I'm not well versed on the business aspects, but I'm not sure there is one, at least not anywhere close to current sizes. I think it basically amounts to "is there enough payload to justify regular launches". My gut says so long as it's heavily reusable, the bigger the rocket the more capacity you have for redundancy and safety systems. And the bigger the payload, the less payload needs to be wasted on inter-module connection systems. Especially for inflatable habitats and the like, where enclosed volume goes up with the cube of diameter, while surface area/mass increases much more closely to the square.
As for risk management, I don't think it actually makes a whole lot of difference - if your rocket has, say, a 2% chance of exploding during the launch, then you expect to lose an average of 2% on *every* launch, and insurance costs reflect that. Moreover, if you have say a 5-module system, and you need all five modules to be useful, then launching them separately instead of all at once dramatically increases your chance of losing part of it. I.e with a single launch you have a 98% chance of success, and a 2% chance of total failure, while with a 5-part launch you have only a 90% chance of success, with a 10% chance of partial failure. If the lost payload was something "off the shelf" - food, fuel, etc., then maybe that's no big deal, just buy more and send it up next week. But if it's something that took many years to build, then you've got a bunch of your assembly floating uselessly in space using up its design life while you rebuild the lost module(s) (no guarantee you only lost one rocket).
As for lower frequency/greater complexity - I'm not certain either holds - at least at this point. Bigger engines doesn't necessarily mean they're any more complex - and if going bigger means they can go fully reusable more easily and reliably, then the individual launch costs may actually be substantially lower than a much smaller partially-reusable rocket with more expensive refurbishment costs.
Basically the cost breakdown of a typical expendable launch is something like 80-90% vehicle construction, 2-5% fuel, and the rest manpower overhead. A bigger rocket takes more fuel, and pretty much the same manpower, so it mostly comes down to how the refurbishment and amortized construction costs compare per pound of payload.
That said - assuming the BFR is everything they hope, then once it's perfected I would fully expect them (or someone else) to build a scaled down version as well - just to increase the profit margins on all those little niche jobs that the BFR is overkill for.