Business and consumer surveys: a user perspective Joint EC - OECD Workshop Brussels 15 November 2005 Benoît Robert Research Department National Bank of Belgium Use of survey indicators at the NBB mainly Belgian indicators (surveys conducted by the NBB), also EC-OECD indicators every-day use for short-term analysis contemporaneous assessment: activity, demand, turning points ex-post explanation: specific questions, case studies forecast: indicators included in short-term models p.m. not in the quarterly model of the Bank Business confidence in Belgium: overall synthetic curve (seasonally adjusted data) 10 Identification of turning points (removing irregular components) 5 0 -5 -10 Latest developments -15 -20 -25 -30 1980 1982 1984 1986 1988 Smoothed series Source: NBB. 1990 1992 1994 1996 1998 p.m. average 1980-2004 2000 2002 2004 Gross series Smoothing of Belgian business confidence 0 -2 Oct -4 Moving average -6 of median values Jul-Aug -8 -10 Jun -12 -14 2004 2005 Smoothed series Source: NBB. Lower forecast Upper forecast Gross series Major properties of survey indicators for economic analysis correlated with activity and demand aggregates, employment, inflation rapidly and regularly available no revision specific type of information: qualitative, expectations, explanatory factors Trade-off: low volatility ↔ timeliness Overall business confidence and GDP 6 15 5 10 4 5 3 0 2 -5 1 -10 0 -15 -1 -20 -2 -25 1996 1997 1998 1999 2000 GDP (y-o-y growth, left-hand scale) Sources: NAI, NBB. 2001 2002 2003 Smoothed series 2004 2005 Gross series Consumer confidence and private consumption 5 20 4 15 10 3 5 2 0 Oil 1 -5 -10 0 -15 -1 9/11 Iraq Lay-offs -20 -2 -25 1996 1997 1998 1999 2000 2001 Private consumption (y-o-y growth, left-hand scale) Sources: NAI, NBB. 2002 2003 2004 2005 Consumer confidence Uses of components of the overall indicators questions for industry assessment of total order book: contemporaneous for demand demand forecast : leading for demand employment forecast and quarterly shortages of skilled labour (DCU): leading for employment selling price expectations (EC consumer): inflation forecast case studies (shocks): perceived inflation following introduction of euro, impact of lay-offs on consumer confidence, explanation for weak business investment (DCU) GDP forecast: Survey indicators used in two models with different characteristics Bridge models Generalized dynamic factor model Individual simple equations Extensive and integrated model Limited number of variables Very large number of variables "Arbitrary" chosen Bridge models Monthly indicators → quarterly NA aggregates. Two steps: 1. quarterly equation: ∆ GDP = f (∆ hard data, ∆ soft data) VAT, industrial production business surveys additional information of surveys: limited but significant 2. monthly ARIMA-equations to forecast missing months of indicators: surveys = more readily available → fewer months to forecast Release calendar: October 2005 Industrial production August (13 Oct) 6 GDP flash Q3-2005 (26 Oct) 3 3.0 0 2.5 -3 2.0 Y-o-y Q-o-q Overall business confidence October (25 Oct) 1.5 1.0 0 0.5 -3 -6 -15 Y-o-y Smoothed Gross Q-o-q Sep-05 Jun-05 Mar-05 Dec-04 Sep-04 Mar-04 -12 Jun-04 0.0 -9 Generalized dynamic factor models (GDFM) Search for common component (CC) in a very large number of series business cycle e.g. business and consumer surveys CC idiosyncratic Generalized dynamic factor models: Uses cyclical properties of indicators: degree of commonality: how large is the business cycle component? lead or lag with respect to CC of GDP (= reference)? forecast of GDP: use of CC of indicators Generalized dynamic factor models: Data set 509 indicators for the Belgian economy Non-Survey Survey 185 Activity 55 Labour market 32 Prices 47 Financial 51 -Interest rates 22 -Asset prices 29 Business confidence Consumer confidence 324 -Overall -Manufacturing -Trade -Building -Capacity -Belgium -Non-Belgium -NBB -IFO -ISM -INSEE -CBS -BOJ -OECD -Belgium -Non-Belgium -NBB -EC 247 77 GDFM results: Cyclical behaviour (1) Coincident Leading Lagging Total 51% 22% 27% Non-survey 38% 25% 37% 0% 97% 3% Survey 59% 20% 21% Business confidence 66% 24% 10% Consumer confidence 35% 9% 56% Asset prices GDFM results: Cyclical behaviour (2) Most leading surveys: ISM indicators OECD indicators (US, UK, Japan) IFO building sector NBB building sector Most lagging surveys: EC consumer confidence indicators Individual results Important to assess the value of an indicator Conclusion: needs of the surveys user General need (analysis-forecast): deliverance of an unambiguous message this means timely and non-revised data but also release of only one (best) serie which is the common reference for all users, including general public danger of double publication based on same underlying data