The impact of NIRP on inflation: some empirical evidence

Another day, another round of lousy US data, another sell off at the long end of the curve, and another big move in the SPX.  Only this time the equity market, perhaps buoyed by tomorrow's quarterly option expiration, upset the recent playbook by surging to the upside.  It's probably not worth getting too worked up about the change from the recent market dynamic, given the proximity to triple witching.   What we can say is that yet again, hopes for a bang up quarter of US growth appear to be slowly ebbing away; the Atlanta Fed's Q3 GDP Now forecast has now slipped to 3%, the lowest level of this forecast cycle.

Hmmm, at this rate it won't be long before the represserati start calling for negative interest rates.  Oh wait, Ben Bernanke has already begun laying the groundwork for just such a call.  Macro Man had to laugh when he read Bernanke's piece; not only did he trot out the usual arguments from the modern central banking playbook, including a touching belief in the efficacy of real interest rates, but he failed to furnish any proof that they actually work.

Given that we now have a reasonable sample of negative interest rates in different countries, Macro Man thought that it might be useful to investigate the track record of negative rates in boosting inflation.   He looked at headline CPI in five economies: Switzerland, Denmark, the Eurozone, Sweden, and Japan, measuring it before and after the introduction of negative rates.

He looked at the 12 month moving average of m/m changes in CPI rather than y/y changes because it made it easier to adjust for Japan's consumption tax hike in April 2014.  It's also worth recalling that Switzerland has had a complicated policy regime, targeting both interest rates and the currency at various times, and intervening in both markets regardless of the ostensible target.   Macro Man dates the imposition of NIRP from the time when the 2 week repo rate went decisively negative in October 2011.

CPIs for the countries in his sample are displayed on the chart below; dotted lines represent the CPI before negative interest rates, and solid lines afterwards.

 
The chart is a little busy, with plenty of overlap.  A summary of the data is set out in the table below.

As you can see, there is scant evidence that negative rates have boosted CPI; in 4 of the 5 countries in our sample, inflation has been lower since the imposition of NIRP than it was in the two years prior.   The margins aren't small, either; the average delta of 0.083% per month works out to more than 1% lower on annualized CPI inflation.   Small wonder that central bankers prefer to talk about their theoretical frameworks when the empirical evidence is so poor!

 Now, a valid criticism of this approach is that trends in headline CPI are dominated by oil prices, which are frankly out of the control of individual national central banks.   As such, a friendly theoretician might claim that the results above tell you more about spare capacity in oil production than they do about the effectiveness of NIRP.  While looking at trends in core CPI might mitigate the issue, the varying calculation methods/weights might render it something of an apples-to-oranges comparison.  (Also, Macro Man didn't have all the data to hand.)

Instead, he decided to look at trends in CPI data relative to a control group, preferably one not impacted by tax or currency moves of a single country.  He decided upon the aggregate OECD CPI index, which represents a range of large economies.  Surely if NIRP were effective in pushing underlying inflation higher, we should see CPI in NIRP countries rise more (or, more to the point, fall less) than in the OECD as a whole.

Sure enough, the chart of relative CPI does indeed look different to that of headline CPI, with all of the inflation rates further from their nadir than in the previous iteration of the study.

When we crunch the numbers, however, the results are still not particularly impressive.   Again, four of the five countries in our sample have seen their inflation rates decline relative to the OECD as a whole since the imposition of NIRP, though the exception (Sweden) manifests the largest absolute value delta in the study.

At best, you could say that the results are mixed, and even that's being very charitable.   Sweden is, after all, a country with strong GDP growth, a large professional immigrant population, and a massive housing shortage in its largest city.  But of course, the authorities don't include house prices in their inflation measures, and didn't that work out well for the US last time?

That four of the five countries have seen inflation fall not only in absolute terms, but also relative to a group of large advanced economies that for the most part still have positive interest rates, does little to suggest that negative rates generate the intended impact upon inflation.   If anything, even the second study probably understates the issue, since the lousy CPIs of the NIRP countries are also included in the OECD measure.

The negative side effects, meanwhile, have been notable, even if Mr. Bernanke and other academics are happy to wave them away with a brush of the hand.   Then again, given that many of them enjoy juicy government-sponsored pensions, why should we expect them to know or care about the (as yet largely unpublicized) negative impacts on private sector pensions?

The most effective arrow in many policymakers' rhetorical quivers since the crisis has been the counterfactual.  ("Yes, things are terrible, but just imagine how much worse they'd be if we hadn't had the courage to charge savers a tax rather than paying them interest!")  In the case of negative rates, however, there is some empirical evidence that we can measure against some sort of control group.   And the verdict is not only have these courageous measures not made things better, they appear to have made them a little worse.