The implications of myopic policy-making for macroeconomic performance

International Journal of Forecasting 5 (1989) 49-58

North-Holland 49

The implications of myopic policy-making for macroeconomic performance

Saade N. CHAMI and David W. BUTTERFIELD * American University of Beirut, Beirut, Lebanon

McMaster University, Hamilton, Ont., Canada L8S 4M4

Abstract: This paper evaluates the implications of myopic behaviour in policy-making on intertemporal economic performance by applying an optimal control approach to Canada. Myopia is characterized in two ways; first by the rate at which the future is discounted and second by the policy maker’s planning horizon. The optimal decision rules, which correspond to various degrees of time preference and different time horizons, generate an intertemporal tradeoff curve of economic performance. This tradeoff tends to disappear with an excessive preoccupation on the present vis a vis the future. That is, an extremely myopic policy results in poor economic performance in the future without any improvement in the present.

Keywords: Macroeconomic policy, Time preference, Time horizon, Interemporal tradeoff.

1. Introduction

It is often believed that governments, interested primarily in staying in power, behave myopically in macroeconomic policy-making. The purpose of this paper is to evaluate the implications of such behaviour for economic performance by applying an optimal control approach to Canada. Optimal policies are derived by maximizing a presumably agreed-upon fixed social objective function subject to the constraint of a macroeconometric model for Canada. The optimal decision rules derived here correspond to varying degrees of myopia in policy making. This myopia is characterized in two ways; first by the rates at which the future is discounted and second by the length of the planning horizon of the policy maker. The results of the control experiments show that there exists an intertemporal tradeoff of economic performance which, as suggested by the shape of the tradeoff curve, tends

* The authors wish to thank Professor F.T. Denton and A.A.

Kubursi for helpful comments.

to disappear with an excessive preoccupation on the present vis a vis the future. ’

The plan of the paper is as follows. In the next section we discuss briefly the structure of the macroeconomic model, its estimation and simulation. In section 3, we describe the specification of the objective function. Next, in section 4, the control experiments are discussed and the results are analysed. The final section summarizes the major findings of the paper.

2. The model

It should be emphasized at the outset that the model proposed here is intended to be an illustra- tive model which can serve as a vehicle for the examination of shortsightedness in macroeconomic policy. 2 It is. therefore, aggregated, simple

’ To avoid any confusion, the term ‘economic performance’

referred to in the paper has a specific meaning: the deviations of the target variables from their desired values. The

smaller these deviations are, the better is the performance of

the economy.

’ The present model is similar in some respect to the models

developed by Buiter and Owen (1979), Pindyck and Rubin- field (1981).

0169-2070/89/$3.50 0 1989, Elsevier Science Publishers B.V. (North-Holland)

50 S.N. Chami and D. W. Butterfield / Myopic policy-making for macroeconomic performance

in structure and expectations are assumed to be static. 3 Nonetheless, it captures many distinctive characteristics of the Canadian economy.

The model combines both short-run and long- run characteristics. For example, although effec- tive demand determines the level of output in the short run, the production function in the model determines the level of capacity output in the long-run. 4 Any increase in one of the components of GNP will induce a rise in wages and prices, and so limit the growth of demand for output. An increase in exports, for example, will raise actual output relative to capacity or potential output, thus inducing a price increase and therefore limit- ing the growth of foreign demand for home output. Such a price increase can also result from an increase in consumer expenditures, which in turn leads to an increase in interest rates and, conse- quently, to a decline in these expenditures. How- ever, total output is not constrained by the aggregate production function at every point in time but can deviate from it in the short-run. Wages and prices are sticky in the short-run but respond to demand and supply pressures in the long-run. In addition, the present model goes beyond tradi- tional short-run analysis by allowing for the accumulation of capital stock, the accumulation of

The assumption of static expectations admittedly represents

a shortcoming in the model especially after the Lucas cri-

tique (1976). The reason for ‘neglecting’ this criticism is the

desire to keep the structure of the model as simple as

possible in order to facilitate the compulation of optimal

control solutions, and to concentrate exclusively on the issue

of myopic policymaking without getting into the issue of

time-inconsistency of optimal plans in the presence of ra-

tional (or forward looking) expectation as discussed by

Kydland and Prescot (1977) and in rmmerous subsequent

papers (see for example Calvo (1978) Buiter and Marston

(1985), Currie (1985), Lucas (1986)).

Capacity or potential output is generated by fitting the following Cobb-Douglas production function:

where K is the stock of capital, UR* and UR are the ‘full

employment’ and actual unemployment rates, respectively, t is a time trend and LF is the total labour force. CR* is

generated by linking the troughs in the unemployment rate series which occurred in 1966 and 1974, and then extrapolat-

ing for the rest of the data period. Once this equation is

estimated, potential output is generated by equating UR and

I/R* and using the estimated coefficients a, /3 and y. Thus

QP, = aKP((l- lJRF)LF,)pe”.

financial assets via the government budget constraint, and technical progress. The channels of transmission of economic policy in the model are typically Keynesian. That is, the effect of fiscal policy on output is direct, while monetary policy exerts its influence via its effect on the rates of interest. As for the foreign sector, the model as- sumes that Canada is a small open economy. In estimating the model the exchange rate is treated as an endogenous variable, even though the Canadian dollar was fixed from 1962 to 1971. In the optimal control solutions, either the change in reserves or the exchange rate (but not both) can be chosen as a policy instrument since they are re- lated through the balance of payments constraint. The exchange rate has been chosen because the solution algorithm converged more quickly than when the change in reserves was used.

The model consists of twelve behavioural equations and nine identities. A full listing of the model is provided in the appendix. The model can be organized conceptually around six interrelated blocks: a domestic components of gross national expenditure block; a wage-price-unemployment block; a monetary sector block; a balance of payments block; a potential GNP block; and a government sector block. A considerable degree of interrelationship exists among the endogenous variables as can be easily seen from the block structure diagram shown in fig. 1.

2. I Estimation of the model

Initially, the model was estimated by Ordinary Least Squares (OLS). However, in simultaneous equation models, such as the present one, inconsistency of the OLS estimator of the structural parameters may arise because the assumption of zero correlation between the right-hand-side variables and error terms is often violated. The use of the Two-Stage Least Squares (2SLS) method is usually recommended. In the present model, however, the number of predetermined variables ex- ceeds the sample size, so that the first stage of the 2SLS method breaks down. A solution to this difficulty is to reduce the number of predetermined variables through the technique of principal components. 5 A set of six principal compo-

5 Johnston (1984).

S.N. Chami and D. K Butterfield / Myopic policy-making for macroeconomic performance 51

Fig. 1. Block structure of the model.

nents was computed accounting for 95 percent of the total variation among the predetermined variables. This set was used in the first stage as instru- mental variables for all endogenous variables ap- pearing on the right-hand-side of the simultaneous equations. The fitted values of the endogenous variables from the first stage were then used in the second stage of the Two-Stage Least Squares principal component method to obtain the results shown in the appendix. Those equations with sig- nificant autocorrelation and lagged endogenous variables were re-estimated using the method proposed by Fair (1970) which consists of aug- menting the list of instruments by adding to it the lagged dependent and independent variables in order to obtain consistent estimates. The test for serial correlation was based on the standard Durbin-Watson statistic DW. Since this test loses its power somewhat in equations involving lagged dependent variables, the Durbin (1970) h-statistic was also calculated.

2.2 Simulations with the model

However well single equations may explain the behaviour of individual components of aggregate economic activity, the ultimate test of a structural model remains the fit of all equations taken to- gether in simulation. There are various types of simulations which can be performed depending on the objective for which these simulations experiments are conducted. Here, we are mainly interested in assessing the ability of the model to reproduce the historical values of the major endogenous variables (historical simulations) and its stability properties. Two types of historical simulation were carried out. The first was a static simulation in which the endogenous variables were calculated with all predetermined variables taking on their actual values. The second was a dynamic simulation where the lagged endogenous variables take the values simulated by the model. While the former gives an indication of the model’s ability to

52 S.N. Chami and D. W. Butterfield / Myopic policy-making for macroeconomic performance

match the behaviour of the system in the short-run. the latter assess its long-run tracking behaviour. In both simulations the tracking record of most endogenous variables was remarkably good as shown

in table A.1 in the appendix. Since the model is designed to examine inter-

temporal tradeoffs in policy analysis it must ex- hibit stable behaviour and converge to a steady state solution in the long-run. The model can be represented by AY(t) =f[Y(t), X(t), 01, where A is a difference operator, Y(t) is the vector of endogenous variables, X(t) the vector of exogenous variables and 8 is the set of structural parameters. This system of difference equations is said to have a steady-state solution if a particular solution y(t) = q*ePr’ exists for all variables i,

when each exogenous variable grows at a constant rate, that is, X,(t) = Xj*exlr. X* and Y” are the steady-state levels of the exogenous and endogenous variables, respectively, at r = 0, and hi and pi are the corresponding growth rates. 6 The steady-state solution that we derived is a special case of the above particular solution. It is a zero- growth steady state where p, and X, are set equal to zero, that is, every variable of the model stays constant (a stationary state). This stationary state calls for the imposition of some restrictions on the structural coefficients, 8, in order to avoid any inconsistency. Once the stationary state is derived, the stability properties of the model where then analysed by displacing the model from this state by means of shocks. If the model has a tendency to return to the initial stationary state after the shock, then one would conclude that the model is locally stable. ‘After many experiments, the model proved to be stable except in the case where the government finances its deficit by issuing bonds. *

3. The objective function

A control experiment consists of minimizing a loss function subject to the constraint of the mac-

Knight and Wymer (1978). Since the present model is nonlinear, one can argue that the

simulation technique is the most plausible way to examine

the stability question. Another way would consist of lineariz-

ing the model and then computing its characteristics roots.

However, this would only indicate a local stability and

therefore has no advantage over the method used here. More details of the simulation experiments can be found in

Chami (1986).

roeconomic model. Having specified the model, the next step is to specify the objective function. In this paper we use a quadratic loss function which penalizes the weighted sum of squared deviations of the target variables from their desired values. It is of the form

T

L = c (Y, -yr* )‘(Q~‘)(Y, -yr* 1, 1=1

where yI and yI* are vectors of computed and desired values of the targets, Q is a diagonal positive semi-definite weight matrix, and X is a discount factor that can be given different values

reflecting different rates of time preference. Note that, as formulated by Chow (1975) the vector of target variables, yr, includes both state and control variables. From a practical point of view, the inclusion of policy instruments in the objective function is essential in order to avoid the problem of instrument instability, which is a major problem in optimal control applications. The target (state) variables chosen to be included in the objective function are the rate of growth output (5%), the unemployment rate (4%) the inflation rate (4%) and balance of payments equilibrium. The instrument (control) variables that would steer these targets toward their desired values are the level of government expenditures (a 4.1% rate of growth), the change in the money base (729.3 million $), a surtax (the intercept in the tax function, -2841.4 million $) and the exchange rate (1.0724 per U.S. dollar). A penalty is also imposed on government borrowing in order to avoid the instability problem which occurs when government spending is financed by borrowing. The target rate of growth of government debt is set at 15 percent which implies that the ratio of government debt to GNP follows its historical trend.

The desired trajectories for the target variables were chosen in such a way as to eliminate the undesirable effects of the symmetric objective function which penalizes the deviations of its arguments from their desired values equally regardless of their direction. Hence a trial method was used in which the desired values for the targets were shifted gradually from their historical averages until a solution was reached in which the welfare cost represented only unfavorable deviations. As for the weights imposed on the variables in the objective function, they are equal to the

S. N. Chami and D. W Butterfield / Myopic policy-making for macroeconomic performance 53

inverse of the historical variances of these varia-

bles. This choice solves the scaling problem and implies that the greater is the historical variation of a target the smaller is the weight attached to it.

4. The implications of myopia on economic performance

4. I Time preference in policy making

5.0

4.0

3.0

2.0

1.0

0.0 .5+.=& , , , , 2 4 6 8 10 12 14 16 18

TIME

Fig. 2. Total Welfare Cost. One way of representing myopia in policy mak-

ing is to manipulate the discount factor h in the objective function by giving it different values representing different degrees of emphasis or preference for the present over the future (X < 1). For sake of comparison, we will assume that there exists a far-sighted policymaker (X > 1) who pre- fers the future over the present. 9 A third policymaker is assumed to follow a ‘ time neutral’ (X = 1) policy plan. The outcome of this ‘time neutral’ plan can serve as a benchmark against which we can compare the other two plans.

The choice of the discount factor h has been made in order to attach the same degree of impor- tance to the first period versus the last period for the myopic planner as to the last period versus the first period for the far-sighted planner. While the intertemporal weights are changing by the factor h, the relative structure of the weights at each point in time remains the same implying that any difference in the results may be attributed solely to the differences in time preference.

4. I. I Results The optimal control solution for the three pro-

grams (time-neutral, myopic and farsighted) are compared by the sums of the squared deviations of the targets in the objective function from their desired values weighted by the matrix of weights Q. ia That is

Lt=(y,-yt*)Q(y,-y,*) for t=l...T.

9 This assumption might be regarded as unrealistic especially when T is large. However, one may argue that governments

resort in some circumstances to such policies in order to

redirect some targets toward their desired long-run paths. Or, an ‘honest’ government may opt for this type of

farsighted policy if it is considered to be the optimal out-

come.

The results of these computations are shown in fig. 2 where time runs along the horizontal axis and the welfare cost L, along the vertical axis. i1 Observe that the far-sighted program (X = 1.1364), denoted by dots, has the lowest cost in the second lo-year period of the planning horizon, while the opposite is true for the short-sighted program (h = 0.8799) denoted by squares. As for the time- neutral program (X = l), denoted by circles, the welfare loss time-path lies between the other two. In figs. 3 and 4 the welfare cost is shown for only the target and control variables separately. It is interesting to note that the gain realized in terms of target achievement in the myopic program is obtained by incurring more cost in the use of instrument variables. On the other hand the far- sighted planner uses his instruments less inten- sively but in return he is not as successful as the short-sighted planner in achieving his targets, except in the last 5 years.

Five more experiments were carried out, each based on a different degree of emphasis on the present vis a vis the future. In all these experiments, the time profiles of welfare losses of the three programs, cross at the same point, namely, the 10th period. Thus the welfare loss before and after this period can be accumulated for each program, and can be translated into a single point in a two-dimensional graph, in which the horizontal axis represents the sum of the losses in the first 10 years, and the vertical axis the sum of the losses

lo All the control experiments carried out in this paper are

based on the algorithm developed by Chow (1975, 1981).

” Note that the welfare scale on the vertical axis has been

adjusted by a multiplicative factor for graphical conveni-

ence.

54 S.N. Chami and D. W. Butterfield / Myopic policy-making for macroeconomIc performance

2.0

1.5

1 .o

0.5

2 4 6 8 10 12 14 16 18

TIME

Fig. 3. Target Welfare Cost.

in the second lo-year period. In other words, for each program, we computed two values

LI = : (Y,-yt*)‘Q(yt-Y,*) and t=l

L,= 5 b-yt*)'Q(~t-Y,*). t=10

These two values generate a single point on the tradeoff curve.

The results of the eight experiments, expressed in terms of Root Mean Squared (RMS) weighted deviations, i.e. [C, ( y, - y,* )‘Q( y, - y,* )/T],“2 are presented in table 1, in which the degree of emphasis refers to the first period vis a vis the last period, or vice versa. The eight points corresponding to the eight experiments are plotted in fig. 5 and joined by straight lines to form an intertemporal tradeoff curve for macroeconomic performance. The tradeoff curve appears to have a smooth and continuous shape and tends to ap-

I I I (

2 4 6 8 10 12 14 16 18

TIME

Fig. 4. Instrument Welfare Cost.

Table 1

Root mean squared deviations of the optimal levels from their

targets in the myopic, time-neutral and far-sighted programs.

Discount Degree of

factor emphasis

(A)

0.6813 1000

0.7742 100

0.8046 50

0.8466 20

0.8799 10

1.0000 -

1.1364 10

1.1810 20

Welfare cost in Welfare cost in

the first 10 the second 10

years (&) years (L, )

2554 6162 2578 5825

2591 5793

2617 5751 2645 5715

2848 5567 3565 5361 4030 5286

preach the two axes asymptotically becoming al- most parallel to the verical axis. The slope of the tradeoff curve increases from - 0.16 to - 14 sug- gesting that an extremely myopic view in for- mulating economic policy would lead to a vertical intertemporal tradeoff curve. That is, the performance of the economy would be much worse in the future with no improvement in the present. A similar conclusion can be drawn if the emphasis is shifted to the future as shown in fig. 5. Thus, there are limits on short-run performance which cannot be surpassed, and over-emphasizing the performance of the economy in the short-run will only restrain the economy in the longer run.

4.2 Time horizon effets in policy-making

A more realistic representation of myopia in policy making is the length of the time horizon over which governments try to achieve their objec-

Economic Performance

9 6200

8 3 6000

I b:: za

5800

ZY ;g 5600

4

; 5400

2 a

2500 3000 3500 4000

RMS DEVIATION IN THE FIRST 10 YEARS

Fig. 5. Intertemporal Tradeoff Curve of Economic Perfor- mance.

S.N. Chami and D. W. Butterfield / Myopic policy-making for macroeconomic performance 55

tives. It has been a frequent criticism of economic planning that the time horizon is short and that

policies focus attention on the outcome for the economy in the present and the near future with little consideration of their long-run implications. In order to investigate this in the present context, three experiments, each corresponding to a different time horizon, are carried out. Both a four-year time horizon, which coincides more or less with the duration of most Canadian governments in office, and a two-year time horizon, which is even more short-sighted, are considered; while the far- sighted policy maker is assumed to have a lo-year time horizon. The optimal policies are then derived in the following way. First, we start the process by optimizing over the specific time horizon - say a 4-year time horizon - using the actual first period historical values as initial conditions. Second, we assume that in the second year the policy maker is free to change his strategy and to choose a new policy, given the optimal solution of the first year. Thus, the values of the endogenous variables (targets and instruments) are reinitial- ized and the optimization is repeated for a second 4-year time horizon (sequential planning revision). That is, if the planning period is 1964-1967 in the first case, it becomes 1965-1968 in the second case, and so on. This implies that decisions are revised sequentially year by year, rather than being made once for all years of the planning horizon. It is more realistic to assume that short-sighted policy makers to not always adhere to decisions they have made in the past, but rather that they revise their decisions frequently in light of the new infor- mation that became available to then each year. This process is repeated for 10 years (1964-73) i.e., the final run was made over the period 1973-1976 for the 4-year program. The same procedure is followed in the case of the other two

programs. Assume now that a new administration came to

power in 1973. It is believed that this administration does not concern itself only with the im- mediate effects of its policy but takes account of the longer term future as well. It sets out a far- sighted plan for the next 10 years, taking as an initial condition that situation inherited from the previous administration. Now, the question is what implications the three previous programs have on the present far-sighted plan. To answer this question, three more experiments are carried out, each

2 4 6 8 10 12 14 16 18 20

TIME

Fig. 6. Total Welfare Cost.

one taking the values of all variables in the I973 optimal solution that correspond to the 2-, 4- and lo-year time horizon as initial conditions. The difference in the solutions of these three experiments are attributable to the difference in the initial conditions, which are in turn due to the effects of the planning horizon used by the policy makers in the first ten years of the optimization process. In other words, these differences measure the damage done to the far-sighted planner by the economic conditions he inherited from the previous administration. If we compute these differences in terms of weighted squared deviations, then we can measure the long-run costs of pursuing a myopic economic policy. On the other hand, the weighted squared deviations from the first 10 period represent the short-run costs of pursuing more or less far-sighted policies.

4.2.1 Results

Following the same procedure used in the first experiment, the weighted squared deviations of the three programs are computed and their time paths are plotted in fig. 6. The 2-year horizon program has the lowest cost in the first 10 years and the highest cost in the last 10 years. The opposite is true for the lo-year horizon program, while the cost path of the 4-year horizon program lies between the two. The Root Mean Squared weighted deviations shown in fig. 6 are summed for the first 10 years and for the last 10 years for each of the three programs. Thus we calculate 6 values which are represented by 3 points in fig. 7. By connecting these three points we obtain an intertemporal tradeoff curve between optimal policies representing various degrees of myopia.

S. N. Chami and D. W. Butterfield / Myopic policy-making for macroeconomic performance

2200 2400 2600

RMS DEVI~ICN~THE FIRST

Fig. 7. Intertemporal Tradeoff Curve of Economic Perfor-

mance.

The highest point on the plot represents the 2-vear The highest point on the plot represents the 2-year horizon plan, the middle point the 4-year horizon plan, and the lowest point the lo-year horizon plan. The slope of the tradeoff curve increases as the horizon of policy makers become shorter. Thus it is expected that, as the planning horizon becomes shorter, the cost in the future will increase considerably with little or no benefit in the present.

These results are not counterintuitive. If the planner did not take into consideration the future he would be less constrained, and therefore more successful in achieving his desired targets in the present. On the other hand, if the planner had a sufficiently long view of the future, then he would have to make allowance for some anticipated shocks which would constrain him in the present, and therefore make his policies less successful in the present. In other words, if the planner suffers from myopia, he will not be concerned with the consequences of his present actions in the years beyond his short time horizon and may thus achieve good performance in the near future, but only at the cost of poor performance further down the road.

5. Summary

In this paper an attempt is made to investigate empirically the implication of myopia in policy making on macroeconomic performance in the Canadian context. Two approaches were used. The first consists of varying the relative structure

of the weights over time by changing the rate of time preference in the objective function. An intertemporal tradeoff curve for economic performance, with points corresponding to different discount factors, was then derived. The second approach aimed at measuring and comparing the costs implied by different time horizons used by policy makers. Tradeoff relationships were then derived between the performance of the economy in the present and the future, expressed in terms of the weighted deviations in the objective function. The results of both approaches confirm the existence of an intertemporal tradeoff between the effects of macroeconomic policies and indicate that as more and more emphasis is put on the present (future), the higher is the cost incurred in the future (present) relative to the gain realized.

References

Buiter, W.H. and R.C. Marston. 1985, International Economic

Policy Coordination, (Cambridge University Press, Cam-

bridge).

Appendix

Table A.1

Results of historical simulations (1964-1982) *

Variable Mean Dynamic Static

RMS RMSP RMS RMSP

Q 99199.2 2429.60 0.02 2274.50 0.02

CON GFI

NFINV

M

X

RS

RL

UR

B

I+

T

CF

AR

;

K

AH

60475.9 2751.30 0.05 1588.30 0.03

22294.0 803.60 0.03 680.50 0.03

531.1 833.47 2.11 1336.87 3.29

24640.4 1061.50 0.04 749.60 0.04

22401.3 1867.60 0.09 1615.80 0.08

0.0727 0.01 0.17 0.01 0.13

0.0824 0.01 0.11 0.01 0.06

0.0618 0.01 0.17 0.01 0.11

0.0652 0.02 0.30 0.02 0.30

0.0847 0.02 0.25 0.01 0.17

46023.3 2613.50 0.09 2885.40 0.09

2870.3 4286.30 5.40 3592.90 2.76

88.0 2204.90 30.90 2627.67 29.50

67737.5 3379.60 0.06 3229.50 0.06

102625.0 2234.50 0.02 1573.40 0.01

245889.0 4784.90 0.02 0.06 0.001

666.4 637.72 1.25 789.30 1.05

* RMS and RMSP are the Root-Mean-Square and the Root-

Mean-Square-Proportionate simulation errors.

S.N. Chami and D. W. Butterfield / Myopic policy-making for macroeconomic performance 57

Table A.2

Equations of the model.

Estimated Equations *

(1) CON = 4087.02+ 0.571973 YD - 57060.8 RS +0.375024 CON_,

(2.67) (2.14) (- 3.52) (1.16)

R2 = 0.9956; SER = 1079.47; D W = 1.92; h = 1.1; 1963-82

(2) GFI= 631.537-18046 (RL - i)_, +0.2222 Q

(0.60) ( - 2.55) (23.0) Rz = 0.988; SER = 573.39; DW = 1.63; p = 0.43; 1964-82

(3) KFINV= -1184.25+0.2715 NFQ -3548.75 A(RS - ?)+0.5382 NFINV-,

(- 3.35) (5.23) ( - 0.44) (2.39) ?i* = 0.6277; SER = 748.0; D W = 2.02; p = - 0.51; 1964-82

(4) &‘= 0.02695 - 0.4044 UR + 0.5012 i + 0.6217 Ii’

(2.59) (- 1.27) (3.02) (3.07) i?* = 0.8627; SER = 0.0116; DW=1.69; h =1.15; 1964-82

(5) p = - 0.0576 +0.3043 + + 0.3924 Pk 1 + 0.0756 (Q/QP)m,

(- 0.27) (0.59) (1.71) (0.33)

i?’ = 0.5654; SER = 0.0241; D W = 1.99; 1964-82

(6) UR = 0.2325-0.2510 (Q/QP)+O.O018 t +0.3506 UR_,

(3.93) (- 3.73) (2.54) (1.43)

i?* = 0.8566; SER = 0.0070; D W = 1.41; p = 0.65; 1963-82

(7) RS=0.3102-4.6660 (H/(Q.P))+O.5918 j

(- 2.65) (- 2.54) (2.15)

i?’ = 0.7307; SER = 0.0202; D W = 1.85; p = 0.51; 1963-82

(8) RL = 0.0071+ 0.2749 RS + 0.1230 ARS + 0.7031 RL_, (1.51) (2.90) (1.27) (5.21)

i?* = 0.9819; SER = 0.0038; D W = 1.98; h = 0.05; 1963-82

(9) M = - 17678.6 + 0.3529 Q + 7774.01 (P/PM)

(- 2.78) (42.65) (1.16)

it* = 0.9895; SER = 923.91; D W = 1.13; 1962-82

(10) X = 35725.3 + 282.83 IPU - 1098.95 (PX/PU. e) - 47068.8 (Q/QP)

(2.91) (8.03) ( - 0.21) (3.46)

i?* = 0.9801; SER = 1092.75; DW = 1.77; 1962-82

(11) CF/Q.P=O.O172+0.5961 A(RS-RSU)+0.8009A(RL- RLU)+O.l759A(GFl/Q)

(6.98) (2.20) (0.85) (0.56) i?* = 0.3377; SER = 0.0108; DW = 1.78; 1963-82

(12) T = 0.3259 (Q P)

Identities * *

(13) Q=CON+GFIiNFINV+FINV+G+X-M

(14) yD= Q.P-6(K.P)-T+TRP+IPPD

(15) K=(l-a)K_,+GFI_;

(16) AR=PX.X-PM.M+CF+TRAN

(17) Qp = e1.21289~0.35~(1 _ us* ~LF~0.65e0.CQ7246”2.r

(18) IPPD=RL.B

(19) AH=P.G+TRPiIPPD+AR-T-AB

(20) PM=e.pf

(21) NFQ=Q-FQ

* The numbers in parentheses under each coefficient esti-

mate are the corresponding t-statistics.

** A residual term has been added to each identity (except

(17) and (20)) in order that the identity hold exactly in the

data.

58 S. N. Chami and D. W. Butterfield / Myopic policy-making for macroeconomic performance

Table A.3 Definition of variables.

Endogenour

CF = Net Capital Inflow

CON = Real Consumption

GFI = Real Investment

H = High Powered Money

IPPD = Interest on the Public Debt

K = Real Capital Stock

M = Real Imports

NFINV = Real Non-Farm Inventories

NFP = Real Non-Farm GNP

P = General Price Level (GNP Implicit Deflator)

PM = Import Price

Q = Real GNP

Qf’ = Real Potential GNP

R = Official International Reserves

RL = Long-Term Interest Rate

RS = Short-Term Interest Rate

T = Total Taxes

UR = Unemployment Rate

W = Wage Rate

X = Real Exports

YD = Real Disposable Income

Exogenous

B

e

FINV

FQ G

IPU

LF

PF

PX

PU

RLU

RSU

t

TRP

TRAN

UR*

s

A

= Total Bonds

= Exchange Rate

= Real Farm Inventories

= Real Farm GNP

= Real Government Expenditures on Goods

and Services

= US Industrial Production Index

= Labour Force

= Foreign Price Level

= Price of Exports

= US Price Level

= US Long-Term Interest Rate

= US Short-Term Interest Rate

= Time Trend

= Transfer Payments to Persons

= Current Account Net Transfer

= ‘Full-Employment’ Unemployment Rate

= depreciation rate

= denotes backward difference operator;

AX=X-X_,

= denotes a proportional rate of change;

ri= AX/X_,

Buiter, W.H. and R. Owen, 1979, “How successful has stabili-

zation policy been in the Netherlands? A neo-keynesian

perspective”, De Economist, 17, 58-104.

Calvo, G.A., 1978, “On the time consistency of optimal policy

in a monetary economy”, Econometrica, 46, 1411-28.

Chami, S.N., 1986, “The short-run versus long-run tradeoff m

macroeconomic policy: An optimal control analysis in the

Canadian context”, Unpublished Ph.D. Dissertation, Mc-

Master University.

Chow, G., 1975, Analysis and control of Dynamic Economic

Systems (Wiley, New York).

Chow, G.. 1981, Econometric Analysis by Control Methods

(Wiley, New York).

Currie, D., 1985, “Macroeconomic policy design and control

theory - A failed partnership?“, The Economic Journal, 95,

285-306.

Durbin, J., 1979, “Testing for serial correlation in least squares

regression when some of the regressors are lagged depen-

dent variables”, Econometrica 38, 490-21.

Fair, R.C., 1970, “The estimation of simultaneous equations

models with lagged endogenous variables and first order

serially correlated errors”, Econometrica, 38, 507-16.

Johnston, J., 1984. Econometric Methods. Third edition (Mc-

Graw-Hill, New York).

Knight, M.D. and C.R. Wymer, 1978, “A macroeconomic

model of the United Kingdom”, International MonetaT

Fund Staff Papers, 25, 742-78.

Kydland, F.E. and E.C. Prescott, 1977, “Rules rather than

discretion: The inconsistency of optimal plans”, Journal o/

Political Economy, 85, 473-91.

Lucas, R.E., 1976, “Econometric policy evaluation: A critique”,

Journal of Monetary Economics, suppl., 19-46.

Lucas, R.E., 1986, “Principles of fiscal and monetary policy”,

Journal of Monetar)? Economics, 17, 117-34.

Pindyck, R.S. and D.L. Rubinfield, 1981, Econometric Models

and Econometric Forecasts (McGraw-Hill, New York).

Biography: Saade N. CHAMI is an Assistant Professor at the

Graduate School of Business and Management at

the American University of Beirut. He holds a

Ph.D. in Economics from McMaster University,

Canada. His research interests include macroeco-

nomics, finance and quantitative methods.

David W. BUTTERFIELD is an Associate Profes-

sor of Economics at McMaster University. He re-

ceived his Ph.D. from the University of

California-Berkeley in 1975. His research interests

include policy modelling and development plan-

ning.

Documents

The implications of myopic policy-making for macroeconomic performance