Time to bring some science to financial regulation

The Global Financial Crisis showed financial markets consistently let down individuals and society. SEIU/Flickr, CC BY-SA

The recent report by the Senate Inquiry into ASIC documenting wide-ranging illegal and unethical conduct by financial advisers, is just another piece of evidence suggesting that regulation of the financial system is often highly ineffective.

Designing effective regulatory interventions faces a severe problem: the models or theories we currently have of individual financial behaviour, financial institutions or financial markets – and on which regulation is typically based – are highly inaccurate.

There are two major reasons for this lack of knowledge. On the one hand, financial behaviour and markets are highly complex and as a result very difficult to understand. On the other hand, financial behaviour and markets keep evolving and so knowledge is out-dated very quickly.

Basing regulation on inaccurate theoretical models can have severe adverse consequences, as the Senate inquiry shows.

Another example is related to inadequate retirement savings in Australia. Over the past few years, regulation has been introduced to encourage proliferation of choice options in superannuation funds. Indeed, superannuation funds in Australia have significantly extended the set of investment options available to their members. Extension of the choice set of investment options in superannuation accounts is thought to improve welfare because individuals can better tailor outcomes to their personal situation.

Scientific evidence, however, suggests that a greater number of investment options available is associated with lower participation rates. Moreover, an increase in the number of investment options is related to a reduction in participation in the equity market, with potentially detrimental impacts on long-term welfare.

A third example is related to the inefficiency of financial markets. There is a strong push by law-makers to centralise trading by eliminating over-the-counter financial markets (such as the Dodd-Frank Act in the US, MiFID II in Europe). This policy is based on the assumption that exchange can be efficient only if everyone trades through a centralised system where all participants can see orders and transaction prices.

Yet, evidence from controlled experiments shows that this assumption is wrong. Worse, evidence recently gathered in laboratory experiments suggests centralised markets reduce incentives for participants to expend effort to collect and verify information about the quality of the securities traded. Everyone trades passively, and prices become increasingly meaningless.

Australia is facing inadequate retirement savings and inefficient financial markets, along with the major public and private costs associated with them. Right now, we are lacking the right interventions to address these problems.

What’s the real cause?

The lack of effective interventions is due, at least partly, to a significant mismatch between the mainstream economic models and empirical evidence, suggesting that we lack knowledge of the problems and their underlying causes.

In particular, we do not have a good understanding of the ways in which various factors cause changes in financial behaviour and outcomes.

The latter suggests the need for greater use of experimental techniques in research on financial behaviour and markets as well as in the design and testing of interventions. Both are necessary for development of more effective public policy.

Overall, financial regulation today is too much based on the premise that the knowledge (of how to do things right) already exists, and that existing knowledge is sufficient to effectively develop effective solutions to society’s financial challenges.

In other fields, such as medicine and aerospace engineering, there is no presumption that all is already known. Instead, these fields acknowledge the complexity of the problems being addressed, and the incompleteness of existing knowledge. As a result, in those fields ideas, proposed regulatory interventions are tested properly before they are implemented widely – because many ideas will turn out to have unpredicted and unintended consequences.

Seeking evidence

Evidence-based regulation is needed in finance, and should replace current model-based regulation.

Introduction of evidence-based regulation in finance would include the establishment of strict formal protocols and institutions for testing of new regulatory initiatives, taking inspiration from policy-making in other fields such as medicine and aerospace engineering.

Regulation should start from an analysis of data from the field. If not available, controlled experimentation could be used to generate data. The data would allow regulators to formulate the right hypotheses, which could then be tested with controlled experiments – either in the laboratory or in the field.

Evidence-based rule-making has been tried successfully in the economic sphere before. The latest wireless spectrum auctions in the United States, for instance, were based on a complex combinatorial auction mechanism that was designed and tested in the laboratory at the California Institute of Technology by Jacob Goeree (now heading the newly created Centre for Market Design at the University of Technology Sydney) and Charles Holt. Significantly, the design deviated from proposals based on theoretical models that had been widely applied elsewhere (such as in the United Kingdom). The Goeree-Holt design was an enormous success and netted the U.S. federal government almost US$20 billion.

With evidence-based financial regulation, approval of new financial products or changes in financial regulation in Australia would be based on an empirical evaluation of benefits and risks of the proposed product or regulatory change, which would be assessed using randomised controlled trials. Any proposed product or regulatory change would have to be subjected to such trials in order to be approved and, crucially, before being deployed at large scale.

Such procedures are necessary because our knowledge of what we are trying to interfere with – new financial products or a change in regulation – is incomplete. There is a significant likelihood that the actual effects are different from our predictions.

Without a new approach, unintended severe adverse consequences on individuals and society are only likely to continue.