Skip to main content

Financial Markets and Regulation: A Genetic Programming Approach

Final Report Summary - FIMAGP (Financial Markets and Regulation: A Genetic Programming Approach)

Project objectives:

The project's objective is to contribute to the understanding of the impact of financial regulation on securities markets through the development and application of an interdisciplinary approach which combines finance, economics and computer science to study order book and OTC markets. Our research aim is to provide a new methodology to support regulators in forecasting and quantifying the effect of regulations. The insights gained in this project are highly relevant for policy makers as European competitiveness and growth can be negatively affected if regulations are poorly designed.

The project addresses in particular the lack of scientific insights into the long-term effects of the regulatory measures of financial transactions taxes and leverage constraints such as short-sale bans. These regulations are aimed at curbing speculation and furthering market stability. Both have strong political support in Europe. On the market design side, OTC markets, whose lack of transparency is believed to have deepened the financial crisis, are under scrutiny with the aim to centralise trade.

The project objective includes dissemination of the new methodology to a wider audience of researchers and practitioners. This will be achieved through the publication of a book, which explains and applies the methodology in conjunction with peer-reviewed publications and a website, which allows public and free access to the software developed on the project. Both will allow verification of the project's findings and encourage further research by regulators, practitioners and the scientific community.

Work performed:

Work performed within this project falls into three main research categories: modelling, calibration / simulation and analysis / interpretation, and three dissemination activities: publication in scientific peer-reviewed journals and a book, presentation at international conferences and outreach to practitioners and general audiences.

The programming work on the implementation, simulation and empirical analysis, which constitutes a major part of the project, has produced a software package which will be made publicly and freely available. The scientific results are published in four articles / working papers. The fellow has co-edited as guest-editor two special issues on behavioural/evolutionary and stochastic financial models in international peer-reviewed scientific journals. A book manuscript is currently in the draft stage and will be finished in summer 2013. Its aim is to present the new methodology and the scientific results in a format suitable for a broader audience. Summaries of the project research for the general public have been published in two freely accessible outlets. Work on columns for VoxEU.org is in progress. The project work was presented at several international conferences as well as showcased through co-organisations of a conference and conference sessions on behavioural and stochastic financial market models during the project period.

The project work was able to go beyond the original aims and scope of the project. This has been made possible through an incentive grant awarded by the Norwegian School of Economics (NHH). The grant, which is financed by the Norwegian government, matches the EU project funding and allows an increase by about 50 % of the effective duration of the research work. The additional original research which results from the grant, will allow us to expand the coverage of topics in the book. As a result publication will be later than originally envisaged.

Main results:

A new financial market model has been developed on this project. The model comprises a general description of investor behaviour with a detailed specification of the market microstructure in institutionally rich environments. The model combines portfolio management with trading strategies as suggested by Parlour and Seppi (2008, Handbook of Financial Intermediation and Banking, North-Holland).

The project has delivered new results on the long-term impact of regulatory reform in order-driven markets. The findings show that many of the conclusions drawn from the 2008 short-sale bans are short run effects that disappear in a long-run equilibrium. We find that short-sale restrictions reduce short-term volatility and long swings in asset prices, while transaction taxes do more harm than good for market quality and stability. In particular, contrary to claims made by some proponents of the financial transaction tax, we find no evidence that the tax reduces long swings in asset prices, as measured by the mean stock price decline from a peak in an expansion to a trough in the next recession. By curbing speculative bear runs, the short-sale ban reduces the depth of recessions and dampens long swings in asset prices. This yields a calmer market with slightly improved price efficiency, substantially lower volatility and a 7 % increase in stock prices.

Our results caution against the use of a financial transaction tax for anything but fiscal objectives. However, our on-going research finds that current EU impact assessments overestimate tax revenues. Our estimates show that the impact of a financial transaction tax on total tax receipts, which takes into account the reduction in GDP, is likely to be negative.

Further research results are obtained on additional applications of the new methodology as well as the further development of the theoretical underpinning of evolutionary finance models and methods for their formal analysis.

We analyse in particular the nexus between market fragmentation, market stability and investment skills. This line of inquiry offers a new perspective on the evaluation of market stability and the degree of market fragmentation. The model describes a quote-driven derivative market in which traders have to form individual asset price theories and act as market makers. A further set of applications is concerned with financial engineering. In particular, we have developed hedging strategies under transaction costs using a parallel-processing approach to solve optimisation problems using genetic programming.

Impact is fostered through the publication of a book (working title 'Computational Evolutionary Finance,' expected publication in 2013) and publication of the source code of an accompanying software package. Public and free availability the software produced during the project will allow practitioners and the scientific community to verify the project results and to explore related research questions.

The main expected impact is through further applications of our new methodology. The project demonstrates the potential of the approach by offering new insights into the impact of financial regulation. No previous model apparatus offers this capability.

Impact on society is closely related to the use of these insights and further application of the methodology by regulators. To foster this impact, columns reviewing work on short-sale bans and transaction taxes, which are invited submissions to VoxEU.org will showcase our project work. VoxEU.org draws a very large audience from outside academia; its columns are often cited in the Financial Times.