A wider ‘new normal’ property to bond yield spread? - Capital Economics
UK Commercial Property

A wider ‘new normal’ property to bond yield spread?

UK Commercial Property Focus
Written by Amy Wood
Cancel X

We think that the all-property to bond yield spread will narrow from its current elevated levels, although this is not likely to happen before late 2020. But, structural changes mean that the property to bond yield spread won’t narrow to the early-1990s to mid-2000s average of 150bps. Nevertheless, there appears to be room for a 100bps or so increase in bond yields to be absorbed by the spread without a major rise in property yields.

  • We think that the all-property to bond yield spread will narrow from its current elevated levels, although this is not likely to happen before late 2020. But, structural changes mean that the property to bond yield spread won’t narrow to the early-1990s to mid-2000s average of 150bps. Nevertheless, there appears to be room for a 100bps or so increase in bond yields to be absorbed by the spread without a major rise in property yields.
  • The spread of property initial yields to the risk-free rate, as proxied by 10-year gilt yields, is a useful measure to assess how under or overvalued property is. However, the property to bond yield spread is not fixed over time as it is buffeted by cyclical and structural forces. Indeed, in the 1980s, the yield spread was negative, as high interest rates were needed to keep inflation at bay, pushing up bond yields to a greater extent than property yields. During the early-1990s to mid-2000s, inflation and interest rates declined to low levels and the yield spread turned positive, averaging 150bps.
  • But, the onset of the financial crisis saw the yield spread widen to over 400bps. Since then it has remained elevated at between 300bps and 400bps. This has reflected that events, such as the sovereign debt crisis and the Brexit vote, have reduced interest rates and bond yields, while increasing the risk premium associated with holding property and lowering expected rental value growth.
  • In our view, the next material move in the spread will be a narrowing. However, in the UK this is not likely until at late 2020, when economic growth is expected to pick up sufficiently causing the cyclical factors that have pushed up the spread to wane. Indeed, stronger economic growth would be consistent with an improvement in rental value growth and thus rental expectations. In addition, stronger growth will entice the Bank of England to raise interest rates.
  • Nevertheless, there are reasons to think that the spread will not narrow to its early-1990s to mid-2000s level. This reflects various structural factors which we think have increased the risk premium associated with holding property and therefore will keep the yield spread permanently wider. Firstly, while lease lengths have not reduced much in the post-crisis period, the use of break clauses has become more widespread, reducing the income security of landlords. Secondly, tighter regulation and reduced willingness to lend on commercial property in the post-crisis period appears to have structurally increased the illiquidity premium.
  • Our assessment of these cyclical and structural factors suggests that the ‘new normal’ level of the property to bond yield spread could be around 70bps higher than the early-1990s to mid-2000s average level of 150bps.
  • Admittedly, given the uncertainties in assessing the extent to which factors, such as the property risk premium, have risen, it is difficult to be precise about what level the yield spread will eventually return to. But, we take comfort in the fact that, even if property yields are unchanged, there appears to be room to absorb around a 100bps increase in bond yields without a major rise in property yields.
  • In conclusion, although the exact level the property to bond yield spread will return to in ‘normal’ times is uncertain, it is likely to be narrower than the 340bps it is currently. But, there are also plausible reasons to suggest that it will be wider than the early-1990s to mid-2000s average level of 150bps.

A wider ‘new normal’ property to bond yield spread?

The spread between property initial yields and government bond yields, which is used as a proxy for the risk-free rate, is a measure which is often used to analyse how under or overvalued property is.

However, the yield spread is not fixed over time as it is buffeted by cyclical and structural factors. In fact, its current level, around 340bps, is elevated relative to history. This Focus looks at why the yield spread has widened and where we might expect it to go in the longer term.

Calculating the property to bond yield spread

To start, we briefly cover some issues that arise when calculating the yield spread. First, there are questions as to which government bond term to use. The term should reflect the lease length of the property being valued, which for the all-property average is currently around 7 years. However, for simplicity it is more common to use a 10-year gilt yield as a proxy. In any case, given that the yield curve is near its post-crisis low, the choice of term would actually make little difference to the analysis.

Second, there is an argument that government bond yields have been distorted in the post-crisis period by the unprecedented monetary stimulus provided by quantitative easing. But, since 2018 at least, any distortion appears to have diminished. Indeed, the shadow rate, which uses the yield curve to measure the (nominal) stance of monetary policy at the lower bound, has moved back in line with Bank Rate. (See Chart 1.)

Chart 1: Bank Rate and Shadow Short Rate (%)

Sources: Refinitiv, Krippner ( 2011 )

Third, there are reasons to believe that, over time, property initial yields and gilt yields have been distorted by inflation and therefore perhaps a better way to calculate the yield spread would be to consider these variables in real terms.

One approach which can be used to account for the effects of inflation is to use the equivalent yield to proxy the real initial yield as it takes into account the changing level of market rents in the first five years of the lease.

That said, some argue that because property income is essentially real, initial yields don’t need to be adjusted at all. However, at the all-property level there is little evidence that property incomes are an inflation hedge, at least for unexpected inflation, as rental values have fallen over time in real terms.

Nevertheless, as Chart 2 shows, although the level of the spread is different, the inflation adjustment would only have a material effect on the spread when inflation is very high or negative, like in the 1980s and between 2015 and 2017 respectively.

Chart 2: All-Property Yields Less 10-Year Government Bond Yield (Bps)

Sources: MSCI, Refinitiv

As such, because we do not expect inflation to move much from the 2% target over the forecast horizon, for the rest of this analysis we use the difference between the nominal initial yield and the 10-year gilt yield to calculate the spread.

The property to bond yield spread over history

Chart 3 shows that there have been three distinct periods in which the nominal yield spread has held broadly stable. The first is the 1980s, where high interest rates were needed to keep inflation at bay, pushing up bond yields to a greater extent than property yields. Nevertheless, as shown in Chart 2, this period was unusual compared to the more recent decades.

Chart 3: All-Property Initial Yield Less 10-Year Government Bond Yield (Bps)

Sources: MSCI, Refinitiv

The second period encompasses the early-1990s to mid-2000s, the so-called Great Moderation. During this time, inflation and interest rates declined to low levels and the yield spread turned positive, averaging around 150bps. Given the stability of inflation over this period, this level appeared a reasonable approximation of fair value.

The third period began at the onset of the financial crisis where the yield spread widened to over 400bps. Since then, the spread has fluctuated between 300bps and 400bps.

This was in contrast to what we expected. Indeed, in 2010, we estimated in a Focus that the yield spread would remain elevated for a few years before narrowing towards the 150bps level as economic growth improved.

But, what we were unable to factor in was the impacts of the sovereign debt crisis and the Brexit vote, which pushed down 10-year gilt yields to historical lows. Property yields also fell, but not to the same extent. In other words, property yields did not fully adjust to the low interest rate environment. So do we still expect the spread to narrow and, if so, to what level?

Why has the yield spread widened?

To estimate what level the yield spread could go to, it is important to distinguish whether the factors that have widened the property to bond yield spread in the post-crisis period are cyclical and therefore will wane over the cycle or are structural and therefore will change its ‘normal’ level.

As a starting point, we use a rearranged version of the simple investment pricing equation:

Required initial yield – risk free rate =

expected rental value growth

+

transaction costs

+

depreciation

+

risk premium

According to the equation, the factors that could be holding the property to bond yield spread above its ‘normal’ level are:

  • Lower expected rental value growth
  • Higher transaction costs and/or depreciation
  • A higher risk premium

We will now step through how each factor has behaved in the post-crisis period in more detail.

Expected rental value growth has fallen recently…

It is simplest to start with expected rental value growth, where estimates can be attained from the IPF Consensus Survey. Notably, rental value growth expectations have held up for much of the post-crisis period. (See Chart 4.) But from around 2016, the average five-year expectation has fallen to 0.6%, well-below the pre-crisis average of 2.5%. In turn, lower expected rental value growth appears to have been one reason why the yield spread has been elevated over the past few years.

Chart 4: All-Property Expected Rental Value Growth (%y/y, Average, Next Five Years)

Source: IPF

…but is expected to improve from late 2020

Given the downside risks to UK economic growth this year and concerns about the rental value growth prospects in the retail sector, it is possible that expectations of rental value growth could deteriorate further, keeping the yield spread elevated in the near term. Indeed, the average IPF consensus expectation for rental value growth over the next two years has fallen to minus 0.1%.

But, from late 2020, we expect economic growth to pick up. Stronger growth would be consistent with an improvement in rental value growth and thus rental expectations. Our longer-term GDP growth forecast implies that expected rental value growth could increase from 0.6% to around 1.5%. (See Chart 5.)

Chart 5: Expected Rental Value and GDP Growth

(% y/y)

Sources: IPF, Refinitiv

Nevertheless, there has been a step shift in the relationship between expected rental value growth and the yield spread in the post-crisis period. (See Chart 6.) Admittedly, this could just reflect a structural break in the series. But, we think that this provides an indication that there are other factors that have also widened the spread.

Chart 6: Yield Spread and All-Property Expected Rental Value Growth

Sources: MSCI, Refinitiv, IPF

Little post-crisis change in costs and depreciation

The second group of factors that might have widened the yield spread include higher transaction costs (such as taxes and legal and agency fees) and greater depreciation. In the past, these have explained a reasonable proportion of the yield spread. Indeed, in 2008 we estimated in a Focus a total allowance for these factors of 1.7%.

Since the financial crisis, there has only been a marginal change in the key transactional tax, stamp duty. Indeed, between 1997 and 2007, the top rate of stamp duty increased from 2% to 4%, while in the post-crisis period it has only risen by a further 1%. In turn, changes to transaction taxes are unlikely to have had a material impact on the yield spread. Moreover, the impact of higher competition on other transaction costs have probably provided an offset.

Similarly, any change in depreciation – the fact that property is a physical asset that declines in quality over time without maintenance or improvement – has probably been marginal. Indeed, the main factor affecting depreciation currently is the move to greener building. For example, research by BNP Paribas highlighted that around 16% of buildings in the City of London still don’t meet the Minimum Energy Efficiency Standards introduced last year.

But the impact of climate change on commercial property is difficult to measure. In turn, there is little evidence that it has increased depreciation rates to a greater extent than technology, which had been the main driver of higher depreciation in the past.

In turn, we see little reason to change our 2008 view that the total allowance for these factors should be around 1.7%. Nevertheless, in the longer term there is a risk that changes to climate change regulations could cause the depreciation rate to increase.

The implied risk premium has risen post-crisis

The final factor that may have contributed to the wider yield spread is the additional risk associated with holding property over bonds, the risk premium. This can be calculated as a residual to the simple investment equation.

Table 1 shows that, given our assumptions for the other components, the implied level of the risk premium is around 2.3%. This is lower than our 2010 estimate of 3.4%, but above our 2005 estimate of the long-term risk premium of 1.6%.

Table 1: The Implied Risk Premium (%)

2010

2019

Property to bond yield spread

3.5

3.4

+ expected rental value growth

1.6

0.6

– transaction costs and depreciation

1.7

1.7

= implied risk premium

3.4

2.3

Sources: MSCI, Refinitiv, Capital Economics

Does such an estimate seem reasonable? In general, indicators of the appetite for risk suggest that investors should be willing to invest. Indeed, the spread between BBB-rated corporate bonds and government bond yields is currently 140bps, well below the financial crisis peak, but in line with its pre-crisis average. (See Chart 7.)

Chart 7: Corporate BBB Less Govt. Bond Yields (Bps)

Source: Refinitiv

Nevertheless, there are reasons to think that the riskiness of holding property is higher than the pre-crisis period. For one, property leases have become more flexible and therefore landlords’ income is less secure. Indeed, as an indicator of implied rental security, the share of over-rented less the share of reversionary leases is higher than the pre-crisis period. (See Chart 8.)

Chart 8: Yield Spread and Share of Over-Rented Less Reversionary Leases

Sources: MSCI, Refinitiv

In turn, there is a greater risk that income from over-rented leases could be lost to units let at (or below) market rent.

Admittedly, since the crisis, average lease lengths have not reduced that much compared with their pre-crisis decline, most of which would already have been priced into the risk premium. (See Chart 9.)

Chart 9: Average Lease Length (Unweighted)

Source: MSCI

Nevertheless, the share of leases with a break clause has doubled since the financial crisis. (See Chart 10.) The fact that break clauses have continued to increase, regardless of the economic cycle, suggests that this trend could continue. However, shorter leases can also be an incentive for tenants, so if lease lengths were to fall, the trend could stabilise.

Chart 10: Share of Leases with Break Clauses (%)

Source: MSCI

In addition, there are reasons to believe that the illiquidity premium has increased in the post-crisis period. Admittedly, over time the property market has become more liquid as access to trading has been opened to a wider range of investors, such as through cross border investment, REITs and the use of derivatives. But there are factors which we think have provided more than an offset.

The most important is the introduction of more stringent regulation on lending for commercial property, including requiring differing risk weights for loans depending on the financial strength of the borrower and the characteristics of the asset. We believe that this has limited the availability of finance, at least for more risky lending, and increased the illiquidity premium.

Indeed, according to the Bank of England, maximum loan-to-value ratios have averaged 60% since the financial crisis, compared to an average of 78% between 1986 and 2006. Further, the spread on new lending has not returned to its early-1990s to mid-2000s level of around 130bps, holding steady at around 260bps since 2013. (See Chart 11.)

Chart 11: Lending Terms for Commercial Real Estate

Source: Bank of England

This has been exacerbated by the fact that commercial property has become more debt financed over time. Indeed, in 2018, average property gearing was 30%, higher than the 20% gearing that was the norm in the 1990s.

That said, there has been an increase in non-bank lending in recent years, which has improved the availability of finance. But this still only accounts for 20% of total lending.

The upshot is that there are reasons to believe that the risk premium should be higher than its early-1990s to mid-2000s level. But in the long term, as regulation changes, we think that it will reduce a bit. We have pencilled in a fall towards 2%.

What does this all mean for our property forecasts?

Overall, since many of the drivers of the wider property to bond yield spread appear cyclical, we expect the spread to narrow as economic growth picks up pace in late 2020. In turn, we think that expected rental value growth will increase. In addition, stronger growth will entice the Bank of England to raise interest rates.

Nevertheless, with the risk premium structurally higher, the spread is not likely to narrow to its early-1990s to mid-2000s average of 150bps. In fact, given our assumptions that expected rental value growth will increase to 1.5%, the combined contribution from transaction costs and depreciation is perhaps 1.7%, and the risk premium could reduce to 2%, the yield spread in the long run could be around 70bps higher than the early-1990s to mid-2000s average of 150bps.

The upshot is that, as interest rates rise, property could start looking overvalued more quickly. However, with 10-year gilt yields currently at 1.2%, even if property yields are unchanged, there appears to be room to absorb around a 100bps increase in bond yields without a major rise in property yields.

But what does this mean for property yields in the long run? Looking beyond 2023, we expect 10-year gilt yields to rise to 3.25%. Adjusting for a long-run yield spread of around 220bps, this implies a long-run required initial yield of 5.45%. (See Table 2.) This compares to the all-property initial yield in Q1 this year of 4.6%.

Table 2: Future Required Property Initial Yields (%)

Long-run

Bond yield

3.25

+ transaction costs and depreciation

1.7

+ implied risk premium

2.0

– expected rental value growth

1.5

= required initial yield

5.45

Sources: MSCI, Refinitiv, Capital Economics

Of course, these estimates should be taken with caution. For one, it is difficult to estimate the effects of structural changes. Further, if economic growth and interest rates do not increase as we expect, there is little reason to think that the yield spread will narrow much from current levels. Nevertheless, the bigger picture is that, when the yield spread does narrow, it seems plausible that it won’t be back to its early-1990s to mid-2000s level.


Amy Wood, Property Economist, +44 20 7808 4994, amy.wood@capitaleconomics.com