A Small Macroeconometric Model of the Bangladesh Economy By Mohammad Mahbubur Rahman, and Rabeya Khatoon * October, 2011 Abstract This paper describes a macroeconometric model of the Bangladesh econ- omy using annual time series data from FY-1980 to FY-2006. The model is constructed with seven macroeconomic blocks, consumption, investment, pro- duction, government, trade, money, and price, capturing transmission among blocks. Structural equations under each block are estimated using short-run error correction model, where long-run equations into error correction terms represent economic theory. Hendry’s general to a specific procedure is fol- lowed to get final short-run error correction equations. Validity of the model is checked both within the sample and out of sample cases. Results from validity study mark that the model is reasonably useful for forecasting and policy analysis. JEL Classification: C51, E17 Key words: Macroeconometric model; Forecasts; Simulations 1 Introduction Over the last two decades, the Bangladesh economy that is growing moderately, has drawn a lot of attention to researchers for macroeconomic policy analyses. There- fore, a tradition of macromodeling has been developed where most of them are Computable General Equilibrium (CGE) models (see e.g., Mujeri and Khondker * The University of Manchester, UK (email: [email protected], Rabeya. [email protected]). The original version of this paper titled “Monetary Policy Impacts in Bangladesh: A Macroeconometric Model Approach” was presented in ECOMOD conference 2009 at Ottawa, Canada. We are grateful to Bureau of Economic Research, Department of Economics, University of Dhaka and Planning Commission, Bangladesh for their research grants.
28
Embed
A Small Macroeconometric Model of the Bangladesh … Small Macroeconometric Model of the Bangladesh Economy By Mohammad Mahbubur Rahman, and Rabeya Khatoon October, 2011 Abstract This
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
A Small Macroeconometric Modelof the Bangladesh Economy
By Mohammad Mahbubur Rahman, and Rabeya Khatoon∗
October, 2011
Abstract
This paper describes a macroeconometric model of the Bangladesh econ-omy using annual time series data from FY-1980 to FY-2006. The model isconstructed with seven macroeconomic blocks, consumption, investment, pro-duction, government, trade, money, and price, capturing transmission amongblocks. Structural equations under each block are estimated using short-runerror correction model, where long-run equations into error correction termsrepresent economic theory. Hendry’s general to a specific procedure is fol-lowed to get final short-run error correction equations. Validity of the modelis checked both within the sample and out of sample cases. Results fromvalidity study mark that the model is reasonably useful for forecasting andpolicy analysis.
Over the last two decades, the Bangladesh economy that is growing moderately, has
drawn a lot of attention to researchers for macroeconomic policy analyses. There-
fore, a tradition of macromodeling has been developed where most of them are
Computable General Equilibrium (CGE) models (see e.g., Mujeri and Khondker
∗The University of Manchester, UK (email: [email protected], [email protected]). The original version of this paper titled “Monetary Policy Impacts inBangladesh: A Macroeconometric Model Approach” was presented in ECOMOD conference 2009at Ottawa, Canada. We are grateful to Bureau of Economic Research, Department of Economics,University of Dhaka and Planning Commission, Bangladesh for their research grants.
(1998), Rahman (2001), Raihan (2010)). Though CGE models have their own ap-
peal for analyzing distributional aspects, they have their limitations in structural
modeling and thereby macroeconomic policy analyses (see e.g. Hall (1995)). The
early static type of CGE models can provide only a snapshot of the macro economy.
Even recent development of the dynamic CGE models cannot capture the dynamics
of the adjustment process which is the core of formal macroeconometric modeling.
For the Bangladesh economy, a number of macroeconometric models listing
Rashid (1981), Hossain (1995), Rahman and Shilpi (1996), Hossain and Razzaque
(2003), Quin, Razzaque and Rahman (2006), Haque (2007) was also constructed.
However, most of these models can be criticized in the ground of both the application
of time series econometrics in estimating behavioural equations and the construction
of such equations in the models. For example, the early macroeconometric models
Behavioural equations are run applying Error Correction Model (ECM) following
Engle-granger (1987). Here, we assume that all series are non-stationary, and there-
fore no unit root test is done. Long-run static equations have been run based on
long-run economic theories. If economic theories did not hold in some cases, then
theoretically plausible parameters have been imposed into ECM. Hendry’s ‘gen-
eral to a specific’ procedure (Hendry, (1995)) is applied to reach final short-run
equations. The use of time dummies is extremely low, as these reduce stability of
parameter. In addition, parameter stability is tested jointly as well as individually
for each parameter in the final short-run equations. Autocorrelation test, specifica-
tion test, heteroscedasticity test are also done to select the best specific model. All
behavioural equations are run in PcGive and PcGets.
6 Validity of the model
Validity of the model is checked through both within-sample and out-of-sample
forecasts. Within-sample validation is done via the mean percentage errors (MPE)
and the root mean square percentage errors (RMSPE). Out-of-sample forecast is
validated by stochastic simulations. For validity checks, all required simulations
(static and dynamic simulations within-sample, and stochastic simulations out-of-
sample) are run in Winsolve (see Pierse (2001)).
15
6.1 Within-sample validation
Time series data of 27 years from FY-1980 to FY-2006 is used to generate both
static and dynamic solutions. The results are then compared to the actual data
to calculate MPE and RMSPE. Table 1 reports the calculations of them for some
selected variables. As is shown in the table, the errors are considerably small and
the model predicts historical data reasonably well. The actual values are plotted
against the static and dynamic simulation results for 6 key variables in Figure 1.
The figure also shows that predicted series (both static and dynamic) are very close
to actual series which again indicate good forecasting power of the model. Previous
models did not have such power, rather there were huge gaps between actual and
predicted series.
6.2 Out-of-sample validation
The model is checked for out of sample validation using stochastic simulation exer-
cises. Stochastic simulation, applying the bootstrap method, exerts random shocks
from individual equation residuals into each estimated equation for a specified pe-
riod, and thereby introduces uncertainty into model forecasts. The magnitude of
uncertainty is shown using quantiles. Figure 2 represents the 96 percent confidence
band for uncertainty in out-of-sample simulations for 6 selected variables with the
2 percent and 98 percent quantiles generated from 100 simulations. In addition to
16
Table 1
Statistics for Validity Check of the Model: FY-1986 to FY-2006a
Variable
Static Dynamic
MPE RMSPE MPE RMSPE
X1c -0.0140 0.0203 -0.0272 0.0429
X2c -0.0010 0.0063 0.1331 0.1593
X3c -0.0133 0.0141 -0.1064 0.1200
Cc 0.0035 0.0155 0.0217 0.0510
Ic 0.0005 0.0310 0.0277 0.0786
GC -0.0220 0.0431 -0.0950 0.1154
E -0.0312 0.0685 -0.0802 0.1343
M -0.0222 0.0641 -0.0842 0.1137
R -0.0159 0.0358 -0.1247 0.1506
Ddb -0.0181 0.0531 0.0041 0.0863
M1 -0.0159 0.0382 -0.0073 0.0796
M2 -0.0099 0.0475 0.1057 0.1129
N bf 0.3837 1.5590 1.6168 3.1421
Py -0.0081 0.0130 -0.0538 0.0664
Pcd -0.0166 0.0250 -0.0674 0.0743
Yscs -0.0453 0.0463 -0.0616 0.0658
Dd 0.0000 0.0116 0.1886 0.2413
Df 0.0109 0.0727 0.2153 0.2347
rl -0.0013 0.0228 -0.0087 0.0281
Dbb 0.0372 0.1569 0.5089 0.6935
aThe MPE and RMSPE are computed as follow :
MPE = 1T
∑Tt=1
(Y st −Y a
tY at
), RMSPE =
√1T
∑Tt=1
(Y st −Y a
tY at
)2where Y s
t and Y at are the simulated and actual values of an endogenous variable respectively in period t
and T is the number of simulated periods.
bAs there are negative values in actual data, the calculated MPE and RMSPE come out to be high in
magnitude.
this, the 50 percent quantile is shown to represent the mean simulated value. As
shown in figure 2, out-of-sample performance of the model appears to be reasonably
good.
1000
1500
2000
2500
3000
3500
Bill
ion
Taka
GDP in constant prices
1000
1500
2000
2500
3000
3500
Bill
ion
Taka
Aggregate Demand in constant prices
0
100
200
300
400
500
600
Bill
ion
Taka
Narrow Money (M1)
050
100150200250300350400450
Bill
ion
Taka
Net Foreign Asset
10
11
12
13
14
15
16 Lending Interest Rate
0.5
0.7
0.9
1.1
1.3
1.5
1.7
1.9 Consumer Price Index
Actual Static Dynamic
Figure 1: Static and Dynamic forecasting: a few key variables
0100020003000400050006000700080009000
Bill
ion
Taka GDP in constant prices
0
1000
2000
3000
4000
5000
Bill
ion
Taka Investment in constant prices
10
11
12
13
14
15
16 Lending Interest Rate
0
1000
2000
3000
4000
5000B
illio
n Ta
kaNarrow Money (M1)
-10000
0
10000
20000
30000
40000
50000
60000
70000
Bill
ion
Taka Net Foreign Asset
02000400060008000
1000012000140001600018000
Bill
ion
Taka
Domestic credit of Deposit Money Bank
2% Quantile 50% Quantile 98% Quantile
Figure 2: Stochastic forecasting: a few key variables
7 Conclusion
In spite of small sample period, the model presented here is a robust macroecono-
metric model of the Bangladesh economy. Compared to the previous models, the
use of time series econometrics in the model is in advanced level. In addition, inten-
sive attention is given to hold economic theory while running long-run equations.
The use of time dummies is kept minimal to make the model stable. Thus, both
within-sample and out-of-sample simulation show that the model has high ability of
forecasting and policy simulations, indicating the usefulness of the model.
References
[1] Bhuyan, A. R., Rashid, M.A., (1993). Trade Regimes and Industrial Growth:A Case Study of Bangladesh. Bureau of Economic Research, Dhaka Universityand International Centre for Economic Growth, San Francisco, USA.
[2] Chowdhury, O. H., (1995). Fiscal Performance of Bangladesh. Chapter 4 inRehman Sobhan (ed), Experiences with Economic Reforms: A Review ofBangladeshs Development 1995, Centre for Policy Dialnue and the UniversityPress Limited, Dhaka.
[3] Doornik, J. A., Hendry, D. F., (2001). Empirical Econometric Modelling UsingPcGive. Timberlake Consultants Ltd., London.
[4] Ducanes, G., Cagas, M. A., Qin, D., Quising, P., Magtibay-Ramos, N., (2005).A Small Macroeconometric Model of the Philippine Economy. ERD WorkingPaper No.62, Asian Development Bank, Manila, the Philippines.
[5] Fischer, S., Easterly, W., (1990). The Economics of the Government BudgetConstraint. World Bank Research Observer, vol. 5, pp.127-142.
[6] Enamul Haque, A. K., (2007). Impact of World Oil Price Rise on the Economyof Bangladesh. BANSIM II, REP, UNDP, Bangkok and Economic ResearchGroup, Dhaka.
[7] Engle, R. F., Granger, C. W. J., (1987). Co-integration and error correction:Representation, estimation and testing. Econometrica, 55(2), 251-276.
[8] Hall, S., (1995). Macroeconomics and a bit more reality. The Economic Journal,105(431), pp. 974988.
[9] Hendry, D. F., (1995). Dynamic Economics. Oxford: Oxford University Press.
[10] Hossain, M. A., (1995). Inflation, Economic Growth and the Balance of Pay-ments in Bangladesh: A Macroeconometric Study. Oxford University Press,Delhi.
[11] Hossain, A., (2002). Exchange Rate Responses to Inflation in Bangladesh. IMFworking paper WP/02/166.
[12] Hossain, M. I., Razzaque, A., (2003). A Macroeconometric Model ofBangladesh: Specification and Estimation. Technical Paper No. 06, MIMAP-Bangladesh Micro Impacts of Macroeconomic and Adjustment Policies inBangladesh.
[13] Islam, N., (1965). A Short-Term Model for Pakistan Economy: An EconometricAnalysis. Oxford University Press, Dhaka.
[14] Klein, L. R., Welfe, A., Welfe, W., (1999). Principles of MacroeoconometricModeling. Advanced Textbooks in Economics 36, North-Holland.
[15] Mahmud, W., (1995). Recent Macroeconomic Development. Chapter 2 inRehman Sobhan (ed), Experiences with Economic Reforms: A Review ofBangladeshs Development 1995, Centre for Policy Dialnue and the UniversityPress Limited, Dhaka.
[16] Mallick, S. K., (1999). Modelling Macroeconomic Adjustment With Growth inDeveloping Economies: The Case of India. Ashgate.
[17] Mujeri, M.K., Khondker, B.H., (1998). Impact of Macroeconomic Policy Re-forms in Bangladesh A General Equilibrium Framework for Analysis. PaperPresented at the Micro Impacts of Macroeconomic and Adjustment Policies(MIMAP), Third Annual Meeting, November 2-6, 1998, Kathmandu, Nepal.
[18] Musila, J. W., (2002). An Econometric Model of the Malawian Economy. Eco-nomic Modelling, vol.19, pp.295-330.
[19] Pierse, R., (2001). Winsolve Manual. Department of Economics, University ofSurrey, United Kingdom.
[20] Quin, D., Razzaque, A., Rahman, M., (2006). A Small Quarterly Macroecono-metric Model of Bangladesh. Asian Development Bank (ADB), Manila.
[21] Rahman, M., (2002). Bangladeshs External Sector in FY2001: Review of Per-formance and Emerging Concerns, in Sobhan, R. (ed) Bangladesh Facing theChallenges of Globalisation: A Review of Bangladeshs Development 2001, Cen-tre for Policy Dialnue, Dhaka.
[22] Rahman, S.M., (2001). A Computable General Equilibrium Analysis of Alter-native Economic Policy Strategies for Agriculture in Bangladesh. UnpublishedPhD dissertation, University of Manchester.
[23] Rahman, S. H., (1994). Trade and Industrialisation in Bangladesh: An Assess-ment in G.K. Helliner (ed.) Trade and Industrialisation in Turbulent Times,Chapter 8, pp.259-291, Routledge.
[24] Rahman, S. H., Shilpi, F. J., (1996). A Macro Econometric Model of theBangladesh Economy: Model, Estimation, Validation and Policy Simulation.Research Monograph No. 17, Bangladesh Institute of Development Studies,Dhaka.
[25] Raihan, S., (2010). Welfare and Poverty Impacts of Trade Liberalization: ADynamic CGE Microsimulation Analysis. International Journal of Microsimu-lation, 3(1), pp. 123-126.
[26] Raihan, S., Fatema, K., (2007). A Review of the Current Hypotheses onInflation in Bangladesh. Working paper # 02/2007, Shamunnay, Dhaka,Bangladesh.
[27] Rashid, M. A., (1981). A Macro Econometric Model of Bangladesh. BangladeshDevelopment Studies, Vol. IX, No. 3.
[28] Valadkhani, A., (2004). History of macroeconometric modelling: lessons frompast experience. Journal of Policy Modeling, 26, pp. 265281.
[29] World Bank, (2003). Bangladesh: Achievements and Challenges. DevelopmentPolicy Review, World Bank, Washington, D.C.
Appendix A : Variables and Sources of Data
Endogenous Variables
Variable
Name
Definition Source Units
B Current account balance Economic Trends,Bangladesh Bank
Million Taka
Cc Private consumption inconstant prices
Bangladesh Bureau ofStatistics
Million 1995Taka
C Private consumption incurrent prices
Bangladesh Bureau ofStatistics
Million Taka
Cp Credit to private sector Bangladesh Bureau ofStatistics
Million Taka
Db Domestic borrowing ofgovernment
Bangladesh Bureau ofStatistics
Million Taka
Dbb Domestic credit of centralbank
Economic Trends,Bangladesh Bank
Million Taka
Ddb Domestic credit of depositmoney bank
Economic Trends,Bangladesh Bank
Million Taka
Dd Domestic Debt Ministry of Finance,Government ofBangladesh
Million Taka
Df Foreign Debt Ministry of Finance,Government ofBangladesh
Million Taka
D Total Debt Ministry of Finance,Government ofBangladesh
Million Taka
Dg Government budget deficit Bangladesh Bureau ofStatistics
Million Taka
Ec Exports in constant prices Bangladesh Bureau ofStatistics
Million 1995Taka
E Exports in current prices Bangladesh Bureau ofStatistics
Million Taka
Gc Government consumptionin constant prices
Bangladesh Bureau ofStatistics
Million 1995Taka
Variable
Name
Definition Source Units
G Government consumptionin current prices
Bangladesh Bureauof Statistics
Million Taka
Ge Government expenditure incurrent prices
Bangladesh Bureauof Statistics
Million Taka
Ig Government investment incurrent prices
Bangladesh Bureauof Statistics
Million Taka
Ipc Private investment inconstant prices
Bangladesh Bureauof Statistics
Million Taka
Ip Private investment incurrent prices
Bangladesh Bureauof Statistics
Million Taka
Ic Total investment inconstant prices
Bangladesh Bureauof Statistics
Million 1995Taka
I Total investment in currentprices
Bangladesh Bureauof Statistics
Million Taka
id Deposit interest rate Economic Trends,Bangladesh Bank
il Landing interest rate Economic Trends,Bangladesh Bank
Kc Capital in constant prices MIMAP Million 1995Taka
K Capital in current prices MIMAP Million Taka
Mc Imports in constant prices Bangladesh Bureauof Statistics
Million 1995Taka
M Imports in current prices Bangladesh Bureauof Statistics
Million Taka
M0 Money circulation Economic Trends,Bangladesh Bank
Million Taka
M1 Narrow money Economic Trends,Bangladesh Bank
Million Taka
M2 Broad money Economic Trends,Bangladesh Bank
Million Taka
Nf Net foreign asset Economic Trends,Bangladesh Bank
Million Taka
Nfac Net factor income fromabroad in constant prices
Bangladesh Bureauof Statistics
Million 1995Taka
Variable
Name
Definition Source Units
Nfa Net factor income fromabroad in current prices
Bangladesh Bureauof Statistics
Million Taka
Pcd Consumer price index Economic Trends,Bangladesh Bank
1995=1
Pcw World consumer priceindex (trade-weighted CPIfor 20 major tradingpartners of Bangladesh)
InternationalFinancial Statistics,Directorate of TradeStatistics
1995=1
Py GDP deflator Bangladesh Bureauof Statistics
1995=1
Pi Investment deflator Bangladesh Bureauof Statistics
1995=1
Pm Import deflator Bangladesh Bureauof Statistics
1995=1
Pob Oil price per barrel Taka
Pol Oil price per litre Taka
Pp Producer price index Economic Trends,Bangladesh Bank
1995=1
Pe Export deflator Bangladesh Bureauof Statistics
1995=1
Pa Administered price of perlitre oil
Petro Bangla Taka
R Government total revenue Bangladesh Bureauof Statistics
Million Taka
T Tax revenue Economic Trends,Bangladesh Bank
Million Taka
X1c Value added in primarysector in constant prices
Bangladesh Bureauof Statistics
Million 1995Taka
X1 Value added in primarysector in current prices
Bangladesh Bureauof Statistics
Million Taka
X2c Value added in secondarysector in constant prices
Bangladesh Bureauof Statistics
Million 1995Taka
X2 Value added in secondarysector in current prices
Bangladesh Bureauof Statistics
Million Taka
Variable
Name
Definition Source Units
X3c Value added in tertiarysector in constant prices
Bangladesh Bureauof Statistics
Million 1995Taka
X3 Value added in tertiarysector in current prices
Bangladesh Bureauof Statistics
Million Taka
Ysc GDP in constant prices Bangladesh Bureauof Statistics
Million 1995Taka
Ys GDP in current prices Bangladesh Bureauof Statistics
Million Taka
Ync GNP in constant prices Bangladesh Bureauof Statistics
Million 1995Taka
Yn GNP in current prices Bangladesh Bureauof Statistics