Viv Davies: Hello and welcome to "Vox Talks," a series of audio interviews with leading economists from around the world. I'm Viv Davies from CEPR. It's the second of June 2011; and I'm in London, talking to Professor Viral Acharya of the New York University Stern School of Business about measuring systemic risk, Basel III and capital requirements. Professor Acharya describes the Stern systemic risk rankings of US financial institutions and what he considers to be the failure of the Basel III capital requirements in addressing systemic risk. He discusses shadow banking, and also how banking and sovereign risks are becoming intertwined.

I began the interview by asking Professor Acharya to explain how systemic risk arises in the first place.

Professor Viral Acharya: I would say; in order to give a very short but maybe useful definition, systemic risk arises when a significant part of the financial sector gets under capitalized all at once. Another way of saying it would be that, a common shock hits a large part of the financial sector that's being funded with short term debt. Different crises have either more emphasis on the shock, sometimes the emphasis is more on the short term debt aspects of it. But I think the key factor is that it has to hit a large part of the financial sector all at once, so that there is some loss of intermediation.

And through the loss of intermediation, there is either a credit crunch for the real sector, there's a reduction in debt for the household sector, capital markets could freeze up all at once. In the extreme, payment and settlement systems could collapse altogether so that there could even be loss of trade globally as a consequence.

Viv: So systemic risk is one of the main reasons why regulation of financial institutions is so important, yet there has never been a formal or organized method, established by regulators, to measure or quantify systemic risk. Why do you think that is?

Prof. Acharya: It's partly because regulators very often follow the letter of the law that's laid out. And unfortunately, most regulation - at least the way it was formerly written down - had always been about what I would call micro-prudential regulation; focused on supervision of individual banks, ensuring that they don't fail, ensuring that there's no sort of fraud in the accounting, that they're rightly recognizing losses in due course of time, and so on. Systemic risk was often justified as a reason for regulating banks. But as you put it, there hadn't been a good formal attempt to measure or quantify systemic risk in an interesting way.

In the Western economies, I would say there hadn't been an episode of a fully blown banking crisis. Scandinavia had one in early '90s. You could argue savings and loans crisis in the United States was also a systemic crisis in the sense that it cost the taxpayer a lot. A big part of the mortgage industry was at risk at that point.

But I think it was partly really because the Great Depression was followed by a suite of regulations, like the Deposit Insurance, the SEC, the Glass Steagall Act, which seemed to have contained banking crises for a while, until these regulations became anachronistic in one way or the other, so that the government guarantees remained in place, but all the restrictions to ring fence its users became weak over time. Amd then gradually the seeds of a full blown systemic crisis were sown in the Western economies.

Viv: Maybe you could explain for us what you and your colleagues at the NYU Stern School of Business have been doing in relation to measuring systemic risk.

Prof. Acharya: Our measurement of systemic risk is based on publicly available data. It's very close to the definition of systemic risk that I made earlier, which is that systemic risk arises when a big part of the financial sector gets under capitalized all at once. As you can see from this definition, the key ingredients are: first what's the leverage of an institution? Second what is its exposure to some sorts of common shocks out there in the economy? And three its size is going to be important because the larger it is, the greater is going to be its share in the overall under capitalization of the system.

At NYU, we have an academic paper on this. I've co authored this paper with Lasse Pedersen, Thomas Philippon and Matt Richardson. My colleague Rob Engel then has an econometric implementation in there with Christian Brownlees. But as a result of this collaboration ended in a website which is called Vlab - Volatility Lab - something that Rob Engel maintains on his Volatility Institute website.

Here, on a weekly basis, we update our ranking of the systemic risk contributions of the top 100 financial firms in the United States. Now, the idea is not to say that systemic risk contributions change on a week to week basis, but it helps you understand why these measures are moving.

Put simply, what our measure does is the following. It says that “I'm going to define a stress scenario for the economy as a 40% crash in the aggregate market, per se. S&P 500, for example”. Why such a simple scenario? Because we don't have the advantage that regulators have of knowing every single asset that banks are sitting on what's currently the most important asset class.

But by and large, in every crisis, if it is systemically important, you would expect a pretty large correction to the stock market, of the economy as a whole, because a systemic crisis should be something that actually spins over to the real economy to households and so on.

Our criterion is…the question we want to ask is: will firms be well capitalized when there is a large common shock to the economy as a whole? So, the question would be, for example, will Bank of America have a capitalization of not more than $8 of assets per dollar of equity? Conversely, which means its leverage should not exceed 12.5 to 1 in this stress scenario.

Therefore, we are not talking about current leverage. We are talking about leverage in a potentially severe stress scenario down the road. So we need to be able to project, given its current leverage, and given its current equity capitalization, how much of their equity capitalization will erode in a 40% correction to the market?

In a simple sort of market language, this is a concept of what is the downside beta of Bank of America? Beta is more of a linear concept. Probably risk increases somewhat more exponentially on the downside, so we do a more sophisticated estimation of this downside risk.

But we basically come up with a measure, which we call MES, or marginal expected shortfall, which is “what your expected shortfall, or loss, when the market is in a bad scenario this 40% shock?”

For example, if Bank of America's shortfall against market, on a per dollar basis, is, say, two, it means that when market loses 40% of its equity capitalization, Bank of America is going to lose 80% of its market capitalization.

Now, therefore, if I know this downside risk measure, the MES, I can project how much equity Bank of America is going to be left with in the scenario. I know what the liabilities of Bank of America are, based on its current balance sheet. I can project its leverage ratio, therefore, in that systemic stress scenario, and now, most likely, it's going to be short of being at the leverage of 12.5 to 1.

So I can calculate by how many dollars is its equity short of ensuring that its leverage doesn't exceed 12.5 to 1. This we call the capital shortfall of each firm in this case Bank of America. I can repeat this exercise for J P Morgan, for Citigroup, for Goldman Sachs. We do this for the top 100 financial firms.

The important thing is we are subjecting them not to what they believe is their own stress scenario, which is usually different across different players. We are subjecting all of them to a common shock, because we are really interested in the system as whole becoming under capitalized at the same time.

So now we can add up the capital shortfalls of each of these firms, and then the proportion or the percentage of that which Bank of America contributes is its own systemic risk contribution. If Bank of America's capital shortfall is $35 billion, the overall shortfall is $75 billion, then Bank of America's systemic risk contribution in the U.S. economy is about 45%, as per our measure.

We update this measure on an ongoing basis, so it reflects changes in assets that these banks might be undertaking, it reflects changes in their leverage, it reflects changes in market volatility, and so on. Potentially, we could even reflect changes in how likely the scenario is, where the market actually gets a 40% correction in the first place.

Viv: So you update weekly, or monthly?

Prof. Acharya: We update it weekly. I would say it's more interesting to study a long time series than try and understand when big changes take place, and try to add some economic meaning to what really happened around those times. One thing I should stress is, as I said we are not in the regulatory shoes; we can't have the supervisory intelligence that they have on specific assets that they have. So our approach to systemic risk is very top down. We are coming up with a measure of systemic risk at the level of an institution itself, whereas the current regulatory approaches are more bottom up. They try to get a more granular knowledge about their assets. In some sense, for example, the Basel requirements attach risk rates to each asset. They add those up, and then they try to come up with a capital requirement.

Viv: So, would this mean that it could become easier now to design efficient regulation that discourages the build up of excessive risk?

Prof. Acharya: In principle, yes. Let's suppose, for the time being, that our measure has some economically sound properties in reflecting systemic risk. You could think about several ways we could easily transform our measure into a capital requirement, because I'm actually telling you how much short you're going to be in a systemic crisis and that therefore tells me how much capital you need to raise today to ensure that you will not exceed a 12.5 to 1 leverage in a systemic risk scenario. You could even, in principle, design a levy or a surcharge that's based on this contribution. Clearly, there will be something that's about the level of regulation that needs to be decided as to how high the surcharge or the levy is going to be. Or, in principle, regulators could even just ensure that the systemically most important institutions are being supervised better, try and ensure that the resolution authority…whether they have sufficient knowledge of what's going to happen when they need to wind them down, are their living wills good enough in prescribing what set of actions need to get triggered, etc.

Viv: So, to what extent have the Basel III capital requirements been successful, or not, in helping to mitigate systemic risk?

Prof. Acharya: I have mixed feelings about Basel III capital requirements, in the following sense. Clearly, it's reasonable to think about capital requirement as a response to dealing with systemic risk. And this is because, by and large, systemic risk arises when there's loss of intermediation. Loss of intermediation generally happens when there is something like a bank run, or at least significant draw down on liabilities of a bank or a bank like institution. Where I think Basel capital requirements have failed and I would say have failed, in some sense, quite royally is in, really, their approach, which is really this bottom up approach of assigning risk weights to each asset class and then aggregating it up. As we've been discussing, what you really care about is whether the system is resilient to a large, common shock. But, what the Basel risk weights are based on is the individual historical risk profile of asset by asset. What you really want to know is the risk of mortgages is in a recession? What you really want to know is the risk of a corporate loan is in a recession? What is the risk of a credit card receivable in a recession? That common conditioning on a recession or a large financial sector crisis is what the risk weights should be based on. Instead, the risk weights are based on individual assessments of risk of different asset classes, but they are not aggregated upon.
So there are several problems with this, I would say at least three. One you are looking at individual risk of assets rather than their systemic risk. So because you have ignored systemic risk, implicitly, systemically-risky assets are being subsidized relative to their risks that they impose on the system.

Two the risk weights are often historical looking, they're backward looking in their risk assessments. And so, because the Basel risk weights don't change that frequently, they completely miss any dynamic aspects of risk taking in the economy. They miss the fact that an asset class could be emerging to be a systemically important asset class, like mortgages were in the last decade, whereas prior to that they had been, historically, the most stable asset class. And I would argue that, taking my first point, because we gave a lower capital requirement to mortgage backed securities in Basel capital requirements, in one form or the other, implicitly we are subsidizing this asset class by saying, "No, if you want to go and lend more on this asset class, please go and do it."

And that brings me to the third point which is that because you are not conditioning on a common shock, there's no sense in which the Basel risk weights are capturing a risk of build up of concentrated, common exposure across balance sheets of institutions. So, historically, mortgages have been a stable asset class; they weren't as dramatically large part of bank balance sheets as they became in the last decade. But during 2003, 2004, 2005, 2006, this was the biggest source of credit creation that banks were engaged in, either as direct mortgage origination or as holdings of mortgage backed securities or the other.

So Basel risk weights were telling you that, oh, banking sector is extremely safe because its risk weighted assets were very small, because they were loading up on an asset class that had historically been very stable. But of course, because they were loading up so dramatically on one asset class, there was lending down the quality curve; there was a common factor exposure that was building up. And add to all of this that Basel capital requirements don't really distinguish that much between short term debt, long term debt, even though Basel III proposes some liquidity capital requirements to get around it.

So I would say the broad approach of Basel capital requirements is problematic because of risk weights. Risk weights are not inherently about systemic risk, the way I see them.

Basel III, in contrast, put simply, is some liquidity requirements, which I think is going to deal with short term debt issues a little better than it has done in the past. But by and large, I'm not sure if Basel capital requirements are really what is needed to contain systemic risk in the economy. They are capital requirements, but unfortunately they are not capital requirements geared to deal with systemic risk.

Viv: And what about the shadow banking sector? There's been some criticism of Basel III in that, whilst it was designed to make the banking sector more resilient, it has perversely created new risks by favoring the development of the insufficiently regulated shadow banking system. Would you agree with that?

Prof. Acharya: I would say that's probably point number four against Basel capital requirements, but I would say this is actually a broader criticism of the regulation of financial sector in general, which is that it's often by form rather than function. So, for example; we regulate depository banks, but we were not regulating the investment banks, which, increasingly, after the repeal of Glass Steagall, started performing very similar functions under the same jurisdiction. One was under bank holding companies; the others were with SEC. Money market funds seem to be doing maturity transformations, just like depository banks do, but they are being treated sort of as a halfway between a mutual fund and a bank. So, I think there were these special purpose vehicles which were, again, doing maturity transformation, but because they were off balance sheet, they were not recognized as being part of the banking sector in one way or the other.

So there's a sense in which the letter of the law is important, which is that regulators often have a tendency to take what the law requires them to do and then go and then enforce those requirements. If the law doesn't allow them to do something, because unless the regulation is very principle based, they don't really go and actually exert their discretion or judgment about what they should be doing. They sort of tend to go in more of a box ticking approach.

So I think, in some sense, the only way around this is to regularly revisit the architecture of the financial system, see what new forms of financial firms are coming up, what their function is, and if their function is inherently looking to be the same as that of a regulated entity, both in what it does and what its capital structure is, I think there's no reason to call it something else, to just put it in sort of common parlance. I think, if a money market fund or a conduit faces a run, there is something about it which looks like a bank, and let's better regulate them, as we regulate banks.

Viv: Finally, what would be your strap line advice to the G20 with respect to regulatory reform in the banking sector?

Prof. Acharya: I would say ensure that regulation is addressing systemic risk, both in an ex ante sense in terms of charging more capital to those institutions that are likely to become under capitalized in a stress scenario. Ensure that those institutions, there is a good resolution authority for dealing with them when things go bad so that they don't become too big or too systemic to fail. And I think the last piece of advice I would give is that clearly sovereign credit risk is emerging as a big source of systemic risk in its own right. Banks own a big chunk of bonds of their own governments. Governments, when they are in fiscal difficulties, are known to push their debt onto the balance sheets of banks, either implicitly, through moral suasion, or explicitly, through higher statutory liquidity requirements.

Clearly, this means that the banking balance sheets and the sovereign balance sheets are getting quite intertwined. Ireland is a prime example. But to the extent that European banks are exposed to, say, debt of Greek banks, how safe Greek banks are depends upon how safe Greece is as a sovereign in its own right. These risks are becoming intertwined, and I think my bottom line on this would be that it's not clear that just macro prudential of private financial sector going forward is going to be sufficient for ensuring financial stability. We may have to think hard about fiscal prudence on part of the governments, to ensure that their problems don't spill over to the private financial sectors.

Viv: Viral Acharya, thanks very much for speaking to us today.

Prof. Acharya: Thank you, Viv.