Discussion: What drives metropolitan house prices in California? Dr Kirdan Lees Tuesday, December 9, 2017 Sense Partners
Structure comments around three areas Brief outline of what the paper does How to improve the paper Make a case for using this type of model
What the paper does
What the paper does Dynamic Factor model to analyse Californian house prices Sets to one side forecasting for structural questions : the impact of regional vs local housing shocks the role of credit vs monetary policy shocks impact on house prices of regional demand and inflation Advocates specific identification scheme to identify shocks
Approach Aggregate US time series Californian time series 𝑍 𝑡 𝐶 𝑡 𝐻 𝑡 𝑖 𝑡 = 𝜆 𝑍𝑍 0 0 𝜆 𝑍𝑖 𝜆 𝐶𝑍 𝜆 𝐶𝐶 𝜆 𝐶𝐻 𝜆 𝐶𝑖 𝜆 𝐻𝑍 𝜆 𝐻𝐶 𝜆 𝐻𝐻 𝜆 𝐻𝑖 0 0 0 1 𝑓 𝑡 𝑍 𝑓 𝑡 𝐶 𝑓 𝑡 𝐻 𝑖 𝑡 + 𝜉 𝑡 𝑍 𝜉 𝑡 𝐶 𝜉 𝑡 𝐻 0 𝑓 𝑡 𝑍 𝑓 𝑡 𝐶 𝑓 𝑡 𝐻 𝑖 𝑡 =Φ(𝐿) 𝑓 𝑡−1 𝑍 𝑓 𝑡−1 𝐶 𝑓 𝑡−1 𝐻 𝑖 𝑡 + 𝜉 𝑡 𝑍 𝜉 𝑡 𝐶 𝜉 𝑡 𝐻 0 Metropolitan house prices Interest rates Variables load on factors contemporaneously US variables contemporaneously unrelated to Californian variables US variables contemporaneously unrelated to Californian variables Californian variables load contemporaneously on US series
Identification scheme Shocks Response at horizon 𝑗=1,…,𝐽 MP AS AD AC RS RD Aggregate inflation 𝑍 𝜋 - + * Aggregate output 𝑍 𝑦 Money aggregate 𝑍 𝑚 HY spread 𝑍 𝑠 ++ Defaults 𝑍 𝛿 Californian inflation 𝐶 𝜋 -- Californian output 𝐶 𝑦 Federal funds rate 𝑖 𝑡
Strengths/ weaknesses of approach Pragmatic approach for structural lens across data Regional and metropolitan angles useful Weaknesses Results hang on plausibility of identification scheme Unclear how well the data supports the model Unclear how to think about expectations
Key findings of the paper Regional demand shocks help drive house prices in California Little role for monetary policy shocks Credit shocks appear important… …and not just for house prices, loan quality and employment more generally Regional supply shocks appear relatively unimportant…?
How to improve the paper
Suggested improvements for this paper Are the results robust to mild changes in the identification structure, choice of input data and interest rate specification? Tighter integration with literature, eg Stock and Watson (2008) who find a West-Coast housing market, not much of a regional impact and a significant change in volatility using consents Clarifying what is true for California market vs other markets Some consistency issues across sections where work is ongoing, eg response to Fed funds, interest rate equation etc.
Suggestions for paper with broader audience Another literature (growing) thinking about the relative impact of (tight) urban regulation and geography (eg Glaeser and Gyourko 2017, Saiz 2010, Hsieh and Moretti 2017) A Dynamic Factor model approach with appropriate restrictions could help identify local housing supply shocks (?) 3. Connect land use regulation to economic structure: - Do markets with tight geography load on factors differently? - Does local land regulation exacerbate boom/bust periods? 4. Could quantify local spillovers from tightly regulated markets
Strong case for using approach
Case study: Reserve Bank of New Zealand Macroprudential (and monetary policy) impact of Auckland housing market critical feature of the past five years Tangly to introduce housing markets and macroprudential in a small open economy (eg Neroli Austin 2016) Useful to incorporate for policymakers that want advice from a range of modelling approaches Approach likely to shine some light on Auckland vs rest of New Zealand issues for macroprudential