Forecasting Tax Revenues in Latvia: Analysis and Models. Velga Ozolina, Astra Auzina-Emsina, Remigijs Pocs Riga Technical University, Latvia

Transcription

1 Forecasting Tax Revenues in Latvia: Analysis and Models Velga Ozolina, Astra Auzina-Emsina, Remigijs Pocs Riga Technical University, Latvia

2 CSB data Data Analysis Ministry of Finance data State Revenue Service (SRS) data Eurostat data

3 Tax Burden in Latvia in , % of GDP 35 % Data Source: CSB database

4 Tax Burden in the EU Countries in 2012, % of GDP % Denmark Belgium France Austria Sweden Italy Finland EU-28 Germany Luxembourg Netherlands Hungary Slovenia United Greece Croatia Cyprus Czech Republic Portugal Malta Spain Estonia Poland Ireland Romania Slovakia Latvia Bulgaria Lithuania Data Source: Eurostat database

5 Tax Revenues in Latvia (ESA95 methodology), m EUR m EUR 2500 m EUR Value Added Tax Customs Duties Excise Taxes Personal Income Tax Corporate Income Tax Social Contributions Other Taxes Nominal GDP (right axes) Data Source: CSB database

6 Social Contributions in m EUR Latvia, m EUR ESA95 methodology National methodology Ratio (right axes) Data Source: CSB database, Ministry of Finance data

7 Analysis of Legal Aspects The main laws in the group of direct taxes are: On State Social Insurance, On Personal Income Tax, On Corporate (Enterprise) Income Tax, Micro-enterprise Tax Law.

8 Analysis of Legal Aspects Employed persons by professional status, thsd thsd Employers (owners) Family workers Employees (workers), right axes Self-employed thsd 1000 Self-employed in the second job thsd thsd I III I III I III I III I III I III I III I III I III I III I III Employed Self-employed, right axes Overall statistics (CSB) Taxpayers (SRS) Data Source: CSB database, State Revenue Service data

9 Analysis of Legal Aspects The main laws in the group of indirect taxes are: Value Added Tax Law (before 2013 law On Value Added Tax), On Excise Duty.

10 Seasonality Analysis Revenues of Direct Taxes, m EUR m EUR Social Contributions Personal Income Tax Enterprise Income Tax Data Source: Ministry of Finance data

11 Quarterly Seasonal Indexes for Social Contributions Q1 Q2 Q3 Q Data Source: Ministry of Finance data

12 Quarterly Seasonal Indexes for Personal Income Tax Q1 Q2 Q3 Q

13 Quarterly Seasonal Indexes for Corporate Income Tax Q1 Q2 Q3 Q

14 Seasonality Analysis Revenues of Indirect Taxes, m EUR m EUR Value Added Tax Excise Duty Data Source: Ministry of Finance data

15 Quarterly Seasonal Indexes for Value Added Tax Data Source: Ministry of Finance data Q1 Q2 Q3 Q4

16 Quarterly Seasonal Indexes for Excise Duty Q1 Q2 Q3 Q Data Source: Ministry of Finance data

17 Productivity and Economic Activity Analysis Labor productivity and real GDP growth rate ( ) ( ) ( ).

18 Methodology Modelling Approaches Models and Equations Monthly Quarterly Annual

19 Monthly Data Seasonality Indexes Corporate Income Tax Revenues

20 Corporate Income Tax Revenues CIT revenues = coef monthly * CIT revenues lag *(1 + + PCI infl /100)/12 + coef may *PROF lag / where CIT revenues corporate income tax revenues, CIT revenues lag annual corporate income tax revenues with 17-month lag, coef monthly corporate income tax advance payments coefficient, PCI infl annual growth rate of private consumption price index in the previous year, coef may corporate income tax revenues coefficient applied only in May, PROF lag annual profit in the previous year % Corporate income tax revenues coefficient, % m EUR Annual profit in the previous year, m EUR

21 Corporate Income Tax Revenues SEE = 0.32 RSQ = RHO = 0.45 Obser = 196 from SEE+1 = 0.28 RBSQ = DW = 1.11 DoFree = 190 to MAPE = 9.70 Variable name Reg-Coef Mexval Elas NorRes Mean Beta 0 LTAX_UIN intercept D_5*@log(IM[6])

22 Identities Quarterly Data Econometric Equations

23 Identities tax_rev = taxr_coef*taxr*tax_base, Tax rate coefficient of social contributions where tax_rev tax revenues, Tax rate coefficient of personal income tax 0.6 taxr_coef tax rate coefficient, taxr tax rate, tax_base tax base I III I III I III I III I III I III I III I III I III I III I III I III I III I III I III I III I III I III I III I III I III I III I III I III I III I III I III I III I III I III I III I III I III Tax rate coefficient of the value added tax Tax rate coefficient of excise duty, right axes

24 Corporate Income Tax CIT revenues = coef q * CIT revenues lag *(1 + PCI infl /100)/ coef II *PROF lag /100, where CIT revenues corporate income tax revenues, CIT revenues lag annual corporate income tax revenues with 2- year lag (quarter 1), with 1- year lag (quarters 3 and 4) or weighted average of the 1-year and 2-year lag (quarter 2), coef q corporate income tax advance payments coefficient, PCI infl annual growth rate of private consumption price index in the previous year, coef II corporate income tax revenues coefficient applied only in the quarter 2, PROF lag annual profit in the previous year I III I III I III I III I III I III I III I III I III I III I III I III I III I III I III I III I III Advance payments coefficient Tax revenues coefficient, right axes

25 Social Contributions Revenues SEE = RSQ = RHO = 0.43 Obser = 48 from SEE+1 = RBSQ = DW = 1.14 DoFree = 45 to MAPE = 2.39 Variable name Reg-Coef Mexval Elas NorRes Mean Beta 0 TAX_SOC intercept TAXR_SOC*((EMPL*W_NOM*3)/100000) D_EU

26 Personal Income Tax Revenues SEE = RSQ = RHO = 0.14 Obser = 48 from SEE+1 = RBSQ = DW = 1.72 DoFree = 45 to MAPE = 3.24 Variable name Reg-Coef Mexval Elas NorRes Mean Beta 0 TAX_INC_PERS intercept TAXR_IIN*((EMPL*(W_NOM-TAX_NMIN))/1000-TAX_SOC*TAX_SOC_E) D_

27 Corporate Income Tax Revenues SEE = 0.26 RSQ = RHO = 0.21 Obser = 72 from SEE+1 = 0.25 RBSQ = DW = 1.58 DoFree = 67 to MAPE = 5.23 Variable name Reg-Coef Mexval Elas NorRes Mean Beta 0 LTAX_INC_CORP intercept D_2*@log(INV_CP[1])

28 Value Added Tax Revenues SEE = 0.10 RSQ = RHO = 0.13 Obser = 76 from SEE+1 = 0.09 RBSQ = DW = 1.74 DoFree = 72 to MAPE = 1.37 Variable name Reg-Coef Mexval Elas NorRes Mean Beta 0 LTAX_VAT intercept D_EU

29 Excise Duty Revenues SEE = 0.08 RSQ = RHO = 0.06 Obser = 60 from SEE+1 = 0.08 RBSQ = DW = 1.87 DoFree = 57 to MAPE = 1.25 Variable name Reg-Coef Mexval Elas NorRes Mean Beta 0 LTAX_EXC intercept D_

30 Annual Data Calculations in ESA95 Identities Transformation Coefficients to forecast national data

31 Dynamics of Estimated II Pillar Rates in Estimated II pillar rate (based on assets) Estimated II pillar rate (based on differences in national and ESA95 methodologies) Estimated II pillar rate (combined) Max II pillar rate

32 Tax Rate Coefficients Tax rate coefficient of personal income tax Tax rate coefficient of corporate income tax Tax rate coefficient of social contributions Tax rate coefficient of the value added tax Tax rate coefficient of excise duty, right axes

33 Forecasts Evaluation of precision Numbers

34 Corporate Income Tax (Monthly Data) MAPE = 14.3% Modified dummy MAPE = 3.6%

35 MAPE Values, % (Quarterly Data) Tax Type 2014 I 2014 I and II Social Contributions Personal Income Tax Corporate Income Tax Value Added Tax Excise Duty

36 Comparison of Forecasts Tax Type Identity-based approach Econometric equations Monthly Quarterly Annual Monthly Quarterly Social Contributions Personal Income Tax Corporate Income Tax Value Added Tax Excise Duty

37 Conclusions Using annual data, identity-based approach should be prefered, however quarterly and monthly data can give similar results and thus the choice is in hands of the model-user Identity-based approach allows for a greater flexibility in scenario-building process. Econometric approach involves less assumptions and thus may seem to be more objective Forecasts depend very much on the values of exogenous indicators, therefore modelling approaches should be tested regulary to find the most reliable ones

38 LAIMA Latvian Interindustry Model (Aggregated/Annual) Goddess of destiny

39 Thank You for Attention Questions?