We use Hazard Rate Model to price our CDS instruments. Every day we input the CDS spread curve (one or more points), from that and the recovery rate we derive the hazard curves and from that our PnL.Let's assume I only know one CDS spread 5Y point for my CDS. As a consequence, the hazard rate will be constant for all t. Our traders will argue that this is unrealistic and will try to enter "made up" (lower) spread points for 4y, 3y, 2y to make the hazard rate increasing. Are they correct ? Is there a better solution to this ?Thank you.

If you only know 5Y CDS spread, this does not mean that the hazard rate h(t) is constant. It is just easier to assume a constant h(t) = h and solve for this "h" by matching the only observable you have, which is the 5Y CDS spread. Other solutions (in fact, an infinite number of solutions) are possible as well.If you (or your traders) have a view on how the CDS curve should behave for shorter maturities, you can then convert it to a model for the hazard rate that would be richer than a simple constant h(t) = h.

In order to let them realize the nonsense of putting lower spreads for shorter maturities jsut ask them "Why lower and not greater"? I guess in recent times the percentage of inverted curves in the market was a considerable one. Anyway any assumption is quite arbitrary. Any curve shape other than flat would have some parameters which you cannot determine (imply form the market) with just one point.

Right - I did not mean to imply that given that we have only one spread point the h(t) must be constant. The h(t) will end up constant in our model.Let me restate the question: what is a better - more realistic - assumption for the hazard rate assuming we have only one point ?

Making up Market Data points dosen't sound like a good idea. Do you really want to deviate from the market standard model?As a starting point you could start by stressing your market data to see the impact of adding 'fictious' quotes to your P&L and Risk numbers - if the numbers look to good to be true then I would be wary of making any changes.......

I agree. I would not generally think that "making up" data points would be a good idea. However, the traders' assumptions (upward sloping spread curve) are much more common than an inverted curve as with increased time to maturity brings increased risk of uncertainty. "Inverted" (downward-sloping term structure) spread curves are very tough because you can actually reach a point where you can get negative hazard rates. I believe the market prices these as "hey if they can survive to 2 years past this housing crisis I think they will survive longer more easily". The issue is how you price these in practice. If the hazard rates go negative you might just let them go negative also it's not reassuring to see probabilities of default decreasing over time.

Like AJSKJ says, all you are doing by adding fictitious points is a bit like stress testing, just as if you use different shapes/assumptions for h(t). In an important sense this all is very meaningless, fortunately though for simple CDSs it is not dangerous. However, if you then go on to use your h(t) to price things with optionality in the hazard rate then you are risking life and limb!P