Hi there,I've got two questions on inflation modelling:1.) Market Data comes from ICAP, i.e. break even rates (zero coupon) and cap/floor (YoY) prices.Using the break even rates it's possible to derive approximations for the YoY rates by neglecting convexity adjustments. On the other side, one can use cap and floor prices for a given strike (1.5%) and apply put-call-parity in order to calculate implied YoY rates. Problem is, that the YoY rates computed with these two approaches, differ a lot, i.e. up to 15 basis points, which is imho too much.So, did I make a mistake here? Or is the market too illiquid, so that break even rates and cap/floor prices are out of sync?2.) Inflation model is Jarrow/Yildirim as described in Mercurios inflation paper (Mercurio's inflation paper).I try to fit the model to a chosen set of cap/floor (YoY) prices, for example caps and floors with strike 1.5% and maturity up to 15y. I'm not overly excited about the quality of the fit. And, in addition, calibration turns out to be a pain, since there are many local minima.Should I try to "harmonize" the market data (i.e. apply something like smoothing so that break even rates and cap/floor prices correspond to each other), in order to get a better calibration result? Would it be a good idea to work with another model instead of Jarrow/Yildirim?Thanks in advance!

1. Have you tried making your calculations using bids and offers? IMHO, what you're seeing is, in fact, a function of illiquidity, rather than anything more fundamental, but it should be reasonably easy to test this.2. I guess it depends on what you're trying to achieve. In general, it's a thankless task, fraught with all sorts of issues, which will force you, willy-nilly, to make all sorts of arbitrary assumptions.

Problem with the Jarrow-Yildirim model is that several parameters from the real rate process have to be calibrated, while in fact the real rate process is not really necessary.It has advantages to model the inflation directly, which is done in Jackel & Bonneton and in Dodgson & Kainth.

Thank you, Church and Martinghoul!Now I'm reading the paper by Dodgson and Kainth, which I find is very well written.I have a small problem with a remark that they made to the convexity correction formula:Dodgson and Kainth say "As the correlation parameter tends to be positive, we see that this covariance is negative, and the 'convexity correction' reduces the value of the year-on-year swap."But obviously the last term in bracketscan get so negative that the convexity correction is positive.So can it be that they have a typo in their formula or have they just made a little mistake with that comment? I'm just trying to understand the paper.Thanks in advance,BustopherJones