Total assets down $30Bn: the biggest weekly decline since May 1, 2019, and down $27Bn from peak on Jan. 1, 2020. All of the decline is on the back of repos, down $43Bn on the week, and 70Bn from peak. At $186Bn, repos were last time here in mid October 2019.
On the liability side, bank reserves declined by $64Bn. But the liquidity dropped more because both the TGA and domestic reverse repos rose by $31Bn and $15Bn respectively. At $412, TGA is at its highest level over the last 12 months. On the other hand, and as expected, FRP continues to decline and at $250Bn is close to the low end of the 12m period. Currency in circulation also dropped for third week in a row, posting the biggest cumulative decline from its peak over the last 12m. The decline in FRP and currency in circulation cushioned the otherwise drop in overall liquidity.
Going forward, there is no doubt that the bulk of the central bank’s increase in balance sheet is behind us for the moment, ceteris paribus. The Fed will continue shifting from repos to T-Bills and probably coupons (especially if it hikes the IOER/repo rate next week). The effect on liquidity will depend on the liabilities mixture, though. Expect TGA to slowly start decreasing ($400Bn has kind of been its upper limit, rarely going above it by much).
FRP has a bit more to go on the downside but I think it will struggle to break $200Bn, probably settle around $215Bn.
That should help liquidity. If the Fed buys more securities than the decline in repos, under that scenario, bank reserves/liquidity go up. If not, it really depends on the net effect of the change in autonomous factors.
If you are trading Fixed Income, expect a bit more pressure on the curve to continue flattening. If you are trading equities, none of this matters to you. At the moment, the only thing the equity market cares about is the size of the gamma cushion.
I am late in this debate, at least in writing, because at first, I thought it did not matter; it is all semantics. Last week I read John Authers’ article in Bloomberg in which he referenced a chart from CrossBorder Capital that showed that the Fed had recently injected the greatest liquidity boost ever. That got me really curious, so I did some digging in the Fed’s balance sheet and I concluded, notwithstanding that I am not privy of how CrossBorder Capital defines and measures liquidity, it is unlikely that the Fed’s actions led to the ‘greatest liquidity boost ever’. And then yesterday Dallas Fed President Kaplan said he was worried about the Fed creating asset bubbles. This pushed the ‘old’ narrative that CBs’ liquidity/NIRP/ZIRP is creating a mad search for yield and a rush in risky assets out of the woodwork again on social media. So, that got me thinking that whatever the Fed did since last September, whether it is QE or not, actually matters.
So, just to refresh, since September 2019, the Fed’s balance sheet increased by about $400bn, of which more than half came from repos, the other from mostly T-Bills, with the increase in coupons more than offset by the decline in MBS. On the liability side, there was a similar breakdown: about 50% came from an increase in bank deposits, the other 50% came from an increase in currency in circulation and the TGA account. This 50/50 in both assets and liabilities is important to keep in mind.
During QE1, the increase in securities held was more than 3x the increase in Fed’s total assets. That was mostly because loans and CBs swaps declined to make up the difference. On the securities side, the Fed bought both coupons and MBS. T-Bills remained the same, while agencies declined. However, 75% of the increase in assets came from a rise in MBS (from $0 to almost $1.2Bn). The Fed had begun to extend loans to some market players even before September 2008, but immediately after Lehman Brothers failed, the Fed extended loans to primary dealers (PD) as well as asset-backed/commercial paper/money market/mutual fund entities to the tune of about $400Bn. These were very temporary loans, pretty much making sure that no other PD or any other significantly important player failed. By the time QE1 finished the loans had gone back to almost pre-Lehman-time sizes. In a similar fashion, the Fed had already put in place CBs swaps even before September 2008, but immediately thereafter, the CB swap line jumped to more than $500Bn, and by the time QE1 finished it had gone to $0. Finally, repos actually decreased during QE1. Bottom line is, as far as Fed’s assets are concerned, September 2019 had absolutely no resemblances at all to September 2008.
On the liability side, the differences were also stark. Unlike 2019, during QE1 bank reserves contributed to 95% of the increase. The FRRP account remained pretty much flat for the full duration of QE1, while the TGA account was unchanged but it did exhibit the usual volatility during seasonal funding periods.
QE2 was much more straightforward than QE1. The Fed’s assets increased only on the back of coupon purchases (around $600Bn), while the Fed continued to decrease its MBS and loans portfolio. On the liability side, bank reserves continued to contribute about 95% of the increase. The rest was currency in circulation. Bottom line here again, really no resemblance to 2019.
QE3 was similar in the sense that Fed’s reserves increased 100% on the back of securities purchases (around $1.6Tn), but this time split equally between coupons and MBS. On the liability side, at 80% of total, bank reserves contributed slightly less towards the overall increase. The rest was split between currency in circulation and reverse repos. During QE3, unlike QE1 and QE2, less of the Fed’s balance sheet increase went towards higher liquidity (bank reserves), but still nothing like in 2019. For one reason or another, the market was willing to give some of the liquidity back to the Fed in the form of reverse repos even before the Fed started tapering (reverse repos were prominent after QE3 when the Fed stopped growing its balance sheet but before it actually started tapering it).
No, you can’t call whatever the Fed has been doing so far, starting in September 2019, QE. There are simply no comparisons with any of the previous QEs: The largest increases on the Fed’s balance sheet in 2019 was T-Bills and repos; the Fed never bought T-Bills or engaged in repos in any of the previous QEs – the asset mix was totally different. On the liability side, while in the QEs almost all of the increase went directly into bank liquidity, in 2019 only 50% did. FRRP was more or less unchanged, at around $100Bn between QE1 start and the end of QE3 – by September 2019 it had tripled! TGA averaged around $60Bn before the end of QE3; thereafter the average increased 4x!
As to the second issue of how much of the Fed’s liquidity injection since the crisis has boost asset prices? Not much.
According to the Fed’s own flow of funds data, real money has been a net seller of equities and buyer of risk-free assets since the 2008 financial crisis. If there is a rush into risky assets, it is not obvious from the data. There is also this argument that the Fed’s consistent boost of liquidity, combined with low interest rates, provides the proverbial put for prices and, therefore, the search for yield can be implemented by selling vol/gamma. This could indeed be the case. The problem is that I have not seen any data which shows exactly what the $ notional (in cash equities) equivalent of that vol selling flow is.
Moreover, given that both ECB and BOJ have engaged in even bigger balance sheet expansions, plus their interest rates are negative, the case could be made for a similar exercise in Europe and Japan. However, both European and Japanese equity markets have been languishing for years, underperforming US equity markets. Finally, even if this indeed were the case, the more likely explanation for the reasons people would be selling vol is the relentless bid from corporates engaging in share buybacks. This would also explain the underperformance of equity markets abroad relative to US ones despite higher CBs’ liquidity boosts there.
But how much liquidity did the Fed provide since the 2008 financial crisis?
Equities bottomed in March 2009. Fed’s assets increased by about $2Tn thereafter. But only 37% of that increase went to bank reserves. 40% went towards the natural increase of currency in circulation, 14% went to TGA and 9% went to the FRRP (drawing liquidity out). It is slightly better if one does the comparison since QE1, but even there, at most, 50% went directly to bank reserves.
Finally, one has to take into account that banks’ reserves needs have also substantially increased since the 2008 financial crisis on the back of Basel III requirements. According to the Fed itself, the aggregate lowest comfortable level of reserve balances in the banking system ranges from $600Bn to just under $900Bn. At $1.6Tn currently, there is not much excess liquidity left in the system. In fact, banks Fed deposits were already at around $800bn in March 2009. Given that most of the regulations were implemented thereafter, one could claim that no additional liquidity was really added to the banking system since.
The market has been predicting the coming collapse of China ever since it joined the WTO in the early 2000s and people started paying attention to it*. The logic being, with the collapse of the Soviet Union in 1989, China would be next: the free market must surely assure the best societal outcome. But with the re-emergence of China thereafter, and, especially, that it is still ‘going strong’ now, on top of the slowdown (‘secular stagnation’ or whatever you want to call it) of the developed world, I think the verdict of what the best model of resource optimization is, is still out.
The Soviet models of resource optimization in the 1960s and 1970s were very sophisticated for their time (and even ours): In the late 1950s, Kitov proposed the first ever national computer network for civilians; in the early 1960s, Kantorovich invented linear programming (and got the Nobel prize in Economics); shortly after Glushkov introduced cybernetics.
Kitov’s idea was for civilian organizations to use functioning military computer ‘complexes’ for economic planning (whenever the latter are idle, for example during the night). Kantorovich brought in linear programming which substantially improved the efficiency of some industries (he is the central character in a very well written book about the Soviet planning system called ‘Red Plenty’). Glushkov combined these two ideas and his OGAS (The All-State Automated System for the Gathering and Processing of Information for the Accounting, Planning and Governance of the National Economy) was intended to become a real-time, decentralized, computer network of Soviet factories. The idea was very similar to a version of today’s permissioned blockchain: the central computer in Moscow would grant authorizations but users could then contact each other without going through Moscow.
The Soviet planning system failed not necessarily because it would not work (limited, though, as it was in terms of computational power and availability of data) but because of politics: Khruschev, who had taken over after WW2 and denounced the brutality of Stalin, was ousted by Brezhnev. The early researchers were pushed aside (in fact, those Brezhnev years were characterized by fierce competition among scientists for preferential political treatment). One could say, the Soviets’ model of resource optimization failed because it was not socialist enough (compared to how the Internet took root in the US on the back of well-regulated state funding and collaboration amongst researchers). In other words, the 1970s Soviet Union was a political rather than a technical failure.
I should know, I guess. I grew up in one of the Soviet satellites. My father was in charge of a Glushkov-style information data centre within a large fertilizer factory. When we were kids, we used to build paper houses with the square punched cards he would sometimes bring home from work. Later on, when I became a teenager, my father would teach computer programming as an extracurricular activity in my local school (I never learned how to program – I preferred to spend my time playing Pacman instead!). At that time, Bulgaria used to produce the PC, Pravetz (a clone of Apple II), which was instrumental in the economy of all the countries within the Soviet sphere of influence.
By the time I was graduating from high school, though, things had begun deteriorating significantly: even though everyone had a job, ‘no one was working’ and there was not much to buy as the shops lacked even the essentials. Upon graduation and shortly after the ‘Iron Curtain’ fell down, I left to study in America.
Eventually, I ended up spending much more time in the ‘trial and error’ economy of the developed world, working at the heart of the ‘free market’ in New York and London. I am certainly not unique in that sense as many people have done this exact same thing, but it does allow me to make an observation about the merits of the planned economy vs the free market.
My point is the following. The problem of the planned economy was not so much technical misallocation of resources, but, ironically, one of proper distribution of the surplus. The Soviet system did not exactly create an extreme inequality, like the one there is now in America (even though some people at the top of the Party did get exorbitantly rich) but instead of using the production surplus for the betterment of the life of the population NOW, politicians continued to be obsessed with further re-investment for the future. There was perhaps a justification for that but it was purely ideological, a military industrial competition with America, nothing to do with reality on the ground.
So, while the Soviets were perhaps winning that competition (Sputnik, Gagarin, Mir, etc.), the plight of the common people was not getting better. And while they ‘couldn’t’ simply go in the street and protest or vote the ruling party out, they expressed their anger by simply pretending to work. Of course, that eventually hurt them more as the surplus naturally started dwindling, productivity collapsed and the quality of the finished products deteriorated. The question is, given a chance, would the planned optimization process have worked? If Glushkov’s decentralized network with minimum input from humans had been developed further, would the outcome now be different?
There is a lesson here somewhere not just for China but also America. Both have created massive surpluses using the two opposing optimization solutions. And both are running the risk of squandering that surplus, in a similar fashion to the Soviet Union of the 1980s, if they don’t start distributing it to the population at large for general consumption. In both cases this means transferring more income to ‘labour’: in China away from the state (corporates), in America away from the capital (owners). But because the differences at the core of the two systems, it is easier for China to do this consciously; in America, the optimization process of the free market, unfortunately, ensures that the capital vs labour inequality goes further to the extreme.
So, can China then pull it off?
While I am not privy to the intricacies of their ‘planned’ resource optimization model, just like in the Soviet Union, the risks there seem more political. But after an additional 50 years of Moore’s law providing computational power and after digitalization has allowed access to data the Soviets could never even dream of, China stands a much better chance of making it than the Soviet Union ever did.
*I actually use “The Coming Collapse of China”, Gordon Chang, 2001 as a reference point
According to John Authers at Bloomberg, data from CrossBorder Capital, going back to January 1969, shows that we have been recently experiencing Fed’s greatest liquidity boost ever. I have no reason to doubt CrossBorder Capital or their proprietary model of measuring liquidity. But there are many ways of defining, as well as measuring, liquidity. So, I decided to simply look at what exactly the Fed has done since it started expanding its balance sheet in September last year.
Fed has indeed been doing more than $100Bn worth of repo operations on a daily basis recently, but those operations are only temporary, i.e. they can not be taken cumulatively in ascertaining the effect on liquidity. In fact, the Fed’s balance sheet has increased by $380Bn, and only 55% of which came from O/N and term repo operations ($211Bn). The other 45% came from asset purchases. On the asset purchases, the Fed bought mostly T-Bills ($182Bn), some coupons ($55Bn) while letting its MBS portfolio slowly mature (-$81Bn).
However, not all of that increase went towards interbank liquidity. In fact, only about 50% of that increase ($198Bn) went towards bank deposits. The TGA account increased by $167Bn; that drained liquidity. Reverse repos decreased by $20Bn (FRP by $17Bn and others by $3Bn), which added liquidity. Finally, $37Bn went towards the natural increase in currency in circulation.
Fed actually started increasing its T-Bill and UST portfolio already in mid-August, three weeks before the repo spike. Part of that increase went towards MBS maturities. But by the end of August, Fed’s balance sheet had already started growing. By the third week of September, also the combined assets portfolio (T-Bills, USTs, MBS) started growing as well, even though MBS continued to decrease on a net basis.
Fed’s repo operations started the second week of September. They reached a high of $256Bn in the last week of December. At the moment they are at the same level where they were in the first week of December ($211Bn).
On the liability side, the TGA account actually bottomed out two weeks before the Fed started buying USTs and T-Bills, while the FRP account topped the week the Fed started the repo operations. Could it be a coincidence? I don’t think so. My guess is that the Fed knew exactly what was going on and took precautions on time (we might find eventually if it did indeed nudge foreigners to start moving funds away from FRP).
Finally, while currency in circulation naturally increases with time, bank deposits also bottomed out the week the Fed started the repo operations in September, but strangely enough, they topped the first week of December (for the time being).
So, while the Fed’s liquidity injection since last September was substantial relative to both the decrease in liquidity before that (starting in 2018 when the decrease in the Fed’s balance sheet became consistent) and, to a certain extent, since the end of the 2008 financial crisis, it is difficult to make a claim that this is the greatest liquidity boost ever. The charts below show the 4-week and 3-month moving average percentage change in the Fed’s balance sheet. The 4-week change in September was indeed the largest boost in liquidity since the immediate aftermath of the 2008 financial crisis. The 3-month change though isn’t.
The Fed pumped more liquidity in the system during the European debt crisis. In the first four months of 2013, not only the growth rate of the Fed’s balance sheet was higher than in the last four months now since September 2019, but also the absolute increase in Fed’s assets and US bank deposits. Moreover, there were no equivalent increases in either the TGA or the FRP accounts.
Final note, if the first week of January is any guide, it might be that a big chunk of the Fed’s balance sheet increase might be behind us, if only for the time being. Fed’s balance sheet decreased by $24Bn, which is the largest absolute decrease since the last week of July 2019, i.e. before the start of the most recent boost in liquidity. I actually do expect the Fed’s balance sheet to keep growing but at a much smaller scale and mostly through asset purchases rather than repos.
ETFs are not like subprime CDOs but they come close. Direct access to the Fed’s balance sheet will become essential for fund managers’ survival during the next financial crisis.
According to Bloomberg Magazine, the largest asset managers in the world, BlackRock, Vanguard and State Street, hold about 80% of all indexed money.
“Some 22% of the shares of the typical S&P 500 company sits in their portfolios, up from 13.5% in 2008…BlackRock, Vanguard, and State Street combined own 18% of Apple Inc.’s shares, up from 7% at the end of 2009… The phenomenon can be even more pronounced for smaller companies.”
This high concentration is the most serious danger to stock market bulls. Though, it is not obvious what the trigger for a market decline could be, when it happens, the present market structure could make it a much worse experience than the 2008 stock market decline.
Unlike 2008, however, the risks are on the buyside and the market doesn’t seem leveraged. But like 2008, the Fed is probably in the dark to the actual risks in the system, because the buyside is like shadow banking: no one knows what is going on/off fund managers’ balance sheets. Like the broker-dealers of pre-2008, the buyside today does not have access to the Fed’s balance sheet. The Prime Dealers Credit Facility (PDCF), which allowed access to borrowing from the Fed, was only created after Bear Stearns ‘failed’. Still, PDCF did not help Lehman Brothers, even though the latter did have good collateral at hand.
The buyside now may not be leveraged that much indeed but the extreme concentration of positions leads to the same effect on liquidity under stress as in 2008. This concentration is worsened by the fact that ETF sizes are many times larger than the underlying assets/markets in many cases. And though in 2008 broker-dealers ‘could’ possibly get some liquidity if they had good collateral, now the interbank market is much trickier as banks are in a regulatory straitjacket and it is not obvious (barring de-regulation) how they can provide much more liquidity even under normal conditions.
Does this make ETFs as dangerous as subprime CDOs of the 2000s indeed? I don’t know, but it makes them not that different at the same time. For example, we know that at least 40% of S&P 500 companies are money losing and we know that there is very high concentration of risk in them as per the Bloomberg Magazine article above (for comparisons, the percentage of subprime mortgages in 2000s CDOs varied between 50% in 2003 to 75% in 2007).
There are also similarities in the way the two markets emerged. In the late 1990s the Clinton administration decided not only to close the budget deficit but to also run a surplus. Bad things happen in financial markets when the US runs a budget surplus and reduces the flow of safe assets (thankfully, it does not happen often). The market responded by creating fake safe assets, like (subprime) CDOs.
In a similar fashion to the US Treasury actions of the late 1990s, US corporates have been buying back their shares, significantly reducing US public share count in the process. As the financial sector kept growing, it was starved of options where to put its money. How did the market respond? Fund managers (mostly, but also some banks) started creating ETFs. Just like CDOs, with the ETFs you buy the pure-bred stallions and the broken carriage as a package – you don’t have a choice. And as the passive/indexing trend spread, concentration soared further.
Non-financial corporate issuance has indeed been negative (corporates bought back shares) since 2008 at $4.6Tn, cumulative, but ETFs issuance is a positive $3.2tn. So, net there is still a reduction in public equity flow but nothing as dramatic as some sell side analysts claim (excluding US equity issuance abroad – see below).
And the flip side of that is retail, which has indeed been selling equities direct but also buying indirect through ETFs – so, similarly, households have sold equity risk down but not as much as claimed – in a sense, retail buying is ‘masked’ in the flows (it is simply an incomplete ‘wash’ from owning equities direct in an active form to owning equities indirect in passive ETFs).
Actually, in 2019 households bought the most equities direct since the crisis. Foreigners and mutual funds, on the other hand, sold the most equities since the crisis. And equity issuance was the most negative (in the chart below, it is shown as a positive number to signify equity share buybacks).
Equity issuance above is comprised of non-financial corporate, ETF and new issues abroad. When most sell side analysts report share buybacks, they only take into consideration net domestic non-financial corporate issuance. But ETFs and new issues abroad also matter from a flow perspective. The big change in 2019 was, in fact, the new issuance abroad which was negative – the last time it was negative was in 2008 (the data for 2019 is as of Q3, but it is annualized for comparison purposes).
Unlike 2008, however, it is not obvious to me where the trigger for the unwind of the ETF flow would come from. With the CDOs it was ‘easy’ – all it took was for rates to rise to ‘cripple’ both the mortgage payer and the leveraged CDOs owner. The debt overhang today is actually even bigger than in 2007, but the Fed’s failed experiment between 2016-2018 pretty much assures rates will stay low for the foreseeable future. Still, even though there is not much leverage in the system now, the high concentration of risk could produce the same effect on liquidity as if there were. But you still need a ‘seller’ to start the carriage rolling down the hill, don’t you?
This seller could be retail as an unwind of the 2019 inflow. Or the ‘selling’ could come from US corporates themselves (in the form of less buying back of their own shares – or no buying at all as a response to a regulatory change – but the latter is a 2021 event, post US presidential election, most likely). Or it could be a natural decline in share buybacks as a response to a drop in corporate Free Cash Flows (FCF) on the back of top-line revenues having peaked already (for the 17 out of 20 largest share buyback companies that is indeed the case).
The need for liquidity from fund managers is unlikely to be provided by the banking system, which is even now, without any stress, constrained by Basel III regulations to expand sufficiently its balance sheet. The Fed could be either forced to start buying equities to stem the slide and allow fund managers to meet redemptions, or it could extend a direct line of liquidity to them in a similar fashion to the creation of PDCF in 2008 for the primary dealers. My bet is on the latter as a politically more acceptable solution.
No financial crises are alike but a precondition for all of them is an extreme build-up of either economic or structural market imbalance. The next crisis is more likely to be a function of the latter one, namely high concentration of risk in institutions without direct access to liquidity from the Fed.
As for the trigger for the crisis, we can only speculate. Anyway, there isn’t usually one trigger per se but rather a combination. And we are only meant to figure that out in retrospect. But it is important to do away with two wrong narratives. First, that retail accounts have been large sellers of equities, and second, that the supply of equities has been hugely negative.
This note showed that neither of them holds water. ETF flow is at the core of the matter here. On one hand, it has facilitated the issuance of plenty equity-like instruments to make up for some of the lack of direct equity issuance, and, on the other, it has allowed retail to partially switch from active and direct equity management to passive and indirect one. In doing so, it has created the said market imbalance.
“For a little reflection will show what enormous social changes would result from a gradual disappearance of a rate of return on accumulated wealth.”
~ John Maynard Keynes
Lately, not a day passes by without someone commenting on the pernicious effect of negative rates and how they are an aberration which cannot and should not be ‘allowed’ to continue. Reality is slightly more nuanced.
To start with, low interest rates are the norm, not the exception throughout history. Second, while indeed a rarity, negative rates have existed in the past, and, depending on circumstances, have lasted longer than initially expected. Third, to ascertain their effect, we must first understand their cause and purpose. All these will eventually allow us to forecast how long they would be around. Still, even then, we must be cognizant that a switch to higher rates will most likely only happen after the economy has first gone through one or a combination of: social unrest, debt jubilee, large increase in the money supply or natural disaster.
Negative interest rates are a result of past accumulation of surplus capital (and its mirror image, large stock of debt) combined with previous persistently high interest rates on that debt relative to the growth rate of the money supply (new money).
The forces that could push rates structurally higher, therefore, would logically be either a reduction of the surplus capital/debt, or a massive increase in the money supply. As neither of this has happened yet, negative interest rates are effectively the market’s response to this status quo: on a long enough timeframe, they reduce the debt stock and they allow the money supply stock to outpace interest payments on the debt and have some left for economic growth.
Considering that interest rates measure the cost of capital, when capital is abundant, ceteris paribus, interest rates should be low. For example, normally, periods of peace bring about an accumulation of surplus capital, either directly because money is not spent on wars, but more importantly, indirectly, as people innovate and bring about technological advancements which increase efficiency and reduce the need for more capital in general. As a result, interest rates trend lower. That’s exactly what happened, indeed, during the almost century of peace in the time of Pax Britannica in the 19th century.
Source: BeyondOverton, BOE Three Centuries of Data
At the same time, wars, conflicts, or even big natural disasters, deplete the capital stock and force interest rates to rise. Sometimes, when these negative supply shocks turn out to be ‘one-off’ occurrences (the 1970s oil crises), we could get just a spike in interest rates; sometimes, if the conflict persists, the increase in interest rates can last much longer (the Cold War).
On a long enough timescale, one would then expect to see periods of low interest rates inevitably followed by periods of high interest rates in a kind-of mean reversion pattern. Actually, that is not the case. In fact, according to Paul Schmeizing, real rates have been falling for over 500 years on a variety of regression measures:
“…over the entire timeframe 1313-2018, I find 19.7% of advanced economy GDP experience negative long-term real rates on an annual basis…the general trend of an even higher frequency of negative rates is independent of the establishment of central banks and active monetary policy.”
Mean reversion of interest rates is not a given from a historical point of view largely because certain events, the so-called paradigm shifts, have such a profound effect on the production function that no war or natural disaster can easily reverse. For example, the Agricultural Revolution ushered the first large (and ‘permanent’) resource surplus which lasted humanity indeed a long time. We came close to depleting it during the centuries of the Dark Ages in Europe but, with the help from the Renaissance, the Industrial Revolution couldn’t come soon enough to change the paradigm shift once again. After that, Aggregate Supply (AS) had been consistently running above Aggregate Demand (AD).
While on the face of it, an imbalance of this sort, AS > AD, is much better than the reverse, managing it, has proven quite difficult over the years. Especially in the more modern times when the changes affect both the production function and the mode and nature of consumption (from a physical to a digital medium – more on this later).
Feudalism, socialism, capitalism, etc., are all examples of how society is designing an institutional framework to help distribute these surpluses in the most optimal way. However, because of the inertia of the past and the numerous vested interests, such institutional changes may take much longer than the production breakthroughs to feed through. Therefore, as the capital surpluses keep adding up while their distribution mode remains the same, the economy becomes even more imbalanced.
If capital does not flow naturally through the income channel to raise the purchasing power of the majority, aggregate demand starts to lag. Debt becomes then the lever which transfers purchasing power, in a way substituting for rising wages. However, as debt comes with the additional burden of positive interest rates, it pushes up inequality to an unsustainable level thus closing even that avenue of balancing the economy. Therefore, AS continues to increase at the expense of AD, and a deflationary spiral ensues. A temporary solution to this problem in the past has indeed been a form of negative rates, called demurrage money.
Demurrage money is not unusual in history. Early forms of commodity money, like grain and cattle, were indeed subject to decay. Even metallic money, later on, was subject to inherent ‘negative’ interest rates. In the Middle Ages in Europe coins were periodically recoiled and then re-minted at a discount rate (in England, for example, this was done every 6 years, and for every four coins, only three were issued back). Money supply though, did not shrink, as the authorities (the king) would replenish the difference.
In 1906, Silvio Gesell proposed a system of demurrage money which he called Freigeld (free money), effectively placing a stamp on each paper note costing a fraction of the note’s value over a specific time period. During the Great Depression, Gesell’s idea was used in some parts of Europe (the wara and the Worgl) with the demurrage rate of 1% per month.
The idea behind demurrage money is to decouple two of the three attributes of money: store of value vs medium of exchange. These two cannot possibly co-exist and are in constant ‘conflict’ with each other: a medium of exchange needs to circulate to have any value, but a store of value, by default, ‘requires’ money to be kept out of circulation.
Negative interest rates solve this issue by splitting these two functions. The problem though is that negative rates are not a very efficient tool for reducing the capital surplus because the whole process takes a really long time. Absent any other changes in the institutional framework, the general pattern of the past has therefore been for a military conflict, either a revolution or a war, to literally obliterates the capital surplus.
As mentioned before, indeed, the Industrial Revolution was followed by the century of peace of Pax Britannica during which neither low rates nor the gold standard managed to close down the inter-country economic imbalances, thus we got two very violent world wars. The period between the end of WW2 and now is considered one of general world peace. And indeed, relative to the horrors of the war which preceded it, it was.
But despite the fact that there were no major traditional global wars after 1945, the Cold War was a major global war of ideologies, in which few shots were fired, but one which caused a large capital outlay (military build-up, but also huge government investments in space exploration, and in general, technology). In addition, there were a lot of proxy wars (Afghanistan, Korea, Vietnam) and conflicts (for example, the 1970s oil crises were a result of such a proxy conflict). It is not surprising then, that during that time interest rates tended to stay high.
(Incidentally, it is with sadness that I heard of the passing of Paul Volcker the other day, but reality is that he presided over a Fed which orchestrated the largest aberration in the history of interest rates since Babylonian times. In my opinion this was totally unnecessary and a complete overkill.)
Source: Business Insider; original data from ‘History of Interest Rates’ by Homer and Sylla
By the end of the Cold War, when it was clear that capitalism had gained the upper hand as the main institutional framework of the time, interest rates started to subside. It also became obvious that the USA was to become the undisputedly dominant global power. In addition, all these (mostly government) investments of the Cold-War time started to pay off, eventually ushering the Digital Revolution which is still ongoing. To a certain extent, one could think of this period as Pax Americana, in reference to the global dominance of Britain during most of the 19th century.
The Digital Revolution has heralded a similar paradigm shift to the Agriculture and Industrial Revolutions in the past. Indeed, the resulting capital surplus has not only completely reversed the previous spike in interest rates but has brought about a strong disinflationary environment pushing real interest rates in negative territory.
This period is called by some the biggest and longest bond bull market in history. It is probably the biggest because interest rates have gone down from double digits in the 1980s to almost 0% now. But it is certainly not the longest. As seen in the chart on page 1, interest rates in Britain trended down from 6% to 2% for almost 100 years in the 19th century. And it doesn’t yet look like there is an end to this bull market as there are no signs that anything is being done on the institutional side to take into account the changing modes of production and consumption caused by the paradigm shift of the Digital Revolution.
Moreover, low/negative interest rates are only really applicable to the government sovereign market. In the current debt-backed system, the majority of money is still loaned into circulation at a positive interest rate. Even in Europe and Japan, where base interest rates and sovereign bond yields are negative, the majority of private debt still carries a positive interest rate. This structure inherently requires a constantly growing portion of the existing stock of money to be devoted to paying solely interest. Thus, the rate of growth of the money supply has to be equal to or greater to the rate of interest; otherwise more and more money would be devoted to paying interest rather than to economic activity.
Source: BeyondOverton, US Federal Reserve
This is indeed problematic if one considers that the long-term average growth rate of US money supply in modern times is around 6% (chart above), which is only slightly higher than the average interest rate on US government debt but it is below the average interest rate on both US household and corporate debt. To reach this conclusion, I used the US Treasury 10-year yield as the average yield on US government debt (the average maturity of US debt is slightly less than that), and allowed very generous estimates for both US corporate debt and household debt.
In addition, I have not included the much higher yield on US corporate junk bonds which comprise a growing proportion of overall corporate debt now. I have not used either credit card/consumer debt, which has a much higher interest rate, or student loan debt, which carries approximately similar interest rate to auto loans rates used in the calculation. Just like for BBB and lower rated US corporates, credit card and student loan debt are a much higher proportion of total US household indebtedness now compared to before the 2008 crisis.
Finally, I estimated the long-term average economy-wide interest rate as a weighted average of government, corporate and household debt – with the weights being their portions of the total stock of debt. With the caveats mentioned above, that average rate since the early 1980s is about 7% – higher than the average money supply growth rate.
Over the last four decades, US money supply has not only not grown enough, on average, to stimulate US economic growth, but has been, in fact, even below the overall interest rate of the economy. Needless to say, this is not an environment that can last for a long time. It is surprising it did go on for so long.
Indeed, if one calculates the above equivalent rates for the period 1980-2007 only, the situation would be even more extreme (see chart above). In fact, until the late 1990s, money supply growth had been pretty much consistently below the economy-wide interest rate. Only after the dotcom crisis, but really after the 2008 crisis, money supply growth rate picked up and stayed on average above the economy-wide interest rate.
What is the situation now? The current money supply growth rate is just above the average economy-wide interest rate: above the government and corporate interest rates but below the household interest rate (data is as of Q1’2019, chart below). It is also still below the combined average private sector interest rate.
So, even at these low interest rate, US money supply is just about ‘enough’ to cover interest payments on previously created money. And that is assuming equal distribution of money. Reality is that new money creation is only just ‘enough’ to cover interest payment on public debt. Moreover, money distribution is very skewed in the private sector: corporates have record amount of cash but that cash normally sits only in the treasuries of few corporates. The private sector, overall, can barely cover its interest payment, let alone invest in CAPEX, etc.
Seen from this angle, negative interest rates may not be a temporary phenomenon designed just to spur lending. On the opposite. It is almost counter-intuitive from what we learn in economics where we are accustomed to think that a rising GDP is associated with higher interest rates because of the need to suppress potentially inflationary pressures. Reality is that a rising GDP also produces more excess capital which tends to naturally put pressure on interest rates lower. If this increase in AS is not fully offset by a rise in AD, inflationary pressures may not develop and interest rates may not rise. In fact, they may start falling if the debt build-up becomes excessive. In that regard, the purpose of negative interest rates may be to help reduce the overall debt stock in the economy and to escape the deflationary liquidity trap caused by the declining marginal efficiency of capital.
Could they work? Sure, they could, but unless they are deeply negative, it will take a really long time and, most likely, the fabric of society would come apart either way. So, what could cause this massive bull market in rates then to reverse?
Well, it is unlikely to see those signs of reversal in any economic variable on the demand side, like lower unemployment or even higher wages, as the surpluses are just too large. At least not from a structural point of view: for example, a pop in real wages could see a pop in real rates but that will quickly reverse as the supply side will adjust almost ‘instantaneously’. Instead, we should look for signs of any pressure on AS which would come about from institutional changes. Anything that suddenly reduces the capital/debt surplus, such as a debt jubilee, or permanent increases of the money supply, such as ‘helicopter money’.
In the absence of such changes, we could either see a prolonged period of negative interest rates to address the above imbalances or, in the worst-case scenario, for example in the US where there is strong institutional pushback against them, social unrest. The process of de-globalization, which started already with Brexit and Trump’s US tariffs, is another supply side force which would take its time but could eventually erode the global resource surplus.
In the end, if all else fails, nature would have the final say as climate change could cause a massive natural disaster, leading to such a destruction of capital, that interest rates would be bound to go much higher from there!
 Eight Centuries of Global Real Interest Rates, R-G, and the ‘Suprasecular’ Decline, 1311-2018, 24 Nov. 2019
 For corporate debt I used the average yield on Aaa and Baa bonds and for household debt I used mortgage debt and auto loans
The stock market is a forward discounting mechanism.
How much forward?
It seems quite a lot judging by this chart above.
At 28.6% return YTD, the S&P Index is on course to one of its best years on record. Going back to 1927, only 10 years have seen more than 30% returns. It will also wrap up a spectacular decade for the index: with still a few days left in 2019, it can still beat 2013, the best year of this past decade, which finished just shy of 30%, at 29.6%.
While US corporate profits have remained flat since 2012, the S&P Index has doubled. That’s a lot of discounting. But that pales in comparisons to Apple’s share price which has doubled from the lows this year, while its revenue is down during that period (latest numbers for Apple are only as of Q3, so things may change with Q4 numbers released).
To be fair, the spectacular performance of US stocks this last decade may be just catching up with the previous decade (2000-2009) during which S&P 500 Index declined while US corporate profits increased 2.5x. So, starting point matters.
Therefore, it is best to look at the long term-chart, at the beginning of the post, going back to 1947. S&P Index and US corporate profits moved hand in hand until the early 1980s when US financialization kicked in (shareholders primacy and the share buybacks era). There had been, of course, some going back and forth between the two even before 1980 but at the beginning of the 1980s they were ‘on top of each other’.
The volatility between the two sharply increased thereafter. Stocks raced higher and while corporate profits also grew, the growth rate of the S&P index outperformed. Then in the mid-1990s, corporate profits stopped growing but the S&P index continued to race higher. The dotcom crisis and the 2008 financial crisis served as a reset but it still took until 2012 for US corporate profits to ‘catch up’ to US stocks.
The time post 2012 seems very similar to the 1995-2000 period whereby corporate profits did not grow but US stocks rose nevertheless. Only this time, at the end of 2019, the disconnect between the two seems to be the most extreme ever.
The stock market is a forward discounting mechanism indeed, but judging from history a reset is more likely to ‘justify’ that discounting. One problem though: no one can claim to know when that will happen.
Apple is up more than 70% this year; Microsoft – 50%. Together they have contributed almost 40% to the more than 30% rise in the Nasdaq.
Facebook, Google and Amazon, the other heavy weights in the index, have together contributed about 19% to this year’s performance.
Just five companies have contributed almost 60% to this year’s Nasdaq returns!
Apple and Microsoft together have bought back almost $85Bn of their own shares this year. Amazon does not buy back its shares (yet). Google started in 2015 and has bought back $8Bn this year. Facebook started buying back its shares last year and has bought back about $13Bn this year.
Intel, Comcast, Nvidia, Costco, Charter Communications bring up the top 10 with a combined share buyback of around $20bn. Costco, Charter Communications and, to a certain extent, Nvidia, are actually, good examples of Yardeni’s claim that companies mostly buy back shares to avoid dilution. Sadly, that is not the case for the heavy hitters in the list, the ones doing most of the buybacks, and thus the ones with a disproportionate influence on index performance. For example, just among those five, Intel and Comcast have done 3/4 of the combined buybacks.
There is a big difference when it comes to Apple and Microsoft, however.
Since 2007 Apple has outperformed Microsoft in both Revenue growth (former’s increased 10x, latter’s – about 2.5x) and EPS growth (20x increase vs. 3.5x). In 2007 Apple’s revenue was about half of Microsoft’s. In 2010 they were the same. Today Apple’s revenue is more than 2.5x bigger than Microsoft’s, however it topped in Q3’2018, while Microsoft’s is still growing.
Apple only started paying dividend in 2010 and buying back its stock in 2011. Since then, both have spent about 25% of their top-line revenues, on average, on shareholders payouts. However, Microsoft spends considerably more on R&D as a % of revenue than Apple.
Apple managed to reduce its share count by almost 30% since 2012. Microsoft reduced its share count by a ‘mere’ 22% since 2007. And it shows. Apple’s share price has massively outperformed Microsoft’s since 2007. Of course, there are other factors in place (the IPhone came in 2007) and maybe it is just a coincidence that the out-performance started in 2010, when Apple began giving cash back to shareholders!
Bottom line: Make sure you have at least AAPL and MSFT share price on your screens all the time into year end.
Home Depot cuts its 2019 forecast after sales miss; shares drop the most since 2008.
Since 2007, Home Depot’s top-line revenue has increased by 37%, while its EPS has increased by 249%, almost 7x more.
Since 2007, Home Depot has bought $72Bn shares back, one of the largest share buyback programs out there, cutting its share count in the process by almost half; yet its dividend payments increased by more than 3x.
Emphasis on the fact that Home Depot topped analysts’ expectations for earnings but sales fell short. Which is exactly the story of our times: Easier to push up EPS through the share count (buybacks) than through top-line revenue growth. Management also gets paid on EPS, not revenue.
And the cherry on the cake: Home Depot’s share price has increased by more than 700% since the end of 2007; the S&P Index has increased by about 111% since then.
One would have thought that after so many years of Home Depot using the same technique to ‘score a home run’, investors would have understood the tactic and re-priced its valuation.
The idea about this post came after listening to Ben Hunt’s and Ed Yardeni’s recent Ritholz Wealth Management podcasts. While, in general, I do agree with the conclusion Ben reaches, I thought looking at just two individual companies’ share buyback programs, as he did in two separate posts, is perhaps a bit too narrow. At the same time, I was doubtful whether Yardeni’s approach of going to the opposite extreme of looking at all S&P 500 companies and taking the average is equally that ‘practical’.
Ed Yardeni is a top strategist but his claim in “Exposing the big lie about stock buybacks” that the majority (2/3) of US stock buybacks go to cover employee stock issuance, even if true on an aggregate level, does not change the fact that buybacks are possibly the main driver of US stock out-performance.
“The current source for equity issuance data in the Financial Accounts of the United States does not fully incorporate issuance to employees by public corporations. Staff is exploring how best to reflect such issuance activity in future releases.”
Emphasis on fully. What does this mean? Look at p.68, table F.223 of the latest Flow of Funds data. So, line 2 (‘Nonfinancial corporate business’) does not include ESOPs (and other equity compensation plans)? I don’t think that changes the demand-supply dynamics. Share grants are different from IPOs and buybacks as they are a non-cash flow item, as in the former there is no exchange of cash at issuance and the effect on the supply of shares happens only when those are sold by the employees in the secondary market. They are a balance sheet item. So mot likely, they will be registered either in line 16 (‘Household sector’) or in line 31 (‘Broker and dealers’) at the point of sale.
In any case, in the Flow of Funds data, ‘Net Issue‘ and ‘Net Purchases‘ have to balance, if we start tinkering with the net issuance (line 2), as in, assume that actually US corporates net bought back fewer shares to account for the ESOP (and other equity compensation programs), then we have to also change something on the net purchases side (line 15 and below). My guess is that ESOPs are incorporated in the Flow of Funds data but not in the net issuance section (with a positive sign), rather in the net purchases section (negative sign). I think this is the more likely scenario in what the Fed official meant as “not fully incorporating ESOP” in that data.
I could be wrong on this, so looking forward to some clarity from the Fed in the future publications of the Flow of Funds data.
I have no reason to doubt indeed that on average for all the companies in the S&P 500, 2/3 of share buyback activity goes towards avoiding dilution from share issuance on the back of employee compensation plans. However, this is not true for the largest corporate share buybacks programs in terms of dollar amounts conversion. And because of their market weight, this is where the big impact from buybacks on the S&P 500 comes from (on average, the price change of these companies’ share price is double that of the S&P 500 in the last 8 years).
If we take the 20 largest share buybacks programs and look at the same period 2011-2018 which also Yardeni data covers, employee stock compensation is only about 14% of the value of the corporate share buybacks. The ratios vary from company to company but just to give you an example, for Apple, which has, by far, the largest share repurchase program of all companies in the US, employee stock compensation is on average about 11% of total buybacks. The banks have the largest ratio, at around 35%, while Boeing and Visa have the lowest, around 4%.
Those large share buyback programs have reduced the companies share count by about 18% on average between 2011 – 2018. In some of the larger buybacks, like Apple, Pfizer etc., the share count is reduced by almost 25%. So, even if indeed some of the share buyback activity is on the back of employee compensation, there is still substantial reduction in the share count which should have an effect on share prices and on performance indicators per share.
ESOPs are actually not that prevalent in US businesses. In fact, only about 10% of the private workforce is involved in such programs and only about 10% of the publicly traded companies have them. Public companies, however, have other equity compensation programs but they are mostly geared to senior management. There are three kinds of equity awards that public companies give: restricted shares, share options and performance shares.
Share options had been the original and the most popular form of executive compensation until the mid-2000s or so. Their use was limited in the 1960-1970s, however, as the stock market languished and went nowhere (which is when companies started giving out restricted shares, which, even though they vest, they do not require either any upward movement of the share prices to ‘make money’ – as options do, or beating any performance targets – as performance shares do).
However, by the late 1990s, with the stock market rising substantially on the back of the financialization of the US economy which started in the 1980s with the doctrine of shareholder primacy and the accompanying surge of corporate share buybacks (SEC Rule 10B-18), the popularity of stock options surged to 75% of all executive compensation. By the mid-2000s, though, three things happened which affected adversely the use of stock options in executive compensation programs: 1) in 2006 the accounting rules changed, resulting in charge against earnings for the grant of options; 2) during the crisis of 2008, stock markets fell substantially; 3) The Dodd-Frank Act of 2010 now requires a say-on-pay vote. As a result, organizations such as Institutional Shareholder Services, now have a much greater influence over executive pay. Therefore, performance shares have become the dominant source of executive compensation.
There are two points I want to make here. First, the concern over exuberant executive pay is not something new – regulators have been looking into it since at least the 2000s. Second, because the performance targets are tied to companies’ EPS, the changes made to the executive compensation plans, unfortunately, fall short of their target: senior management can legally find a loophole to still continue to disproportionately increase their compensation. What do I mean ‘disproportionately’? I mean relative to other employees’ compensation and to companies’ top line growth.
Which brings me to a second claim made by Yardeni, namely that S&P 500 companies’ EPS growth is only about 1% lower than their total earnings growth. Again, I have no reason to doubt that, as a whole, this is correct: on average across all the S&P500 companies, there may not be much ‘smoothing out’ of EPS through share buybacks . But looking at the largest share buybacks programs, that’s not the case.
Of those, on average over the 2011-2018 period, top line net operational income growth is about half of EPS’s growth. What I found is that while top line income growth is somewhat similar to the average of the whole S&P 500 index (11.1% vs 8.1%), EPS growth in the largest share buyback companies was much higher.
Again, the numbers vary. For some (Microsoft, Nike), actually, average top-line growth is slightly higher than EPS growth. On the other spectrum, companies like Pfizer and Merck have had much smaller top line growth rates than EPS ones. Apple, having the largest share buyback program, has a more respectable 23% average annual top line operational income growth rate vs 27% EPS growth rate. But still very far from the average for the S&P 500 index company.
In fact, for majority of the companies in that sample, their revenues had peaked already some time ago. That did not stop companies continuing with their share buyback programs. In fact, some like Apple, Cisco, Amgen, BAML, Merck, Pfizer have substantially increased their spend on share buybacks thereafter.
Given the above, it is not unreasonable to assume that indeed senior management in these companies is ‘wrongly’ incentivized to authorize more buybacks despite lackluster top-line operational performance in order to benefit from the obviously better EPS performance.
But it is not all doom and gloom when it comes to share buybacks. Employment
has risen on average by 26% over that period for those companies. However, that
is really concentrated in a few big companies like Apple, Microsoft and Nike
(for example, Apple has doubled its employment between 2001-2018). A lot of
companies, the industrials (Pepsico, Merck, Pfizer, Texas Instruments) and the
banks (JPM, BAML) have actually marginally reduced employment.
Finally, while it is reasonable to suppose that a lot of the increased activity
in buybacks is linked to senior executives’ compensation packages, it is a lot
more difficult to make the claim that they happen at the expense of R&D. For
example, comparing Apple to Microsoft – massive difference when it comes to
R&D spending vs buybacks. While Apple has spent a lot less on R&D vs
buybacks in the last 8 years (in fact $175Bn less!), Microsoft has spent a lot
more ($19Bn). And while Apple spends just about 3% of its revenue on R&D, Microsoft
spends about 13%.
There is a divergence among the industrialists as well, with Texas
Instruments, Boeing and Amgen spending substantially more on buybacks relative
to R&D, while Pfizer, Merck and Cisco have spent substantially more on R&D
than on buybacks.
What’s the right way to analyse US share buybacks? I don’t think there is one right way. Ben Hunt has done great investigative work looking at a couple of the large corporate share buybacks. Ed Yardeni has done the same looking at the whole S&P 500 index. I have taken the middle road by looking at the 20 or so largest share buybacks. Probably each one of these three approaches is prone to some selection bias geared to the conclusions one may want to reach!?
From being a niche discussion and research topic when I first started looking at it in 2015 to being now in the popular media on a daily basis, share buybacks do deserve this attention because they could be potentially affecting things like income and wealth inequality, corporate investment, employment, productivity, economic growth, etc. So, it is only good if more and more people look at the the numbers to try to figure what exactly is going on.