Transcript Macro Conference IV

Chapter 13
Empirical Models
of Stabilization
© Pierre-Richard Agénor and Peter J. Montiel
1




Central macroeconomic policy challenge: how to
achieve stabilization and adjustment while minimizing
the cost measured in terms of real income.
In the past, this refered to reducing inflation and
improving the current account while avoiding short-run
income losses arising from deficient aggregate demand.
Recently, disappointing medium-term growth experience
of many developing countries has turned attention to
the maintenance or reactivation of the economy's longrun growth momentum.
In this chapter, analytical tools available to address
interaction among stabilization, adjustment, and growth
in practical developing-country applications are
examined.
2



Short-run macroeconomic models assume that
productive capacity is exogenous.
In contrast, the study of the interaction between
stabilization and growth requires the specification of
medium-term models.
Two essential features of these models:
 Productive capacity is treated as endogenously
determined.
 Explain the rate of accumulation of productive factors
and the rate of change in total factor productivity as
functions of the current and expected future values of
macroeconomic variables and policy instruments.
3
Interaction among stabilization, adjustment, and
growth in medium-term models is:
 Given the predetermined values of total factor
productivity and stock of productive factors, economy's
short-run equilibrium simultaneously determines output,
employment, price level, the current account, and rate
of net investment in new productive factors.
 Rate of net investment determines the stock of
productive factors in the next period, which together
with updated values of total factor productivity
determine the next period's level of productive capacity.
 Thus, growth of productive capacity between this period
and the next depends on
 characteristics of this period's short-run equilibrium,
 rate of net investment generated in that equilibrium.
4


Four types of models: “Bank-Fund” models, “gap”
models, macroeconometric models, and computable
general equilibrium models.
Success of such models in shedding light on the
interactions between stabilization and growth in
developing countries depends on three features:
 Specification of determinants of productive capacity.
 Description of the forces determining the rate of
accumulation of productive assets and total factor
productivity.
 Quality of the model's description of the economy's
short-run equilibrium.
5




Bank-Fund Models.
Three-Gap Model.
Macroeconomic Models.
Computable General Equilibrium Models.
6
Bank-Fund Models



The IMF Financial Programming Model.
The World Bank RMSM Model.
A Simple Bank-Fund Model.
8
The IMF Financial
Programming Model



IMF provides advice to developing countries on
macroeconomic policy.
The Fund extends financial support to stabilization
programs that meet certain criteria: they must
 be consistent with the principles set out in the
institution's articles of agreement;
 offer a convincing prospect of repayment.
This assistance is conditioned on the borrowing
country's compliance with a set of quantitative policy
performance criteria drawn up in consultation with the
Fund and embodied in a financial program.
9



Design of such a program and specification of such
criteria rely on “financial programming.”
Simplest financial programming model determines the
magnitude of domestic credit expansion required to
achieve a desired balance-of-payments target under a
predetermined exchange rate.
Balance sheet identity for the banking system,
M = D + ER,
E: nominal exchange rate;
D: credit to the nonbank sector;
R: claims on foreigners;
M: monetary liabilities.
10


R and M are endogenous and D is an exogenous policy
variable under the control of the monetary authorities.
Velocity:
 = Y/M,

Y: nominal GDP.
Money market is required to be in flow equilibrium:
M = -1Y - -1
-1Y-1.

Assume: nominal exchange rate and velocity are both
constant and that nominal output is exogenous.
11

Model can be solved for the change in the stock of
international reserves R as a function of  and Y, as
well as of the monetary policy instrument D:
ER = -1Y - D.

(4)
Given a target value for the change in reserves and
projections for  and Y, required expansion in the stock
of credit can be derived from
D = -1Y - ER.

Expanded version (“Polak model”) makes nominal
output endogenous (Polak, 1957).
12

Balance-of-payments identity
R = X - (Y-1+Y) + F,



0 <  < 1. (6)
X - (Y-1+Y): net exports.
X: autonomous component of net exports.
F: exogenous net capital inflows.
Figure 13.1: interaction between the money-market
equilibrium condition (4) and the balance of payments
identity (6) in determining nominal income and the
balance of payments.
(4) is positively sloped MM locus, while (6) is negatively
sloped locus RR.
E: equilibrium values of the balance of payments and
13
change in nominal income.
F
i
g
u
r
e
1
3
.
1
T
h
e
P
o
l
a
k
M
o
d
e
l
R

M
M
R
*

E
R
R
y
*

y

14



Increase in the rate of expansion of credit causes the
balance of payments to deteriorate and nominal income
to rise.
Increase in exogenous receipts of foreign exchange
improves the balance of payments and raise nominal
income.
“Polak” financial programming model can be given
 “classical” closure (solved for the domestic price
level, taking real output as exogenous);
 “Keynesian” closure (solved for changes in real
output, taking the price level as given).
15
The World Bank RMSM Model



IMF financial programming model:
 Short-run model of stabilization and adjustment.
 “Adjustment” refers to the balance of payments.
 “Stabilization” refers to the price level in classical
mode and to real output in Keynesian mode.
 It contains no aggregate production function and
does not determine capacity output.
World Bank's Revised Minimum Standard Model
(RMSM): model of capacity output.
Used to generate macroeconomic projections in country
economic reports at the Bank.
16





Emphasis in RMSM: ascertaining whether the domestic
and external financing available to a particular country is
adequate to achieve a target for economic growth.
Key analytical feature of RMSM: link between financing
and capacity growth.
(6): interpreting Y as real output, if autonomous net
~
exports are taken to be exogenous (X = X), the
economy's growth is determined by the volume of
external financing F - R.
RMSM is then used to calculate the volume of domestic
saving required to sustain this increment to output.
Assume fixed-coefficients Harrod-Domar production
function in capital and labor, with capital taken to be the
scarce factor.
17

Increase in capacity requires domestic investment
I = Y,
(7)
: incremental capital/output ratio (ICOR).
18

Letting C and S denote domestic consumption and
saving, we can use the national income accounting
identity
~
I = (Y - C) + (Y - X),
(8)
and the definition of savings
S = Y - C,
(9)
to find the volume of domestic saving required to
achieve a given increment to productive capacity:
~
S = X - Y-1 + (-)Y.

(10)
S is an increasing function of Y, since  is a number in
19
the neighborhood of 3-4, while  is a fraction.

With saving itself linearly related to output in Keynesian
fashion,
S = SA + sY,

0 < s < 1.
(11)
Resulting framework can be used to derive the level of
autonomous saving required to sustain an increment to
capacity output:
~
SA = X - Y-1 + [ - ( + s)]Y.
(12)
20
A Simple Bank-Fund Model





The model can be used in a variety of ways, with the
“financing needs” mode described above being the most
common.
With R and Y exogenous, while F and S are
endogenous, the model is solved in “financing needs”
mode.
With R bounded from below, F and S exogenous, Y
endogenous, and (8) setting a minimum value for
investment, the model is solved in “two-gap” mode.
Alternative application of the RMSM is to combine it with
the financial programming model to derive a simple
Fund-Bank model of adjustment and growth.
Assume: Y is real output and SA is exogenous.
21




(12) can be solved for the rate of capacity growth Y.
Growth of productive capacity is determined in
neoclassical fashion by
 availability of saving (domestic and foreign, via SA
~
and X);
 productivity of investment, given by .
This capacity growth rate can be fed into the Polak
model given by (4) and (6).
Since Y is now interpreted as real output, and (4)
involves nominal income, it must be rewritten as
ER = -1[(P-1 + P)Y + Y-1P] - D, (13)
P: domestic price level.
22
Resulting model can be solved in two ways:
 If capital inflows are exogenous and the domestic price
level is endogenous, (6) yields the balance of payments
R and (13) the domestic price level.
 If the domestic price level is exogenous and capital
inflows are endogenous, (13) yields the balance of
payments and (6) the level of capital inflows.
23
Three-Gap Models
“Financing gap” method incorporated in the RMSM is
the most venerable approach to the projection of real
output growth in developing countries.
 “Gap” models are close contenders.
 “Two-gap” models focus on foreign exchange and
domestic saving as alternative constraints on growth,
date back to Chenery and Strout (1966).
 “Three-gap” models includes also fiscal gap.
Bacha (1990):
 Foreign exchange availability is linked
 directly to the rate of growth of productive capacity;
 indirectly to the level of actual real output.

25

ICOR relationship described in (7) is retained, but
investment now is assumed to require imported capital
goods:
Z = I,

0 <  < 1,
Z: level of capital goods imports.
From (14) and the balance-of-payments identity:
I = (1/)[X + (F - J)],

(14)
(15)
X: level of exports net of other imports;
J: sum of external debt service, transfers, and changes
in foreign exchange reserves.
Suppose that X is subject to the upper bound X*,
26
determined by external demand.

Then (15) becomes the inequality
I  (1/)[X* + (F - J)],

(16)
which represents the “foreign exchange” constraint on
investment, and thus, by (7), on capacity growth.
“Savings” constraint is derived as follows: from the
balance-of-payments and national income accounting
identities (6) and (8), we have
I = Y - C - G - (F - J),
G: government spending.
27

Defining private saving as Sp = Y -  - C, where  is
government net tax revenue, the previous equation
becomes
I = Sp + ( - G) - (F - J).


Sp is an increasing function of output and is bounded
~
above by private saving at full capacity output, Sp,
 taking private consumption to be exogenous;
 noting that Y is bounded above by full-capacity
output.
This means that I is also bounded from above, so that
~p
I  S + ( - G) - (F - J),
(17)
28
which represents the saving constraint on investment.

To derive the “fiscal constraint,” assume that base
money is the only financial asset available for the
private sector, so that private sector's budget constraint
can be written as
Sp - Ip = M/P,

Ip: private investment;
M: stock of base money.
Change in M is assumed to be given by
M = M(, ),
: rate of inflation;
: “propensity to hoard.”
0 <  < 1,
29

In this case, all foreign capital flows accrue to the
government, and the budget constraint of the
consolidated public sector can be written as
Ig = M(, ) + ( - G) + (F - J),

(20)
Ig: public investment.
Total investment:
I = Ip + Ig.
(21)
30

Assume: private and public investment are
complements, so that private investment is bounded
above by the level of public investment:
Ip  k*Ig,

(22)
k*: ratio of private to public capital in the composite
capital stock.
From (20) to (22), fiscal constraint on total investment
takes the form
I  (1 + k*)[M(, ) + ( - G) + (F - J)]. (23)
31






This model simultaneously determines the level of
output, the current account, the rate of growth of
productive capacity, and the rate of inflation.
Focus of “gap” models is on the implications for such
variables of alternative levels of foreign financing (F J).
Figure 13.2: illustrate the mechanisms at work.
Central endogenous variable I is plotted against (F -J).
(16) and (17) are plotted as the loci FF and SS,
representing the foreign exchange and saving
constraints.
Slope of SS is unity, as can be verified from (17), while
that of FF is 1/, which is greater than unity, since  is a
fraction.
32
F
i
g
u
r
e
1
3
.
2
T
h
e
T
h
r
e
e
G
a
p
M
o
d
e
l
I
F
G
(
1
+
k
*
)
[
H
(
,
)
+
(
g
)
]




S
I
'
G
S
~
p
S
+
(
g
)

(
1
/
m
)
x
*
F
0
S
o
u
r
c
e
:
A
d
a
p
t
e
d
f
r
o
m
B
a
c
h
a
(
1
9
9
0
,
p
.
2
9
1
)
.
F
J
)
'


F
J

33






Relative positions of the two curves rely on the
~
assumption that (1/)X* < S + ( - G).
Hatched areas beneath the curves represent the
feasible regions for I.
If net foreign inflows are (F -J)’, both constraints are
binding and investment is I’.
To the left of (F - J)’, foreign exchange constraint
binds.
Investment is determined by foreign exchange
availability.
Economy will suffer from Keynesian excess capacity
 since investment will therefore be less than the level
that would satisfy (17) as an equality;
 since the other components of aggregate demand
34
are fixed.

Actual output will be given by
Y = C + (1-)Ic + X*,



Ic: actual constrainted level of investment.
If (F - J) exceeds (F - J)’, economy will be
constrained by domestic saving.
Investment will now be determined along SS, and
output will be at full capacity.
Slack variable in this case is net exports, which are
squeezed by domestic demand and are given by
X = Y* - C + (1-)Is,
(25)
Is: savings-constrained level of investment.
35
This part of the analysis reproduces the two-gap model
of Chenery and Strout.
How does the ``fiscal gap'' fit in?
 Inequality (25) is represented by an area bounded
above by a locus GG with slope 1 + k* and vertical
intercept (1 + k*)[M(, ) + ( - G)].
 Quantity 1 + k* may be greater or less than 1/, so GG
may be steeper or flatter than FF.
 Curves GG and SS have the same slope.
 But, their relative heights depend on the values of  and
k*.
 Although the private sector budget constraint (18)
~p
implies that S > M(, ) as long as Ip is positive, the
~p
difference between S and M(, ) decreases with .

36






Thus a larger value of  raises the height of GG relative
to SS.
Larger value of k* has a similar effect.
Incorporate the fiscal constraint in the model: treat  as
an endogenous variable that ensures that (23) holds as
an equality.
In this case, the role of the fiscal constraint is merely to
determine the rate of inflation.
Given the value of I, (21) and (22) holding as an equality
would determine the levels of Ip and Ig, and given the
latter, (23) holding as an equality would determine .
Endogenous changes in  move GG to intersect
whichever of the two other loci happens to be binding at
a point directly above the relevant value of (F - J).
37



Increase in (F - J) would
 increase the rate of capacity growth by raising I,
 reduce the rate of inflation by permitting the
government to finance itself externally, rather than
through the inflation tax.
If  is an exogenous policy variable, GG serves as an
independent constraint.
If the fiscal constraint does not bind
 k is the slack variable;
g
p
 given I and I , private investment I is determined,
and the actual value of k implied by Ip and Ig may be
smaller than k*.
38



If fiscal constraint binds, increase in (F - J) increases
capacity growth, because receipt of foreign financing
results in higher public investment, which in turn
induces more private investment.
Actual level of output will rise, through Keynesian
demand effects emanating from higher levels of both
private and public investment.
This brings the economy closer to full capacity
utilization, and net exports will fall.
39
Macroeconomic Models
Problems related to models considered in Section 1:
 Omit substantial amount of economic structure and
behavior.
 They are deficient as medium-term models both
because of
 their specification of the determinants of productive
capacity;
 rate of accumulation of productive factors.
 Production function is of the fixed-coefficients HarrodDomar type.
 They do not contain an independent investment function
describing the behavior of agents who actually make the
decision to accumulate productive factors.
41



Investment in physical capital is treated as a residual.
 Bank-Fund model treats investment as determined
by the available saving;
 three-gap model derives it residually from saving,
foreign exchange availability, or the government
budget, depending on which is the binding constraint.
To analyze complicated dynamic macroeconomic
phenomena, macroeconometric models are used.
This section provides a brief overview of the structure of
“representative” macroeconometric models for
developing countries.
42



Structure of Production.
Aggregate Supply.
Aggregate Demand.
43
Structure of Production




Majority of macroeconometric models for developing
countries have been built along the lines of the MundellFleming production structure.
Economy specializes in the production of a single home
good, which is an imperfect substitute for the foreign
good.
Domestically produced good can be either consumed at
home or exported.
Domestic residents consume both the home good and
the foreign good.
44
Aggregate Supply


Family of models dubbed RMSM-X retains the constant
ICOR assumption.
Capacity output is specified as a function of capital K,
labor N, and an imported intermediate good O:
Y = F(N, O; K),


with standard neoclassical properties.
Empirically, this function is given a Cobb-Douglas or
CES form.
Capital stock is predetermined in the short run and
grows endogenously as the result of net investment.
45





Both the level of total factor productivity and the labor
force grow exogenously.
Imported intermediate goods are available from the
world market in infinitely elastic supply at an exogenous
foreign-currency price.
Determination of the domestic-currency price of the
home good depends on how the supply side of the
economy is modeled in the short run.
In the simplest macroeconometric models, the supply
side is described along
 Keynesian fix-price or
 classical flex-price lines.
Keynesian version has no dynamics arising from the
expansion of productive capacity, since it takes output
46
to be entirely demand driven.

Classical version:
 continuous full employment is assumed;
 output of the home good is determined by the
inherited capital stock, the size of the labor force, and
the price of the imported intermediate good;
 short-run supply behavior is depicted via a variable
mark-up price equation.
47

Price of the home good is taken to be determined by
unit costs plus a mark-up factor that depends on the
rate of capacity utilization in the economy:
+
+
+
P = P(w, PO, Y/Yc),

(27)
w: nominal wage;
Yc: capacity output;
PO: domestic-currency price of the imported
intermediate good.
Short-run aggregate supply curve is an upward-sloping
function of level of real output, and it is displaced
vertically by changes in w or in PO.
48



In (27) w and Yc are predetermined, while Y and P are
solved out simultaneously.
Price of imported intermediate goods depends only on
their exogenous foreign-currency price and the
exchange rate.
Behavior of nominal wages is based on an
expectations-augmented Phillips curve:
+ +
+
N(Y,
K,
P
)
w
O
,  a]
=[
w
Ns
Ns: labor force;
a: expected inflation rate;
N(): actual employment.
49

Most common approach to the modeling of expectations
formation in macroeconometric models for developing
countries has been the use of adaptive expectations.
50
Aggregate Demand
Demand side of macroeconometric models for
developing countries tends to have an IS-LM flavor with
ad hoc adjustments of behavioral relationships.
“IS” side:
 Set of determinants of consumption includes
disposable income Ydisp, real interest rate r, and some
type of real wealth variable a = A/P:

+
- +
C = C(Ydisp, r, a),

0 < C/Ydisp < 1. (29)
There is no particular consensus on the specification of
private consumption behavior, and the specifications
range.
51


A is a predetermined variable that evolves over time in
response to private saving, derived from substituting
(29) in the private sector's budget constraint.
Set of determinants of private investment spending
includes real output, a cost of capital term q, and lagged
capital stock:
+ - -
I = I(y, q, K-1).

Problems related to investment functions:
 omission of forward-looking variables;
 failure to account for the likely irreversibility of
investment in physical capital.
52


Some structural aspects of potential significance for
understanding capital accumulation in developing
nations are not dealt with in the investment functions:
 effect on investment of the existence of an external
debt overhang;
 role of complementary factors such as infrastructure,
human capital, and imported inputs;
 role of financial repression.
Treatment of the trade sector:
 Exports are determined by foreign income and the
relative price of the home good.
 Imports depend on the same relative price as well as
on domestic real GDP.
53
Some models include a measure of foreign reserve
availability in the import demand function to capture
the intensity of rationing in official foreign exchange
market.
“LM” side:
 Institutional disaggregation includes the central bank
and the deposit money banks.
 The central bank's balance sheet is at the heart of the
financial sector, with the change in its liabilities linked to
changes in its assets.
 Demand for money depends on real GDP and a
domestic nominal interest rate.
 Careful modeling of financial repression and informal
markets are not found in macroeconometric models.

54



Common treatment of capital mobility leaves foreign
interest rates out of domestic asset demand functions,
taking capital flows to be exogenous.
Both specification of domestic financial markets and
their interactions with world capital markets play
important roles in determination of cost of capital.
Result: macroeconometric models fall short of providing
a suitable framework for the analysis of the interactions
among stabilization, adjustment, and growth.
55
Computable General
Equilibrium Models
Economy-wide modeling approach for developing
countries is the construction of computable general
equilibrium (CGE) models.
 CGE models are applied microeconomic models,
designed to study issues such as the effects on
resource allocation and income distribution of various
types of external shocks or policy interventions.
Key distinction between CGE models and
macroeconometric models: level of aggregation
assumed for both goods and factors.
 CGE models are highly disaggregated on both their
demand and supply sides.
On the supply side:
 Contain a large number of distinct production activities
and factors of production.
57

Supply of imports is disaggregated into different types of
commodities that can be purchased on the world
market.
 Profit-maximizing behavior by firms generates sectoral
supply functions for individual domestically produced
commodities and demand functions for various types of
factors.
On the demand side:
 Different types of households are distinguished by the
nature of their ownership of factors of production.
 Utility maximization by individual classes of households
generates demand for domestic and foreign goods, and
supply functions for different types of factors.

58
World demand for domestic goods exhibit different price
elasticities for different goods, permitting the domestic
economy to exercise market power over some
commodities.
 Relative price adjustments play a central role.
Distinction between macroeconometric models and
traditional CGE realted to their dynamic properties.
 CGE models are oriented to the solution of period-byperiod comparative-statics exercises.
 Since CGE models represent actual economies,
macroeconomic phenomena of an explicitly
intertemporal nature appears.
 But, these phenomena are modeled in rudimentary
fashion.

59




Private households are assumed to save a fixed fraction
of their incomes.
Public consumption is exogenous.
Investment is often either exogenous or determined by
saving rate.
Implications of these assumptions:
 with households not spending all of their income
each period;
 with independent investment and public consumption
decisions;
CGE models require a period-by-period macroeconomic
mechanism to reconcile aggregate saving and
investment (“closure” rule).
60




Dewatripont and Michel (1987) classify potential
closures into four types:
 Keynesian,
 Johansenian,
 Kaldorian,
 classical.
Sensitivity of numerical results to the particular choice of
a closure rule is not always clear, and may vary
depending on the issue at hand.
None of these mechanisms leaves traditional CGE
models well suited to the analysis of the links among
stabilization, adjustment, and growth.
Reason: all of them have the effect of leaving
investment exogenous or determined by saving.
61




Addressing links between stabilization and growth
requires the inclusion of an independent investment
function, that responds endogenously to current and
expected future macroeconomic variables.
Since these intertemporal links are omitted, dynamics in
traditional CGE models are too simple to capture
interactions among stabilization, adjustment, and
growth.
Walrasian problem solved each period is static.
Recently, CGE models have begun to incorporate more
sophisticated macroeconomic relationships:
 “New structuralist'' CGE models in Taylor (1990),
 “micro-macro” models patterned after Bourguignon et
al. (1992),
incorporate traditional macroeconomic relationships.62




This modification brings recent CGE models close to
“dependent economy” macroeconomic simulation
models.
Although these innovations enrich the macro dynamics
exhibited by CGE models, the macroeconomics of these
models remain relatively simple.
In contrast to the static optimizing behavior assumed for
within-period sectoral supply and demand functions,
dynamic behavior is left rather simple and ad hoc.
Intertemporal optimization on the part of either
households or firms based on forward-looking
expectations remain absent.
63
Appendix: The Khan-Knight
Monetary Disequilibrium
Model

Figure 13.3.
64
F
i
g
u
r
e
1
3
.
3
L
o
g
i
c
a
l
S
t
r
u
c
t
u
r
e
o
f
t
h
e
K
h
a
n
K
n
i
g
h
t
M
o
d
e
l
D
e
m
a
n
d
f
o
r
m
o
n
e
y
R
e
a
l
o
u
t
p
u
t
E
x
p
e
c
t
e
d
i
n
f
l
a
t
i
o
n
r
a
t
e
F
o
r
e
i
g
n
i
n
f
l
a
t
i
o
n
r
a
t
e
D
o
m
e
s
t
i
c
i
n
f
l
a
t
i
o
n
r
a
t
e
T
a
x
e
s
F
i
s
c
a
l
d
e
f
i
c
i
t
I
m
p
o
r
t
s
G
o
v
e
r
n
m
e
n
t
s
p
e
n
d
i
n
g
P
r
i
v
a
t
e
s
e
c
t
o
r
c
r
e
d
i
t
E
x
p
o
r
t
s
C
u
r
r
e
n
t
a
c
c
o
u
n
t
T
o
t
a
l
d
o
m
e
s
t
i
c
c
r
e
d
i
t
C
a
p
i
t
a
l
f
l
o
w
s
N
e
t
f
o
r
e
i
g
n
a
s
s
e
t
s
E
n
d
o
g
e
n
o
u
s
v
a
r
i
a
b
l
e
s
M
o
n
e
y
s
u
p
p
l
y
65