(Article originally published Summer 2018)
Part one: What do recent rises in interest rates mean for property prices?
US 10-year Treasury Bill rates have been at an all-time low, having fallen below 2% for the first time in 2017.
UK government bond (10-year) yields have also been lower than they have been at any time and 10-year Treasuries in Australia are trading at all time low yields. Short-term interest rates have also been at all-time lows. We habitually (and for good reason – see below) connect low interest rates and bond yields to low property yields and high prices.
However, recent rises in US treasuries have got us all wondering. Is this the start of a downturn for property values? Let’s get one thing straight – I don’t know, and nor does anyone else. That’s because I do not know where bond yields and interest rates are going from here, But if I did, I would be able to hazard more than a guess. And if, when you read this, they have already moved, there may still be time to use the information I am going to share to protect yourself.
Real risk free rates, proxied by the yield on government-issued inflation indexed bonds, are also trading at very low or negative rates. US Treasury Inflation Protected Securities have been issued offering negative yields since 2010 and UK index linked gilts have been priced at negative yields since 2011. Australian Treasury Indexed Bonds have never offered negative yields, but are now being issued at all-time lows of around 0.5%.
There is much debate over how long such low yields will continue, and the implications for real estate prices. Real estate capitalisation (cap) rates (the inverse of price-earnings ratios) are low in many markets, albeit not – yet – at all time lows.
How strongly have real estate capitalisation rates been connected or correlated with conventional bond yields, short term interest rates and indexed bond yields? What will happen to real estate prices if bond yields revert back to more ‘normal’ levels over the next few years? And what will happen to prices if they do not?
In order to gain an insight into these questions, we need to go back to re-explore the theory supporting the determination if the yield on bonds. A good starting point is the work of Irving Fisher in the 1920s.
The Fisher equation
The Fisher equation (Fisher, 1930) considers the components of total return delivered by an investment. It states that:
R = l + i + RP (1)
Where:
R is the total required return
L is a reward for liquidity preference (deferred consumption)
i is expected inflation
RP is the risk premium
Index-linked government bonds are widely considered to be risk free and, because the coupon is enhanced by the inflation rate in the previous period, investors do not need to earn an inflation reward in that coupon or running yield. Hence ‘l‘ is given by the coupon on index-linked government bonds (for the purpose of building an example, let us assume 0.5%).
‘l + i‘ is the required return on conventional government bonds (for simplicity, ignoring an inflation risk premium, let us assume 2.5%). These may be regarded, respectively, as the real and the nominal risk free rates (RFR and RFN). If we assume that there is no inflation risk premium in the pricing of the bond RFN= RFR + i, and RFN – RFR = i, so i appears to be 2.5% – 0.5% = 2%.
inflation will lead to lower than expected returns; and second, there is a general discounting of investments where they are risky. In a market dominated by investors with real liabilities, risky (in real terms) conventional gilts would be discounted, meaning the required return would be higher. If required returns equal the available return in an efficient market, then the 2.5% available on the conventional gilt must include a risk premium. Following Fisher again, the full explanation of a required return is:
R = l + i + RP (1)
If an inflation risk premium of 0.5% in the pricing of the conventional gilt is assumed, the rate of expected inflation implied by a comparison of index-linked and conventional bond yields is 1.5%.
For assets other than government bonds, further factors contribute to the risk premium. For example, investors in corporate bonds might require a higher premium to reflect the greater risk of default associated with such bonds. Meanwhile, cash flows and prospective sale prices from equity and real estate investments are subject to volatility, which will push the required risk premium even higher.
For illustration, we will assume that an additional risk premium for prime real estate of 3.5% is required. (Note that this includes the inflation risk premium of 0.5% in the pricing of the conventional gilt). The Fisher equation can then be re-written as:
R = RFN + RP (2)
R is 2.5% + 3.5% = 6% in this case.
However, for an investor interested in real returns (say a defined benefit pension fund) conventional gilts are less attractive than index linked gilts. There is a risk of inflation expectations not being realised, so that higher than expected inflation will lead to lower than expected returns; and second, there is a general discounting of investments where they are risky.
In a market dominated by investors with real liabilities, risky (in real terms) conventional gilts would be discounted, meaning the required return would be higher. If required returns equal the available return in an efficient market, then the 2.5% available on the conventional gilt must include a risk premium. Following Fisher again, the full explanation of a required return is:
R = l + i + RP (1)
If an inflation risk premium of 0.5% in the pricing of the conventional gilt is assumed, the rate of expected inflation implied by a comparison of index-linked and conventional bond yields is 1.5%.
For assets other than government bonds, further factors contribute to the risk premium. For example, investors in corporate bonds might require a higher premium to reflect the greater risk of default associated with such bonds. Meanwhile, cash flows and prospective sale prices from equity and real estate investments are subject to volatility, which will push the required risk premium even higher.
For illustration, we will assume that an additional risk premium for prime real estate of 3.5% is required. (Note that this includes the inflation risk premium of 0.5% in the pricing of the conventional gilt). The Fisher equation can then be re-written as:
R = RFN + RP (2)
R is 2.5% + 3.5% = 6% in this case.
Gordon’s growth model
Investors are clearly prepared to accept a lower initial return from an investment in cases where the cash flow and value of an investment are expected to grow over time.
Expected income growth became embedded in the pricing behaviour of equity and property investors by the late 1950s in the US and UK. It became necessary to extend the simple cash flow model and this was first achieved by introducing a constant rate of growth in nominal income (GN). Following Gordon, 1960,
K = R – GN (3)
where K is the initial rate of return or capitalisation rate
(income / price).
GN can be driven by inflation (i) or real growth (GR) so that GN = I + GR.
Finally, this analysis can be extended by introducing a constant rate of depreciation (D. This gives (approximately):
K = R – GN + D (4)
which, by reference to equation 1, can be expanded to:
K = RFRN + i + Rp – GN + D (5)
If growth of 1.5% was anticipated, in line with expected inflation, and depreciation was 0.5% per annum, this would suggest that an initial yield (K) of 6% – 1.5% + 0.5% = 5% for prime real estate should be observed. The gap between this yield and the return on conventional government bonds would then be 5% – 2.5% = 2.5% with the size of this gap driven by the property risk premium minus growth expectations (3.5% minus 1%).
This is a simple framework which assumes that property behaves as a pure equity investment with annually reviewed rents. The real world is more complicated, not least since rents might be fixed for periods of time, indexed or stepped. Nonetheless, this framework can be used to give insights into how property cap rates should behave through time.
Empirical evidence
Given that property cap rates or yields are determined by a number of factors, it is not clear that any rise in bond yields would translate into a rise in property yields. Whether property yields rise depends on why bond yields are increasing. Conventional government bond yields might rise because of a rise in expected inflation. In this case, index-linked yields would stay unchanged and property yields might not change either depending on how nominal cash flow growth responds to inflation.
For real estate pricing, a presumed correlation between rents and inflation means that there should be a stronger link between real estate yields and index-linked bond yields than between real estate yields and conventional government bond yields. Is there a correlation between rents and inflation? How strong in practice is the link between real estate yields and index-linked bond yields – and between real estate yields and conventional government bond yields? We will examine this evidence in our next contribution.