loans and money

Should the time frame used be based on the term of individual loans? This may make sense when looking at the pricing of individual loans but makes production of consolidated numbers difficult. It may make sense also to look at default probabilities within a specified time frame independent of the term of individual loans. A problem here is that extrapolation over longer time intervals is problematic as changes in loss rates are not random and serial correlation is clearly present. A common time frame taken is one year and this may be appropriate in developed markets but in the context of emerging markets seems short and is definitely considerably shorter than the economic (as opposed to legal) term of many bank exposures.