# Why are KRDs calculated by shifting the par curve?

When we compute Key Rate Duration, why is the par curve the right curve to shift instead of the spot curve?

KRDs can be calculated by shifting either the par or the spot (zero coupon) curve and many vendor systems will provide both out of the box. Both approaches are “correct”.

Typically the par curve is favored since this is directly observable, but again this is a matter of preference. Major benchmarks (Bloomberg Barclays, FTSE/Citi) publish only par curve durations and KRDs.