Down the rabbit-hole : Does monetary policy impact differ during the housing bubbles?

Transcription

1 Down the rabbit-hole : Does monetary policy impact differ during the housing bubbles? T. Reichenbachas 1 1 Bank of Lithuania and Vilnius University Vilnius, Lithuania Recent trends in the real estate market and its analysis 2017 edition, November 2017 Tomas Reichenbachas Monetary policy and housing bubbles 2017 November 22 1 / 22

2 Disclaimer The views expressed herein are solely those of the authors and do not necessarily reflect the views of the Bank of Lithuania. Tomas Reichenbachas Monetary policy and housing bubbles 2017 November 22 2 / 22

3 Outline Motivation Literature overview Identifying housing price bubbles: some descriptive statistics The methodology Empirical results Model fit Regimes fit Regime-dependent dynamic effects of structural shocks Conclusions Tomas Reichenbachas Monetary policy and housing bubbles 2017 November 22 3 / 22

4 Motivation (1) Tomas Reichenbachas Monetary policy and housing bubbles 2017 November 22 4 / 22

5 Motivation (2) Should monetary policy react to asset prices? Bernanke and Gertler (2001) Given a strong commitment to stabilizing expected inflation, it is neither necessary nor desirable for monetary policy to respond to changes in asset prices, except to the extent that they help to forecast inflationary or deflationary pressures. Sveriges Riksbank (2013) One problem with not taking financial imbalances into account when considering monetary policy is that target attainment may appear to be good in the short term, at the same time as one misses the fact that financial imbalances are building up that increase the risk of poor target attainment in the longer term. Tomas Reichenbachas Monetary policy and housing bubbles 2017 November 22 5 / 22

6 Motivation (3) Svensson (2017)...benchmark estimates and reasonable assumptions the result is that the costs of leaning-against-the-wind exceed the benefits by a substantial margin. Tomas Reichenbachas Monetary policy and housing bubbles 2017 November 22 6 / 22

7 Motivation (4) Housing market booms start gathering pace again... It remains unclear if tighter monetary policies could have been effective in containing housing and credit bubbles and sparing the economies from economic and financial fallout associated with them. The convenience of linear models is not a sufficient cause to maintain that no fundamental changes in economic dynamics occur when the economy goes into a different state (Hamilton (2016)) Tomas Reichenbachas Monetary policy and housing bubbles 2017 November 22 7 / 22

8 Literature review (1) Monetary policy shocks with credit market and asset prices. Incorporating the housing market is crucial for the identification of monetary policy shock transmission channels (Musso et al., (2011); Iacoviello and Minetti, (2008); Elborne, (2008)) Macroeconomic regimes and regime shifts. The economy can be governed by few separate regimes: pre-volcker and post-volcker monetary policy (Murray et al. (2015)); monetary policy under the zero-lower-bound constraint (Hirokuni (2016)); monetary policy during financial stress (Hubrich et al. (2014)) Regime switches in monetary policy. Regime-switching monetary policy block (Bernanke and Mihov (1998); Sims and Zha (2006)) Tomas Reichenbachas Monetary policy and housing bubbles 2017 November 22 8 / 22

9 Literature review (2) Asset prices and regime shifts. identify boom-bust regimes in real estate market (Ceron et al. (2006); Corradin et al. (2013); Nneji et al. (2013)) Housing price prices bubbles and monetary policy. just two attempts (Simo-Kengne et al. (2013) and Chang et al. (2011)) to analyse the differences of monetary policy shocks during different housing price regimes: Simo-Kengne et al. (2013) examine asymmetries in the impact of monetary policy on the middle segment of the South African housing market.; Chang et al. (2011) utilize the MS-VAR approach to analyse the impact of monetary policy on housing returns for the US. Tomas Reichenbachas Monetary policy and housing bubbles 2017 November 22 9 / 22

10 Data All data are quarterly. Data are taken from the OECD database with the exception of the housing price and credit data, which are taken from the BIS databases. The length of the data sample is country dependent: Switzerland (from 1980q1 to 2017q1); Norway (from 1992q1 to 2017q1); United Kingdom (from 1978q1 to 2017q1); Sweden (from 1993q1 to 2017q1) Canada (from 1981q1 to 2017q1) It is crucial to note, that the housing and credit data are not necessarily fully harmonised. Tomas Reichenbachas Monetary policy and housing bubbles 2017 November / 22

11 Identifying housing price bubbles: historical trends (1) Tomas Reichenbachas Monetary policy and housing bubbles 2017 November / 22

12 Identifying housing price bubbles: historical trends (2) Tomas Reichenbachas Monetary policy and housing bubbles 2017 November / 22

13 Identifying housing price bubbles: historical trends (3) Tomas Reichenbachas Monetary policy and housing bubbles 2017 November / 22

14 The methodology - model Following Sims and Zha (2006), Sims et al. (2008) and Lhuissiers (2017), in our paper we use the Markov-switching structural BVAR models to capture regime-switching in real estate market. The model can be expressed in following form: y ta 0 (s t ) = p i=1 y t i A i(s t ) + C(s t ) + ε tξ 1 (s t ), t = 1,..., T, In our baseline specification we use 5 variables: quarterly real GDP Consumer Price Index respective interbank lending rate quarterly real household credit real housing prices index We are applying a Cholesky decomposition to recover structural shocks. Tomas Reichenbachas Monetary policy and housing bubbles 2017 November / 22

15 Empirical results - Model fit Model: UK NO SE CH CA with time invariant coefficients with 2-regimes in all equations with 2-regimes in RE equation with 2-regimes in RE and credit equations Note: The marginal data densities (MDDs) are computed based on Sims et al. (2008). For the constant parameter model, we use the Chib (1995) procedure. It is worth to emphasise that the MDDs in that table is provided in Log-likelihood scale so that differences of one or two in absolute value mean little, while differences of ten or more imply significant odds in favour of the higher-marginal-data density model. Tomas Reichenbachas Monetary policy and housing bubbles 2017 November / 22

16 Empirical results - Regimes fit (CH) Tomas Reichenbachas Monetary policy and housing bubbles 2017 November / 22

17 Empirical results - Regimes fit (NO) Tomas Reichenbachas Monetary policy and housing bubbles 2017 November / 22

18 Empirical results - Regimes fit (UK) Tomas Reichenbachas Monetary policy and housing bubbles 2017 November / 22

19 Empirical results - Regimes fit (SE) Tomas Reichenbachas Monetary policy and housing bubbles 2017 November / 22

20 Empirical results - Regimes fit (CA) Tomas Reichenbachas Monetary policy and housing bubbles 2017 November / 22

21 Empirical results - Regime-dependent dynamic effects of structural shocks Tomas Reichenbachas Monetary policy and housing bubbles 2017 November / 22

22 Conclusions Main results: Data favour the models associated with changes in real estate equation coefficients in United Kingdom, Switzerland, and Canada. We do not find systemically significant differences in the responses of monetary policy shocks. Some additional thoughts: TV-VARs vs MS-VARs? The Great Mortgaging (Jorda et al. (2014)) and monetary policy transmission Tomas Reichenbachas Monetary policy and housing bubbles 2017 November / 22