Over-Evaluation
Consider the case of the blue guy shown below.
People who trust his self-evaluation will see his request as considerably more
important than other people's, and so be inclined to work more to try and satisfy it.

Now let us suppose that someone fulfils his request.
He can now submit a post-hoc evaluation
of how the transaction actually influenced his welfare.
If his original prediction was an over-evaluation,
this presents him with a problem that can be summarised by looking
at the following pair of options:
(1) Publishing a post-hoc evaluation as inflated
as his original prediction. This would
be to publicly acknowledge receipt of a benefit he did not receive.
(A loose parallel exists with someone in an auction
whose bid for an item exceeds the item's value to him).
(2) Publishing a true post-hoc evaluation, less than his prediction.
Other people could compare this with his prediction,
and it would impact on the
credibility of his future predictions, creating
a kind of 'inflation' problem.
(A parallel exists with someone who is consistently late
- after a while people learn to accomodate for this and plan for him
to turn up later than he says).