I think you'll find that it varies enormously.
Once upon a time we were a small company and took on pretty much any development that came our way. Some of these went on to sell good volumes for years, others led to one production batch and then the customer (often a new startup who couldn't get assistance from someone larger) folded or vanished. This approach is a bit like rolling a dice - sometimes you'll get lucky, other times you won't. Really the only way to assess it is: is the overall activity/income that it generates covering the costs to your business and generating a profit?
One potential advantage is that you can get lucky and back someone small who wouldn't be picked up by a larger business, and then have their loyal custom if they are the right product / right time and it takes off significantly. I'd genuinely attribute some of my employer's growth in the early years to getting lucky with a couple of these, but it is of course a total gamble and may get you nowhere. As commercial justification for an approach it's probably a bit tenuous - certainly if I was a shareholder/investor then it wouldn't fill me with confidence 
As the business became larger we moved on to charging for some development - larger accounts still got free development work, but smaller customers had to pay a "day rate" (which covered our costs and generated a modest profit), with extra fees if they also wanted to have e.g. taste panel sessions hosted in our NPD facilities etc. Want to mess about tweaking your recipe endlessly? It's no problem as we will keep charging you for doing so. Only actually buy a small amount of the final product? Again not an issue as we've made some income from the development process itself (as long as it isn't taking up time that could be used more profitably elsewhere).
This approach can strike a pretty good balance - as long as the fees aren't exorbitant then you'll still pick up smaller new clients that can grow into something bigger, your own costs are covered either way, and you can still provide the "free" service where you have a reasonable degree of confidence that it's going somewhere.
Nowadays we're part of a much larger organisation and don't charge, but instead simply say no unless you're on a list of prequalified clients we think are worth our time - usually based on turnover with us, or potential turnover given the scale of their established business in other sectors of the industry.
This obviously means that we're generally only using our time for something that has a better probability of success, with a client who'll be spending enough for it to not matter that only e.g. 30% of their projects actually progress to launch.
But I think you've got to be in a fairly confident financial position to use this approach, as it is effectively turning an amount of business away. It possibly also risks alienating clients for whom you'd previously have done work but now won't - overall it has a bit of an arrogant corporate vibe to it, but there is perhaps a place for that in parts of the industry, as some clients expect that almost as a sign of credibility. If your customers are large multinationals then this approach will probably be the one to favour, but otherwise it wouldn't be my personal preference.
If your business is neither tiny nor enormous then I think the middle option strikes a reasonable balance, and indeed in the UK at least I've noticed quite an increase in the number of independent product development consultants doing paid-for NPD for startups and small to medium sized brands, so whilst some of your colleagues may feel awkward about charging for the service, there is an increasing likelihood that your customer would have to pay for it if they went elsewhere anyway.
QAGB and Glenn have both made very good points - obviously you'll need to do the analysis to work out what it actually costs you at present, and how much it makes for you. Defining who actually owns the recipe is also relevant, as that can have value in its own right. Again you can make more of your costs back via this route, as if a client wants exclusivity then that has a cost to your business (in the value of sales you can't get from using the recipe elsewhere) so it is entirely reasonable to look to recoup that either via development costs, a licensing arrangement, or as an extra cost calculated into the sell price of the product.
