Endogenous Life-Cycle Housing Investment and Portfolio Allocation

Transcription

1 Endogenous Life-Cycle Housing Investment and Portfolio Allocation Denis Pelletier and Cengiz Tunc Discussion by Florian Exler HFCN Conference, ECB Florian Exler (University of Mannheim) Tunc - Housing and Portfolio Allocation HFCN Conference, ECB 1 / 9

2 Summary Motivation Housing investment might crowd out risky assets Shed light on 1 Low stock market participation 2 Low equity holdings Other studies set household preferences exogenously Contribution Epstein-Zin preferences (EZ): disentangle relative risk aversion (RRA) and elasticity of intertemporal substitution (EIS) Estimate RRA and EIS Florian Exler (University of Mannheim) Tunc - Housing and Portfolio Allocation HFCN Conference, ECB 2 / 9

3 Summary Model Life-cycle model Households consume numeraire and housing goods 3 asset classes 1 Save bond (& mortgages) 2 Risky equity 3 Risky housing 3 types of risk 1 Labor income 2 Equity prices 3 House prices Florian Exler (University of Mannheim) Tunc - Housing and Portfolio Allocation HFCN Conference, ECB 3 / 9

4 Summary Calibration Choose RRA and EIS to match 1 Equity market participation 2 Equity investment shares 3 Homeownership Quantitative Findings Housing investment crowds out investment in risky assets Fit level of equity market participation and homeownership EZ preferences are not too important Florian Exler (University of Mannheim) Tunc - Housing and Portfolio Allocation HFCN Conference, ECB 4 / 9

5 Discussion Important question, powerful model to tackle it Focus / Contribution Crowd-out is established in literature Focus on quantitative contribution (Ñ see later) Emphasize difference to literature, especially Yao and Zhang (2005) Florian Exler (University of Mannheim) Tunc - Housing and Portfolio Allocation HFCN Conference, ECB 5 / 9

6 Calibration 1 Improved fit of low participation rate and low risky investment share ñ Directly targeted 2 EZ does better than CRRA ñ Additional degree of freedom in over-identified SMM ñ Exogenous moments: debt/income ratios etc.? What is calibrated vs. what drives results? ñ Preference parameters vs. frictions & inefficiencies Discipline exogenous processes ñ Price correlations, mortgage interest rate,... Income process controlling for gender, education, marriage? EIS parameter noisy ñ Not too different from inverse of RRA parameter Florian Exler (University of Mannheim) Tunc - Housing and Portfolio Allocation HFCN Conference, ECB 6 / 9

7 Calibration 1 Improved fit of low participation rate and low risky investment share ñ Directly targeted 2 EZ does better than CRRA ñ Additional degree of freedom in over-identified SMM ñ Exogenous moments: debt/income ratios etc.? What is calibrated vs. what drives results? ñ Preference parameters vs. frictions & inefficiencies Discipline exogenous processes ñ Price correlations, mortgage interest rate,... Income process controlling for gender, education, marriage? EIS parameter noisy ñ Not too different from inverse of RRA parameter Florian Exler (University of Mannheim) Tunc - Housing and Portfolio Allocation HFCN Conference, ECB 6 / 9

8 Data Fit Risky Asset Share UNCOND (%) Risky Asset Share COND (%) Participation Rate (%) 1 Homeownership Rate (%) SCF Data 0.4 EZ CRRA Florian Exler (University of Mannheim) Tunc - Housing and Portfolio Allocation HFCN Conference, ECB 7 / 9

9 Model Better fit homeownership Moving shock More concave bequest motive Empty budget sets for low house prices and low income? Cannot stay in house (mortgage rate ą income) Cannot move (mortgage ą house price) Timing: Does household observe current wage when choosing consumption and investment? Florian Exler (University of Mannheim) Tunc - Housing and Portfolio Allocation HFCN Conference, ECB 8 / 9

10 Misc Given the complex asset structure, it would be very interesting to explore the trade-offs of different assets in more detail. Reverse mortgage should increase old-age homeownership rates (and thus works against you). I think higher risk in house prices lets young households buy houses because in some cases prices drop low enough to afford a decently-sized house. I do not believe it is because young households are risk-loving. That would mean the value function is not concave. I don t understand Table 7. Which cost are varied and which rates are reported? Florian Exler (University of Mannheim) Tunc - Housing and Portfolio Allocation HFCN Conference, ECB 9 / 9