No, the past may not provide a reliable long-term future indicator. However, the current mere “spot” position can't be better than the information from the whole “sample”.
We've heard it said that “today” is the only evidence we have. Well, that is simply not true, as there is clearly far more. Indeed, so far as we can trace, such an approach has never actually been tested against reality for long-term financial entities (until recently). Looking back at any series of financial parameters, we have observed huge variability. Ignoring the variability excludes almost all of the data!
Why assume that future experience must be worse? Is that what actuaries (or clients) really think their job is? This does seem to be the regulatory attitude, which is worrying.