People often move from one project or job to the next. If they rent their software, they'll unsubscribe when they no longer need it, and cease being your customer. But with a perpetual license, your customers will never cease being your customer.
This makes it much more likely they'll pick up my app again, when they need it again. For SaaS, I presume that the number of people resubscribing after unsubscribing is very low.
Now why is that important? Because a big user base is necessary for growth.
Word of mouth and personal recommendations are extremly important for a small developer like me. The biggest problem I have isn't convincing people to buy a license; my biggest problem is that my prospective customers have never heard of me.
For example I pay for the intellij monthly subscription, if I changed job it would be to a similar role and I'd keep paying for intellij (or my employer would).
Since JetBrains switched there hasn't been any stagnation that I've seen in their development speed or addition of new features.
JetBrains is competing against free in many areas of it's product range so they have an incentive to ensure that users are happier paying than not paying.
I think that subscription pricing is a perfectly fair deal for software you use everyday. But lots of software has a long tail of casual users. I really want to make my apps affordable for those less invested users as well.
Pricing and monetization models are often strategic decisions that can have a number of different short and long-term implications for a company and for its product development.
@maxpupmax has a great point about aligning incentives with referral structures.
@compumike has another good point about b2b customers many times preferring a recurring subscription in return for certain enterprise type features like Support and SLAs.
on the P&L side, @lukebennett makes a great point about ongoing cost with cloud services. This is generalizable to any sort of ongoing cost incurred to provide the service (e.g. support, guarantees on service, security, or even processing costs in ecommerce).
Additionally, subscription contracts provide some means making cashflow predictable as some flexibility in creating more immediate cash flow via prepayments or in letting companies borrow against existing contracts. Tomasz Tunguz (http://tomtunguz.com/) has a bunch of fantastic posts on this.
Finally, on the product side, I'd actually argue that in many cases a subscription model actually better aligns incentives between customer and company. In one-time purchase models, new customers are king. In subscription models, churn is a huge concern and customer success eventually becomes paramount and I think there are more than a few cases where that creates a better product. Tomas explains this better and in more detail than me: http://tomtunguz.com/renewals_percent_rev/
Not saying that subscription models are categorically better, just saying that it's a decision that might make a lot of strategic sense and shouldn't be disregarded out of hand.