One of the most provocative observations from a Federal Reserve Bank came our way earlier this week from Richmond. Often curmudgeonly, the Bank issued a note we analyzed on Thursday which essentially says that central banks should be very, very scarce with discount-window advances and confine these to the seasonal liquidity shortages cited in the 19th century by Bagehot that became the foundation of modern central banking. If the Richmond Fed is Ebenezer Scrooge, the New York Fed is Santa Claus – its president has consistently suggested that Reserve Banks should not only be forthcoming lenders of last resort (LOLR), but also even market-makers at the end of the road for financial institutions that come up short, be they banks or not. This might seem an academic issue fought out in competing Reserve Bank briefs, but it’s on the front line next week when Senate Banking sits down to start a fundamental discussion of what the Fed should do, when it should do it, and behind whom the central bank should stand.

The Richmond paper uses a 1985 incident to make a point that is very, very timely in 2015, as it well knew when it released the brief a week before the mark-up. At issue is the inability in 1985 of the Bank of New York to handle an operational break-down, forcing the Fed to fork over the then-unprecedented sum of $22.6 billion in discount-window support. The paper argues, as Chairman Volcker and New York Fed President Corrigan did at the time, that this case was far afield from the Bagehot-defined liquidity shortages at solvent banks necessitated by monetary-policy objectives.

In essence – my words, not theirs – the FRB’s LOLR backstop covered up a glaring failure by the bank to spend shareholder money to shore up its systems. Absent the aid, the bank would have been forced to face more than a few irate customers sure to take their business elsewhere. But, whether this would have harmed anyone other than BNY is far from clear.

This leads the Richmond Fed to argue that central banks should never provide LOLR support for operational failures. I would, though, contrast the 1985 operational lapse to another – unmentioned by Richmond – that forced the FRB to provide over $83 billion in emergency liquidity. Bank of New York was at the heart of this rescue too, but this time it wasn’t totally its fault – the cause was September 11, 2001.

Had the bank put in place better contingency operations and recovery backstops, the damage might not have been so grievous, at least to the payment system, but it didn’t nor did anyone else. The threat was unanticipated and the macroeconomic danger so acute after the attack that the Fed’s intervention was unquestioned — even, I would guess, by the Reserve Bank of Richmond.

So, some operational lapses should be supported by the LOLR and others not. This I get, but which is which? One could argue this –indeed, Senate Banking surely will. However, there’s one critical issue illuminated by the confluence of Richmond’s paper and the 2010 flash-crash anniversary that is indisputable: operational risk is not only more acute since the crisis, but much of it is moving outside the banking system.

See, as a case in point, high-frequency trading and the analysis we provided last week on what bank regulators fear and how little they can do about it. See also our new white paper, in which we lay out growing operational risk as one unintended, perverse result of the new regulatory framework – when risks move outside regulated institutions, they by definition increase in part because there is no LOLR backstop.

Hence the Federal Reserve Bank of New York’s solution: the market-maker of last resort. But, as many at the central bank even outside the Richmond Bank readily acknowledge, this would create awesome opportunities for regulatory arbitrage born of a new class of financial players with an iron-clad backstop from a taxpayer-supported central bank. Given all the constraints on banks, trading and market-making would fly to this new comfy corner unless or until a framework of prudential and resolution standards governs all activities that pose systemic operational risk, not just those housed in banks. Mr. Bagehot, call your office!