You said the incentive to do so, not an incentive to do so. If, as it turns out, there are many different competing incentives, the easiest way to find out what affect switching to subscriptions has on the addition of features to software is to look at software that has switched to subscription. If you do look at it, your concern has little to no practical impact; subscription software gets new features all the time.

Which still doesn't tell you which features are missing because of the missing incentive, or how much more you then have to pay to get the new features (which is a different way to say: which features you wouldn't get if you were to pay the same price as otherwise).

Yes, of course, essentially nothing is monocausal, but that doesn't mean my point was fundamentally wrong: The subscription model (and by that I don't mean the setup where you have a contract for regular updates that you could just cancel and keep using the version you have now) removes the old version you already have as a competitor to the new version that you could pay for. That is most likely not going to be to the advantage of the customer, whichever form that disadvantage may take.

No, it isn't. It is not a matter of logic that the incentive is removed. One could logically say that if the company wanted people to continue subscribing, they would have to keep up with the bugfixes and patches.

Only if you ignore migration costs and assume that alternatives exist in the first place.

With the model where you can just keep using the old version without any further payments, you as the customer have at least two options two choose from: keep using the old version (doesn't cost anything), or buy the new version (you have to pay). Those two options effectively compete for marketshare, and as such, the latter option has to convince you with something that you consider valuable enough for you to pay the higher price. There might be further options in the marketplace, but that's not guaranteed, and also, migrating to a different solution usually has costs associated with it (learning the new software, lower productivity in the beginning, downtime, paying for the export/conversion/import of the data, ...), while the migration costs between versions of the same software tend to minimal, in particular lower than migrating to a different product altogether.

Now, with the subcription model, the first option is removed from the market. You only can buy the new version, or pay the migration costs of switching to a new product. Which is also not lost on the company selling you the software: If their market research people are any good, they'll set prices such that it's a bit cheaper than the competition plus migration costs, effectively turning a large part of your potential migration costs into their profit.