Conference ”Estimating and Interpreting Financial Cycles” Opening by S J Koopman http://sjkoopman.net Department of Econometrics, Vrije Universiteit Amsterdam CREATES, Aarhus University De Nederlandsche Bank, 2 September 2016, Amsterdam
W E L C O M E to DNB in Amsterdam Many thanks to the organizers: • Gabriele Galati (Monetary Policy) • Peter van Els (Econometrics and Modelling) • Jasper de Winter (Econometrics and Modelling) • Irma Hindrayanto (Econometrics and Modelling) • Gita Gajapersad (Secretariat) • Claudia Trivieri (Travel and Conference Services) 2 / 26
Workshop on Estimating and Interpreting Financial Cycles 2 September 2016 De Nederlandsche Bank Program 9:00-9:30 Registration with coffee/tea 9:30-10:00 Siem Jan Koopman (Vrije Universiteit Amsterdam) Challenges in estimating financial cycles SESSION 1 10:00-11:00 Mathias Drehmann (Bank for International Settlements) Methodologies to measure the financial cycle Discussant: Andreas Pick (De Nederlandsche Bank and Erasmus University Rotterdam) 11:00-11:30 Coffee 11:30-12:30 Mikael Juselius (Bank of Finland) Monetary policy and ultra-low interest rates: A financial cycle perspective Discussant: Marente Vlekke (Centraal Planbureau) 12:30-13:30 Buffet Lunch SESSION 2 13:30-14:30 Gerhard Rünstler (European Central Bank) Business, Housing and Credit Cycles Discussant: Gabriele Galati (De Nederlandsche Bank) 14:30-15:30 Irma Hindrayanto (De Nederlandsche Bank) Modeling the business and financial cycle in a multivariate structural time series model Discussant: Gabriel Pérez-Quirós (Bank of Spain) 15:30-16:00 Coffee 16:00-17:00 Oscar Jorda (Federal Reserve Bank of San Francisco) Macrofinancial History and the New Business Cycle Facts Discussant: Marco Terrones (International Monetary Fund) 17:00-17:30 Stijn Claessens (Federal Reserve Board) Using financial cycle estimates in policy 17:30-18:30 Farewell drinks 3 / 26
Financial Cycle (FC) The financial cycle captures systematic patterns in the financial system that can have important macroeconomic consequences, as formulated by Borio, Furfine and Lowe (2001). FC is typically characterized by co-movements of medium-term cycles in credit, the credit-to-GDP ratio and house prices while its peaks tend to coincide with onsets of financial crises, see Drehmann, Borio and Tsatsaronis (2012). Several methods are developed for measuring FCs including • variations of the Burns and Mitchell (1946) turning-point analysis, see Claessens, Kose and Terrones (2011, 2012) • nonparametric bandpass filters, see Aikman et al. (2015) and Sch¨ uler et al. (2015). • UC methods based on state space methods, see Hindrayanto et al. (2016), R¨ unstler and Vlekke (2016), Koopman et al. (2016). 4 / 26
Acronyms • BIS – Bank of International Settlement • DNB1 – De Nederlandsche Bank (Division FS) • DNB2 – De Nederlandsche Bank (Division Ebo) • BOE – Bank of England • BOJ – Bank of Japan • BOL – Bank of Lithuania • OF – Ortec Finance • BB – Bundesbank • BOF – Bank of Finland • FC – Financial cycle • RTC – Real Total Credit • CGR – Credit to GDP Ratio • RRP – Real Residential Property Prices • REP – Real Equity Prices 5 / 26
Estimates of the U.S. Financial Cycle BIS_FC BIS_FC DNB1_FC DNB1_FC DNB2_FC? DNB2_FC? 0.1 0.25 0.0 0.0 0.00 -0.1 -0.1 -0.25 1980 2000 1980 2000 1980 2000 0.10 BOE_FC BOE_FC BOJ_FC BOJ_FC BOL_FC BOL_FC 0.1 0.1 0.0 0.05 -0.1 -0.1 1980 2000 1980 2000 1980 2000 0.2 OF_FC OF_FC 0.1 BB_FC? BB_FC? 0.2 BOF_FC BOF_FC 0.0 0.0 0.0 1980 2000 1980 2000 1980 2000 6 / 26
Estimates of the U.S. Financial Cycle: all in one 0.3 BIS_FC BIS_FC DNB1_FC DNB1_FC DNB2_FC? DNB2_FC? BOE_FC BOE_FC BOJ_FC BOJ_FC BOL_FC BOL_FC OF_FC OF_FC BB_FC? BB_FC? BOF_FC BOF_FC 0.2 0.1 0.0 -0.1 -0.2 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 7 / 26
Correlation matrix method: BP CF UC NP HP PCA NP Wav CVAR BIS DNB1 DNB2 BOE BOJ BOL OF BB BOF BIS 1.00 DNB1 0.70 1.00 DNB2 0.92 0.56 1.00 BOE 0.40 0.81 0.30 1.00 BOJ 0.65 0.90 0.48 0.83 1.00 BOL 0.06 0.41 -0.15 0.61 0.47 1.00 OF 0.91 0.43 0.92 0.14 0.40 -0.14 1.00 BB 0.06 0.33 -0.10 0.57 0.45 0.44 -0.17 1.00 BOF -0.27 -0.64 -0.04 -0.58 -0.65 -0.73 -0.08 -0.26 1.00 8 / 26
Principal Components PC# Eigenvalues % variation % cumulative 1 4.687 52.08 52.08 2 2.587 28.74 80.83 3 0.8089 8.99 89.81 4 0.4091 4.55 94.36 5 0.2602 2.89 97.25 6 0.09800 1.09 98.34 7 0.09497 1.06 99.39 8 0.03320 0.37 99.76 9 0.02129 0.24 100.00 9 / 26
Eigenvectors PC1 PC2 PC3 PC4 BIS 0.37032 -0.34884 0.017974 -0.21917 DNB1 0.43254 0.025954 -0.020163 0.42316 DNB2 0.30043 -0.45222 0.084131 -0.063972 BOE 0.38104 0.23780 0.19841 0.30476 BOJ 0.43169 0.085877 0.074512 0.26785 BOL 0.23275 0.42527 -0.31832 -0.55359 OF 0.26683 -0.47238 -0.072259 -0.37049 BB 0.18427 0.34883 0.72755 -0.39822 BOF -0.30764 -0.28904 0.55803 0.043308 10 / 26
Estimates of the U.S. Financial Cycle: First 4 PCs 5.0 PC1 PC1 PC2 PC2 2.5 2.5 0.0 0.0 -2.5 -2.5 -5.0 1970 1980 1990 2000 2010 1970 1980 1990 2000 2010 PC3 PC3 PC4 PC4 2 1 1 0 0 -1 -1 1970 1980 1990 2000 2010 1970 1980 1990 2000 2010 11 / 26
Estimates of the U.S. Financial Cycle: First PC 5 4 3 2 1 0 -1 -2 -3 -4 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 12 / 26
Estimates of the U.S. Financial Cycle: First two PC 4 2 0 -2 -4 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 13 / 26
Estimates of the U.S. Financial Cycle: First three PC 6 4 2 0 -2 -4 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 14 / 26
Estimates of the U.S. Financial Cycle: cf PPC123 PPC123 BIS_FC BIS_FC 0.10 0.05 0.00 -0.05 -0.10 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 15 / 26
Estimates of the U.S. Financial Cycle: cf PPC123 PPC123 DNB2_FC DNB2_FC 0.3 0.2 0.1 0.0 -0.1 -0.2 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 16 / 26
Common dynamic factors We can also decompose the nine FC estimates by four dynamic factors : FC i , t = a i ψ 1 , t + b i ψ 2 , t + c i ψ 3 , t + d i ψ 4 , t + u i , t , where ψ j , t is a cyclical dynamic component and u i , t is an error, for i = 1 , . . . , 9 and j = 1 , . . . , 4. We have used the STAMP program to carry out the estimation. 17 / 26
Goodness-of-fit R 2 BIS 0.99873 DNB1 0.85312 DNB2 0.99203 BOE 0.97467 BOJ 0.84678 BOL 0.59206 OF 0.99947 BB 0.45042 BOF 0.56713 18 / 26
Prediction residuals 2 5.0 2 2.5 0 0 0.0 -2.5 -2 -2 1980 2000 1980 2000 1980 2000 2 5.0 2 2.5 0 0 0.0 -2 -2.5 -2 1980 2000 1980 2000 1980 2000 2 2 2 0 0 0 1980 2000 1980 2000 1980 2000 19 / 26
Estimates of Financial Cycle for Netherlands 0.3 0.50 BIS_FC BIS_FC 0.1 DNB1_FC DNB1_FC DNB2_FC DNB2_FC 0.2 0.25 0.1 0.0 0.00 0.0 -0.25 -0.1 1980 2000 1980 2000 1980 2000 0.10 0.2 0.2 BOE_FC BOE_FC BOJ_FC BOJ_FC BOL_FC BOL_FC 0.1 0.0 0.05 0.0 -0.1 -0.2 1980 2000 1980 2000 1980 2000 0.2 0.1 OF_FC OF_FC BB_FC BB_FC BOF_FC BOF_FC 0.25 0.1 0.00 0.0 0.0 -0.25 -0.1 1980 2000 1980 2000 1980 2000 20 / 26
Estimates of Financial Cycle for NL: all in one 0.3 0.2 0.1 0.0 -0.1 -0.2 -0.3 BIS_FC BIS_FC DNB1_FC DNB1_FC DNB2_FC DNB2_FC BOE_FC BOE_FC BOJ_FC BOJ_FC BOL_FC BOL_FC -0.4 OF_FC OF_FC BB_FC BB_FC BOF_FC BOF_FC 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 21 / 26
Correlation matrix – NL method: BP CF UC NP HP PCA NP Wav CVAR BIS DNB1 DNB2 BOE BOJ BOL OF BB BOF BIS 1.00 DNB1 0.70 1.00 DNB2 0.76 0.66 1.00 BOE 0.33 0.50 0.07 1.00 BOJ 0.37 0.58 0.17 0.81 1.00 BOL -0.03 0.31 0.10 0.45 0.47 1.00 OF 0.81 0.37 0.85 -0.10 0.04 -0.16 1.00 BB -0.02 0.39 0.08 0.16 0.10 0.25 -0.20 1.00 BOF -0.06 -0.49 -0.00 -0.75 -0.72 -0.66 0.28 -0.48 1.00 22 / 26
Principal Components PC# Eigenvalues % variation % cumulative PC1 3.844 42.71 42.71 PC2 2.742 30.47 73.18 PC3 1.064 11.82 85.00 PC4 0.6638 7.38 92.37 PC5 0.2272 2.52 94.90 PC6 0.1991 2.21 97.11 PC7 0.1423 1.58 98.69 PC8 0.1060 1.18 99.87 PC9 0.01170 0.13 100.00 23 / 26
Estimates of Financial Cycle for NL: First 4 PCs PC1 PC1 PC2 PC2 2 2.5 0 0.0 -2 -2.5 1970 1980 1990 2000 2010 1970 1980 1990 2000 2010 3 PC3 PC3 PC4 PC4 2 1 1 0 0 -1 -1 -2 -2 1970 1980 1990 2000 2010 1970 1980 1990 2000 2010 24 / 26
Conclusions • Financial Cycle : it is an unobservable variable !! • Only noisy indicators associated with FC can be observed. • Problem and challenge is comparable to Business Cycle. • However, the definition of the FC is more diverse. • Indicators are more noisier and not necessarily typical Gaussian variables. • Many challenges and more work required !! 25 / 26
Conclusions • Financial Cycle : it is an unobservable variable !! • Only noisy indicators associated with FC can be observed. • Problem and challenge is comparable to Business Cycle. • However, the definition of the FC is more diverse. • Indicators are more noisier and not necessarily typical Gaussian variables. • Many challenges and more work required !! Let’s make some progress today !! 25 / 26
Recommend
More recommend
