The vector clock is an important mechanism to track logical time and causality in distributed systems. Vector clocks incur an overhead of n integers on each message, where n is the number of processes in the system. The incremental vector clock technique attaches only the changed components of the vector clock to a message. This technique to reduce the size of the message overhead is popularly used. We evaluate the performance of the incremental vector clock technique via extensive simulations under a wide range of network loads and communication patterns. Our simulations confirm the intuition that this technique shows marked gains when application processes communicate with locality patterns. In addition, the simulations revealed the following behaviour: (i) the message overhead is not much dependent on the number of processes, (ii) a higher multicast frequency, as opposed to unicasting, lowers the message overhead, and (iii) a slower network speed relative to the inter-message generation time lowers the message overhead.