Content uploaded by J. Scott Armstrong
Author content
All content in this area was uploaded by J. Scott Armstrong on Feb 10, 2021
Content may be subject to copyright.
!
"" !
#$
!%&'()*+,
•-!.!!/0
•#1!2!!!
!
&
!.!3!
!!!
!4!
!2
-!!
5.56+!
#!!!
672
8
(
•*2!2$
!.!
62
•*& 2
62
•! !most accurate
forecasts9:
%
*!!;
no beer
;
*&
<
9.=>?!$
!$. $
@)A+'B2!,
Expert polical judgment).&''<,
&+%$ !+&(C 6
!! &'9
!
C
!
4.!
;
9.6
;2
D
"
6E 2
!;
; 6
F!2
5.2$ will happen
6:
+
•!92!!
6
•!92!
!)0FEThe CureBG
H!2,
•H2
!6$
!!6
A
#
•*!6<(
•I9!!=
–5656H
–56 H6
–Journal of Forecasng
–Internaonal Journal of Forecasng
–ForecasngPrinciples.com
•.!! !!9
2! 6A+'
•F!!!!Long-Range Forecasng
)AD+A+<) ! Principles of Forecasng )&'',
•--2-$5>".6 @
$
•H6! 62K
!
•6.2 !!
.)! !!2
=>5J*6@,
•H2!!2 6!
!L2!! .!
!92!62 !
*9! 5*
LH$ 66K!."!!6
56K!66K2
&
%
•9!
•?6K!
–M6
–M0!*H6
•M!6!02
•0696K 6;
!!
(
&'
*6 N!!
!92!!!=
J92
& 26
( I9O6 !
All three are necessary
!6K 6
%
()
-!6!5II
!D&+A =
!!!!!
& 2!!!
( I !O!
4. Be conservave 6
! $
=02
=H266KH
<
()detrimental
eects
• 6 :
•5:
•5.:
C
()cost-
eecve policies
•6 !6 2
!!
•Findings: H%P6
!
•Examples:
–H ''!2!!
K)6,
–-=>0 H!!
-J6"@
–* !!!
•=- 6HB#
D
&*$+,
•H2.
–*6
–-
•J
•!
•62!.!
–*
–"$
•6.!
–4!92!!
+
-
•8
4 6=
–I $
–#
–Q.22!
•I6=
–0
–5!
A
-./"
H!26
.!!=
•6
•H6!!!
!6
&'
%
•26)!B!
2,
•!!
•4$ 6!
•I!
•I2!!!
!!!
&
&
•!92!6!!!
6
•!!!R!
•" !6
2
• .!2!6
O
•I2!!!!
!!
&&
0!1%%&*"
•5II !!0
! !.!0*
0!>I.@
–?6 !.
!
•7!
.5.'
–&<&D!R!!2
–Q5II!&<
•=0!*H6)&'(,
&(
-&*"
• !
62
•H .
•H!!
8
&%
1&*.2/
#!.!O:
Q $
-. 62
4!>.
2@
&<
&
<'
&C
*
• !6
6N:
•# !!!
2
• !
•- !
&D
34%
)=G&''A,
-
&<
!
)S'DTIB
S''(
,
&+
$
!!!
-N62!:
!
!:
I
0!*0!=
J. ! !
5!
!)942
6!,
&A
5+6&.,/
• !20!
need !6!)G66=
!24!#!,
•Terms:
–'02G
•Posions:
–0=>6 @>@
–='9
•0!U2!
!6
• !!2
('
(
7
! $!9
>2.!@
•!!25II
)''(TI9 9,/GQV#!IE
!+<'9&''D
•H9)''!,9
!E89:5II!E
)D<<' !6,
•HFA''9!E
898;5II!E
=0W)&''A,
(&
5 >6@)6
,
& !R.2#!!
( I!! !!9
% I2!92!
#!K !
G 9!N
)G4<'9!'&%I,
((
!
HW/L! =
!$$ 6!
5II &'9!
& $!92!! 2
6
( !! $!$ !
!
LH!*W/F?)&',
M!6!6!Internaonal Journal of
Forecasng27AC+8AA<)26,
(%
$<=
! ! $!6
& !26!
( I!!!O!29
96
% I2!!92!!
< I!!!6!.!
C !!!
D !!!! 2!
! N62
#.!>N62@#!!
(<
$<=>
>?@
Horizon (years) Forecasts per
model
Improvement
vs. naïve
1–20 1,190* -2%
11–20 500* -8%
LH !!2 N
(C
M!6 2
&''+#!
!2 !! 2 6
-$! ! !2
G 6 2 6!
9O
4!
2 6!)G
,
(D
$,
+
0W)0&'(9.,
I!
& " !! $!
( !R!6!.2
% 6!6!
< 52!6
(+
$&6?;A8'@
5 !!)2!
! ,2P9
! 99!
9!69
!!6!6!
9!
HN
–5
:
(A
%+&
an)
•2
>!@
•!!
2
)Madness of Crowds,
%'
•QX!2
•!.!
..2!
!
%
1. Generate analogies=
.!$ !2!!
2B!
2. Assess similarity=
I! 2!!! !
06B! !!
3. Thumbnail sketches=
!2! 6
4. Assess outcomes:
Q2!!E 2!
5. Forecast for target=
"!202!
41
%&
I!!!6!+16
Example=6)I=DPB='PB&Y=<'P,
H!AD!!'C
!!$ =
# B
0 &+
4$ !!R! (&
<C
=0!)&''D,
%(
6&-
•D !)B$ ,
•&C
–I
–G!
–#
•?2!6K!
•06!!
%%
! !6!!6
),8++()GI=H,
& !68A(%
( ""!8AC&
% 6!8AC+
< 028AD<
C 4!8ADA
D GK8&''%
%<
C
Z0.6&(&C
Z? !!
ZG!!!
Z0$ !! X
0 !.
!2
6
=0!
>4O02@)&',
%C
-8DD;
!
6!2 6
6K2R6
#!!!E
!.
32
%D
1
!! !!:
•5 likelihood2!
E
!B 26! 6
W[.)&',
•-O 6!
$
%+
12
7
* 66
& 46!O6
3
( * 66O6 !2
!B!
% 4!2 6
.
< >E@)J,
H!92! 2!
Persuasive Adversing
%A
1
•Q99!
•J
<'
-;AA::
•? 2
•0229
! 2!
•M66F
2\(2!!
2
<
$
!2!""
..
Q >4!92!6
@! 6Q2
2 !2
. !
K!