Special Report
Highlights Rates are going higher ... : Flight-to-quality episodes aside, the bond bear market that began in July 2016 remains in force. Investors should maintain below-benchmark Treasury duration. ... but that doesn't necessarily spell immediate trouble for stocks: Consistent with our work on the fed funds rate cycle, it appears that the level of rates matters more for equity returns than their direction. Empirical evidence of a rates tipping point is elusive ... : The notion that investors migrate from stocks to bonds at a particular level of rates exerts a powerful intuitive appeal, but the data fail to validate it. ... but a 10-year yield Treasury of 3.75 - 4% might halt the bull market in its tracks: Higher rates reliably slow equities only when they rise enough to slow the economy. We estimate that the pinch point is somewhere in the neighborhood of a 3.75 - 4% 10-year Treasury yield. Feature A share of stock is a pro rata claim on the future earnings of the company that issued it. Holding future earnings constant, the price an investor will be willing to pay for a share is wholly a function of the rate used to discount its earnings back to the present day. The simplicity and ubiquity of this valuation approach suggest that equity returns should be predictably related to moves in interest rates. It may also point the way to a tipping point - either in the level of rates, or the magnitude of their rise - at which capital and savings migrate from stocks to bonds. This Special Report reviews the historical record to see how U.S. equities have interacted with real 10-year Treasury yields. It considers the key variables that would logically seem to bear on equity performance and investors' propensity to rotate between asset classes. We find that the relationship between rates and equity returns is conditional, depending on which crosscurrent dominates in any given episode. We did not uncover any predictable rotation pattern. Do The Math As noted above, valuing a stream of future cash flows is a simple mechanical process once one settles on an appropriate discount rate for converting future dollars to current dollars. According to the security analysis textbooks, then, moves in stock prices are inversely related to changes in interest rates. But the textbooks leave out one key point: changes in interest rates don't occur in a vacuum. When they change, earnings estimates are likely to change, too, most often in the same direction as real rates. To be sure, the denominator discounting future cash flows rises when real rates rise, but the future-earnings numerator most likely rises, too. If real rates are rising, the economy is probably gaining momentum, and earnings estimates should probably be revised higher as well. Conversely, falling rates lead to a higher earnings multiple (ex-the not insignificant animal-spirits wild card), but will regularly be accompanied by downward revisions in future earnings. The net effect is uncertain, and depends on whether the multiple change outweighs the change in earnings or vice versa. Bonds Are A Snap Compared To Stocks It's far simpler to compute the impact on a bond portfolio from a given increase in interest rates because the denominator is the only variable that changes. The future-cash-flows numerator is contractually fixed, and it takes a big shift in the state of the economy to spark an economy-wide change in perceived repayment potential.1 This is why bonds' sensitivity to changes in interest rates can be captured in a single universal metric (duration). Stocks are pulled in so many different directions by factors affecting future cash flows that duration has no equity analogue. Investors should therefore be cautious about pinning too much on interest rates as they relate to equities. Bonds move in fixed orbits around the interest-rate sun, according to strictly ordered rules that establish a very clear cause-and-effect relationship. Equities improvise as they go along, taking their cues from a rotating cast of variables that interact differently over time. Attempts to stretch the concept of interest-rate sensitivity beyond bonds regularly trip up equity investors; we cannot know in advance how rates will come together with the other factors that influence equities. Confounding Intuition, Part 1: Equities Prefer Rising Rates (And Multiples Don't Care) U.S. postwar history makes it clear that equity investors need not run from rising rates. The S&P 500 has fared considerably better when real 10-year yields have risen by at least 100 basis points ("bps") than it has when they've declined by that magnitude (Chart 1), gaining 9.4% and 5%, respectively (Chart 2). Rates do not exhibit any sort of a consistent relationship with either forward (Chart 3) or trailing (Chart 4) S&P 500 multiples, though extremely high and extremely low real yields are both associated with lower trailing P/Es. Negative real yields carry an unwelcome whiff of deflation, and their scatterplot data points tend to cluster at below-the-mean forward and trailing multiples. Chart 1Stocks Actually Do Better When Rates Rise ... Chart 2... Considerably Better When Do Higher Rates Hurt The Economy? Charts 3 and 4 show that both forward and trailing multiples almost always decline when real 10-year Treasury yields cross above 5%. What's bad for multiples isn't necessarily bad for earnings, however, and a 5% real threshold is irrelevant to today's cycle. The steady decline in the average fed funds rate over the last several completed cycles (Chart 5) makes it clear that neutral rate thresholds are not constant across time periods. Assessing interest rates' impact on the economy over time requires a sliding scale. Chart 3Hard To See A Trend Through The Windshield ... Chart 4... Or The Rear-View Mirror Estimates of potential economic growth provide a useful yardstick for measuring the impact of real yields. Comparing real long rates to potential output offers insight into the burden of servicing debt across the economy. If real rates exceed the economy's potential growth rate by a material amount, several marginal borrowers are likely to be gasping for air, and their travails will weigh on the economy. Conversely, servicing debt should be easy when real rates are below potential growth, and investors are more likely to invest, businesses are more likely to expand, and consumers are more likely to spend. Chart 5One Size Does Not Fit All There have been 22 instances in the postwar era when real 10-year Treasury yields have increased by at least 100 bps, and Table 1 lists all of them, grouped by their relationship to real GDP's potential five-year growth rate. There are three possible states for interest rate increases in relation to potential output: starting and ending below trend growth, starting below trend growth and ending above it, and starting and ending above trend. The S&P 500 comfortably tops its overall postwar returns when rates go from Below-to-Below and Below-to-Above, but declines outright when rates start above potential growth and go even higher. Earnings consistently rise when rates start below potential growth, making multiples the swing factor - when they expand, S&P 500 gains tend to be very large (Box 1). Table 1Real Rates Versus Potential GDP Growth Box 1 Decomposing S&P 500 Returns Table 2 details the decomposition of S&P 500 returns during rising real rate episodes occurring after S&P 500 earnings estimates began to be compiled in 1979. Except in the crucible of 2009, when they were flat, forward earnings estimates have always risen when rates rise from a below-trend starting point, putting a tailwind behind the S&P 500 that regularly overcomes the multiple contraction that occurs in half of the Below/Above instances. Multiples are the swing factor; when they expand in conjunction with rising earnings estimates, U.S. equities soar. They always contract when rates go from high to higher, dragging stocks down against a mixed earnings expectations backdrop. The action is consistent with our fed funds rate cycle work: stocks do best when rates are below equilibrium and falling because earnings and multiples expand in tandem in that setting, but they do nearly as well after rate hikes commence, in spite of multiple contraction. Earnings surge when the Fed is confident enough about the economy to embark on a tightening cycle, but has not yet hiked enough to choke off the expansion. Multiple expansion in a majority of the Below/Above instances reveals that investors do not rotate out of equities en masse when rates rise, even by a considerable amount. The rotation story has intuitive appeal, but it doesn't show up in these data. Table 2Decomposition Of S&P 500 Returns During Rising Rate Periods A Little More Slicing And Dicing (Potential GDP Matters) Chart 6Mind The Gap Defining Below-to-Below and Below-to-Above states is easy in hindsight, but an investor cannot know in real time where a rising-rate instance that begins with rates below potential output will end. Earnings rise no matter where rates end relative to potential GDP, but re-rating in Below/Below can flip to de-rating in Below/Above, slamming the brakes on phase gains. The empirical data say investors should lighten up on S&P 500 exposure when real rates cross above real potential GDP. S&P 500 returns trounce their overall postwar gain when rates rise from below potential GDP to potential GDP but lag it once rates cross above potential GDP (Chart 6). Confounding Intuition, Part 2: Institutional Investors Don't Rotate Even if S&P 500 returns fail to demonstrate any consistent relationship with interest rates, one would expect that professional investors' asset-class positioning would. Bonds and stocks are alternatives for one another, and institutional investors presumably shift their allocations in line with the asset classes' relative prospects. We examine Pension Funds', Life Insurers', and Mutual Funds' asset-allocation profiles over time using balance-sheet data from the Federal Reserve's quarterly Flow of Funds report. The data show that asset-allocation decisions are made without apparent regard for relative valuations, at least as proxied by the equity risk premium. Pension funds' steady increase in equity allocations across the '90s appears to have been less a function of rate moves than buying into the bull market (Chart 7). Since the dot-com bubble burst in 2000, bond and equity allocations have mainly reflected the performance tides. The extended trend in pension funds' equity-to-bond allocation ratio suggests that the funds set a long-range goal and grind steadily toward achieving it, regardless of relative valuation movements. It also suggests that the funds may not bother with rebalancing, much less dynamic asset allocation. Life insurers kept their fixed income and equity allocations more or less fixed across the '70s (not shown) and most of the '80s. They then reduced equity exposure for three years after 1987's Black Monday, assiduously built it up across the '90s, and have more or less let it drift since the millennium (Chart 8). The equity risk premium does not appear to have been a consideration. Asset-allocation stasis may simply be a reflection of life insurers' stringent regulatory constraints, but their portfolio managers' limited discretion precludes opportunistic allocation shifts. Mutual fund allocations tend to depend much more on past events than future expectations. Equity holdings peak when the equity risk premium bottoms and bottom when the equity risk premium peaks (Chart 9). The problem is that mutual fund managers are structurally hostage to their investors' whims. They are sorted into narrow silos and then straitjacketed by the rigid allocation rules written into their fund prospectuses. Even if they think asset-class rotation is a great idea, only a tiny minority of fund managers can act upon it. Chart 7Pension Funds Don't Allocate Based On Yields Or The ERP ... Chart 8... While Life Insurers Appear To Allocate In Defiance Of Them Chart 9Mutual Funds##BR##Obey Their Owners ... Confounding Darwin's Intuition: Human Investors Never Learn Chart 10... Who Act On Real Emotion, Not Real Yields Kahneman and Tversky's groundbreaking research into decision-making under uncertainty revealed that our species is wired to make suboptimal investment decisions. Prospect theory, loss aversion and an unhealthy fixation on recent data all encourage retail investors to repeatedly shoot themselves in the foot. When it comes to asset allocation, households appear to focus exclusively on the action in the rear-view mirror (Chart 10). Retail investors as a group rotate between equities and fixed income retroactively, in response to recent past returns, not proactively in response to cues about future relative-return prospects. Investment Implications Despite the compelling intuition that investors should set their course by the interest-rate stars, there is no evidence in the flow of funds data that they have done so in the past. We posit that structural constraints on institutional investors, combined with humans' durable cognitive biases, offer no reason to expect that they will do so in the future. While there may not be any predictable rotation pattern, rising rates have given rise to a predictable performance pattern. Equities reliably perform better when real rates are rising by at least 100 basis points than they do when they're falling. Decomposition of S&P 500 returns indicates that the pattern holds because earnings rise a good bit more in rising-rate periods than multiples decline. And multiples don't always decline when rates rise, anyway; sometimes emotion overrides cash flow discounting mechanics. Investors should lighten up on Treasury allocations, while keeping the exposures they do hold at below-benchmark duration. They should not flee equities, however. Rates have not yet risen enough to cool off the economy in any material way, and we judge that they won't until somewhere around a 3.75% 10-year Treasury yield.2 Tight supplies in labor and goods markets will eventually stoke realized inflation and provoke the Fed into tightening enough to cut off the rally, but it hasn't happened yet, and it is far too early to de-risk portfolios on account of interest rates. Doug Peta, Senior Vice President U.S. Investment Strategy dougp@bcaresearch.com 1 An unusually large drop in rates may well be associated with economic distress, but default-adjusted bond payment streams are much less variable than near- and intermediate-term earnings estimates. 2 Based on the evolution of the Congressional Budget Office's longer-run estimates of real potential GDP growth, and the trend in our own model of long-term inflation expectations, it appears as if nominal potential GDP growth will be somewhere in the neighborhood of 3.75-4% next year. This is a much lower estimate than one would get from adding the Fed's 2% inflation target to the current rate of GDP growth, but we need to look past the immediate boost of the stimulus package to get a read on its longer-run effects. As with all of the estimates produced by our models, we look to it for a general guide to the future, not a precise point estimate.
