 |
|
© 2000 John Petroff |
|
No rated * * * * *
|
Resize -A
+A
|
1)- Econometric models
Several universities (e.g. Wharton, University of Michigan,
Brookings Institute) and several financial institutions (e.g.
Chase Manhattan Bank) have constructed econometric models which
they use to simulate what will happen to the economy given a set
of assumptions and key variable estimates. The
model can be made of 6, 20, 50 or more equations explaining some
or all major economic aggregates such as consumption, investment,
gross domestic output, profits, tax revenue, disposable income,
saving, money supply, imports, and so on. For each endogenous
variable, assumptions are made about the specific variables that
best explain them, and regressions are used on past data to estimate
coefficients of the explanatory variables. For instance, consumption
Ct in year t can be assumed to be explained by previous
year consumption Ct-1, and by current year disposable
income Yt . OLS regression explained in Chapter
5 Section E, can produce coefficients such as
Ct = a + 0.75 Ct-1 + 0.25 Yt
The key variables which are not
explained by the system of equations, i.e. exogenous variables,
and for which estimates are entered manually, are, for instance,
current reserve ratio of banks, recent average of Fed discount
rate, volume of exports, total labor force, government purchases
for the coming year as stated in the federal budget, and so on.
They reflect the latest conditions. In some models, among these
exogenous variables, attitude variables are present. One will
recall from Chapter 14 Section
E that these are compiled from surveys conducted by the Conference
Board and by the University of Michigan. The entire system of
simultaneous equations is run one or several periods beyond the
sample period to produce forecasts for next year or several years.
Econometric models have the advantage of allowing making not
just one forecast but a number of predictions by varying assumptions
and exogenous variables. Each system must however maintain an
internal logic and must not have conflicting assumptions. Repeated
simulations can produce a range of values from which a most likely
outcome can be calculated. In the end
analysis, the quality of the approach depends primarily on how
well the assumption reflect the forthcoming reality.
|
The following is an
example of econometric system known as the Revised Klein-Goldberger
Model as listed by Michael
Evans, which is considered as a relatively small model with
only 20 equations and 34 variables. For the sake of the standard
deviations of the estimated equations are omitted abut can be
found on page 498 in Michael
Evans "Macroeconomic Activity".
The estimated equations are
- Cd - 0.7 Cd-1 = 0.230 (Y - 0.7Y-1)
- 0.105 Cd-1 - 4.51
-
- Cns = 0.228 Y + 0.752 Cns-1 - 1.468
-
- Ih = 0.0517 Y - 0.042is-1 + 0.33Ih-1
- 1.853
-
- Ii = -.137(X-dIi) + 0.396Ii-1
- 24.702
-
- Fi = 0.0284X - 10.14(pi - p) + 0.463Fi-1
- 0.942
-
- (X - (Wg/p) )- 0.95(X-Wg/p))-1
= 0.364(Ip + Ih) +3.532((Nw
- Ng + Ns) -
- 0.95(Nw - Ng + Ns)) + 1.335(h
- 0.95h-1) -6.483
-
- h = -0.450dw - 1.996(NL - Nw -Ns)
+ 1.157
-
- ((W - Wg)/p) = 0.413(X - (Wg/p)) +
0.282((W-Wg)/p)-1 - 10.607
-
- dw = -1.697(NL - Nw - Ns)
+ 1.116(dp)-1 + 0.184
-
- iL = 0.157is + 0.835(iL)-1
+0.335
-
- RE = 0.788(Pch - Tc ) -0.667(Pch
- Tc -RE)-1 - 0.148
-
- PB = 0.0107pX + 0.89(PB)-1 + 0.674
-
- RI = 0.0623 p(Ip + Ih) - 0.0230diL
+ 0.938(RI)-1 + 0.394
-
- Ip - 0.95Ip-1 = 0.0656(X-Wg)-1
- 2.11(iL)-1 - 0.590Ip-1 + 9.329
-
- D = 0.0492(p0(Ip + Ih)0
+ ... + p20(Ip + Ih)20)
+ 0.0856Du+1.411
-
- is = 1.145id - 0.815RR-1
+ 0.533 Du - 0.511
-
- X = Cd + Cm + Ip + Ih
+ dIi + G + Fe - Fi
-
- pY = pX - D - Ti - RE - Tc - T
-
- Pch = pX - D - Ti - W - RI - PB
-
- W = whNw
-
- Definitions of variables
-
- Cd = consumption of durables, billions of 1954
dollars
- Cns = consumption of nondurables and services,
billions of 1954 dollars
- D = capital consumption allowances (depreciation), billions
of 1954 dollars
- Du = dummy variable: 0 for 1929-1946; 1 for 1947-1962
- Fe = exports, billions of 1954 dollars, billions
of 1954 dollars
- Fi = imports, billions of 1954 dollars, billions
of 1954 dollars
- G = government purchases of goods and services, billions
of 1954 dollars
- h = index of hours worked per week, 1954 = 100
- id = average discount rate at all Federal Reserve
Banks, percent
- Ih = residential construction, billions of 1954
dollars
- Ii = stock of inventories, billions of 1954 dollars
- iL = average yield on corporate bonds (Moody's),
percent
- Ip = investment in plant and equipment, billions
of 1954 dollars
- is = yield on prime commercial paper, four to
six months, percent
- Ng = government employees, millions
- NL = total labor force, millions
- Ns = self-employed workers, millions
- p = implicit GNP deflator, 1954 = 100
- PB = proprietor's income, billions of current dollars
- Pch = corporate profits including inventory valuation
adjustment, billions of current dollars
- pi = implicit price deflator for imports , 1954
=100
- RE = retained earnings including inventory valuation adjustment,
billions of current dollars
- RI = rental and net interest income, billions of current
dollars
- RR = year-end ratio of member banks' excess required reserves
- T = personal taxes + contributions for social insurance -
government and business transfer payments - interest on government
debt, billions of current dollars
- Tc = corporate profits taxes, billions of current
dollars
- Ti = reconciling item between net national product
and national income, billions of current dollars
- W = wages and salaries and supplements, billions of current
dollars
- w= annual wage rate of all employees, thousands of dollars
per year
- Wg = wage bill of government employees, billions
of current dollars
- X = GNP, billions of 1954 dollars
- Y = personal disposable income, billions of 1954 dollars
|
See review questions Q-15C1.1
through Q-15C1.6.
See research assignments R-15C1.1 and R-15C1.2.