As was also mentioned elsewhere, players are always hungry for new items, and my guess is that Zynga has plenty of metrics to indicate that new items is where the money goes (a hefty dose of behavioral economics like scarcity is used on those ones). Simply leaving a game open, or closing it to new players, is probably leaving money on the table. That said, I think there's a big risk in teaching players that their virtual items really do just disappear into the ether.
My guess is it's the right move for Zynga to make, but Indiana Jones and Mafia Wars 2 is a huge blow to them. They were flagship products, and both of them are only a year or so old.
If you somehow knew I was going to buy exactly one book, you'd do your best to make it a hardcover and pull the paperback off the shelf.
The assumption in that though is that game $Y cannibalizes $X. Which is to say if you also release a new game $Y your returns on $X go away.
And while I see that as true with similar/derivative games "CashVille 2" is just like "CashVille 1" but with unicorns or something, I am not sure I'm persuaded that we do accrue an opportunity cost against different game. This is the root of the reasoning in my print money comment.
The reasoning is that if this config pays for the development cost and then generates marginal $ over time, you just add another stack to increase the marginal dollars you get. A writer releasing a second book doesn't 'cut into' sales of the first book, if nothing else they can sometimes increase the revenue of the first book by getting wider visibility.
I'm interested though in how we might reason about the opportunity cost here.