What do we want to get out of regulations?. Jon Danielsson. Modelsandrisk.org

What do we want to get out of regulations?

October 6, 2022
Financial regulations are vital. But do they give us what we need?

The objective of financial regulations can only be to maximise high level social objectives, like economic growth, subject to us not suffering too many costly crises and abuse of clients (macro and micro), as well as broader social objectives like protecting the environment and social cohesion.

$$\max \frac{\text{ Growth and other social objectives}}{\text{Undesirable outcomes}}$$

We are not doing this joint social constrained optimisation.

Instead, the focus of regulations only on the denominator — preventing undesirable outcomes. Which means only controlling the risk of something terrible happening and building up buffers, in practice de-risking.

For that to work, three conditions have to be met.

First, risk has to be measurable. That a riskometer exists, one that we can plunge deep into the bowels of the City of London and get an accurate reading of risk. We can then design a feedback mechanism so that if risk is too low, increase it and decrease it if it is too high. Like the temperature in my office is 21°, and the thermostat keeps it at 21°.

Except, risk isn't measurable in that way. It is easy to measure the risk of day-to-day events because we have a lot of data to train our models on. But we don't care about this day-to-day risk. What matters is the extremes. What is the chance that ten year interest rates in the United Kingdom increase from 3.2% to 4.5% in one day, as happened last week, or the stock market will drop by 25%? Since we have practically no data on these events, we estimate the risk of day-to-day events (say 10 basis point changes in yields or 5% market drops) and assume the distribution applies to the extremes. The vicious margin feedback loops the pension funds suffered as a consequence of the 130 basis point increase shows what can happen when we get that wrong.

The second condition is that risk has to be well defined, a monolithic block we can shape to achieve our objectives. But it isn't. What is risk to me is different to the risk to my parents or my students. What is risk to a pension fund is not risk to a high frequency trader or a bank. What is risk to a small bank is very different to what it is for a large bank. Every individual and institution has a different view of what important risk is. So by treating it as a monolithic block, a considerable nuance is missed. And forcing the institutions to behave based on this monolithic risk block is inefficient — a waste of resources that might not achieve much. A risk theatre.

The final condition is that the buffers need to be sufficiently large, and they can never be. I wrote about buffers recently on these pages, arguing that it's much better to build shock absorption. But unfortunately, those buffers that protect against the most extreme outcomes need to be so large is as to be economically unviable, so they can not protect against what matters most. That means we also need the ability to absorb shocks.

And dependence on risk being monolithic and measurable has misled the architects of regulations. A long time ago, regulations used to be based on activity. Financial institutions were restricted to particular activities. Today, we are much more likely to tell them what level of risk is acceptable, how they are to measure that risk and what they are supposed to do when that risk is too high.

The way shock absorption works is to ensure that when some adverse shock hits the system, the financial institutions disperse it instead of amplifying it, just like the shock absorbers in your car do. The way to achieve that is to make the institutions of the financial system diverse. A shock comes along, and some institutions buy and others sell. That way, the shock is dispersed. Unfortunately, this is not the way we do things today. Because everybody has to be prudent, when a shock comes along, the prudent institutions have two reduce risk immediately, which means selling risky assets into a falling market, amplifying the shocks. Like the pension funds last week.

My contention is that financial regulations, especially macro prudential regulations, have amplified procyclicality since 2008, thereby increasing and not decreasing systemic risk.

So coming back to my problem:

$$\max \frac{\text{ Growth and social objectives}}{\text{Undesirable outcomes}}$$

If we focus too much on risk it is not possible to hit the optimum, if only because risk is so imprecisely measured.

So what should we do instead? First, increase the diversity of the institutions in the financial system. Make them as different from each other as possible.

We can't, of course, force them to be diverse. Still, a very good start would be for the financial authorities to stop being anti-diversity and instead actively grant licences for new types of financial institutions, with new business models, new ideas and new ways of doing intermediation.

The standard objection is: "we have to protect the users of the system". Of course, micro is essential. But, I don't think these users are well served by regulations that limit diversity, and therefore access to those financial activities best suited for them, while being very costly, where those costs are, of course, borne by those same users.

If we make the system more diverse, systemic risk will decrease, financial crises will become milder and less frequent, and financial services will become cheaper and more efficient. Both help with economic growth. Both help with meeting the objectives of financial regulations.

A win-win-win not The Illusion of Control.

Systemic risk in 2008 and today
Who believes risk can be measured?

Models and risk
Bloggs and appendices on artificial intelligence, financial crises, systemic risk, financial risk, models, regulations, financial policy, cryptocurrencies and related topics
© All rights reserved, Jon Danielsson,