Quantitative Modelling-part-1 By Tarun Das

Quantitative Modeling Presented by

Dr Tarun Das

1.1 Purpose of Quantitative Modeling Policy planning versus physical planning Indicative planning versus target setting Optimizing versus consistency approach Partial/sectoral versus economy wide model Selective versus all purpose model Learning versus blue print approach Consultancy versus dirigist approach Monitoring trends versus forecasting future Sound public investment versus overall investment planning 2

1.2 Dimensions of Quantitative Models Macro models (National level) Meso (Middle level) Spatial (over space) Regional (over regions) Sectoral (over industries/ sectors) Micro (at unit levels) Inter temporal (over time) Intergenerational (over generations)

3

1.3 Types of Models • Static or dynamic models • Short, medium, long term model • Consistency or Behaviouristic or Optimizing or Econometric models • Closed or open economy model • Economy wide or sectoral model • Selective or All purpose model • Comprehensive or partial model • Deterministic or indeterminate 4

1.4 Desirable characteristics of an Ideal Model Internal consistency Comprehensiveness with two-way feed backs Dynamic to take care of change over time Unique and stable solutions Easy testing and calibration of the model with the help of available data, computer capacities and algorithms, statistical techniques • Model should be continually tested, reviewed, Calibrated, monitored, updated, simulated and improved to make it more realistic • • • • •

5

1.5 Advantage of Econometric Modeling An Econometric Model helps:  to formalize the system,  to establish inter-relations among major policy variables that can be specified, calibrated, tested, monitored, updated, simulated and predicted with certain degree of confidence, and  to identify trade-off among alternative policy options. 6

2.1 Steps in Econometric Modeling  Specify objectives and purpose  Have a sound theoretical basis  Identification of potential variables  Specification of equations  Identification of equations  Calibration of parameters  Testing- Various goodness of fit statistics- R-sq,  -sq, Theil index etc.  Simulation and policy planning Monitoring and updating Prediction and Projection. 7

2.2 Econometric Modeling Types of Data • Time series • Cross section • Panel - Combination of time series and cross section data • Pooled- Combination of sectors/ regions

8

2.3 Types of variables •

Endogenous variables (determined within the model) Ct, Yt • Exogenous/ predetermined variable Wt • Parameters  , , , ,  • Lagged variable Ct-1 • Instrumental variable Wt • Dummy (binary, categorical, indicator, qualitative, dichotomous) variable Dt • Omitted variables • Catch all variable (could be time trend) Ct =  +  Yt +  Wt +  Ct-1 +  Dt Dt = 1 if t is 1991 to 2003, 0 otherwise 9

2.4 Types of Equations • • • • •

Technical such as production function Behavioral such as consumption function Definitional such as capital/output ratio Identities such as Y = C + I + G Structural form Ct =  1 +  1 Yt + Ut It = 2 + 2 Yt + 3 Yt-1 + Vt Yt = Ct + It + Gt

• Reduced form Yt =  3 +  4 Yt-1 + 5 Gt + Wt 3 = (1 + 2 )/ 6, 4 = 3 / 6, 5 =1/ 6, 6 =1 1 - 2

10

3.1 General Disagreements by Modelers • • • • • • • • •

“Let all the flowers flourish” General purpose versus single purpose model Normative versus descriptive model Policy oriented versus general understanding Short run versus long run model Large versus small model Top down versus bottom up model To guess or leave out crucial missing data How to deal with future economic agents, technology change, prices and population? Whom should we address? 11

3.2 General Agreements by Modelers -1 • State your biases and intuitive arguments • Computer models of social systems should not be expected to produce precise results. • Qualitative judgements from econometric models are useful for policy formulation. • Models should be selected to fit problems. • Social environment and political economy may be treated as given in the model. • Models should be tested rigorously for the real world with full range of policies. • Substantial portion of resources should be used for full documentation of the model.

12

3.3 General Agreements by Modelers -2 • Part of the model documentation should be technically complete so that any other group can test, calibrate and run the model. • Part of the documentation should be clear and free from jargons for general understanding by the nontechnical audience. • Modelers should specify data sources and share their data. • Users and policy makers should be involved in the modeling process. • It is necessary to continually review, monitor, update, upgrade, and simulate the model.

13

Thank you Have a Good Day

14