It is hard not to conclude that the exceptional Lib Dem incumbency effect is at least partly related to MPs' performance. The Lib Dems have certainly had the best policies for the whole of my 44 year voting life but "thanks" to the unproportional electoral system only had policy setting influence in the Lib Lab pact and the current coalition.
Today – the incumbency benefit sitting MPs can build up, based on an analysis of the 1983-2010 general elections:
This note adapts two models commonly used to estimate the incumbency advantage that US members of Congress enjoy – the ‘slurge’ and the Gelman-King Index – to provide comparable estimates for UK MPs. The results show that Liberal Democrats enjoy extremely large such advantages on a par with those of US Congressmen of between 5% and 15% of the vote. Labour and the Conservatives have incumbency advantages at around 2% and 1% respectively. The note estimates that effects could have changed the outcome in as many as 25 seats in some elections, and that they cost the Conservatives the chance to govern alone after the 2010 election.
The piece goes on to explain those two methods of measurement in more detail. The slurge is:
It was observed that in districts where a sitting Congressman stood down and was replaced by a newcomer the sitting party would underperform. When the new Congressman had completed his or her ﬁrst term, the in party’s vote would surge back up. Slurge takes the mean of this observed retirement slump and the ﬁrst term ‘sophomore surge’ seen by the parties at each election.
As for Gelman-King, it is a regression equation based on changes in vote share, whether there is an incumbent and who wins.
Aside from the point quoted above about Liberal Democrat MPs having by far the biggest incumbency bonus, the analysis also finds a strong double-incumbency effect for new MPs who have taken seats off other parties. When up for re-election for the first time they now both have their own incumbency bonus and also are up against a rival party which has lost its incumbency bonus, giving a double bonus:
The double incumbency term is particularly strong at the 1997 election, showing that ﬁrst term Labour MPs who took a seat off the Conservatives in 1992 outperformed other ﬁrst term Labour MPs by 2.9%, and gaining a ﬁrst term surge of 4.3%. Conservative MPs who took seats off Labour in 1983 and 1987 did similarly well at the end of their ﬁrst terms.