There was a time when it was nothing short of economic blasphemy and statist apostasy to suggest three things: i) that the Fed's canonic approach to monetary policy, in which Stock not Flow was dominant, is wrong (as we alleged, among many other places, here); ii) that the Fed is monetizing the deficit, thus enabling politicians to conceive any idiotic fiscal policy: the Fed will always fund it no matter how ludicrous, converting the Fed effectively into a political power and destroying any myth of its "independence" (as we alleged, among many other places, most recently here in direct refutation of Bernanke's sworn testimony); and iii) that by overfunding bank reserves, the same banks are left with one simple trade - to frontrum the Fed in its monetization of the long-end, in the process destroying the bond curve's relevance as an inflationary discounting signal, with more QE, leading to tighter 10s, flatter 10s30s, even as the propensity for runaway inflation down the road soars, in the process eliminating any need for the massively overhyped, and much needed to rekindle animal spirits "rotation out of bonds and into stocks" trade (as we explained, first, here). Well, that time is now officially over, with that stalwart of statist thinking, JPMorgan, adopting all of the above contrarian views as its own, and admitting that once again, the Fed and conventional wisdom was wrong, and fringe bloggers were right all along.
And while we recreate the piece in its entirety below, here is the punchline:
Since the Lehman crisis, the Fed has been purchasing Treasuries and Agencies at a $500bn per year pace. This flow, which is equivalent to around 3.5% of US GDP, has offset more than a third of the government deficit since the end of 2008. In other words, QE purchases meant that the QE-adjusted government deficit has averaged 5.8% of GDP since the end of 2008 instead of 9.3% for the actual government deficit. This week’s Fed announcement means that this QE flow will double from a $500bn pace currently to $1tr. Coupled with a projection of a lower government deficit next year, to around 6% of GDP, this means that QE will offset almost all of next year’s government deficit.
Who knew that in the internal JPM thesaurus, "offset" was equivalent to "monetize"... But we'll take it.
From JPM's Nikolaos Panigirtzoglou:
Flows & Liquidity: QE's Stock Effect
- The Fed announced this week an extension of operation twist bringing the total amount of net bond purchases expected for next year close to $1tr.
- The excess reserves are likely to increase by the same amount, from $1.4tr to $2.4tr, adding 70% extra liquidity into the banking system.
- This extra liquidity has the potential to suppress bond yields in addition to the traditional demand effect of QE. This demand effect is well documented and understood. As the Fed buys Treasuries and Agencies it offsets government supply, exerting downward pressure on yields.
- What is often overlooked and surely less well understood is the liquidity effect of QE. What is this liquidity effect?
- As we highlighted in F&L Oct 5th, QE by itself does not affect the overall supply of collateral, as government bonds are replaced with reserves. But it does affect the relative pricing of collateral by creating a shortage/expensiveness of government bonds vs. cash (reserves). QE is a good example of the distinction between collateral scarcity and collateral shortage. QE does not create a shortage, as the overall supply of collateral is unchanged, but it does create scarcity by making one form of collateral (government bonds) more expensive relative to another (zero yielding reserves).
- The most important difference between the liquidity effect and the demand effect is that that the former operates via stocks while the second operates via flows. As the Fed buys Treasuries and Agencies both effects are in operation.
- The Fed offsets government supply via bond purchases and at the same time the stock of zero-yielding excess reserves rises relative to the stock of government and government related bonds held outside the Fed.
- If the Fed were to stop purchasing bonds on net, the demand effect would disappear but the liquidity effect would remain. For as long as the Fed maintains its stock of QE and its zero-interest-rate policy, the stock of zero-yielding excess reserves will continue to affect the relative pricing of government bonds.
- Since the Lehman crisis, the Fed has been purchasing Treasuries and Agencies at a $500bn per year pace. This flow, which is equivalent to around 3.5% of US GDP, has offset more than a third of the government deficit since the end of 2008. In other words, QE purchases meant that the QE-adjusted government deficit has averaged 5.8% of GDP since the end of 2008 instead of 9.3% for the actual government deficit. This week’s Fed announcement means that this QE flow will double from a $500bn pace currently to $1tr. Coupled with a projection of a lower government deficit next year, to around 6% of GDP, this means that QE will offset almost all of next year’s government deficit.
- What about the liquidity effect? The liquidity effect is also set to intensify next year. To quantify this liquidity effect we follow the approach by Krogstrup-Reynard-Sutter “Liquidity Effects of Quantitative Easing on Long-Term Interest Rates”, January 2012. This paper proxies the liquidity factor using the ratio of excess reserves divided by the stock of government securities held outside the Fed.
- Excess reserves should rise by $1tr, to $2.4tr by the end of 2013. At the same time the stock of Treasury and Agency bonds should rise by around $675bn (net supply of $900bn for Treasuries and shrinkage of $225bn for Agency debt and Agency MBS). This means that the stock of government and governmentrelated bonds held outside the Fed will likely decline by $375bn ($1tr of QE purchases minus $675bn of net issuance for Treasuries and Agencies).
- As a result, the liquidity effect ratio, i.e. the ratio of excess reserves divided by the stock of government securities held outside the Fed, is set to rise from 10% currently to 17% by the end of 2013.
- What does this mean for Treasury yields?
- We have tried to quantify the impact of the government deficit and other factors on bond yields in the past by fitting a regression model using data going back to 1959 (see “A Fair Value Model for US Bonds, Credit, and Equities”, Panigirtzoglou and Loeys, Jan 2005). The model values the 10-year UST yield as a function of long-term inflation expectations, the real Fed funds rate, inflation volatility, and three major components of the demand for capital in the US market, the government, corporates, and emerging market issuers in dollars. We measure these by the government deficit, the corporate financing gap (i.e. the difference between capex and cash flows), and the EM current account balance, all as a % of US GDP. Higher deficits by governments and corporates push up yields as overall demand for capital rises. External surpluses of EM countries push US yields down because of the repayment of dollar-denominated debt and the dollar asset accumulation by their central banks.
- As explained above, we capture the QE flow effect by subtracting bond purchases by the Fed from the overall government deficit (as % GDP). We capture the liquidity effect by the ratio of excess reserves divided by the stock of government securities held outside the Fed. In theory, these two effects depress bond risk premia, i.e. term premia.
- Beyond bond purchases, the Fed is also affecting bond yields via its communication. That is, via its communication the Fed affects the path of expected future short rates. The literature has tried to quantify this affect via event studies. This is because, in theory, signaling should be mostly present at announcement times.
- Given the difficulties that these event studies have with choosing the appropriate window around the announcement date, we instead prefer to proxy this signaling/communication effect more directly via Fed’s forward guidance. In particular, the variable we use is the time period (in months) over which the Fed has committed to keep rates low and is being constructed and used by US Fixed Income Research team, Terry Belton et al.
- The coefficients of the 10y UST regression model are shown in Table 1 while the fitting errors are shown in Figure 2. The model has provided a good fit to quarterly averages of the 10-year UST yield over the past 50 years with an average absolute error of 57bp and R-sq of 86%.
- All coefficients are statistically and economically significant, suggesting that both QE flow and QE stock effects are present. This is admittedly a mechanical exercise with imperfect proxies for QE flow, QE liquidity and Fed communication effects. But the model illustrates how important and persistent the liquidity effect of QE can be as the Fed continues to expand the stock of excess reserves next year. It also casts doubt to the idea, advocated by Professor Michael Woodford via his speech in Fed’s annual Jackson Hole conference, that the portfolio balance effects of QE purchases are minimal and that the Fed’s impact on bond yields stems almost exclusively from forward guidance.
* * *
Are these the first rumblings of mutiny on the Titanic, we hear?