Richard Suvak, MSF, CFA

Investing is all about seeing the future. Fortunes can be made with perfect clairvoyance, but as we know, such clairvoyance can be difficult, if not impossible, to come by. Sometimes, however, seeing the future can be as simple (if there’s such a thing as simplicity in the stock market) as connecting the dots.

Here then follows a step-by-step method to see the future.

We know (although I won’t go through the analysis) that interest rates effect the stock market. Largely this is due to the competing returns interest bearing instruments offer relative to the stock market as well as the discount rate used in present-value formulas for equity earnings and/or dividends. High or rising interest rates lead to lower equity returns while low or falling interest rates lead to higher equity returns.

__Competitive Choices__

Using the table and chart below, we can see that investor preferences between the S&P 500 and 10-Year T-Bond yields becomes challenging as average bond yields begin to match or exceed those of equity returns. As a result, investors are faced with the choice of similar returns albeit with significantly different risk profiles (not shown). That is, would investors prefer 10.7% equity return +/- 15% (or more), or 8.4% bond returns +/- 10%. Most rational investors would choose the latter.

__Discount Rate__

Pushing investors further away from equities in high interest rate environments is the rate at which future cash flows are discounted back to the present. As interest rates rise, the discount rate rises. The 10-Year T-Bond yield is most often used as a discount rate, although not exclusively. The table and chart below show the impact a changing discount rate has on the present value of $1 – the higher the discount rate, the lower the present value. As a significant portion of equity valuation models are based on discounting of earnings or dividends, interest rates are therefore a significant factor in investors perception of future equity market returns. In short, in high interest rate environments the equity market appears more expensive because tomorrows dollar is worth less today.

__1 ^{st} Conclusion__

Interest rates are critical when determining equity market value!

Therefore, our first dot will be to understand interest rates and their direction. Starting with the simplest (again, if there is such a thing), short-term rates are largely driven by changes in inflation and Federal Reserve interest rate decisions (which are largely driven by changes in inflation and economic growth). Therefore, it is reasonable to conclude that short-term interest rates are heavily influenced by inflation and the Fed Funds rate (the rate the Fed manipulates when setting policy). Using these two factors as input, we can see the efficacy of the predicted 2-Year T-Bill yield relative to actual 2-Year T-Bill yield. The model has an R-Squared of 0.95, or predictive power of about 95%. Further, we can see that the model predicted that current rates would be 1.97% (versus the actual 1.84%) and predicts December 2018 yields to be 2.29%.

__2 ^{nd} Conclusion__

The second dot in our connected conclusion is that our model thinks short-term interest rates are rising, albeit not dramatically.

So, if short-term rates are rising, what are long-term rates going to do? Generally speaking, the longer you hold an investment, the more risk you accept. As a result of this increased risk, investors in longer-term investments must be paid to accept additional risk. In other words, the return of longer-term instruments is generally higher (although the opposite can be said preceding recessions – but that’s another topic) as compensation for the higher risk longer-term investors must take. Nonetheless, (typically) long-term rates are heavily influenced by short-term rates. Therefore, it’s reasonable to build a model incorporating actual (and our projections) for short-term rates to predict long-term rates. Doing so produces the following predictions for the 10-Year T-Note. It has an correctly predicts the actual rate 97% of the time. From the line graph, we can also see the model has been predicting increasing interest rates since 2012, currently suggesting 10-year rates should be closer to 3.4% versus the current 2.4%. I attribute this difference to the Fed’s lingering program of Quantitative Easing that they have employed since the Global Financial Crisis. By all accounts, they are in catch-up mode as the economy and inflation heat up. The model also suggests further increases (to 3.8%) by December 2018 as short-term rates and inflation rise throughout the year.

__3 ^{rd} Conclusion__

Long-term interest rates are rising.

Our connected dots so far suggest inflation, the Fed Funds rate, short-term rates and long-term rates are all on the rise. We made the case at the beginning of this paper that equity prices are heavily influenced by interest rates (among other things). Can we connect that dot as well?

Adding our projections for long-term rates along with 12-month earnings growth to our model, we can reasonably forecast market valuation (based on Shiller P/E). While the prediction accuracy drops dramatically ( to 61%), it is still an excellent forecast of market valuation. From the historic chart you can see that it missed the late 1990’s internet bubble, as well as more minor differences elsewhere. One concern with this result is that the model does not (possibly can not) predict market bubbles like that of the late 1990’s. The question then is, are we at the beginning stages of a similar bubble and if so everything we do here today becomes useless.

If not, the models suggests the current Shiller P/E should be 29.75, while it is currently 31.04. Looking ahead to December 2018, the market suggests the Shiller P/E to be 23.07. That’s a dramatic drop from current levels and can be possible by an increase in earnings, a decrease in prices or a combination of both.

__4 ^{th} Conclusion__

Market valuation will fall. The source of the fall in Shiller P/E (decrease in price or increase in earnings growth) is yet to be determined however. Earnings growth therefore will be a significant determinant in future market returns. Our model will deal with earnings shortly.

The series of charts below show the recent history and projected values (along with the 95-percent range of likely values) for each of the above components. As you can see, the models predicts increasing short-term and long-term interest rates, following inflation and Federal Reserve policy. To account for the increased economic growth, low unemployment, recent tax cuts and overseas tax reversion program, I have used projected earnings growth based on the 90^{th} percentile of modeled earnings growth – which is significantly higher than recent growth, ranging between 20-30% growth over the next year – roughly two- to three-times the current earnings growth rate. Nonetheless, despite this increased earnings growth, market valuation falls about 10 points (from 32 to 23).

Based on these projections, we can build best, worst and average expected returns over the next year. In an attempt to be as conservative as possible, I have assumed only half of the reduction in valuation will translate into return. I have also assumed that Trump’s economic growth projection is (almost) correct, adding 3.5% to return as well as adding an additional 2.5% to account for dividend payments. We are therefore getting a “free” 6% return from economic growth and dividends. The difference between this 6% and the actual return we will get will be accounted for by valuation contraction or expansion. In the chart below, I have provided the distribution possibilities of the best, worst and average case for 2018. As you can see, the best-case scenario translates into low single-digit returns. No doubt, that would be a fantastic year, particularly after the recent bull market run the market has seen over the last several years. The worst-case scenario is much bleaker, with an expected return nearing -20%. The market would be in full bear-market mode if this were to happen. Finally, based on the projections and distributions of projections outlined above, the most likely case is a more modest -7% return.

__Final Conclusion__

Rewriting Shakespeare to suit our needs, the question is; to see or not to see. Predicting the future is near impossible. However, it is possible to build a series of connect-the-dot predictions based on known relationships and range of possible outcomes. Doing so allows some confidence in predictions which might otherwise stretch credibility. In this analysis, there’s no doubt that a positive outcome is possible. However, the skew of possible outcomes leans toward the negative.

It’s always more fun to be bullish, however blind bulls can be a danger to themselves and others.