Economic Projection at NESDB International Seminar on Timeliness
Economic Projection at NESDB International Seminar on Timeliness Methodology and Comparability of Rapid Estimates of Economic Trends Ottawa, Canada 27 -29 May 2009 Wichayayuth Boonchit Economic Modeling & Projection Division Macroeconomic Office of the National Economic & Social Development Board Thailand wichayayuth@nesdb. go. th 1
NESDB’s Functions and Responsibilities Economic projections 1. To study and analyze the national economic and social condition for development planning, and recommend related policy issues to the government 2. To follow-up the performance of the National Plan and monitor and evaluate some major development programs and projects 3. To coordinate with all agencies concerned in implementation of the development plan 4. To appraise and evaluate development programs/projects of public agencies 5. To undertake any assignments by the government 2 Impact analysis
NESDB and Economic Projections Long-term projection Medium-term projection (guideline formulating development plan & development strategy) Potential GDP (subjected to strong assumptions) Medium Term Macroeconomic Frameworks (Trends and targets) Medium Term Expenditure Framework: MTEF 3
The Quarterly Economic Report Contents • Current economic conditions • Outlook for remaining of the year • Yearly GDP projection (Quarterly GDP forecast is part of yearly GDP projection) • Policy guidelines Release to the public (release on the same date with actual QGDP) Submit to cabinet for consideration 4
NESDB and Economic Projections o Forecast economic growth (focuses mainly on expenditure side) o Current account balance o Inflation Annual GDP forecast o Update on quarterly basis (Assumptions & databases, revise if necessary) o Requires quarterly forecast o Projection in the range of 1% in February and May o Projection range will be reduced to 0. 5% in August o Point estimate in November (the projection for following year will be also released) 5
Tools for Quarterly and Yearly Economic Projection - Current Quarter Model: CQM An old style macro-econometric model in the tradition of Lawrence Klein that are heavily reliance on econometric estimation (Other models of this types are Project Link, DRI, Wharton model) - Quarterly Financial Model: QFM The newer style macro econometric models with greater reliance on economic theory. 6
Adjust Annual forecast (CGE + Fin. Programming) CQM’s forecast (2 quarters) comparison Current Quarter Model CQM QFM’s forecast (4 quarters) Quarterly Financial Model (QFM) Assumptions & exogenous variables (Quarterly) High frequency data (Monthly) Beginning Annual forecast is the mixtures of CQM and QFM forecast 7
Yearly GDP Projection Release date Actual CQM forecast QFM forecast 23 rd February Q 4 of previous year Q 1 Q 2 -Q 4 of previous year Q 1_Q 2 Q 3 -Q 4 Q 1 Q 2 -Q 3 Q 4 Q 2 Q 3 -Q 4 - Q 3 Q 4 – Q 1 (next year) - March-April 25 th May Jun-July August September-July November December-January 8
Current Quarter Model: CQM Tool for rapid estimation CQM is first developed at the University of Pennsylvania by Noble Laureate Lawrence R. Klein Concept • Utilize high but different frequencies information (indicator variables) to estimate immediate future values of GDP both on demand supply sides • The estimates are made on the basis of bridge equations that link high frequency data (indicator variables) to low frequency data (NIPA). • The procedure is first to predict the future value of high frequency data (indicator variables) by using time-series analysis (ARMA process) and then estimate future values of NIPA by using bridge equations. • The estimate values of GDP will be updated on the rolling basis, when new piece of information or new figure for one of indicator variable become available (not late than 15 th of each month). 9
o Projected NIPA data (National estimate income and product account data) “bridge” equations o NIPA data (National income and product Account data) o Indicator variables ( Monthly) with projected period ARMA and ARIMA techniques o Indicator variables (Daily, Weekly , and Monthly) Econometric techniques 10
Jan. Feb. Mar. Apr. May. Jun. Jul. Aug. Sep. Q 4 of previous year Q 1 Q 2 Oct. Nov. Q 3 Dec. Jan. Feb. Mar. Q 4 11
How to Construct CQM? • Search for suitable high frequencies information (indicator variables) (availability on timely basis, reliability or variability) • Plot time series that are selected to obtain broad ideas of trend, cyclical variability and seasonality of variable. • Test for stationarity by applying Augmented-Dickey Fuller Test and Philips Perron Test • Select appropriate ARMA Term (p, d, q) for each indicator variable • Estimate Bridge equations that link high frequencies information (indicator variables) to low frequencies variables (NIPA) 12
Example: relationship between indicator variables and NIPA variables PCE at current prices and Value added tax PCE Prices deflator and CPI 13
Example: relationship between indicator variables and NIPA variables Manufacturing at 1988 prices and MPI Deflator of Manufacturing and PPI Manufac. 14
Example: relationship between indicator variables and NIPA variables Hotel and Restaurant at current market prices and VAT Deflator of Hotel and Restaurant and CPI Recreation 15
Monthly indicators (Indicator variables) Jan. Feb. A A On June 15: Monthly indicators as of April 31 become available Mar. Apr. May. Jun. Jul. Aug. Sep. A A P P P Forecast indicator variables by using ARMA equation 2. Transform from monthly indicators to quarterly indicators 3. Use bridge equations to estimate NIPA variables 1. Q 1 X A Q 2 X P Bridge Equation QNIPA = a +b. QX Q 3 X P NIPA A E E A= Actual value P= Projected value E=Estimated value 16
NIPA & Indicator Variables (A) Expenditure Side NIPA Monthly Indicators Nominal/Real Expenditures Price Deflator Private Consumption Expenditure Retail Sales Index Sales of Passenger Car CPI Government Expenditure Exogenous Gross Fixed Capital Formation Identity - Construction • Construction area permitted (lag terms) • Cement consumption Construction price index - Machinery and Equipment • Commercial car sales • Import volume index of capital goods PPI of capital equipment Exports of Goods Export of goods (BOP) Unit value of exports Exports of Services Receipts of services income and transfer (BOP) Weighted avg. of CPIs Imports of Goods Imports of goods (BOP) Unit value of imports Imports of Services Payments of services income and transfer (BOP) Nominal imports of goods 17
(B) Production Side NIPA Monthly Indicators Nominal/Real Expenditures Price Deflator Agriculture, Hunting and Forestry Crop production index Crop price index Fishing Export Volume of Fishery PPI Fishery product Mining & Quarrying Production index of natural gas PPI Mining Manufacturing production index PPI Manufacturing Electricity, Gas & Water Supply Electricity consumption CPI Electricity, fuel and water Construction • Cement consumption • Construction area permitted Construction price index Wholesale and Retail Trade VAT for this sector CPI Hotel and Restaurant VAT for this sector CPI Recreation Transportation and Communication VAT for this sector CPI Transportation Financial Interest rate spread CPI Real Estate & Business activity VAT for this sector CPI Shelter 18
Example: CQM Forecast for Q 1/2009 2008 2009 Q 1 Q 2 Q 3 Q 4 Q 1 F Q 2 F Private Consumption 2. 7 2. 5 2. 7 2. 2 -1. 3 -2. 1 -0. 5 -2. 6 Government Consumption -0. 4 -3. 7 -2. 9 10. 4 7. 5 2. 0 11. 2 9. 5 2. 8 Gross Fixed Capital Formation 5. 4 1. 9 0. 6 -3. 3 -7. 8 -4. 3 -13. 7 -9. 7 -15. 8 - Private 6. 5 4. 3 3. 5 -1. 3 -7. 6 -4. 2 -14. 0 -12. 0 -17. 7 - Public 1. 9 -5. 2 -5. 5 -10. 2 -8. 5 -4. 5 -12. 7 -2. 3 Export of Goods and Services 8. 9 11. 2 -8. 6 -16. 1 -18. 6 -17. 1 -21. 1 -16. 4 - Export of Goods 8. 3 13. 2 12. 6 -8. 9 -19. 7 -22. 4 -17. 8 -20. 9 -17. 9 11. 1 5. 6 4. 9 -7. 5 -2. 4 -0. 1 -14. 6 -21. 9 -11. 0 9. 3 6. 7 13. 1 1. 0 -19. 4 -22. 6 -23. 6 -28. 7 -31. 4 10. 0 5. 2 12. 5 0. 1 -23. 7 -27. 0 -29. 3 -34. 7 -36. 1 - Import of Services 6. 9 13. 7 16. 2 4. 5 -2. 0 -3. 1 -0. 5 -2. 6 -12. 3 Gross Domestic Expenditure 6. 2 5. 5 3. 8 -4. 3 -7. 2 -8. 2 -6. 5 -5. 8 -7. 1 Agricultural Sector 3. 1 8. 6 9. 6 1. 8 1. 6 2. 7 0. 8 1. 3 3. 5 Non-Agricultural Sector 6. 2 5. 0 3. 5 -5. 0 -8. 0 -6. 7 -7. 2 -6. 4 -8. 1 Manufacturing Sector 9. 5 7. 7 6. 1 -6. 8 -15. 9 -14. 9 -13. 6 -12. 7 -14. 9 Construction 1. 1 -3. 4 -4. 5 -12. 8 -10. 5 -5. 2 -9. 7 -12. 3 -7. 9 Wholesale and retail trade 4. 1 3. 4 3. 1 -3. 0 -1. 9 -2. 2 -0. 6 -1. 3 -4. 0 Gross Domestic Product 6. 0 5. 3 3. 9 -4. 3 -7. 2 -6. 0 -6. 5 -5. 8 - Export of Services Import of Goods and Services - Import of Goods * With 1 month actual indicator variables, ** With 3 months actual indicator variables Q 1 F** Q 2 F Actual -9. 1 -7. 1 19
Quarterly Financial Model: QFM • QFM can be classified as a newer style macro econometric models. • QFM comprise of 29 endogenous aggregate variables and 28 exogenous aggregate variables • QFM forecast is used to reconcile with CQM forecast and to form annual projection • QFM is also used for analyzing shocks in financial sector 20
List of Endogenous Variables • • • • Ctp = Private consumption expenditures CAD = Current account deficit CFt = Net capital inflows EX = Aggregate exports GRt = Total government revenue GSt = Government surplus Itp = Private investment IMi = Import classified by commodity groups, (where i=1, 2, 3, …, 10) IM = Aggregate imports Ms = Money Supply MB = Money base NFA = Net foreign asset NES = Net exports of services PXDi = Relative price of export to domestic price index (classified by commodity groups), where i=1, 2, 3, …, 10 • • • • Ptd = Consumer price index PIMDi = Relative price of import to domestic price index (classified by commodity groups), where i=1, 2, 3, …, 10 rd = Domestic interest rate (MLR) St = Saving TAXt = Government tax revenue Xi = Exports classified by commodity groups, where i=1, 2, 3, …, 10 VAT = Value added and business tax revenue Yt = Gross domestic product YA = GDP from agriculture sector YC = GDP from YE = GDP from electricity and water supply YM = GDP from industrial sector Yother = GDP from other sectors YD = Disposable income YD_er = Disposable income in USD 21
List of Exogenous Variables • • • • Ctg = Public consumption expenditures CONP = Claims on nonfinancial public enterprise (collected since January 1995) CREDIT = Export credit (USD) DISt = Statistical discrepancies et =Bilateral exchange rate (baht/USD) EOS = errors and omission portions in balance of payments Get = total government expenditure GRWTHUS = Growth of US GDP Itg = Public capital formation expenditures real NCOG = Net claims on Government by bank of Thailand Pid = Domestic price index classified by commodity groups, where i=1, 2, 3, …, 9 Pt. E = Price expectation in period t (adaptive) Ptf = Foreign price index (USCPI) Pt. IM = Import price index Pt. X = Export price index Pt. YA = Price index of agriculture sector • • • Pt. YC = Price index of construction Pt. YE = Price of electricity and water supply Pt. YA = Price index of industrial sector PIMi = Import price index classified by commodity groups, where i=1, 2, 3, …, 10 PXi = Export price index classified by commodity groups, where i=1, 2, 3, …, 10 ri = Interbank rate rf = Foreign interest rate (LIBOR) Qi = Time trend of export i classified by commodity groups, where i=1, 2, 3, …, 10 Qie = Expected export i (time trend of export i classified by commodity groups classified), where i=1, 2, 3, …, 10 Qt = Output capacity (time trend of aggregate exports) Ytf = world gross domestic products (USGDP) Yte = Expected output (time trend of Yt) 22
Tools for Impact Analysis and Medium & Long-term Projections Recursive Dynamic Applied General Equilibrium Model Exogenous saving & exogenous growth Used for analyzing the impacts of some certain shocks and policy changes as well as long-term potential GDP projection Ramsey-Cass-Koopmans Dynamic General Equilibrium Model Perfect foresight dynamic CGE model Endogenous saving but exogenous growth Mostly used for analyzing of economic shocks, i. e. oil shock and agricultural TFP shock 23
Recursive Dynamic AGE model § Based on exogenous-saving growth model discussed in Solow (1956) and Swan (1956) § 76 production sectors, 1 representative household and 2 primary production factors Example: results from recursive dynamic AGE model (Potential GDP Growth) GDP (% growth) 51 -55 56 -60 61 -65 66 -70 Avg. Base case 4. 96 4. 75 4. 35 3. 87 4. 48 High case 5. 42 5. 65 5. 55 5. 21 5. 46 24
Ramsey-Cass-Koopmans Dynamic CGE Model • Base on Neoclassical growth theory of the Ramsey-Cass. Koopmans type • Extension to incorporate investment adjustment cost and embodied Q theory • Captures both macro (intertemporal) and micro (intratemporal) efficiencies • Solve one for all period such that both intertemporal and intratemporal efficiencies are satisfied • • It is a dynamic CGE model for a small open-economy Single household, 12 production sectors, one government Solve for 100 period horizon with totally 36, 988 single equations Can be divided into five blocks, households, firms, foreign trade, within period equilibrium conditions and steady-state terminal conditions. 25
Example: results from R-C-K Dynamic CGE Model (Impacts of Oil Prices Shock on the Thai Economy) GDP C I G X M CPI %change from 2003 2004 -0. 156 -0. 552 -1. 627 3. 309 -1. 169 -1. 814 0. 400 2005 -0. 554 -1. 483 -2. 771 7. 910 -3. 041 -3. 883 1. 159 2006 -0. 866 -2. 037 -3. 490 10. 504 -4. 131 -4. 951 1. 616 2007 -1. 027 -2. 277 -3. 658 11. 275 -4. 596 -5. 358 1. 816 %change from previous year 2004 -0. 156 -0. 552 -1. 627 3. 309 -1. 169 -1. 814 0. 400 2005 -0. 401 -0. 940 -1. 159 4. 452 -1. 897 -2. 113 0. 757 2006 -0. 312 -0. 563 -0. 741 2. 404 -1. 119 -1. 108 0. 455 2007 -0. 161 -0. 240 -0. 177 0. 692 -0. 486 -0. 427 0. 197 %change from previous year over 1 % change in crude oil price 2004 -0. 016 -0. 047 0. 096 -0. 034 -0. 052 0. 012 2005 -0. 010 -0. 023 -0. 028 0. 109 -0. 046 -0. 052 0. 019 2006 -0. 018 -0. 032 -0. 043 0. 137 -0. 064 -0. 063 0. 026 2007 -0. 049 -0. 067 0. 004 0. 065 -0. 131 -0. 101 0. 057 Average 04 -07 -0. 020 -0. 035 -0. 029 0. 102 -0. 069 -0. 067 0. 028 26
Example: results from R-C-K Dynamic CGE Model (impacts of Oil V. S agricultural TFP shocks during 2004 -2005) Table C 1: The impacts of oil and agricultural TFP shocks on aggregate variables GDP C I G X M CPI Oil shock (% from previous year) 2004 -0. 157 -0. 630 -1. 159 3. 528 -1. 260 -1. 739 0. 503 2005 -0. 403 -0. 953 -1. 063 4. 683 -1. 910 -2. 029 0. 769 Average -0. 280 -0. 792 -1. 111 4. 106 -1. 585 -1. 884 0. 636 Agricultural TFP shock (% from previous year) 2004 -0. 860 -0. 777 -0. 850 -0. 312 -0. 928 -0. 678 0. 614 2005 -0. 862 -0. 757 -0. 702 -0. 348 -1. 005 -0. 679 0. 610 Average -0. 861 -0. 767 -0. 776 -0. 330 -0. 966 -0. 679 0. 612 Sources: NESDB-RCK-CCB CGE simulation 27
Experiences Analytical tools for economic projections: no single tool suit for all purposes • For NESDB, CQM is the most available efficient tool for shortrun & rapid estimation (times & resources) • Nevertheless, QFM is needed (for the purpose of both reconciliation and longer-term projection) • Under some certain conditions (shocks to the variables that are not included in QFM, structural models (i. e. CGE models) are useful 28
Experiences Key success factors • Technical skills • Understanding of economic structures and economic conditions • Databases, models & assumptions Problems • To release advanced estimates (base on QFM and CQM forecasts), their precision is the main obstacle. • Strong and abrupt shocks reduce precision of CQM forecasts • Fast changes in global condition made it more difficult to update exogenous variables in QFM and thus reduce its precision. • Judgments rise with projection horizon 29
Challenges • To officially release advance estimated of quarterly GDP • To improve/construct/ find a better econometric model for quarterly and yearly forecast • To form an efficient forecasting team 30
- Slides: 30