What can we learn from the “equity risk premium” about the prospects for equities vs bonds

Introduction

The “equity risk premium” could stake a claim to being the most important number in investment. There are different ways to measure it but conceptually they all come down to the same thing: assessing the return pickup from investing in equities compared with bonds. In general terms a risk premium can be thought of as a measure of the additional return that investors demand or expect for taking on a particular kind of risk, relative to some alternative. Other examples include a credit risk premium, for corporate bonds compared with government bonds, or an illiquidity premium, for illiquid private assets compared with more easily tradable public assets. These assets are the core building blocks for the vast majority of portfolios, most famously in the classical 60% equity/40% bond portfolio. Their valuations and outlooks also have a bearing on most other asset classes, including private assets. That’s why it matters so much.

The rationale

Buy a bond and hold it until it matures and you know what you will get back. Invest in equities and the range of outcomes is wide. You could make a lot of money, but you could lose a lot. Equities have to have a higher expected return to compensate investors for taking on this risk. Otherwise why bother? And the “equity risk premium “is one way to assess this extra payback.

If it’s high, and you have conviction, it can be an argument for allocating more to equities and less to bonds, and vice-versa.

Importantly, this is all about expectations. There is no way of knowing how equities or bonds will perform until it happens. You can balance the probabilities in your favour but just because you expect equities to do better doesn’t mean they will. Risk means more things can happen than will.  That risk is the price of the entry ticket to the equity market.

In this article we look at three of the most popular ways of assessing the equity risk premium and what they say about the prospects for equities compared with bonds today:

1. The historical approach
2. The simple approach
3. The “what’s priced in” approach

We focus on the US for reasons of data availability but also provide some comparisons with Europe and the UK.

The historical approach

This looks at the past performance of equities compared with bonds over a long time horizon. And, with disregard for compliance disclaimers, uses this as an estimate of what they might be expected to earn in in future.

For example, US equities have outperformed long-term US government bonds by 4.5% since the year 1871 (to March 2023), a 152-year period covering wars, depressions, booms, busts, and everything in between.

Adherents to this approach would plug a figure of 4.5% into their asset allocation models, for the assumed outperformance of equities over bonds.

Two challenges with this approach are (1) the answer you get depends on the length of history you are able to analyse and (2) by being backward looking, it is insensitive to whether equities or bonds have better prospects at this point in time.

On the first of these, Figure 2 highlights the significant variability in the historical estimate of the ERP, depending on how far back you look. It could be as low as 2.3% (over the last 23 years) or as high as 6.7% (over the past 91 years). Even a difference of a few years could make a big impact if those years cover big market moves.

Plugging 2.3% or 6.7% into an asset allocation model would result in a very different equity/bond split than if 4.5% was used. The length of history available to analyse isn’t always within your control. Not all markets have such a long time series as the US, especially emerging markets. This makes any historical estimate of the ERP hostage to data availability. And, as shown above, that roll of the dice can yield very different results.

The issue about ignoring relative valuations is potentially even bigger. Under this approach, periods of very strong outperformance lead to a higher estimate of the ERP. In the 50 years to 31 December 1999, US equities had outperformed bonds by 7.6% a year. In the 100 years to that date outperformance was 5.8% a year. Both were close to their all-time highs (Figure 3) and would have resulted in elevated equity allocations if used as an input to asset allocation modelling.

This was just before the Dotcom crash, when equity valuations were at record levels of expensiveness and 10-year Treasuries yielded more than 6%. Equities went on to underperform bonds by 7.4% a year in the 10 years which followed (a period made to look worse by the Global Financial Crisis but, even prior to that, equities were underperforming bonds by more than 3% a year)

Using the historical average may seem like the easy way out but it is not necessarily helpful for deciding on asset allocation. The next two approaches attempt to overcome this shortcoming.

The simple approach: the yield-gap

An easy to calculate, and hence popular, approach to assessing the relative prospects for equities and bonds is to compare the earnings yield on the equity market (the inverse of the price/earnings multiple) with the yield on the 10-year Treasury. When the earnings yield is high relative to bond yields, this approach argues that equities are cheap relative to bonds, and hence more appealing. The opposite is also true.

This is sometimes referred to as “the Fed model” even though it has never been officially endorsed by the Federal Reserve. A variant compares the earnings yield with the real yield on 10-year Treasury inflation-protected securities (TIPS). Another compares it with Treasury bills/cash. The current conclusions (discussed later) are the same whichever approach you take.

The yield-gap’s historical track record of providing insight on the future difference in equity and bond returns is mixed (Figure 4). Over the very long-term, there has been a positive relationship between the yield-gap and subsequent returns on a 10-year horizon. Periods when it has been high have been more likely to be followed by periods of stronger long-term outperformance from equities over bonds.

The relationship is much weaker when the yield-gap has been closer to zero or negative. There are lots of instances when a low or negative yield-gap has preceded a period of very strong performance from equities, most obviously the 1980s and 1990s (Figure 6).

There are two main reasons why this indicator can underestimate equity prospects. First, it ignores the earnings growth and dividend income components of equity returns. Second, and harder to estimate, even if equity valuations are expensive, that doesn’t mean they can’t become even more so, boosting returns in the process.

When we look back over historical periods when equities have done well from the starting point of a low yield-gap, a near-condition has been real earnings growth. In many such cases, valuations have fallen but equities have still done much better than bonds – because of strong real earnings growth. It has been more of a rarity for real earnings growth to be negative but valuations ride to the rescue.

The strength of this relationship also weakens as the investment horizon shortens (Figure 5). It has not been helpful in giving a steer on short-term market movements.

How should we interpret the current reading?

The yield gap approach can add some value to setting strategic asset allocation, but almost none for tactical.

It has fallen to a depressed level of only 0.6%. This has been driven by bond markets repricing much faster and further than equities. Since December 2021, the 10-year Treasury yield has risen by 2.2%, from around 1.5% to 3.7%. The equity earnings yield has only risen by 0.1%,  from 4.2% to 4.3%.

Real yields on 10-year TIPS have risen by slightly more, and cash rates have risen by more than 5%, so the broad conclusions are similar if we calculate the yield-gap using real yields or cash rates.

The TINA trade – There Is No Alternative – was a popular rationale for strong equity performance in the low-interest rate environment. But now there is an alternative. Bond yields are dramatically higher. Cash has also become a more viable alternative, with US cash rates now exceeding bond and equity yields – albeit bonds deliver a yield over a longer time horizon whereas cash rates are unlikely to stay at current levels for such a prolonged period.

This doesn’t have to mean that US equities will struggle versus bonds. Earnings could grow strongly or valuations rise further. But, with corporate profit margins and equity valuations both still-elevated, both face headwinds.

Although US equities cannot be written off, our analysis suggests the outlook for equities is gloomier versus bonds than many investors will have had to contend with for a long time. With the yield gap around 1%, the reward for taking US equity risk has diminished.

The same is not true of other markets though. The yield-gap has also come down for Europe ex UK equities compared with German bund yields, and UK equities compared with UK gilt yields, but not by as much (Figure 7). The yield-gaps for Europe ex-UK equities and UK equities were both 4.3% at the end of May.

European and UK yield-gaps are also within their ranges of the past 15 years (Figure 7), rather than having dropped well below them, as has happened in the US. Relative to their own histories, European and UK equities continue to offer reasonable value compared with bonds.

Both yield-gaps are a lot higher than the US in absolute terms, although this does not capture their relative growth outlooks (see next section). It could also be flipped around and interpreted to mean that investors are demanding a higher risk premium for investing in European and UK equities compared with the US. It should not be interpreted as a “free lunch”.

The “what’s priced in” approach

This approach looks at equity prices and consensus expectations for earnings growth and “backs out” the return assumption that is priced into the equity market. In simple terms, the equity market price equals the sum of discounted future cashflows from the market. That discount rate can be thought of as the return demanded by investors (it is the internal rate of return).

If the equity price falls, you need a higher discount rate to set the present value of cashflows equal to that new lower price, all else being equal (which it rarely is but that’s not important for this framing). In other words, a lower price leads to a higher equity return, all else equal. That is why this can be thought of as looking at the return assumption that is priced in to equity markets.

It is possible to get more granular by coming up with a set of assumptions based on one’s own view of the outlook. But the aim here is not to work out what is “most likely” in one’s own opinion, but what the market is expecting/ what is priced in.

The ERP can then be calculated as the difference between this forward looking equity assumption and the risk-free rate, such as the yield on 10-year government bonds.

We take a multi-stage approach to assessing the ERP on this basis. For the current and next two calendar years we use consensus analyst earnings growth forecasts from I/B/E/S, the Institutional Brokers' Estimate System. For subsequent years we assume that earnings grow in line with consensus expectations for 10-year real GDP growth and inflation, sourced from the Survey of Professional Forecasters. This is a simplification as earnings growth and domestic GDP growth do not move in tandem e.g. some earnings are earned overseas so depend on international growth. However, adding such additional complexity only has a small impact on the equity return outlook and hence would not materially impact any of our conclusions. A normalised payout ratio of 50% is assumed. This is higher than the dividend payout ratio to reflect the popularity of share buybacks.

Figure 8 shows the evolution of the equity return priced into the US market since 1992 (the date when I/B/E/S consensus earnings forecasts first became available) alongside bond yields. This shows the nominal equity return outlook. The real equity return outlook has not risen by as much in recent years, as part of that move has been down to higher inflation expectations.

The drop for the recent equity return figures is primarily because consensus expectations for long-term US real GDP growth have recently been cut from 2.3% to 2.0%, and inflation from 3.0% to 2.4%. As with the yield-gap approach, Figure 8 highlights that the bond market has repriced a lot more than the equity market. This would have been true even without the latest cuts to the US growth and inflation outlook.

Figure 9 shows how the ERP has varied over time using this approach. It has collapsed to is lowest level for twenty years.

How reliable is this measure as an indicator?

As with the yield-gap approach, this indicator has a reasonable, if slightly mixed, track record of success. A high ERP has been associated with better future 10-year equity performance vs bonds but the relationship is weaker at low levels (mainly from the 1990s).

As with our other estimates of the ERP, there has been very little relationship between the ERP that is priced in and subsequent returns over shorter time horizons (Figure 11 shows this on a five year horizon).

How should we interpret the current reading?

This is a slightly more damning assessment than the yield-gap approach, because this takes account of consensus expectations for earnings growth. The ERP was lower in the 1990s yet equities performed very well compared with bonds, but that was helped by soaring valuations. Given the starting point and outlook today, that seems a less likely outcome, relying on hope rather than expectation.

As with the yield-gap, the ERP priced into European markets should not be as worrying for European equity investors (Figure 12). It has also fallen sharply, pointing to reduced reward for bearing equity risk, but remains above pre-GFC levels.

Conclusions

There are different ways to assess the outlook for equities compared with bonds. None is perfect but all can be useful. Historical estimates may seem like the easy option but they take no account of current market valuations and are sensitive to the time period assessed, which depends on availability of data. For non-US markets this can be particularly problematic and lead to potentially misleading conclusions.

Our more forward looking measures tell a consistent story. Bond yields have re-priced more than equities. US equity investors today are being rewarded with a smaller return premium for bearing equity risk than at any time in recent memory, at a time when macroeconomic risks are high and central banks are in less supportive mood. More risk, less reward.

The US looks particularly bad on this basis with things not as worrying in Europe and the UK. The US may have been the strongest performing market for much of the past 15 years, but our analysis of the ERP suggests that it will struggle to repeat that feat. And, with the US having risen to now make up 68% of the global developed stock market, global equity investors are highly exposed to US performance. Long-term investors may be better served by allocating more to non-US markets in the decade to come.

Importantly, our analysis demonstrates that these frameworks are only useful when setting strategic asset allocation on a long time horizon, such as 10 years. In the shorter term, other factors can be more in the driving seat. There will be shorter term periods when equities (US or elsewhere) could do much better, or much worse, than bonds. But identifying those requires a different toolkit.