So for the issues I raised above, even quarterly aggregation would be an improvement on the current situation but in terms of what would really be useful:
I know that a number of countries request monthly data from donors, so the maximum aggregation in any case should be monthly, if we want IATI to at least align with the granularity of data countries are already getting / requesting.
In general, I cannot really see an argument against aggregating low-value transactions (by month). However, for larger transactions, I think it is probably important to publish actual transactions given that some currencies can be quite volatile (for example, the price of the Liberian Dollar has fallen from around LD 135 to LD 120 against the US dollar in the last week, a fall in value of about 10% (see Central Bank survey). For large financial values, this can clearly make a big difference.
For loans, I think even small transactions should be published as actual values given that the precision is so much more important for calculating precisely debt and repayment amounts; though the Standard probably needs a few more fields to be really useful for debt management systems. I have been looking at debt management systems here the last couple of weeks. At the moment staff have to read through (~40 page) contracts and manually key data in to debt management systems. Being able to pull this data in automatically would clearly save time and probably improve the quality of data captured.