The SMCCF, Noise Trader Risk, and Samuelson’s Dictum
May 2020
On May 11th 2020, the New York Fed published a 66-page ‘Investment Management Agreement’ with BlackRock – a document that converts the Fed’s ability to make secondary corporate credit purchases from a signaling device into a contractual reality. The Fed is thus following the ECB and BOJ in pushing the frontier of how central banks interpret their mandates, increasing its footprint in non-sovereign risk markets. This adds to a scope and scale of monetary and fiscal intervention amid corona for which we lack any precedent, both in ‘stable’ times or amid the currently sharply lower employment, output, and corporate profitability expectations. Nonetheless, especially equity valuations in the US have proven resilient, leading to wide debate on the divergence of risk asset prices and the real economy. I will explore mechanics of the Fed’s new secondary corporate credit facility, and then suggest a framework that attempts to rationalize ex ante why central bank liquidity provision leads stylized investors to pay rich multiples across asset classes regardless of fundamentals. I will close by considering how this ‘Fed put’ aligns micro-efficiency with macro-inefficiency.
Overview
By decreasing yield across an increasingly exhaustive list of fixed income classes, central banks are crowding investors out of safer instruments and pushing them into riskier ones. Zero interest rate policy (ZIRP) thereby creates a rotation that moves investors seeking yield with previously more defensive allocations up the risk spectrum. In a 2015 CNBC interview, Icahn analogized this to Fink (BlackRock CEO) and Yellen (former Fed Chair) pushing investors towards a cliff, given the amount of funds redeployed through accessible, low-fee, passive US stock and bond ETFs. Even amid corona, 2020 saw $90bln in YtD inflows as of May 1st. Through its agreement with BlackRock, the Fed is now creating a secondary market corporate credit facility (SMCCF) that allows it to follow investors into such bond ETFs.
Interestingly its stated goals do not obviously justify the means; the language proposes as desirable ‘order’ and ‘normality’ in markets, with only the chosen means telling us the Fed has ensuring prices, not ensuring price discovery, in mind. SMCCF candidates will range from investment grade to high yield, representing another central bank squeeze on risk premia – this time by pushing down credit spreads – that will intensify hunt for yield by investors spanning pension funds to ‘mom and pop’. However, while the proliferation of passive equity investing has made us comfortable with the laissez-faire convenience of market capitalization weightings, we will see that for credit, market-weighting prioritizes different variables. The Fed’s approach does not circumvent higher exposure to more levered segments of the market, resulting in a risk profile that may be harder to reconcile with stated SMCCF goals, despite risking Treasury equity. On the whole, the SMCCF thereby requires a quite generous interpretation of how the Fed’s dual mandate is pursued in the corona crisis.
More generally, the SMCCF adds to a growing source of excess return that is largely decoupled from fundamental risk: the Fed signaled as a buyer of last resort. Investors in a market with Fed intervention are simplistically receiving a free put option at a strike of what price the Fed deems “orderly”. This may create noise trader risk, which exists when rational investors expect deviations from fundamental values to increase in the short run, such that they are incentivized to buy an asset and sell it later on to someone accepting even greater overvaluation. The ‘greater fool’ ends up owning the asset into its eventual convergence to fundamental value. I will explore whether this framework explains the lofty valuations paid by investors who, given Fed liquidity and low-fee passive vehicles, are pushed into higher risk asset classes despite having heterogeneous preferences that previously solved for more fragmented holdings across the risk spectrum. My analysis will suggest a simple chain effect, justifying the expectation of always finding a ‘greater fool’ and boosting prices without regard to fundamentals even in markets the Fed does not engage directly. As a result, this varied investor base is incentivized to ignore its risk aversion, creating reward for bearing not only socially useful fundamental but also noise trader risk – the risk they themselves create.
The preceding assessment of the method and implications of the Fed’s expanded intervention will leave us with a practical consideration: do we need to be more granular in our claim that prices stop reflecting fundamentals? Samuelson’s dictum says stock markets are macro- inefficient but micro-efficient – we will explore what logically enables this apparent paradox, as well as an empirical proof via dividend yields. We will then connect the aforementioned crowding of heterogeneous investors into homogeneous risky asset exposure, and associated noise trader risk, with the proposed state of Samuelson’s incorrect aggregate valuation but correct relative valuation. As investors use passive vehicles to move into riskier products despite varied risk preferences, market-weighted forms of ownership can see largest divergence from fundamentals. If the Fed reduces liquidity provision and belief in a ‘greater fool’ fades, active management should benefit from its by-design lower delta to the Fed put.
Fed intervention in secondary corporate credit markets
“If current [central bank] measures are the Himalayas, then 2008 was the Rockies and anything before that the foothills of Vermont.” – Larry Summers, April 2020
The Fed’s intervention in secondary credit markets will be managed jointly with the primary facility (PMCCF, a separate funding backstop for direct corporate debt issuance) through a structured product vehicle (SPV). The SPV is funded by a loan from the Fed secured by all SPV assets, and a $75bln equity investment from the Treasury (initially allocated 2:1 across primary and secondary facilities). It can buy corporate bonds either individually or through ETFs, for a combined cost basis of up to $750bln together with primary issuances.
On the outset, we should reflect on BlackRock’s reward and compensation– the SMCCF agreement has a four-legged fee schedule (BlackRock’s March 25th agreement for agency CMBS is solely on AUM), and while we can debate what constitutes ‘fair’ basis points on tiers ranging from $10bln to >$350bln in assets (does variable cost justify increments?), more importantly we need to consider information effects. Fed endorsements still matter, of both the things it buys and the institutions it uses to buy these things. BlackRock, as the world’s largest asset manager, benefits from both: it gets a big stamp of approval (helps BlackRock raise more funds), along with intimate and real-time influence on and knowledge of Fed activity (helps Blackrock deploy funds). While the implied winner-takes-all effect on the asset management industry makes the Fed’s choice of intermediation impactful, I’d like to focus on how the agreement hints at why the Fed thinks it needs to provide liquidity at all.
By purchasing corporate credit, the Fed is taking a qualitative position on what makes a market ‘good’ or ‘bad’. The stated goal of the SMCCF, as per Annex A, is to ensure “orderly repositioning and pricing of risk”, help issuers raise at “normalized” funding costs, and reduce incidence of “market dysfunction”. The purported divide is therefore between how things should be (orderly/normalized = good) and how they could be (dysfunctional = bad). When considering means to this end, we are confronted with messy semantics around “order” and “normality”. Do we mean orderly price discovery (price reflects fundamentals), or orderly prices (prices reflect stability and hence partial risk)? Modern Monetary Theory is popular nowadays, so in US Treasury markets, where central bank purchases have become entirely conventional, these two questions may actually be one and the same. But here we are dealing with corporate credit: real capital structure tradeoffs and debt spirals – with credit spreads supposedly reflective of that. Given the opaqueness of bond markets compared to equities, starting with the absence of centralized and regulatable exchanges, a consideration of creating “order” by improving market structure and technology doesn’t seem farfetched.
The Fed, however, cares about the pricing engine’s output, not the engine itself; as a qualitative judgement, it is labeling as “normal” funding cost levels for issuers when they do not reflect risks like pandemics. Its planned purchases of market-weighted bond ETFs therefore reveals two viewpoints: that average credit spreads are too high, but also that relative valuations are fine. While we are used to a market-based weighting mechanism from equities, we need to keep in mind that stocks and bonds are fundamentally different claims: perpetual ownership with unlimited return potential versus time-bound payment obligations with limited return potential. As a result, bond market weightings are a function of risk andissuance. ETFs like LQD (investment grade) or HYG (high yield), partially mitigated through their 3% issuer cap, are victims of this fixed payoff profile: they prioritize firms that issued more debt over firms that markets like best. Hence capital structure decisions, rather than operating asset value creation, can dominate in driving expected demand by the SMCCF. This feels counterproductive – we’re buying market values based on high leverage and low credit spreads, although we wanted to support financing of firms most impacted by corona.
Meanwhile, the agreement hints at fairly generous leverage: maximum loan-to-value ratios for the secondary purchase facility range from 67% (high yield) to 90% (investment grade) cost basis. If programs were to hit this full capacity and halt purchases, ensuing volatility can cause quick pain to the Treasury’s SPV equity stake – not an impossible outcome, given the aforementioned tendency of bond ETFs to lean into pockets of the market with highest leverage. When instead assessing the program’s reward, aside from making it easier to pitch refinancing, a higher bond price in the secondary market does not relieve a company of future interest and principal repayment burdens. The SMCCF simply floors bond investors’ price returns, putting pressure on credit spreads to reflect probability distributions of cash flows that don’t include shocks like corona. As a result, the benefit seems to be higher prices… period. The Fed signals it will be a buyer of last resort, making it possible for investors to capture privatized gains and socialized losses. We will now explore a theoretical framework on how this liquidity provision causes prices to diverge from fundamentals even in other asset classes, with broader implications on market efficiency and active vs passive investing.
Noise trader risk and investor crowding
This prospect of buying assets on the way up and selling them to the Fed on the way down creates an incentive to buy securities not because they look attractive versus fundamentals, but out of the expectation that someone else may pay an even higher price in the future regardless of fundamentals. The 1986 paper “Noise Trader Risk in Financial Markets” by De Long, Schleifer, Summers, and Waldmann models a market that reminds of this phenomenon – it finds that “Irrational noise traders with erroneous stochastic beliefs affect prices and earn higher expected returns”, because their behavior deters rational arbitrageurs who trade based on fundamentals. Arbitrageurs are assumed to be risk averse and have short time horizons. Their positioning is limited not only by fundamental risk – uncertainty on their calculated over- or undervaluation – but also by the risk that price deviations from their fundamentals- based fair value become more extreme. As a result, if bullish noise traders “overestimate returns or underestimate risk, they…earn higher average returns” in risky assets. The key result is therefore that “noise traders can earn higher expected returns solely by bearing more of the risk that they themselves create…their own destabilizing influence” (italics added).
And so, by analogy to what is discussed above, if investors believe a facility like the SMCCF will bid for securities in pursuit of ‘orderly’ prices rather than price discovery, it becomes rational to buy overvalued securities in expectation of offloading them to a ‘greater fool’ later on. They are relying on the Fed honoring the put option it thereby sells to investors, with a strike at a ‘orderly/normal’ price level. This encourages overvaluation beyond markets directly impacted by Fed purchases, given the cascading effect of targeted assets being sold and re-allocated into other higher risk assets – although admittedly with much lower realized volatility than the traditional noise trader model would predict. I can buy stocks at a high P/E multiple, because someone else will sell their investment grade bonds to the Fed and create incremental demand for stocks, such that I can later sell my stocks at an even higher multiple.
Prescient about the 35 year-old paper is the authors’ prediction on what market characteristics further magnify noise trader risk: this includes “assets of long duration which promise fundamentally uncertain as opposed to immediate and certain cash payouts”, and a higher share of passive investments as “a large fraction of investors allocate a constant share of their wealth to stocks”. Technology stocks amid corona have outperformed, with Microsoft, Apple, Facebook, and Google the S&P500’s four largest holdings at roughly 20% as of May 13th. The first two pay below-market dividends and the latter two pay none at all. And passive management in 2019 was reported to exceed half of US equity and one quarter of US bond AUM. It feels intuitive that today’s environment of passive strategies and delayed cash flow prospects is conducive to fundamental arbitrageur deterrence and noise trader price impact.
As result, central bank intervention causes Icahn’s risk “party-mobile” that is not insulated within asset classes. Instead, investors rotate across assets to retain yield with little regard to fundamentals, causing crowded ownership that used to be fragmented across a wider spectrum of investment products. And so riskier products like stocks and high yield bonds are now owned by an investor base with much more heterogeneous beliefs and institutionally and psychologically varied risk preferences, which previously solved for other allocations. As this asset-rotating population is incentivized by the Fed to reallocate exposure and ignore its risk aversion, we should expect lower conviction both on relative price convergence with and the absolute level of fundamentals. This should increase compensation not only for bearing socially useful fundamental risk but also for bearing thus created noise trader risk.
Assessing market efficiency and the future of active vs passive management
The idea that Fed liquidity encourages noise trading across asset classes, with heterogeneous investors crowding into homogeneous riskier holdings, requires us to consider if we can broadly dismiss stocks and bonds as being mispriced (whether with short- or without long- term justification), or if we need to be more granular in our claim. Specifically, we need to consider market efficiency– where prices are unbiased estimates of true value – in the context of portfolios, with potentially different predictability for single-issuer than for index products.
Simpson’s paradox applies when a relationship exists between variables in partitions of data but becomes insignificant or reverses when the data are pooled. Or, put differently: that it is possible a relationship between two variables can take the opposite sign in every possible subpopulation. For example, medicine X can be better than Y individually for category A and B – yet be worse for the union of A and B due to the lurking variable of weighting. This is important for investing, as the quest to predict returns means we may draw conclusions on a security- or sub-portfolio-basis that don’t hold/are the reverse on an index basis. “Simpson’s Paradox and Investment Management” by Bassett illustrates an active manager who beats benchmarks in every sector, but still underperforms the aggregate market benchmark.
Samuelson’s dictum reveals this partitioning problem when we test the efficient market hypothesis, claiming micro-efficiency but macro-inefficiency in equities: predictability of future excess returns is present for individual stocks but not necessarily for portfolios that they comprise. The 2006 paper “Samuelson’s Dictum and the Stock Market” by Jung and Nobel laureate Shiller cites a private letter from Samuelson, proposing that in stocks “the minority who can spot aberrations from micro efficiency can make money from those occurrences… [but we have] in no contradiction…long waves in the time series of aggregate indexes of security prices below and above various definitions of fundamental value”. Jung and Shiller test this claim by regressing future dividend growth on current dividend-price ratios. Without technical constraints, we should see “assuming the simple efficient markets model, that the slope in the regressions should be minus one and the intercept be the average return in the market”. Given technical constraints, however, the authors are mainly testing for the right sign at a statistically significant level – does a lower dividend yield indeed predict higher future dividend growth? The answer is ‘yes’ in pooled data representing the micro case, with relevant betas of -0.6 to -0.7 with statistically significant t-statistics. But when combining everything into one regression by using an equally weighted portfolio representing the macro case, the betas of 0.3 to 0.7 are suddenly positive with mostly statistically insignificant t-statistics. The authors conclude that the simple efficient markets model is the “result of a faulty extrapolation to the aggregate of a model that did indeed have some value for individual firms”; the relationship between fundamentals and price thus provably holds, paradoxically, depending on whether or not we look at the whole portfolio or its constituents.
It’s intuitively appealing that the ability of ‘rational arbitrageurs’, as they are called in the Noise Trader paper, to assess fundamental value and monetize it against market prices should be a function of scale. Of course this does not mean individual stocks are free of anomalies – factors from small-vs-large to betting-against-beta are well documented. But more pertinent here is the relative ease with which investors, through passive, market-weighted vehicles, can contribute to aggregate valuation without changing relative valuation. When a retail investor sells LQD to the SMCCF and increases risk exposure by rotating to SPY (S&P500 ETF), she can retain S&P500 micro-efficiency while contributing to macro-inefficiency. We should therefore feel more confident in valuation differentials of Target versus Walmart compared to small-cap stocks versus investment grade bonds. This aligns Samuelson’s dictum with impact of measures like the SMCCF and forces us to reconsider claiming sweeping inefficiency.
In this light, the micro-macro paradox has important implications for the active versus passive debate. We established that market-weighted index products are more susceptible to noise trader risk, as they allow prolonged, self-sustained price deviations from fundamentals. ETF providers are thus among greatest beneficiaries of the ‘Fed put’, given ETFs are such a convenient product for yield-seeking investors who, as discussed earlier, have varied risk preferences they are now incentivized to ignore. If, however, the Fed stops pressuring yields and investors lose faith in a ‘greater fool’ ready to buy down the line, then these opportunistic investors may quickly backtrack and restore asset allocations to reflect their true risk appetite. It also seems reasonable that this will put active management back into the spotlight, as the absence of Fed support will by above logic be felt strongest in market-weighted portfolios, something that active management by design avoids. The catalyst for such an unwind – the Fed put disappearing – is exogenous to markets and thus does not require a change in fundamentals. We therefore would expect a return to less crowded allocations that provide better matching of investor preferences with assets’ QE-free risk-return profiles, whereby any resulting change in valuations will still be consistent with Samuelson’s dictum.
Conclusion
On May 8th, the WSJ listed the S&P500 P/E ratio at 25x TTM versus 22x exactly a year ago. While this can partially be explained by the effective Fed Funds rate collapsing from 2.39% to 0.05% during this period, it also feels counterintuitive to pay more for past earnings even though the risk of a global pandemic changed from distant fear to harsh reality. And so, the divergence of risk asset valuations (with notable exception of commodities) and the real economy has generated much debate in the media. Populist reporting often suggests that the Fed’s forcing function is controlled by the right tail of the wealth distribution, which owns the lion’s share of financial assets. It was my goal to try and develop a framework that spells out why mechanically, and perhaps rationally, we can see a sustained rise in valuations no matter if the economy grinds higher like 2010-2019 or is brought to a standstill like in 2020. While my treatment of investor beliefs and preferences is largely stylized, it tries to depict how Fed efforts to suppress yields – now extended by the SMCCF – generate a rotation up the risk spectrum and incentivize investors to ignore their risk aversion, creating value they themselves can capture. This seems aligned, by crude measure, with micro-efficiency and macro-inefficiency. It also implies that any reversal in the ‘Fed put’ can position active management to temporarily regain ground lost to passive management in the past decade.
“Investment Management Agreement (Secondary Market Corporate Credit Facility)”
https://www.newyorkfed.org/markets/secondary-market-corporate-credit-facility
“Noise Trader Risk in Financial Markets”
De Long, Shleifer, Summers, and Waldmann
“Simpson’s Paradox and Investment Management”
Bassett
“Samuelson’s Dictum and the Stock Market”
Jung and Shiller
Synthetic Friday 