(UNTITLED)
Document Type:
Collection:
Document Number (FOIA) /ESDN (CREST):
CIA-RDP88T00096R000800960002-8
Release Decision:
RIPPUB
Original Classification:
C
Document Page Count:
75
Document Creation Date:
January 12, 2017
Document Release Date:
May 20, 2011
Sequence Number:
2
Case Number:
Publication Date:
December 1, 1987
Content Type:
REPORT
File:
Attachment | Size |
---|---|
![]() | 2.87 MB |
Body:
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Directorate of
Intelligence
ILLEGIB
Economic Analysis
The Israeli Econometric
Model: A Key Tool for
A Technical InteWgence Report
NESA 87-10051
December 1987
copy 3 4 7
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Economic Analysis
The Israeli Econometric
Model: A Key Tool for
Division, NESA,
Office of Near Eastern and South Asian Analysis.
Comments and queries are welcome and may be
directed to the Chief, Issues and Applications
This paper was prepared by
Confidential
NESA 87-10051
December 1987
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Confidential
Summary
/rt/ormation available
as of 1 September 1987
was used in this report.
Economic Analysis
Model: A Key Tool for
The Israeli Econometric
The econometric model of Israel, developed in the late 1970s, incorporates
and its supporting data files have been revised and updated.
information on government policies, international economic conditions,
resource availabilities, and other factors. As the economy of Israel has
evolved and underlying conditions and policies have changed, the model
less reliable are the results of the model.
The model, written in the TROLL language, now contains almost 200
equations, each of which describes a different facet of the economy. These
equations can be solved for each of the last 10 years or the next 10 years.
Because projections depend on data inputs that are inherently less certain
than those pertaining to the past, the more distant the year projected, the
Most of the equations in the Israeli model, such as those computing the
balance of payments and disposable income, are definitional. These
equations describe identities that are true for all economies at any time.
Twenty other equations in the model represent behavioral and technologi-
cal relationships that give the Israeli economy its character and distinguish
it from other economies. For example, fuel imports appear to be a function
of their average price, Israeli GNP, and capital availability. The specific
form of each of these equations has been estimated from observations of
the Israeli economy during the last 10 to 20 years or is based on the
judgment of economic analysts. Because such relationships are subject to
change over time, projections based on these equations assume that the
relationships they represent will remain constant for the projection period.
of events over time.
The model can be divided into several sections, each describing a sector of
the Israeli economy. The most important of these deal with the availability
and use of resources in the economy, including private and government
consumption, investment, exports, and imports. Linkages among the
various sections of the model reflect linkages among the different sectors of
the Israeli economy. Moreover, the model is dynamic, showing the impact
The Israeli model lends itself to a variety of analyses. For example,
historical simulations of the Israeli economy in the first half of the 1980s
suggest that, as expected, monetary growth was a major contributor to the
Confidential
NESA 87-10051
December ~/ 987
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Confidential
25X1
Figure 1
Israeli Econometric Model
TRANSFERS-~ CURRENT ACCOUNT--~ CAPITAL ACCOUNT
l t
EXPORTS IMPORTS INFLATION
1
~a ~
MILITARY
MONEY
GNP ~ GOVERNMENT
triple-digit inflation in Israel at that time. A baseline projection of the
Israeli economy over the next two years reveals sluggish growth of output,
declining growth of private consumption and investment, and worsening
balances of trade and payments as imports rise faster than exports.
Scenario analysis with the model shows that increases in investment can
speed up growth of production during the next two years but would have
little effect on private consumption. Consumption would benefit from a
reduction in taxes. Moreover, tax cuts appear to reduce the budget surplus
less than matching increases in public investment. The model reveals that
increased investment is an expensive way to improve labor productivity.
Other approaches-such as better training and placement and productivity
inducements-should be explored.
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Confidential
The Israeli model was evaluated by ex post simulation of 1977-85
economic activities. A comparison of the simulation with actual observa-
tions suggests that the model and the supporting data base provide a
consistently accurate description of the Israeli economy during that period.
The model was also evaluated by comparing ex ante projections of 1985-86
economic activities with actual developments in those years. Performance
in these cases, which depended additionally on the accuracy of external
projections of certain economic activities, was not as consistently good as in
the case of the historical simulation. Performance was weakest for
projections of rates of growth and inflation, tax revenues, and international
payment balances.
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Confidential
Summary
iii
Scope Note
ix
The Structure
of the Model
1
GNP
1
Private Income and Consumption
4
Investment and Capital Stock
5
Labor
6
Government Finances
8
Foreign Trade
10
Exports
11
Imports
12
Balance of Payments
15
Applying the Israeli Model
17
Historical Analysis-Israeli Hyperinflation
19
Projecting the Future-The Baseline Case
19
Scenario Analyses
23
Policy Analysis
23
Goal Analysis
29
Evaluating the Israeli Model
31
Historical Tracking
31
Past Projections
32
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Confidential
Scope Note The Israeli economy, like other economies, is a complex set of interacting
phenomena. Some of these interactions occur quickly; others take place
over longer periods of time. Some involve only one or two factors; others
consist of many. The econometric model allows the analyst to study the
economy and to gain insights at a level of complexity that is impossible
with other approaches. Past, present, and future economic activities can be
simulated so that the analyst can look at the roots of current problems, ex-
periment with possible solutions, and project future implications.
The Israeli econometric model has been in use since the late 1970s.' This
paper describes the latest version, explains how it can be used, and presents
some illustrative findings. It also explains some of the limitations of this
type of analysis and provides an assessment of the validity of model results.
25X1
25X1
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Confidential
Model: A Key Tool for
Economic Analysis
The Israeli econometric model, written in the TROLL
language, consists of almost 200 equations.Z Each
describes a different aspect of the Israeli economy.
Twenty of these equations contain parameters that
have been estimated by econometric analysis of Israeli
economic data for the last 10 to 20 years. The model
is divided into sections describing Israeli production,
private consumption, investment and capital growth,
labor supply and employment, government revenues
and spending, exports and imports, and Israel's inter-
Figure 2
Israel: Real GNP
ISGR
ISEMI~ ~.
~
-~- ISYRp
ISKSY'78E
ISNETFR~ \
/~
l
/sl
\
\\
national financial position.
GNP
Israel's GNP was $21 billion in 1985. This equaled
Egypt's output of goods and services in that year,
although the Egyptian population then totaled 48
million while Israel's stood at only about 3.9 million.
The model estimates total Israeli output of goods and
services in 1980 shekels (figure 2) as aCobb-Douglas
function of capital (ISKSM78E), employment
(ISEMP), and real government consumption (ISGR):'
Log(ISYR) _ -1.288 + 0.04*DUM81
- 0.031*DUM84 + 0.366*Log(ISKSM78E)
+ 0.597*Log(ISEMP) + 0.037*Log(ISGR)
' ACobb-Douglas function is one in which the variable to be
explained, here ISYR, is expressed as a multiplicative function of
the explanatory variables:
ISYR = a*ISKSM78Eb*ISEMP*ISGRd
The exponents add to 1 (b+c+d=1). The equation implies that
ISCR: R*.l Co~*rnmmt Con.umpUon
1SEYP: 1Smplo~moat
tslcs>.Tea: c.plw sloak
LSYRp: cxp
ISYRa Roounu A~ait.Dlo
ISllnt: Rs*1 ImperL
ISIY7R: R*ol ImpoK Tu*.
ISIX: R**1 Into.tmont
YRRRS: C6on3* b Stoeb
IS100t: R*.l [ryorla
ISmfSR: Roal 1[zport Subddl*.
Variables in black are computed by identities.
Variables in blue are computed by historically
derived econometric equations.
Variables in red are exogenously determined.
1973-85 are explained by the equation specified here, the given
parameter estimates, and changes in the values of ISKSM78E,
ISEMP, and ISGR during that period.
DUM81 and DUM84 represent dummy variables for the years
1981 and 1984, respectively. They are used because Israeli output
of goods and services in these years is not fully explained by the
other variables in the model. By including a variable for each year
explicitly in the model, it is possible to estimate coefficients for
them that indicate the influence of the years themselves (or rather
the factors implicitly associated with these years) on the value of
The function is estimated from data for 1973-85 in its
logarithmic form in which the explanatory variables are additive
and the "a" term is separated into a constant and two dummy
variables. The coefficient of determination-commonly called the
R'~f this estimate equals 0.998, indicating that the function
explains 99.8 percent of the variation in ISYR during that time.
That is, almost all of the changes in the value of ISYR during
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Confidential
The macroeconometric model is a toollor studying
the economy of a country by simulating its activities.
It consists of a number of equations, each describing
a separate facet of the economy.
Most of the equations are accounting identities,
equally true of any country at any time. Thus, goods
and services used must equal those produced or
acquired in other ways, say imported from abroad or
drawn from inventories. If'uses include consumption
(CJ, investment (IJ, government services (GJ, exports
(XJ, and net additions to inventories (VJ and sources
consist of domestic production (Y) and imports (MJ,
one identity in the model would be:
Other model equations represent behavioral or tech-
nological relationships among the various compo-
nents of the economy. The exact form of such a
relationship is likely to be peculiar to a given country
and subject to change over time as government poli-
cies, social institutions, economic conditions, technol-
ogy, and other factors change.
Consumption, for example, may be a function of
population (PJ and disposable income (YdJ. A possible
form of this relationship is:
C/P = bl + b2*Yd/P
Here C, P, and Yd are variables, the values of which
change from year to year; bl and b2 are parameters
that describe the specific relationship among these
variables for a given country and time period and are
expected to change little, 4f at all, over time.
When the model is run, the economy is simulated one
year at a time for the entire simulation period. For
each year, each equation in the model is solved for
one distinct variable. To do this, the computer must
know the values of all the other variables and any
parameters in the egi:ation.
Variables computed by equations in the model are
said to be endogenous. Each must appear in at least
one distinct equation, and the model must have as
many equations as there are endogenous variables.
The values of other variables, termed exogenous, are
taken from external sources and must be supplied to
the model for each year of the simulation period.
Estimates of these values are taken from several
sources:
? Extrapolation of current trends.
? The judgment and research findings of other ex-
perts, both within the Intelligence Community and
at other government, academic, and research
institutions.
? Official country publications.
Projections of exogenous variables involve dlfjerent
degrees of uncertainty. In general, the most confi-
dence resides in those input values subject to little, if
any, policy manipulation and those that reflect long-
term trends not likely to be quickly reversed. The
least certain input data are those strongly
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Confidential
ir~uenced by factors such as policy decisions and
international market forces that may change
abruptly.
Where uncertainty is greatest, alternative input val-
ues would indicate the sensitivity ojmodel output to
the value ojthe variable in question. When irtforma-
tion for an input variable is absent, the model can be
used to estimate likely values by indicating the range
in which they must fall to be consistent with other
known data.
Econometric modeling encourages the collection of a
diverse body ojir~formation concerning the economy
and provides a means of checking its consistency.
This helps to assess the value ojexisting intelligence
and research activities and to suggest where addition-
al eh`orts may be needed.
Parameters are values that are expected to change
slowly or not at all during the simulation period.
These may be coefficients ojvariables in behavioral
and technological equations or values appearing in
identities. Projections with such equations are valid
only for the period in which the parameters are
expected to remain constant.
While coe,~cient estimates may be based on analyst
judgment or taken from other sources, most are
derived by econometric techniques from recent his-
torical data. The period involved varies from equa-
tion to equation, the goal being to find estimates that
cause the equation toht the historical data best
while meeting various consistency, bias, and other
criteria. In general, parameter estimates are consid-
ered unsuitable unless the resulting equation explains
at least 95 percent ojthe variation in the value ojthe
endogenous variable over the historical estimation
period.
Estimates attained in this way are likely to be
incorrect to some extent because of problems in the
underlying historical data base and limitations in the
estimation procedure. The procedure itseU, however,
assesses the probability ojany deviation ojthe true
coefficient valuejrom its estimated value.a It there-
fore is possible to estimate the probability that the
true value falls within any given range around the
estimate. In most cases an estimate is accepted only iT
there is less than a S percent chance that the devi-
ation is so large that the true value may be zero or of
the opposite sign.
a Statistical estimation of a parameter also yields the standard error
of the estimate. In general, there is less than a 5-percent probability
that the true value of the parameter is more than two standard
errors from the estimated value. This statistical rule of thumb
allows us to assess the validity of the parameter estimate. If the
standard error is greater than half the value of the parameter itself,
the probability that its true value is zero or of the opposite sign is
greater than 2.5 percent.
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Confidential
Government consumption is not a factor of produc-
tion. It is a demand variable. The public sector,
however, plays an important role in the Israeli
economy:
? It directly produces goods and services that account
for a large share of the total output of the economy.
? Government agencies provide entrepreneurial ser-
vices that facilitate the efforts of other sectors of the
economy, making their capital and labor more
productive, thus helping to increase their total out-
put along with that of the economy as a whole.
? Government spending, as large as it is in the
economy, almost certainly influences the decisions
of other producers. Increases in such spending prob-
ably encourage those producers to increase output.
It is, therefore, reasonable to expect the level of
government activities to be significant in explaining
the level of GNP.
It might be more appropriate to measure these activi-
ties by some government variable that reflects output
rather than demand. Government consumption, how-
ever, appears to be a good proxy for that variable. It
significantly improves the fit of the model to the
historical data, raising the expectation of more accu-
rate GNP forecasts.
Each coefficient in aCobb-Douglas function provides
an estimate of the relative change in GNP associated
with a 1-percent change in a factor of production.
Thus, a 1-percent increase in capital stock is associat-
ed with a 0.37-percent rise in real output. The
coefficients also indicate the share of total output
attributable to the individual factors. This equation
portrays an economy where labor inputs play a bigger
role than capital in determining output. Not surpris-
ingly, government spending is also important.
Net real additions to stocks (YRRES) are derived
from the difference between total availability of goods
and services~utput plus imports and import taxes
(ISIMR and ISIMTR) and total uses-private con-
sumption (ISPCR), total investment (ISIR), govern-
ment spending, exports (ISEXR), and export subsidies
(ISEXSR):
ISYR =ISPCR + ISIR + ISGR + ISEXR
+ ISEXSR -ISIMR -ISIMTR - ISNETFR
+ YRRES
Real additions to stocks are converted to current
terms using the investment deflator. Additions in
current terms along with consumption, investment,
government spending, exports, and imports in current
terms are used to arrive at GNP in current shekels
(ISY). A comparison of real and nominal estimates of
GNP yields the GNP deflator (ISYDEF) and the rate
of inflation:
ISYDEF = ISY/ISYR
Private Income and Consumption
Private consumption in Israel was two-thirds the level
of GNP, $13.7 billion, in 1985. In the model real
consumption in the private sector is estimated by a
behavioral equation from real disposable income
(ISDISPR), lagged private consumption, and the in-
flation rate for imports (DDIMPDEF) (figure 3):?
ISPCR = -2.74 + 0.216*ISDISPR
+ 0.825*ISPCR(-1) - 0.029*DDIMPDEF
Here the marginal propensity to consume out of
additional disposable income appears to be quite low.
But the rate of inflation in the import sector is a
significant inducement to private spending.
The rate of inflation for private consumption
(DDPCDEF) is estimated as a behavioral function of
the rate of growth of the money supply (DDM2) and
that of total military consumption (DDG2), both of
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Confidential
which are projected outside the model by analysts.
The rate appears to have trended upward during
1973-85 (DUM7385O) apart from the influence of the
other explanatory variables (TIME is an index for
year):5
DDPCDEF = 0.767*DUM7385O*TIME
+ 143.395*DUM7385E - 21.363*DUM78
+ 0.697*DDM2 + 1.216*DDG2
Changes in military spending appear to be twice as
important as those in the money supply in determin-
ing the rate of inflation. The deflator derived from the
inflation rate is used to determine the nominal value
of private consumption.
Private disposable income in current shekels (ISDISP)
is arrived at by subtracting direct taxes (ISDT) and
those on domestic production (ISTDPFY), public-
sector business income (ISGBUSIN), and private-
sector loans to the government (ISNETCL) from
GNP (figure 4). Subsidies to business (ISBUSSUB),
other government transfers to the private sector (ISG-
TRAN), and personal remittances from abroad (ISF-
TRANP) are then added:
ISDISP = ISY - ISTDPFY - ISGBUSIN
- ISDT -ISNETCL + ISBUSSUB
+ ISTRAN + ISFTRANP
This is converted to real disposable income by using
the deflator for private consumption.
Investment and Capital Stock
Total Israeli investment in 1985 was $4.1 billion-
about 20 percent of GNP. Capital stock was valued at
almost $50 billion, giving an average capital-output
ratio of 2.4.
The model calculates total investment, in current
shekels, as the sum of private and public housing
investment (ISINHI and ISPUBH), private and pub-
lic nonresidential investment (ISINS and ISPUBI),
` DUM78 and DUM7385E are dummy variables for 1978 and
1985, respcctively. R' = 0.998; estimated with 1975-85 data.
investment in ships and planes (ISVSP), and changes
in stocks (ISINSTOK):
ISI =ISINHI +ISPUBH +ISINS +ISPUBI
+ ISVSP + ISINSTOK
The private housing and nonresidential investment
components are both computed in real terms (figure
5). Real private housing is a behavioral function of
past values of real personal transfers from abroad and
government transfers to the private sector:b
0.569*(ISFTRANP(-1) + ISFTRANP(-2)1 25X1
ISIDEF(-U ISIDEF(-2) J
+ 0.127*rISGTRAN(-1) + ISGTRAN(-2)1
ISIDEF(-1) ISIDEF(-2) J
Changes in foreign transfer payments are far more
important than those in government transfers in de-
termining the level of private housing investment.
The investment sector deflator (ISIDEF) is derived
from the sector rate of inflation (DDIDEF), a behav-
ioral function of the rate of growth of the money
supply and of total military consumption (figure 6):'
DDIDEF = 104.67*DUM7385E
+ 0.841 *DUM7385O*TIME + 0.685*DDM2
+ 1.369*DDG2
As with the rate of inflation for private consumption,
changes in military spending are twice as important as 25X1
changes in the money supply in determining inflation
in the investment sector. The investment deflator is
used to convert current private investment into real
investment.~~ 25X1
b DUM7385 is a dummy variable for all years after 1972. Rj =
0.930; estimated with 1967-85 data.
' RZ = 0.996; estimated with 1975-85 data
25X1
25X1
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Sanitized Copy Approved for Release 2011/05/20 :CIA-RDP88T00096R000800960002-8
Confidential
Figure 3
Israel: Private Consumption
DD1~2 DDG2
ISDISPR
DDIMPDEF
Figure 4
Israel: Disposable Income
ISY ISYR
~ ISGTRANt ISGTRANR
ISPC
ISBUSSUT}--i ISDISP~~
~ ISFTRAND
~ISFTRAN~
DDM2: Money Supply Rate of Growth
DDG2: Percent Chaa`e in Total Military Consumption
ISTDPPY: DemasUe Production Tasee ISDISPR-- ISPCDEF
TIME: lade= for Yesr
ISCBUSIN: Pubao-Sector Business Income
DDPCDEF: Private Coaaumptton Inflation Rate
ISYR: Real CNP
ISPTRANP: Personal Tramfen from Abroad
ISDISPR: Real Disposable Income
ISYD6r: GNP DaMlor
ISPTRAND: Penenal Tnnsfen lrom Abroad (Dollars)
DDIMPDEF: Inflation Rate is the Import Sector
ISN6TCL? Nat Compubeq bane
13XR: E: