Two performance statistics almost everyone knowsFirst, a Happy New Year to those who have been following these posts and articles over the past year.As 2022 ended, markets closed December much as they had spent last year – with yields higher and equity prices lower. This pattern cemented a 2022 return pattern with two dismal features familiar to almost every investor. One was that every major market except Cash, Commodities and the US dollar delivered negative returns (chart 1). The other was that a balanced 60/40 portfolio of Equities and Bonds delivered its worst returns in almost a century (chart 2). This article is about a long-term issue relevant to the next 12 months: mean reversion in risk premia, and what today’s starting levels imply about year-ahead returns for Bonds, Credit, Equities, Currencies and Commodities.I rarely write about valuations as a stand-alone factor, because markets respond to a balance of several forces. But at the start of the year, knowing where risk premia are in long-term perspective is useful for determining markets’ biases in the year ahead. (January 2022, for example, presented unusally low risk premia in all major markets but Commoditise, as evidenced by negative real yields, below-average credit spreads and above-average equity multiples). After a year of Fed-driven repricing, these same indicators plus a recession baseline favour Bonds over Cash; only some parts of Credit (MBS, High Grade) over Bonds and High Yield; Fixed Income over Equities; a lower trade-weighted USD; lower Oil but higher Gold. Chart 1: As occurred during stagflations of the 1970s/early 80s, 2022’s best performers were Cash, Commodities and the US dollar. 2022 returns. Source: YCharts.Chart 2: 2022’s 18% losses on a 60/40 portfolio were the worst since 1937. Annual returns on 60% US Equities/40% US Treasuries portfolio. Shading indicates US recessions. Source: Aswath Damodaran (New York University).Mean reversion through feedback loopsIf investors had no more information on hand but 100 years of return data, they would be reasonable in expecting a much better outcome in 2023. In the past century, Bonds have never posted three consecutive years of losses, and have only delivered back-to-back annual declines three times (1955-56, 1958-59, 2021-22). This statistical pattern owes to the casual dynamic of a feedback loop, as higher yields trigger a recession and an eventual Fed policy reversal. US High Grade Credit has only posted two consecutive annual losses on one occasion (1979-80), as it benefits after Bonds from a Fed reversal. Multi-year losses in Equities are somewhat more common (1929-32, 1939-41, 1973-74, 2000-02), however, partly due to the variable length of recessions and recessions’ impact on earnings. Multi-year stretches of Emerging Markets underperformance are even more common, as this asset class confronts both DM spillovers and its organic crises (1997-98 Asia Crisis, 2013-15 EM Credit Crunch). Commodities like Oil rarely post more than two consecutive years of gains (1994-96, 2003-07), again given a feedback loop from high price to demand destruction and sometimes higher supply.The causal channel matters as much as the statistics, so the next five sections provide macroeconomic context for mean-reversion concepts that should inform the five most important allocations that investors make in public markets: (1) Cash vs Bonds; (2) Bonds vs Credit; (3) Bonds vs Equities: (4) USD vs non-USD currencies (a proxy for DM vs Emerging Markets); and (5) Commodities (real assets) vs nominal assets.Bonds: Average risk premium, outperformance vs Cash on Fed rethinkFor Bonds, the most common measures of risk premia are real yields (yield premium over expected inflation) and the term premium (yield premium over cash). Current US 10Y real yields near 1.5% are very high by the standards of the past decade, when QE suppressed these to an average of only 0.5% per annum (chart 3). To appreciate the extent of financial repression over the past decade, note that for 30 years prior to the QE era (1980 to 2008), real rates generally tracked the trend rate of real GDP growth (chart 3). Thus, US real yields would have averaged 1.5% to 2% over the past decade (rather than 0.5%) had it not been for the distortive effects of central bank asset purchases.This suppression of the real risk-free rate, in turn, fuelled the mainstreaming of several other previously non-core asset classes (EM Fixed Income in 2009-13, Private Equity throughout the 2010s, innovation assets from about 2017-2021), some of which have experienced spectacular deratings as the Fed normalised real rates. Current real yields of 1.5%, which is near most estimates of US potential GDP growth, represent fair value given the long-term relationship between cyclical momentum and the cost of capital. For Bonds to be cheap, real rates would need to trade meaningfully above trend growth (so a real yield of 2% or more), as they did in the 1980s after the Volcker interest rate shock.Chart 3: Before the QE era, US real 10Y yields generally tracked the trend rate of real GDP growth. Real US 10Y yields versus US real GDP growth year-on-year. Real yields calculated from TIPS since 2003 and from nominal yields minus headline PCE inflation before 2003. Shading indicates recessions. Source: Federal Reserve. Positive real yields are not a sufficient condition to guarantee Bond outperformance versus Cash in 2023, however. The scatter plot highlights the relationship since 1960 between starting real yields in January of each year and realised outperformance of Treasuries versus Cash over the following 12 months. The higher the starting real yield, the greater the outperformance from owning duration. The model fit is weak, but nonetheless implies that owning the 10Y will generate about the same return as owning Cash, so about 5%. The constraints on Bonds beating Cash this year are fairly obvious: cash rates are high (4.25% – 4.5% Fed funds) and will rise further in Q1 (to about 5%), while the Fed funds-10Y curve is inverted. So unless the money market curve prices another Fed ease for late 2023, Cash and Bonds will generate comparable returns.annual data since 1960I think this repricing is likely this year as the following sequence of events begins: corporate profits begin showing a year-on-year decline during the January earnings season, companies begin shedding labor mid-year to restore profit margins, and the US economy enters a mild recession in H2. So to take exposure to this baseline, I would rather own duration than cash even when the curve is inverted. (See also these posts on the duration trade from October, November and December.) I realise that this view is now widely held across banks and asset managers, judging from the number of 2023 outlooks carrying taglines like “Bonds are Back”, “Bonds are a Buy” or “The Year of Income”. Despite this tone, some surveys suggest that investors are more overweight of Cash than they are of duration (chart 4). Chart 4: Individual investors have accumulated Cash as they sold Equities this year, but they have not increased their Bond allocation yet. Percent allocation to Cash, Bonds and Equities in monthly AAII survey. Shading indicates recessions. Source: American Institute of Individual Investors.Credit: Low-to-average risk premia, so a higher-yield duration option on a mild recessionBased on valuations alone, it is difficult to rationalise Credit outperformance versus Bonds this year. In 2022, spreads indeed widened across the curve (High Grade and High Yield) and across regions (DM and EM), but the only credit premia that reached recessionary levels by Q4 were US 30Y mortgages, UK High Grade Credit and 10Y Italy. By comparison, most other corporate benchmark levels – US High Grade at 140bp, US Baa at 200bp, US High Yield Spreads at 460bp and EM Corporates at 300bp – trade at half to two-thirds of their typical recession peaks (chart 5). It’s true that credit fundamentals are somewhat unique in that corporate leverage (non-financial debt to GDP) is unchanged during the current expansion while household indebtedness has fallen over the past two business cycles. But corporate balance sheets were also improving in the 1970s and household ones in the early 1980s. That feature did not prevent Credit from underperforming Bonds during recessions in those decades, given the income stress created by higher rates, contracting earnings and rising unemployment.Chart 5: Spreads have widened across the credit curve and across regions, but only a few markets (US 30Y MBS, UK High Grade, Italy 10Y) trade at recession levels. Spreads to Treasuries (basis points) for US High Grade, Baa corporates, US 30Y MBS and EM Corporates, Source: Federal Reserve, YCharts. The scatter plot highlights the breakeven problem for parts of Credit in 2023: current spreads of 140bp on a High Grade, 200bp on Baa’s and 480bp on High Yield imply flat-to-negligible outperformance vs Bonds this year. So for a buy-and-hold investor for the next 12 months, the approach to credit needs to be quite differentiated: only own the cheapest sectors like US MBS and UK High Grade that have already discounted a recession.annual data since 1960Other sectors (US High Grade) are worth owning more for optionality on the possibility of a very mild recession in which spreads widen no more than about 50bp, which would be offset by a rally in underlying Treasuries. I don’t see value in owning spread products any riskier than High Grade given current valuations and by baseline view for a mild recession.Equities: Because Bonds cheapened too in 2022, Equity/Bond relative value has not improvedSome claim that Equities offer good value entering 2023, because they have posted one of their largest-ever declines from peak, before the feared recession has actually begun. This view is reasonable on a single-name basis but not at the index level, for two reasons. One is that broad indices’ maximum drawdown in 2022 was only ever consistent with the mildest of previous recessions, so carried no risk premium for uncertainty over the duration and depth of the next recession. The second is that such statements do not address the relative attractiveness of Equities versus Fixed Income, which also cheapened in 2022.Charts 7 traces the relative value question through the standard comparison of stocks’ earnings yield (inverse of the P/E ratio) to the real US 10Y yield. So while it is true that Equities have cheapened in absolute terms through a fall in P/E ratios (Shiller trailing measured used below to extend the sample period to 1960s), the attractiveness of Equities relative to Bonds has declined because of the spike in real US 10Y yields. The same dynamic occurred during the Volcker rate shock of the early 1980s: Equities cheapened in absolute terms but become more expensive relative to Bonds.Chart 7: Oddly, Equities have become more expensive relative to Bonds despite a drop in P/Es, because US real yields have spiked. US earnings yield (inverse of P/E ratio) based on Shiller trailing measure versus US real 10Y yield. Shading indicates US recessions. Source: Robert Shiller (Yale University), Federal Reserve.Unfortunately, the earnings yield to bond yield ratio is a very poor predictor of year-ahead return differentials between these two markets. The poor fit owes partly to the multi-year nature of previous equity market bubbles, like that of the late 1990s. There is, however, much more consistency to the behaviour of Equity vs Bonds during recessions. Three features are notable: (1) Equities tend to fall about 30% from peak around a recession; (2) they bottom about two-thirds of the way through a recession, around the time that manufacturing surveys like the ISM and PMI inflect higher (chart 8); and (3) they recoup most of their recession losses within about a year of troughing.I won’t be stickler for each of these criteria being met before changing the negative view on Equities that I have had for much of 2022, because I didn’t see a path to inflation normalisation that could avoid a Fed-induced recession. (Balance sheets weren’t the issues; the inflation overshoot was.) But given how much Equities have already derated and how underweight some positioning metrics have become (for institutional investors, not retail ones), I would own Equities versus Bonds on cheap valuations alone, if S&P500 drawdown reached 30-35%. Current drawdown from peak is about 20%. Chart 8: As a timing tool for buying the lows in Equities, an inflection higher in the ISM or PMI manufacturing index is quite reliable. S&P500 returns year-on-year versus level of ISM manufacturing survey. Source: YChartsCurrencies: Near-record expensiveness heading into a recessionMost investors acknowledge that that US dollar is expensive on all standard measures, like the real exchange rate’s level versus long-term average, the trade-weighted index’s level versus the US terms of trade, and pairwise valuations versus interest rate differentials. The degree of overvaluation has reduced over the past three months given the trade-weighted index’s 5% drawdown from its peak, but the misalignment persists. Think of this overvaluation as a risk premium on non-US economies, either for their vulnerability to the Ukraine/Russia war or tighter financial conditions globally.As shown in chart 9, the trade-weighted index is a reasonably mean-reverting asset, where peaks in the currency typically follow peaks in Fed funds. The problem is that the lag between a peak in Fed funds and a peak in the dollar are not uniform. Volcker’s early 1980’s hiking cycle ended in 1981, but the dollar did not turn lower until 1985. The lags between monetary policy and currency turning points were shorter after tightening cycles ended in 2000, 2006 and 2018, but oncoming recessions meant that the dollar did not peak in trade-weighted terms for another 18 months at least, even if leadership within the index did rotate.Chart 9: USD peaks follow Fed peaks, but with sometimes long and variable lags. USD real effective exchange rate versus Fed funds rate. Shading indicates recessions. Source: Federal Reserve.The problem with extrapolating this currency pattern into 2023’s Fed peak is that the dollar has never entered a recession so strong (chart 9), and its expensiveness is particularly pronounced versus the yen (bar chart below).Mechanically, a dollar this strong at the beginning of the year tends to be followed by a dollar decline (scatter plot), in part due to a negative feedback loop from a strong currency to expectations of Fed easing. I don’t expect the Fed to ease this year, but I do think that 10Y rate differentials between the US and Europe plus Japan will narrow as the ECB and BoJ play catch-up.annual data since 1970Neither Europe nor Japan are in recession, despite the terms of trade shock from energy prices. Yet both the ECB and BoJ continue to maintain negative real policy rates (cash rates minus core inflation), which look increasingly anomalous. Another 100bp of hikes from the ECB and another 50bp increase in the BoJ’s 10Y yield target should be sufficient to narrow bond spreads given how advanced the Fed cycle is. And if EUR and JPY are slightly higher in 2023, it will be difficult for the trade-weighted dollar to make a new cycle high. A broader and more sustained mean reversion lower in the dollar will require Fed easing, which won’t come until 2024. Hence why investors can afford to rotate gradually out of USD and into non-USD currencies over the next year or two.Commodities: Structural or not, commodity price cycles always succumb to the business cycleCommodities were a consensus overweight since 2021, both outright and via proxies in Equities and Currencies, for anyone familiar with the stagflation playbook from the 1970s and early 1980s. How much Commodities bulls were able to monetise this view in 2022 is debatable, because some supercycle pushers seemed to ignore subsequent chapters in the stagflation playbook. Those later lessons concern the diversification failure of Commodities once high price trigger demand destruction and price reversal in resources most sensitive to the business cycle, like Energy and Base Metals. Hence the need to rotate into defensives like Gold.I don’t question the ability of Commodities to generate extraordinary returns in the early, mid and late phases of economic expansions, because I understand the volatility implications of short-run inelasticity in supply and demand. Elasticities, realised returns and volatility are just empirical points. But I have never been much the fanboy of supercycle theses positing some indefinite period of extraordinary price gains, because I recognise that a macroeconomic feedback loop (commodity price surges, CPI overshoot, restrictive monetary policy, recession) has been aligning commodity price cycles with the business cycle for at least the past 50 years (chart 10). Given this dynamic, forecasts for indefinite periods of extraordinary price gains are somewhat incoherent, as what makes the price cycle extraordinary is also what makes is unsustainable. Indeed, the mid-2020 to mid-2022 commodity price upswing has been the shortest ever, precisely because the monetary response to the CPI overshoot is threatening to make this business cycle one of the shortest of the past half-century. This article from November discusses the supercycle dynamic in more detail.Chart 10: Most commodity price upswings last 2 to 3 years; the current cycle’s rally, which ended in June, was the shortest ever. Inflation-adjusted commodity prices, indexed to 100 in 1970. Shading indicates recessions. Source: EIA, IMF.Because expensive commodities are vulnerable to mean reversion lower via demand destruction, a useful cross-check on 2023 price forecasts is current valuations. As the scatter plot shows, the Oil price could have about 10% downside (to about $70/bbl on Brent) since the price in real terms is above average but not by much (less than a standard deviation). So there is a modest risk premium in energy markets for supply stress, but no risk premium in these markets for recession.annual data since 1970This result will disappoint the supercyclists, who believe that underinvestment in carbon-based energy during the net-zero transition will generate above-average long-term prices and recurring price spikes. Their long-term view is correct, but energy prices are no longer cheap on long-term measures (as they were in 2020 and 2021), and demand destruction is usually more acute than the supply response. So the more reasonable way to reconcile short, medium and long-term perspectives is the forecast lower-than-average Oil price declines in 2023 rather than price resilience, and to focus instead on defensive Commodities like Gold that will benefit from falling real yields and a weaker trade-weighted dollar. #2023Outlook #riskpremia #meanreversion #assetallocation #Bonds #Credit #Equities #Currencies #CommoditiesLinks to previous articles on global asset allocation, geopolitics and sustainable investingLate-2022 volatility compression, yesterday’s Fed realism and 2023’s labor market reckoning (Dec 15, 2022)The Fed’s next moves are obvious, but 2023 remains a duration play (Dec 5, 2022)China’s COVID protests and the super-short Commodities supercycle (Nov 28, 2022)Presentation: Working in Finance, Investing in Macro Markets & Applying Philosophy to ESG (Nov 21, 2022)Trump’s back for 2024 — thin US voting margins but higher China risk premium (Nov 16, 2022)Cryptocurrencies, fraud and philanthropy (Nov 13, 2022)Further evidence that 10Y bond yields have peaked, and what that path means for other markets (Nov 11, 2022)Meta (Facebook) as a victim of weak governance rather than just high interest rates (Nov 6, 2022)A week ahead of the next UN climate summit, not a single indicator of progress towards netzero is on track (Oct 26, 2022)US mid-term elections, the RedWave and implications for 2023 & 2024 (Oct 25, 2022)What to do with the UK’s cheap markets (Oct 18, 2022)Bonds are worth owning, even after yesterday’s strong US CPI report (Oct 14, 2022) Louisiana dumps Blackrock for ESG investing – the strong and weak points in the Treasurer’s arguments (Oct 6, 2022).Companies that left Russia for the right & the woke reasons (Oct 2, 2022)What to do with the second-strongest US Dollar ever (Sep 30, 2022)