This varies year by year. In recent years, prop desks have either gained (Goldman) or lost (Lehman) huge amounts of money.

But this isn't always true. Sometimes services are the big moneymakers. Prime brokerage (allowing hedge funds to outsource their back office) was big up until the crisis nuked many hedge funds, for example. In a good year, classical IBanking (IPOs, M&A, etc) can be big. Big banks run many desks at either a small loss or small profit just for the purpose of keeping the lights on. When the market changes and that desk becomes important, they rake it in.

Is the economy, say, twice as well off from a financial allocation standpoint compared to 25 years ago? If not, how are the trading desks pulling in twice as much money without being extractors?

They could capture a larger portion of the new value being created. Suppose they created 100 units of value in the past, and captured 25% of it. Now suppose they create an extra 50 units of value, but capture 50% of it. Before this change, the banks captured 25 units of value, the world 75 units. After, the banks capture 50 units of value, the world captures 100 units.

As for payday loans, they always existed to some extent (loansharks were always present, as were pawnshops), but they only got into full swing in the 90's. The internet made tracking defaulters easier, competition in electronic banking made the transfers cheaper, and the Clinton-era wave of bank deregulation eliminated many state level interest rate caps (California's was lifted in 1996, for example).