With the passing of 30 June we have entered another busy period for year-end valuations. One of the most common questions we are asked at these important balance dates is “why is there a difference between the bank valuation and the Hedgebook valuation (or any other system’s valuation for that matter)?” The question is most commonly posed by auditors. It is probably not surprising that auditors want a perfect reconciliation between the client’s information and their own independent check but alas it will never come to pass. In this article we consider a selection of reasons that can lead to differences in valuations.

Although the modelling of interest rate swap valuations is relatively unchanged over many, many years there are subtle differences that will result in no two valuations being the same. From an interest rate swap perspective the most likely source of valuation differences is the construction of the zero curve. The zero curve is used to estimate the future cashflows of the floating leg of the swap, as well as the discount factors used to net present value the future values of the cashflows (both fixed and floating legs).

The underlying interest rate inputs into the zero curve construction (deposit rates, bank bills, LIBOR, futures, swap rates, etc.) may be slightly different between one system and another. Unlike official rate-sets such as BKBM, BBSW, LIBOR, EURIBOR, CDOR, etc. there is no one source for zero curves.

The mathematical technique to combine the various inputs into a zero curve can also differ (linear interpolation, cubic spline). These types of differences can lead to discrepancies between one valuation and another. Although on a percentage of notional basis the discrepancies are small, the monetary differences can become material if the notional of the swap is big enough i.e. a $500 difference on a $1 million interest rate swap becomes a $50,000 difference on a $100 million swap – a number that will draw attention but in reality is still immaterial.

The timing of the market snapshot for closing rates can be different, too. For example, Hedgebook uses New York 5pm as the end of day for valuation purposes, therefore, if the Hedgebook valuation is compared to a system that uses, say, Australia 5pm as its rate feed, then any movement in the intervening period will cause differences in valuations.

What we have talked about above is premised on the fact that two identical deals are being valued against each other. By far the most common reason for different valuations lies in human error around the inputting of a deal. From an interest rate swap perspective, the rate-set frequency (monthly, quarterly), accrual basis, business day conventions, margins on the floating leg are all possible areas which can result in valuation differences. The most common input error we come across relates to amortising interest rate swaps (changing face value and/or interest rate over the life of the swap). Very often there is no way for an auditor to realise that a swap is an amortising structure just by looking at the bank valuation. Often it is just the face value of the swap at the current valuation date that is shown on the bank valuation. It is the schedule at the end of the original bank confirmation that is required to accurately input and value such a structure.

Of course the true valuation of a derivative is determined by the price at which it can be sold/closed out which will be different to a valuation for accounting purposes. Valuations for accounting purposes are based on mid rates and, therefore, take no account of bid/offer spreads. Some of the changes we are seeing in the International Financial Reporting Standards are trying to provide greater consistency and more explicit definitions of fair value (IFRS 13). At least the “risk-free” component of an interest rate swap is a well-established methodology. The same cannot be said for the credit component (CVA), for which there is a myriad of approaches. It will be interesting to see how differences are reconciled and treated by auditors, as there is even less likelihood of two valuations being the same.