Transcript A composite leading indicator for a small transition economy– the

A CLI for a small transition economy –
the case of Croatia
Amina Ahec-Sonje
Katarina Bacic
Presentation will consist of…
 Explanation of what the Croatian composite prognostic index is …
 How the system’s reference series was chosen
 What methods were used to develop/review the Croatian composite
prognostic index
 How the prognostic index is computed and interpreted
 Final remarks and future directions/tasks
A CLI for a small transition economy > the case
of Croatia
2
CROatian LEading Indicator (CROLEI) today
• average lead time is 5,7 months
• calculated on monthly basis
• record in forecasting turning points in the Croatian business
cycle:
1995 – 2 mths. ; 1998 – 4 mths.;
1999 - 3 mths.; 2002 - 8 mths. (link)
11 components:
-monetary and financial (5)
-real (4)
-fiscal (1)
-external trade (1)
A CLI for a small transition economy > the case
of Croatia
3
Development of CROLEI

NBER methodology applied in 1993/94, first published in 1995

Main obstacle> time series > short, with breaks, frequent revisions

Prognostic system with identified lead, lag and coincident indicators

Methodological revision of the system every 2 yrs.

Importance of a meticulously run database, 104 time series in 2004
PROCESS‫׃‬
1.
Identifying reference series > volume of industrial production
2.
Adjusting series for inflation and seasonal adjustment
3.
"SCORING “ method
4.
Calculation of the index
A CLI for a small transition economy > the case
of Croatia
4
Reference series: INDUSTRIAL PRODUCTION
 Best option >GDP unsuitable because of Q. basis
 Second best choice >ind. production index, moth. Basis
 Industry overlaps with GDP, both as seasonally adj. series and in
trend/cycles
 One shortcoming > represents only 28% of TVA
1997=100
130
125
125
120
120
115
115
110
110
105
105
100
100
95
95
90
90
1Q97
2Q97
3Q97
4Q97
1Q98
2Q98
3Q98
4Q98
1Q99
2Q99
3Q99
4Q99
1Q00
2Q00
3Q00
4Q00
1Q01
2Q01
3Q01
4Q01
1Q02
2Q02
3Q02
4Q02
1Q03
2Q03
3Q03
4Q03
1Q04
2Q04
130
GDP
1Q97
2Q97
3Q97
4Q97
1Q98
2Q98
3Q98
4Q98
1Q99
2Q99
3Q99
4Q99
1Q00
2Q00
3Q00
4Q00
1Q01
2Q01
3Q01
4Q01
1Q02
2Q02
3Q02
4Q02
1Q03
2Q03
3Q03
4Q03
1Q04
2Q04
1997=100
Industrial Production
A CLI for a small transition economy > the case
of Croatia
GDP
Industrial Production
5
2. GRAPHIC AND SCORING method
GRAPHIC METHOD>

graphically comparing potential indicators against industrial trend-cycle

Results in a list of coincident indicators, leads and lags

Subjective evaluation, complemented with objective…
SCORING METHOD> N=25

Evaluation and analysis of the role & quality of series

Follows NBER methodology, except in one step

Series are scored according to their>
1.
2.
3.
4.
5.
Economic significance
Statistical adequacy
Economic significance of the link with the ref. series
Smoothness
Currency of statistics
A CLI for a small transition economy > the case
of Croatia
20%
20%
30%
20%
10%
6
…HOW WE SCORED SERIES…
1. Score of economic significance > key activities in the cyclical eco. process
2. Statistical adequacy > the quality of statistical methods of collecting data
3. Economic significance of the link with the ref. series > instead of the conformity
to the business cycle, we use bi-variate Granger causality test (Auerbach, 1982):
H: Leading indicator does not Granger cause CROLEI (ß=0)
IND t  A 0Dt 

k
j 1
αjIND t  j 

k
j 1
βjLI t  j  εj
if H is rejected (ß≠0), then series are possible LI and R2 is used for scoring
4. Score of smoothness > seasonal ARIMA models and the measure MCD-months for
cyclical dominance
5. Currency of statistics > promptness of data and data availability
A CLI for a small transition economy > the case
of Croatia
7
SCORES AND THE SELECTION OF COMPONENTS
Score
Criterion
I. Economic significance (20%)
20
II. Statistical adequacy (20%)
20
III. Economic significance of the link (30%)
30
(2)r16
IV. Smoothness (20%)
V. Currency of the statistics (10%)
6
VI. Total score (I+II+III+IV+V)
92
Narrower selection of leads is founded on 4 requirements:
 a non dominant, ignorable irregular component;
 high total score, with 70 points as a threshold for entering the index;
 long enough lead time, starting from t-4;
 priority is given to economic aggregates
(plus to be careful that series that are representatives of the same activity, but on a
different degree of aggregation do not enter the index)
A CLI for a small transition economy > the case
of Croatia
8
RESULTS
 25 indicators scored with scores ranging from 46 to 92 points
 13 series that have traditionally shown to be good indicators kept their
good scores / again are eligible
 11 components chosen following the principle of wide/complete coverage
of different areas an economy
Series
Code
S43
S48
S54
S58
S92
S95
S107
S109
S112
S123
S136
Indicator
Nominal net wage per employee
Employed with the help of the Croatian Employment Agency during a month
Tourist nights, total
Retail trade, real
Machinery and transport equipment import
Unconsolidated revenues of state, county and municipal budgets
Broadest money M4
Reserve money
Time and Sav. Deposits with Dep. Money Banks, in dom. currency
Reserves on time and notice deposits
Money market rate, daily
A CLI for a small transition economy > the case
of Croatia
Lead time
(t-4)
(t-4)
(t-9)
(t-8)
(t-8)
(t-4)
(t-4)
(t-5)
(t-4)
(t-8)
(t-5)
9
Scores and significance weights
Series
Lead
time
Economic
Statistical
significance
adequacy
(20%)
(20%)
Smoothness
(20%)
Currency of
the statistics
(10%)
Economic
significance of
the link
score
(30%)
(Si)
Total
Weights
(Wi)
S43
-4
100
95
60
60
100
87
1,04
S48
-4
100
100
40
100
100
88
1,05
S54
-9
100
100
40
60
100
84
1,00
S58
-8
100
95
60
60
100
87
1,04
S92
-8
80
85
0
60
90
66
0,79
S95
-4
100
100
0
100
100
75
0,89
S107
-4
100
100
100
60
30
75
0,89
S109
-5
100
100
80
60
100
92
1,07
S112
-4
80
100
100
60
100
92
1,10
S123
-8
80
100
80
60
100
88
1,05
S136
-5
100
100
80
60
100
92
1,10
926,0
11,00
TOTAL
Average
84,18
-5,7*
A CLI for a small transition economy > the case
of Croatia
10
Computation of the CROLEI
 computed by following the NBER/BEA instructions (BEA/USDC, 1977;
Zarnowitz i Boschan, 1975; Gapinski, 1982; Shiskin, 1961) that consist of
five methodological steps and those are:
-
Computing symmetric (Shiskin’s) percentage changes;
Standardization of the amplitude;
Weighting of the standardized changes;
Standardization of Rt
Turning monthly changes into the index
RESULT:
a prognostic expression that may be seen as a weighted
average of the chosen leads
A CLI for a small transition economy > the case
of Croatia
11
Interpretation of the CROLEI
Few types of interpretation>
Vaccara and Zarnowitz (1977) : consecutive three falls/rises (first difference
between current and the previous month) in the composite index value
signals a recession/recovery of the total economic activity
Stekler and Schepsman (1973): 4 month fall (or rise) below/above the last
lowest/highest point in the index to be considered as an interpretation
criterion for turns
Additional measure - the diffusion index complements the interpretation of
the prognostic expression. Its value may fall into three possible ranges:
a)
0 – 50 range> a recession or a fall;
b)
50 – 75 range> a moderate growth
c)
75 – 100 range> an acceleration or an expansion
A CLI for a small transition economy > the case
of Croatia
12
CONCLUSION
 Construction of CROLEI follows the original NBER method to the extent that
it is possible
 Throughout the revisions of CROLEI, results of the graphic and scoring
procedure remain consistent in that the same series obtain good scores
(list)
 In the modified step in the scoring procedure, we follow recommendations
on what to do from the economic literature – result: CROLEI predicts
movement of the reference series and not only its turns
 Revised CROLEI indices appear to have a very similar cyclical pattern to the
old CROELI despite the introduction of some new components (figure)
 The new CROLEI contains several series that were already a part of the old
indices Shorter lead time does not imply a loss in the prognostic power (list)
 Quality of the index improves with every new revision because of the
introduction new series in the database and because time series are getting
longer
A CLI for a small transition economy > the case
of Croatia
13
Future directions and tasks
 A new revision of CROLEI is due in 2007
 Priority will be given to determining turns in a business cycle using
a scientific method instead of only using subjective evaluation
 Carefully monitoring when the moment will be right for applying
all of the original steps of the barometric method
 In due time developing prognostic indices for the Croatian regions
A CLI for a small transition economy > the case
of Croatia
14
CROLEI from 1999 and CROLEI from 2004
A CLI for a small transition economy > the case
of Croatia
15
13 persistent leads

Employed with the help of the Croatian Employment Agency during a month


Users of unemployment compensation
Nominal net wage per employee


Tourist nights, total
Foreign tourist nights, total


Retail trade, real
Unconsolidated revenues of state, county and municipal budgets


Broadest money M4
Reserve money


Money M1
Money M1a


Deposit Money Banks’ Claims
State budget expenditures, total
A CLI for a small transition economy > the case
of Croatia
(return)
16
6 series that appear in CROLEI from ’99 and from ‘04
 Employed with the help of the Cro. Employment Agency during a month
 Tourist nights, total
 Retail trade, real
 Unconsolidated revenues of state, county and municipal budgets
 Broadest money M4
 Time and Sav. Deposits with Dep. Money Banks
A CLI for a small transition economy > the case
of Croatia
17
CROLEI and INDUSTRY 1/95-7/04
2000=100
A CLI for a small transition economy > the case
of Croatia
(link)
18