.0 used 4or signi4i$an$e le>els |+---------------------------------------------+
+---------+--------------+----------------+--------+---------+----------+
|3aria&le | Coe44i$ient | Standard Error |&?StEr|5|@|70 | Mean o4 8|
+---------+--------------+----------------+--------+---------+----------+
Constant .(.9"2#)( !....1# 2!9 (."2
REAL%5 #"!2". ("9219" .#912 (((( ""!.!19)
RHO 99)).). (("!#""2 2))")9 ((((
-
7/25/2019 Econometrics I 25
18/25Part 25: Time Series5-18/25
Detecting 1utocorrelation
?se residuals Dur!in8Watson d
1ssumes normall# distri!uted distur!ances strictl#
e6ogenous regressors
Jaria!le addition Godfre#4 #t 6t 7 Kt8&7 ut
?se regression residuals etand test $1ssumes consistenc# of !3
2
2 1
2
1
( )2(1 )
T
t t t
T
t t
e er
e
=
=
-
7/25/2019 Econometrics I 25
19/25Part 25: Time Series5-19/25
1 ?nit *oot+
=o, to test for &+
B# construction: Kt> Kt8& 8 &4Kt8&7 ut
Test for @ 8 &4 $ using regression+ Jariance goes to $ faster than &T3 /eed a ne, ta!leL
cant use standard t ta!les3
Dic9e# > -uller tests
?nit roots in economic data3 1re there+4 /onstationar# series
Implications for con"entional anal#sis
-
7/25/2019 Econometrics I 25
20/25Part 25: Time Series5-20/25
*einterpreting 1utocorrelation
1
1 1 1 1
t
t
Regression form
' ,
Error Correction Form'( ) ( ' ) , ( 1)
' the equilibrium
The model describes d!ustment of " to equilibrium #hen
$ chnges.
t t t t t t
t t t t t t t
t
y x u
y y x x y x u
x
= + = +
= + + =
=
-
7/25/2019 Econometrics I 25
21/25Part 25: Time Series5-21/25
Integrated Processes Integration of order P4 ,hen the Pth differenced
series is stationar#
Stationar# series are I$4 Trending series are often I&43 Then #t> #t8& #t
is I$43 AMost macroeconomic data series3
1ccelerating series might !e I243 Then#t> #t8&48 #t> #t8&4 2#tis I$4 AMone# stoc9 in
h#perinflationar# economies3 Difficult to find
man# applications in economics
-
7/25/2019 Econometrics I 25
22/25
Part 25: Time Series5-22/25
-
7/25/2019 Econometrics I 25
23/25
Part 25: Time Series5-23/25
Di"ergent Series+
O bs erv.#
4.30
8.48
12.67
16.85
21.04
.12
20 40 60 80 1000
YT %T
Variable
-
7/25/2019 Econometrics I 25
24/25
Part 25: Time Series5-24/25
-
7/25/2019 Econometrics I 25
25/25
P 25 Ti S i5 25/25