This is just one of many examples out there discussing the implication of very low interest rates on bonds. What I have never seen from these same people is a discussion on when exactly, or even generally bonds become money-like, and the implications on evaluating monetary policy prior to 2008. I am going to run through this pretty sloppily, the assumptions and numbers here are to get a feel for the scale, and not to put a final number or even range on it.Think of a liquidity trap as a situation where investors view central bank money and treasury debt as perfect substitutes. Such a condition likely never holds exactly in reality, but that's neither here nor there. Empirical relevance is possible even if liquidity trap conditions hold approximately. In any case, if money and bonds are close to perfect substitutes, then the composition of total government debt (between money and bonds) has little economic significance (in the same way that the composition of the money supply between $5 and $10 bills does not matter).
I doubt that anyone would defend the position that bonds are 0% like money at 0.001% and an almost perfect money substitute at 0.000%. Nor do I think that anyone would defend the position that bonds with a 20% yield are good money substitutes (though under some conditions they might be, but for now I am talking about the US from 2000-present).
So my first assumption is that the moneyness of bonds starts to become significant around a 2.5% yield. This is somewhat arbitrary, but a defensible starting position. The Fed has an inflation target around 2%, which makes bonds paying that rate or less (expected) losers in real terms. If there is an inflection point where bonds start to become more like money the expected inflation rate is a reasonable guess as to that point. I use 2.5% because inflation had typically been above 2%, and because it makes it nice and easy with the fed dropping rates by a quarter point at a time along with assumption #2.
Second assumption, debt is distributed pretty evenly by duration. Short term bonds should cover 8-10% of all debt in the US, and only publicly held debt matters for this exercise. I am going to use 10%, and the US had a publicly held debt of between 3.5 and 4.0 Trillion dollars from 2000-2003. So 3.75 trillion for an average.
So starting with bonds being 0% money at 2.5%+ and becoming 100% money at 0%, each drop of 0.25% represents 10% more "moneyness" (yes, laziness and linearness, they go arm in arm)- which effects 10% of the total debt. So a 0.25% drop below 2.5% means 1% of total publicly held debt becomes money. 1% is 35-40 billion dollars.
Using the Federal funds rate (monthly data)- October 2001 was the first breach of the 2.5% fake barrier since 1962. The rate fell to 1%, so the BotE calculation implies a 210-240 billion dollar increase in the money supply over that time period. To put that in perspective the adjusted monetary base went up by less than 200 billion dollars during that time period. This effect, if real and within half an order of magnitude of this calculation would mean the Fed loosened during this period 2 to 4 times as much as conventional estimates.
What else does it imply though? Well when the Fed started raising rates in Mid 2004 they were tightening twice as much as conventional estimates (until the rate broke 2.5%, then it magically stopped).
The real kicker though comes in 2008 when the Fed again drops below 2.5%, because total government debt was 50% higher in 2008 than in 2003 the effect would be ~50% higher. So a 50-60 billion dollar increase per 0.25% rate drop. By the time the Funds rate was bouncing around below 0.25% in December 2008 this effect would have added $600 billion dollars or so in "money" in addition to the $800 billion or so the monetary base increased.
The final implication is even more dramatic as federal debt held by the public has shot up to almost $13 trillion in 2014. That would be another $700 trillion in money "created" on top of the $1.6 trillion added to the monetary base.
This seems to me to be a very straight forward consequence of the idea that bonds and money are excellent substitutes near the zero bound. You can argue against some of the methods I used, for instance you could disagree with using the federal funds rate instead of actual US Treasury rates, or disagree with the %s used. Those disagreements would only effect the timing and magnitude of the loosening/tightening, and not the core concept.