I admit that I’ve been lurking in the shadows for more than might seem becoming. In 2011, FedFin issued the first paper we know of that highlighted the challenges posed by the cumulative impact of the post-crisis regulatory framework. Our policy and proprietary analytics then and now conclude that – with some variations based on activity or charter – the bigger and bankier you are, the badder the rules will bite. This conclusion isn’t based on whether that’s good or bad for public policy. Rather, it’s dispassionate business analysis based on our best projection of return on investment correlated with regulatory direct and indirect cost. Some have suggested that cost projections are disguised big-bank pleas for mercy. You could read them that way, but bottom lines don’t lie and a new IMF study shows why.

As noted in our client alert, the IMF paper – a staff brief, not IMF policy pronouncement – is the first empirical analysis of which we are aware that looks at the impact of new macroprudential regulations to see if they work as desired when the goal is to curb risky credit expansion. And, a good question that is too as the Federal Reserve finalizes its controversial counter-cyclical capital buffer and considers still more standards to short-circuit boom-bust cycles.

As I’ve noted in the past, including in a paper requested by the Federal Reserve Bank of Chicago in 2014, macroprudential rules might well be a strong antidote to procyclicality, but only if only banks dictate credit availability. If non-banks exempt from macropru play a role, then bank-centric standards backfire.

The IMF finds that this is in fact the case. Macroprudential tools have varying effects based on whether they try to affect the quantity of credit or its rate, but all of them taken together are found to reduce bank credit by an average of 7.7 percent two years after a tightening macropru tool is implemented. Credit availability overall thus drops as regulators might like, but nowhere near as much once non-banks and market financing (i.e., corporate bonds) are taken into account. When they are, then overall credit availability drops only 4.9 percent – and, again, that’s on average.

When one gets past the averages and looks more specifically at advanced economies like the U.S., one finds that macroprudential tools still sharply reduce bank credit offerings, but have a statistically-insignificant impact on overall market credit because non-banks offset the drop in bank lending. The FRB may try to take the proverbial punch bowl away through macroprudential regulation, but non-banks keep it full and frothy.

Importantly, the IMF studies only what happens when regulators try to restrain credit growth. This is, though, just one side of the macroprudential life-cycle. If these tools are genuinely counter-cyclical, then they have not only to curtail credit, but also to restart it after recessions, financial crises, or other stress events.

I suspect that the bank-centric nature of macroprudential regulation makes it even less suited for stimulus than sanction, meaning that recovery will – as we see now – be far slower and more tentative than it should be taking into account both natural economic factors and awesomely-accommodative monetary policy. The reason for this lies in the inter-connection between micro- and macro-prudential regulation.

Micropru rules such as the supplementary leverage ratio, liquidity standards, surcharges, and the like – largely apply only to banks and mostly also only to the biggest among them. These standards don’t come and go – they are as they are and, indeed, stress tests demand that big banks hold more than the micropru rules require to ensure resilience under even severely-adverse scenarios. As a result, there’s a rock-hard floor under the ability of large U.S. banks to promote credit availability without fear of sanction.

In contrast, many non-banks depend on market funding or secondary-market appetite that can and often has come and gone in a blink of the fickle market’s eye. Look for example at the travails of on-line marketplace lenders well before anything has gone really bad and see a precursor of how liquidity stress can stifle non-bank credit availability. Corporate-bond markets are also perched atop the head of a pin, as occasional liquidity squeezes already demonstrate.

Based not only on these market trends, but also on the IMF findings, it is clear that the U.S. credit market no longer lives or dies by the hand of big banks. As a result, big-bank macroprudential rules will often fall far short of their intended counter-cyclical effect.

This might – might – be an acceptable price to pay if macropru prevented systemic risk. Several overseers, including FRB Vice Chairman Fischer, have conquered their macropru willies by concluding that, even if these rules don’t conquer pro-cyclicality, they still prevent systemic risk. The IMF study – with which I concur – begs to differ.

What the IMF says is that non-banks don’t pose systemic risk if they aren’t backed by taxpayers or otherwise inter-connected with the financial system in risky ways. If theirs is an expectation of bail-out – as the markets continue to infer for large non-banks – or if non-banks undertake maturity or liquidity transformation in inter-connected or leveraged ways, the IMF does see systemic risk, as well it should.

The reason non-bank credit capacity is good is that it gives banks time to lick their wounds without undermining economic recovery; the reason it’s risky is that, when banks are told before a crisis to cut back, non-banks only power up, increasing volatility and, quite possibly, the odds of very, very hard landings.