Over the last 1-2 decades so-called DSGE (dynamic stochastic general equilibrium) models have become the dominate research tool for central banks around the world. These models certainly have some advantages, but it is notable that these models generally are models without money. Yes, that is right the favourite models of central bankers are not telling them anything about money and the impact of money on the economy. That is not necessarily a major problem when everything is on track and interest rates are well above zero. However, in the present environment with interest rates close to zero in many countries these models become completely worthless in assessing monetary policy.
I was therefore pleasantly surprised this week when I discovered a relatively new working paper – “The impact of QE on the UK economy — some supportive monetarist arithmetic” from the Bank of England (BoE) in which the authors Jonathan Bridges and Ryland Thomas estimate what they call a “broad” monetarist model and use their model(s) to evaluate the impact on the UK economy of BoE’s quantitative easing over the past four years. Here is the paper’s abstract:
“This paper uses a simple money demand and supply framework to estimate the impact of quantitative easing (QE) on asset prices and nominal spending. We use standard money accounting to try to establish the impact of asset purchases on broad money holdings. We show that the initial impact of £200 billion of asset purchases on the money supply was partially offset by other ‘shocks’ to the money supply. Some of these offsets may have been the indirect result of QE. Our central case estimate is that QE boosted the broad money supply by £122 billion or 8%. We apply our estimates of the impact of QE on the money supply to a set of ‘monetarist’ econometric models that articulate the extent to which asset prices and spending need to adjust to make the demand for money consistent with the increased broad money supply associated with QE. Our preferred, central case estimate is that an 8% increase in money holdings may have pushed down on yields by an average of around 150 basis points in 2010 and increased asset values by approximately 20%. This in turn would have had a peak impact on output of 2% by the start of 2011, with an impact on inflation of 1 percentage point around a year later. These estimates are necessarily uncertain and we show the sensitivity of our results to different assumptions about the size of the shock to the money supply and the nature of the transmission mechanism.”
I draw a number of conclusions from the paper. First, the authors clearly show that monetary policy is highly potent. An increase in the money supply via QE will increase nominal GDP and in the short-run also real GDP. Second, the paper has a very good discussion of the monetary transmission mechanism stressing that monetary policy does not primarily work through the central bank’s key policy rate, but rather through changes in a number of asset prices.
The authors’ discussion of the transmission mechanism and the empirical results also clearly refute that money and other assets are perfect substitutes. Therefore, unlike what for example has been suggested by Steven Williamson open market operations will impact nominal income.
I particularly find the discussion of the so-called buffer stock theory of money interesting. The Buffer stock theory, which was developed by among others David Laidler, has had a particularly large impact on British monetarists and in general Bridges and Thomas seem to write in what Tim Congdon has called the British monetarist tradition which stresses the interaction between credit and money more than traditional US monetarists do. British monetarists like Tim Congdon, Gordon Pepper and Patrick Minford – as do Bridges and Thomas – also traditionally have stressed the importance of broad money more than narrow money.
Furthermore, Bridges and Thomas also stress the so-called “hot-potato” effect in monetary policy, something often stressed by Market Monetarists like Nick Rowe and myself for that matter. Here is Bridges and Thomas:
“A further key distinction is the difference between the individual agent’s or sector’s attempt to reduce its money holdings and the adjustment of the economy in the aggregate. An individual can only reduce his surplus liquidity by passing that liquidity on to someone else. This is the genesis of ‘hot potato’ effects where money gets passed on among agents until ultimately the transactions underlying the transfers of deposits lead to sufficient changes in asset prices and/or nominal spending that the demand for money is made equal to supply.”
Even though I think the paper is extremely interesting and clearly confirms some key monetarist positions I must say that I miss a discussion of certain topics. I would particularly stress three topics.
1) A discussion of the property market in the UK monetary transmission mechanism. Traditionally UK monetarists have stressed the importance of the UK property market in the transmission of monetary policy shocks. Bridges and Thomas discuss the importance of the equity market, but the property market is absent in their models. I believe that that likely leads to an underestimation of the potency of UK monetary policy. Furthermore, Bridges and Thomas use the broad FTSE All Shares equity index as an indicator for the stock market. While this obviously makes sense it should also be noted that the FTSE index likely is determined more by global monetary conditions rather than UK monetary conditions. It would therefore be interesting to see how the empirical results would change if a more “local” equity index had been used.
2) The importance of the expectational channel is strongly underestimated. Even though Bridges and Thomas discuss the importance of expectations they do not take that into account in their empirical modeling. There are good reasons for that – the empirical tools are simply not there for doing that well enough. However, it should be stressed that it is not irrelevant under what expectational regime monetary policy operates. The experience from the changes in Swiss monetary and exchange rate policy over the last couple of years clearly shows that the expectational channel is very important. Furthermore, it should be stressed that the empirical results in the paper likely are strongly influenced by the fact that there was significant nominal stability in most of the estimation period. I believe that the failure to fully account for the expectational channel strongly underestimates the potency of UK monetary policy. That said, the BoE has also to a very large degree failed to utilize the expectational channel. Hence, the BoE has maintained and even stressed its inflation target during the “experiments” with QE. Any Market Monetarist would tell you that if you announce monetary easing and at the same time say that it will not increase inflation then the impact of monetary easing is likely to be much smaller than if you for example announced a clear nominal target (preferably an NGDP level target).
In regard to the expectational channel it should also be noted that the markets seem to have anticipated QE from the BoE. As it is noted in the paper the British pound started to depreciate ahead of the BoE initiating the first round of QE. This presents an econometric challenge as one could argue that the start of QE was not the time it was officially started, but rather the point in time when it was being priced into the market. This of course is a key Market Monetarist position – that monetary policy (can) work with long and variable leads. This clearly complicates the empirical analysis and likely also leads to an underestimation of the impact of QE on the exchange rate and hence on the economy in general.
3) The unexplained odd behavior of money-velocity. One of my biggest problems with the empirical results in the paper is the behaviour of money-velocity. Hence, in the paper it is shown that velocity follows a V-shaped pattern following QE. Hence, first velocity drops quite sharply in response to QE and then thereafter velocity rebounds. The authors unfortunately do not really discuss the reasons for this result, which I find hard to reconcile with monetary theory – at least in models with forward-looking agents.
In my view we should expect velocity to increase in connection with the announcement of QE as the expectation of higher inflation will lead to a drop in money demand. So if anything we should expect an inverse V-shaped pattern for velocity following the announcement of QE. This is also quite clearly what we saw in the US in 1933 when Roosevelt gave up the gold standard or in Argentina following the collapse of the currency board in 2002. I believe that Bridges and Thomas’ results are a consequence of failing to appropriately account for the expectational channel in monetary policy.
A simple way to illustrate the expectational channel is by looking at Google searches for “QE” and “Quantitative Easing”. I have done that in Google Insights and it is clear that the expectation (measured by number of Google searches) for QE starts to increase in the autumn of 2008, but really escalates from January 2009 and peaks in March 2009 when the BoE actually initiated QE. It should also be noted that BoE Governor Mervyn King already in January 2009 had hinted quite clearly that the BoE would indeed introduce QE (See here). That said, M4-velocity did continue to drop until the summer of 2009 whereafter velocity rebounded strongly – coinciding with the BoE’s second round of QE.
Despite my reservations about parts of Bridges and Thomas’ paper I think it is one of the most insightful papers on QE I have seen from any central bank and I think the paper provides a lot of insight to the monetary transmission mechanism and I think it would be tremendously interesting to see what results a similar empirical study would produce for for example the US economy.
Josh Hendrickson has a great post on his blog The Everyday Economist on the monetary transmission mechanism.
See also my earlier post “Ben Volcker” and the monetary transmission mechanism.