A new arrow in the quiver of the ECB and the BoJ.

Both the European Central Bank and the Bank of Japan are struggling to regain price stability, defined as inflation approximating 2%, as the prospect of reaching that objective moves further and further into the future.

The Bank of Japan has recently brought its interest rate to a negative level, surpassing recent opposition to this measure, while the ECB has nearly promised more action at its March meeting.

Both Draghi and Kuroda-san seem to be confident that they have many arrows in their quiver that they could still use to achieve their objective. To an external observer, the quiver seems, instead, to contain only two tools: more QE and more cuts in interest rates, pushing them further into the unexplored domain of negative rates.

While the two central banks may indeed need to use more QE and further cuts in interest rates to pursue their objectives, there is no doubt that both tools have drawbacks and that their use is not guaranteed to bring back price stability.

As regards QE, the main drawbacks are two: first, the risk of a confusion between monetary and fiscal policy; second, the financial stability consequences of the huge amount of liquidity thrown into the money market. In terms of effectiveness, while it is generally accepted that financial conditions have been impacted in the desired direction by QE, the evidence about an effect on inflation is not overwhelming, at least as yet.

As regards pushing interest rates further down into the negative domain, there are also two main drawbacks: first, the unprecedented and “unnatural” condition of lenders paying borrowers to part with their money creates a general situation of uncertainty, as we do not know its possible consequences, not having seen negative rates for about the last 5 millennia 1; second, very low rates for a very long period risk feeding financial instability. In terms of effectiveness, there is a clearly a dimensionality problem: we know that interest rates cannot be pushed ad libitum into the negative domain, even if we found out that the limit is somewhat lower than zero, and we also know that the elasticity of inflation to interest rate changes is fairly small 2. A limited room to further lower rates together with a small impact of interest rate changes on inflation raise doubts about the progress towards price stability that negative rates can bring.

Since we agree that the two central banks should remain faithful to their mandate of price stability and we are not sure that QE and further cuts of interest rates will be sufficient to achieve this objective, we proposed some time ago in a post of ours a new arrow to be added to the quiver of the two central banks: interventions in the market for inflation derivatives 3.

We are not sure why neither the ECB nor the Bank of Japan have added this tool to their panoply as yet. There are indeed four possible reasons for this:

  1. They are more confident than us on their ability to regain price stability with more QE and/or further interest rate cuts,
  2. They are not so worried by the fact that inflation is and is projected to remain for quite long lower than the “close to 2 per cent” objective,
  3. They have other tools, which we ignore,
  4. They see the balance between advantages and drawbacks of intervention in the market for inflation derivatives as unfavourable, possibly because they have doubts about the transmission mechanism between the inflation derivatives market and the real economy.

We tend to discard the first two possible explanations, we are totally ignorant about the third and therefore we concentrate our attention on the fourth one, trying to analyse what are the advantages and drawbacks of such an innovation in the conduct of monetary policy.

Advantages of intervening in the market for inflation derivatives.

  1. The hawks of the ECB legitimately stress the risk that QE blurs the borders between monetary and fiscal policy. This problem would, of course, not arise with interventions on the market for inflation derivatives. Indeed in terms of “legitimacy” of central bank action, we do not see a difference between intervening in the money market or the FX market – where central banks are known to resort to derivatives 4 – and intervening in the market for inflation derivatives.
  2. There would be practically no creation of additional liquidity.
  3. The interventions would impact directly on inflation expectations and, if they work, on inflation, the objective of the central banks. Analytically their effect could be seen as a shift to the right and up of the Phillips curve and as a cut in the real rate of interest.
  4. The effect on firms and household behaviour could be more direct as it would depend less than QE or interest rate changes on banks intermediation.

Drawbacks of intervening in the market for inflation derivatives.

  1. The main drawback, as we see it, is that the novelty of this instrument means that there are no precedents on which to build and calibrate the interventions. Basically, we do not know whether it will work, including, as mentioned above, because of uncertainty about the transmission mechanism from the market for inflation derivatives to the real economy. This is an important drawback, but it could be lessened by experimenting with the tool and, in any case, using it, if needed, as a complement rather than as a full substitute of QE and interest rate changes.
  2. A related drawback is that the market for inflation derivatives is small and not particularly liquid and therefore the entry of a potentially very big player could destabilize it. However, there is evidence that the market for inflation derivatives is subject to some important biases 5 and therefore central bank interventions could contribute to a more efficient functioning rather than disturbing its equilibrium.
  3. Another drawback is that the central bank would incur substantial financial losses if it would commit, say writing inflation options at a strike price of 1.9%, to a price stability outcome but would eventually not achieve it. But this is just the other side of the coin of putting the central bank´s hand where its mouth is and reinforcing its commitment to achieve its objective of price stability. Of course, the question of whether and to what extent the central bank should be ready to sustain financial losses in pursuit of its price stability mandate is not trivial. However, the experience of many central banks shows that they have performed well in policy terms despite suffering substantial losses, sometimes operating even with negative equity (including Czech National Bank, Bank of Israel, as well as central banks in Chile and Mexico). What these experiences suggest is that – unlike other institutions – the central bank can get away with financial weakness as long as it retains public trust. Thus, while intervening in inflation derivatives markets could eventually lead to losses, this needs not undermine central bank’s credibility. Instead, not intervening and allowing inflation to run below its target for an extended period of time constitutes a direct threat to central bank credibility. This seems to be yet another manifestation of Walter Bagehot’s adage that in times of crisis only the brave plan of the central bank can be the safe plan.

In conclusion, one may have different views about the evaluation of the net effect between advantages and drawbacks of intervening in the market for inflation derivatives. We are of the view, and this is evident from the formulation above, that the net is positive. We are even more convinced, however, that the topic deserves serious consideration and that the current difficulties in regaining price stability, in both the €-area and in Japan, justify considering and, after due consideration, possibly implementing, the new tool that we propose.


This post was jointly written by Juliusz Jabłecki and Francesco Papadia, with the assistance of Madalina Norocea.

We thank J. Alzola for a useful exchange of views about the benefits and drawbacks of interventions in the market for inflation derivatives.


ANNEX

The market for inflation derivatives and commodity prices.

Darvas and Hüttl document, as previous authors before them, that oil price changes today influence inflation expectations, also in the quite distant future. For instance they show that a fall in oil price today can influence annual inflationary expectations up to 6 years in the future. This is, as they note, puzzling since a change in the price of oil today should not have anything to do with inflation far into the future. The only two reasons that could explain a correlation between inflationary expectations today and today´s change in the price of oil are:

  1. Changes in the price of oil have a positive autocorrelation pattern, so that if the price of oil has gone down today it is likely to continue to go down in the future, thus pushing inflationary expectations down.
  2. The price of oil today and inflation expectations may be affected by a common factor (say economic activity).

These two reasons, especially the second one, may have a short-term effect but it stretches imagination that they could be so persistent as influencing expectations over a long time horizon.

To explore the issue a bit further consider the following simple conceptual framework.

Lets define:

          πt  = actual inflation at time t

           πt,c    = core inflation at time t

        Et(πt+τ) = inflation expected at time t for period t+τ , that we assume is equal to breakeven inflation derived from inflation swaps.

We can reasonably assume further that:

Et(πt+τ) ≡Et(πt+τ,c)

i.e. today`s expected headline inflation is the same as today´s expected core inflation, as we cannot properly forecast volatile items like commodities.

We can define further

Et(πt+τ)t+τ ≡ ut,t+τ

i.e. the forecasting error is defined as the difference between inflationary expectation at time t and the actual inflation at time t+ τ.

We can also, simplifying, reasonably assume:

ut+τ= ΔPoilt,t+τ

Where   Poil   is the price of oil, or commodities, and the inflation forecast error is approximately equal to the change in the price of commodities between t and t+τ. 6 It also follows that investing in the variable leg of an inflation swap should give a similar return and variance as investing in a basket of commodities (in both cases one makes, or loses, money depending on whether actual inflation is higher, or lower, than expected inflation because of a change in commodity prices).

It does not follow, though, that Et(πt+τ)and  ΔPoilt,t =ut,t  , are correlated,  i.e. that the expectation of inflation today is correlated with today´s inflation error, which is, in our simplified setting, equal to the change in the price of oil/commodities between yesterday and today. Indeed for  τ large enough, the correlation should be 0, since any autocorrelation that may be present in  ΔPoilt,t =ut,t   should die out over a long enough period. Similarly any common factor impacting both  Et(πt+τ) and  ΔPoilt,t  , i.e. inflation expectations and the change of commodities prices today, should not have an effect on long maturity expectations (with τ large).

So, if indeed we find, like Darvas and Hüttl, a correlation between long horizon inflation expectations and today`s change in the price of oil/commodities we can say that there is something suspicious in inflationary expectations derived from inflation swaps.

To further explore the issue from an empirical perspective we had a quick look at the relationship between inflation expectations derived from inflation swaps and contemporaneous commodity prices changes for the €-area, the US and the UK. To account for a range of commodities – as not only oil impacts inflation developments – we looked at oil futures but also at the JP Morgan Aggregate commodity price index.

The results are shown in the Table 1 below.

Table 1. Regression of changes in inflation expectations against changes in oil/commodity prices*. (January 2014 – January 2016)

 

JP Morgan Commodities Index (% change) Generic 1st Brent futures (% change)
Tenor Slope R-squared Correlation Slope R-squared Correlation
EA 2Y 1.48 0.34 0.59 0.65 0.36 0.60
5Y 1.05 0.32 0.56 0.45 0.33 0.57
10Y 0.74 0.28 0.53 0.30 0.28 0.53
5Yx5Y 0.38 0.16 0.40 0.15 0.15 0.39
US 2Y 3.01 0.65 0.80 1.27 0.66 0.81
5Y 2.00 0.58 0.76 0.87 0.61 0.78
10Y 1.58 0.54 0.73 0.68 0.55 0.74
5Yx5Y 1.32 0.41 0.64 0.54 0.40 0.64
UK 2Y 1.80 0.39 0.62 0.74 0.38 0.61
5Y 1.27 0.40 0.63 0.53 0.40 0.63
10Y 1.02 0.38 0.62 0.41 0.37 0.61
5Yx5Y 0.69 0.27 0.52 0.27 0.25 0.50

* Weekly data. JP Morgan Aggregate Commodity Price Index and oil future prices.

We find that commodity/oil prices explain between 16% and 65% of variation in breakeven inflation rates, with the results differing by currency area and inflation swap tenors. The inflation swap market in the US appears to be most strongly related to developments in commodities prices – with average correlation between weekly changes of inflation expectations and commodities of 0.74 – while markets in the €-area and the UK exhibit average correlations of 0.52 and 0.60 respectively. In line with intuition, correlation levels tend to fall with the maturity of the swap. Indeed, the longer the contract, the less is inflation affected by current movements in commodities prices. Yet, even the correlation levels observed for longer maturities are definitely too strong when judged against the argument that commodity prices today should not affect inflationary expectations. This is all the more evident when one takes into account that, as one could expect for a highly speculative market as that of commodities 7, changes in commodity prices do not exhibit any statistically significant autocorrelation.

Of course, the correlation between oil/commodities price changes and swap-based inflationary expectations identifies a puzzle: why don´t do investors establish trading strategies based on this correlation to gain extra profits? In current conditions, in particular, given the recent falls in the price of oil/commodities, inflation expectations are too low and it would pay to sell aggressively the fixed leg to receive the variable leg, which should provide a profit. If the central bank promises to pay the variable leg against the fixed leg it pushes break-even inflation up (which is the desired outcome) and corrects a market bias.

One possible explanation for the bias could be that the market for inflation derivatives is heavily populated by institutional investors with liabilities linked to inflation that express a structural demand for inflation protection. In order to convince counterparties to sell them the variable leg of the inflation swaps, to be protected against inflation, they have to offer a premium compensating for the variability of inflation. The “structural gap” between demand and supply of inflation protection would be filled, at least in part, by the central bank interventions.

We did not embark into the search of a possible common factor that could influence both inflationary expectations and the change in commodity prices as we expect this search to be futile, given the difficulty of explaining commodity price changes with general macroeconomic factors. We would welcome, though, if any of our readers would embark in the exercise and come up with significant results.

  1. Should we worry that we have about the lowest interest rates in the history of humanity?[]
  2. Will lower rates help the ECB in regaining price stability?[]
  3. A help for the ECB from the derivative market[]
  4. The results of a recent BIS questionnaire indicate that out of a total of 22 central banks queried as many as 8 occasionally intervene in forward markets (2 claim to do so “regularly”) and 7 admit to using also futures and other derivatives, including interest rate swaps and options (with 3 entering these markets “regularly”[]
  5. As documented in “Oil prices and inflation expectations” by ZSOLT DARVAS and PIA HÜTTL,  January 21, 2016 and further analyzed in the Annex to this post[]
  6. To be precise the change in the price of commodities/oil should be multiplied by a parameter k, representing the share of commodities in the consumer price index.[]
  7. Indeed, even if one does not believe in the „Efficient market hypothesis“ the general empirical evidence is that changes of prices established in speculative market show no significant autocorrelation.[]