Towards
Realistic
Risk & Return Estimates
for Real Estate
Dr. Thies Lindenthal
Photo ©
Notes: * Thank you very much for having me, I am excited to be here with you. * I am looking forward the the Q&A, so let me jump right to the main research question.
Is real estate a good investment?
We should know: It’s the world’s largest asset class.
* How much investment risk and return can we expect? * How does its risk/return profile compare to other investments? * How much more risky are individual assets? Notes: * Most of my research centers around a straight-forward question: Is real estate a good investment? We should know, it's the largest asset class in the world. * Still, property investments are often misunderstood as being fundamentally different from more mainstream asset classes such as stocks or bonds. Simplistic advice such as "get on the housing ladder!" is typical for a view in which real estate is exceptional: returns exceed a fair compensation for risk. * As an applied financial economist, I strive for a realistic estimates of real estate’s risks and returns, both at the market and at the asset level. To do so, I combine theory and data, traditional econometrics and innovative machine learning techniques, dusty archives and big data sensors. * To come back to the main question: Academics, industry players, policymakers and private households— they should be familiar with real estate’s risk–return profile to make better decisions.
Economy-wide risks
Bank of England Financial Stability Report (
2019
)
*
“
The housing market can be a key source of risk to UK financial stability
. In the UK, mortgages are households’ largest financial liability and lenders’ largest loan exposure in aggregate. Housing accounts for nearly half of the total assets of UK households. [...] Historically, the rapid build-up of household debt has been a key source of risk to financial and economic stability — and not just in the UK.”
Photo ©
Notes: * The global financial crisis of 2008/2009 has highlighted the importance of real estate for our economy and financial system once more. * The sheer size of the asset class dominates national accounts and household-level balance sheets alike, both in terms of assets and liabilities. * A large section in the Bank of England’s 2019 Financial Stability Report is devoted to real estate, modelling how the housing market can be a key source of risk to UK financial stability.
Just an alternative?
Real estate is a sizeable component of institutional portfolios
* Sovereign wealth funds, insurance companies, pension funds - “Our” pension scheme USS holds £4.3 billion of private RE, for instance.
Notes: * The risk/return profile of real estate is absolutely relevant to the industry. * Sovereign wealth funds, insurance companies, pensions funds and other institutional investors, they are all invested in real estate. * Our pension scheme, the USS, for instance, owns private property worth 4.3 billion at the moment. * Beside the shrinking pensions, real estate affects the bottom line of most households directly.
Households
...face the same challenge as sophisticated professional investors
3 BR late Victorian terraced house in East Cambridge. Sold in August 2019.
How much return can the new owners expect?
How much risk do they face?
Notes: * Estimating the risk and return of real estate remains a real challenge. It's a challenge because it's very difficult to do and it's real because it affects not only sophisticated big investors, but also normal households. * It's even more difficult at the asset-level. For example, this 3 bedroom late Victorian terraced house in East Cambridge was sold in 2019. If you think about this particular building, how much return can the owners expect over the coming years and how much risk do they face? * When we bought our house, my family assumed that I would conduct a very elaborate financial analysis. It's my job, I should be able to do it. Let me confess, I did not. I made a few bold assumptions and then we made an offer.
## Academia to the rescue?
Broad literature, surprisingly few answers. Data are scarce.
* **A risk premium puzzle?** - “Arguably the most surprising result of our study is that long-run returns on housing and equity look remarkably similar.
Yet while returns are comparable, residential real estate is less volatile on a national level, opening up new and interesting risk premium puzzles.
”
*The Rate of Return on Everything, 1870–2015* (
Jordà et al., QJE 2019
) Notes: * And I am not the only one struggling to come up with helpful answers. * As a discipline, we lack good quality data. * Without reliable data, the academic literature remains stuck in "puzzles". A good example is a recently published QJE paper that compared the risk-return profiles of housing to stocks. * They, again, repeat this widely-held misconception that real estate is somehow magical: Ig offers the same returns as stocks but at a lower risk. * That cannot be true—and it isn't.
## Better data!
“The Total Return and Risk to Residential Real Estate”
(
Eichholtz, Korevaar, Lindenthal, Tallec, RFS 2021
)
* Asset-level total return: Rents, costs, taxes, prices for residential RE in Paris (1809–1943) and Amsterdam (1900–1979)
Notes: * In a recent paper, Piet Eichholtz, Matthijs Korevaar, Ronan Tallec and I show that this puzzle completely disappears with better data. * We collected 170K rent and price observations for Paris and Amsterdam and were able to calculate total returns at the property level. This is the first study that offers asset-level total return data for more than just a few years. * Two things stand out from this paper: First, risk is a lot higher than previously thought. That means that the risk-adjusted returns of real estate are no positive outliers but comparable to stock returns. * Second, almost all of the return comes from rents. Rents are really making or breaking property returns in the long run.
## "Normal" Sharpe Ratios
Asset-level total returns (
Eichholtz, Korevaar, Lindenthal, Tallec
)
* Paris (1809–1943) and Amsterdam (1900–1979)
## Commercial Real Estate
Total returns dominated by income, not capital gains.
* City of London, 1920–2010
Source: Scott (1996); MSCI/IPD
Notes: * And this not just a residential real estate story. * For commercial space we see a similar pattern. This figure gives you the returns for CRE right in the City of London. The green line shows that very limited long-run capital gains have been achieved since WWII—in real terms. But the total return was solid due to the income delivered by these properties. * So, rents are crucial. But our understanding of rents is even worse than our understanding of prices. Giacoletti, Sagi etc. they all work on capital returns but hardly anybody investigates rents. * That's why I think that a new working paper that I co-authored is so exciting.
Realistic Rent Expectations
“Growth and Predictability of Urban Housing Rents”
* New working paper with Piet Eichholtz and Matthijs Korevaar - Currently under review at JEEA * **Core question: How risky are rental income streams?**
Reliable Micro-Data
Rent records of institutional investors, from 1500–2020
*
Amsterdam, Antwerp, Bruges, Brussels, Ghent, London, and Paris
*
436,000 rental cash flow observations
*
Repeat observation indices: Same methodology in all cities, quality-adjusted.
Half a millennium
500 years of Western European economic history
* Rent indices (nominal) for 7 cities
Notes: * and here the are: Rent indices for 7 cities covering more than 500 years. * Such a long time series on urban housing markets might be a world record. I am not sure.
Real Rents
Inflation is a relatively modern phenomenon
* Rents deflated with a general consumer price index
Notes: * This is the same indices, but deflated with a general consumer price index. * It turns out that real rent indices are great sensors of the cities' economies, they track the ups and downs of our cities very closely.
Rents are a sensor
... of a city's fortune
* Disasters: Sack of Antwerp ("Spanish Fury"), 1576
Notes: * Let's start with a drastic example. * It is reassuring to see that a massive disasters like the sack of Antwerp in 1576 are clearly identifiable in our indices.
War & regulation
Another disaster: Siege of Paris, 1590
* First example of rent control in our sample.
Notes: * I won't go through 500 years of European history, but let me give one more example: Rents collapsed during the siege of Paris in 1590. This was, of course, directly caused by Henry the fourth bombarding the city from the outside. * But in addition, rents were cut during the first instance of rent regulation that we observe in our sample. This rent regulation regime was short-lived and we basically observe free markets up to WWI. * Having several centuries unregulated rent data is unique and can really help to come up with counterfactuals for more regulated housing markets.
Shared trends
Centuries of stability, followed by fundamental change
* Rents doubled during the industrial revolution
Notes:
Long-term rents: growth
Annual growth, 500-year average: 0.12–0.30%.
* Modest growth ($\mu$), despite rapid urbanisation
Notes:
Long-term rents: risk
Modest growth ($\mu$), substantial risk ($\sigma$)
* $\sigma$ of 8–13% (total returns are roughly 4–5%)
Notes:
Superstar bias?
In 1500, the Belgian cities were the stars. Not London or Paris.
* Don't focus on the successful survivors, only
Notes:
Market-wide predictability
Can we predict rents with economic fundamentals?
* Population growth is linked to economy and housing demand.
* `$ \tiny \Delta_{25} r_{it} = \mu_i + \beta_1 \Delta^+_{25} pop_{i,t-1} +\beta_2 \Delta^-_{25} pop_{i,t-1} + \gamma_1 \Delta^+_{25} pop_{i,t} + \gamma_2 \Delta^-_{25} pop_{i,t} + \gamma_3 \Delta_{25} w_{it} + \varepsilon_{i,t} $` Notes: * Next we test for predictability at the market level and try to predict rents with economic fundamentals. * We tried wages, prices,
Market-wide predictability
Past population changes do not explain rents well
* Coefficients are not significant. Model has hardly any predictive power.
* `$ \tiny \Delta_{25} r_{it} = \mu_i + \beta_1 \Delta^+_{25} pop_{i,t-1} +\beta_2 \Delta^-_{25} pop_{i,t-1} + \gamma_1 \Delta^+_{25} pop_{i,t} + \gamma_2 \Delta^-_{25} pop_{i,t} + \gamma_3 \Delta_{25} w_{it} + \varepsilon_{i,t} $` Notes:
Market-wide predictability
Current economic fundamentals determine rents
* Rents are strongly linked to population changes and wages
* `$ \tiny \Delta_{25} r_{it} = \mu_i + \beta_1 \Delta^+_{25} pop_{i,t-1} +\beta_2 \Delta^-_{25} pop_{i,t-1} + \gamma_1 \Delta^+_{25} pop_{i,t} + \gamma_2 \Delta^-_{25} pop_{i,t} + \gamma_3 \Delta_{25} w_{it} + \varepsilon_{i,t} $` Notes:
Any economic relevance?
If there is predictability, can market participants act upon it?
* Yes! - Term-structure of rents in Paris: contracts are typically for 3, 6 or 9 years - We compare the agreed annual rents for short-term contracts to longer-term leases - Without predictability, we should observe no differences (beyond a general premium/discount for longer durations) * Tenants and landlords successfully predict future rents and negotiate rental contracts accordingly.
Asset-Level Risk
Tenants rent single units, not an index.
* Landlords cannot fully diversify. How much idiosyncratic income risk do they face? * Excess growth rate $\Delta_e$ is the difference between the rental growth rate at the asset- and at the market level. * $\Delta_e$ is large: 0.093 in absolute terms.
Notes:
Asset-Level Risk
Modelling the deviations from the city-wide trends
* `$ \tiny abs(\Delta_e R_{i, t_0, t_{\textit{future}}}) = \alpha + \beta_1 abs(\Delta_e R_{i, t_{\textit{past}, t_0}}) + \beta_2 \textit{MAE}_{i, -3y} + \beta_3 abs(\Delta_e R_{i, t_{\textit{past}, t_0}})\times\textit{MAE}_{i, -3y}+\epsilon_{i,t_0} $` * We try to predict growth in excess of the market
Notes:
Asset-Level Risk
Modelling the deviations from the city-wide trends
* `$ \tiny abs(\Delta_e R_{i, t_0, t_{\textit{future}}}) = \alpha + \beta_1 abs(\Delta_e R_{i, t_{\textit{past}, t_0}}) + \beta_2 \textit{MAE}_{i, -3y} + \beta_3 abs(\Delta_e R_{i, t_{\textit{past}, t_0}})\times\textit{MAE}_{i, -3y}+\epsilon_{i,t_0} $` * explanatory variables: past excess growth + market-wide uncertainty
Notes:
Risky assets and risky times
Persistence in excess growth: Long-lived out-/underperformance
* Only a third (0.317) of previous deviations get corrected at rent revision.
Notes:
Risky assets and risky times
Persistence in market uncertainty
* Large deviations at *other* homes increase excess future rent growth (0.694)
Notes:
Risky assets and risky times
In volatile years, all homes experience idiosyncratic risk.
* Low-risk properties are safe in good times only.
Notes:
Summing up
Rents are key when evaluating real estate investments
* Long-term rent indices are a powerful indicator of the success of cities * Real rental growth rates have been lower than previously assumed - Asset pricing implications: e.g., Gordon growth model * Volatility in rents is high: Lower Sharpe ratios for RE * Predictability both at the market and the asset level can reduce risk. Notes:
Next steps
Interdisciplinary work in three main directions
1. More data, especially on commercial real estate and affordability 2. Richer data: 3D models (Lindenthal, REE 2020), remote sensors (Lindenthal, 2017), street-level images (Lindenthal & Johnson, JREFE 2021), dynamic preferences (Lindenthal, Schmidt & Wan, hopefully soon) 3. New empirical approaches to analyse these data: applied ML, spatial ANN (Clapp & Lindenthal, JHE R&R) Notes:
Q & A
Surely, there are questions?
Photo ©
Notes: