Regional Capacity Building Workshop Formulating National Policies and Strategies in Preparation for Graduation from the LDC Category: Macroeconomic Modelling for SDGs in Asia and the Pacific Santi Chaisrisawatsuk 16 November 2017 Thimpu, Bhutan
Outline Introduction Fair s Model Key modifications Macroeconomic model Preliminary results for discussions SDG variables integration
Introduction Changing economic dynamic and environments Growing important of service sector Different mechanism fostering economic growth Economic development process Emphasis place on growth versus other economic objectives to improve living standard such as inequality and environmental concerns The role of international economics An increasing role of international economic activities and economic integration Possible linkages of SDG variables in macroeconomic modeling
Fair s Model (Mini version) Traditional macroeconomic model consists of a large number of structural equations (behavior + identities equations) Completed Fair s model (30 behavior equations and 158 identities) Mini Fair s model covers fundamental economic variables in aggregate without going too much into more sectoral details Easier to integrate SDG variables into the model Fair s mini version model (10 behavior equations and 19 identities)
Key Modifications Total wealth variable is approximated by Gross Domestic Saving data on total wealth is not available and wealth is mostly held in the form of saving in most developing economy (data is available in World development Indicators by World Bank) Data on financial wealth and housing wealth are not available so they are not included in the model Integration of durable and nondurable consumption to use total final consumption expenditure (data is available in World development Indicators by World Bank) Consumption in service still remains in the model (data is available in World development Indicators by World Bank)
Key Modifications Population structure is adjust according to the World Bank base (Fair s 26-55, 56-65, 66+) (WB s 20-49, 50-59, 60+) (Data is available in World development Indicators by World Bank) Depreciation rate is exogenous variable and assumed to be fixed at 6% Capital gain is proxied by total tax on income capital gains and profits tax on goods and services Physical depreciation rate of capital stock assumed to be fixed at 10% (Data available at NESDB Thailand) Export and Government Expenditure are exogenous variable (Data is available in World development Indicators by World Bank)
Key Modifications Residential investment data is not available (not include in the model) Import equation remains as in Fair s model (Data is available in World development Indicators by World Bank) Housing stock and durable good stock are not available (not include in the model) Inventory investment uses change in inventory variable from the World Bank database Capital stock variable use data from the World Bank database GDP deflator in the model use data from the World Bank database
Key Modifications Mortgage rate is replaced by lending rate 3-month T-bill rate is replaced by saving rate Financial saving of household sector replaced by Gross Domestic Saving Tax and Tax ratio is available in the World Bank database Stock of inventory is proxy by change in inventory variable Calculation of Minimum Capital required based on change in economic structure instead of peak to peak approach Calculated Potential Output base on time series method
Macroeconomic Model 7 behavior equations and 10 identities The Behavior Equation: Final Goods Consumption, Service Consumption, Change in capital stocks, GDP potential, Import, Price, and Saving Rate The Identities Equation : Nominal GDP, Nominal disposable Income, Growth of disposable Income, Nominal Depreciation, Nominal Tax, Nominal Saving, Real Disposable Income, Minimum Capital, Output Gap, Inflation, and GDP Growth
A Case of Thailand: A Work in Progress The initial model is tested by using Thailand data set for any possible problems Annual data are used from 1980-2015 according to the availability of the data set for Thailand (and later for CLM) in the World Bank database All 7 behavior equations are estimated with some reliability of the results Later, some of the SDG Variables will be introduced to the model
Preliminary results for discussions: Final Goods Consumption equation Dependent Variable: LNCFPOP Included observations: 37 after adjustments Variable Coefficient Std. Error t-statistic Prob. C 1.193252 0.862716 1.383135 0.1772 AG1 8.96E-11 3.26E-09 0.027465 0.9783 AG2-2.58E-08 1.31E-08-1.975743 0.0578 AG3 6.29E-08 2.93E-08 2.148333 0.0402 LNCFPOP(-1) 0.846774 0.128953 6.566507 0.0000 D(LOG(YD/POP)) 0.453660 0.162106 2.798536 0.0090 RS -0.005775 0.002306-2.504484 0.0181 D(LNAAPOP(-1)) 0.093114 0.040756 2.284659 0.0298 R-squared 0.994292 Mean dependent var 7.959909 Adjusted R-squared 0.992915 S.D. dependent var 0.312644 S.E. of regression 0.026317 Akaike info criterion -4.248424 Sum squared resid 0.020084 Schwarz criterion -3.900118 Log likelihood 86.59585 Hannan-Quinn criter. -4.125630 F-statistic 721.7068 Durbin-Watson stat 1.698701 Prob(F-statistic) 0.000000
Preliminary results for discussions: Service Consumption equation Dependent Variable: DLNCSPOP Included observations: 37 after adjustments Variable Coefficient Std. Error t-statistic Prob. C 0.086288 0.034128 2.528362 0.0172 AG1-2.56E-09 1.15E-09-2.220857 0.0343 AG2-3.05E-10 7.05E-09-0.043279 0.9658 AG3 7.42E-09 1.49E-08 0.499775 0.6210 DLNCSPOP(-1) 0.229247 0.079341 2.889376 0.0072 D(LOG(YD/POP)) 0.781907 0.119986 6.516673 0.0000 RS -0.002758 0.001168-2.361530 0.0251 DLNAAPOP 0.014676 0.034480 0.425649 0.6735 R-squared 0.862765 Mean dependent var 0.028327 Adjusted R-squared 0.829640 S.D. dependent var 0.035523 S.E. of regression 0.014662 Akaike info criterion -5.418316 Sum squared resid 0.006234 Schwarz criterion -5.070009 Log likelihood 108.2388 Hannan-Quinn criter. -5.295522 F-statistic 26.04524 Durbin-Watson stat 2.556321 Prob(F-statistic) 0.000000
Preliminary results for discussions: Import equation Dependent Variable: LNIMPOP Included observations: 38 after adjustments Variable Coefficient Std. Error t-statistic Prob. C -7.368308 1.794379-4.106326 0.0002 LNIMPOP(-1) 0.309045 0.138081 2.238138 0.0319 LOG((CS+CF)/POP) 1.470921 0.322554 4.560234 0.0001 LNPPIM(-1) 0.072843 0.132743 0.548751 0.5868 R-squared 0.979132 Mean dependent var 7.458130 Adjusted R-squared 0.977291 S.D. dependent var 0.694322 S.E. of regression 0.104632 Akaike info criterion -1.577435 Sum squared resid 0.372227 Schwarz criterion -1.405057 Log likelihood 33.97126 Hannan-Quinn criter. -1.516104 F-statistic 531.7585 Durbin-Watson stat 1.060337 Prob(F-statistic) 0.000000
Preliminary results for discussions: Change in Capital Stock equation Dependent Variable: D(LNKK) Included observations: 37 after adjustments Variable Coefficient Std. Error t-statistic Prob. C 0.100636 0.045852 2.194780 0.0353 LNKKKKMIN(-1) -0.008441 0.004888-1.726974 0.0935 D(LNKK(-1)) 0.260735 0.142678 1.827438 0.0767 D(LOG(Y(-1))) 0.516941 0.182882 2.826632 0.0079 R-squared 0.511201 Mean dependent var 0.096674 Adjusted R-squared 0.466765 S.D. dependent var 0.050721 S.E. of regression 0.037038 Akaike info criterion -3.651924 Sum squared resid 0.045271 Schwarz criterion -3.477770 Log likelihood 71.56059 Hannan-Quinn criter. -3.590526 F-statistic 11.50414 Durbin-Watson stat 2.291668 Prob(F-statistic) 0.000026
Preliminary results for discussions: Price equation Dependent Variable: LOG(P) Included observations: 38 after adjustments Variable Coefficient Std. Error t-statistic Prob. C 0.250697 0.212861 1.177752 0.2473 T 0.001042 0.002340 0.445403 0.6589 LOG(P(-1)) 0.924606 0.063271 14.61335 0.0000 LOG(PIM) 0.022740 0.038470 0.591095 0.5585 GAP -5.49E-13 3.59E-13-1.527019 0.1363 R-squared 0.997848 Mean dependent var 4.386535 Adjusted R-squared 0.997588 S.D. dependent var 0.428096 S.E. of regression 0.021026 Akaike info criterion -4.764001 Sum squared resid 0.014590 Schwarz criterion -4.548529 Log likelihood 95.51601 Hannan-Quinn criter. -4.687337 F-statistic 3826.146 Durbin-Watson stat 1.680867 Prob(F-statistic) 0.000000
Preliminary results for discussions: Saving rate equation Dependent Variable: RS Included observations: 38 after adjustments Variable Coefficient Std. Error t-statistic Prob. C -0.980484 0.535458-1.831113 0.0761 RS(-1) 0.876345 0.058486 14.98382 0.0000 PCP 0.406432 0.091886 4.423246 0.0001 GAP -6.35E-12 1.92E-11-0.330441 0.7432 D(GAP) -8.47E-12 3.31E-11-0.255996 0.7995 R-squared 0.911814 Mean dependent var 6.998092 Adjusted R-squared 0.901124 S.D. dependent var 4.512122 S.E. of regression 1.418814 Akaike info criterion 3.659600 Sum squared resid 66.43013 Schwarz criterion 3.875071 Log likelihood -64.53239 Hannan-Quinn criter. 3.736263 F-statistic 85.30183 Durbin-Watson stat 1.745865 Prob(F-statistic) 0.000000
Preliminary results for discussions: Potential GDP equation Dependent Variable: LOG(Y) Included observations: 38 after adjustments Variable Coefficient Std. Error t-statistic Prob. C 0.359343 0.206940 1.736459 0.0915 LOG(Y(-1)) 0.910496 0.015250 59.70377 0.0000 LOG(X) 0.077105 0.015461 4.987023 0.0000 V(-1)/1000000000-0.000861 0.001891-0.455173 0.6519 R-squared 0.998039 Mean dependent var 25.90115 Adjusted R-squared 0.997866 S.D. dependent var 0.594930 S.E. of regression 0.027481 Akaike info criterion -4.251371 Sum squared resid 0.025676 Schwarz criterion -4.078994 Log likelihood 84.77606 Hannan-Quinn criter. -4.190041 F-statistic 5769.088 Durbin-Watson stat 1.404658 Prob(F-statistic) 0.000000
Way Forward LDCs Data challenges (availability, reliability, accessibility) Simple macro model not to focus too much on the accuracy of prediction of economic variables but will have more emphasis on interactions of variable to explain the impacts Integration of SDGs into macro model (CLM): To be treated as exogenous VS Endogenous variables
SDG variables integration Consideration based on availability of the data and priority of national public policy objectives Poverty (Deprivation reduction) Health (Improving quality of health services) Education (Productivity improvement) Gender in equality (Wage Gap)? Access to electricity (Access to necessary infrastructure) Energy Intensity (Efficiency used of resources) Co 2 emission (Environmental concerns) Inequalities (Income inequality as measured by GINI coefficient)