Operational Manual for “A Macroeconomic Framework for ......Operational Manual for “A Macroeconomic Framework for Quantifying Growth and Poverty Reduction Strategies in Niger”

Operational Manual for “A Macroeconomic Framework

for Quantifying Growth and Poverty Reduction Strategies in Niger”

Nihal Bayraktar* and Emmanuel Pinto Moreira**

First complete draft: February 2, 2005 This version: October 09, 2005

Abstract

This operational manual provides detailed information on the simulation of a macroeconomic model linking aid, public investment (disaggregated into education, health, and infrastructure), and growth, developed by Agénor, Bayraktar, and El Aynaoui (2005) and applied to Niger by Pinto Moreira and Bayraktar (2005). The manual explains how the model is specified, the parameters are calibrated, and the program is run. It also explains the different steps to follow to introduce policy shocks, analyze the output table, and derives policy implications. In order to help readers not familiar with Eviews to get started, we provide some basic information on EViews 4.0.

⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯

*Penn State University - Harrisburg and World Bank. E-mail address: [email protected]. **World Bank.

2

Table of Contents I. INTRODUCTION

II. INPUT DATA FILE

III. PARAMETERS 1. Econometric Estimation of Some Parameters

2. Given Parameter Values

3. The Parameter Calibration Part of the Program

IV. EXOGENOUS VARIABLES PROJECTED WITHIN THE PROGRAM

V. CALCULATION OF RESIDUALS TO EQUATE ESTIMATED VARIABLES WITH THEIR

ACTUAL VALUES

VI. PARTIALLY ADJUSTED VARIABLES

VII. THE SIMULATION PROGRAM

1. How to Install the Simulation Package

2. The Setup of the Simulation Package

3. Details about the EViews Simulation Program

3.1. The Basic Information about Running the Program in EViews

3.2 EViews Commands Used in the Program

4. Details about the Excel Output Files

5. Details about the Summary Table File

VIII. Simulating Shocks

SHOCK 1 - Increase in Foreign Aid

SHOCK 2 - Reallocation of Public Investment

SHOCK 3 - Reduction in Tariffs

The Non-Neutral Case – Shock 3a

The Neutral Case: Adjustment in Direct Taxation – Shock 3b

The Neutral Case: Adjustment in Indirect Taxation – Shock 3c

IX. SENSITIVITY ANALYSIS

X. LINKING THE MODEL WITH THE MILLENNIUM DEVELOPMENT GOALS XI. LINKING THE MODEL WITH THE DECOMPOSITION OF PUBLIC CAPITAL EXPENDITURE TABLE APPENDIX A – Definitions APPENDIX B – List of Variables and Parameter Estimates APPENDIX C – Estimation Results APPENDIX D – EViews Commands Used in the Program and Their Meanings APPENDIX E – List of Equations APPENDIX F – Simple Example Model APPENDIX G – Tables of Simulation Results APPENDIX H – Calculation of Variables and Projection of Exogenous Variables

3

I. INTRODUCTION, BACKGROUND, and OBJECTIVES

Pinto Moreira and Bayraktar (2005) applied a macroeconomic model which

analyzes the linkages between aid, public investment, and growth developed by

Agénor, Bayraktar, and El Aynaoui (2005) to Niger.1 The macroeconomic model has

been simulated using the software Eviews 4.0. The simulation program has been

used to create the baseline results and to investigate the effects of alternative policy

shocks on the economy, including an increase in aid/GDP ratio, a reallocation of

public investment toward investment in infrastructure, and neutral and non-neutral

reduction in effective tariff rate.

The objectives of this operational manual is twofold: (i) help the user of the

Niger’s model understand the technical aspects of the modeling exercise carried out

in Niger and, (ii) familiarize a reader interested in macro-modeling with some basics,

including modeling procedure, methods, and requirements using the Niger’s model

as an example.

The remainder of the manual is organized as follows. Section II presents

information on the input data file. Section III describes the calculation procedures and

methods used to compute the parameters used in the model. Section IV, presents

the methods used to project exogenous variables within the model. Section V

explains how the residuals are defined. Section VI describes the procedure of

introducing partially adjusted variables in the model. Section VII presents detailed

information about the simulation program written in Eviews 4.0. Section VIII describes

how shocks are run in the model. Section IX concludes.

PRELIMINARY REMARKS TO GET STARTING

Required Programs and the Files in the Package

• Two software programs are required to run this simulation program:

1 This paper applies the dynamic macroeconomic framework developed originally by Agénor, Bayraktar, and El Aynaoui (2005) to Niger.

4

(a) EViews Version 4 or higher2.

(b) Microsoft Excel.

• Four different files are used to simulate the model. . a. “Niger-Data.xls” Excel data file: This is the input data file. It contains

initial values of both exogenous and endogenous variables, and

projections for exogenous variables.

b. “Niger-Simulation.prg” EViews program file: This file is used to run the

simulation program.

c. “Niger-output.xls” Excel data file: It reports the simulation results

created by “Niger-Simulation.prg”. The names of the output files are

“OUTPUT-NIGER.xls” for the baseline output data; “OUTPUT-NIGER-

SHOCK1.xls” presenting the “Shock 1” output data; “OUTPUT-NIGER-

SHOCK2.xls” presenting the “Shock 2” output data; “OUTPUT-NIGER-

SHOCK3A.xls” presenting the “Shock 3A” output data; “OUTPUT-

NIGER-SHOCK3B.xls” presenting the “Shock 3B” output data;

“OUTPUT-NIGER-SHOCK3C.xls” presenting the “Shock 3C” output

data.

d. “Niger-table.xls” Excel data file: This table summarizes the simulation

results. The names of the output files are “NIGER-Output Table -

BASELINE.xls” for the baseline summary table; “Niger-Output Table-

SHOCK1.xls” for “Shock 1” summary table; “Niger-output-Aid-Shock-

Table 4.xls” presenting the deviation of the “Shock 1” values from the

baseline values; “Niger-Output Table-SHOCK2.xls” for “Shock 2”

summary table; “Niger-output-Aid-Shock-Table 5.xls” presenting the

deviation of the “Shock 2” values from the baseline values; “Niger-

Output Table-SHOCK3A.xls” for “Shock 3A” summary table; “Niger-

output-Aid-Shock-Table 6.xls” presenting the deviation of the “Shock

3A” values from the baseline values; “Niger-Output Table-

SHOCK3B.xls” for “Shock 3B” summary table; “Niger-output-Aid-Shock-

2 This software is created by Quantitative Micro Software. The new versions of EViews have tools for programming and solving simulation models.

5

Table 7.xls” presenting the deviation of the “Shock 3B” values from the

baseline values; “Niger-Output Table-SHOCK3C.xls” for “Shock 3C”

summary table; “Niger-output-Aid-Shock-Table 8.xls” presenting the

deviation of the “Shock 3C” values from the baseline values; “NIGER-

Output Table – BASELINE-table 9.xls” for the baseline summary table

in case of lower public investment efficiency; “Niger-Output Table-

SHOCK4.xls” for “Shock 4” summary table (aid shock with lower public

investment efficiency parameter); “Niger-output-Aid-Shock-Table 10.xls”

presenting the deviation of the “Shock 4” values from the baseline

values.

II. INPUT DATA FILE Data entry and Location • The input file is named as “Niger-Data.xls” (an Excel file). Variables in the model

are separated into two groups: exogenous and endogenous variables. While the

exogenous variables are determined outside the model, the endogenous

variables are determined inside the model.3 The values of these variables and

parameters are presented in the input file.

• The location of the endogenous variables is on “ENDO” sheet, the exogenous

variables on “EXO” sheet, and the parameters on “PARAM” sheet. Data sources

of the variables are given in Appendix H.

• While the names of the variables are reported in column A, their definitions are

given in column B. In the following columns, the data points are presented starting

from 1999. The base year is 2004 in the model. But the data file starts in 1999

due to the presence of lagged variables in the model. All data points between

1999 and 2004 are actual numbers.

6

Base year and Simulation period

• The base year is 2004 in the model. But the data file starts in 1999 due to the presence of lagged variables in the model. All data points between 1999 and 2004

are actual numbers.

• The model is simulated for the years starting from 2004 until 2015. As specified above, the simulated values of endogenous variables for this period are

determined within the model. But we have to project the values of exogenous

variables for these years since they are determined outside the model.4 The

projected values of exogenous variables are also reported on “EXO” sheet for the

years starting from 2005 until 2015. Detailed information on projections is reported in

Appendix H.

III. PARAMETERS

Three types of procedures have been used to determine the values of the

parameters in the model:

• Estimation running regression equations;

• Use of parameters provided in various studies as given; or,

• Calibration within the model5.

1. Econometric Estimation of Some Parameters

• The parameters of the three “fiscal” equations and private investment equation

(IP), which are listed below, are obtained by running econometric regressions. The

estimation technique is the ordinary least squares. But in order to correct for serial

correlation, the equations are estimated with autoregressive processes of order one

3 Exogenous and endogenous variables are listed in Appendix B and the list of equations is given in Appendix E. 4 It should be noted that some exogenous variables (DB, FP, ERROR_OMM, AID, LE_G, and WG) are projected within the model since they are projected as a constant share of endogenous variables. 5 Different types of parameters are used in Pinto Moreira and Bayraktar (2005). Their definitions are given in Appendix A. In this appendix, we also define the production and transformation functions used in the model.

7

and/or two, denoted AR(1) and AR(2) below, where needed. All regressions are

based on annual data for the period 1982-2002. The E-views program used to run

the regression equations are given in “niger-regression-ig-indtxr-ip.prg”. The input file

is “Niger-dataset-regression-IG-IP-INDTXR.xls”

• The estimated equations are

INDTAX = INDTXR(INDTXR-1, AID/NGDP)·PQ·Q (1)

PQT·IG/NGDP = IG[(TAX/NGDP)-1, AID/NGDP, (AID/NGDP)2] (2)

PQT·IP/NGDP = IP((∆Y/Y-1) -2, PQT*KGINF/NGDP, ER·FP/NGDP). (3)

• The definitions of variables and equations are given in Appendix B and E,

successively. The estimation results are reported in Appendix C. It should be noted

that the coefficient of (TAX/NGDP)-1 in the PQT·IG/NGDP equation in the model is

adjusted downward since the high value of the coefficient was causing the simulated

value of government investment, IG, to be extremely sensitive to changes in the tax

rate. Similarly, the coefficient of (AID/GDP)2 in the same equation is adjusted

downward since the higher value of the coefficient was leading IG to be less

responsive to changes in aid. The estimated coefficient of PQT*KGINF/NGDP in the

PQT·IP/NGDP equation is adjusted upward to make changes in the level of public

capital stock in infrastructure more effective on private investment.

• The regression results are:

INDTXR = 0.008 + 0.706*INDTXR-1 - 0.029*AID/NGDP

(2.865) (7.717) (-2.068)

Adjusted R2 = 0.742; Durbin-Watson statistic = 1.861

AR(1) = -0.350 (-1.347)

PQT·IG/NGDP = -0.174 + 0.649*(TAX/NGDP)-1 + 1.549*AID/NGDP

(-2.333) (3.380) (2.363)

-3.261*(AID/NGDP)^2

8

(-2.175)


AR(1) = 0.527 (1.762); AR(2) = -0.420 (-1.679)

PQT·IP/NGDP = 0.001 + 0.056*(∆Y/Y-1) -2 + 0.15*PQT*KGinf/NGDP

(0.058) (2.099) (1.413)

+ 0.033* ER·FP/NGDP -0.028* Dummy87 -0.027*Dummy92_95

(0.186) (-3.871) (-4.042)


AR(1) = 0.829 (2.723); AR(2) = -0.353 (-1.271)

2. Imposed Parameter Values

Some of the parameters in the model are determined either by dwelling on the

scant literature for Niger, or by using plausible values for low-income developing

countries in general—including the estimates used by Agénor, Bayraktar, and El

Aynaoui (2005) for Ethiopia in a similar framework. The values of these parameters

are reported in the “GIVEN PARAMETER VALUES” section of the simulation

program.

• In Eviews, all parameters must be specified as scalars. For this purpose, the

“scalar” command is used. The general Eviews syntax is the following:

scalar name of the parameter = its value

Example. The value of the parameter σDE is introduced in the program as

follows:

scalar sigma_DE = 0.3

9

• Some of the parameters need to be calculated using the values of other

parameters as it is the case for ρDE, which is a function of σDE. In the program, it is

coded as follows:

scalar rho_DE = 1+(1/sigma_DE)

Elasticity of substitution

• The elasticities of substitution on the production side were kept at relatively low

values. For instance, the elasticity of substitution between T and KP, σJ, was set to

0.3; the elasticity of substitution between LE and Kghea/POPθH, σT, to 0.3; and the

elasticity of substitution between J and KGinf, σY, to 0.4. σZ is taken as 0.2. The

corresponding substitution parameters are calculated by using these values of

elasticity of substitution. How they are calculated is presented in the “Substitution and

Transformation Parameters” section of the simulation program.

Elasticity of transformation

• The elasticity of transformation in domestic production was set at 0.3, whereas

the elasticity of transformation between domestic and imported goods at 0.7. The

corresponding transformation parameters are calculated in a similar way (see the

“Substitution and Transformation Parameters” section of the simulation program).

Shift parameter

• Most of the shift parameters are calibrated within the model as explained in

detail below. However, three of them are given, including AJ, AT, AZ, and AKGZ

which are taken 1. The reason is that we need to calibrate J, T, and Z variables,

which are not empirically observable, by assigning some values to their shift

parameters.

Share parameters

• βZ is taken as 0.5. βE has been initially calibrated, but since the calibrated

value was not proper, it is taken as 0.3. βT and βY are taken as 0.85. βJ is

10

0.6. βDM, which is equal to 0.75, is calculated as the share of DOM in the sum

of DOM and M. βDE, which is equal to 0.15, is calculated as the share of X in

the sum of DOM and X.

Depreciation rates

• For lack accurate information on the depreciation rate of capital stocks are

taken, the values used were taken from Agénor, Bayraktar, and El Aynaoui (2005).

The rate of depreciation of public capital (education, health, infrastructure, and other),

delta_h, is set at 0.035. The depreciation rate of the private capital stock, delta_P, is

0.06.

Congestion parameters

• Parameters capturing congestion effects were difficult to estimate, given the

lack of information for developing countries in general. Since congestion effects seem

to be quite significant in Niger, we have chosen relatively high parameter values

compared to the values of congestion parameters chosen for Ethiopia in Agénor,

Bayraktar, and El Aynaoui (2005). The parameter capturing congestion effects in the

education system, theta_KGE and theta_KGI, are set at 0.9; that determining the

strength of congestion effects in the provision of health services, theta_H, at 0.4; and

for the parameter capturing congestion effects in infrastructure capital, theta_I, we

chose a value of 0.3.

• The savings rate is taken as 10 percent.

3. The Parameter Calibration Part of the Program

• The remaining parameters are calibrated within the model. One important

advantage of our simulation program is that this calibration is coded within our

simulation program. In this way, as the values of variables or parameters change, the

calibration of the remaining parameters will be done automatically. Thus we do not

need to use any other program. The related section in the simulation program is

“PARAMETERS CALIBRATED WITHIN THE MODEL”. Most of the share and shift

11

parameters are calibrated in this section. First of all, we calibrated the share

parameters, and then the shift parameters are calibrated using these values of the

share parameters.

• It is worth noting that the variables that are used in calibrating parameters need

to be specified as series. In order to declare series, we use the “series” command in EViews, which is followed by the name of the variable. The reason for this

transformation is that the calibrated parameters are scalar and in order to calculate a

scalar in EViews, all variables must be specified as “series”. The general form of this

command is

series variable name

For example, output, Y, is defined as series using the following command:

series Y

After variables are defined as series, we can refer to specific data points more

easily. For example, Y(5) means the value of Y in period 5, which corresponds to

year 2003.

• βKGZ is calibrated as a share of KGinf in the sum of KGinf and KGedu.

Calibration of the shift parameters

• We calibrated some of the share and shift parameters of the following constant

elasticity of substitution functions.

Y(J, KGinf-1,Y-1) = AY·[βY·J-ρY + (1 - βY)(KGinf-1/Y-1θI) -ρY]-1/ρY

∆LEN = AE·[βE·(LR-1)-ρE + (1 - βE)(Z)-ρE]-1/ρE

In these equations, the calibrated shift parameters are AY and AE.

12

The shift parameters of the following equations are obtained by solving them for ADE

and ADM, using the given values of βDE and βDM.

Y = ADE·[βDE·XρDE + (1 - βDE)DOMρDE]1/ρDE

Q = ADM[βDM·DOM-ρDM + (1 - βDM)M-ρDM]-1/ρDM.

• The last parameter calibrated within the model is APQ, which is the shift

parameter of the composite price level, PQ.

IV. EXOGENOUS VARIABLES PROJECTED WITHIN THE PROGRAM

Projections of variables

• Most of the exogenous variables are projected outside of the model using

different techniques (See Appendix H). Some exogenous variables, including the

number of educated workers in the public sector, LE_G, the wage rate in the public

sector, WG, domestic borrowing, DB, errors and omissions, ERROR_OMM, and

private capital flow, FP, are projected within the model by taking them as a constant

share of other endogenous or exogenous variables. For example, the government’s

domestic borrowing, DB, is projected as a constant share of NGDP (1 % of NGDP)

within the model using the following code

niger.append DB = NGDP*0.01

The “niger.append” command is defined below.

• Some of these constant shares are calculated using 2004 values of the

variables. For instance, errors and omissions are projected as a constant share of

GDP, where this constant share is named as ERRORGDP_const. Using this share,

13

ERROR_OMM is calculated as ERROR_OMM = ERRORGDP_const*(NGDP/ER).

Foreign aid is also projected within the model. The ratio of AID to NGDP (AID_const),

which was 0.1067 in 2004, is multiplied by NGDP to construct the aid series between

2005 and 2015. The number of educated labor in the public sector is defined as a

constant share of total number of educated labor. This ratio (LE_G_const) was equal

to 0.0191 in 2004. The share of FP in % of GDP in 2004 is 0.49 percent. V. CALCULATION OF RESIDUALS TO EQUATE ESTIMATED VARIABLES WITH

THEIR ACTUAL VALUES Calculation of residuals

• As mentioned earlier, some of the parameters are obtained running regression

equations. The simulated values of variables such as IG are defined using these

regression results. Since it is required that all identities must hold in the base year,

we constructed residuals to equate the estimated values of variables to their actual

values in 2004. These residuals are defined for IG, INDTXR, and IP. IG regression

equation is taken as an example. While the actual value of IG in 2004 is

83,631,764,436 in LCU, the estimated value of IG is equal to

Estimated IG = (-0.174921+0.649353*(TAX(-1)/(NGDP(-

1)))+1.549799*(AID/NGDP)-3.26115*(AID/NGDP)^2)*NGDP/PQ.

• This estimated IG is defined using 2004 values of the right hand side

variables, which is equal to 27,679,573,152 in LCU. The difference between the

actual and estimated IG is defined as residual which is equal to 55,952,191,283 in

LCU. IG_RES is the name of this residual.

• The residual for INDTXR (INDTXR_RES = 0.005) is calculated by taking the

difference between the actual and estimated values of INDTXR in 2004:

INDTXR_RES = Actual value of INDTXR – [0.00792+0.706572*INDTXR(-1)-

0.029834*(AID/NGDP)]

14

• Another residual is calculated for private investment, IP, using regression

results. IP_REGRES is equal to 110,380,533,916 in LCU and calculated as

IP_REGRES = Actual value of IP – [(0.001554 + 0.056452*((Y(-2)-Y(-3))/Y(-

3))+0.15304*(PQT*KGINF/NGDP)+0.033393*(ER*FP/NGDP))*NGDP/PQT]

• In addition to these residuals, we also defined one more residuals for AID

variable in order to make identities hold in the base year.

• The data source of foreign aid is OECD, which can be considered a reliable

source. But the values of this series are different from the one reported in the balance

of payment table prepared by IMF. For the balance of payments to hold, we define a

residual which equates the values of foreign aid from these two sources. This

residual is obtained by subtracting the values of AID from the balance of payment

tables from the values of AID taken from OECD sources. AID_RES is equal to -

$225,407,778. This residual is used to calculate the change in net foreign assets of

the central bank, delta_NFA, in the simulation program:

delta_NFA = PXstar*X - PMstar*M - RGstar*FdebtG(-1) - RPstar*FdebtP(-1) +

UTR$ + (AID$ + AID_RES) + FG + FP + ERROR_OMM

• In the model, a residual for private consumption, CP, is also calculated. The

historical CP series is calibrated as Qd – CG – IG – IP. In the model, CP is supposed

to be equal to Ydisp*(1-s). Thus, in order to make CP equal to Ydisp*(1-s), we

introduce a residual, which is equal to CP*PQT – Ydisp*(1-s). The residual is

calculated using the data points in 2004.

• A residual for NGDP is calculated as well. In the model, NGDP =

PQT*Qd+PX*X-PM*M +PMstar*tm*ER*M. In order to equate the right hand side of

the equation to the historical value of NGDP in 2004, we calculate the NGDP residual

which is equal to NGDP in 2004 – (PQT*Qd+PX*X-PM*M +PMstar*tm*ER*M).

15

VI. PARTIALLY ADJUSTED VARIABLES

We assumed that some endogenous variables are adjusted gradually. This

means that they follow a partial adjustment process. For example, total output, Y, is

simulated following a partial adjustment process. In this case, the total output

equation can be written as

Y(J, KGinf, Y-1) = λY.[AY·[βY·J-ρY + (1 - βY)(KGinf/Y-1θI) -ρY]-1/ρY] + (1- λY).Y-1,

where λY (lambdaY in the simulation program) is the adjustment parameter. This

parameter captures a low propensity to adjust total output in the short run. Its value is

0.4, which means that the adjustment rate is 40 percent per year.

Similarly, real imports, M, and domestic sales, DOM, are assumed to follow a

partial adjustment process. These equations are redefined as follows

DOM = λDOM*(X/(((PX/PD)*((1 - βDE)/ βDE))σDE)) + (1-λDOM)*DOM-1,

M = λM*(DOM*(((1 - βDM)/ βDM)*(PD/PM))σDM) + (1-λM)*M-1,

where λM=0.9 and λDOM=0.2 are the partial adjustment parameters.

It is assumed that PD exhibits a disequilibrium price mechanism, adjusting

partially towards its equilibrium value, EQPD:

PD = λPD.EQPD + (1-λPD).PD-1, (43)

where λPD is a parameter measuring the speed of price adjustment towards its

equlibrium value. λPD is taken as 0.4.

16

VII. THE SIMULATION PROGRAM

The simulation program is written and run in EViews and it is connected to

outside input and output Excel files. This section describes: (i) how the simulation

package is installed; (ii) how it is coded; and (iii) how it is run.

1. How to Install the Simulation Package

There are 4 files in the package: “Niger-data.xls”, “Niger-Simulation.prg”,

“Niger-Output.xls”, and “Niger-Output-Table.xls”. The package is installed following

these two steps.

a. Create a directory named “Niger” on the C drive of your computer.

b. Copy all these files into the newly created directory.

2. The Setup of the Simulation Package The execution of the simulation program consists of the following steps:

Step 1: The data for the variables are put in the excel file named “Niger-Data.xls”. When we run the simulation program, the values of exogenous and

endogenous variables will be imported into the program. The details about the input

file are given in Section II.

Step 2: Running of the simulation program is the second phase of the simulation process. It is executed in EViews.

Step 3: When the simulation program is completed, the output file, in which the simulated variables are stored, will be created. It is named as “Niger-Output.xls”.

It should be noted that this output file is automatically generated by EViews and after

each execution, the program overwrites the existing output file.

17

Step 4: In this stage, the simulated variables stored in “Niger-Output.xls” will be used to generate tables summarizing the simulation output. This excel file is

named as “Niger-Output-Table.xls”. When you open this file, you are asked whether

you want to update the information in the file. If you choose “update”, the summary

table will be updated by using newly created values of the simulated variables from

“Niger-Output.xls”.

These steps are presented in Figure 1.

3. Details about the EViews Simulation Program

• Before explaining the setup of the simulation program, the following points

must be emphasized related to programming in EViews.6 EViews can work with

square systems. It means that each equation in the model must have only one

endogenous variable assigned to it. Thus the number of independent equations

excluding exogenous variables which are projected within the model must be equal

to the number of endogenous variables in the model. The solution provided by an

EViews program consists of values for endogenous variables given exogenous

variables.

• EViews is a quite user friendly program. If your only aim is to investigate the

effects of shocks on the economy or to recalibrate the model with new values of

variables and parameters, it is not necessary for you to be familiar with Eviews

programming. But if you want to make any structural change, you may need to have

more experience with Eviews programming.

• Our EViews program is executed by double-clicking on the “Niger-

simulation.prg” file. It will be automatically launched and the simulation starts

immediately. After the completion of running of the program, EViews generates a

workfile named “Niger-Simulation.wk1”, in which simulated variables are stored.

5 A simple example model is presented in Appendix F.

18

3.1. The Basic Information about Running the Program in EViews

This subsection provides basic information on how we can run our simulation

program in EViews.

FIGURE 8

Figure 8 shows how the simulation program looks like when you open the

simulation program file in EViews. In order to run the program, you click on the “run”

bottom (shown in a black circle in Figure 8). When you click on this bottom, the

following window opens. After you click on “OK” bottom, the program starts running.

19

• If there is no error in the program, the workfile of our program will be opened

automatically right after the program stops. The name of this work file is “Niger-

Simulation.wf1”. A sample workfile created by the simulation program is presented in

Figure 9.

FIGURE 9

20

• In this file, one can see the list of all variables and parameters. Baseline

variables (exogenous or endogenous) which are simulated within the program are

named with “_0” extension. Our model, which is named as “Niger” also appears in the

list. When you double click on “Niger”, you can see the details about our model. First

of all, the list of equations appears as default. This is shown in Figure 10. By clicking

on the “solve” icon (shown in a black circle in Figure 10), you can change your

solution method. Figure 11 shows the “solve” window. Our model is solved by using

the deterministic simulation technique.7

FIGURE 10

7 See the EViews Help Manual for details.

21

FIGURE 11

• As specified in the EViews Help Manual, the following steps are taken while

running a deterministic simulation model in EViews:

a) The block structure of the model is analyzed.

b) The variables in the model are restricted to series in the workfile.

c) The equations of the model are solved for each observation in the solution

sample. During this process, an iterative algorithm is used to compute

endogenous variables.

d) The results will be rounded to their final values.

• As it can be seen in Figure 11, “Dynamics” option is used to specify how the

values of the lagged endogenous variables are determined. This means that the

lagged endogenous variables in the model are calculated using the solutions

calculated in previous periods, not from actual historical values.

• If you change the simulation type or the options related to dynamics, the

model will be simulated again by clicking on “OK” bottom in Figure 11. The old values

of the simulated variables will be replaced by the new values of them and these new

values will be stored in the workfile. If you want to store new values, you should save

the workfile by clicking on “save” bottom as shown in a black circle in Figure 9.

22

• You may want to run our program using different parameter values. In this

case, you need to specify these new values in the simulation program file. If you

make any changes in the program, you should save it before you run it again. In

order to save a file, you need to click on the “save” icon (shown in a red circle in

Figure 8).

3.2 EViews Commands Used in the Program

• We use different EViews commands in the simulation program. These

commands and their meanings are given below. Detailed information is presented in

Appendix D. It should be noted that since Eviews reads codes only in the text format,

the program can be written either in the Eviews environment directly or in Microsoft

Word but then saved as a text file with a “prg” extension, which stands for EViews

program.

a. The Create Command

• Whenever one runs an EViews program, a workfile will be created, which

contains data we used and all results created by the program. Detailed information

on workfiles is given in Appendix D. In order to create this workfile, one uses the

create command. The general syntax for this command is as follows:

create workfile_name frequency start end

• Any workfile name can be chosen. The frequency of data can be annual,

monthly, etc or undated. While “start” specifies the starting date of the data, “end” is

the last year in our data file. In the program, this command is coded as follows create Niger-simulation U 17

• Here “Niger-simulation” is chosen as name of our workfile, which will be

created by our simulation program after we run it. “U” stands for undated data

23

frequency. Since our data file and the simulation program covers the years starting in

1999 up to 2015, the number of observations is 17.

b. The smpl Command

• The “smpl”, which stands for sample, command specifies the time period that

we are working on. It is generally used after the “create” command. The general

syntax of this command:

smpl sample_name start end

• It is optional to name your sample. The sample range must be given using the

starting and ending dates. One example of “smpl” command in our code is

smpl 1 17

• Note that we have not given any sample name. This code specifies that we will

work with the sample covering the periods from 1 to 17. This means that all the

following calculations and simulations will be done for this period as far as we do not

change our sample range. Some of our calculations require a smaller sample range.

In this case, we redefine our sample range such as “smpl 3 17”.

c. The Read Command

• As it is specified before, we need to use an external data file. When this is the

case, we use the “read” command to import data from an external file. The general

syntax is:

read(options) path\file_name variable names

• After the “read” command, we have to specify our options. Our external data

file is an Excel file. Thus the options are defined in a way that the program is

24

importing data from an Excel file. We explain these options below. Then we have to

specify the location of our data file and its name. At the end, we write the names of

the variables that we want to import. The order of the names of the variables must

match with their order in the Excel file.

The “read” command, which imports exogenous variables in the program, is

specified as follows:

read(ae4, t, s=EXO) C:\Niger\Niger-Data.xls AID$ CG DB ER ERROR_OMM FP kappa kappa_edu kappa_hea kappa_inf kappa_oth LAND LE_G n PMstar PXstar RD RGstar RPstar tm UTR$ WG tmnew

• We want to import data from an excel file; thus we need to specify our options

accordingly. Options will be presented within the parenthesis.

a) Provide information about the coordinates of the upper left cell of the data

matrix (excluding names and other definitions) in the Excel spreadsheet. In our

example, c4 stands for the cell number, at which point data that we want to import

starts.

b) Write “t ” when our data series are in rows rather than in columns. In this

columns. In this way, the observations will be read in rows. s=sheet_name option

shows the sheet in the Excel workbook from which data is to be read. Thus, s=EXO

means that we want to import data from the sheet named “EXO”. The location of the

data file is “C:\Niger\” and the name of the file is “Niger-Data.xls”.

c) List the name of variables that will be imported. These will be the names

that will be used if we need to refer to them in our program. It should be noted that

the order of the variables must follow that of the listing of the variables in the Excel

file. Figure 3 shows how the EXO sheet of our data file looks like. The projected data

are highlighted in blue.

• We follow exactly the same procedure to import endogenous variables using

this command: read(ae4, t, s=ENDO) C:\Niger\Niger-Data.xls AID CP DdebtG delta_LE_N Delta_NFA DITAX DITXR DOM FdebtG FdebtP FdebtTot FG GBAL GTOT IG IGedu IGhea IGinf IGoth INDTAX INDTXR IP J KGedu KGhea KGinf KGoth KP LE LE_P LR M NGDP PD PM POP PQ PQT PX PY Q Qd SP T TAX X Y Ydisp YTOT Z KGZ

25

• Now, we refer to the sheet name “ENDO” and we have a new list of variables.

Figure 4 shows how the ENDO sheet of our data file looks like. Since these variables

will be determined within the model after 2004, we do not project them.

d. The Write Command

• The “write” command exports variables from EViews into an external file. The

way it is coded is quite similar to the “read” command. The general syntax:

write(options) path\file_name variable names

This command appears in our code as follows

smpl 3 17 'baseline' write(t=xls,b4,t) C:\Niger\output-niger.xls AID_0 CG_0 CP_0 PQ_0 DdebtG_0 DOM_0 ER FdebtG_0 FdebtP_0 FdebtTot_0 GBAL_0 GTOT_0 IG_0 IGedu_0 IGhea_0 IGinf_0 IP_0 J_0 KGedu_0 KGhea_0 KGinf_0 KP_0 LE_0 LR_0 M_0 PD_0 PM_0 POP_0 PX_0 PY_0 Qd_0 Q SP_0 T_0 TAX_0 X_0 Ydisp_0 Y_0 kappa_edu kappa_hea kappa_inf DB_0 FG_0 FP LAND n PMstar PXstar RD RGstar RPstar tm UTR$ Delta_NFA_0 ERROR_OMM_0 NGDP_0 KGedu_0 LE_P_0 LE_G_0 DITXR_0 INDTXR_0 YTOT_0 kappa_oth WG_0 IGoth_0 DITAX_0 INDTAX_0 PQT_0 Z_0 KGZ_0

• Here we want to export variables for the period 3 to 17 corresponding to the

years from 2001 to 2015. The first of the options specified in the code is t=xls, which

means that the type of file, in which we want to write the outcome, is an Excel file.

Then we write the coordinate of the cell, at which the exported series will start in the

output file. If we want to export our series in rows, we have to include “t” while

defining our options. Then we specify the desired location of the file that will be

created and the name of the file. If this file does not exist, the program is going to

create it automatically. On the other hand, if it already exists, the program is going to

overwrite it. At the end, we list the names of the series that we want to export. Figure

4 shows an output file.

e. The Model Command

26

• The “model” command creates a model. We have to declare our model before

we start coding our equations. The general syntax:

model model_name

In our simulation program, the name of our model is “Niger” and it is declared

as follows

model Niger

f. The Append Command

• We use the “append” command to specify our equations. The general syntax

is:

model_ name.append equation

• For instance, population is defined by the following equation in the program:

niger.append POP = (1+n)*POP(-1)

• It specifies that the equation POP = (1+n)*POP(-1) is going to be added to the

model “Niger”.

g. The Solve Command

• The “solve” command triggers Eviews to solve a model. While running this

command, the Eviews will find a solution to a simultaneous equation model using

available data. This command needs to be placed after equations are listed. The

general syntax is:

solve(options) model_name

27

• There are many options that we can use within the “solve” command. Details

are given in Appendix D. In our simulation program, we use the default solution,

which is dynamic simulation.

• The name of the model that will be solved must be specified. The solution

method may be modified by changing the options. The code in the program is:

smpl 7 17

solve(m=100000, c=.001) niger

• We set our sample range between 7 and 17. This means that the model will be

solved for the years 2005 to 2015. “m = integer ” indicates the maximum number of

iterations to be executed. “c = number ” specifies the convergence criterion for the

solution of the dynamic simulation. “Niger” is the name of our model.

h. The Statusline Command

• This command enables a message to be displayed on the status line at the

bottom of the Eviews window. The general syntax is:

statusline message

We use this command as follows

Statusline iteration number: !IDX

• This means that the current iteration number for the current period will be

written as Eviews runs the simulation program.

i. The genr Command

• This command generates new series, which are calculated using available

series. The general syntax is:

28

genr ser_name = expression

• For instance, the following statement generates the T series using different

series and parameters:

genr T = AT*(beta_T*(LE_P^(-rho_T)) + (1 - beta_T)*((Kghea/(POP^theta_H))^(-rho_T)))^(-(1/rho_T))

4. Details about the Excel Output File

• All simulated endogenous variables and exogenous variables projected within

the program are named with “_0” extension. For example, AID_0 is the simulated AID

series. All exogenous variables and historical endogenous variables preserve their

original names; they do not take any extension. The output file is created for the

period from 3 to 17. This corresponds to the years from 2001 to 2015. The historical

values of the variables will be reported between 2001 and 2004. After these years,

the simulated values of endogenous variables and the projected values of the

exogenous variables will be presented.

5. Details about the Summary Table File

• This table is directly linked to the “Niger-Output.xls” file. It has to be updated if

we have a new output file. In order to update this table, the file must be opened and

then the “update” option must be chosen when the Excel program asks whether you

want to update this file or not. In this table, variables are presented either in levels (in

millions of LCU) or in percent of other variables, especially in percent of NGDP.

VIII. Simulating Shocks

29

• This section explains how we can implement shock in the simulation program.

In Pinto Moreira and Bayraktar (2005), there are three different types of shocks introduced:

1. Shock 1: Permanent increase in the ratio of foreign aid to GDP by 5 percent.

2. Shock 2: 12 percentage point reduction in investment in “other” category, which is fully reallocated to investment in infrastructure.

3. Shock 3: Permanent cut of 10 percentage point in the effective tariff rate.

a. Case 1 - The Non-Neutral Case: No change in the indirect and direct tax rates.

b. Case 2 - The Neutral Case (Adjustment in Direct Taxation): the effect of the tariff cut on revenue is offset, ex ante (that is, at

initial baseline values), by an increase in direct taxation.

c. Case 3 - The Neutral Case (Adjustment in Indirect Taxation): the effect of the tariff cut on revenue is offset, ex ante (that is, at

initial baseline values), by an increase in indirect domestic

taxation.

• In order to run these shocks we have to make some simple changes in the

simulation program. All we need to do is to open some of the lines in the program,

which need to be closed during the baseline simulation, and to close some of the

lines if they will not be used while running the program to investigate the effects of

the shocks. In order to open a line in EViews, all we need to do is to remove the “ ' ” sign at the beginning of the line. We do the opposite to close a line: add “ ' ” at the beginning of the line. In this way, the program is not going to read these lines when it

is executed.

SHOCK 1 - Increase in Foreign Aid

30

• Our first shock on the economy is a permanent increase in the aid-to-GDP

ratio by 5 percentage points (i.e. 0.05). The original value of the ratio of aid to GDP

was 10.67 %. Its value will increase to 10.67% + 5% = 15.67% after the shock to aid

is introduced. In order to apply this shock:

a) Open the line named as “line1shock1” in the simulation program under the

SHOCKS section. When we open this line, EViews reads this line. The following

example shows how we can open this line.

A CLOSED Line 'line1shock1'scalar AID_const = 0.1067 + 0.05 How to OPEN the line 'line1shock1' scalar AID_const = 0.1067 + 0.05

How to re-close the line 'line1shock1'scalar AID_const = 0.1067 + 0.05

• In order to prevent the program from overwriting on the existing output file, we

assign a new name to the output file that will be generated after we run the program.

In order to do this, we need to open the following line:

'line2shock1'write(t=xls,b4,t) C:\Niger\output-niger-shock1.xls AID_0 CG_0 CP_0 PQ_0 DdebtG_0 DOM_0 ER FdebtG_0 FdebtP_0 FdebtTot_0 GBAL_0 GTOT_0 IG_0 IGedu_0 IGhea_0 IGinf_0 IP_0 J_0 KGedu_0 KGhea_0 KGinf_0 KP_0 LE_0 LR_0 M_0 PD_0 PM_0 POP_0 PX_0 PY_0 Qd_0 Q SP_0 T_0 TAX_0 X_0 Ydisp_0 Y_0 kappa_edu kappa_hea kappa_inf DB_0 FG_0 FP LAND n PMstar PXstar RD RGstar RPstar tm UTR$ Delta_NFA_0 ERROR_OMM_0 NGDP_0 KGoth_0 LE_P_0 LE_G_0 DITXR_0 INDTXR_0 YTOT_0 kappa_oth WG_0 IGoth_0 DITAX_0 INDTAX_0 PQT_0 Z_0 KGZ_0

The solution to our model will be stored in the “output-niger-shock1.xls” output file.

• Since we will create a new output file, we have to close the line containing the

original “write” command. In EViews, lines are closed adding “ ‘ ” at the beginning of

the line. After you close this line, it looks like as follows:

‘write(t=xls,b4,t) C:\Niger\output-niger.xls AID_0 CG_0 CP_0 PQ_0 DdebtG_0 DOM_0 ER FdebtG_0 FdebtP_0 FdebtTot_0 GBAL_0 GTOT_0 IG_0 IGedu_0 IGhea_0 IGinf_0 IP_0 J_0 KGedu_0 KGhea_0 KGinf_0 KP_0 LE_0 LR_0 M_0 PD_0 PM_0 POP_0 PX_0 PY_0 Qd_0 Q SP_0 T_0 TAX_0 X_0 Ydisp_0

31

Y_0 kappa_edu kappa_hea kappa_inf DB_0 FG_0 FP LAND n PMstar PXstar RD RGstar RPstar tm UTR$ Delta_NFA_0 ERROR_OMM_0 NGDP_0 KGoth_0 LE_P_0 LE_G_0 DITXR_0 INDTXR_0 YTOT_0 kappa_oth WG_0 IGoth_0 DITAX_0 INDTAX_0 PQT_0 Z_0 KGZ_0

• After these changes are made, the simulation program must be saved before

we run it. A new output file will be created which will store the new values of

simulated variables after the shock. The results are shown in Table 4 in Appendix G.

In this file, the results are displayed as percentage changes (for variables in levels) or

absolute differences (for variables already in percentage form) from the baseline

scenario.

Note: The lines that we had opened or closed to run the shocks have to be re-closed

or re-opened again after the output file is created.

• Percentage change = 100*(new value – original value)/original value.

Example: From the baseline table (See Table 3 in Appendix G), gross

domestic product at market prices is simulated as 5246.2 billion CFA Franc in

2015 and it is equal to 8116.3 billion CFA Franc in 2015 when there is a 5%

increase in aid to GDP ratio, the first shock.

Percentage change = 100*(8116.3 -5246.2)/ 5246.2 = 54.71%

• Absolute Deviation from the baseline = absolute value of (New value –

Original value).

Example: From the baseline table, the current account balance is simulated as

2.4% (in percent of GDP) in 2015 and it is equal to 5.8% (in percent of GDP) in

2015 when there is a 5% increase in aid to GDP ratio, the first shock.

Absolute deviation from the baseline = absolute value of (5.8%-2.4%) = 3.39%

32

SHOCK 2 - Reallocation of Public Investment

• The second shock is a 20 percentage point reduction in investment in “other”

categories, which is fully reallocated to investment in infrastructure. In order to run

this shock, all we need to do is to open some of the lines in the simulation program.

The lines that need to be opened in the program are:

'line1shock2'smpl 6 17 'line2shock2'genr kappa_oth = kappa_oth - 0.20 'line3shock2'genr kappa_inf = kappa_inf + 0.20

'line4shock2'write(t=xls,b4,t) C:\Niger\output-niger-shock2.xls AID_0 CG_0 CP_0 PQ_0 DdebtG_0 DOM_0 ER FdebtG_0 FdebtP_0 FdebtTot_0 GBAL_0 GTOT_0 IG_0 IGedu_0 IGhea_0 IGinf_0 IP_0 J_0 KGedu_0 KGhea_0 KGinf_0 KP_0 LE_0 LR_0 M_0 PD_0 PM_0 POP_0 PX_0 PY_0 Qd_0 Q SP_0 T_0 TAX_0 X_0 Ydisp_0 Y_0 kappa_edu kappa_hea kappa_inf DB_0 FG_0 FP LAND n PMstar PXstar RD RGstar RPstar tm UTR$ Delta_NFA_0 ERROR_OMM_0 NGDP_0 KGoth_0 LE_P_0 LE_G_0 DITXR_0 INDTXR_0 YTOT_0 kappa_oth WG_0 IGoth_0 DITAX_0 INDTAX_0 PQT_0 Z_0 KGZ_0

• The new values of the capital shares will be used only during the period that we

simulate the model. It corresponds to the sample period from 7 to 17 (the years

between 2004 and 2015). “genr kappa_oth = kappa_oth - 0.20” generates the new

public investment share in “other” capital, which is 20% less than the original level.

This drop in the share of public investment in “other” will be allocated to public

investment in infrastructure. Thus its value will be 20% higher. The new share

parameter is defined as “genr kappa_inf = kappa_inf + 0.20”.

• The location of the last line is at the end of the program. The output file that

will be created after we run the program will be “output-niger-shock2.xls”. The results

of this shock are presented in Table 5 in Appendix G. Since the output will be

exported to a new output file, we must close the line containing the original write

command. This can be achieved by adding “ ‘ ” at the beginning of the line which

writes the output in “output-niger.xls” file.

When opening these lines, the code will look like as follows:

'line1shock2' smpl 6 17

33

'line2shock2' genr kappa_oth = kappa_oth - 0.20 'line3shock2' genr kappa_inf = kappa_inf + 0.20

'line4shock2' write(t=xls,b4,t) C:\Niger\output-niger-shock2.xls AID_0 CG_0 CP_0 PQ_0 DdebtG_0 DOM_0 ER FdebtG_0 FdebtP_0 FdebtTot_0 GBAL_0 GTOT_0 IG_0 IGedu_0 IGhea_0 IGinf_0 IP_0 J_0 KGedu_0 KGhea_0 KGinf_0 KP_0 LE_0 LR_0 M_0 PD_0 PM_0 POP_0 PX_0 PY_0 Qd_0 Q SP_0 T_0 TAX_0 X_0 Ydisp_0 Y_0 kappa_edu kappa_hea kappa_inf DB_0 FG_0 FP LAND n PMstar PXstar RD RGstar RPstar tm UTR$ Delta_NFA_0 ERROR_OMM_0 NGDP_0 KGoth_0 LE_P_0 LE_G_0 DITXR_0 INDTXR_0 YTOT_0 kappa_oth WG_0 IGoth_0 DITAX_0 INDTAX_0 PQT_0 Z_0 KGZ_0

• After the program is run, we need to close the lines that we had opened, and

open the lines that we had closed.

SHOCK 3 - Reduction in Tariffs

• In this section, the aim is to study the impact of a permanent cut of 12

percentage point in Niger’s effective tariff rate. We first examine the case where the

cut is non-neutral. We then study the case where the authorities offset the adverse

revenue effect of lower tariffs by either an increase in direct or indirect taxes.

The Non-Neutral Case – Shock 3a

• The purpose is to investigate the response of the economy to a drop in tariff

rate. In the non-neutral case, we keep the direct and indirect tax rates at their year

2004 values throughout the simulation process. We need to slightly change the

simulation program to run this shock. First of all, the following lines in the program

must be opened:

'---------------------------------------------------------------------------------------------------------- 'Shock to tm (DITXR and INDTXR fixed) '---------------------------------------------------------------------------------------------------------- 'line1shock3a'smpl 7 17 'line2shock3a'genr tm = tmnew 'line3shock3a'scalar DITXR_ALT = 0.019826305 'for tm shock 'line4shock3a'scalar INDTXR_ALT = 0.023180876 'for tm shock '---------------------------------------------------------------------------------------------------------- 'Shock to tm (DITXR and INDTXR fixed) '---------------------------------------------------------------------------------------------------------- 'NOTE: Before running this shock, don't forget to close the DITXR and INDTXR equations above.

34

'line5shock3a'niger.append DITXR = DITXR_ALT 'line6shock3a'niger.append INDTXR = INDTXR_ALT 'line7shock3a'write(t=xls,b4,t) C:\Niger\output-niger-shock3a.xls AID_0 CG_0 CP_0 PQ_0 DdebtG_0 DOM_0 ER FdebtG_0 FdebtP_0 FdebtTot_0 GBAL_0 GTOT_0 IG_0 IGedu_0 IGhea_0 IGinf_0 IP_0 J_0 KGedu_0 KGhea_0 KGinf_0 KP_0 LE_0 LR_0 M_0 PD_0 PM_0 POP_0 PX_0 PY_0 Qd_0 Q SP_0 T_0 TAX_0 X_0 Ydisp_0 Y_0 kappa_edu kappa_hea kappa_inf DB_0 FG_0 FP LAND n PMstar PXstar RD RGstar RPstar tm UTR$ Delta_NFA_0 ERROR_OMM_0 NGDP_0 KGoth_0 LE_P_0 LE_G_0 DITXR_0 INDTXR_0 YTOT_0 kappa_oth WG_0 IGoth_0 DITAX_0 INDTAX_0 PQT_0 Z_0 KGZ_0

• Since the shock is applied only during the simulation period, we set the sample

range between 7 and 17. “genr tm = tmnew” defines our new tariff rate, which is equal to

the half of the old tariff rate. “scalar DITXR_ALT = 0.019826305” and “scalar INDTXR_ALT =

0.023180876” guarantee that the direct and indirect tax rates will be kept at their original

levels in 2004.

• We add two new equations to our model: “niger.append DITXR = DITXR_ALT” and

“niger.append INDTXR = INDTXR_ALT”. These equations specify that the direct and

indirect tax rates will be kept constant at their original levels throughout the

simulation process.

• The last line in the set writes the results into the “output-Niger-shock3a.xls” file.

It should be noted that the locations of the lines in this set are different. The first set

of lines is located in the “shocks” section of the program. The ones in the middle are

located in the section we define the equations. The last line is at the end of the

program. After we open these lines, they will look like as follows

'---------------------------------------------------------------------------------------------------------- 'Shock to tm (DITXR and INDTXR fixed) '---------------------------------------------------------------------------------------------------------- 'line1shock3a' smpl 7 17 'line2shock3a' genr tm = tmnew 'line3shock3a' scalar DITXR_ALT = 0.019826305 'for tm shock 'line4shock3a' scalar INDTXR_ALT = 0.023180876 'for tm shock '---------------------------------------------------------------------------------------------------------- 'Shock to tm (DITXR and INDTXR fixed) '---------------------------------------------------------------------------------------------------------- 'NOTE: Before running this shock, don't forget to close the DITXR and INDTXR equations above.

35

'line5shock3a' niger.append DITXR = DITXR_ALT 'line6shock3a' niger.append INDTXR = INDTXR_ALT 'line7shock3a' write(t=xls,b4,t) C:\Niger\output-niger-shock3a.xls AID_0 CG_0 CP_0 PQ_0 DdebtG_0 DOM_0 ER FdebtG_0 FdebtP_0 FdebtTot_0 GBAL_0 GTOT_0 IG_0 IGedu_0 IGhea_0 IGinf_0 IP_0 J_0 KGedu_0 KGhea_0 KGinf_0 KP_0 LE_0 LR_0 M_0 PD_0 PM_0 POP_0 PX_0 PY_0 Qd_0 Q SP_0 T_0 TAX_0 X_0 Ydisp_0 Y_0 kappa_edu kappa_hea kappa_inf DB_0 FG_0 FP LAND n PMstar PXstar RD RGstar RPstar tm UTR$ Delta_NFA_0 ERROR_OMM_0 NGDP_0 KGoth_0 LE_P_0 LE_G_0 DITXR_0 INDTXR_0 YTOT_0 kappa_oth WG_0 IGoth_0 DITAX_0 INDTAX_0 PQT_0 Z_0 KGZ_0

• In addition to these changes, the line containing the original “write” command

must be closed, as we did before, since we will create a new output file to study the

effects of the shock. This can be achieved by adding “ ‘ ” at the beginning of the line.

• The results are presented in Table 6 in Appendix G. After we obtain this output

file, we need to close the lines that we had opened and open the lines that we had

closed.

The Neutral Case: Adjustment in Direct Taxation – Shock 3b

• In this scenario, we consider the case where the effect of the tariff cut on

revenue is offset by an increase in direct taxation. Given the magnitude of the

reduction in the effective tariff rate, this requires an increase in the direct tax rate of

4.89 percentage points.

• The aim is to keep the total tax revenue and total indirect tax revenue fixed

and to adjust the indirect tax rate in a way to compensate any tax loss caused by the

lower tariff rate. The new direct tax rate is calculated using the following equation:

New direct tax rate = (TAX – INDTAX – (tm/2).ER.M.PM*)/YTOT

The values of the variables are giving in the following table.

36

2004TAX 1.717E+11INDTAX 42670999624tm 0.239834512ER 528.2848PMstar 0.000713135M 1.06446E+12YTOT 1.65688E+12

Alternative direct tax rate 0.049

• The preparation of the program to run this shock is quite similar to the

changes that we made to run the previous shocks.

• In the “Shocks to tm and DITXR” section (given below), the last two lines must

be opened. The first line helps us reduce the value of the tariff rate to half of its

original value. The last line increases the direct tax rate to 4.89% in order to

compensate the reduction in tax revenue caused by decreased tariff rates.

'---------------------------------------------------------------------------------------------------------- 'Shock to tm and DITXR (INDTXR fixed) '---------------------------------------------------------------------------------------------------------- 'NOTE: When tm drops to tmnew, the new value of DITXR must be equal 'to 0.048850383, keeping total tax revenue and INDTXR fixed. 'It is calculated for 2004. 'line1shock3b'genr tm = tmnew 'line2shock3b'scalar DITXR_ALT = 0.048850383

• We also need to open the last line of the following section. This line equates

the value of DITXR to its new higher value.

'---------------------------------------------------------------------------------------------------------- 'Shock to tm and DITXR (INDTXR fixed) '---------------------------------------------------------------------------------------------------------- 'NOTE: Before running this shock, don't forget to close the DITXR equation above. 'line3shock3b'niger.append DITXR = DITXR_ALT

• Since DITXR is defined using the new equation specified above, we need to

close the line containing the original equation determining DITXR. The line that we

need to close is:

37

niger.append DITXR = DITXR_const

• Its location is the section, in which we list the equations. In addition to these

changes, we must also open the line that generates our new output file, which is

named as “output-Niger-shock3b.xls”.

'line4shock3b'write(t=xls,b4,t) C:\Niger\output-niger-shock3b.xls AID_0 CG_0 CP_0 PQ_0 DdebtG_0 DOM_0 ER FdebtG_0 FdebtP_0 FdebtTot_0 GBAL_0 GTOT_0 IG_0 IGedu_0 IGhea_0 IGinf_0 IP_0 J_0 KGedu_0 KGhea_0 KGinf_0 KP_0 LE_0 LR_0 M_0 PD_0 PM_0 POP_0 PX_0 PY_0 Qd_0 Q SP_0 T_0 TAX_0 X_0 Ydisp_0 Y_0 kappa_edu kappa_hea kappa_inf DB_0 FG_0 FP LAND n PMstar PXstar RD RGstar RPstar tm UTR$ Delta_NFA_0 ERROR_OMM_0 NGDP_0 KGoth_0 LE_P_0 LE_G_0 DITXR_0 INDTXR_0 YTOT_0 kappa_oth WG_0 IGoth_0 DITAX_0 INDTAX_0 PQT_0 Z_0 KGZ_0

As we did before, we have to close the line containing our original “write” command in

order to prevent the program from overwriting the baseline output file.

• The results are presented in Table 7 in Appendix G.

After we obtain our results, all the changes that we have done must be adjusted

to their original setup.

The Neutral Case: Adjustment in Indirect Taxation – Shock 3c

• In this final shock scenario, we consider the case where the effect of the tariff

cut on revenue is offset by an increase in indirect domestic taxation. Given the

magnitude of the reduction in the effective tariff rate, this requires a 4.93 percentage

point increase in the indirect tax rate on domestic sales of goods and services.

• The aim is to keep the total tax revenue and total direct tax revenue fixed and

to adjust the direct tax rate in a way to compensate any tax loss caused by the lower

tariff rate. The new indirect tax rate is calculated using the following equation:

New indirect tax rate = (TAX – DITAX – (tm/2).ER.M.PM*)/(PQ.Q)

38

The values of the variables are giving in the following table.

2004

TAX 1.717E+11DITAX 32849901454tm 0.239834512ER 528.2848PMstar 0.000713135M 1.06446E+12YTOT 1.65688E+12

Alternative indirect tax rate 0.049

• The preparation of the program to run this shock is quite similar to the

changes that we made in the model to run the previous shocks.

• In the “Shocks to tm and INDTXR” section (given below), the last two lines

must be opened. These two lines help us reduce the value of the tariff rate to the half

of its original value. The last line increases the indirect tax rate to 4.93% in order to

compensate the reduction in tax revenue caused by decreased tariff rates.

'---------------------------------------------------------------------------------------------------------- 'Shock to tm and INDTXR (DITXR fixed) '---------------------------------------------------------------------------------------------------------- 'NOTE: When tm drops to tmnew, the new value of INDTXR must be equal 'to 0.049305362, keeping total tax revenue and DITXR fixed. 'It is calculated for 2004. 'line1shock3c'genr tm = tmnew 'line2shock3c'scalar INDTXR_ALT = 0.049305362

• We also need to open the last line of the following section. This line equates

INDTXR to its original value.

'---------------------------------------------------------------------------------------------------------- 'Shock to tm and INDTXR (DITXR fixed) '---------------------------------------------------------------------------------------------------------- 'NOTE: Before running this shock, don't forget to close the INDTXR equation above. 'line3shock3c'niger.append INDTXR = INDTXR_ALT

39

• Since INDTXR is redefined using this new equation, we have to close the line

containing the original equation determining INDTXR. The line that we need to

close is:

niger.append INDTXR = 0.00792+0.706572*INDTXR(-1)-0.029834*(AID/NGDP)+ INDTXR_RES

The location of this line is the section in which we list the equations.

In addition to these changes, we must also open the line that creates our new

output file (“output-Niger-shock3c.xls”).

'line4shock3c''write(t=xls,b4,t) C:\Niger\output-niger-shock3c.xls AID_0 CG_0 CP_0 PQ_0 DdebtG_0 DOM_0 ER FdebtG_0 FdebtP_0 FdebtTot_0 GBAL_0 GTOT_0 IG_0 IGedu_0 IGhea_0 IGinf_0 IP_0 J_0 KGedu_0 KGhea_0 KGinf_0 KP_0 LE_0 LR_0 M_0 PD_0 PM_0 POP_0 PX_0 PY_0 Qd_0 Q SP_0 T_0 TAX_0 X_0 Ydisp_0 Y_0 kappa_edu kappa_hea kappa_inf DB_0 FG_0 FP LAND n PMstar PXstar RD RGstar RPstar tm UTR$ Delta_NFA_0 ERROR_OMM_0 NGDP_0 KGoth_0 LE_P_0 LE_G_0 DITXR_0 INDTXR_0 YTOT_0 kappa_oth WG_0 IGoth_0 DITAX_0 INDTAX_0 PQT_0 Z_0 KGZ_0

One should not forget to close the line containing the original write command.


As it is noted before, after the new output file is created, all the changes that

we have done must be turned back to their original setup.

IX. SENSITIVITY ANALYSIS

In this case, we assume that public investment is partially efficient such that αh

is equal to 0.5. In order to obtain the baseline results, open the following lines in the

program:

'line1baseline 2'scalar alpha_h = 0.5 'line2baseline2'write(t=xls,b4,t) C:\Niger\OUTPUT-NIGER4.xls AID_0 CG_0 CP_0 PQ_0 DdebtG_0 DOM_0 ER FdebtG_0 FdebtP_0 FdebtTot_0 GBAL_0 GTOT_0 IG_0 IGedu_0 IGhea_0 IGinf_0 IP_0 J_0 KGedu_0 KGhea_0 KGinf_0 KP_0 LE_0 LR_0 M_0 PD_0 PM_0 POP_0 PX_0 PY_0 Qd_0 Q SP_0 T_0 TAX_0 X_0 Ydisp_0 Y_0

40

kappa_edu kappa_hea kappa_inf DB_0 FG_0 FP LAND n PMstar PXstar RD RGstar RPstar tm UTR$ Delta_NFA_0 ERROR_OMM_0 NGDP_0 KGedu_0 LE_P_0 LE_G_0 DITXR_0 INDTXR_0 YTOT_0 kappa_oth WG_0 IGoth_0 DITAX_0 INDTAX_0 PQT_0 Z_0 KGZ_0



Similarly, in order to run a aid shock (5 percent increase in aid to GDP)

together with partially efficient IG, one needs to open the following lines in the

program:

'line1shock4'scalar alpha_h = 0.5 'line2shock4'scalar AID_const = 0.1067 + 0.05 'line3shock4'write(t=xls,b4,t) C:\Niger\output-niger-shock4.xls AID_0 CG_0 CP_0 PQ_0 DdebtG_0 DOM_0 ER FdebtG_0 FdebtP_0 FdebtTot_0 GBAL_0 GTOT_0 IG_0 IGedu_0 IGhea_0 IGinf_0 IP_0 J_0 KGedu_0 KGhea_0 KGinf_0 KP_0 LE_0 LR_0 M_0 PD_0 PM_0 POP_0 PX_0 PY_0 Qd_0 Q SP_0 T_0 TAX_0 X_0 Ydisp_0 Y_0 kappa_edu kappa_hea kappa_inf DB_0 FG_0 FP LAND n PMstar PXstar RD RGstar RPstar tm UTR$ Delta_NFA_0 ERROR_OMM_0 NGDP_0 KGedu_0 LE_P_0 LE_G_0 DITXR_0 INDTXR_0 YTOT_0 kappa_oth WG_0 IGoth_0 DITAX_0 INDTAX_0 PQT_0 Z_0 KGZ_0



As it is noted before, after the new output file is created, all the changes that

we have done must be turned back to their original setup.

X. LINKING THE MODEL WITH THE MILLENNIUM DEVELOPMENT GOALS (MDGS)

This section explains how we can link the macroeconomic framework to the

MDGs.8 It is also shown how we can run the simulation program to create simulated

data files which are used to construct the MDG tables.

8 Details are given in Agenor, Bayraktar, Pinto Moreira, and El Aynaoui (2005).

41

Six of the MDG indicators are integrated: the poverty rate, the literacy rate,

infant mortality, malnutrition, life expectancy, and access to safe water. Because the

model can directly calculate values for the poverty and the literacy rates, we only ran

regressions to estimate the equations for infant mortality, malnutrition, life

expectancy, and access to safe water. The estimation method is ordinary least

squares. We use cross-section data, obtained by taking average values of variables

for each country for the period 1965-2003, depending on the availability of data

series. Our sample consists of Sub-Saharan countries. The regression results are

presented in the following table.

42

Cross-Section Regression Results(All sub-Saharan countries are included unless otherwise indicated)

Dependent variables MALNUTRITION ln(MORTALITY) ln(LIFE_EXP) WATER 2/ Constant term 75.415

(6.055) 5.485 (10.761)

3.428 (27.187)

6.711 (0.299)

HEA_P_GDP 1/ -4.790 (-3.961)

-0.091 (-1.949)

0.048 (2.802)

…

ln(CPPC2003$) -7.951 (-4.126)

… … …

POVERTY 0.144 (1.635)

0.011 (3.247)

-0.002 (-2.771)

…

ln(GDPPC2003$) … -0.191 (-2.820)

0.078 (4.189)

6.921 (2.458)

INF_GDP … … … 1.702 (1.718)

ln(POP_DENSITY) … … … 4.076 (1.551)

Number of observations

28 31 20 31

Adjusted R2 0.552 0.479 0.739 0.292 Note: The estimation technique is OLS. Data points of independent variables in each country correspond exactly to the years in which dependent variables are available. First, averages at the country level are calculated, then the regression equations are run using these cross sectional data. t-statistics are reported in parenthesis. MALNUTRITION is malnutrition prevalence, weight for age (% of children under 5); HEA_P_GDP is public health expenditure in % of GDP; CPPC2003$ is private consumption per capita (in constant 2003 dollars); POVERTY is the percent of population living under $2 per day; MORTALITY is infant mortality rate (per 1000 live births); GDPPC2003$ is GDP per capita (in constant 2003 dollars); LIFE_EXP is life expectancy at birth, total, years; INF_GDP is public infrastructure expenditure in percent of GDP; WATER is percentage of population with access to safe water; POPDEN is population density (people per km square). 1/ While the data source of public heath expenditure is Government Financial Statistics in the life-expectancy regression, the data source of public heath expenditure is World Bank African Database in other regressions. 2/ Due to insufficient number of data points for sub-Saharan African countries, all developing countries are included depending on data availability.

After estimating these coefficients, we calculate the residuals of each

regression equation for Niger, which are going to be used in calculating predicted

values of the MDG indicators. For example, in case of malnutrition prevalence, the

estimated equation is:

ACTUAL value of MALNUTRITION = - 4.79*(HEA_P_GDP) -

7.951*LN(CPPC2003$) + 0.144* POVERTY + MALNUTRITION RESIDUAL

43

where MALNUTRITION is malnutrition prevalence, weight for age (% of children

under 5); HEA_P_GDP is public health expenditure in % of GDP; CPPC2003$ is

private consumption per capita (in constant 2003 dollars); POVERTY is the percent

of population living under $2 per day. MALNUTRITION was 40.1 percent in Niger in

2000 (the latest available data point); HEA_P_GDP was 1.28; CPPC2003$ was

$175; and POVERTY was 63 percent. Plugging these numbers in the equation given

above, we can calculate the residual as

MALNUTRITION RESIDUAL =40.1+4.79*(1.2801)+7.951*LN(175.0706)-0.144*63

Then, we can calculate the predicted values of MALNUTRITION by using the

simulated data as follows:

PREDICTED VALUE OF MALNUTRITION = -4.79*(100*IGhea*PQT/NGDP) -

7.951*LN([CP/(ER in 2003)]/POP) + 0.144*POVERTY + MALNUTRITION RESIDUAL

where POVERTY is defined with the partial elasticity of -1.0.

Similarly, we calculate the residual of infant mortality for Niger by using the

following equation:

LN(ACTUAL value of INFANT MORTALITY) = - 0.091*(HEA_P_GDP) -

0.191*LN(GDPPC2003$) - 0.011* POVERTY + INFANT MORTALITY RESIDUAL

where MORTALITY is infant mortality rate (per 1000 live births) and GDPPC2003$ is

GDP per capita (in constant 2003 dollars). INFANT MORTALITY was 155 in Niger in

2002 (the latest available data point); HEA_P_GDP was 0.84; GDPPC2003$ was

$220; and POVERTY was 63 percent. Plugging these numbers in the equation given

above, we can calculate the residual as

44

INFANT MORTALITY RESIDUAL=LN(155) +0.091*0.84+0.191*LN(220)-0.011*63

Then, we can calculate the predicted values of INFANT MORTALITY by using the

simulated data series. For example in 2005

PREDICTED VALUE OF INFANT MORTALITY = EXP(-

0.091*(100*IGhea*PQT/NGDP) -0.191*LN([(NGDP/PQT)/(ER in 2003)]/POP) +

0.011*POVERTY + INFANT MORTALITY RESIDUAL)


Similarly, the residual of life expectancy for Niger is:

LN(ACTUAL value of LIFE EXPECTANCY) = 0.048*(HEA_P_GDP) +

0.078*LN(GDPPC2003$) – 0.002*POVERTY + LIFE EXPECTANCY RESIDUAL

where LIFE_EXP is life expectancy at birth (total, years). LIFE EXPECTANCY was

46.19 in Niger in 2002 (the latest available data point); HEA_P_GDP was 0.84;

GDPPC2003$ was $220; and POVERTY was 63 percent. Plugging these numbers in

the equation given above, we can calculate the residual as

45

LIFE EXPECTANCY RESIDUAL= LN(46.19)-0.048*(0.84227)-

0.078*LN(219.6256)+0.002*63

Then, we can calculate the predicted values of LIFE EXPECTANCY by using the

simulated data as follows:

PREDICTED VALUE OF LIFE EXPECTANCY =

EXP(0.84227*(100*IGhea*PQT/NGDP)

+0.078*LN([(NGDP/PQT)/(ER in 2003)]/POP)

+ 0.002*POVERTY + LIFE EXPECTANCY RESIDUAL)


Access to safe water is calculated as follows. The residual of access to safe

water for Niger is:

(ACTUAL value of WATER) = 4.0767*LN(POPDEN) +

6.9219*LN(GDPPC2003$) + 1.7024*(INF_GDP) + WATER RESIDUAL

where INF_GDP is public infrastructure expenditure in percent of GDP, WATER is

percentage of population with access to safe water, and POPDEN is population

density (people per km square). WATER was 59 in Niger in 2000 (the latest available

data point); INF_GDP was 1.712; GDPPC2003$ was $212; and POPDEN was 8.48

per km square. Plugging these numbers in the equation given above, we can

calculate the residual as

46

WATER RESIDUAL= =59-4.0767*LN(8.48)-6.9219*LN(212)-1.7024*1.712

Then, we can calculate the predicted values of WATER by using the simulated

data as follows:

PREDICTED VALUE OF WATER =

1.7024*(100*IGinf*PQT/NGDP)

+6.9219*LN([(NGDP/PQT)/(ER in 2003)]/POP)

+ 4.0767*POPDEN + WATER RESIDUAL.

Calculation of the Composite MDG Indicator

This indicator summarizes the changes in the MDG indicators. It is calculated

by combining poverty (neutral case), literacy rate, life expectancy, access to safe

water, malnutrition prevalence, and infant mortality. The index is assumed to be

equal to 100 in 2005. An increase indicates an improvement. First of all, we calculate

the value of each MDG indicator with respect to their values in 2005.9 For example,

Relative poverty rate in 2006 = Poverty rate in 2006/Poverty rate in 2005.

It should be noted that while calculating the index, we take the inverse of

poverty, malnutrition prevalence, and infant mortality since an increase in indicators

are supposed to show an improvement toward achieving the MDGs.

After calculating the relative values of the MDG indicators, their geometric

average is the composite MDG indicator.

Creating Simulated Data Files

9 “Niger-MDG-table-baseline.xls” shows how this index is calculated.

47

The baseline MDG table is given in “Niger-MDG-table-baseline.xls” file. This

file is linked to “OUTPUT-NIGER.xls” and “Niger-Output Table-BASELINE-MDG.xls”

files.10 The baseline MDG table is given in Appendix I.

SHOCK 5 – Effects of an Increase in Foreign Aid on the MDG Indicators

• The first shock is a permanent increase in the aid-to-GDP ratio by 5

percentage points (i.e. 0.05) starting in 2006. The original value of the ratio of aid to

GDP is taken to be equal to 10.67 % up to 2006. Its value will increase to 10.67% +

5% = 15.67% in 2006. The variable introducing this shock in the simulation program

is AID_GDP_MDGSHOCK.

• In order to apply this shock, open the following lines in the simulation program.

'line 1 MDG AID SHOCK after 2005'niger.append AID = NGDP*AID_GDP_MDGSHOCK 'line 2 MDG AID SHOCK after 2005'write(t=xls,b4,t) C:\Niger\output-niger-shock5.xls AID_0 CG_0 CP_0 PQ_0 DdebtG_0 DOM_0 ER FdebtG_0 FdebtP_0 FdebtTot_0 GBAL_0 GTOT_0 IG_0 IGedu_0 IGhea_0 IGinf_0 IP_0 J_0 KGedu_0 KGhea_0 KGinf_0 KP_0 LE_0 LR_0 M_0 PD_0 PM_0 POP_0 PX_0 PY_0 Qd_0 Q SP_0 T_0 TAX_0 X_0 Ydisp_0 Y_0 kappa_edu kappa_hea kappa_inf DB_0 FG_0 FP LAND n PMstar PXstar RD RGstar RPstar tm UTR$ Delta_NFA_0 ERROR_OMM_0 NGDP_0 KGedu_0 LE_P_0 LE_G_0 DITXR_0 INDTXR_0 YTOT_0 kappa_oth WG_0 IGoth_0 DITAX_0 INDTAX_0 PQT_0 Z_0 KGZ_0

• In order to run this aid shock, the following line in the simulation program

should be closed as indicated in the program: niger.append AID = AID_const * NGDP



simulated data after the shock.

10 Details about how “OUTPUT-NIGER.xls” is constructed are given in Section VII.

48



• The output file, “OUTPUT-NIGER-SHOCK5.xls”, is used to create the

summary table, “Niger-Output Table-SHOCK5.xls”, and the MDG table, “Niger-MDG-

table-withaidshock.xls”. “Niger-output-Aid-Shock-MDG Table 1.xls” shows the

deviation from the summary baseline table.

• “Niger-MDG-table-withaidshock-deviation.xls” presents the deviation from the

baseline MDG table. This table is given in Appendix I.

SHOCK 6 – Effects of Cancellation of External Debt on the MDG Indicators

• The second shock is cancellation of external debt. For this experiment, we

assume that the outstanding stock of Niger’s external debt is cancelled in 2006, and

that in the following years “new” borrowing occurs only at a very low effective interest

rate, of 0.2 percent. We assume that the savings associated with lower interest

payments (which represent about 0.52 percent of GDP in 2006) are reallocated

entirely to public investment.

• This additional revenue is calculated as follows using the simulated data given

in “OUTPUT-NIGER.xls” file.

100*FdebtG in 2005*RG* in 2006*ER in 2006/NGDP in 2006 = 0.52 percent

• In order to run this shock, new variables are introduced in the simulation

program. IGRESIDUALFORDEBT is additional income from debt relief, which is

entirely allocated to public investment. DUMMY1 is a dummy variable which is equal

to 0 in 2006 (in which public external debt is cancelled) and 1 otherwise. It

guarantees that external public interest payment is going to be zero in 2006.

RGSTARlow series is equal to its original values up to 2006, but it is equal to 0.2

49

percent after that, corresponding to a lower effective interest rate on public external

debt.


'line 1 - debt relief shock'niger.append GTOT = WG*LE_G + PQT*(CG + IG) + DUMMY1*RGstarlow*ER*FdebtG(-1) + RD*DdebtG(-1) 'line 2 - debt relief shock'niger.append IG = (-0.174921+0.649353*(TAX(-1)/(NGDP(-1)))+1.549799*(AID/NGDP)-3.26115*(AID/NGDP)^2)*NGDP/PQT + IGRESIDUALFORDEBT 'line 3 - debt relief shock'niger.append delta_NFA = PXstar*X - PMstar*M - DUMMY1*RGstarlow*FdebtG(-1) - RPstar*FdebtP(-1) + UTR$ + (AID$ +AID_RES) + FG + FP+ERROR_OMM 'line 4 - debt relief shock'niger.append FdebtG = FG + DUMMY1*FdebtG(-1) 'line 5 - debt relief shock'write(t=xls,b4,t) C:\Niger\output-niger-shock6.xls AID_0 CG_0 CP_0 PQ_0 DdebtG_0 DOM_0 ER FdebtG_0 FdebtP_0 FdebtTot_0 GBAL_0 GTOT_0 IG_0 IGedu_0 IGhea_0 IGinf_0 IP_0 J_0 KGedu_0 KGhea_0 KGinf_0 KP_0 LE_0 LR_0 M_0 PD_0 PM_0 POP_0 PX_0 PY_0 Qd_0 Q SP_0 T_0 TAX_0 X_0 Ydisp_0 Y_0 kappa_edu kappa_hea kappa_inf DB_0 FG_0 FP LAND n PMstar PXstar RD RGstar RPstar tm UTR$ Delta_NFA_0 ERROR_OMM_0 NGDP_0 KGedu_0 LE_P_0 LE_G_0 DITXR_0 INDTXR_0 YTOT_0 kappa_oth WG_0 IGoth_0 DITAX_0 INDTAX_0 PQT_0 Z_0 KGZ_0

• In order to run this shock the following lines in the simulation program should

be closed as indicated in the program: 'NOTE: Close the following line when running the debt relief shock (Shock 6) niger.append GTOT = WG*LE_G + PQT*(CG + IG) + RGstar*ER*FdebtG(-1) + RD*DdebtG(-1)

'NOTE: Close the following line when running the debt relief shock (Shock 6) niger.append IG = (-0.174921+0.649353*(TAX(-1)/(NGDP(-1)))+1.549799*(AID/NGDP)-3.26115*(AID/NGDP)^2)*NGDP/PQT + IG_RES

'NOTE: Close the following line when running the debt relief shock (Shock 6) niger.append delta_NFA = PXstar*X - PMstar*M - RGstar*FdebtG(-1) - RPstar*FdebtP(-1) + UTR$ + (AID$ +AID_RES) + FG + FP+ERROR_OMM

'NOTE: Close the following line when running the debt relief shock (Shock 6) niger.append FdebtG = FG + FdebtG(-1)

'baseline' write(t=xls,b4,t) C:\Niger\output-niger.xls AID_0 CG_0 CP_0 PQ_0 DdebtG_0 DOM_0 ER FdebtG_0 FdebtP_0 FdebtTot_0 GBAL_0 GTOT_0 IG_0 IGedu_0 IGhea_0 IGinf_0 IP_0 J_0 KGedu_0 KGhea_0 KGinf_0 KP_0 LE_0 LR_0 M_0 PD_0 PM_0 POP_0 PX_0 PY_0 Qd_0 Q SP_0 T_0 TAX_0 X_0 Ydisp_0 Y_0 kappa_edu kappa_hea kappa_inf DB_0 FG_0 FP LAND n PMstar PXstar RD RGstar RPstar tm UTR$ Delta_NFA_0 ERROR_OMM_0

50

NGDP_0 KGedu_0 LE_P_0 LE_G_0 DITXR_0 INDTXR_0 YTOT_0 kappa_oth WG_0 IGoth_0 DITAX_0 INDTAX_0 PQT_0 Z_0 KGZ_0







summary table, “OUTPUT-NIGER-SHOCK6.xls”, and the MDG table, “Niger-MDG-

table-withdebtshock.xls”. “Niger-output-Debtrelief-Shock-Table 2.xls” presents the

deviation from the baseline summary table.

• “Niger-MDG-table-withdebtshock-deviation.xls” presents the deviation from the

baseline MDG table. This table is given in Appendix I.

SHOCK 7 – Effects of 0.52 percent Increase in Foreign Aid on the MDG Indicators

• This policy experiment aims to investigate the question of whether the impact

of debt relief compares favorably with a permanent increase in aid of the exact same

magnitude, of about 0.52 percentage of GDP, beginning also in 2006.

• Given that the original value of the ratio of aid to GDP is equal to 10.67 %, it

will increase to 10.67% + 0.52% = 11.19% in 2006. The variable introducing this

shock in the simulation program is AID_GDP_MDGSHOCK2.


51

'line 1 MDG AID SHOCK of 0.52 % after 2006'niger.append AID = NGDP*AID_GDP_MDGSHOCK2 'line 2 MDG AID SHOCK of 0.52 % after 2006' write(t=xls,b4,t) C:\Niger\output-niger-shock7.xls AID_0 CG_0 CP_0 PQ_0 DdebtG_0 DOM_0 ER FdebtG_0 FdebtP_0 FdebtTot_0 GBAL_0 GTOT_0 IG_0 IGedu_0 IGhea_0 IGinf_0 IP_0 J_0 KGedu_0 KGhea_0 KGinf_0 KP_0 LE_0 LR_0 M_0 PD_0 PM_0 POP_0 PX_0 PY_0 Qd_0 Q SP_0 T_0 TAX_0 X_0 Ydisp_0 Y_0 kappa_edu kappa_hea kappa_inf DB_0 FG_0 FP LAND n PMstar PXstar RD RGstar RPstar tm UTR$ Delta_NFA_0 ERROR_OMM_0 NGDP_0 KGedu_0 LE_P_0 LE_G_0 DITXR_0 INDTXR_0 YTOT_0 kappa_oth WG_0 IGoth_0 DITAX_0 INDTAX_0 PQT_0 Z_0 KGZ_0

• In order to run this aid shock, the following line in the simulation program

should be closed as indicated in the program: niger.append AID = AID_const * NGDP







summary table, “Niger-Output Table-SHOCK7.xls”, and the MDG table, “Niger-MDG-

table-withaidshock0.52.xls”. “Niger-output-Aid-Shock-MDG Table 3.xls” shows the

deviation from the summary baseline table.

• “Niger-MDG-table-withaidshock-deviation-aidshock=0.052.xls” presents the

deviation from the baseline MDG table. This table is given in Appendix I.

Sensitivity Analysis (Partial Efficiency) – Baseline MDG Table

• In this case, we assume that public investment is partially efficient such that αh

is equal to 0.5 in equation (A33) in Appendix E. In order to obtain the simulated data

52

series that will be used to construct the baseline MDG table, open the following lines

in the program:

'line1baseline 2'scalar alpha_h = 0.5 'line2baseline2 - MDG tables'write(t=xls,b4,t) C:\Niger\OUTPUT-NIGER8.xls AID_0 CG_0 CP_0 PQ_0 DdebtG_0 DOM_0 ER FdebtG_0 FdebtP_0 FdebtTot_0 GBAL_0 GTOT_0 IG_0 IGedu_0 IGhea_0 IGinf_0 IP_0 J_0 KGedu_0 KGhea_0 KGinf_0 KP_0 LE_0 LR_0 M_0 PD_0 PM_0 POP_0 PX_0 PY_0 Qd_0 Q SP_0 T_0 TAX_0 X_0 Ydisp_0 Y_0 kappa_edu kappa_hea kappa_inf DB_0 FG_0 FP LAND n PMstar PXstar RD RGstar RPstar tm UTR$ Delta_NFA_0 ERROR_OMM_0 NGDP_0 KGedu_0 LE_P_0 LE_G_0 DITXR_0 INDTXR_0 YTOT_0 kappa_oth WG_0 IGoth_0 DITAX_0 INDTAX_0 PQT_0 Z_0 KGZ_0

• The

Operational Manual for “A Macroeconomic Framework for ......Operational Manual for “A Macroeconomic Framework for Quantifying Growth and Poverty Reduction Strategies in Niger”

Documents