Saturday, July 23, 2016

Macro | The Aerodynamics of Helicopter Money

As a former rotor-craft specialist I do have some experience with helicopters and its dynamics. It is a machine not supposed to fly, but somehow it does. And for some missions it is immensely more useful than the traditional stuff - fixed wing aircraft.

The next best thing to QE is already in town. Ever since former Fed chairman Ben Bernanke had a discussion with Japanese leaders last week, this has captured the attention of mainstream media. Although the BoJ Gov. Kuroda has effectively ruled out "helicopter money" (HM) on Monday, nobody missed the phrase "at this stage" in his statement. With increasing market frustration with the now-standard QEs, HM appears a real possibility in future policy adventure should things get much worse.

As is famously known, the term was originally described by the famous monetarist economics Milton Friedman to describe a permanent money creation and direct distribution to general population by central bank. In recent context, the meaning has changed more to monetary financing of fiscal stimulus. Nonetheless, it is interesting to see how this policy compares to other central bank tools like policy rates or QE.

There are two ways to look at, one from the accounting perspective and the other from economic perspective. From accounting point of view, HM is markedly different than other tools like policy rates or QEs that goes through what is known as open market operation (OMO). A central bank balance sheet, very roughly, can be thought as below. 


In traditional policy operation, the central bank announces a target rate and use standard OMO to adjust the level of treasury holding (asset side) to affect corresponding changes in commercial bank reserves (liability side). Tight monetary policy reduces the available reserves and hence put pressure on the fed fund rate (the rate at which commercial banks lend reserves to each other). QE in operation is similar to this, only the central bank buys a much larger quantity (and longer maturity) of treasuries (with a corresponding large increase in bank reserves). While the operations are similar, the channels through which they impact the economy are quite different. In case of regular OMO, the channel is mostly interest rate channel, where the long term interest rates are assumed to be affected by short term rates. In case of QE however, there are multiple channels, with the most important ones being inflation expectation, interest rate (portfolio re-balancing) and wealth effect. See here for a more detailed view.

HM is quite different than either of these. In the original scenario propose by Milton Friedman, the central bank simply prints money and distributes to the public. From accounting angle, this means an increase in currency in circulation (liability). It is clear the only change that can balance this is a corresponding decrease in capital of the central bank. Technically a central bank can run a negative capital indefinitely, as it can print money to fund it. However, in practice this may be limited due to legal rules (if any) and public and political perception among other things.

The current avatar of HM is different. The proposed method is government issuing perpetual zero coupon bond (appearing on the asset side of the central bank against a balancing liability entry for government account) and then using the proceeds to fund tax cuts or pay for infrastructure programs (ultimately money in government account from the last step disappearing in to accumulating commercial bank reserves). Prima facie the net effect has the appearance of a QE process as outlined above (treasury holding goes up, reserves goes up), But the dynamics is quite different. In QE, the money created will hit the commercial bank reserves directly. Now it is up to the lending intention of the commercial banks (and of course the ability and willingness of the general public to borrow) if this will just sit at the reserve or will actually enter the real economy. However, for HM, it is the other way around. The money created first goes to (via government) the general public and finds its way back to the banking system and reserves as the public either spend or save it. In this sense this monetary financing of fiscal expenditure is closer to the original HM concept in spirit.

The key difference is that QE or other OMOs are essentially asset swaps, swapping treasury for bank reserves - a swap between the two sides of the balance sheet. While HM is essentially swapping central bank capital for base money (currencies in circulation or bank reserves). In the above example, technically we recognized the zero coupon perpetual bonds issued by the government on the asset side at acquisition cost. But clearly such a bond has zero value, and a fair value treatment will create a hole in the capital, exactly like the original HM. The other key point to observe is that while QE is an increase of monetary base, its permanence is a function of central bank's credibility. Some may legitimately believe the central bank will withdraw this (sell QE assets) once the situation normalizes and hence factor that in into today's decision. However, HM is fundamentally an irrevocable permanent increase in base money. There is no way to reverse it unless central bank destroys currencies in circulation (reverse HM?) or forces the government to redeem those zero coupon perpetual bonds. Both seems highly unlikely under most scenarios conceivable.

Now on the impact of this policy on the broader economy - well since economics is not an exact science (and many assumptions are not even falsifiable), you can pretty much successfully argue for whatever you believe in. An HM operation can cause the interest rates to go down, as this means a large money supply in the economy. It can make things even worse if more people choose to save the money they get than to spend it, fearing an even lower interest rate and trying to keep interest income constant (think of retirees). You can argue for an increase in interest rates as well, as an injection of money in such a manner may increase inflation expectation. You can postulate that HM will cause GDP to increase - as a result of the direct fiscal expenditure and also through the fiscal multiplier. Or you can invoke the crowding out (and with some labor even the Ricardian equivalence) to assume no change at all.You can follow the thread of a heated argument here. However to give some method to the madness, we can arrange our thoughts in the IS-MP framework.

HM can be explained in this framework (see the figure below). The story is, in the beginning the aggregate demand is such that the output (GDP) is at y, below natural rate (y*). This causes inflation to fall. The central bank responds with a rate cut, to reduce the real rate, and pushes MP to right (expansionary policy), but hit the nominal zero lower bound. Then HM comes along and jacks up the inflation expectation (assuming that is the dominant dynamics, see above). This pushes the MP curve further to the right to MP1, beyond the possibility of zero lower bound. Then the fiscal stimulus component kicks in and moves the IS to the right at IS1 as well, bringing the output back to potential level of y*. Note the model suggests a final (real) interest rate levels higher than a pure play monetary policy response (only MP shifting to the right).


Theories apart, from market perspective a few things are more certain than others. Firstly, unless there is a crisis of confidence (or potential), fiscal stimulus is usually good for an economy, especially so at a zero rates environment when traditional monetary policy faces serious constraints, and at a time when economy can do with a booster dose or two. The fiscal stimulus component of HM therefore should be positive for markets and economy. One can argue why monetary financing is necessary when the government can borrow at such low rates. This is an excellent argument which the BoJ governor seems to like, at least for the time being. Nonetheless this part is positive for equities and risk assets. For FX markets, note the possibility of both the rates going down and up as noted earlier. Interestingly, this affects different parts of the curve differently. The part that will tend to go down will be short dated rates and long end will tend to push up. As a result FX (which is mostly influenced by the shorter end of the curve) will go down. And as for rates, assuming the market perception of HM is positive, this will mean 1) a re-pricing of the terminal rate upward as well as 2) increase in inflation expectation pricing. This will mean a bear steepening of the curve (increase in rates led by the long end on the balance).

The other aspect is of course the political risks of monetary finance. Some central banks absolutely abhor monetary financing (Bundesbank!), and many are potentially legally unable to do so. But leaving aside the muddled politics and economics, the key takeaway here is that in case of the next Lehman Brother scenario or a China bust, this talk about HM should assuage investors' collective concern that central banks are running out of options.

Wednesday, July 20, 2016

Off Topic | If you Read Only One Book From Every Country

This is a mighty interesting list. Appears here more as a bookmark for myself and a target before setting out on the final journey to the monosyllabic perdition.


Saturday, July 2, 2016

Markets: The Rise of The Vol Tourists

Since the Great Financial Crisis, the volatility market has undergone some significant changes. One major driver was an increased awareness about tail risk hedging. This was further aided by increasing acceptance of volatility as an asset class. Following the correlation one period during the crisis, the trend among asset managers has been risk factors based investment, moving away from traditional asset class diversification. This, along with the rising popularity of exchange traded funds and exchange traded notes, has given rise to a whole new set of demands for volatility products as an asset class.

Another impact came via the central bank reaction function route. The profound changes and the new normal condition following the crisis brought in a new set of players ready to supply (short) volatility - including those so called "vol tourists". But the appeal of systematic short volatility strategy has been strong following the crisis. As the unprecedented monetary stimulus created a huge yield chasing pressure, shorting volatility has become an important source. I have written about this quite a while back from rates perspective, but this is generally applicable to any asset class.

The left hand side chart below shows why shorting volatility systematically has been so popular. This tracks performance of a strategy that shorts the nearest IMM VIX futures and rolls just before expiry. The size is determined to match a margin of 10% of the invested capital (the approximate worst case loss). After the crisis, apart from a few hiccups (notably during the 2011 US debt ceiling crisis), the performance has been quite impressive. 


The result has been a discernible dynamics in the VIX futures market. The right hand side chart above shows the typical nature of VIX positioning that we have seen in recent time.

On one hand we have the asset managers managing the various ETNs linked to VIX. The left hand chart below shows the flows in to such ETNs (the short VIX ones are added with sign, reflecting net flow in to equivalent long VIX funds). These flows have typically been negatively correlated with VIX level itself. And the positions of these asset managers in the futures market have pretty much followed these flows - as shown in the right hand chart.


This has led to a situation where dominant players are the swap dealers (large banks) and leveraged money managers - the hedge funds - either discretionary or systematic short vol players. In fact, given the fact that after the introduction of tighter regulations since the crisis, most of the swap dealers positioning will be driven by hedges. So this leaves the leveraged managers as the only discretionary players in the VIX markets. 

This particular development in volatility markets - fundamentally driven by ZIRP policy of central banks, new regulations and the paradigm of risk factor investing - has resulted in an overall low volatility and high contago environment, even over and above what one can expect with a central bank puts. Apart from the China fear back in Aug 2015, the VIX level has remained remarkably tamed - below 25 almost always. Also the spread between front month VIX futures and the VIX levels itself has widened significantly since the crisis, as the most discretionary players have been systematically short in futures. The futures curve has been so steep that it is now very costly for long players to systematically roll macro hedges in VIX futures. In a normal market in a mean-reverting asset class like volatility, you would expect just the reverse.

The second impact, arguably, has been the feedback loop to S&P itself. As we have seen above, the VIX funds are flow driven. This means the leveraged managers are short against the large banks. The fact that most banks will have a hedged position, especially after the new regulations, make this positioning quite asymmetric. For the short VIX players, it is a linear position in volatility. However for the swap dealers - the opposite long VIX position will also mean a short option position as hedge. It is not important whether the short option position is the trade and long VIX is the hedge or vice versa. What is important that, a long VIX positioning will also mean a short gamma position. And the act of delta hedging will feed this into the underlying, i.e. S&P. If most hedgers are short gamma, as the underlying moves and the hedgers buy or sell to re-balance delta, they will tend to amplify the move. On the other hand, if most of the hedgers are long gamma, their delta hedging will introduce a stabilizing effect on the underlying. And this is captured in the following chart.


The chart shows the 20 day correlation (kernel-smoothed to capture the trend) of S&P 500 opening moves vs trading hours moves. This can be treated as a measure of the gamma effect above. We can treat the opening move as an impulse from overnight news. If the day move tends to counter that systematically, it is highly probable that the long gamma dealers are introducing a stabilizing effect. This means you would expect to show this up as an accompanying short VIX position for the swap dealers under such condition. Whereas if the day move amplify the open, this points to a short gamma position of the street (long VIX). So this correlation measure should move in steps with swap dealers positioning if we are right. And as we can see this is indeed the case, especially since 2014.

For a few days prior-to UK referendum, you must have noticed this phenomenon in practice. Taking a cue from the European markets, the S&P would open down more often than not, only to recover and more almost with statistical consistency during the trading hours. 

The rise of the vol tourists (and the short vol players in general) means watching VIX positioning and tallying it with the underlying moves has now become an important input for investors, even if you have nothing to do with VIX itself.


all data from CFTC reports and Bloomberg