Many phylogenetic tools can calculate each type of mutation (transisions, transversions; G - A, G - C, G - T, A - C etc...; insertion, deletion). Most methods of phylogenetic reconstruction do not count the insertions and deletions in the distances, because there are different types of insertions or deletions that need to be considered differently. A deletion of 15 bases is probably one event, not 15 single-base deletion events. Insertion or deletion of bases in a homopolymer run (AAAAA -> AAAAAAA) is different than duplication of a repeating pattern (ACACACACA - > ACACACACACAC) for just two examples. Another difficulty with insertion/deletion events is the multiple sequence alignment step can become a problem, with no single unambiguous solution.
As far as I know, PHYLIP does not produce a count of insertion/deletion events in pairwise distance calculations.
@Yong Shi, most of the phylogenetic packages such as MEGA, PHYLIP, DAMBE etc include distance methods which produce a matrix of all pairwise distances as part of the process of tree building. In PHYLIP there are DNADIST and PROTDIST for DNA or protein distances, for example. From that matrix, is it fairly easy to calculate the mean, average, std deviation or other values for all or any subset of the pairs of interest to you. Also, the programs allow a range of options for which model of evolution to use when calculating the distances. The simple % identity between pairs is often called a "hamming distance". If you do a GOOGLE search for [MEGA pairwise distances] you will find several tutorials on exactly how to do all this in MEGA.