Forecasting at the Ministry of Finance

Forecasting at the Ministry of Finance Åsa Andersson, PhD Senior Advisor, Division for Public Finances Anna Breman, PhD Deputy Director, Forecasting Division

Welcome to the Ministry of Finance

Outline 1. Forecasting at the Ministry of Finance Forecasting organisation Different types of forecasts; Macro and Public finances 2. The Macroeconomic Forecast Forecasting process Key variables forecasted Example: household consumption expenditures 3. Public Finances Forecast 4. The macroeconomic and public finances forecasts and the budget process

Economic Affairs Department Forecasting Division Division for Public Finances Division for Income Distribution Division for Macroeconomic Analysis Division for Labour Market and Education Analysis

Forecasting at the Ministry of Finance The Forecasting Division makes the macroeconomic forecast Division for Public Finances is responsible for the real forecasts of the local and central government income and consumption, production and labour market Budget Department: central government expenditure Forecasts are done by the civil servants, who present the forecasts to the political leadership

Forecasting at the Ministry of Finance 4 major forecasts per year December-January March ==> Spring Bill June August ==> Budget Bill Spring Bill and Budget Bill are published externally Several minor updates of forecasts if necessary

Forecasting at the Ministry of Finance Spring Bill and Budget Bill are written by the civil servants are co-ordinated in the Government Offices are negotiated by the politicians / the leadership in the coalition parties are decided by the government and handed over to the parliament

Spring Bill and the Budget Bill Spring Bill includes guidelines for coming economic policy and structural reforms Budget Bill presents the full government budget for the coming years and new economic policy and structural reforms

The Macroeconomic Forecast Responsibilities of the Forecasting Division Macroeconomic forecasts (t+4 years) Detailed, approx. 1600 economic variables are forecasted Include structural analysis of government policies Support the budget process by providing key numbers for the Budget Department Provide analysis and recommendations for economic policy

The Macroeconomic Forecasts is a collaboration between several departments Forecasting Division coordinates the forecasting process Forecasting Division has responsibility for forecasting: Swedish GDP and its components, productivity, output gaps, resource utilization, U.S. and Euro area GDP Macroeconomic Forecast get input from the following divisions Division of Public Finances (government consumption) Division for Labour Market and Education Analysis (labour market variables) Division for Financial Market Analysis (interest rates, exchange rates, oil prices)

Macro forecasting method framework Data (National Accounts) Quick indicators Quantitative models Judgement Forecast Short-term: year t, t+1 Medium-term: year t+2, t+3, t+4 General Equilibrium Model, including assumptions about potential growth etc.

Forecasting process Judgement-based forecasting Forecasts are an iterative process Models and Data support the process Final forecast Consistent with national accounts Provide an analysis of the medium- and long-term macroeconomic development Take into account information about structural economic reforms Judgement-based

Forecasting process Sweden is a small open economy Iterations, in the following order: 1. International and Financial variables 2. Short term forecast; Swedish GDP, inflation, wages, labour market 3. Medium-term Macro Forecasts and Output Gaps 4. Revised short-term and Medium-term GDP, Labour Market, Inflation and Wages 5. Iterations

The Macroeconomic Forecast Key variables forecasted 1. International and Financial variables United States and Euro-area GDP, World Market Demand, Interest rates, Exchange rates, Oil prices 2. Swedish GDP Consumption, Exports, Imports, Public Consumption, Investment, Inventories, Production 3. Output gap and Resource Utilization Potential GDP, NAIRU 4. Labour Market Unemployment, Labour market participation, Productivity 5. Inflation and Wages

Key variables In-house forecasts Large set of variables - approx. 1600 Quarterly data Externally published forecasts (Spring Bill and the Budget Bill): Include key variables Yearly forecasts for 5 years (currently 2011-2015)

Example: Forecasting household consumption expenditure National Accounts Quick indicators Fundamental factors Target variable Latest outcome Econometric models Evaluation of previous forecasts

Example: Forecasting household consumption expenditure National Accounts Quick indicators Fundamental factors For example: Retail sales Car registrations Household surveys House prices Stock Exchange Econometric models Evaluation of previous forecasts

Example, Household expectations about the economy 12 months ahead Households' expectations: Sweden's economy 12 months ahead Better - worse % 80 60 40 20 0-20 -40-60 80 60 40 20 0-20 -40-60 % -80 90919293949596979899000102030405060708091011-80 Source: Statistics Sweden's Consumer Survey The breaks in the time series are due to changes of method in Jan. 2000 and 2002.

Example: Forecasting household consumption expenditure National Accounts Economic theory Quick indicators Fundamental factors Econometric models i. Wealth position (financial and real) ii. Disposable income (employment) iii. Interest rates Evaluation of previous forecasts

Example, OLS regressions using actual consumption and quick indicators Jämförelse indikatormodeller av prognos på konsumtion (konsekutiv tillväxttakt) 0,025 0,020 0,015 0,010 0,005 0,000-0,005-0,010 2009Q1 2009Q3 2010Q1 2010Q3 2011Q1 2011Q3 BP12 juni DHI bilar CCI DHI bilar CCI 4 lag kons DHI bilar CCI 1 lag kons DHI bilar CCI 2 lag kons DHI bilar CCI 4 lag kons börs DHI bilar CCI 4 lag kons börs boindikator DHI bilar CCI 4 lag kons börs boindikator huspris DHI bilar 4 l ag kons börs 2 lag DHI 4 lag kons DHI bilar 4 l ag kons börs huspris 2 lag CCI 4 lag kons 2 lag CCI 4 lag boindikator kons AR-modell 4 lag utfall

Example, OLS regressions using actual consumption and quick indicators Comparison of average predictions of OLS models and our consumption forecast (quarterly data) 0,025 0,020 0,015 0,010 0,005 0,000-0,005 10Q1 10Q3 11Q1 11Q3 BP12 AR soft indicators hard indicators soft and hard indicators outcome

Example, Baysian VAR model using fundamental factors 1,2 1 0,8 0,6 0,4 0,2 0 2011Q1 2011Q3 2012Q1 2012Q3 2013Q1 2013Q3 2014Q1 2014Q3 2015Q1 2015Q3 Prognos Scenario 1 Scenario 2 Scenario 3

Medium-term forecast General Equilibrium Model KIMOD developed at the National Institute for Economic Research (NIER) Structural model Potential variables Consistent with national accounts Provides support for medium-term forecasts Calculate effects of structural reforms Policy Experiments

Example: Forecasting household consumption expenditure National Accounts Quick indicators Evaluation of forecasts Follow up forecast errors Fundamental factors Econometric models Evaluation of previous forecasts

Output Forecasting Division Macroeconomic Forecast for the Spring Bill and Budget Bill, including Key variables, yearly data (t, t+4) Analysis of macroeconomic scenario and structural reforms Alternative risk scenarios Detailed In-house Macroeconomic Forecasts Key numbers to the Division of Public Finances and the Budget Department

Division for Public Finances Tax forecasts general government tax revenue total income of the central government budget Local government sector real and financial forecasts Central government sector real forecasts Public finances compilation of public finances follow-up of budget goals (surplus target) reporting to EDP Other forecasts for household disposable income, operating surplus for businesses, unit labour costs (ULC) updating of Swedens convergence program to EU

Tax forecasts

Tax revenue as per cent of GDP Per cent of GDP 54 52 50 48 46 44 42 40 80 82 84 86 88 90 92 94 96 98 00 02 04 0 6 08 10 12 14 Skatt ekvot

Tax revenue as per cent of GDP 2006 2007 2008 2009 2010 Tax on labour 29,1 27,9 28,4 28,1 26,4 Tax on capital 6,5 6,7 5,1 5,2 5,7 Tax on consumption 12,7 12,8 13,1 13,7 13,6 Other taxes -0,1 0,0-0,1-0,1 0,1 Total tax revenue 48,3 47,4 46,4 46,8 45,8

Why tax forecasts? Important input in public finances ==> important input to the government in policy making Follow-up of fiscal goals surplus target expenditure ceiling

How tax forecasts? The tax forecasts depend on tax time horizon tax assessment (monthly or yearly)

How tax forecasts? Macroeconomic forecast Reforms and proposals Tax forecasts Other information

How tax forecasts? Strong connection to macroeconomic variables Wage bill, employment, hours worked ==> Income tax and SSC Household consumption ==> VAT Weak connection to macroeconomic variables Corporate income tax Capital gains tax

Example: VAT Normal tax rate: 25 % Reduced rates: 12 %: on foodstuffs, hotel accommodation, works of art supplied by the artist, import of art and antiques 6 %: books, papers, magazines, cultural activities (concerts, theatres, movies etc.), transportation of passengers 0 %: pharmaceutical products on prescription, ships for commercial traffic, aviation fuel No VAT: rents, health care and financial services

VAT revenue in Sweden 2010 SEK Billion Private sector Household consumption expenditure 179 55% Investments 42 13% Input consumption 38 12% Total VAT from private Botswana.xls sector - Blad3!A1 259 80% Public sector Municipalities 40 12% Central government authorities 27 8% Total VAT from public sector 66 20% Total VAT revenue 325 100%

Inputs in VAT forecast Macroeconomic forecasts Quarterly outcomes from Statistics Sweden (national accounts) Monthly outcomes from the central government accounting system

Example: household consumption Ministry of Finance forecasts (10 groups of goods) Statistics Sweden outcome (100 groups of goods) * Quarterly growth rates = Tax base Tax base * Tax rates = Theoretical VAT-revenue * Calibration factor = VAT revenue

Calibration factor Proportion between theoretical VAT and actual VAT outcome SEK Billion 2008 2009 2010 Theoretical VAT from private sector 254 248 267 Debited VAT (outcome from NFMA) 240 237 258 Calibration factor 94,5% 95,6% 96,6%

Following and adjusting the forecast Monthly outcomes from the Swedish Tax Agency Forecast calibration (mainly during the current year)

Forecasting in the budget process The macroeconomic forecast affects the budget process in several ways: The forecast of public finances The scope for policy reforms and the cyclically adjusted budget balance The calculations of various parts of the budget, e.g., through CPI and deflators

Macro forecast feeds into budget, directly and indirectly, examples Household consumption, current prices VAT income CPI Index-linked social security Nominal GNI Foreign aid; EU fees Interest rates Government debt interest payments Wages Taxes, pensions Unemployment Size of labour market policy programmes

Public finances are important in the budget process Key variables when describing an economy net lending debt cyclically adjusted measures Important in policy making scope for reforms follow-up of domestic fiscal framework (surplus target, expenditure ceiling) follow-up of requirements in the stability and growth pact

Summary Forecasting at the Ministry of Finance The Forecasting process at the Ministry A collaboration between different departments Key variables affect the other departments forecasts The Forecasting Division: Macroeconomic forecast Division for Public Finances: Local and central government income Budget Department: Central government expenditure