Price discovery on traded inflation expectations: does the financial crisis matter?

Abstract

We analyze contributions of different markets, related by an approximate arbitrage relationship, to price discovery on traded inflation expectations and how it changed during the financial crisis. We use a new high-frequency data-set on inflation-indexed and nominal government bonds as well as inflation swaps to calculate information shares of break-even inflation rates in the euro area and the USA. In the euro area, for maturities up to 5 years new information comes from both the swap and the bond markets. For longer maturities, the swap market provides less and less information in the euro area. In the USA, the bond market dominates the price discovery process for all maturities. The severe financial crisis that spread out in Autumn 2008 drove a wedge between bond and swap break-even inflation rates in both currencies. Price discovery ceased to take place on the swap market. Disruptions coming from the short-end of the market even separated price formation on both segments for maturities of up to 6 years in the USA. Against the backdrop of the most severe financial crisis in decades, contributions to price formation concentrated a lot more on the presumably safest financial instrument: government bonds.

Keywords:

• inflation
• bond markets
• swaps
• price discovery
• financial crisis

JEL Classification:

• E43
• F37
• G12

Acknowledgements

We thank Christoph Fischer, Joachim Grammig, Joseph Haubrich, Thomas Laubach, Franziska Peter, Stefan Reitz, the two referees as well as seminar participants at the Banco de Espana-Bank of Canada workshop on Advances in Fixed Income Modelling 2011, the RES in Surrey 2010, the IFC in Basel 2010, the EEA in Barcelona 2009, the International Conference on Macroeconomic Analysis and International Finance in Crete 2009 and at Deutsche Bundesbank for helpful comments. All remaining errors are ours. The opinions expressed in this paper are those of the authors and do not necessarily reflect the opinions of the Deutsche Bundesbank or its staff.

Notes

1. See Campbell and Shiller (1996) for an overview of early linkers, including issues from emerging markets.

2. As of January 2009. Relative to its outstanding marketable debt, the UK is still the largest issuer, with a share of of 28%.

3. A repo or repurchase transaction is a standard technique to fund purchases of financial instruments, which serve as collateral themselves. See Buraschi and Menini (2002) for a discussion of specialness.

4. Liu, Longstaff, and Mandell (2006) deal comprehensively with liquidity and default risk in interest rate swaps; to our knowledge, no similar study exists for inflation swaps.

5. See Armann, Benaben, and Lambert (2005, p. 94) and Deacon, Derry, and Mirfendereski (2004, chapter 9) for a lucid treatment.

6. France as a vanguard on issuing inflation-linked debt in the euro area has set the standard of linking claims on an index excluding tobacco products, thus controlling for administered prices to some degree.

7. As reported on Bloomberg. For longer horizons, e.g. the French bond expiring in 2040, we would need to interpolate between infrequently traded 30- and 40-year inflation swap rates, which is prone to errors.

8. All bonds except for the 5-year tenors are off-the-run.

9. Intervals with only either a bid or an ask entry are eliminated.

10. Calculating true zero coupon yields for our high-frequency bond price data is nearly impossible, as necessary interpolations are prone to contaminate the marginal price change of a single bond.

11. Only on coupon dates, there is no bias as inflation is paid out. For an explanation and visualization of seasonality in CPI, see, for example, Peat and Segregeti (2008, pp. 183).

12. See Eijsing, Garcia, and Werner (2007) for further explanations of the adjustment method.

13. See Hasbrouck (1995), Baillie et al. (2002), Mizrach and Neely (2008) or Grammig and Peter (Forthcoming) for derivations and a discussion of both measures.

14. The covariance matrix is and the lower-triangular matrix is

15. Reestimation of the VECM and the Hasbrouck information shares with the overnight returns substituted with the mean return of the following day showed virtually no influence on the parameters. We thank Franziska Peter and Joachim Grammig for performing the estimation using their Gauss procedures.

16. Since not all time series showed linear trends, we did not include them in our cointegration analysis. Nevertheless, once included the results did not change qualitatively.

17. Wald tests on the equality of the ratio of adjustment coefficients or the ratio of the γ, respectively, are rejected at conventional test sizes.

18. This could be traffic infrastructure projects or hospitals. Many are regulated to adjust their prices by the inflation rate or receive a share of their contract payments directly inflation linked, typically the remuneration for operating expenses. See Grath and Windle (2006).

19. We performed unit root tests for all series. The number of lags recommended by the Schwarz information criterion did not exceed 15 or 22 lags, where a number of intermediate lags were excluded.

20. Wald tests on the equality of the information shares or the ratio of the γ for the summer and autumn period, respectively, are rejected for both currency areas.

21. Numbers and more information can be found in Schulz and Stapf (2009).