Description

We recommend shorting the Russell 2000 through shorting a Russell 2000 ETF such as IWM.  We will discuss the overall valuation metrics for the index, then touch on the index’s return on capital, margin structure and leverage levels.  Then, we will discuss the impact of interest rates on valuation.  At points, the write up below will make use of S&P 500 data, since the S&P 500 has better long-term data than the R2000.

Valuation

Analysts most often use P/E ratios to measure the R2000’s valuation.  The Russell 2000 is often quoted as P/E excluding negative earnings for which it sports a 21x P/E ratio.  If we are to include companies that lose money, the P/E increases to 62.7x.  There’s no precise way to measure the index’s valuation given the issues that money losing companies (~35% of the index by number and ~25% by market value) present.   Bulls touting the 21x PE commit a sin of omission, while bears citing a 62.7x P/E capitalize negative earnings for companies that investors judge worth significant positive values. The following example shows the issue with a simple framework that takes index earnings / index market cap (62.7x). Let’s assume that the index is composed of 2,000 companies, each with a $1bn value.  Further, let’s assume that 1,999 companies earn $100m and 1 company loses $100bn. The total index earnings would be $100bn for an index P/E of ~20x. However, if we assume that the money losing company has a value of $0, the index P/E would be ~10x. While we are quick to point out the flaws with the two most common valuation methods, we do not have an obviously better solution. However, through looking at the valuation in a number of ways, we feel comfortable that we are “approximately right rather than precisely wrong.”

P/E

• The median P/E of companies that have positive net income = 21.8x.
• We set companies with PEs >40 (high P/Es) and <0 (money losing) to 40x.  This produces a weighted average P/E for the index of 28.9x P/E.  We think this is the most fair way to deal with the negative earnings issue.
• The index PE (total income / total market value) = 62.7x. 

 EV/EBIT

• The EV/EBIT of the index equals 27.8x excluding financial firms.
• The EV/EBIT (excluding negative earnings) of the index is 20.9x excluding financial firms.

Thus, we think the index trades at rich multiples of ~29x P/E and an EV/EBIT of >20x.  It’s important to note that we are in the 9^th year of a bull-market and an economic expansion.  These multiples capitalize late-cycle earnings. 

Return on Capital and Margins

Stocks should trade at high multiples when margins are set to rise or the business can grow at returns on capital above the cost of capital.  As an aside, the index short eliminates much of the individual security risk.  Shorting a speculative biotech may end in disaster but shorting a diversified index of 2,000 companies eliminates the risk that you may misjudge the merits of an individual security.  Moreover, it’s difficult to suspend the laws of economics for a large and diversified set of companies. 

One would expect a set of 2,000 companies to produce average return metrics.  The median ROE of profitable companies is 9.7% and the weighted average ROE is 17.4%.  Including money losing companies at a -5% ROE (average ROE of money losing companies is actually -85% ROE) reduces the weighted average ROE to 11.8%.

The average return on invested capital (NOPAT / common equity plus long-term liabilities) for profitable companies equals 8.8%, with a median of 6.6%.  A more generous definition of return on capital (NOPAT / PPE + CA - CL) for profitable companies produces an average and median return of 17.4% and 11.9%, respectively (we round down ROIs >100% to 100% to avoid distorting the averages). Theses returns are based on 2017 earnings relative to 2016/2017 IC.  These returns came during a strong economic year and are likely to be lower on a mid/through the cycle basis. Nevertheless, as stated earlier, the returns generated by the R2000 appear to be near or slightly above the cost of capital.  This is not a portfolio of 2,000 Amazon’s and Netflix’s, but a diversified collection of average businesses.  Even if you believe that we are in the early stages of an economic expansion, the multiples for the group as a whole do not make sense. 

With respect to margins, we assume that the Russell 2000’s margins are correlated with the Fed’s corporate margin data.  The Fed’s data set dates back to 1947 and shows a clear mean reverting trend. The average margin over this period is 6.9% compared to today at 9.5% (86.6% percentile).

Our case does not rest on margins falling, but we think that the risk is clearly to the downside.  Margins have averaged 9.8% over the past 10 years.  This ranks in the 96.7% percentile for 10-year periods.  If margins were to fall to historical averages, the Schiller P/E would rise from 32.8x to 48.5x.

Moreover, we hear that margins should be higher today given the shift towards capital light companies in the index.  However, to produce a similar return on capital, economic theory suggests that margins should be lower for capital light companies than for capital heavy ones.  The R2000 does not have many world-beating firms that have high barriers to entry and that can grow quickly organically without significant capital investment.  Anecdotally, private equity has taken out many of the higher quality, smaller companies, leaving the R2000 with a negative survivor bias.  This is difficult to prove, but the pedestrian return on capital metrics show that the index does not deserve a stratospheric multiple

Leverage

These companies are quite levered. For non-financial companies only, the Net Debt to EBIT = 6.3x for the index, 4.7x excluding the losses of making companies, and 4.6x excluding the net debt and losses of loss making companies. Should earnings fall, margins compress or rates rise, the average company’s debt burden will increase from these already high levels.

Interest Rates

Bulls tend to cite low rates to justify current trading multiples. 

Schiller P/E During Similar Nominal Interest Rate Periods

Interest rates have been below four percent in 54% of months since 1871.  When interest rates have been under 4%, the Schiller P/E has averaged 16.7x. Excluding this bull market, the Schiller P/E averages drop to 15.7x. Today the Schiller PE equals 32.8x. The current 32.8x P/E ranks as the 99.8% percentile relative to the other 963 months where interest rates were under 4%.

Interest Rate Cycles

Historically, interest rates have moved in long cycles. There were bull markets in interest rates from 1873-1899, 1920-1946, and 1981-?, and there were bear markets in interest rates from 1899-1920 and 1946-1981. These market cycles have lasted between 21 and 35+ years. Interest rates have been rising since 2016, however, we do not know whether or not 2016 marked the end of the 35-year interest rate bull market starting in 1981. This is not being discussed for the purpose of declaring that the cycle is turning. We mention this as we believe long term historical context is important and is left out of most analyses.  

Global Markets

If we are incorrect and current interest rates in the US justify valuations, then the logical assumption to us would be that even lower interest rates in other markets would justify even higher valuations in those markets. Below are some of the markets with the most extreme interest rates. We use StarCapital’s Stock Market Valuation data for CAPE’s for other markets - https://www.starcapital.de/en/research/stock-market-valuation/.

While, the table above doesn’t necessarily say that the US is overvalued, there is a clear relative valuation discrepancy between the US and the other markets. In the other three markets shown, interest rates are significantly lower while valuations are less demanding.

Compared to July 1, 2016

The Russell 2000 is up 46.3% from July 1, 2016 despite the fact that the 10 Year Treasury has gone from 1.47% to 2.88% and the 10 Year TIPs increased from 0.09% to 0.79%. While both rates and the index price in future expectations not current/trailing numbers, valuations have gone up while interest rate expectations have also moved up.

Real Interest Rates

Lastly, we find most of the analysis done using nominal interest rates to be unsatisfactory. While interest rates are nominal, history shows that earnings of businesses adjust over time to inflation. Therefore, comparing interest rates (nominal) with P/Es or earnings yields (real) is like comparing apples and oranges. Comparing a 10-year treasury yielding 2.9% today vs. 12.4% in December 1980 in the context of stock market valuation or the attractiveness of buying the 10-year treasury is misleading. Inflation in 1980 was 13.5% and inflation today is 2.1%. Assuming inflation stayed the same in both periods for the ensuing 10 years, the 2.9% treasury today would actually be the more attractive one in real returns. Its real pre-tax return would be better and real after-tax returns tax would be significantly better than the 1980 bond despite the lower nominal yield. Obviously using current inflation for the next 10 years is a stretch and ideally, we could compare TIPS rates which price in expected inflation, however TIPs were only introduced in 1997. We believe real yields (TIPS) relative to stock earnings yields is not perfect but a better comparison than comparing to nominal yields. Based on this assertion we are not surprised to find essentially no correlation between nominal interest rates and stock market Schiller P/Es. We would expect a stronger correlation between TIPS rates and P/Es but unfortunately no such data exists. We believe this disproves any logic saying that 3% nominal interest rates justify the current valuations.

Catalyst

Recession

Interest Rate Hikes

Quantatitve Tightening

China

Emerging Markets