Dixon 2001

Dynamic General Equilibrium Modelling for
Forecasting and Policy: A Practical Guide and

Documentation of MONASH
An Illustrative Application ofMONASH: the Australian Motor Vehicle Industry
from 1987 to 2016
Peter B. DixonMaureen T. Rimmer
Downloaded by UNIVERSITY OF EXETER At 05:59 28 October 2015 (PT)
Article information:
To cite this document: Peter B. DixonMaureen T. Rimmer. "An Illustrative
Application ofMONASH: the Australian Motor Vehicle Industry from 1987 to
2016" In Dynamic General Equilibrium Modelling for Forecasting and Policy:
A Practical Guide and Documentation of MONASH. Published online: 10 Mar
2015; 37-111.
Permanent link to this document:
http://dx.doi.org/10.1108/S0573-8555(2002)0000256004
Downloaded on: 28 October 2015, At: 05:59 (PT)
References: this document contains references to 0 other documents.
To copy this document: permissions@emeraldinsight.com
The fulltext of this document has been downloaded 91 times since NaN*
Users who downloaded this article also downloaded:
Peter B. Dixon, Maureen T. Rimmer, (2001),"Aims, Background, Innovations
and Presentation", Contributions to Economic Analysis, Vol. 256 pp. 1-36
http://dx.doi.org/10.1108/S0573-8555(2002)0000256003
Peter B. Dixon, Maureen T. Rimmer, (2001),"Extensions, progress and the
future", Contributions to Economic Analysis, Vol. 256 pp. 279-308 http://
dx.doi.org/10.1108/S0573-8555(2002)0000256008
Peter B. Dixon, Maureen T. Rimmer, (2001),"Mathematical Structure,
Solution Algorithm and Computer Implementation: an Overview",
Contributions to Economic Analysis, Vol. 256 pp. 113-136 http://
dx.doi.org/10.1108/S0573-8555(2002)0000256005
Access to this document was granted through an Emerald subscription

provided by emerald-srm:162036 []
For Authors
If you would like to write for this, or any other Emerald publication,
then please use our Emerald for Authors service information about how
to choose which publication to write for and submission guidelines are
available for all. Please visit www.emeraldinsight.com/authors for more
information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit
of society. The company manages a portfolio of more than 290 journals
and over 2,350 books and book series volumes, as well as providing an
extensive range of online products and additional customer resources and
services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization
is a partner of the Committee on Publication Ethics (COPE) and also works

with Portico and the LOCKSS initiative for digital archive preservation.
*Related content and download information

correct at time of download.
CHAPTER2
An Illustrative Application ofMONASH:

the Australian Motor Vehicle Industry
from 1987 to 2016
4. Introduction
MONASH is a framework for:

• estimating changes in tastes and technology and for generating up-to-date
input-output tables (historical simulations);
• explaining periods of economic history in terms of driving factors such as
policy changes, changes in world commodity prices and changes in tastes and
technology (decomposition simulations);
• generating forecasts for industrial, occupational and regional variables using
detailed extrapolations of trends in tastes and technology together with a wide
variety of projections from organizations specializing in macro, export, tourism
and policy forecasting (forecast simulations); and
• calculating the deviations from explicit forecast paths for macro and micro
variables which would be caused by the implementation of proposed policy
changes (policy simulations).
AH four components of the framework are illustrated in sections 5 to 7 where we
analyse the Australian motor vehicle industry for the period 1987 to 2016. We use
an historical simulation to estimate changes in tastes and technology for 1987 to
1994 with special emphasis on motor vehicles. We use a decomposition simulation
for the same period to assess the significance of changes in tariffs relative to
changes in other variables (e.g. tastes and technology) as determinants of the
performance of the motor vehicle industry. We use a forecast simulation to project
output and employment for the motor vehicle industry to 2016 in the absence of
further reductions in tariffs beyond those planned to 2001 and we use a policy
simulation to work out the deviations from the forecast paths for macro and
industry variables which would be caused by proposed further reductions in motor
vehicle tariffs.
Tariff policy for motor vehicles was a major political issue in Australia in 1996
and 1997. MONASH results of the type presented here and results from two rival
models, SCM and AE-CGE, were the subject of newspaper reports and
37
38 Dynamic General Equilibrium Modellingfor Forecasting atA Policy
Parliamentary debates for many months.1 However, we do not require readers to be

interested in the intricacies of Australian tariff policy. We choose this application
because it is a good illustration of how dynamic CGE modelling can play a key role
in detailed policy making. We hope that the early presentation of such an
illustration will motivate readers to work through the technical material in later
chapters.
To help readers obtain an early impression of the role of various MONASH
technicalities, this chapter contains frequent forward references to material in
Chapters 4 and 5. In reading the present chapter there is no need to follow these
forward references in any detail.
The chapter is organized as follows. Section 5 describes the historical and
decomposition simulations. Sections 6 and 7 describe the forecast and policy

simulations. Section 8 contains concluding remarks concerned with: (a) the policy
implications of our results; and (b) the use of back-of-the-envelope calculations in
analysis of results.
5. Historical and decomposition simulations: the Australian motor vehicle

industry from 1987 to 1994
In this section we use MONASH historical and decomposition simulations in a

description of developments in the Australian economy over the period 1987 to
1994 with special reference to the motor vehicle industry. Subsection 5.1 contains
the motor vehicle input data to the MONASH simulations. Results and conclusions
are in subsections 5.2 and 5.3.
5.1. Information for the period 1987 to 1994

As explained in subsections 16.1(j) and 23, MONASH can absorb economic data in
different industry/commodity classifications and at different levels of
disaggregation. The data absorbed in our historical simulation for 1987 to 1994 of
direct relevance to the motor vehicle industry are listed in Table 5.1 and in the 1987
columns of Tables 5.2 and 5.3.2 Some of these data refer to the Motor vehicle
industry (MONASH industry 68 and MONASH commodity 70) in isolation and
some refer to a broader industry of which motor vehicles is a part. Our data on
investment, for example, refer to the transport equipment sector (industries 68-71),
i.e. motor vehicles, ships, trains and planes.
1
Successive sets of MONASH results are in Industry Commission (1996, 1997) and in Dixon et al.
(1997a). The SCM results are in Murphy (1997) and the AE-CGE results are in Access Economics
(1997).
The 1994 columns of Tables 5.2 and 5.3 are outcomes of the historical simulation.
An illustrative application ofMONASH
Table 5.1. Growth in Motor Vehicle related variables from

1987 to 1994: Shocks used in the Historical Simulation^
Variable Shock used Source Note
in
MONASH
Output of motor 14.50 Estimate supplied by
vehicles the Industry
Commission taking
account not only of
the increase in die
quantity of output but
also its quality.
Tariff on motor -30.28 Unpublished data The impact effect is a
vehicles from the Industry reduction in landed-duty

Commission. -paid price of motor
vehicle imports of 6.57
per cent.
Exports of motor 50.97 Published ABS data This is less growth than
vehicles (volume) (Cat 5215.0) and for total exports (63 per
unpublished ABS data cent).
atthe5-digitSITC
level mapped to input-
output commodities
using an unpublished
ABS concordance.
Imports of motor 64.01 As above. Total import growth was
vehicles (volume) 59 per cent.
Foreign currency 30.87 Unpublished ABS Exchange rate fell by
export price of motor data on merchandise 4.29 per cent. Therefore
vehicles export price deflators in domestic currency the
by 4-digit IOIC group. f.o.b. export price rose
by 36.48 per cent. The
GDP deflator increased
by 34.42 per cent.
Foreign currency 24.20 Unpublished ABS In domestic currency the
import price of motor data on merchandise c.i.f. import price rose
vehicles import price deflators by 29.53 per cent.
by 4-digit IOIC group.
Investment by the -26.24 ABS published data in Investment growth is
transport equipment Cats 5221.0 and allocated to industries in
sector (includes Motor 5626.0 on private this sector according to
vehicles, Ships, Trains gross fixed capital their relative rates of
and Aircraft) expenditure. growth of capital. This
gives the percentage
change in investment by
motor vehicles as
-28.49.
.... continued
40 Dynamic General Equilibrium Modelling for Forecasting and Policy
Table 5.1 continued
Variable Shock used Source Note

in
MONASH
Capital stock of the 0.00 Estimate supplied by
motor vehicles the Industry
industry Commission.
Labour input to the -17.50 Estimate supplied by
motor vehicles the Industry
industry Commission.
Household NA(22)-36.05 Unpublished National Most of motor vehicle
consumption in NA(24)« 36.04 Accounts consumption is in NA
National Accounts consumption data for commodity 22 and

(NA) categories. The 38 commodities NA(22) is almost
bulk of motor vehicle supplemented by entirely motor vehicles.
expenditures are in unpublished Despite this, MONASH
categories 22 concordance matrix implies motor vehicle
(Purchases of motor between these 38 consumption growth of
vehicles) and 24 commodities and 115 64.07 per cent. This
(Operation of motor MONASH reflects slow growth of
vehicles nee.) commodities. mechanical repair
consumption in NA(24).
* Our data and simulation results refer to financial years, that is years ending on June 30.
f
ABS catxxx refers to catalogue xxx published by the Australian Bureau of Statistics.
Investment growth in this sector was -26.24 per cent for the period 1987 to 1994.
In our MONASH historical simulation, we impose this sectoral investment growth,
i.e., we assume that
2:?J68S(i)y(i) =-26.24 (5.1)
where
S(i) is the share of industry i in the investment of the transport-equipment
sector; and
y(i) is the growth in investment in industry i from 1987 to 1994.
Then we allocate the individual y(i)s by assuming a common movement in the
investment/capital ratios of all industries in the sector. The result for motor vehicles
is a decline in investment of 28.49 per cent, i.e y(68) = -28.49.
So that we can make use of sectoral data and associated allocation rules, we
include in MONASH many definitions of sectoral aggregates and of ratios of
industry variables, see for example (16.52) which defines sectoral employment.
5.2. The historical simulation

In the historical simulation, we exogenized all of the observed variables and
shocked them with their observed movements. Thus the results are consistent with
An illustrative application ofMONASH 41
all our statistical information. For example, we set investment growth in the
transport sector exogenously at -26.24 per cent, growth in motor vehicle output at
14.50 per cent and growth in motor vehicle exports at 50.97 per cent. To allow
MONASH to hit these and many other targets, we endogenized variables concerned
with: primary-factor-saving technical change; intermediate-input-saving technical
change; preferences for imported goods relative to domestic goods; household
tastes (the form of the utility function); and rates of return on industrial capital. In
the case of motor vehicles, we found for the period 1987 to 1994:
• that total-factor productivity growth was approximately zero. The numbers in
Table 5.1 imply strong primary-factor-saving technical change in the motor
vehicle industry. Using the cost information in Table 5.3, we see that the
primary factor input (the share-weighted average of the percentage changes in

capital and labour input) to the industry declined over the period 1987 to 1994
by about 15 per cent. With an output increase of 14.5 per cent this gives
primary-factor productivity growth of about 35 per cent [consistent with the
MONASH result for the variable APF(68) appearing in (16.6) and (16.7),
discussed in subsection 16.1(a)]. On the other hand, the industry suffered a
considerable reduction in intermediate-input productivity. Our calculations
imply that intermediate inputs to the industry increased by 23.38 per cent
giving a decline in intermediate-input productivity of 7 per cent ( =
100x(l.145/1.2338 - 1)). Taking account of the shares of primary factors and
intermediate inputs in motor vehicle costs (about 23 and 77 per cent, see Table
5.3), a back-of-the-envelope estimate is that total-factor productivity growth in
the industry was close to zero (=100x{1.145/[0.23x0.85 + 0.77x1.2338] - 1}).
This was borne out in our historical simulation which implied that the industry
in 1994 could produce any given level of output with 0.41 per cent less of all
inputs (primary and intermediate) than were required in 1987.
• that there was a strong twist in the industry's technology towards the use of
capital. The result for the MONASH variable TWLK(68) [(16.6) and (16.7)
discussed in subsections 16.1(a) and 17.5] was -17 per cent. This means that
the industry's technology changed so that at any given ratio of the wage rate to
the rental rate on capital, the industry would choose a capital/labour ratio 17
per cent higher in 1994 than in 1987.
• that there was a shift in consumer preferences towards the purchase of motor
vehicles. The result for the MONASH variable A3(70) [(16.13) discussed in
subsection 16.1(a)] was 28.80 per cent. This means that at any given set of
prices and per capita income, consumption per household of motor vehicles
would be about 28.80 per cent3 higher in 1994 than in 1987.
3
More precisely, the consumption per household of motor vehicles in 1994 would be 28.80*(1 - S70)
per cent higher than in 1987 where S70 is the share of motor vehicles in household expenditure.
42 Dynamic General Equilibrium Modellingfor Forecasting and Policy
Table 5.2. Sales Shares for the Australian

Motor Vehicle Industry
1987 1994
Intermediate input sales to:
Motor vehicles and parts 0.22 0.21
Mechanical repairs 0.05 0.05
Defence 0.03 0.04
Other 0.04 0.06
Investment 0.37 0.27
Households 0.22 0.28
Exports 0.07 0.09
Total 1.00 1.00

Import share in domestic market 0.38 0.46
• that there was a shift across industries towards the use of motor vehicles as an
intermediate input and as a capital good. Via results for variables such as
Al(68) [appearing in (16.5), discussed in subsection 16.1(a)], MONASH
indicated that motor vehicle input per unit of output and per unit of capital
creation averaged over all industries was 18.3 per cent higher in 1994 than in
1987.
• that there was a shift in the preferences of users of cars towards the imported
product. The result for the MONASH variable ftwist(70) [appearing in (23.3)
and discussed in subsection 23.1] was 28.20 per cent. If ftwist(70) is x per
cent, then at any given ratio of import/domestic purchasers' prices and level of
demand pressure4, users of motor vehicles increase the import/domestic ratio in
their demands for motor vehicles by x per cent.
5.3. The decomposition simulation

Having completed the historical simulation, we then adopted the decomposition
closure in which technology and taste variables [APF(j). Al(j), A3(i), TWLKfl) and
ftwist(j)] are exogenous. By setting these variables at their values estimated from
the historical simulation, we obtain results in the decomposition simulation for
output, employment and other endogenous variables identical to those in the
historical simulation. However, with technology and tastes exogenous in the
decomposition simulation we can answer questions about the effects of changes in
these variables. More generally, we can decompose history into the parts
The idea of demand pressure is discussed in connection with (5.13) and (23.3).
Table 5.3. Cost Shares for the Australian

Motor Vehicle Industry
1987 1994
Commodity inputs
Motor vehicles and parts 0.42 0.51
Rubber 0.06 0.05
Iron and steel 0.04 0.04
Other 0.25 0.21
Labour 0.20 0.16
Returns to capital 0.02 0.02
Other costs 0.01 0.01

Total 1.00 1.00
attributable to changes in variables such as those identified in the column headings

of Tables 5.4 and 5.5.5
The decomposition results in these tables are analysed in subsection 5.3(c).
Necessary background for understanding them is in subsections 5.3(a) and (b).
Subsection 5.3(a) describes our modelling in decomposition simulations of
accumulation relationships and momentum effects. Subsection 5.3(b) sets out the
macroeconomic assumptions underlying decomposition simulations.
(a) Momentum and accumulation in MONASH decomposition simulations
Data collection for the two end points (1987 and 1994) of our study was a
formidable task. It was not feasible to put together data sets for intermediate years.
Thus our analysis is necessarily comparative static. We compare different versions
of the 1994 economy: with and without the technological changes of the period
1987 to 1994; with and without the taste changes of this period, etc.
A fundamental assumption of comparative static simulations is that solutions for
year t can be computed from a set of equations linking variables only for year t.
This can present problems in the treatment of accumulation relationships such as
that connecting the stock of net foreign liabilities (NFL) with annual flows of
domestic saving, investment and interest Because movements in NFL affect
national income, they also affect private and public consumption expenditure. This
in turn affects the real exchange rate and import-competing industries such as
motor vehicles. Thus, for analysing the effects on the performance of the motor
vehicle industry of changes in technology and of other potential explanators, it is
important to model the links between these explanators and NFL.
3
Earlier CGE applications which have attempted to explain historical episodes include Dervis et al.
(1982, ch. 10), Parmenter et al. (1994), Kehoe et al. (1995) and Gehlhar (1997). Relative to these earlier
attempts, we used data on many more variables and explain a broader range of events.
To facilitate the handling of accumulation relationships in a comparative static

framework, we formulate MONASH in decomposition simulations as:
F(X,Q) = 0 (5.2)
where
F is an /w-vector of differentiable functions;
X is an w-vector of variables for the year of interest, with n > m\ and
Q is a vector of initial conditions, treated as parameters.
In our decomposition simulation for 1987 to 1994, X includes prices and quantities
applying in 1994, and the m equations in (5.2) impose the usual CGE conditions
such as: demands equal supplies; demands and supplies reflect utility and profit
maximising behaviour; prices equal unit costs; and end-of-year capital stocks equal
depreciated opening capital stocks plus investment. The vector Q includes 1987
values of net foreign liabilities and borrowing.
Equation (5.2) does not include values of variables for years between 1987 and
1994. In our computations for 1994, we handle accumulation relationships by
making smooth growth assumptions.6 Via these assumptions, stock and flow values
for 1987 enter our model as parameters (part of the Q vector) and those for 1994
enter as variables (part of the X vector), but stock and flow values for intermediate
years are substituted out. For example, consider an accumulation relationship for
NFL of the form
N F W , = NFL,+S(1+R) + K,^, - Kt+S(l-D) - APS* GNP^ (5.3)
where
NFLj is the accumulated value of NFL at the beginning of year i;
R is the interest or dividend rate applying to net foreign liabilities;
Ki is the quantity of capital at the beginning of year i (for simplicity we assume
here, but not in MONASH, that there is only one type of capital and its price is
1 in all years);
D is the depreciation rate;
APS is the average propensity to save out of gross national product (treated
here, but not in MONASH, as a parameter); and
GNPj is gross national product in year i.
In (5.3), net foreign liabilities at the beginning of year t+s+1 equal net foreign
liabilities at the beginning of the previous year inflated by interest and dividend
6
There is a long tradition of using these assumptions to introduce a dynamic element to single period
general equilibrium simulations, see for example Marine (1966), Johansen (1974), Evans (1972), and
Dixon and Butlin (1976).
An illustrative application o/MONASH 45
payments plus investment expenditure during year t+s minus domestic saving.
From (5.3) we obtain
NFLt +T =NFL t *(1+R) T + I [Kt+s+1 -K t + s (l-D)-APS*GNP t + s ](l+R) T - 1 - s .

s=0
(5.4)
Under smooth growth assumptions,7
£ £
K t + s = K t * f * ^ and GNP t+s =GNP t * f ^ t l > . (5.5)
I Kt J \ ON*t ;
Now thinking oft as 1987 and t+T as 1994, we rewrite (5.4) for potential inclusion
in (5.2) as
N F L t + T = N F L t * ( l + R)T
sy J \ £
x_l f K t + x V (v \~ fGNPt+^t
+ I K, - — - ^ T - ( l - D ) -APS*GNP t — — — (1 + R) T -K
s=0 [ K t J ^ Kt J ^ GNPt J
(5.6)
Equation (5.6) is a relationship between the 1994 values of NFL, K and GNP
(components of the X vector) and it includes the 1987 values of NFL, K and GNP
(components of the Q vector).
As mentioned in subsection 2.2, in computing solutions for MONASH we always
start from an initial solution (Xinili,i) of (5.2). Then we compute the movements in m
variables (the endogenous variables) away from their values in the initial solution
caused by movements in the remaining n-m variables (the exogenous variables)
away from their values in the initial solution. In the simulations in this section, the
initial solution is derived mainly from input-output (10) data for 1987. The
movements we impose on the n - m exogenous variables are those occurring from
1987 to 1994. With the initial solution at 1987, our computation then generates
movements in the m endogenous variables which can be interpreted as the effects
between 1987 and 1994 of movements in the exogenous variables.
However, there is one difficulty: 1987 values do not satisfy accumulation
relationships such as (5.6). With the variables NFL94, K94 and GNP94 set at the
NFL, K and GNP values of 1987, we have
' We thank Michael Kohlhass for reminding us that smooth growth assumptions should be applied only
to variables such as capital and GNP that move quite smoothly. They should not be applied directly to
variables such as the balance of trade (investment - savings) that can be either side of zero.
LHS(5.6) = NFL87 (5.7)

and
RHS(5.6) = NFL87*(1 + R) 7 + [K 87 D-APS*GNP 87 ]* I ( 1 + R) 6 _ S . (5.8)
s=0
We overcome this difficulty by replacing (5.6) with:
NFL94 =
NFL 87 + NFL 87 *[(l + R ) 7 - l ] + [Kg 7 D-APS*GNP 87 ]* I 0 + R)6"5 *U
L s=0
. .±r i ^
6 ,6 s I K-94 |7 fv*A^
7
+ S(l + R) - *K 8 7 -î
K
|21| -0-D)
s=o (^ 87j \,K87;
f - T
-K87D-APS*GNP87 f ^ 1 ! 7 -1 (5.9)
^GNPg7 J
V J.
where U is included as one of the n variables in the X vector in (5.2).8 In our initial
solution for (52), U is zero. This allows (5.9) to be satisfied with NFL94, K94 and
GNP94 set at the NFL, K and GNP values of 1987. In our final solution for (5.2),
the solution representing the situation in 1994, we set U at one. This imposes the
original relationship (5.6). Thus, in deriving a final solution for (5.2), we include
the movement in U from 0 to 1 among the movements in the n - m exogenous
variables. As can be seen from (5.9), the movement in U ensures that our final
solution for (5.2) implies a change in NFL between 1987 and 1994 which reflects
not only the variations in K and GNP between these two years but also interest and
dividends associated with NFL87 and increments to NFL which would have
occurred if K and GNP had remained at their 1987 values.
In Tables 5.4 and 5.5, we report the effects of the movement in U from 0 to 1 as
"momentum". These are effects that would have occurred in the absence of shifts in
foreign demands and supplies, in the absence of changes in protection, etc. The
nature of momentum effects will be explained more fully in the next subsection.
Here we to note that MONASH recognizes that shifts in foreign demands and
supplies, changes in protection, etc. are only part of the reason for changes in the
Australian economy between 1987 and 1994. Via momentum effects, MONASH
o
0
U is known as a homotopy variable, see for example, Zangwill and Garcia (1981).
recognizes that except in the unlikely situation of a steady-state equilibrium, stock

and flow variables (e.g. net foreign liabilities and net foreign borrowing) must
change even if foreign demands and supplies, protection, etc. are unchanged.
(b) Macroeconomic assumptions in the decomposition simulation
For understanding the results in Tables 5.4 and 5.5, it is useful to work through
Figure 5.1. This is a flow diagram for a one-commodity CGE model. Consequently
it does not illustrate relative-price or other structural effects. These are important in
the results, nevertheless, Figure 5.1 is a helpful representation of the main macro
assumptions underlying our MONASH decomposition simulation.
Exogenous variables in the decomposition closure are represented in Figure 5.1
by rectangles while endogenous variables are shown in ovals. The change in

aggregate employment between 1987 and 1994 (AL), for example, is exogenous.
Thus we assume that changes in technology (ATECH) and changes in other
exogenous variables between 1987 and 1994 did not effect aggregate employment
in 1994. As is conventional in macro modelling, we assume that employment
effects are eliminated over the medium term by adjustments in wage rates.
Lines (1), (2) and (3) in Figure 5.1 impose a production function: the change in
output (AGDP) between 1987 and 1994 is a function of ATECH, AL and the change
in start-of-year capital (AK) 9 .
We assume that capital earns the value of its marginal product, that is, MPK is the
ratio of the rental price of capital to the price of the product. In a one-commodity
model, the product price can represent the asset price of capital. Consequently,
MPK is the ratio of the rental price of capital to the asset price. We represent the
rental/asset price ratio as the rate of return (ROR). Under constant-returns-to-scale
(assumed in MONASH), MPK is a function of K/L and TECH. Thus, AK is
determined by AL, ATECH and AROR [lines (4), (5) and (6)].
As indicated in the figure, in our decomposition simulation AROR is exogenous.
When we are concerned with analyzing the effects of particular shocks over periods
as long as seven years (1987 to 1994) it is conventional to assume that capital
adjusts to restore rates of return. For example, in isolating the effects of technology
changes between 1987 and 1994, we assume that rates of return are unaffected, i.e.
AROR=0.10
y
We assume that GDP in any year is a function of technology available during the year, labour used
during the year and capital available at the start of the year. In the present context AK is the change in the
capital stock between the start of 1987 and the start of 1994.
10
In practice we have found it useful to damp capital responses in individual industries by introducing a
positive relationship between capital growth and required rates of return. Thus, in MONASH
decomposition simulations it is only the average rate of return across industries that is treated
exogenously. Damping of capital responses is discussed further in subsection 5.3(c), column 8 and in
subsection 16.1(i).
48 Dynamic General Equilibrium Modelling/or Forecasting and Policy
With capital earning the value of its marginal product, labour also is paid
according to the value of its marginal product. Thus, via the factor-price frontier
(the relationship between the MPK, MPL and TECH, Samuelson, 1962), AROR
and ATECH determine the real wage rate. This is indicated in Figure 5.1 by lines
(16) and (17).
Exogenization of the rate of return can be thought of as tying down capital stocks
in 1994. While tying down capital stocks for 1994 ties down aggregate investment
between 1987 and 1994, it does not determine investment in 1994. We link
investment in 1994 to capital in 1994 [line (7) in Figure 5.1]. In isolating the effects
of changes in technology etc., we assume that such changes have no impact on
business confidence. Thus we treat the investment/capital ratio (a reflection of
business confidence) as exogenously determined. This can be done for each
industry by exogenizing the variables IKRATIO(j) defined in (16.48).
Lines (8) and (9) allow for the calculation of the change in gross national product
between 1987 and 1994 (AGNP). This is AGDP less the change in interest/dividend
payments to foreigners (a proportion of the change in start-of-year net foreign

liabilities, ANFL).11
We assume in line (10) that the change in public and private consumption (AC) is
an exogenously given proportion of the change in GNP. [This assumption is
imposed in MONASH by exogenizing the shift variable, APC, in (16.65).] With
AGDP, AC and AI now determined, the change in the balance of trade (ABOT) falls
out as a residual.
Line (11) links accumulated excess savings (S87/94) to AGNP. S87/94 is the
difference between the value of accumulated saving over the period 1987 to 1994
and the value it would have had in the absence of any change over this period in
GNP. In deriving the link between S87/94 and AGNP, we assume that saving in each
year between 1987 and 1994 is a fixed proportion of GNP. Under the smooth
growth assumption applied to GNP, this allows us to specify accumulated excess
saving as a function of AGNP.
By again invoking the smooth growth assumption, we can specify in our model
the excess accumulated cost of investment (Ig7/94) between the beginning of 1987
and the beginning of 1994 in terms of the change between these two dates in the
capital stock (AK). The excess cost of investment is the difference in the value of
accumulated investment and the value it would have had in the absence of any
change over the period in the aggregate capital stock. The link between I87/94 and
AK is indicated in Figure 5.1 by line (12).
The final set of relations in Figure 5.1 are lines (13), (14) and (15). They
determine ANFL as a combination of three components: I87/94 minus S87/94 plus
momentum. Momentum is the change in NFL which would have occurred in the
absence of either excess accumulated savings or excess accumulated investment,
that is the change in NFL that would have occurred in the absence of changes in
GNP and K. Momentum consists of accumulated interest payments between 1987
and 1994 on the net foreign liabilities of 1987 plus depreciation investment (that is
the investment required to maintain the capital stock at its 1987 level) minus static
saving (that is the accumulated value of saving that would have occurred in the
absence of any change in GNP). As explained in connection with (5.9), momentum
is recognized in MONASH simulations via movements from zero to one in the
homotopy variable, U. The connections (13), (14) and (15) together with (11) and
(12) are embedded in MONASH equations such as (16.63) and are discussed in
subsections 5.3(d), 16.1(m) and 22.
(c) Decomposition simulation: results
The last columns of Table 5.4 and 5.5 show outcomes for macro variables and
11
GNP = GDP - ROI*NFL where ROI is the rate of interest or dividend on net foreign liabilities.
Holding ROI constant, we obtain: AGNP = AGDP - ROI*ANFL.
50 Dynamic General Equilibrium Modelling for Forecasting and Policy
industry outputs for Australia between 1987 and 1994. Columns 1 through 10
provide a decomposition of these outcomes computed according to (2.12) with the
closure illustrated in Figure 5.1 and described in detail in section 30.
The outstanding feature of the period 1987 to 1994 was rapid growth in trade
relative to GDP. Whereas the increase in real GDP was 19.6 per cent, export and
import volumes increased by 63.4 and 58.2 per cent Dixon et al (2000) used a
table similar to Table 5.4 to explain this rapid growth in trade. Here our emphasis
will be on the motor vehicle industry. Nevertheless all prominent features of the
tables will be discussed.
In analysing the decomposition results, we start by looking at each of the columns
1 to 10 individually. Then we make a comparison between columns.
Column 1: momentum
Column 1 in Tables 5.4 and 5.5 shows momentum effects, generated by imposing
AU equals one (see Figure 5.1). The column shows what would have happened to
the Australian economy over the period 1987 to 1994 if there had been no changes
in other exogenous variables, that is no changes in the variables identified in the
headings to columns 2 through 10.
With AL, ATECH and AROR fixed at zero in the momentum column, Figure 5.1
indicates that there should be no change in K, I and GDP. That there are changes in
these variables reflects structural effects which cannot be captured in a one-
commodity representation of MONASH. These structural effects will be discussed
shortly.
Consistent with Figure 5.1, column 1 of Table 5.4 shows a large effect for NFL,
with significant consequences for GNP and C (see rows 2, 10 and ll) 1 2 . With
almost no growth in K between 1987 and 1994 (1.4 per cent over seven years, row
6), investment expenditures over the period would have covered little more than
depreciation. In these circumstances, Australia's saving would have comfortably
outstripped investment expenditures leading to a decline in NFL relative to GDP
(61.0 per cent, row 2, column 1). This would have allowed an increase in GNP and
C (3.6 and 3.7 per cent, rows 10 and 11, column 1) even in the absence of any
improvements in TECH or growth in L, K and GDP.
12
Readers may be surprised at the slight discrepancy in column (1) between the results for GNP and C
(rows 10 and 11). As explained in the footnote 12, ch. 1, in decomposition simulations we use a mid-
point database. Thus MONASH generates percentage change results of the form 200(Z F - Z,y(ZF + Zi)
where ZF and Zt are the final and initial values of variable Z. In Tables 5.4 and 5.5 we have scaled each
row to convert these mid-point percentage changes into the more familiar initial-point form, 100(ZF -
Z|VZ|. In columns (1) to (9) the mid-point percentage change in C was assumed to be the same as the
mid-point percentage change in GNP. Because there were differences [recognized in column (10)] in the
growth of GNP and C, identical mid-point percentage changes in columns (1) to (9) translated to slightly
different initial-point percentage changes.
Table 5.4. Macroeconomic and Trade Variables: Decomposition of Changes from 1987 to 1994
Column 2 i 4 5 1 6 1 7 8 9 "16 11
,. ' Chan ECS attributable to:
mom en- shifts in changes technical changes in changes growth apparent shifts in other Totartf)
tumW foreign in change import/ in in changes in export factors
demands protect- domestic household employ- required supply
& import ion preferences tastes ment rates of curves
prices return
1.Change in balance of trade, -Z2 0.8 0.1 -0.9 -0.1 0.0 1.2 0.2 -0.1 2.2 1.2
% of "87 GDP
2.Change in net foreign liabilities -61.0 0.5 5.4 63.2 3.1 6.6 43.6 -71.5 0.6 38.0 28.5
% of "87 GDP
Percentage changes
3.Real exchange rate(a' 6.5 28.9 -2.4 -3.3 •4.4 0.2 -8.6 4.5 2.0 -1.2 22.3
I
4.Nominal exchange rate (<J>) 6.3 32.3 -2.1 -3.3 -4.1 0.3 -8.1 5.1 2.5 0.3 29.2
S.Rcal wage rate, wage 2.1 10.8 1.7 2.9 -1.5 •02 -3.2 -8.5 2.1 -0.6 5.7
deflated by Pg
6.Capitai stock 1.4 5.3 1.0 14.2 0.2 U 9.2 -9.6 0.3 5.6 28.9
7 Real investment, investment
deflated by Pj
8.Employment
9.Real GDP, GDP deflated

byP g
lO.Real GNP, GNP deflated
1.2
0.0
0.5
3.6
4.5
0.0
2.0
5.0
1.0
0.0
0.4
-0.2
13.4
0.0
7.7
5.0
0.1
0.0
0.2
-0.3
1.1
0.0
0.0
-0.4
8.4
10.1
9.8
7J
-10.4
0.0
•2.6
0.0
0.3
0.0
0.1
-0.3
-14.4
0.0
1.5
-1.7
5.2
10.1
19.6
18.0
f
byP c
11.Real public and private 3.7 5.1 -0.2 5.2 -0J -0.4 7.5 0.0 -0.3 3.2 23.3
consumption
12 .Imports, volu me 2.3 14.7 2.8 29.4 8.7 0.4 8.7 -3.9 1.4 -6.3 58.2
13. Ex ports, volume -15.5 -5.2 5.2 35.4 12.4 0.7 23.3 -5.4 3.3 9.1 63.4
.... continued
Table 5.4 continued 53
Column 1 2 3 4 5 6 7 8 9 10 1—17—
Chances attnbutable to:
momen- shifts m changes technical changes in changes growth apparent shifts in other TotaKd)
tum^) foreign in change import' in in changes in export factors
I
prices return
14.Traditional exports, volume -11.4 -29.6 3.7 44.8 9.5 0.3 18.2 -7.4 -0.1 8.5 36.6
15.Non-traditk>nal exports. •26.1 29.5 9.4 17.9 20.6 1.0 35.7 -0.3 13.7 11.8 113.3
volume
16.Tourism, volume
n.Price deflator for GDP (Pg)
18.Price deflator for

-19.8
0.4
0.0
56.2
2.6
0.0
5.0
-0.3
0.0
17.1
0.0
0.0
12.5
-0.4
0.0
1.4
-0.1
0.0
27.8
•0.6
0.0
-8.4
-0.5
0.0
-3.9
-0.3
0.0
7.9
-1.4
0.0
95.9
-0.6
0.0
i
consumption (Pc) W
19.Price deflator for -1.3 -3.5 -0.6 5.7 0.6 -0.3 1.4 -3.5 0.0 -4.8 -6.2
investment (Pj)
20.Terms of trado 3.5 19.3 -I.I -7.6 -2.7 -0.1 -5.2 1.7 -2.2 -2.0 3.6
|
(a) This is the percentage change in 4>*Pg/Pf wheic 4> is die nominal exchange rate, P» is the GDP deflator and Pf is the foreign price level.
1
(b) The price deflator for public and private consumption is the numeraire. Results for other price deflators indicate movements relative to consumption prices.
(c) This reflects initial (1987) levels of net foreign liabilities, capital and savings, see description in subsection 5.3a.
(d) With the exceptions of nominal variables, the results in (his column are actual percentage changes between 1987 and 1994. The results for the nominal variables reflect our
convention that Pc is the numeraire, i.e. the percentage change in P c is 0. The actual percentage change was 35.0. Thus the actual percentage changes in the price variables (l/<t>,
Pg, Pi, and Pc) were 35.0 percentage points more than is shown here. The only other nominal variables are in tows 1 and 2. These are nominal trade magnitudes expressed as
percentages of a fixed number, 1987 GDP. As with the price variables, the results for these variables are sensitive to our convention that the percentage change in PG is 0. The
actual changes fn the balance of trade and net foreign liabilities expressed as percentages of 1987 GDP were 1.0 and 43.4 per cent The percentages that we explain in Table 5.4 (1.2
and 28.1) abstract from purely inflationary valuation effects which are o f no economic significance. I f inflation were to be included in the computations, then it would be
inappropriate to compare inflation-affected nominal trade variables in the numerator with a fixed value for the denominator 1
I
Table 5.5. Sectoral Outputs: Decomposition of Changes from 1987 to 1994
Column 1 2 3 4 5 6 7 8 9 10 II
Percent ace chances attributable to:
Sectoml output momen- shifts in changes technical changes in changes growth apparent shifts in other Total
tum(c> foreign in change import/ in in changes in export factors

prices return
1.Agriculture, forestry, fishing -3.9 -9.6 0.9 15.3 2.1 0.7 10.5 -7.4 2.8 1.1 12.5
2 Mining -8.4 -16.3 3.4 21.1 5.4 0.2 18.6 4.9 0.6 8.7 38.3
3.Food processing -2.7 -0.6 0.6 11.4 0.3 -2.2 11.8 -3.2 -I.I 1.3 15.7
4 Textiles, clothing, footwear -0.9 -6.2 -5.2 9.4 -11.9 -10.3 10.5 -1.9 -0.4 1.7 -15.1
5 Paper, printing -0.1 0.0 -0.1 4.3 -06 0.2 10.5 -1.4 -0.2 2.0 14.4
6 Chemicals, petroleum, coal products -2.7 -4.6 0.2 15.4 0.3 1.3 13.5 -3.1 0.0 3.3 23.6
7 Basic metal products -6.3 -14.2 1.9 34.5 -6.9 0.7 16 3 -11.0 5.8 3.5 24.4
I
S.Transport equipment -4.3 -1.9 -2.8 13.0 •4.6 2.2 14.9 -5.3 -I.I -1.9 8.4
8a motor vehicles •4.9 •4.1 -S.t 24.4 -4.0 4.4 16.7 -7.0 -l.t -19 14.5
9 Electronic equipment -2.0 1.8 -0.1 19 1 -3.9 0.2 12.4 -5.5 0.5 -4.7 17.9
10.Other manufacturing -0.8 0.4 -0.5 3.6 -4.8 -0.1 10.9 -4.6 -0.1 0.6 4.7
II.Electricity, gas, water 0.0 0.0 0.4 119 00 0.9 10.5 -3.4 0.3 2.7 23.2
12 Construction 1.7 4.5 0.7 15.0 0.0 1.1 8.3 -8.3 0.2 -12.0 11.2
1
13. Whole sale trade -t.O 2.4 0.8 6.3 0.3 -1.2 10.7 -3.4 0.8 -0.2 15.4
14 Retail trade, repairs 2.5 4.4 0.2 0.3 -0.6 -2.6 8.0 0.0 -0.1 2.0 14.3
15Transport, storage, communications -1.2 1.7 0.9 25.9 1.8 -0.3 11.9 -2.3 -0.3 2.8 40.9
16 Finance, insurance I.S 2.2 •0.1 69.0 2.5 1.4 12.3 -0.4 -0.5 0.3 88.3
I7.0ther business services 0.0 2.6 0.5 12.2 0.1 -0.1 10.6 -1.6 0.0 1.3 25.7
18 Education, libraries 3.1 5.7 -0.2 5.2 -0.5 7.0 8.2 0.3 -0.3 -2.1 26.4
19 Health, welfare 3.6 3.6 -0.6 6.0 -0.1 l.t 8.1 1.0 -0.7 9.5 31.5
20. Entertainment 3.3 7.6 -0.1 5.7 -0.4 5.5 7.9 1.5 -0.5 -3.2 27.3
21.Personal services 2.8 4.1 -0.6 -5.0 0.1 -7.0 8.2 2.3 -0.9 11.5 15.6
22 Restaurants, hotels 0.6 10.4 0.1 -2.8 1.4 -8.8 9.9 1.2 -0.8 10.6 21.8
23.0wnership of dwellings 3.1 52 0.2 7.6 0.0 3.0 7.9 -9.2 •0.2 6.7 24.2
24.Public administration 3.1 4.4 -0.1 5.2 -0.2 -0.7 8.1 -0.4 -0.3 6.8 25.8
25.Dcfence 3.1 4.8 -0.1 5.0 -0.1 -0.4 7.3 0.0 -0.2 -10.6 8.8
The structural effects in column 1 which produce changes in K, I and GDP can be
understood in terms of the GDP identity across the bottom of Figure 5.1. With a
significant increase in C and little change in I and GDP, the balance of trade (BOT)
deteriorates (2.2 per cent of GDP, row 1, column 1). The mechanism is real
appreciation (6.5 per cent, row 3, column 1) which increases imports and reduces
exports. For exports we assume that Australia faces downward-sloping foreign
demand curves. Thus, the contraction in exports causes an improvement in the
terms of trade (3.5 per cent, row 20, column 1). The terms-of-trade improvement
increases the GDP deflator (Pg) relative to domestic expenditure deflators such as
the investment goods price index, Pi [compare rows 17 and 19 of column 1]. This is
because the GDP deflator includes the prices of exports but excludes the prices of
imports, whereas the opposite is true for domestic expenditure deflators. With the
rate of return (the rental price of capital divided by the asset price) held constant, a
decrease in Pj /Pg generates a decrease in the marginal product of capital (MPK).
This follows from the marginal productivity condition for capital:
MPK=^L = ^ ! ' i U R O R « P L . (5.10)
P P P
g Pi g g
With employment and technology fixed, a decrease in MPK requires an increase in
K (row 6, column 1). This produces an increase in both GDP (row 9, column 1) and
I (row 7, column 1). The real wage rate rises (row 5, column 1) reflecting the
increase in the K/L ratio and the consequent increase in the marginal product of
labour.
There are two features of the BOT deterioration in column 1 worth noting. First,
the bulk of the deterioration is a contraction of exports. This is because the import
content of public and private consumption is low and therefore consumption
expansion requires relatively little imports. Second, non-traditional exports
(manufactures) contract relative to traditional exports (agriculture and mining).
This is because potential manufactured exports are diverted into domestic
consumption to a much larger extent than are potential agricultural and mineral
exports.
The industry results in column 1 of Table 5.5 can be explained in terms of the
macro results in column 1 of Table 5.4. Traditional export sectors (1, 2 and 7) are
harmed by real appreciation and perform poorly. Import-competing manufacturing
sectors (4 to 6 and 8 to 10) also suffer from real appreciation but receive partial
compensation from increases in domestic demand. Most service sectors have little
exposure to international trade. These sectors are advantaged by domestic-demand
expansion and not hurt by real appreciation. However some service sectors have
strong indirect connections with exporting. Restaurants and hotels (22) is adversely
affected by the reduction in tourism exports; Transport (15) is adversely affected by
reductions in tourism and traditional exports; and Electricity (11) is adversely
affected by reductions in electricity-intensive exports such as aluminium.
The Motor vehicle industry (8a) suffers a reduction in output in column 1 of 4.9
per cent. In MONASH this industry has an Armington elasticity of 5.2 and a high
import share in its domestic market (38 and 46 per cent in 1987 and 1994, Table
5.2). This makes the industry particularly susceptible to real appreciation.
Column 2: shifts in foreign export demands and import prices
The second column of Tables 5.4 and 5.5 shows the additional effects (additional to
the momentum effects) of changes over the period 1987 to 1994 in Australia's
international trading conditions. We consider import and export markets
simultaneously. These markets are interrelated: both are affected by world inflation,
the state of the world business cycle and changes in relative prices in major
economies.
In the historical simulation we deduced shifts in the export demand curves from
information on changes in export quantities and foreign-currency f.o.b. prices. For
imports we assume that Australia is a price taker and we treat c.i.f. foreign-currency
prices of imports as exogenous in both the historical and decomposition closures.
Because our 'data are for f.o.b. export prices and c.i.f. import prices, column 2
captures the effects of changes both in traded-goods prices on major world markets
and of changes in transport costs into and out of Australia.
The decomposition simulation shows that the historically estimated shifts in
export demand curves together with the observed changes in import prices were
favourable to Australia. In column 2 of Table 5.4 these changes in international
trading conditions improved the terms of trade by 19.3 per cent (row 20). As in the
momentum column, the terms-of-trade improvement generates increases in K,
GDP, I and the real wage rate (rows 6,9,7 and 5).
By generating an increase in the price deflator for GDP relative to the price
deflator of consumption (Pg /Pc), the terms-of-trade improvement in column 2
causes an increase in real GNP (calculated using Pc as a deflator) relative to real
GDP (calculated using Pg as a deflator). This explains the sharp increase in real
consumption (which is linked to real GNP) relative to real GDP (compare rows 11
and 9). Via the GDP identity, the sharp increase in real consumption, together with
the increase in I, leads to a deterioration in the real balance of trade (export
volumes minus import volumes, rows 13 and 12), facilitated by real appreciation
(row 3). This deterioration is offset by the terms-of-trade improvement, leaving the
change in the balance of trade slightly positive (row 1).
The standard two-good trade model implies that an improvement in the terms of
trade leads to a reduction in the output of importables and an increase in the output
of exportables. With the balance of trade approximately fixed and with normal
preference assumptions, there are increases in the consumption of both importables
and exportables. This produces an increase in imports, a relatively small change in
exports (either positive or negative) and an overall increase in trade. The results in
rows 12 and 13 (a sharp increase in imports and a relatively small decrease in

exports) are consistent with these predictions.
While the decline in total exports in column 2 is only 5.2 per cent, the volume of
traditional exports (agriculture and mining) falls by 29.6 per cent and the volumes
of non-traditional (manufacturing) and tourism exports rise by 29.5 and 56.2 per
cent. The compositional change away from traditional exports is due to weaker
upward shifts in the foreign demand curves for traditional exports compared with
those for non-traditional and tourism exports. This is a continuation of a long-term
trend reduction in the prices of Australia's primary goods on world markets relative
to those of manufactured goods and services.
The sectoral results in column 2 of Table 5.5 follow easily from the macro results
in column 2 of Table 5.4. Traditional export sectors [Agriculture (1), Mining (2)
and Basic metal products (7)] are adversely affected by real appreciation and by the
weak increase in world demand for their products relative to the increases in world
demands for Australia's non-traditional and service exports. Non-trade-exposed
service sectors benefit in column 2 from increases in domestic demand. Restaurants
and hotels (22) and Entertainment (20) are the strongest performers in column 2
because of their connection with tourism. Import-competing sectors gain from
increases in domestic demand but lose from real appreciation. The net effect for
most of these sectors is negative with the most adversely affected being Textiles,
clothing and footwear (TCF, 4). The' products of this sector have high Armington
elasticities (averaging about 4) and high import shares in the domestic market.
For Motor vehicles (8a) the import share is similar to that for TCF and the
Armington elasticity (of 5.2) is higher than the average in TCF. Nevertheless,
column 2 shows less damage to the Motor vehicle industry than to TCF. The main
reason for the relatively mild impact on Motor vehicles over the period 1987 to
1994 is that the c.i.f. price of imported cars increased by about 16 per cent relative
to overall c.i.f. prices of Australia's imports and f.o.b. prices of Australia's exports.
The price of imported cars to Australia rose sharply due to an appreciation of the
Yen. Another positive influence on Motor vehicles in column 2 arises from the
structure of the demand shifts for Australian exports. These shifts strongly favoured
manufacturing (including Motor vehicles) relative to traditional exports.
Column 3: changes in protection
Over the period 1987 to 1994, protection fell in most industries. The impact effect
on landed-duty-paid prices of imported goods was a reduction of about 5.5 per cent.
Taking account of the induced devaluation of the exchange rate (2.1 per cent, row
4, Table 5.4), reductions in protection lowered landed-duty-paid import prices by
about 3 per cent. The effect on purchasers' prices was even smaller. Thus, column
3 of Table 5.4 shows only minor macroeconomic effects.
These effects include a reduction in the cost of capital goods relative to the GDP
deflator (compare rows 19 and 17), reflecting the relatively high share of tariff-
bearing imports in the cost of capital goods. The reduction in Pj /Pg induces an
increase in K (row 6), see (5.10). The increase in K induces increases in GDP (row
9) and in I and NFL (rows 7 and 2). With broad-based reductions in protection,
there is stimulation of imports (row 12) and a real-devaluation-induced stimulation
of all categories of exports (rows 14 to 16). Expansion of exports causes a
reduction in the terms of trade (row 20) leading to a contraction in consumption
(row 11). The real wage rate increases by 1.7 per cent (row 5). However, this is the
real pre-tax wage rate. If lost tariff revenue is replaced by increased income taxes,
the net effect on the real post-tax wage rate is approximately zero13.
As indicated in column 3 of Table 5.5, reductions in protection increased output
in Agriculture and Mining and reduced output in import-competing sectors such as
TCF.
The impact effect of the tariff cut on motor vehicles was to reduce the landed-
duty-paid price by 6.57 per cent (Table 5.1). Results not shown here indicate that
the effect on motor vehicle output of the reduction in the motor vehicles tariff alone
was a contraction of 7.5 per cent. Motor vehicle output was stimulated by 1.9 per
cent by the reductions in non-motor vehicle tariffs, leaving a net effect on Motor
vehicle output of all tariff reductions of 5.6 per cent (row 8a, Table 5.5).
Column 4: technical change
MONASH contains equations of the form14
Z(j) = F(Xl(j)/Al(j), X2(j)/A2G),..., Xn(j)/An(j)) (5.11)
where
Z(j) is the output of industry j ;
Xi(j) is use of input i (115 intermediate inputs, labour, capital and land) by
industry j ; and
Ai(j) is a variable allowing input-i-saving technical change in industry j .
Using information on outputs and inputs, the historical simulation reveals
movements in the Ai(j)s between 1987 and 1994. We refer to these movements as
technical changes although they encompass more than just changes in technology.
For example, we find sharp increases in the Ai(j)s where i equals cars and sharp
decreases where i equals primary factors and j equals agriculture. Rather than being
a change in technology, for cars our findings probably reflect changes in Australian
tax laws which (unintentionally) encouraged firms to buy cars to be used by
'•* In the decomposition simulation, we avoid modelling the revenue replacement tax by assuming (a)
that revenue replacement is via a non-distorting tax (e.g. a tax on labour income with fixed labour
supply) and (b) that consumption is determined in each column of the decomposition table independently
of disposable income as an exogenously given share of GNP.
The details of the production functions in MONASH are given in subsection 17.1.
employees as part of their remuneration. For agriculture, our findings probably

reflect more favourable weather in 1994 than in 1987. On the other hand, we find
increases in Ai(j)s where i equals communication equipment, communication
services, scientific equipment and computers. This probably reflects genuine
changes in technology.
The macro effects of the historically estimated movements in the Ai(j)s are shown
in column 4 of Table 5.4. With fixed employment and fixed rates of return on
capital, technical improvements (reductions in the Ai(j)s) increase the real wage
rate and increase GDP both directly (line 2, Figure 5.1) and indirectly by increasing
the capital stock (lines 5 and 3). In column 4 we find that changes in the Ai(j)s
between 1987 and 1994 increased the real wage rate by 2.9 per cent, GDP by 7.7
per cent and the capital stock by 14.2 per cent With technical change being a major
driver of capital growth, we find in column 4 a large increase in NFL (row 2). The
extra investment induced by technical change over the period 1987 to 1994 was far
in excess of the extra induced Australian savings. As shown by back-of-the-
envelope calculations in subsection 5.3(d), movements in NFL are particularly
sensitive to growth in capital.
Although the increase in GDP in column 4 is only 7.7 per cent, the increases in
imports and exports are 29.4 and 35.4 per cent (rows 12 and 13). The sharp
increase in exports causes a decline in the terms of trade of 7.6 per cent (row 20).
This, together with increased servicing costs associated with increased NFL,
explains the reduction in consumption relative to GDP (compare rows 11 and 9).
There are two reasons that the changes in the Ai(j)s were strongly trade-
favouring. First, between 1987 and 1994 it was the export-oriented industries,
especially in the agricultural and mining sectors, that experienced the largest
increases in total factor productivity. This produced a sharp increase in traditional
exports (44.8 per cent). Second, the movements in the Ai(j)s happened to favour
the use of inputs that are heavily imported. As already mentioned, over the period
1987 to 1994 there was strong growth throughout Australian industry in the use per
unit of output of communication equipment, scientific equipment, computers and
motor vehicles. The import share in the Australian market of all of these products is
very high.
In column 4 of Table 5.5 the traditional export sectors (1,2 and 7) all show large
increases in output reflecting strong cost-reducing improvements in total factor
productivity. Motor vehicles also exhibits a large increase in output, despite having
almost no total-factor-productivity growth (subsection 5.2). The industry benefits in
column 4 of Table 5.5 from the strong shift, mentioned above, in industry
technologies favouring the use of motor vehicles. Other products for which there
were strong increases in industry usage per unit of output include Finance and
insurance, Electronic equipment and Communications. This explains the large
positive entries in rows 16,9 and 15.
Personal services and Restaurants and hotels have negative entries in column 4 of
Table 5.5. There was little total-factor-productivity growth in these sectors. Thus,
in this column, their products became relatively expensive to consumers. An
additional negative factor for Restaurants and hotels was technological change
against the use of their products by industries. This was associated with
introduction in the mid-1980s of a tax on fringe benefits, that is a tax paid by firms
on benefits such as restaurant lunches provided to employees.
Column 5: changes in import/domestic preferences
MONASH uses the Armington specification of import/domestic choice (see
subsections 17.2, 17.4 and 17.5). For the typical agent k the percentage change in
the ratio of import to domestic usage of commodity i is given by:
xmf - xdf = - a f * (prnf - pdf ) + twist; , (5.12)

where
xm-' and xdf are percentage changes in the demand for imported and
domestically produced good i by agent k (e.g. consumers and industries);
pmjt and pd* are percentage changes in the prices to agent k of imported and
domestically produced good i;
G\ is a positive parameter (k's elasticity of substitution between imported and
domestically produced good i) controlling the responsiveness of the
import/domestic mix to changes in relative prices; and
twistj is a variable allowing for cost-neutral changes in preferences between
imported and domestically produced good i. Cost neutrality is imposed by
including twist terms in the demand equations for both domestic and imported
goods in such a way that these terms allow for the replacement of domestic
goods with imported goods of equal cost to the user (see subsection 17.5).
As explained in subsection 23.1, we model import/domestic twists as:
twist, = (xOdom(i) - gdpreal)+ ftwist, (5.13)
where
xOdom(i) is the percentage change in domestic output of commodity i;
gdpreal is the percentage change in real GDP; and
ftwist; is a shift variable.
The first term on the RHS of (5.13) allows for demand pressures. It captures the
idea that when output of commodity i in the domestic economy is growing rapidly,
there is a tendency for demand shifts to occur towards imports. This is explained by
shortages and lengthening queues and is unrelated to movements in relative prices.
Similarly, when output of i is growing slowly there is a tendency for shifts to occur
towards the domestic product. The second term allows for twists in import/domestic
ratios beyond those that can be explained by changes in relative prices and demand
pressures.
In the historical simulation we measured ftwistj using data: on GDP; on demands
for imports and domestically produced goods; and on import and domestic prices.
In the decomposition simulation the ftwists are exogenous. Their effects over the
period 1987 to 1994 are shown in column 5 of Tables 5.4 and 5.5.
In aggregate, the ftwists were in favour of imports. A possible explanation is the
elimination in the mid-1980s of import quotas. These quotas were often applied to
quantities of imports measured in numbers of items. For example, motor vehicle
quotas operated on the number of imported cars, not their value. This limited car
imports to a narrow range of large expensive cars. Similarly, quantity quotas on
textiles, clothing and footwear limited the quality range of these imports. With the
removal of quotas there was a rapid increase in the variety of imported products.
We think this caused a preference shift in favour of imports. To the extent that this
is correct, the results in column 5 can be interpreted as lagged effects of reductions
in protection occurring before the commencement of the period which we are
studying in this chapter.
The effect of the ftwists on import volumes in column 5 of Table 5.4 is an
increase of 8.7 per cent. With L, ROR and TECH held constant, Figure 5.1 suggests
that the ftwists should have little effect on K and I. This is borne out in rows 6 and
7. With little impact on I there is little impact on the balance of trade (row 1). Thus
the ftwists are accompanied by an increase in exports to match the increase in
imports, requiring real devaluation (4.4 per cent, row 3). The percentage increase in
export volumes exceeds that in import volumes because export expansion causes a
decline in the terms of trade. The terms-of-trade decline reduces real GNP and real
consumption.
Another effect of the reduction in the terms of trade is to increase Pj /Pg (rows 19
and 17). Via (5.10), this increases MPK, reducing MPL and the real wage rate (row
5). With an increase in MPK, we might expect associated decreases in K and I.
Nevertheless K and I show small increases. This reflects a structural change
favouring capital-intensive industries, especially in the export-oriented mining
sector (row 2 of column 5 in Table 5.5).
The effect oh the Australian motor vehicle industry of the ftwists for Motor
vehicles alone is strongly negative (-7.9 per cent, not shown in the tables). This is
partly offset (4.2 per cent, again not shown in the tables) by the general equilibrium
effects (e.g. real devaluation) of the ftwists for other products. This leaves a net
effect of-4.0 per cent in row 8a of column 5 in Table 5.5.
Column 6: the effects of changes in consumer preferences, i.e., changes in the
parameters of the household utilityfunction
In the historical simulation, changes in the parameters of the household utility
function are revealed by the values of endogenous shift variables in MONASFTs
household demand functions.15 The values of these shift variables reflect the data
for 1987 to 1994 on household demands, on consumer prices, on household income
and on the number of households. In the decomposition simulation, the changes in
the shift variables are exogenous, that is we exogenize consumer preferences.
Column 6 in Tables 5.4 and 5.5 show the effects of the estimated changes in
preferences between 1987 and 1994.
Over this period, household purchases of motor vehicles increased by more than
can be explained by changes in: the number of households; household income; and
consumer prices. Thus MONASH in historical mode indicates that there was a shift
in consumer preferences in favour of motor vehicles. Reflecting this shift we find a
positive entry in row 8a, column 6 of Table 5.5 (4.4 per cent). On the other hand,
there is a large negative entry (-10.3 per cent) in row 4. This arises from a shift in
preferences away from textiles, clothing and footwear, possibly explained by
increasingly casual styles of dress.
At the macro level, the effects of shifts in consumer preferences were minor
(column 6, Table 5.4).
Column 7: the effects of employment growth and growth in the number of
households
Column 7 of Table 5.4 shows the effects over the period 1987 to 1994 of growth in
employment (10.1 per cent) and in the number of households (10.0 per cent).
With constant returns to scale, fixed rates of return, fixed I/K ratios and no
change in technology, we would expect the system depicted by lines (1) - (7) and
(16) - (17) in Figure 5.1 to transform a 10.1 per cent increase in employment into
10.1 per cent increases in K, I and GDP with no change in the real wage rate.
However, a larger domestic economy produces more exports with an associated
decline in the terms of trade. This increases Pi/Pg(rows 19 and 17), restricting the
increases in K, I and GDP to 9.2, 8.4 and 9.8 per cent [see (5.10)]. With a reduction
in the K/L ratio, there is a reduction in the marginal product of labour and in the
real wage rate (row 5).
By causing an increase in Pc /Pg, terms-of-trade deterioration in column 7 is a
factor in limiting the growth in real consumption to 7.5 per cent, well below that of
real GDP, 9.8 per cent. However, the main factor is accumulation of foreign
liabilities, reflecting rapid growth in K [see subsection 5.3(d)]. The increase in NFL
restricts consumption by restricting the growth in real GNP (7.3 per cent, row 10).
With subdued growth in consumption and investment relative to GDP, column 7
shows a 14.6 percentage point gap between export and import growth. This is
achieved with a sharp increase in exports (23.3 per cent, row 13) facilitated by real
15
These are the shift variables A3(i)/A3AVE which appear in (16.13) and are discussed in subsection
16.1(a).
devaluation of 8.6 per cent (row 3). Sharp growth in exports is required to achieve
the 14.6 percentage point gap because growth in imports (8.7 per cent, row 12) is
close to that of GDP. While real devaluation reduces import growth, imports are
favoured by the change in the composition of GNE away from consumption which
has low import intensity.
All categories of exports exhibit strong growth in column 7. However, non-
traditional and tourism exports expand relative to traditional exports (agriculture
and mining). This is because exportable manufactured goods and tourism services
are freed from domestic consumption by subdued growth in domestic demand
relative to GDP. This occurs to a much smaller extent for exportable agricultural
and mineral products.
In Table 5.5 we find that heavily trade-exposed sectors, 1 to 10, grow relative to
GDP. These sectors benefit from real devaluation. Among the trade-exposed
sectors, Agriculture has a relatively small expansion. This is explained by the fixity
of agricultural land. Motor vehicle output grows by 16.7 per cent (row 8a). Real
devaluation is particularly good for this industry because of its high Armington
elasticity and the high share of imports in its domestic market
Column 8: the effects of apparent changes in required rates of return
MONASH contains rate of return equations which can be summarised as
r o r j = p * ( k j - k a v e ) + fj (5.14)
where
rorj is the percentage change in the rate of return on capital in industry j ;
kj is the percentage change in j's capital stock;
kave is the percentage change in the economy-wide capital stock;
P is a positive parameter; and
fj is a combination for industry j of shift variables that can be used to move j's
rate of return according to j's sector and according to the initial value (value in
1987) ofj's rate of return relative to the economy-wide average rate of return.
For given values of the shift variables making up fj, (5.14) implies that fast growth
is associated with increased rates of return and slow growth with reduced rates of
return. As alluded to in footnote 10, the p term in (5.14) plays a useful damping
role in decomposition and other long-run simulations in which the f)s are
exogenous.
In the historical simulation for 1987 to 1994, we treated the fjs endogenously. In
modelling them, we assumed that there is a tendency for high rates of return to fall
and low rates of return to rise, that is, if industry j ' s rate of return was high (low) in
1987 relative to the economy-wide average rate of return, then, on this account, we
adopted a negative (positive) value for fj. As explained in subsection 16.1(i), the fjs
were also modelled in the historical simulation to ensure that (5.14) was compatible
with sectoral data on capital growth and on movements in rates of return (revealed
by data on prices and costs). Having evaluated the fjs in the historical simulation,
we exogenize them in the decomposition simulation and report the effects of their
movements in column 8 of Tables 5.4 and 5.5.
On average the fjs increased between 1987 and 1994 by 8.3 per cent. In terms of
Figure 5.1, we can think of column 8 as being generated by the imposition of a
positive shock to ROR, holding constant L and TECH. Consistent with this
representation, column 8 in Table 5.4 shows decreases in K, I and GDP. With fixed
L there is a fall in K/L producing a reduction in real wage rates. NFL declines
sharply in response to the decrease in,K [see subsection 5.3(d)], This allows GNP
and consumption to increase relative to GDP. The increase in consumption relative
to GDP is sufficient to cause imports to increase relative to exports, despite the

reduction in investment. The decline in the real balance of trade, which is facilitated
by real appreciation, is tilted towards export contraction rather than import
expansion. For imports, the positive effect of real appreciation is outweighed by the
negative effects of: (a) reduced domestic output (and consequent reductions in the
use of imported inputs); and (b) the shift in gross national expenditure towards
consumption (a low import-intensive category of demand) and away from
investment (a high import-intensive category of demand).
Consistent with real appreciation, column 8 of Table 5.5 generally shows poor
outcomes for trade-exposed sectors relative to those for non-trade-exposed sectors.
However there are two notable exceptions. Mining (an export sector) expands by
4.9 per cent and Ownership of dwellings (a non-traded sector) contracts by 9.2 per
cent.
Mining was a high-rate-of-retum sector in 1987 giving it sharp overall decreases
in its fjs for 1987 to 1994. When these negative fjs were applied as shocks in the
decomposition simulation, the result was a strong increase in the sector's capital
stock and in related variables such as output.
For Ownership of dwellings, the historical simulation implies a sharp increase in
its rate of return and consequently a positive value for its shift variable fj. This
reflects mainly the data on the rental value of houses. In the decomposition
simulation, the positive shock to fj produced a large increase in the price of housing
services with a corresponding reduction in demand.
The entry in column 8 for Motor vehicles is strongly negative (-7.0 per cent, row
8a). The industry is adversely affected in column 8 by three factors: (a) an increase
in its required rate of return from a low value in 1987; (b) real appreciation; and (c)
the decline in economy-wide investment (a vehicle-intensive category of
expenditure).
Column 9: the effects ofshifts in export supply curves
As part of the explanation in the historical simulation of changes in export prices
and volumes (both observed) we endogenize the positions of the export demand
curves and the levels of supply-shifting export taxes and subsidies (phantoms)16. In
the decomposition simulation the demand and supply shift variables are exogenous.
The effects of the demand shift variables have already been discussed in relation to
column 2 of Tables 5.4 and 5.5. The effects of the supply-shift variables (phantom
export taxes and subsidies) are given in column 9.
The main export supply shift deduced from the historical simulation was a
downward movement in the supply curve for non-traditional exports. This is
consistent with McKinsey and Co. (1993) who recorded a large increase since the
mid-1980s in the awareness of Australian manufacturers of exporting possibilities
in niche markets. On imposing the export-supply shifts in the.decomposition
simulation, we generated a sharp increase in non-traditional exports (Table 5.4, row
15).
Overall, the export supply shifts (mainly downward movements) generate a
reduction in the terms of trade of 2.2 per cent and an increase in exports of 3.3 per
cent (Table 5.4, rows 20 and 13). Deterioration in the terms of trade reduces
consumption (row 11). Reflecting the relative capital intensity of export industries,
export stimulation generates increases in the capital stock, GDP and investment
The reduction in consumption and the increase in GDP outweigh the increase in
investment, requiring a movement in the real balance of trade towards surplus
(compare rows 12 and 13). Nevertheless, the real exchange appreciates. In view of
the downward movements in export-supply curves, real appreciation is required to
restrain the improvement in the real balance of trade to that implied by the
movements in the other components of the GDP expenditure identity.
As mentioned earlier, output in the Motor vehicle industry is highly sensitive to
the real exchange rate. The industry's output in row 8a, column 9 of Table 5.5
declines by 1.8 per cent.
Column 10: otherfactors
Decomposition simulations allow us to look at the effects of an overwhelming
number of exogenous variables. Inevitably, we must terminate the process by
having an "other column". Here, this is column 10 of Tables 5.4 and 5.5.
The main shocks in column 10 are to macro ratios. Recall from lines (7) and (10)
in Figure 5.1 that in the decomposition simulation we exogenize I/K ratios and the
average propensity to consume (C/GNP). In columns 1 to 9 of Tables 5.4 and 5.5,
these macro ratios were fixed. In column 10 we introduce, as exogenous shocks, the
changes in these ratios that were endogenously determined in the historical
simulation. Also important in column 10 are shocks to relative sectoral wage rates.
16
The supply-shift variables are modelled as export taxes and subsidies. We refer to them as phantoms.
They are not genuine taxes and subsidies and do not appear in the MONASH specification of the
government accounts. Their role is to allow MONASH in historical mode to "explain" observed export
behaviour. Details are in subsections 16.1(g) and 19.
For most sectors, the I/K ratio was lower in 1994 than in 1987, explaining the
strongly negative result for real investment (Table 5.4, row 7). The C/GNP ratio
was higher in 1994 than in 1987, explaining the positive result for real consumption
(row 11). Between 1987 and 1994, there was a sharp decline in the wage rates of
construction workers, explaining the reduction in the price deflator for investment,
Pi (row 19). With rates of return constant in column 10, the reduction in Pj
generates an increase in the capital stock (row 6) and associated increases in real
GDP (row 9) and net foreign liabilities [row 2, see subsection 5.3(d)].
Together, the changes in real investment, real consumption and real GDP
generate a 15.4 percentage point gap between growth in exports and imports. This
gap is fairly evenly distributed between export expansion and import contraction.
Many of Australia's export-oriented industries are capital intensive and are

favoured by the reduction in Pj. This explains the strong increase in total exports
(9.1 per cent, row 13) despite real devaluation of only 1.2 per cent (row 3). Imports
are reduced in column 10 by investment contraction, an import-intensive
component of aggregate demand.
Some of the sectoral results in column 10 of Table 5.5 follow easily from the
macro results in Table 5.4 while others require us to consider some quite specific
sectoral shocks which are included in the "other" category. Strong growth in
Mining (row 2), a highly capital-intensive export-oriented sector, follows from the
reduction in the cost of capital, Pi. The negative results for Construction and
Electronic equipment (rows 12 and 9) follow from the contraction in investment. In
the case of Motor vehicles, output is dependent on both investment and
consumption. However, while investment is only about 20 per cent of GNE, it
accounts for about 30 per cent of Motor vehicles sales (see Table 5.2). Thus, the
change in the composition of GNE in column 10, which is strongly against
investment, has a negative effect on Motor vehicle output (-2.9 per cent, row 8a).
The positive result for Ownership of dwellings (row 23) follows from the
expansion in consumption and the reduction in the cost of capital.
Among the specific sectoral shocks which have a marked influence in column 10
of Table 5.5 are changes in the commodity composition of government expenditure.
Over the period 1987 to 1994, Public administration (row 24) and Health (row 19)
gained share in public expenditure at the expense of Defence (row 25) and
Education (row 18).
Another group of specific shocks which influence the sectoral results in column
10 are those for phantom consumption taxes. These were introduced endogenously
in the historical simulation to reconcile data on consumption prices with data on
costs in consumption industries (see section 31, step 12). For Personal services and
Restaurants (rows 21 and 22) reconciliation of cost and consumer price data
required significant negative movements in phantom consumption taxes. When
these movements are introduced as exogenous shocks in column 10 of the
decomposition simulation, they produce sharp negative movements in the consumer
prices of the services produced by these two sectors and corresponding stimulation
of demand.
Cross-column comparison: some conclusions for the macro economy and for the
Motor vehicle industry
Comparison of columns in Table 5.4 provides an explanation of some of the main
macro developments in the Australian economy over the period 1987 to 1994,
including: the appreciation of the real exchange rate; the strong growth in imports
and exports; the sharp increases in non-traditional and tourism exports relative to
traditional exports; and the sluggish growth in real wages.
Row 3 indicates that real appreciation arose mainly from favourable movements
in Australia's international trading conditions (column 2). These movements were
also important in explaining import growth (row 12). However, the main cause of
import growth was technical change (column 4). This favoured the use of products
that are heavily imported such as electronic equipment Other important factors in
the growth of Australia's imports were shifts in import/domestic preferences
(column 5) and growth in employment (column 7).
The main cause of Australia's export growth was the same as that for import
growth, technical change (column 4). Technical change stimulated exports relative
to GDP because total-factor-productivity growth was strongest in export-oriented
industries especially agriculture and mining.
Strong growth in non-traditional and tourism exports relative to traditional
exports is explained in Table 5.4 by changes in Australia's international trading
conditions (column 2). In the period 1987 to 1994, prices of non-traditional and
tourism exports grew rapidly relatiye to prices of traditional exports. The growth in
non-traditional and tourism exports relative to traditional exports was slowed by
technical change (column 4) which strongly favoured traditional exports.
For most periods of economic history, we suspect that technical change is the
major source of growth in real wages. However, over the period 1987 to 1994,
technical change in Australia was sluggish, producing little increase in total-factor
productivity. With capital representing about 30 per cent of GDP, total-factor
productivity growth in column 4 is only is 3.4 per cent [= 7.7 - 0.3x(14.2)]. This
produces a comparable increase in real wages, 2.9 per cent. In column 2 the
increase in real wages is 10.8 per cent, indicating that Australia's main source of
real wage growth in the period of our study was improvements in international
trading conditions.
Looking across the columns of Table 5.5, we can identify the sources of growth
for 1987 to 1994 in different sectors of the Australian economy. Here we
concentrate on the Motor vehicle industry, row 8a. Among the conclusions from
this row are the following.
• A major positive influence on Motor vehicle output was technical change
(column 4). Technical change in the Motor vehicle industry itself was slow.
However technical change in other industries favoured the use of Motor

vehicles. By disaggregating column 4 we found, in results not reported here,
that changes in the composition of the intermediate inputs used by industries
accounted for all of the growth in Motor vehicle output in column 4.
• General employment growth in the economy (column 7, Table 5.5) was
another important favourable influence on the Motor vehicle industry.
• Changes in consumer preferences favoured the use of motor vehicles,
contributing 4.4 per cent to the growth in the industry's output (row 8a, column
6, Table 5.5).
• Increases in required rates of return exerted a strong damping influence on
Motor vehicle output (column 8).
• Another major negative influence was cuts in protection. The Motor vehicle
industry benefited from cuts in protection of other industries but suffered from
cuts in its own protection. The net effect was a reduction in the industry's
output of 5.6 per cent (column 3).
• Although c.i.f. prices of imported cars increased relative to the prices of most
imports, Motor vehicle output was reduced by changes in Australia's
international trading conditions (column 2). The favourable effect on the
domestic industry of relatively high prices for imported cars was outweighed
by real appreciation associated with improvement in the terms of trade.
• Shifts in user preferences towards imported motor vehicles had a strong
negative influence on Motor vehicle output. However, the damage shown in
row 8a of column 5 is mild, 4.0 per cent By reducing the exchange rate,
import-favouring preference shifts outside the Motor vehicle industry had a
positive influence on Motor vehicle output.
In the Australian policy debate, protection is often portrayed as the critical
variable determining the welfare of the Motor vehicle industry. An implication of
the decomposition tables is that the health of the Motor vehicle industry depends on
many factors apart from protection. These include international trading conditions,
technology, economy-wide employment growth, import/domestic preferences and
required rates of return on capital.
(d) Appendix: back-of-the-envelope explanation of results for netforeign liabilities
A prominent feature of Table 5.4 is the volatility of the results in row 2 for net
foreign liabilities (NFL). As a share of GDP in 1987, the change in NFL varies
from -71.5 per cent in column 8 to 63.2 per cent in column 4. The MONASH
equation for determining movements in NFL in decomposition simulations is
complex, see (16.63) and the associated explanation in subsection 16.1(m).
However, we need not go to technical material to obtain a good understanding of
the results in row 2. This can be done by back-of-the-envelope calculations.
In each of the columns of Table 5.4, the effect on NFL at the beginning of 1994
of the shocks under investigation is given approximately by
ANFL= 6 * ^ * R I + P i * A K - 6 * A P S * ^ ^ - 6 * - A ^ * G N P - N F L * - | -
2 2 2 100
+ 6*0.05*^^-156*AU (5.15)
2
where the A refers to a change in a variable.
The first term on the RHS of (5.15) recognizes that if the shocks in a column
change the price of investment goods (Pj) then they will affect NFL via the cost of
replacement investment, that is the cost of maintaining the 1987 level of capital
through to the beginning of 1994. Assuming that the change in the price of
investment goods (AP,) occurs in a linear fashion, then the extra cost of replacement
investment over the six years from the end of 1987 to the beginning of 1994 is
APj/2 times the annual quantity of replacement investment (RI) for 6 years.
The second term is the cost of extra capital (AK). We ignore the cross product
term AP,*AK.
The third and fourth terms take account of additional savings arising from the
change in GNP and from the change in the average propensity to save (APS). In the
third term, additional annual saving in the 6 years from the end of 1987 to the
beginning of 1994 is APS times the average increase in GNP, AGNP/2. In the
fourth term the additional annual saving is the average change in the APS, AAPS/2,
times GNP. We ignore the cross product, AAPS*AGNP.
Where ty is the percentage appreciation of the Australian dollar caused by the
shocks under investigation, the fifth term approximates the exchange-rate valuation
effect. We assume that net foreign liabilities are incurred in foreign currency. Thus,
shocks which cause a 10 per cent appreciation of the $A reduce the Australian
dollar value of NFL by 10 per cent.
The sixth term on the RHS of (5.15) allows for extra interest associated with extra
net foreign liabilities. On average, in each of the six years 1988, 1989,...,1993, this
extra interest is 0.05 (the interest rate) times the average increment in NFL,
(ANFL/2).
The final term allows for the momentum effect. It is relevant only for column 1 of
Table 5.4 (AU ** 1). In the other columns AU is zero. The coefficient for AU in
(5.15) is the change in NFL (in billions of Australian dollars) that would have taken
place from the beginning of 1987 to the beginning of 1994 if there were no
changes: in capital prices (APj = 0); in capital stocks (AK = 0); in GNP (AGNP = 0);
in the average propensity to save (AAPS = 0); and in the nominal exchange rate (<))
= 0). Under these no-change assumptions, savings and investment in each year
from 1987 to 1993 would have been the 1987 levels of savings and of replacement
investment. From our database for 1987 we find that these levels were $55.6 billion
and $31.7 billion. The interest payment on the 1987 level of net foreign liabilities
was $4.8 billion (an interest rate of 5 per cent applied to NFL of $97 billion). Thus
under the no-change assumptions, there would have been a reduction in NFL in
1987 of $19.1 billion (= 55.6 - 31.7 - 4.8); a further reduction in 1988 of $19.1
billion plus saved interest of 0.05*19.1 associated with debt reduction in 1987,
giving a total reduction in 1988 of 19.1*1.05; a further reduction in 1989 of $19.1
billion plus saved interest of 0.05*(19.1+1.05*19.1), giving a total reduction in
1989 of 19.1*1.052; and so on up to the end of 1993. The overall reduction in NFL
from the beginning of 1987 to the beginning of 1994 would have been $156 billion.
This is the coefficient on AU in (5.15).
Using values from our mid-point database (see footnote 12 in ch. 1 and section
22) for P.K/GDP (3.48), P.RI/P.K (0.033), GNP/GDP (0.98), NFL/GDP (0.44),
APS (0.19) and GDP (332), we find that (5.15) reduces to
10Q.iggk.ft34."10'*1> 100,AK
-0.56'100*AONP
+3.48»
GDP Pi K GNP
rt^*100*AAPS nt.+ . , r]k ANFL „_* 4TT ,c . , .

-0.56* 0.44*6 + 15* 47*AU . (5.16)
APS GDP
Solving for (ANFL/GNP) gives
1OO.ANFL=041,100*APi+409,100*AK_066,100*AGNP
GDP Pi K GNP
1 0 0 * AAP<1
-0.66* -0.52*<1)-55*AU . (5.17)
APS
We can evaluate the RHS of (5.17) using results from Table 5.4. As shown in
Figure 5.2, equation (5.17) approximately reproduces the results in row 2 of Table
5.4. The only significant discrepancy is in column 8.17
From (5.17) we see that movements in net foreign liabilities are highly sensitive
to capital growth (AK/K). Thus in columns 4, 7 and 8 of Table 5.4, columns with
large capital movements, we find large changes in NFL. Despite significant capital
growth in column 2, the movement in NFL is zero. This reflects sharp appreciation.
In column 1, the large movement in NFL is explained by the AU term.
1' We traced this discrepancy to our measure of capital stock. In row 6 of Table 5.4 we measure
additional capital by a rental-weighted average of the percentage changes in industry capital stocks. A
better measure of additional investment requirements is an asset-weighted average of the percentage
changes in industry capital stocks. We found that it was only in column 8 that there was a significant
difference between these two measures. In column 8 there are noticeable changes in relative rental rates
on capital in different industries, associated with assumed changes in required rates of return. When we
used the asset-weighted measure of total capital in our back-of-the-envelope calculations the result for
column 8 sharply improved.
6. Forecast simulation: prospects for the Australian motor vehicle industry,

1998 to 2016
Table 6.1 provides estimates for 1987 to 1994 and forecasts for 1998 to 2016 for
macro and motor vehicle variables. The purpose of this section is to explain the
motor vehicle forecasts. Our strategy is as follows. In subsection 6.1 we explain the
forecast movement in imports of motor vehicles relative to domestic sales of
domestically produced motor vehicles, that is we explain the ratio M/D. Then in
subsection 6.2 we explain the forecast growth in domestic sales of motor vehicles,
that is we explain S where S = M + D. At this stage we have a complete
explanation of S, M and D. In subsection 6.3 we introduce our forecast for exports
(E) of motor vehicles. This then gives us a forecast for output (Z), where Z= D+E.
A summary of the forecasts is in subsection 6.4.
6.1. Forecast growth in imports relative to growth in domestic sales of

domestically produced motor vehicles (M/D)
As can be seen in Table 6.1, between 1987 and 1994 domestic sales of domestically
produced motor vehicles grew by 1.6 per cent a year and imports grew by 7.3 per
cent a year implying an annual 5.6 per cent (= 100*(1.073/1.016 - 1)) increase in
the import/domestic ratio (M/D) in domestic sales.
This is explained in part by a decrease in the basic price18 of imported motor
vehicles (PMB) relative to that of the domestic product (PDB). Real appreciation
(Table 5.4, row 3, column 11) and reductions in tariffs had significant negative
influences on the ratio of these prices. However, to a large extent these influences
were offset by strong growth in the c.i.f. price of imported motor vehicles relative
to that of other commodities. [Whereas c.i.f. prices in general increased at an

average rate of 1.3 per cent a year, the c.i.f. price of imported motor vehicles
increased at an average rate of 3.8 per cent.] This limited the decline in PMB/PDB to
0.7 per cent a year (=100*(1.028/1.035-1), see Table 6.1). The Armington elasticity
in MONASH for motor vehicles is 5.2, and margins represent about 15 per cent of
purchasers' prices. Assuming that margin costs are independent of basic prices, we
find that the change in the import/domestic basic price ratio generated an annual
increase in the M/D ratio of about 3 per cent ( = 5.2*0.85*0.7).
The remaining 2.6 per cent a year increase in the import/domestic ratio is
explained mainly by a twist in user preferences towards imports.19 With imports
occupying close to half the domestic market (Table 5.2), the twist in user
preferences increased the growth in imports by about 1.3 percentage points a year
and reduced the growth of domestic sales of the domestic product by a similar
amount. The twist in user preferences towards imports reflected increased variety in
imports relative to the domestic product As discussed in subsection 5.3(c), column
5, the elimination of import quotas in the mid 1980s led to increased variety among
imports. At the same time, government policy toward the Motor Vehicle industry
(the Button Plan) led to a reduction in the number of domestic product lines.
Because the variety of foreign cars relative to that of domestic cars on the
Australian market is likely to continue increasing, but at a slower rate, we assume a
continuing twist in preferences towards imports, but at a slower rate. Instead of the
1
The basic price of an import is the landed-duty-paid price and for domestic products it is the factory-
door price. Basic prices are separated from purchasers' prices by sales taxes and margin costs (e.g.
wholesale, retail and transport costs).
19
In the historical simulation the value of twist(70), the variable on the LHS of (23.3), for the seven year
period was about 22 per cent. This contributes slightly more than 2.6 per cent annual growth to M/D. A
small negative influence on growth in M/D was the relatively slow growth in the Motor vehicle industry,
a major user of imported motor vehicle parts.
Table 6.1. Growth rates in macro and motor vehicle variables:

1987 to 2016
Estimates for Basecase'
Percentage annual growth in: 1987 to forecasts for
1994 1998 to 2016
Macro variables
Real GDP 2.6 3.0
Real investment 0.7 2.6
Real consumption 3.3 3.3
C.i.f. price of imports 1.3 1.6
Motor vehicles:
Sales on domestic market (domestic and imported) (S) 4.3 2.6

Domestic sales on domestic market (D) 1.6 1.2
Imports (M) 7.3 3.7
Domestic output (Z) 2.0 1.7
Exports (E) 6.1 4.8
Basic price of domestic product 3.5 1.6
Basic price of imports (includes tariff) 2.8 1.1
C.i.f. price of imports (excludes tariff) 3.8 1.5
* In the basecase forecasts, the scenario for the tariff on motor vehicles (MONASH
commodity 70) is that shown in the first column of Table 7.1.
twist contributing 2.6 percentage points a year to the increase in the M/D ratio for
motor vehicles, we assume that its contribution will be halved to 1.3 percentage
points a year. In addition; our forecasts imply that relative price changes will make
a smaller contribution to M/D growth than they did in history (1987 to 1994). Tariff
cuts in the basecase forecasts are less significant than in history but we assume that
there will be no further increase in the c.i.f. price of imported motor vehicles
relative to the c.i.f. prices of all imports. Our assumptions for tariffs and c.i.f.
import prices, together with the other assumptions underlying our forecasts,
generate an annual decrease in PME/PDB for motor vehicles of 0.5 per cent a year (=
100*(1.011/1.016 - 1)). This is reduced to about 0.4 per cent a year for purchasers'
prices. By multiplying by the Armington elasticity of 5.2, we find that the effect of
relative price changes is to increase M/D for motor vehicles in our forecasts by
about 2.1 (=5.2x0.4) per cent a year. Together, the twist and relative price
contributions explain an annual increase in M/D of 3.4 per cent a year. Our actual
forecast of 2.5 per cent (=3.7 - 1.2) reflects not only twist and relative price effects,
but also changes in the forecast composition of activity throughout the economy.
This favours domestically produced motor vehicles relative to imports mainly
because the motor vehicle industry, which is a major user of imported motor
vehicles20, is forecast to grow slowly relative to GDP.
6.2. Forecast growth in sales of motor vehicles (S)

In history, sales of motor vehicles on the domestic market (S) increased by 4.3 per
cent a year. This is a combination of the 7.3 per cent annual growth in imports (M)
and the 1.6 per cent annual growth in domestic sales of domestic output (D)
discussed in the previous subsection. Growth in GDP in the historical period was
2.6 per cent a year, growth in real consumption was 3.3 per cent a year and growth
in investment was 0.7 per cent a year. These macro growth rates suggest sales
growth for motor vehicles in the Australian market of about 2.0 per cent a year21.
To explain 4.3 per cent a year sales growth, our historical simulation implied a
small shift in consumer preferences and a strong shift in industry technologies
towards the use of motor vehicles [a small positive value for the percentage change
in A3(70)/A3AVE and a large positive value for the percentage change in Al(70),
see (16.13) and (16.5) described in subsection 16.1]. These shifts contributed about
2.3 per cent a year to growth in motor vehicle sales. The strong shift in industry
technologies towards motor vehicles is consistent with their increasing use as part
of remuneration packages.
In our forecasts for 1998 to 2016, annual real growth rates in GDP, consumption
and investment are 3.0 per cent, 3.3 per cent and 2.6 per cent, suggesting an
increase in the rate of growth of motor vehicle sales of 0.7 percentage points, that is
an increase in the annual growth in sales from 4.3 per cent a year to 5.0 per cent22.
However, growth in motor vehicle sales in our forecasts is only 2.6 per cent. In our
forecasts we have assumed no further shifts in industry and consumer preferences
towards the use of motor vehicles. Thus, rather than contributing 2.3 per cent a year
to the growth in motor vehicle sales, as they did in history, in the forecasts these
shifts contribute nothing.
u
The commodity Motor vehicles in MONASH includes motor vehicle parts. The domestic motor
vehicle industry relies heavily on imported parts.
1
In our data for 1987, the shares of investment, consumption and intermediate usage in the sales of
motor vehicles (domestic and imported) are 0.39, 0.22 and 0.40. Using these shares and assuming that
intermediate usage grows in line with GDP we calculate the expected growth in motor vehicle sales as:
expected sales growth • 0.39x0.7 + 0.22x3.3 + 0.40x2.6 = 2.0.
22
•" In our data for 1994, the shares of investment, consumption and intermediate usage in the sales of
motor vehicles (domestic and imported) are 0.30, 0.26 and 0.45. Using these shares and continuing to
assume that intermediate usage grows in line with GDP we calculate the expected extra sales growth in
our forecasts compared with historyfromextra growth in investment, consumption and GDP as:
extra sales growth = 0.30x(2.6-0.7) + 0.26x(3.3 - 3.3) + 0.45x(3.0-2.6) = 0.7.
With sales (S) and the M/D ratio having forecast annual growth rates of 2.6 and
2.5 per cent, and with the import share of domestic sales averaging about 55 per
cent in the forecast period, we calculate that the annual growth rates for M and D
are 3.7 and 1.2 per cent, the values shown in Table 6.1.
6.3. Forecast growth in output of motor vehicles (Z)

We forecast that exports of motor vehicles (E) will grow at 4.8 per cent a year. The
rate achieved in the historical period was 6.1 per cent. The slowdown reflects our
forecast that the rate of growth of manufactured exports will be below the very high
rates of growth since the mid-1980s. Exports in the forecast period account for
about 13 per cent of motor vehicle output Thus, together with our forecast of 1.2
per cent annual growth in domestic sales of the domestic product (D), our export
forecast implies annual output growth for motor vehicles (Z) of 1.7 per cent.
6.4. Summary of forecasts for motor vehicles

Our forecast for annual growth in motor vehicle output is slightly lower than that
achieved in recent history. This is the result of positive influences being slightly
outweighed by negative influences.
Relative to recent history, in our forecasts we expect stronger growth in
investment, consumption and GDP. This will be a positive influence on motor
vehicle output On the other hand, we expect a slowdown in the rate of shift of
consumer preferences and industry technologies towards motor vehicles. This will
be a negative influence on output
Another positive influence on industry output in the future relative to history will
be a slowdown in the rate of growth of M/D. In the forecast period we expect less
reduction in tariffs and less twist in user preferences towards imported vehicles.
Partly offsetting this, we assume that c.i.f. prices of imported vehicles will not
increase relative to c.i.f. prices of other imports. In the historical period, growth in
M/D was inhibited by a sharp increase in the c.i.f. prices of imported vehicles
relative to other c.i.f. prices.
A final negative influence on the rate of growth of motor vehicle output in the
forecasts relative to history is a slowdown in the rate of growth of exports.
7. A policy simulation: the effects of reductions in the tariff on motor vehicles
This section reports results from a MONASH policy simulation concerned with the
effects of reductions in tariffs on motor vehicles (MONASH commodity 70).
Different tariff rates apply to the sub-commodities within commodity 70. As
explained in Dixon et al. (1997a) the average tariff rate applying to commodity 70
in 1997 was 19.13 per cent Government plans in 1997 implied that the average
tariff rate would follow the path shown in column (1) of Table 7.1, that is it would
fall from its 1997 level to 11.69 per cent in 2001 and then stabilize. In 1997 the
Industry Commission (a Government research organization with the responsibility
for advising on micro-policy) recommended that the reductions in motor vehicle
tariffs be continued beyond 2001 according to the path shown in column (2) of
Table 7.1. In the basecase forecasts described in section 6 we adopted the tariff path
from column (1). In the policy simulation to be described in this section we
calculate deviations from the basecase forecasts caused by adopting the tariff path
in column (2), that is we calculate the effects of adopting the Commission's
recommendation.
Our results are set out in Charts 7.1 to 7.16. Before we embark on detailed
explanations, it will be useful to summarize the charts and to hint at the issues that
they raise.
Chart 7.1 shows that the tariff cuts will have: negative short-run and zero long-
run effects on employment; ever expanding negative effects on the aggregate
capital stock; and positive effects on real GDP despite reductions in factor inputs.
Although the capital stock path in Chart 7.1 slopes down, the investment path in
Chart 7.2 exhibits positive deviations for most of the simulation period.
Real consumption (Chart 7.2) and real national wealth (Chart 7.9) are reduced
despite increases in GDP.
Charts 7.3, 7.4 and 7.6 show, as expected, that the tariff cuts stimulate both
exports and imports, reduce the real exchange rate and reduce the terms of trade.
However, we will need to explain the negative slopes beyond 2005 in the graphs
for exports and imports and the positive slopes for the real exchange rate and the
terms of trade.
Charts 7.5 and 7.8 show that the tariff cuts have an expanding negative effect on
real wage rates and require an increasing rate of consumption tax to replace lost
tariff revenue.
In explaining these and other aspects of the charts we make frequent reference to
quantity and share effects. If X is a Divisia or other variable-share index of micro
variables Xj, X can grow faster in a policy simulation than a forecast simulation for
either of two reasons: (a) the XjS grow faster in policy than in forecast (quantity
effect) or (b) the X;S with relatively high forecast growth rates have larger weights
in policy than in forecast (share effect). In modelling exercises under balanced
growth assumptions, share effects are negligible because all of the X;S have the
same forecast growth rates. In MONASH, where there can be wide range of
forecast values for the X;s, share effects are often important.
Apart from their role in share effects, the basecase forecasts have many other
roles in the determination of policy deviations. For example: the forecast path for
motor vehicle imports is important in determining revenue losses from tariff cuts
and the required path for the replacement tax; the forecast path of technical change
Table 7.1. Tariff rates applying to elf. imports of

motor vehicles (MONASH commodity 70)
(1) (2)
Year Basecase forecasts Policy simulation
1997 19.13 19.13
1998 17.27 17.27
1999 15.41 15.41
2000 13.55 13.55
2001 11.69 11.69
2002 11.69 9.83

2003 11.69 7.97
2004 11.69 6.11
2005 to 2016 11.69 4.25
in the Motor vehicle industry is important in determining the effect of tariff cuts on
the overall rate of technical progress in the economy; and the forecast paths for
motor vehicle employment and output are important in determining the likely
adjustment costs from tariff cuts.
The remainder of this section is organized as follows. In subsection 7.1 we set out
the key assumptions underlying the policy simulation. Then subsection 7.2 presents
the results and explains them by the use of back-of-the-envelope algebra.
7.1. Key assumptions

(a) Labour market
We assume that real wage rates are sticky in the short run and flexible in the long
run. This means that cuts in motor vehicle tariffs can lead to changes in aggregate
employment in the short run. However in the long run we assume that real wage
rates adjust so that the tariff cuts have no effect on aggregate employment This
approach to the labour market is consistent with conventional macro-economic
modelling in which the NAIRU is exogenous.
Technically, our labour market assumptions for the current simulation were
implemented via (24.2) and (24.6), described in subsection 24.1.
Initially we conducted the policy simulation with F_W(t), the shift variable in
(24.2), set exogenously on zero in each year. Thus we assumed in each year that the

77

79

81

83
deviation in the real post-tax wage rate23 from its basecase forecast level increases
in proportion to the deviation in employment from its basecase forecast level. The
coefficient of proportionality (cti) was set so that the employment effects of shocks
to the economy are largely eliminated after 5 years. In other words, after about 5
years the benefits or costs of a shock, such as a reduction in the tariff on commodity
70, are realized almost entirely as an increase or decrease in real wage rates.
With F_W(t) exogenized on zero in every year, our initial MONASH policy
simulation produced negative effects on employment in the early years followed by
a cycle. Rather than asymptoting, aggregate employment in the policy simulation
passed through its basecase forecast path. This caused the movement in the gap
between the policy and basecase forecast real wage rate to change sign. Whereas in
the early years of the simulation, the gap became progressively more negative, in
later years it became progressively less negative. The cycles in the deviations
between the policy and basecase forecast levels of both employment and the real
wage rate were damped and of no practical importance. However, they are difficult
to explain and we have found that such cycles are a distraction for policy makers.
To maximize the probability that policy makers will absorb the main messages
from MONASH simulations, we often eliminate cycles by changing our labour-
market closure in mid-simulation. In the policy simulation reported in this section,
we assumed in each year after 2005 a halving in the gap between the policy level of
aggregate employment and its forecast level. As can be seen from Chart 7.1, this
ensured a cycle-free return of employment to its forecast path after the initial
negative effects of the cuts in motor vehicle tariffs. It also ensured a cycle-free
adjustment of real wage rates. As explained in subsection 24.1, we implement the
gap-halving approach by turning off (24.2) [endogenizing F_W(t)] and turning on
(24.6) [exogenizing F(t)].
(b) Public expenditure and taxes
We assume that reductions in the tariff on commodity 70 make no difference to the
path of real public consumption and that the Government imposes an additional
uniform consumer tax designed to replace the lost revenue.
There are several methods we could have adopted for calculating the rate of the
replacement consumption tax. For example, we could have assumed that the tax is
varied each year to prevent the policy shocks (the tariff cuts) from having any
effect on the path of Government revenue. Alternatively we could have assumed no
variation in the path of national saving or the path of real national wealth. However,
Alternatively, we could have conducted the policy simulation with the real pre-tax wage rate
responding to deviations in employment from its basecase forecast level. In the present simulation we
hold income tax rates on their forecast path. Thus percentage movements in real pre-tax and post-tax
wage rates are the same. In these circumstances the choice between pre- and post-tax wage response is
unimportant.
a problem with these methods is that they endow the Government with an
unrealistic level of foresight. They require the Government to take account in each
year not only of the direct effect of tariff cuts on revenue but also of second-round
effects arising from tariff-induced changes in employment, the exchange rate and
other variables.
The method we chose for calculating the rate of the replacement consumption tax
for the years of additional tariff cuts (2002 to 2005) takes account of two factors:
(a) the first-round or impact effect of the tariff cuts on Government revenue and (b)
the evolving effects of the tariff cuts on real national wealth (the value of
Australia's capital stock less net foreign liabilities, deflated by the price index for
investment goods). Assume for example that car imports are $10 billion, aggregate
consumption is $300 billion and that by year t-1 the tariff cuts have reduced real
national wealth (perhaps through their second-round effects on employment and
national saving in previous years) by $0.3 billion. Then via factor (a) we assume
that a reduction in the car tariff of 10 percentage points in year t will induce the
Government to increase the rate of consumer tax by 0.33 percentage points24. Via
factor (b) we assume that in year t the Government attempts to increase national
saving and thus restore real national wealth by imposing a consumption tax of 0.1
per cent (= 100*0.3/300). Thus in total, the additional consumption tax imposed in
year t is at the rate 0.43 per cent. As described in subsection 19.3, the calculation in
each year of the additional consumption tax ( f | ) is achieved through (19.12) with
the LHS set to reflect lagged changes in real national wealth.
In each year beyond 2005, we set f| exogenously at 0.0037, that is, we assumed
further small increases in the tax on consumption (see Chart 7.8). As explained
below, this was necessary to prevent real national wealth in the policy simulation
from declining continuously relative to its level in the basecase forecast
Rather than using a broad-based consumption tax to replace lost tariff revenue,
we could have used additional income tax. However replacement by an indirect tax
seems the more policy-relevant assumption because the current thrust of taxation
policy in Australia is towards indirect taxes and away from direct taxes.
(c) Private consumption
We assume that reductions in tariffs make no difference to the average propensity
to consume, that is we assume that tariff-induced percentage movements in private
consumption are determined by tariff-induced percentage movements in household
24
In MONASH the Motor vehicle industry is treated as if it pays the full tariff rate on all its imports of
automotive products. However via a by-law scheme much of these imports enter Australia dutyfree.We
capture the effects of the by-law allowances via a subsidy on motor vehicle production. Details are given
in subsection 19.3. As can be seen from (19.12), in calculating the revenue-replacing change in
consumption taxes we take account of tariff-induced changes in this subsidy.
disposable income. We do this by exogenizing Fc in (16.69), discussed in

subsection 16.1(o).
(d) Rates of return on capital
In policy simulations, MONASH allows short-run divergences in post-tax rates of
return on industry capital stocks from their levels in the basecase forecasts. Short-
run increases in post-tax rates of return cause increases in investment and capital
stocks, thereby gradually eroding the initial divergences in post-tax rates of return.
In the policy simulation described here, investment in each industry responds to
movements in post-tax rates of return calculated under static expectations. As
detailed in subsection 21.2, static expectations mean that investors make decisions
in year t on the assumption that capital rentals and asset prices will grow between
years t and t+1 at the current rate of inflation and that real interest rates and tax
rates will remain at their current levels. Thus we assume that investment decisions
in each industry in year t are heavily influenced by current profitability.
As an alternative to static expectations, MONASH allows for forward-looking
expectations, that is, it allows us to assume that investors in year t make decisions
on the assumption that capital rentals, asset prices, real interest rates and tax rates
will change between years t and t+1 in the way indicated by our model.
Implementation of the assumption of forward-looking expectations is
computationally difficult (subsection 21.3). In the absence of any strong empirical
evidence in favour of forward-looking expectations, we have in almost all
MONASH applications adopted the assumption of static expectations. Consistent
with this, we use static expectations in the simulation analysed here. Nevertheless,
we have computed a forward-looking MONASH solution for the effects of the
proposed tariff cuts. This shows significantly different short-run results from those
generated under static expectations. As is apparent from Charts 7.1 and 7.2, the
tariff cuts in combination with short-run stickiness in real wages have negative
short-run effects on employment and investment. With tariff cuts continuing over a
number of years, the negative effects in the early years are magnified under
forward-looking expectations. The bad news associated with tariff cuts in year t+1
has a depressing effect in year t. In the long run, consumption under forward-
looking expectations recovers to the same level as under static expectations and
long-run wealth is higher. However, the present value of the time path of deviations
in consumption and the terminal deviation in wealth is lower under forward-
looking expectations than under static expectations.
(e) Production technologies
In the policy simulation analysed in subsection 7.2, we assumed that the rates of
technical progress in production and capital creation in each industry are the same
as in the basecase forecast simulation. In another simulation discussed briefly in
subsection 8.1, we assumed that cuts in motor vehicle tariffs cause a reorganization
in the industry resulting in a technological improvement.
7.2. Results
(a) Macro effects: back-of-the-envelope model
In explaining the macro results we use a back-of-the-envelope (bote) model. In the
bote model we assume that the economy produces one good (grain) and imports
one good (vehicles). Grain production is via a constant-returns-to-scale production
function of capital and labour inputs. Grain and vehicles are both consumption and
investment goods. Units of consumption and investment are formed as Cobb-
Douglas functions of grain and vehicles leading to Cobb-Douglas unit-cost
functions. Finally, we assume that the costs per unit of employing capital and
labour equal the values to the employer of their marginal products. Under these
assumptions we have:
Pc=(PgTgc) a 8 C (PvTvc) a v c . (7.1)
Pi=(PgTgi) a g i (P v Tvi) a v i , (7.2)

W = PgMi , (7.3)
Q = PgMk , (7.4)
W r e al=W/P c , (7.5)
and
R = Q/Pi (7.6)
where:
Pg and Pv are the basic price of grain and the c.i.f. price of vehicles;
Pc and Pj are the purchasers' prices of a unit of consumption and a unit of
investment;
Tgc, Tv„ Tgi and T^ are the powers (one plus rates) of the taxes (including tariffs)
applying to consumption purchases of grain and vehicles and investment
purchases of grain and vehicles;
Q and W are factor payments, the rental rate and the wage rate;
M, and Mkare the marginal products of labour and capital;
Wreai is the real wage rate;
R is the rate of return on capital calculated as the rental or user price of capital
divided by the cost or asset price of a unit of capital; and
the a's are positive parameters reflecting the shares of grain and vehicles in
consumption and investment, such that agc + a^ = 1 and c^ + o^ = 1.
From these equations we find that

fp V*vc
Mi(£) = W * "v real *T C (7.7)
&J
and
M k ( ^ ) = R* P, *Tj (7.8)
p
g;
where Tc and T; are the average powers of the taxes on consumption and investment
defined by
T c =T g c a gc *T v c a vc and Tj =T g i a gi *T vi a vi . (7.9)
In (7.7) and (7.8), we emphasize that the marginal products are both functions of
K/L. M| is an increasing function of K/L and Mk is a decreasing function of K/L.
As explained in subsection 7.1(b), in our MONASH policy simulation we reduce
the tariff on vehicles and replace the lost revenue with a broad-based consumption
tax. In terms of the bote model, this has the effect of increasing the average power
of the tax on consumer goods (Tc) and reducing the average power of the tax on
investment goods (Tj).
In the short run, Wreai is sticky, that is it will adjust only slowly to eliminate
deviations between the policy and basecase forecast levels of employment (see
subsection 7.1(a)). With a cut in tariffs there is an increase in both imports and
exports leading to an increase in Pv relative to Pg, that is a decline in the terms of
trade.25 Thus from (7.7) we see that M|, and consequently K/L, will increase.
Because K moves slowly, there must be a short-run decrease in L. This is
confirmed in Chart 7.1 where we see employment moving below control (basecase
forecast) in the years of additional tariff cuts (2002 to 2005).
Looking now at (7.8), we ask what is the short-run impact of the additional tariff
cuts on the rate of return (R)? With an increase in K/L, Mk falls. As already
mentioned, Pv /Pg rises. Both these effects tend to reduce R. However T( falls and
this tends to increase R. Thus the effect on R is uncertain. In our MONASH
simulation R falls, and as can be seen in Chart 7.1, K edges downwards.
The short-run decrease in employment leads to reduced wage demands and W„,i
moves down [Chart 7.5 and subsection 7.1(a)]. Thus, after its initial increase, M|
moves back down towards control [see (7.7)]. This means that after its initial rise,
K/L must fall towards control. Because K edges down only slowly, the fall in K/L
towards control is accomplished mainly by an upward movement in L towards
*" We adopt the small country assumption for imports but we assume that increases in Australia's exports
reduce their world prices (see sections 18 and 27).
control. This can be seen in Chart 7.1 where L rises steadily from its trough
deviation in 2005. As explained in subsection 7.1(a), after 2005 the percentage gap
between L in the policy and forecast simulations is halved each year.
To allow the movement of L back to control, Wr(B| must continue to fall relative to
control for several years after the period of additional tariff cuts. Even after L has
approximately returned to control, there are further reductions in W ^ . These are
explained mainly by two factors: the continuing increase in taxes (Tc) on
consumption [subsection 7.1(b)] and the continuing decline in K (Chart 7.1).
Beyond 2010, reductions in W ^ offset these two factors allowing L to remain at
control. A slight positive influence on W ^ in the long run is a gradual recovery in
the terms of trade, that is a fall in Py/Pg, (see Chart 7.6 and the explanation in the
next paragraph). However this is insufficient to counteract the effects of the other
two factors.
What explains the movements in the terms of trade in Chart 7.6? In MONASH,
we treat Australia as a small country on the import side, that is we treat c.i.f. import
prices in foreign currency as exogenous. On the other hand, we recognize that
Australia has considerable shares of world markets for several relatively
homogeneous agricultural and mineral products, and that Australia exports
distinctive varieties of manufactured goods. Consequently, we assume that
expansions of Australia's exports reduce their world prices and generate a decline in
Australia's terms of trade. This means that the deviation path of, the terms of trade is
closely connected with the deviation path in aggregate exports. Looking at the
charts, we see that they give results for aggregate exports and other trade variables
consistent with those which would be expected in a model with a fixed balance of
trade: the additional tariff cuts stimulate imports (Chart 7.3); reduce the real
exchange rate (Chart 7.4); stimulate exports (Chart 7.3); and thereby reduce the
terms of trade (Chart 7.6). However, in our policy simulation the balance of trade is
variable. It tends towards surplus (requiring real devaluation with consequent
export expansion and terms-of-trade deterioration) when investment is weak. It
tends towards deficit (requiring real appreciation with consequent export
contraction and terms-of-trade improvement) when investment is strong. Thus,
beyond the period of additional tariff cuts, when the deviation in investment (Chart
7.2) is recovering from its trough level, the real exchange rate strengthens, exports
decline and the terms of trade improve.
At first glance the long-run behaviour of investment is difficult to reconcile with
that of the capital stock. By looking at Charts 7.2 and 7.1 we see that beyond 2007
investment is above control but that K (an asset-weighted index of industry capital
stocks) continues to move further below control. The explanation is a change in the
composition of the capital stock away from low-depreciation components. The cuts
in the motor vehicle tariff reduce consumption sharply relative to GDP. (This is
discussed below.) Consequently the tariff cuts cause a reduction in the housing
share of the nation's capital stock. Housing has a much lower rate of depreciation
than other components. Thus, the shift in the composition of the capital stock away
from housing increases replacement investment per unit of capital producing
apparently anomalous capital and investment results.
Another unexpected result is the long-run decline in the K/L ratio (Chart 7.1). On
the basis of (7.8) we thought initially that the additional tariff cuts would cause a
long-run decline in Mk with a corresponding increase in K/L. This is because R
eventually returns to control [subsection 7.1(d)] and we anticipated that the
permanent reduction in Tj would outweigh the effect of the deterioration in the
terms-of-trade (increase in Pv/Pg). In other words, we anticipated a long-run decline
in Q/Pg [see (7.4)]. On checking the MONASH results we found that the cuts in
motor vehicle tariffs do indeed cause a long-run decline in Q/Pg, so how can there
be a long-run decline in K/L?

The answer is found in sectoral effects in MONASH, not captured in a one-sector
bote model. In a multi-sectoral model, the economy-wide K/L ratio is affected not
only by Q/Pg, but also by the sectoral composition of GDP. K/L is decreased by
changes in the composition of GDP away from capital-intensive industries. This is
what happens in our tariff cut simulation: the negative effect on K/L of a
compositional shift away from capital-intensive industries outweighs the positive
effect of the decline in Q/Pg. The shift away from capital-intensive industries is
associated with the decline in consumption relative to GDP which causes a decline
in the output of housing services. Production of these services is highly capital
intensive.26 To check this sectoral explanation of the long-run decline in K/L we
repeated the tariff-cut simulation with consumption of housing services fixed on its
forecast path by an endogenous artificial shift in consumer preferences. In this
simulation the tariff cuts produced a long-run increase in K/L.
A final aspect of the capital path which we thought worth investigating is its
failure to stabilize. In Chart 7.1, K is continuing to move further below its forecast
path ten years after the end of the tariff cuts. We traced this to two factors: negative
share and quantity effects.
To understand these two factors we start with the equation
kj(t) = Xsj'(t)*k/(t) (7.10)
i
where
k J (t) is the rate of growth from year t-1 to year t in the aggregate capital stock
in simulation j , either the basecase forecast simulation (j = 0 o r the policy
simulation (j = p);
In MONASH housing services are produced by an industry consisting of the housing stock.
kjJ(t) is the rate of growth from year t-1 to year t in industry i's capital stock in
simulation j ; and
S|(t) is the share in simulation j of industry i in the aggregate capital stock for
yeart
Equation (7.10) implies that the difference between the growth rates in year t of
aggregate capital stock in the policy and forecast simulations is given by:
kP(t)-kf(t) = X[sr(t)-Sf(t)l*kfve(t) + Xrkf(t)>kf(t)l*Sfvc(t), (7.11)

where the superscript ave denotes the average of values in the policy and forecast
simulations.
We refer to the first term on the RHS of (7.11) as a share effect. Via the share
effect, the growth rate of aggregate capital in year t in the policy simulation can fall
short of that in the forecast simulation [ k p ( t ) - k (t)<0] if the policy reduces the
capital shares [ S P ( t ) - S j ( t ) < 0 ] of industries with fast rates of capital growth
[high k? v e (t)]. We refer to the second term on the RHS of (7.11) as a quantity
effect. The quantity effect will be negative if the policy reduces capital growth rates
[kf(t)-kf(t)<0].
In our tariff experiment both the share and quantity effects remain negative
through the simulation period. By calculating its industry components, we found
that the share effect is negative because the housing stock has a high growth rate in
our forecast simulation [giving a high value for k? v e (t), i = housing] and at the
same time the policy reduces the share of housing in the aggregate capital 'stock
[ S f ( t ) - s f ( t ) < 0 , i = housing].
In analyzing the quantity effect we looked at k P ( t ) - k [ (t) for all i. In most of

the losing industries (those adversely affected by the tariff cuts, e.g. housing),
kP(t)-kj (t) is negative in the early years and approaches zero from below in
later years, whereas in most of the winning industries (e.g. export-oriented mining),
kP(t) - kj (t) is positive in the early years and approaches zero from above in later
years. The quantity effect is negative because it happens that the approach of
kP(t) - kj (t) to zero is on average slightly slower for the losing industries than for
the winning industries.
Eventually, when kP(t) —kf (t) is close to zero for all i, the quantity effect makes
no further contribution to the gap between k (t) and k (t). However the share effect
can remain significantly negative indefinitely. Thus the policy can permanently
reduce the growth rate in aggregate capital causing an ever expanding negative
deviation of the type shown in Chart 7.1.
The remaining macro results requiring explanation are those for real GDP,
consumption and welfare.

(b) Decomposition of GDP deviations
Throughout the simulation, the GDP path in Chart 7.1 is high in relation to the
paths for capital and labour. For example, with labour contributing about 70 per
cent and capital about 30 per cent to GDP,27 the employment and capital deviations
in Chart 7.1 for 2005 contribute about -0.009 per cent ( = -0.0114x0.7 -
0.0029x0.3) to the GDP deviation. But the actual GDP deviation is +0.007 per cent.
That is, other factors including improvements in resource allocation (welfare
rectangles and triangles) associated with the additional tariff cuts made a positive
contribution of 0.016 per cent. In 2016, the employment and capital deviations
contribute -0.005 per cent (= 0*0.7 - 0.0164*0.3) to GDP but the actual GDP
deviation is 0.018 per cent. That is, other factors make a positive contribution of
0.023 per cent. What explains the size and movement in the contribution of these
other factors?
We answer this question in Chart 7.7 by decomposing the path of the GDP
deviations into various parts. The decomposition starts with the equation28:
gdpreal =
Sum[ (o,h), LABOCCIND, [LAB(o.h)/ GDP ]*employment(o.h) ]
+ Sum[ h, IND, [CAP(h)/ GDP ]*capital(h) ]
+ [TARIFF(70)/ GDP]*impvol(70)
+ Sum[ h, ALLFLOWS, [FLOW(h)/ GDP]*tc(h) ]
+ Sum[ h, ALLFLOWS, [TAX(h)/ GDP]*realfiow(h) ] . (7.12)
^'These are approximate shares in GDP at factor cost. In a two-input (labour and capital) analysis itis
reasonable to assume that GDP moves in line with GDP at factor cost and to use shares in GDP at factor
cost in calculations of factor contributions to GDP growth.
Sum[ h, SET, X(h)] is the sum of all X(h) for h in SET.

In (7.12) the variables denoted by lowercase symbols are percentage changes

between years t-1 andt:
gdpreal is the percentage change in real GDPfromyear t-1 to t;
employment(o,h) is the percentage change in employment in occupation o in
industry h;
capital(h) is the percentage change in the quantity of capital in industry h;
impvol(70) is the percentage change in the volume of imports of commodity 70
(motor vehicles);
tc(h) is the percentage change in the technology variable associated with flow
h;29and
realflow(h) is the percentage change in the quantity of flow h.30

The uppercase symbols are levels of variables:
GDP is the level of nominal GDP;
LAB(o,h) is the level of payments to labour by occupation and industry;
CAP(h) is the level of payments (rentals) to capital by industry;
TARIFF(70) is the level of tariff collection on imports of commodity 70;
FLOW(h) is the value of flow h; and
TAX(h) is the collection of indirect taxes on flow h including tariffs apart from
the tariff on commodity 70.
The italic symbols refer to sets:
COM is the set of all commodities;
LABOCCIND is the set of all occupation/industry categories of employment;
JND is the set of all industries; and
ALLFLOWS is the set of all commodity and primary-factor flows.
To reduce linearization error in (7.12), we use mid-point percentage changes. For
example, gdpreal is g/(l+gt/200) where gt is the percentage growth in real GDP
between years t-1 and t with real GDP in year t-1 as the base. Similarly, the
coefficients (GDP, LAB, etc.) are averages of values in years t-1 and t.
With the paths of GDP defined by (7.12), we demonstrate in section 9 that the
percentage gap (DevT) in year T between GDP in a policy simulation and GDP in a
forecast simulation is given by:
29
tc(h) refers to intermediate- or primary-factor-saving technical change in any industry; input-saving
technical change in the creation of units of industry capital; and changes in the use of margins services
per unit of sales.
30
Flow h refers to intermediate-input and primary-factor flows to industries; and commodity flows to
investment, exports, households, inventories and government.
F F
Dev T = ]TQCONTiT + £SHCONT iT C7-^)
i=l i=l
where
T
QCONTiT = X S Tti(v t P i-v[i) . (7-14)
t=l
T 0
SHC0NTiT = 2 v t i ( S ? t i - S ^ ) , (7.15)
t=l
STti=(Sî+S^)/2 and (7.16)
vti=(vg+vJj)/2 . (7.17)
vjlj and vjj , i=l,...,F, refer to percentage changes between years t-1 and t in
the policy and forecast simulations in the quantity variables on the RHS of
(7.12). For example, vjj for i = employment(o,h) is the mid-point percentage
change in employment(o,h) between t-1 and t in the policy simulation.
Sip . and st*. are modified ratios (as defined in section 9) between years t-1 and
t of the value of i to GDP in the policy and forecast simulations. For example,
SEvj for i = employment(o,h) refers to a modified share in GDP of occupation o
in industry h (LAB(o,h)/GDP) between years t-1 and t in the policy simulation.
We interpret QCONTiT as the contribution to Devr of the differences between the
policy and forecast simulations in the growth rates of the quantity of factor i, and
we interpret SHCONTVr as the contribution of the differences between the two
simulations in the GDP shares of factor i. Thus, we see that Devi will be positive if
• growth in quantity variables is rapid in the policy simulation compared with
the forecast simulation (that is, the vjijs are large relative to the vjjS, giving
positive quantity effects); and
• the modified GDP shares of fast growing quantity variables are greater in the
policy simulation than in the forecast simulation (that is, as we vary i, positive
D* f*
values for S£t- - Syt- correspond to large values for Vti, giving positive share
effects).
In the context of the MONASH model, (7.13) decomposes real GDP deviations
into many thousands of components. In Chart 7.7 we have aggregated these
components into six parts. The first two are the contributions of labour and capital,
with quantity and share effects combined. For example, the labour line in Chart 7.7
shows aggregations for each year T of QCONTYr plus SHCONTiT over all i where i
is a category of labour. The third and fourth parts shown in Chart 7.7 are the
quantity effects of the tariff on commodity 70 and the quantity effects of other
taxes. All share effects associated with indirect taxes including tariffs are
aggregated and form the fifth part of our decomposition of the GDP deviations. The
sixth part is technical change. It is an aggregation of QCONTn- plus SHCONT;T
over all technical change variables. The six parts add to the GDP deviations
represented by shaded columns.
The labour line in Chart 7.7 in the early years has approximately the same shape
as the aggregate employment line in Chart 7.1. In the early years the contribution of
employment to the GDP deviation is approximately 0.61 times31 the percentage
deviation in employment. In the later years this relationship breaks down. In these
years the employment deviation is zero but the labour contribution to the GDP
deviation grows continuously. To understand how this happens we return to the
first term on the RHS of (7.12), the contribution (CL) of labour to GDP growth,
and rewrite it as
CLj=SJ*/j (7.18)
where superscript j refers to simulation j and
S} = AGGLABVGDP, that is S\ is the labour share in GDP, and
/>=Sum[(o,h), LABOCCIND, [LABj(o,h)/AGGLABj ]*employmeirfCch)], that is P
is the rate of growth of aggregate employment calculated as a cost-share-
weighted average of the rates of growth of employment in individual labour
categories. Deviation results for this variable are almost identical to those
shown in Chart 7.1 for the measure of aggregate employment (an hours
measure) used in our policy and forecast simulations.
From (7.18) we find that the labour contribution to the difference between GDP
growth in the policy and forecast simulations in any year is given by
CLP-CL f = ( S / p - S / f ) * / a v e + ( / P - / f ) * S ^ v e . (7.19)
In the later years of our tariff-cut experiment, with the policy-induced deviation in
aggregate employment constant (at zero), P and f are the same. Thus the second
term on the RHS of (7.19) is zero. However, the first term is positive. In both the
policy and forecast simulations aggregate employment growth is positive (fw >0),
and in the policy simulation the labour share of GDP is greater than in the forecast
The share of labour in GDP (at market prices) is approximately 0.61.

simulation ( S? - S, >0) reflecting the policy-induced reduction in the capital stock.

With CLP - CLf greater than zero in each of the later years of the simulation period,
labour makes an ever-increasing contribution to the difference between the levels
of GDP in the policy and forecast simulations. This is despite employment having
returned in the policy simulation to its level in the forecast simulation. More
generally, real GDP is a Divisia (continuously varying weight) index of movements
in factor, technology and tax variables. Even if each of these variables ends up at
the same point in a policy simulation as in the forecasts, real GDP need not end up
at the same point. The deviation in real GDP depends on the paths of the deviations
in the constituent variables not just their endpoints.
In the early years of the simulation period, the contribution line for capital in
Chart 7.7 can be approximated by multiplying the deviation line for the quantity of
capital (real asset value) in Chart 7.1 by 0.3 (the approximate capital share in GDP).
In later years, the capital contribution in Chart 7.7 looks too high relative the
quantity deviation in Chart 7.1. For example, the capital contribution in 2016 is
-0.0035 per cent whereas the capital quantity deviation multiplied by 0.3 is
-0.0049 (= -0.0164*0.3). On investigation we found that the capital contribution is
subject to two forces in addition to the deviation in the quantity of capital. The first
is a policy-induced change in the industrial composition of the capital stock. The
tariff cuts increase the share in the capital stock of export-oriented mining and
reduce the shares of motor vehicles and housing. In the MONASH database mining
has high rentals per unit of capital (high rates of return) and motor vehicles and
housing have low rentals per unit of capital. Thus the policy-induced shift in the
composition of the capital stock tends to increase aggregate rentals and
consequently the contribution of capital to GDP. The second force partially offsets
the first during the simulation period and would eventually overwhelm the first if
the simulation period were extended. This force arises from a term for capital
analogous to the first term on the RHS of (7.19). In the case of capital, S£ - s £ is
negative (reflecting the policy-induced reduction in the capital stock) and kave is
positive (reflecting capital growth in our forecasts). Thus throughout the simulation
period, (S£ - s £ ) * k a v e is negative, imparting an ever-growing negative effect to
the capital contribution to the deviation in GDP.
The tariff(70) contribution in Chart 7.7 covers the familiar triangles and
rectangles from partial equilibrium welfare economics, see Figure 7.1. In each year
T, the tariff(70) contribution is given approximately by
QCONTtarif^yo),! = (TAR70f>T + 0.5 * DIFTAR70T ) * SM70T * m70T (7.20)
where
TAR70f#T is the forecast tariff rate for motor vehicles in year T;
DIFTAR70T is the difference between the tariff rate for motor vehicles in the
policy simulation and the basecase forecasts;
SM70T is the share of imported motor vehicles in GDP (an average between the
forecast and policy simulations); and
m70T is the percentage increase in imports of motor vehicles caused by the
additional tariff reductions.
From our projections for 2005, we find that SM70T is about 0.021 and from our
results for the motor vehicle industry (to be discussed below) we find that m70t is
5.50 per cent Table 7.1 indicates that TAR70f>T is 0.1169 and DIFTAR70T is
-0.0744. Thus (7.20) gives a tariff contribution in 2005 of about 0.01. This is
consistent with the tariff contribution for 2005 shown in Chart 7.7. SM70T and
m70r fall slightly in the period beyond 2005 giving a gradual decline in the
tariff(70) contributions to the GDP deviations.
Chart 7.7 includes a line showing the quantity contributions to the GDP
deviations of indirect taxes and tariffs other than the tariff on commodity 70.
Because most of these taxes are on consumption, our initial expectation was that
the other-tax contribution line in Chart 7.7 would have a similar shape to the
deviation path for consumption in Chart 7.2. In working out why the other-tax
contribution follows an up and down path in Chart 7.7, we found it helpful to
disaggregate this contribution into two parts: the contribution from sales taxes on
motor vehicles and the contribution from other indirect taxes. We found that the
second part follows approximately the expected shape. The positive and increasing
other-tax contribution in the early years in Chart 7.7 is explained by the first part.
During the period of tariff cuts there is stimulation of motor vehicle sales to
households. These sales' are heavily taxed. Stimulation of this heavily taxed
component of consumption is sufficient to cause the upward movement in the
other-tax line in Chart 7.7, despite the downward path in aggregate consumption.
In later years, when there is no further stimulation of motor vehicle demand, the
continuing downward movement in the path of aggregate consumption causes the
downward movement in the other-tax contribution.
For the purpose of understanding the tax share effects in Chart 7.7, it is
reasonable to approximate them by
T T
Tax share effect(T)« £ ( S ? c - s f c ) * c t + j ^ - S ^ * ! ^ (7-21)
t=l t=l
An illustrative application q/MONASH 99
where
Sfc and sfc are the shares in the policy and forecast simulations of
consumption taxes in GDP in year t,
Sjjjj and Stm are the shares in the policy and forecast simulations of motor
vehicle tariff revenue in GDP in year t,
c, is the growth rate, averaged between the policy and forecast simulations, in
real consumption from year t-1 to year t, and
m, is the growth rate, averaged between the policy and forecast simulations, in
the volume of motor vehicle imports from year t-1 to year t.
In (7.21) we ignore the minor modifications discussed in section 9 in the definitions
used in the decomposition formulas of shares (the S*s). We also ignore changes in
the composition of aggregate consumption. Because we assume that lost tariff
revenue is approximately replaced by the imposition of a broad-based consumption
tax [see subsection 7.1(b)],
Sfc + S£n - S tc + SS„ . (7.22)

Using (7.22) we can write (7.21) as
T
Tax share effect(T) - ] > > K n - s t m ) * ( r n t - c t ) . (7.23)
t=l
With tariff cuts, SJ^ is less than SJm for all t in the simulation period. Except for a
few years in the early part of the simulation period, our forecasts growth rates for
motor vehicle imports are larger than those for real consumption. This means that
for most years, including all of the later years in the simulation period, mt is greater
than Ct. Thus, consistent with the tax share line in Chart 7.7, (7.23) implies that the
tax share effect becomes increasingly negative as T increases.
The final part of the decomposition of the GDP deviations is the technical change
effect. As we will see in subsections 7.2(e) and (f), the additional cuts in motor
vehicle tariffs cause a reduction in the share of the economy's resources (capital
and labour) devoted to motor vehicle production and an increase in the share
devoted to export-oriented activities. In our forecasts, these latter activities have
more rapid rates of technical progress than motor vehicle production. Thus,
although we assume no difference between the forecast and deviation simulations
in the rates of technical progress in each industry [subsection 7.1(e)], the overall
rate of technical progress in the policy simulation is greater than that in the forecast
simulation. This gives an increasingly positive technical change contribution in
Chart 7.7.
In comparative static models and models with basecase forecasts exhibiting
balanced growth, GDP deviations induced by policy changes can be explained
largely in terms of welfare triangles and rectangles and changes in employment and
capital usage. Such an explanation is inadequate for MONASH policy simulations.
Our analysis of Chart 7.7 reveals that GDP deviations in these simulations depend
critically on the details of the realistic (unbalanced) basecase forecasts. We found
that the GDP deviations are affected by, among other things:
• the forecast growth rates for aggregate inputs of labour and capital;
• the forecasts for the paths of tariff rates and sales taxes on motor vehicles and
of the share of imported motor vehicles in GDP;
• the forecast growth in imports of motor vehicles relative to that of aggregate
consumption; and
• the forecast rate of technological progress in motor vehicle production relative
to that in export-oriented industries.
(c) Consumption
As can be seen from Chart 7.2, the cuts in motor vehicle tariffs reduce household
real consumption by about 0.05 per cent in 2005. The real consumption deviation
becomes gradually more negative over the next decade reaching -0.06 per cent in
2016.
The reduction in real consumption of 0.06 per cent in 2016 can be explained as
the outcome of three negatives and one positive.
The first negative is the decline in the terms of trade of 0.16 per cent (Chart 7.6).
In our forecasts, the shares of exports and imports in GDP in 2016 are both 0.31.
Thus a decline in the terms of trade of 0.16 per cent is equivalent to a loss in GDP
of 0.050 per cent. With household consumption representing 60 per cent of GDP,
this translates into a loss of real consumption of 0.083 per cent.
The second negative is a reduction in real national wealth. In Chart 7.9, the
deviation path in real national wealth is u-shaped. As explained in subsection
7.1(b), we assume that the government initially takes account only of first-round
effects in setting the rate of consumption tax to replace revenue lost from the tariff
cuts. Because the second-round effects (reductions in employment and
consumption) are negative, the government's replacement tax is inadequate.
Consequently, the government's budget moves towards deficit, with negative
effects on real national saving and wealth. These are reinforced by reductions in
household disposable income leading to reductions in household saving. After the
period of tariff cuts we assume that there are continuing increases in consumption
taxes (Chart 7.8) designed to prevent the negative deviation in real national wealth
from continuously expanding. In our simulation the decline in real national wealth
is arrested in 2012. Beyond 2012, real national wealth in the policy simulation
moves back towards its forecast level. Nevertheless in 2016 there is still a negative
deviation in real national wealth of 0.016 per cent. This reduces the net (post-
depreciation) incomes of Australians by about 0.0018 per cent (= 0.016*0.11, real
national wealth, mainly physical capital, produces 11 per cent of Australia's net
income). The effect on real consumption is a loss of 0.003 per cent (=0.0018/0.6).
The final negative effect is growth in real national wealth. From 2016 to 2017, the
deviation in real national wealth rises from -0.0162 per cent to -0.0138 per cent.
National wealth is about 2.3 times GDP or about 3.8 times consumption. Thus the
accumulation of an extra 0.0024 per cent of real national wealth in the policy
simulation relative to the basecase forecast causes a negative deviation in real
consumption in 2016 of 0.009 per cent (= 0.0024*3.8).
The positive influence on consumption is growth in real GDP net of the
contribution of capital which we have already accounted for in real national wealth.
In 2016 the deviations in real GDP and capital are 0.018 and -0.016 per cent (Chart
7.1). With the gross return to capital in 2016 being about 27 per cent of GDP, the
deviation in the non-capital component of real GDP is about 0.031 per cent [=
(0.018+0.27*0.016)/0.73]. This generates a real consumption gain of about 0.037
per cent (= 0.73*0.031/0.6).
In summary, our back-of-the-envelope calculations for the real consumption
deviation in 2016 are:
Terms of trade effect = -0.083

Wealth effect = -0.003
Wealth growth effect = -0.009
Non-capital GDP effect = 0.037
Total -0.058
The total of-0.058 per cent is very close to the MONASH result in Chart 7.2. The
bote calculations show that the dominant effect on real consumption is the
deterioration in the terms of trade. In our forecasts the share of exports and imports
in GDP grows steadily throughout the simulation period causing growth in the
terms-of-trade effect. This explains the negative slope in the real consumption path
in Chart 7.2 beyond 2005.

(d) Welfare
With real public consumption held on its forecast path [subsection 7.1(b)], it is
tempting to use the deviation in real household consumption (Chart 7.2) to indicate
the welfare effect in each year of the policy change. However, the MONASH
variable (CR) for real consumption is a Divisia index of percentage movements in
the consumption of individual commodities. Thus the deviation in CR for year T
from its forecast value caused by the additional tariff cuts could be non-zero in
circumstances where there is no deviation in the consumption of any individual
commodity.32 This implies that deviations in CR may not accurately indicate
deviations in household welfare.
For measuring the welfare effects of the tariff cuts in any year, we rely on
Paasche and Laspeyres cost differences.33 The Paasche cost difference is the
amount of money that could be taken away from households in the policy situation,
leaving them with just sufficient money to buy the basecase forecast bundle of
commodities, and the Laspeyres cost difference is the amount of money that must
be given to households in the forecast situation so that they can just buy the policy
bundle of commodities. The Paasche cost difference is a lower bound on the
32
Assume for example that the forecast share of good i in the household budget trends downfromyear
1 to year T. Assume also that the policy growth rate in the consumption of i is greater than the forecast
growth rate ( xj* - x[ > 0) in the early years of the policy simulation but the opposite is true in the later
years ( xf - x[ < 0). Then the deviation in CR in year T could be positive even if the consumption of all
goods in year T in the policy simulation is the same as in the forecast simulation. This is because in the
calculation of the deviation in CR, the positive values of x? - x[ in the early years receive higher
weights than the negative values in later years.
33
These cost differences as measures of welfare were proposed by Machlup (1957), see also Ng (1979,
p.89). Calculations of cost differences are performed via equations described in subsection 24.2.
compensating variation measure of welfare change. This is because consumers

faced with the policy prices and with income sufficient to buy the basecase forecast
bundle could achieve at least the basecase level of welfare. The Laspeyres cost
difference is an upper bound on the equivalent variation measure of welfare change
because consumers faced with the forecast prices and with income sufficient to buy
the policy bundle could achieve at least the policy level of welfare.
In Chart 7.10 we show the path of the Paasche cost differences for our tariff-cut
experiment expressed as percentages of the policy levels of household expenditure
and the path of the Laspeyres cost differences expressed as percentages of the
forecast levels of household expenditure. As can be seen from the chart, there is
little difference between these cost-difference paths and between them and the
deviation path for CR. In the present application of MONASH, any of these three
measures is an adequate indicator of welfare change.
(e) Results for the motor vehicle industry
For 2016, Chart 7.11 shows that sales in Australia of domestically produced
commodity 70 are 7.38 per cent below their forecast level and imports are 5.41 per
cent above their forecast level. This is an 12.1 per cent reduction in the domestic-
to-import sales ratio (=100*[(l-0.0738)/(l+0.0541) - 1]). It arises from a change in
relative prices and from a change in the composition of demand for commodity 70.
The direct effect of the tariff cuts is to reduce the landed-duty-paid price (that is
the basic price) of imports of commodity 70 by 6.66 per cent [=100*(1.0425/1.1169
- 1), see the tariff rates in the last row of Table 7.1]. The MONASH results indicate
that there is little effect on the exchange rate and that in 2016 the basic price of
imported commodity 70 (PM70) in the policy simulation is 6.57 per cent below its
basecase forecast value. With a lower import price, there is a reduction of 2.25 per
cent in the basic price of domestic commodity 70 (PD70). The main reason is that
the domestic industry benefits from a reduction in the cost of one of its principal
inputs, namely imported automotive parts34. Bringing the results for PM70 and
PD70 together, we find that by 2016 the tariff cuts reduce PM70/PD70 by 4.4 per
cent (=100*[(l-.0657)/(l-.0225) - 1]). However, for households, basic prices of
domestic and imported cars are only about half of purchasers' prices. Sales taxes
account for about 22 per cent and margins (e.g. retail, wholesale and transport
costs) make up the rest. Because we assume that margin costs are determined
independently of basic prices, the effect of tariff cuts on the import/domestic ratio
of purchasers' prices to households is much less than their effect on the
import/domestic ratio of basic prices. In the policy simulation, the reduction in the
34
Against this, the industry loses much of its by-law subsidy (discussed in footnote 24). Because the
automotive imports by the industry exceed the level allowed under the by-law, the cost-reducing effects
on the industry of reductions in automotive tariffs outweigh the cost-increasing effects of the erosion of
the by-law subsidy.
import/domestic ratio of purchasers' prices to households is only 3.17 per cent. The
substitution elasticity between imported and domestic cars in MONASH is 5.2.
Thus we find in our results for 2016 (Chart 7.12), a reduction for households of 15
per cent (=100*(1-(1-0.0317)52)) in their ratio of purchases of domestic to imported
motor vehicles.35 For other users of commodity 70, margins play a less important
role and the reductions in their import/domestic purchasers' price ratios are closer
to 4.4 per cent. This gives reductions in their domestic/import mixes of close to 21
per cent (=100*(l-(l-0.044)52)). Nevertheless, the reduction in the overall
domestic/import sales ratio is limited to 12.1 per cent This reflects an import-
reducing change in the composition of demand for commodity 70. The Motor
vehicle industry (which suffers a negative output deviation and is a major user of
commodity 70) is a much more intensive user of imported 70 than are other users of
commodity 70 (e.g. households).
Having explained the deviation result for the domestic/import ratio in the sales of
70, we now turn to the deviation result for total domestic sales (domestically
produced plus imported). In Chart 7.11, total domestic sales in 2016 are almost
unaffected by the tariff cuts. Lower prices for commodity 70 increase the demands
of households (Chart 7.12) and most other users. However, as already mentioned,
one of the main users of commodity 70 is the motor vehicle industry. Contraction
of this industry is sufficient to cause the path of total domestic sales of 70 in the
deviation simulation to lie slightly below the forecast path for much of the
simulation period.
From the deviation results for total domestic sales and their domestic/import mix,
we can provide an approximate explanation of the results for domestic sales of the
domestic commodity and for import sales. We use the equations:
%dev(total domestic sales) = Shra* %dev(imports) + Shj* %dev(domestic)
(7.24)
and
" 1 + %dev(domestic) /100
100< = -12.1 , (7.25)
1 + %dev(imports) /100
where Shra and Shd are import and domestic shares of total domestic sales. In our
forecasts for 2016 these two shares have the values 0.51 and 0.49. With the
percentage deviation in total domestic sales being approximately zero, (7.24) and
(7.25) imply that:
35
One curious aspect of Chart 7.12 is the positive slope of the deviation path for overall household
consumption of motor vehicles beyond 2005 despite the negative slopes of the deviation paths of
household consumption for both domestic and imported motor vehicles. We return to this at the end of
our discussion of the motor vehicle results.
%dev(domestic) =-6.6 and %dev(imports) = 6.3

These are reasonably close to the values (-7.4 and 5.4) in Chart 7.11. The
discrepancies arise because (7.24) is only an approximation to the MONASH
calculations. In MONASH, %dev(total domestic sales) reflects not only quantity
deviations of the type appearing on the RHS of (7.24) but also share effects. These
share effects are similar to those discussed in subsection 7.2(b). In the case of car
sales, the share effects reflect differences in the forecast rates of growth of sales of
imported and domestic cars and differences between the policy and forecast
simulations in the import/domestic shares of total sales.
The output deviation path for commodity 70 in Chart 7.11 lies above the
deviation path for domestic sales of the domestic commodity. This is because
exports of 70 are not reduced by tariff cuts.

Charts 7.13 and 7.14 show paths for the levels of motor vehicle output and
employment in the basecase forecasts and in the policy simulation. During the
period of the additional tariff cuts (2001 to 2005), output is projected to have very
slight negative growth in the policy simulation. Employment is projected to decline
more sharply, by about 2.4 per cent a year. Because the motor vehicle industry is
subject to high rates of staff turnover, employment decline at this rate would be
unlikely to cause adjustment problems.
We conclude our discussion of the motor vehicle results by considering the
curious aspect of Chart 7.12 noted in footnote 35: the positive slope beyond 2005
of the deviation path for overall household consumption of motor vehicles despite
the negative slopes of the deviation paths of consumption of domestic and imported
motor vehicles.
In simplified notation the percentage growth rates (c70J(dom,t) and c70J(imp,t)) in
year t in household demands for domestic and imported cars in simulation
j (j = p or f, policy or forecast) are given by
c70j(dom,t) = a»(t) - SH70j(imp,t)*twist (7.26)
and
c70j(imp,t) = a>(t) + SH70i(dom,t)*twist (7.27)
where
SH705(dom,t) and SH70)(imp,t) are the shares of household expenditure on cars
accounted for by the domestic and imported products in simulation j ;
aJ(t) is a variable summarizing income, population and price factors affecting
household demand for cars in simulation j ; and
twist is a preference variable affecting consumer choice between imported and
domestic cars. It imposes a change in the import/domestic mix of car purchases
without affecting overall growth [c70J(t)] in household demand for cars where
overall growth is given by
c70i(t) = SH70j(dom,t)*c70i(dom,t) + SH70j(imp,t)*c70i(imp,t)= j(t) . (7.28)
In (7.26) and (7.27) we have assumed that the same summary variable a*(t) applies
to the consumption of both domestic and imported cars. This is representative of
the forecast and policy simulations beyond the period of tariff cuts. Broadly
consistent with the MONASH simulations, the preference variable, twist, has no
superscripts or subscripts, i.e. we assume that it is the same in both simulations and
for all years.
From (7.26) to (7.28) we obtain the growth rates along the deviation paths for
domestic, imported and overall household consumption of cars as
c70p(dom,t) - c70f(dom,t)
= ap(t) - af(t) - (SH70p(imp,t) -SH70f(imp,t))*twist , (7.29)
p f
c70 (imp,t) - c70 (imp,t)
= ap(t) - af(t) + (SH70p(dom,t) - SH70f(dom,t))*twist (7.30)
and
c70p(t) - c70f(t) = ap(t) - af(t). (7.31)
Beyond the period of additional tariff cuts, ap(t) is very similar to af(t) but slightly
more positive. This explains the slight positive slope in Chart 7.12 in the deviation
path for overall household consumption of motor vehicles. As mentioned in
subsection 6.1, in our forecasts we assume a continuing preference twist in favour
of imported cars. Hence twist is positive. Additional cuts in tariffs mean that
SH70p(imp,t) - SH70f(imp,t) > 0 and SH70p(dom,t) - SH70f(dom,t) < 0 . 36
Therefore the twist terms make negative contributions on the RHSs of both (7.29)
and (7.30). These are sufficient to produce negative slopes in the deviation paths of
household demands for domestic and imported cars beyond 2005.
(f) Results for other industries
Fifteen of the 112 MONASH industries have output deviations of more than 0.4 per
cent by the end of the simulation period. They are all trade-exposed industries and
benefit from real devaluation (Chart 7.4). Most export more than 60 per cent of
their output and/or face considerable competition from imports. Deviation paths for
a selection of the winning industries are shown in Chart 7.15.
Apart from Motor vehicles, only twelve industries have output deviations in 2016
of less than -0.06 per cent The deviation results for a selection of these industries
are shown in Chart 7.16. The negative outcome for rubber products in Chart 7.16 is
mainly due to its dependence on sales to the Motor vehicle industry. The remaining
industries in Chart 7.16 suffer negative output deviations because of overall
contraction in consumption and substitution in the consumption bundle towards
motor vehicles.
36
The high substitution elasticity between imported and domestic cars, 5.2, ensures that SH70(imp) rises
and SH70(dom) falls in response to a decline in the import/domestic price ratio.
8. Policy implications and concluding remarks
8.1. Policy implications

In section 7 we reported results from a MONASH policy simulation on the effects
of implementing an Industry Commission recommendation to lower the tariff on
motor vehicles from 12 per cent in 2001 to 4 per cent in 2005. We found that some
industries would be winners from cuts in motor vehicle tariffs and others would be
losers. However, for all industries apart from Motor vehicles the effects would be
very small. Even for the Motor vehicle industry we found that the proposed tariff
changes would have only minor effects on output and employment This is
consistent with the decomposition analysis in section 5. There we found that several
other factors have been at least as important as tariff cuts in determining the growth
performance of the Australian Motor vehicle industry since the mid 1980s. These
other factors include: movements in world commodity prices; twists in consumer
preferences between imported and domestic cars; Australia's macroeconomic
performance; and the rate of technical progress in the Australian Motor vehicle
industry relative to that in other industries. At the macro level, our analysis in
section 7 indicated that the proposed tariff cuts would have small negative effects.
After a vigorous public debate during 1997, the Australian government rejected
the Commission's recommendation. Superficially, the results in section 7 appear to
support the government's decision. However, there is an alternative interpretation.
The policy simulation reported here was made under two pessimistic
assumptions: (a) that further tariff cuts will not generate efficiency gains in the
Motor vehicle industry; and (b) that these cuts will not affect Australia's access to
foreign markets.
In a separate policy simulation reported elsewhere37, we assumed, contrary to (a),
that the motor vehicle industry would reduce its unit costs by an amount sufficient
to match the reductions in the price of imported vehicles caused by additional tariff
cuts. The industry in Australia is far from competitive. It is dominated by four
vehicle assemblers producing slightly differentiated products with production runs
that are short by world standards. It is reasonable to suppose that further tariff cuts
could eliminate one of the assemblers leading to expanded production runs and
lower unit costs for the remaining three. Under this assumption, the MONASH
policy simulation showed that the proposed tariff cuts would generate significant
macroeconomic benefits.
Contrary to (b), Australia's tariff policy may influence the policies of its trading
partners. Prior to the 1997 debate on motor vehicle tariffs, Australia had announced
See Dixon et a!. (1997a).

in APEC its intention to cut tariffs on all manufactured imports to 5 per cent of
their f.o.b. value (about 4 per cent in MONASH simulations which use c.i.f. prices)
by 2005. It appears that Australia will meet this target for the bulk of its
manufactured imports. The decision not to meet the target for motor vehicles, and a
similar decision made subsequently for textiles, clothing and footwear (TCF), will
reduce Australia's ability to argue effectively for tariff cuts by other countries both
inside and outside APEC. Terms-of-trade effects were responsible for a large part
of the negative macro results obtained in subsection 7.2. These effects would be
eliminated or even reversed by a slightly faster rate of world trade liberalization. If
by carrying out its original intentions Australia induced its trading partners to
accelerate their tariff reforms, then the proposed cuts in motor vehicle and TCF
tariffs could improve Australia's terms of trade. In these circumstances, a

MONASH policy simulation would show a favourable macroeconomic outcome
from further tariff cuts.
Thus, we interpret the results in section 7 as being supportive of the proposed
tariff cuts for motor vehicles. Under pessimistic assumptions, they indicate that the
cuts would have had very minor macroeconomic costs and would have caused little
if any microeconomic disruption. Even if the proposed tariff cuts were
implemented, our forecasts in section 6 show that the Australian motor vehicle
industry would be likely to experience positive output growth over the period 2001
to 2016. Without the tariff cuts, motor vehicle employment is likely to decline
slowly. In our basecase forecasts, motor vehicle employment falls by about 0.6 per
cent a year from 2001 to 2005. With the tariff cuts, the likely rate of decline over
these four years is about 2 per cent a year. Employment decline at this rate can be
handled by natural employment turnover in the industry. Thus we think that there
was almost no chance of a significant negative outcome from implementing the
proposed tariff cuts. On the other hand, there was a reasonable chance of a
significant positive outcome. This could arise from efficiency gains in the motor
vehicle industry and from more open policies by Australia's trading partners.
8.2. Back-of-the-envelope calculations

Throughout this chapter we have supported our results by detailed back-of-the-
envelope (bote) calculations. Such calculations have been important in our
development, understanding and application of MONASH and earlier models.
There are five reasons for our emphasis on bote calculations.
• Bote calculations are a necessary check for data handling and other coding
errors.
• Bote calculations reveal result-affecting theoretical shortcomings. For example,
such calculations applied to earlier versions of the policy simulation reported
in section 7 revealed that the results were heavily influenced by the treatment
in the model of miscellaneous costs (e.g. the costs of holding inventories). In
effect, these minor costs were treated (unrealistically) as production taxes.

Because miscellaneous costs were relatively large for some export-oriented
industries, this treatment led to an overestimation of the welfare gains from
export-stimulating tariff cuts. This problem was avoided in the results
eventually reported in section 7 by recognizing in the model that miscellaneous
costs involve resource usage.
• Bote calculations allow us to identify the principal mechanisms and data items
underlying particular results. Such identification is a necessary part of
explaining the results to business decision makers and policy advisers. We
cannot expect these people to be familiar with the details of a large model such
as MONASH, and we should not expect them to accept the results on a black-
box basis.
• Bote calculations are an effective form of sensitivity analysis. Clients are often
concerned about sensitivity issues. One approach to attempting to answer their
questions is repeated simulations.38 We have found it more informative to use
well-designed bote models. These allow us to help clients to assess the
reasonableness of our results and to work out how these results would be
affected by alternative assumptions and parameter values.
• Bote calculations generate theoretical insights. The CGE models of the last
thirty five years have provided numerical illustrations of well understood
theoretical propositions. For example, they have illustrated the proposition that
pollution problems are better tackled by tradeable emission permits than by
mandated targets. Illustration and quantification of existing propositions is a
valuable role for CGE models. However CGE models can incorporate detailed
structural and dynamic information well beyond that in purely theoretical
analyses. Thus we would expect such models to reveal new theoretical
insights. We have found that bote calculations are an effective way of deriving
these insights. For example, in the present chapter we have used bote
calculations to explain how policy results depend on characteristics of our
basecase forecasts.
9. Appendix: the derivation of theformulas for analysing the difference between

GDP paths in alternative simulations
Growth in real GDP in simulation j from year 0 to year T is
J8
See for example Harrison and Kimbell (1985) and Harrison (1986). For other approaches to numerical
sensitivity analysis see Pagan and Shannon (1985 and 1987), Amdt and Hertel (1997) and DeVuyst and
Preckel(1997).
T
gj T =ri( i + 8t)- 1 (9-!)
t=l
where g| is the proportionate growth rate in real GDP for year t-1 to year t.
A second order approximation to g^T is
4T = £SI + £ I«W • (9-2)

t=l t=l t>t
From (9.2) we obtain
sfc-gST-ZfBF-«0 + 2 Sfrf«?-8t8?) (9-3>

t=l t=l x>t
where the superscripts p and f denote the policy and forecast simulations.
Using the identity
g ? g ^ - g £ g w - f g $ - g v ) ( g w + g w ] / 2 + fgw - 8 w ] ( « v + S v ) / 2 <9-4>
we can rewrite (9.3) as
SOT-SOT =Zfgf-g[]( 1+H tT) <9-*)

t=iv /
where
HtT = Z k + g s ) / 2 . (9.6)
s=l
s*t
From (9.5), we see that the proportionate difference between the policy and forecast
results for real GDP in year T, i.e.,
D e v T - ( g S T - g o T ) / ( 1 + gOT) <9-7)
can be written as
T
Dev
T = Zfgt > -gO h tT • <9-8>
J
t=l
where
h t T = ( l + H t T )/(l + gST) . (9.9)
Next we decompose the g\s into component parts. We start with the formula
m{«5>H (9.10)
i=l
where
mîs the proportionate increase in real GDP between year t-1 and t calculated
from a mid-point base, i.e.,
mtj = gJ/(i+Kg;); (9.1D

v^j is the proportionate increase in quantity variable i [e.g., employment and

capital stock, see (7.12)] between years t-1 and t calculated from a mid-point
base; and
S^j is the mid-point share of quantity variable i in real GDP, i.e., S^ is the
average share of i in real GDP over years t-1 and t.
By using mid-point concepts, we ensure that (9.10) is highly accurate, i.e., the
omitted second order terms are negligible.
On rearranging (9.11) as
f>\=m\/{\-y2mfy (9.12)
and substituting from (9.12) and (9.10) into (9.8) we find that
DevT-SfisftvS-sSvSl. (9-13)
t=lU=l J
where
S
Tti=htTs£/(l-Xmj) • (9.14)
Finally we rewrite (9.13) as
F F
Dev T = £QCONTj T + ^SHCONTix (9.15)
i=l i=l
where QCONTiT, SHCONTiT, STti, and vti are defined by (7.14) to (7.17).

Dixon 2001

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Dixon 2001

Uploaded by

Copyright:

Available Formats

Dynamic General Equilibrium Modelling for

Forecasting and Policy: A Practical Guide and

Access to this document was granted through an Emerald subscription

is a partner of the Committee on Publication Ethics (COPE) and also works

*Related content and download information

An Illustrative Application ofMONASH:

MONASH is a framework for:

Parliamentary debates for many months.1 However, we do not require readers to be

decomposition simulations. Sections 6 and 7 describe the forecast and policy

5. Historical and decomposition simulations: the Australian motor vehicle

In this section we use MONASH historical and decomposition simulations in a

5.1. Information for the period 1987 to 1994

Table 5.1. Growth in Motor Vehicle related variables from

vehicles from the Industry reduction in landed-duty

Table 5.1 continued

Variable Shock used Source Note

National Accounts consumption data for commodity 22 and

5.2. The historical simulation

primary factor input (the share-weighted average of the percentage changes in

Table 5.2. Sales Shares for the Australian

Total 1.00 1.00

5.3. The decomposition simulation

Table 5.3. Cost Shares for the Australian

Other costs 0.01 0.01

attributable to changes in variables such as those identified in the column headings

To facilitate the handling of accumulation relationships in a comparative static

NFLt +T =NFL t *(1+R) T + I [Kt+s+1 -K t + s (l-D)-APS*GNP t + s ](l+R) T - 1 - s .

LHS(5.6) = NFL87 (5.7)

recognizes that except in the unlikely situation of a steady-state equilibrium, stock

by rectangles while endogenous variables are shown in ovals. The change in

payments to foreigners (a proportion of the change in start-of-year net foreign

9.Real GDP, GDP deflated

n.Price deflator for GDP (Pg)

18.Price deflator for

demands protect- domestic household employ- required supply

rows 12 and 13 (a sharp increase in imports and a relatively small decrease in

employees as part of their remuneration. For agriculture, our findings probably

xmf - xdf = - a f * (prnf - pdf ) + twist; , (5.12)

to GDP is sufficient to cause imports to increase relative to exports, despite the

Many of Australia's export-oriented industries are capital intensive and are

However technical change in other industries favoured the use of Motor

rt^*100*AAPS nt.+ . , r]k ANFL „_* 4TT ,c . , .

6. Forecast simulation: prospects for the Australian motor vehicle industry,

6.1. Forecast growth in imports relative to growth in domestic sales of

to that of other commodities. [Whereas c.i.f. prices in general increased at an

Table 6.1. Growth rates in macro and motor vehicle variables:

Sales on domestic market (domestic and imported) (S) 4.3 2.6

6.2. Forecast growth in sales of motor vehicles (S)

6.3. Forecast growth in output of motor vehicles (Z)

6.4. Summary of forecasts for motor vehicles

7. A policy simulation: the effects of reductions in the tariff on motor vehicles

Table 7.1. Tariff rates applying to elf. imports of

2002 11.69 9.83

7.1. Key assumptions

An illustrative application ofMONASH

An illustrative application ofMONASH

An illustrative application ofMONASH

An illustrative application ofMONASH

disposable income. We do this by exogenizing Fc in (16.69), discussed in

Pc=(PgTgc) a 8 C (PvTvc) a v c . (7.1)

Pi=(PgTgi) a g i (P v Tvi) a v i , (7.2)

From these equations we find that

NFLt +T =NFL t (1+R) T + I [Kt+s+1 -K t + s (l-D)-APSGNP t + s ](l+R) T - 1 - s .

rt^100AAPS nt.+ . , r]k ANFL „_* 4TT ,c . , .

kP(t)-kf(t) = X[sr(t)-Sf(t)lkfve(t) + Xrkf(t)>kf(t)lSfvc(t), (7.11)