The author seems to confuse credit risk and fraud risk. Credit risk pertains to the probability of you being unable or unwilling to honour your commitments. Fraud, within this context, pertains to the unauthorised use of funds. Facebook's data may be a uniquely valuable to evaluating the former, but to fighting the latter it is less important than existing transaction history databases.
Financial fraud is an immensely complicated problem. An estimate quoted in The Economist recently put the fraud loss rate at approximately 0.35% of transactions and up to 1.8% for online payments [1]. Cross-border payments are a niche unto themselves - even with margins as high as 10%, the fraud risk still poses a high enough barrier to entry that even the banks handle just 5-10% of remittances [2].
It's 11AM and so far Visa and American Express already have 1-3 data points of swipe activity on me. Over the day they'll get 5-10 swipes apiece (+1 NYC). Facebook has a tremendous amount of value in its dataset, but think for a moment about how much Visa, Mastercard, American Express, or your primary bank know about where you live, what you spend on (and by virtue when and where you spend it), and, by aggregating that data, who you tend to spend with. That is what's valuable for anti-fraud and by proxy payments processing.