What are the benefits of publishing papers in mathematical finance and trading. Let us assume that the primary goal of a person/entity is to make money and reduce losses. Wouldn't publishing techniques/results give away secrets to competition?
There is an interesting article "How Derivatives and Risk Models Really Work: Sociological Pricing and the Role of Co-Ordination" by R. Rebonato answering your question.
In section "3.8 Conferences and Journals" the author formulate his version of the question as follows:
It is worth mentioning one last aspect ... of the ‘institutional ecology’ in which derivatives models are produced and used: the large international conferences devoted to discussing the latest innovation in option pricing. The speakers at these conferences are typically established quants, and the delegates junior quants, or quants from bank that want to become involved in complex derivatives pricing, risk managers and controllers.
The advantage for the quants in doing so is easy to understand: they can showcase their skills and prowess, and can become recognized and marketable in a relatively small industry. But why would a bank allow, let alone encourage, these quants and traders to speak about their ‘trading secrets’ (and, by doing so, to become ‘poachable’ by competing firms)? And why would a bank allow exotic quants and traders to publish the in-house-developed research in refereed academic journals, where, to be publishable, every model variant and improvement must be documented with precise and reproducible formulae?
And then the author answers:
The answer, again, lies in how money is made in derivatives trading (ie, by adding new trades, by taking no directional positions, by using sparingly the hedgeable-risk limits). Given this modus operandi, making money becomes much easier when model uniformity is enforced across the industry.
For the house, there are reputational advantages of publishing (“ we have smart quants”). This may not outweigh the loss of competitive information, depending on the material published. As you imply, the main benefit seems to accrue to the quants themselves, in terms of enhancing their own brand.
Perhaps this paradox arises from the question of what is the objective ? Do we have a communal objective of advancing the overall knowledge level of mathematical finance (as is the case for scientific endeavor)? Or are we trying to advance the knowledge only within our own organizations ? In the case of banks, presumably the latter.
There are different types of publication by banks
paid publication: this is what is called “research” where bank analysts or quants offer the bank client various sorts of market insights or reviews. This is paid research and usually under condition that client does not broacast it to others. Of course some very famous ones get so popular that they effectively end up in the public domain.
academic publications: those can be themselves of various sorts. There can be review papers where some quant will effectively publicly present methodologies which have been more or less used by every desk in the street for some time (in this case there is no competitive reason not to publish), there can be publications that aim to create consensus from the street on how to best price a new sort of risk (think of counterparty risk/CVA which became essential after the crisis of 2008 and Basel III), there can be other publications which are more speculative/unimplemented in practice (like supersymmetric methods in finance maybe ? :) ) .
In addition there is a number of desks which maintain active relationship with people in academia which can be as formal as say the financing of a research chair at some public institution. In those circumstances a researcher will offer support/insight say developping a model and the bank quants will offer the academic the benefit of knowing some practical aspects that make the new model more relevant to industry inn addition to market data etc.... It’s usually win/win situation and the counterpart for the academic is usually to be allowed to make a publication on some simplified form of the model in order to illustrate the main ideas behind. For example I can think of some publications on optimal execution that fall into this category.
If a company has actually built a superior model, they may publish it in order to benefit from the remaining market participants adopting it.
Say you have built a model showing that asset X is overpriced. Hence, prior to publishing, you simply short asset X and wait for the correction when market participants begin pricing according to your model.