Measure what is measurable, and make measurable what is not so

Late last year a (long ago) former Reserve Bank economist closed down his blog just like that. I didn’t have anything quite so dramatic in mind.  But after my elderly mother died last August, I got to thinking about the next stage in my life and how I could best use what skills and talents I have.  And before Christmas I’d decided that today would be my last day of high-frequency blogging.  Why today?  Well, for all practical purposes it is the fifth anniversary of the blog; five years today since I left the Reserve Bank.  It wasn’t that I’d run out of ideas or energy, but there is always the opportunity cost to consider, and I’d begun to map out a series of archival-research-based projects (mostly in New Zealand economic history) I wanted to pursue, as well as making more space for some other interests and priorities.   I’d planned to keep writing here perhaps once a week or so.

That is still what I want to move towards at some stage.  But in the middle of the most dramatic economic developments, and wrenching dislocations of economic policy, of our lifetimes it doesn’t quite seem the time to change course just yet.  We’ll see how things go, but for now, quite probably for at least the rest of this year, normal service continues (and in the meantime thanks to all of you who’ve become readers –  numbers that, for a fairly geeky New Zealand focused blog, I never imagined – and to (almost) all who’ve commented, either on the blog itself or directly to me).

And today I’m indulging the inner geek.

One of the most frustrating (to me) aspects of the response to the massive economic dislocation by the government and the Reserve Bank is their utter complacency, bounded by some deep-seated conventional wisdom, that interest rates can’t, shouldn’t, and won’t move down from here –  the OCR having been cut, in the biggest slump in modern history, by a mere 75 basis points.    The Minister of Finance was at it again in his testimony to the epidemic select committee yesterday highlighting the “certainty provided” by the MPC’s commitment not to change the OCR for some considerable time.  I presume he had in mind the idea that the MPC wouldn’t raise the OCR – as if anyone has supposed such increases were at all likely in the foreseeable future –  but he seems totally uninterested in the idea that considerable relief could be provided, and desirable income rebalancings achieved, with much lower interest rates.

Sadly most of our media also seems more interested in channelling conventional wisdoms rather than challenging them.  So I’ve heard MPs and journalists ask the Minister of Finance about commercial rents, but never once about interest rates, even though commercial rents are based on private contracts, whereas the OCR –  which influences a wide range of variable private interest rates – is directly under official control.   If the Reserve Bank refuses to act –  as it appears they do –  the Minister has existing statutory powers to simply override them.  They aren’t powers that should be used lightly, but they were put in the Act for a purpose, and these are pretty extreme times.  If the Minister refuses to use his powers, he makes himself directly party to the choice to hold New Zealand retail interest rates well above where they should be at present.   All while facilitating/encouraging firms and households to take on more debt at these interest rates.

Still, for the time being at least, the MPC’s floor is the floor, and markets have to take that into account in pricing other securities.  This has been an issue in a variety of other markets for some considerable time, particularly since the last recession when various central banks got a point where they were simply not willing to cut further, no matter the economic conditions, or to make the changes (around physical currency) that would have made further deep cuts –  of the sort Taylor rule estimates in the US, for example, suggested economic conditions warranted –  effective.   Central banks fell back on trying to do what they could –  often not much –  with other “unconventional” instruments.

And years ago, a Reserve Bank of New Zealand researcher Leo Krippner, got interested in the question of how one might represent the effectiveness of such policies in a single number.  Whereas in normal times, the OCR itself readily expresses the stance of policy, if there is a self-imposed floor on the OCR but other things are being tried, how best to express the overall stance of policy.

Leo is an expert in yield curve modelling, and got to try to estimate how different the interest rate yield curve was –  especially in the US –  in the presence of the floor on the Fed funds rate, as a result of unconventional policies (“QE”, loosely).     He has published various technical papers in journals, but I think his first paper in the area was published in 2012 in the Reserve Bank’s Analytical Notes series.  I was the editor of that series, and one of my priorities was to ensure that as much as possible was accessible to intelligent lay readers, and my impression is it still remains a good introduction to the work Leo is continuing to do and to update.

One can think of a yield curve of government bond interest rates (from overnight out to, say thirty years) as, broadly speaking,  a series of market implied expectations about the future setting of the official overnight interest rate (the OCR in New Zealand).  All else equal, if the central bank sets a floor on the OCR –  which is regarded as binding and credible – markets pricing say a 10 year bond will rule out any chance of an OCR below that floor, and the 10 year bond rate will be higher (all else equal) than it would be in the absence of such a floor.  On the other hand, if large scale bond purchases by the central bank –  even actively targeting a bond rate (as Japan and now Australia are doing) –  may succeed in lowering to some extent where the bond yields settle, relative to the floor-constrained normal market price.    Leo’s work, calculating what he calls a Shadow Short Rate, attempts to combine the OCR and this effect into a single number.

This work wasn’t very relevant to New Zealand itself for a long time (there were internal sceptics as to whether it should even be done)….and yet now it is. (In his speech a couple of weeks ago the Governor even suggested the Bank might publish a semi-official series of such a measure.)   Leo’s work has been recognised in various places abroad –  cited in public by at least one Fed Reserve president, and honoured by the house journal of the central banking community, Central Banking magazine (for whom I do some reviewing: my light lockdown reading is here).  Leo left the Reserve Bank last year, but is continuing to update his work and earlier this week circulated a note with a Shadow Short Rate series for New Zealand, now that we operate with a formal OCR floor and in the presence of the MPC’s commitment to buy $30 billion of government bonds over the coming year. (The quote that heads this post, from Galileo, appears upfront in Leo’s note.)

Leo’s modelling estimates that the overall stance of monetary policy is “approximately equal to an OCR of -0.48 per cent”, compared to an OCR itself of 0.25 per cent.    If so, that suggests a small but somewhat useful contribution from the “unconventional monetary policy” or UMP.  Small because even a total of 150 basis points of effective easing is tiny by comparison to the scale of the economic shock.

There are a number of caveats to this modelling.  Leo articulates some of them himself

  1. The SSR is an estimated value rather than a setting like the OCR or an observed market short rate. Hence, any SSR series will (unavoidably) vary with the model specication and data used for its estimation. The choices I have made for the SSR series in this note have produced more favorable properties than alternatives for the United States,4 but the magnitudes of negative SSR estimations can easily vary by fractions of a percentage point on re-estimations, and sometimes more for UMP periods if the lower-bound setting in the model needs revisiting in light of central bank communications and/or actions. My model for New Zealand currently uses a lower-bound setting of 0.25%, consistent with the RBNZs 16 March 2020 forward guidance.
  2. Related to estimation, in the present global UMP environment compared to previous years (e.g. for the United States easing/tightening cycle) yield curves may no longer capture sufficient information to quantify the stance of monetary policy. The reason is that yield curves in New Zealand and around the world are now very at(i.e. longermaturity rates are close to shorter-maturity rates). Only time (and updated analysis) will tell on this aspect, and model refinements may be needed.
  3. The SSR is not a market rate at which borrowers and lenders can transact, particularly in UMP times when the OCR and short rates will remain close to zero while the SSR may become increasingly negative. Hence, SSR declines in UMP times will not result in the same cash flow effects from interest payments and receipts as OCR cuts in CMP times, so the SSR transmission to the economy may differ from at least that perspective.

I went back to Leo and asked about how safe it was to use a floor of 0.25 per cent for New Zealand (after all, we’ve seen more than a few policy lurches from the Reserve Bank over the last year, in fact over the last few weeks, and some other central banks have – eventually –  shown a willingness to take their policy rates lower, in the Swiss National Bank’s case as low as -0.75 per cent.  Leo noted that if one were to assume the New Zealand floor was anything like that low, the SSR for New Zealand would still be much the same as the OCR itself.

He went to say (with his permission to quote)

The longer answer is: should one use an assumption of the effective  lower bound in the model (perhaps something like your -0.7, but also  see further below), or should one use a value that central banks have  indicated they would go to? Across the different economies I’ve  modeled, I’ve noticed that the different yield curves seem to use the latter. For example, the US yield curve data has pretty much respected  the lower point of the 0-25 bp range, and the euro area data (OIS  rates) has pretty much respected the incremental settings of the ECB  policy rate. In particular, neither have had longer-maturity yields  rush to, say, the -75 bps of Switzerland.

So, based on what I’ve seen with the data, I set the lower-bound for  each economy according to the central bank’s indications. And then I  lower that if/when the policy rate gets set lower (which, again, seems  to be how the yield curves behave – perhaps not rational though, I agree).

To which my response in turn was to note that in at least some of those overseas central banks –  the Fed in particular –  the stated floor has been consistently applied for a decade, while the Reserve Bank of New Zealand has no track record in this area at all –  indeed in that same speech a few weeks ago the Governor was openly talking of negative rates as a possibility.  A rational investor in the current climate, with a Governor prone to lurches (and, actually, in the presence of override powers), might not price in a 100 per cent chance of the MPC sticking to its word.  Time will tell.

My bigger unease about this work, stretching all the way back to that 2012 paper (where I recognise a couple of sentences I insisted on being added) is the third in Leo’s own list of caveats above: the SSR is not a rate that can be transacted, and if much of borrowing and lending in an economy is either at (or swapped back to) or very close to a floating rate, the actual OCR (self-imposed constraint and all) will be a lot more important than the SSR estimate might suggest.  In the US, for example, a lot of mortgage lending takes places at very long-term fixed rates (and so what happens to very long-term bond and swap rates has a direct transmission mechanism).  That is much less so in New Zealand (or Australia or the UK).  That is consistent with some of the doubts I’ve expressed in earlier posts around quite what difference the Bank’s large scale asset purchase programme really makes where it matters (thus, on my telling, the main advantange of that programme is that it should reveal quite quickly how severe those limitations are and turn the focus back on the OCR and that self-imposed floor).

This stuff won’t be everyone’s cup of tea, but I’m really glad to see that Leo is continuing with this work –  he says he will provide updated estimates on his website each month for the time being –  and will be interested to see whether the Reserve Bank follows through on the suggestion of a quasi-official series (for what its worth, I would suggest they not do so, and use work such as Leo’s as a reference point for commentary/research when it is relevant).