a No Price Like Home Global House Prices

a) No Price Like Home: Global House Prices, 1870 -2012 (by Katharina Knoll, Moritz Schularick and Thomas Steger) b) New results on households subjective probability of future house prices (by Olympia Bover) c) Social security wealth in Italy: 20 years of pension reforms (by Michele Belloni, Agar Brugiavini, Elena Buia and Giacomo Pasini) Discussion by Francesco Zollino Bank of Italy – Economic outlook and monetary policy studies The Bank of Italy’s Analysis of Household Finances Fifty Years of The Survey on Household Income and Wealth and the Financial Accounts Bank of Italy, Rome 3 – 4 December 2015

No Price Like Home: Global House Prices, 1870 -2012 (by Katharina Knoll, Moritz Schularick and Thomas Steger) The paper develops new historical times series for 14 industrial countries, covering for years since 1870 not only real house prices but also a variety of complementary indicators such as real land prices and construction costs. It is extremely well documented with reference to both the very rich data source they adopt and the methods by which they compute the targets series and check their reliability As houses are heterogeneous assets and available source data vary a lot across countries and time by area covered, type of property, method used to produce indices, the authors seek for preserving three guidelines: constant quality, longitudinal consistency and historical plausibility. 2

Among the main results: i) Global real house prices almost tripled since 1960 after remaining virtually unchanged between late XIX and mid XX century; about 80 per cent of the acceleration is due to land revaluation (with a minor role of construction costs) ii) Although with a different intensity, the trend is common to the 14 industrial countries, marking a break with respect to years 1870 -1950 (when land prices almost stagnated despite the steady increase in population and income) iii) In advanced countries land revaluation fuelled the increase in the wealth-to-income ratio documented by Piketty (2014), possibly more strongly than the capital accumulation itself. 3

Beyond the novel historical data set, the paper valuably casts in a long run perspective the analysis of the role of two different determinants of house prices: a) The cost of land; b)The cost of structures This is a core distinction that helps better understanding the macroeconomics of the housing market on the viewpoint of both structural, cyclical and financial stability analysis Candidate fundamental drivers can affect to different extents and directions the trends in reproducible (structures) and fixed (land) assets’ prices. Volatility of the two components may also be different thus affecting our ability to detect vulnerabilities to observed trends in house prices: similar evolution of “fundamental” may convey different signals depending on the importance of land versus structures on total house prices 4

Comment 1: estimating the land share on house prices is a difficult challenge as it may vary a lot over time, across locations and dwelling types The authors assume that for all countries the land shares is constantly at 50% (or with some local variations)… but, as documented in the paper, shares are significantly different across countries, generally increasing over time, and to different extents This could bias the estimate of such a key parameter: better going along a country specific approach? (In Italy it should stay now at 34%). 5

Comment 2: Composition effects seem to significantly affect also price estimates (e. g. size/population effect) Real House prices OECD Focusing on the pronounced acceleration since late 1960 s, global house revaluation proves largely less pronounced when heterogeneous country developments are controlled by weighting according to population (as shown also in the paper) and, even more so, if a large (and fixed) basket of countries is considered (OECD aggregate) 6

Comment 3 a: Which are the drivers of land revaluation in more modern times? . . . a) Population change b) Land use regulations (cfr trends in Boston in Glaeser-Ward, 2009) 7

Comment 3 b: …and is the increasing role of land prices uniform within and between countries (relevant for equity issues)? On the demand side: do people demand “physical” land or dwellings located in area well equipped with high quality services (school, health care), local transport system, amenities, and clean air, parks…? Accordingly, the increasing shares of land may signal tensions in the equal access to better life conditions (in addition to the wealth inequality addressed in the paper in the vein of Piketty), on which may weigh the design of policies (by the central and local governments) On the supply side: Is the role of land lower if the supply price elasticity of houses (or structures) is higher? And with the efficient allocation of new houses? In this respect it would be interesting to consider historical developments in housing stocks and the possible role of policies (taxation, land use restrictions or, more recently, energy efficiency). 8

New results on households subjective probability of future house prices (by Olympia Bover) The paper exploits (currently) rare information collected by the 2011 Spanish Survey of Household Finances about the individual expectations on house price changes in one year horizon. The topic of house price expectations is becoming increasingly important in order to i) explain consumer behaviour; ii) explore sustainable path of house prices; iii) monitor the quality of information underlying the key decision to invest on housing The paper provides plenty of interesting results, often validated by several robustness checks, either by simply adding/dropping covariates or by switching across completely different approaches (linear regression, quantile regression on subjective probability, treatment effects) 9

A strict selection of results: a) conditioning on a variety of controls, the self-assessed current house price does not significantly affect the formation of expectation for future trends by households; b) households that recently bought their houses show a positively biased subjective probability of future prices c) female are more optimistic than males on house price trends d) pessimism on house price trends play a significant role on the purchases of both houses and other important consumption items 10

Comment 1: which is the quality/extent of control of information on the housing market by (Spanish) households? Nominal house prices Indices 2000=100 Spain – Regional Map 11

a) How much the more optimistic found for women is affected by purely psychological factors as the group proves: i) inconsistently more pessimistic regarding the assessment of the financial assets; ii) less accurate to predict observed trends of house prices; a) iii) likely affectively linked to their homes somehow in line with aged respondents (who prove more reluctant to expect a price drop) b) 12

b) In times of generalized and prolonged steep decline observed in house prices in Spain i) 1/3 of respondents expect with no-uncertainty a further drop in prices but in general less uncertain predictions comes with lower expected decline (loss of accuracy in expectations) ii) based on average probability distribution, a price drop is virtually as likely as the event of price stabilization and/or increase. 13

Comment 2: Sparse survey information available in the US and UK support the idea that expectations of future appreciation are linked with past observed appreciation (Muellbauer, 2012; Case at al, 2012) Almost all approaches (either structural or rent-arbitrage based) in modelling house prices need to compute the user cost of dwellings with an expectation component uc = hp(r + δ + − Δhp*/ hp) and the common empirical practice is that house price expectations are adaptively formed by a multiple lagged rate of changes (generally four-five year moving averages; Himmelberg, et al. 2005) 14

In this respect, evidence that for Spanish households adaptive expectations do not to hold (as the current self-assessed price fails to enter significantly in the estimated model) may be due to non-linearities. For example, a possibility is that only if current house prices exceed the threshold of the perceived transaction and finance costs they affect expectations as the potential speculative gains would become very attractive. Investigating this point would shed light on the phycology of a house price bubble Does the result hold for both owner and tenants? In the empirical literature tenants may form different expectations (demand prices) than the owners (supply prices) and this may also affect the relation between the assessments of current versus future prices. Do the self-assessed current prices fit with the observed prices? (see ensuring SHIW based evidence for Italy in Cannari-Faiella, 2009) 15

Comment 3: Expecting with certainty a severe drop in house prices proves an important driver of consumption decision: confidence o wealth effects? In the estimates reported in Tables 10 -11 it could be interesting to consider not only the certain price drop, but also an intermediate level of “certainty”: when in the answer to the questionnaire, points are fully allocated to adjacent intervals of expected price changes. In this way it could be possible to test for symmetric effects of deep pessimism vs gradually more positive beliefs This could also help to understand if the consumption effects of the “certain” huge drop in house price found in the paper is largely related to a more general loss of confidence about macroeconomic trends rather than to pure developments of house price (and wealth). 16

Social security wealth in Italy: 20 years of pension reforms by Michele Belloni, Agar Brugiavini, Elena Buia and Giacomo Pasini The paper is very ambitious as it aims at eliciting the impact on Italian households’ optimal retirement decisions coming from the multiple reforms of the public social security system that (so far) have been implemented since 1992 For this purpose the authors manage to exploit at the best value pieces of information coming from two different surveys and administrative data, which are used complementarily to trace the evolution of SSW of different cohorts (divide by gender) of Italian workers under the sequence of pension reforms. As a further challenging scope, the authors provide first estimates of the effects of reforms on the financial incentives of Italian households on 17 both the timing and exit route, by cohort and by gender.

The paper has the great merit to refresh interest on the possible economic effects of recent pension reforms beyond their direct impact on fiscal balances, that have been more stressed in recent debate due to urgencies of the crisis In this respect it may re-open a large project for future research, in which also the macroeconomic effects of pension wealth on savings may be reassesed in the new institutional framework. A variety of interesting results have been already found, although still in a preliminary version also due to the complexity of the required computations. A major missing documentation in the current version regards the data and methods adopted to estimate the weight of three ways to retirement considered in the option value model. 18

(Some) main results: a) Based on individual SSW profiles under all possible retirement ages (differently computed either by SHARELIFE or SHIW): i) for old age benefits, incentives to postpone retirement regards only women but not men; ii) for early retirement, for both it becomes convenient to postpone the age (with maximum SSW at age 59 and 60, respectively); b) Including disability treatment as a further exit route, estimates based on an a dynamic programming model show that for men and women born in 1945 -49, the financial gains to postpone retirement are larger for old age than for early or disability benefits; c) For men, the three options become indifferent at around 63; for women, old and disability converge 5 year earlier, while early retirement much later (around 68) 19

Comment 1: The paper focuses on public pensions (first pillar). Should supplementary schemes (and the related regulatory changes) be also included in the computation of pension wealth? And how this could affect the estimated retirement profiles? SHIW – Expected coverage rate at retirement (in black: only from 1^ pillar, in red including supplementary pensions; mean and min-max values by age) 20

Comment 2: Which is the role of uncertainty about expected benefits under the different pension schemes? In Italy the sequence of reforms and the still open debate on the final setting of the social security system may indent confidence of households on their ability to project the streams of expected benefits, to variable extents across different groups of households and exit routes. This could affect the actual responsiveness of the households to regulatory changes as well as their ability to identify the changing incentives about the time and the way of retirement. Accordingly, the option value of postponing retirement imputed by the authors could be somewaht biased …. more so if the SSW profiles include some anticipated effects of reforms according to the household assessments 21

For example: the radical 1992 reform (and the long debate that went before) apparently caused a downward revision of expected pension benefits already couple of year in advance (based on elaborations from SHIW) Wealth of Italian households (percentages of disposable income) 22

Two minor comments/curiosities regarding the estimation of option value model a) SSW is imputed gross of contributions: should it be adjusted for the (actual and expected) streams of contributions under the different schemes to more soundly track the incentive to retire? b) The disability treatment is considered as an option that may be discretionarily chosen by households as an exit route, very much alike the old and early retirement benefits. But, the provision of the benefit is actually conditional on the scrutiny of health eligibilility criteria, so it is hard to think of a financial incentive attached on disability status (at least under credible legal codes) 23

thank you for listening 24