Uncommon Truths Tight But Loose

Uncommon truthsTight but looseIs the Fed too loose or too tight? Recent marketbehaviour suggests there is a belief that it is tootight. However, we continue to believe it is veryaccommodative and find support from the TaylorRule (even when adapted to the reality of the Fed).We suspect Fed rates and bond yields will rise.Those familiar with Led Zeppelin will recognise the titleof this document. Tight but loose could also be used todescribe the collective viewpoint about the state of Fedpolicy. The market implied future path of Fed rates wasnever as aggressive as that suggested by the Fed’sown dot plot, thus suggesting a common belief thatcurrent policy rates were not too far from whatconstituted “normal”. Indeed, those market impliedrates now suggest a belief there will be a rate cut thisyear. However, our own feeling remains that, at currentpolicy settings (Fed Funds upper target rate of 2.5%),the Fed remains extremely accommodative.The analysis has been complicated by the recentsoftening of the Fed’s stance, witnessed by recentpost-meeting statements: notably the emphasis ofpatience in deciding the next change in policy rates, thelowering of the future path of rates (as described by adot plot suggesting no change in rates this year) andthe decision to end quantitative tightening bySeptember. Such a change in tone seemed veryunlikely just six months ago.Any discussion about the appropriateness of thecurrent level of interest rates depends upon somenotion of a long-term norm. Many economists refer tothe notion of R* (R-star), which is basically the real rateof interest that equilibrates the economy in the long run(sometimes called the natural rate). Simple, apart fromthe practical issue of deciding what it should be! Forreasons that I have long forgotten, we were taught thatthe equilibrium real rate of interest is in line with realeconomic growth, which is of course a shifting target.Prior to the global financial crisis, we may have beenconvinced that US real economic growth was normallyin the 3.0%-4.0% range (measured up to 2007,annualised growth was 3.1% over 10 years, 3.1% over20 years, 3.2% over 30 years, 3.4% over 60 years and3.8% over 200 years). However, measuring backwardsfrom 2018, those growth rates become 1.8%, 2.2%,2.5%, 3.0% and 3.8%, respectively. Apart from thevery long-term analysis, the judgement about what is anormal growth rate is very much impacted by start andend-dates.Of course, the future may not look like the past and oneimportant difference versus the post-WW2 era will bepopulation growth. UN estimates suggest theannualised working age population growth in the USwas 1.2% in the 1950-2015 period but that it will beonly 0.2% in the 2015-2100 period (using 20-64 yearsas the definition of working age). That implies quitesome slowdown in economic growth, which is of coursecritical in our attempt to define R*. Interestingly, theformulation of the Taylor Rule implicitly assumes thatR* is 2%, which may be a good reflection of thediminished growth potential (see Figure 1).Figure 1 – Fed Funds Rate: actual and according to Taylor Rule (%)Note: quarterly data from 1990 Q1 to 2018 Q4. The Taylor Rule states that r p 0.5y 0.5(p-2) 2 (r is the federal funds rate, p is the rateof inflation and y is the GDP output gap). The amended version uses a coefficient of 1.0 on y (output gap), as suggested by Ben Bernanke in“The Taylor Rule: A benchmark for monetary policy?”. Source: CBO, BEA, Datastream and Invesco.01 April 2019For professional/qualified/accredited investors only1

The basic Taylor Rule (developed by John Taylor)suggests that when the economy is in equilibrium(output in line with potential and inflation in line with theFed’s policy target of 2%), the Fed’s policy rate shouldbe in line with nominal GDP growth (if real GDP growthaverages 2%). The Taylor Rule policy rate then variesas output varies from its potential rate (the output gapis non-zero) and inflation differs from the Fed’s 2%target. Put simply, the rule states that the Fed shouldraise rates by 50 basis points above the equilibrium(R*) level for each percentage point of positive outputgap (GDP exceeding potential GDP) and also for eachpercentage point that inflation exceeds the Fed targetof 2%. The same adjustments are to be made inreverse.On the basis of this “standard” Taylor Rule formulation,Figure 1 suggests the Fed was too “tight” (rates toohigh) during much of the 1990s but was too “loose” inthe run-up to the financial crisis (in the early 2000s) andfor most of the post-crisis period. It is currentlysuggesting that Fed rates should be around 4.25%,well above what the market and the Fed now believe(the Fed’s current dot plot suggests the long-run neutralrate should be around 3.00%).Is it possible that the Fed has been running so “loose”for so long after the financial crisis? In a 2015 paper,Ben Bernanke suggested that Fed policy makers havehistorically been more sensitive to swings in the outputgap than suggested by the Taylor Rule. He suggestedusing a coefficient of 1.0 on the output gap, rather thanthe standard 0.5, and we show this amended version inFigure 1 (he also suggested using the core PCEmeasure of inflation, rather than the GDP deflator,which we have done throughout).Fed policy seems to have been closer to this“amended” version over recent decades. This versiondidn’t suggest conclusively that the Fed should startraising rates above the 0.25% post-crisis level until thesecond-half of 2014 (the Fed waited until December2015). That the amended rate was well below zeroduring the financial crisis was the justification for theuse of non-conventional tools to support the economy.Unfortunately, this amended version now suggests Fedrates should be around 4.8% (the amended rate is nowhigher than the standard rate because the output gap ispositive). This again suggests that Fed policy isextremely easy at the moment, judged by the level ofinterest rates.However, many investors would argue that the Fed hastightened by more than the 2.25% rise in policy ratessince December 2015 because it has shrunk itsbalance sheet. We agree but would argue that weshould also consider the loosening done during thequantitative easing (QE) phase. Figure 2 shows ourregular attempt to do this by constructing a syntheticpolicy rate that adjusts the actual policy rate forchanges in the size of the Fed’s balance sheet (using arule provided by the Fed itself). Our synthetic policyrate touched a low of -4.9% during 2015 and is nowaround -1.9%, suggesting a total tightening equivalentof a 3% rise in policy rates since 2015.So, Fed policy rates do underestimate the extent of Fedtightening in recent years but whether we look at theactual or synthetic policy rates, they are well belownominal GDP growth, which is where they should bewhen the economy is in equilibrium. Whether we statethis fact using the Taylor Rule or express it moresimply, we believe Fed policy rates are too low.Figure 2 – Fed policy rates and US nominal GDP growth (%)From July 1954 to March 2018. Synthetic policy rate is the actual policy rate adjusted for Fed asset purchases using the rule that each 150bn- 200bn of assetpurchases is equivalent to a 25bp rate cut. Source: Datastream and Invesco.01 April 2019For professional/qualified/accredited investors only2

However, there is one nagging doubt: if the Fed hasbeen too loose over a period of at least five years,where is the evidence in terms of economic or financialmarket behaviour? From an economic perspective, M2money supply growth of around 4% is hardly indicativeof a credit boom (though that could be due to supplyfactors, rather than demand). Further, it seems to usthat both the housing and auto cycle have, at best,flattened and may have turned lower. Again, hardlysuggestive of an economy being inflated by an overlyaccommodative central bank.Hence, though the Fed’s accommodative policies seemto have had little impact on consumer (and investment)spending, they do seem to have had an effect onfinancial markets, either directly via debt purchases orindirectly via corporate sector financial engineering.Both the Fed and financial markets have adopted amore cautious attitude in recent months (as have we –see Dialling down the risk) and there has been anoticeable softening in US economic data, both surveyand real. No matter what we or the Taylor Rule sayabout Fed rates, we expect the US central bank to cutThere are nevertheless signs of excess in the corporate rates if there is a hint of recession (and perhaps buysector, especially when it is considered that corporateassets again if the recession is deep).sector debt is at a record high relative to GDP (in themidst of an economic expansion) and that most of theWe do not expect recession this year, believing that thenew debt in the last 10 years seems to have goneUS economy is taking a mid-cycle break. Rather, wetoward share buybacks (see our earlier piece Shouldexpect recession sometime in 2020/21, which wewe be worried about US corporate debt).believe gives the Fed scope to raise rates closertowards what the Taylor Rule and our intuition tells usThis has its counterpart in financial markets with a USwould be more reasonable at this stage of the cycle.equity market that seems very expensive to us, not inWe suspect the psychological damage caused duringrelation to today’s profits but in relation to what wethe financial crisis (to the Fed and the rest of us) willmight consider a normal level of profits (for instance, aprevent policy rates rising to the Taylor Rule levelShiller PE above 30 is worrying).during this cycle but nonetheless believe there isupside. On that basis, we expect a reversal of theFinally, the Fed has had a direct effect on fixed income recent decline in bond yields later this year asmarkets: by changing the balance of supply andeconomic surprises again turn positive.demand over the last 10 years, we think it has distortedtreasury yields lower (a real 10-year yield of 0.6%Unless stated otherwise, all data as of 29 March 2019.seems well below the economic growth that mostcommentators would expect over the next decade).01 April 2019For professional/qualified/accredited investors only3

Multi-asset researchUncommon truthsFigure 3 – Asset class total returnsData as at 29/03/2019EquitiesWorldEmerging MarketsUSEuropeEurope ex-UKUKJapanGovernment BondsWorldEmerging MarketsUS (10y)EuropeEurope ex-UK (EMU, 10y)UK (10y)Japan (10y)IG Corporate BondsGlobalUSEuropeUKJapanHY Corporate BondsGlobalUSEuropeCash (Overnight LIBOR)USEuro AreaUKJapanReal Estate (REITs)GlobalEmerging MarketsUSEurope ex-UKUKJapanCommoditiesAllEnergyIndustrial MetalsPrecious MetalsAgricultural GoodsCurrencies (vs USD)*EURJPYGBPCHFCNYIndexCurrentLevel/RY1wTotal Return (USD, .5-8.8-4.0-7.1-3.9-6.3-----12mTotal Return (Local Currency, %)1w1mQTDYTD12mNotes: *The currency section is organised so that in all cases the numbers show the movement in the mentioned currency versus USD ( veindicates appreciation, -ve indicates depreciation). Past performance is no guarantee of future results. Please see appendix for definitions,methodology and disclaimers.Source: Datastream and Invesco01 April 2019For professional/qualified/accredited investors only4