• C++ Programming for Financial Engineering
    Highly recommended by thousands of MFE students. Covers essential C++ topics with applications to financial engineering.
    Python for Finance with Intro to Data Science
    Gain practical understanding of Python to read, understand, and write professional Python code for your first day on the job.
    An Intuition-Based Options Primer for FE
    Ideal for entry level positions interviews and graduate studies, specializing in options trading arbitrage and options valuation models.

FT: Banks cut budget for maths and models as rules change

Andy Nguyen

Member
Nice reading for QuantNet audience. Use it to have a more informed idea on career path.

Banks cut budgets for maths and models as rules change - FT.com

Banks are cutting back their nine-figure budgets for PhDs and complicated risk software ahead of a regulatory clampdown on lenders that use models to make themselves look safer and reduce their capital needs.

Investment banks have been spending up to €150m a year building and maintaining complex internal models to measure the risk from everything from deadbeat homeowners to interest rate rises and market movements.

The models help lenders spot looming problems, but up to now they have also come with big financial benefits. In the years before the financial crisis, global regulators rewarded institutions that could show a better understanding of risk by allowing them to hold less capital against their lending and trading.

Now those benefits are fading fast, as policymakers at the Basel Committee for Banking Supervision (BCBS) strip away the capital advantages of using many of the models,.

“Part of the … philosophy was to incentivise banks to have better risk modelling methodologies through capital,” says Andres Portilla, head of regulation at industry group the Institute of International Finance. “That has been abandoned.”

Policymakers changed their mind after the financial crisis. The wide variation in banks’ risk calculations stoked fears that some lenders were gaming the system to make themselves look less risky.

The Basel group has already banned the use of internal models for calculating operational risk, and they are reducing the capital benefits banks can get from using their own models to calculate the riskiness of their securities and loans.

“The banks say that if everything is going to be standardised, why spend a lot of resources to maintain the models that inform the advanced approach,” says Guy Moszkowski, New York-based analyst at Autonomous.

Right now most large banks use internal models to calculate at least some of their risk-weighted assets — essentially computer programs assign risk scores to the loans and securities each lender holds, with 100 per cent being the highest.

Individual banks use quite different calculations, although some of it has to do with their mix of assets: home loans have much lower risk scores than things like derivatives.

UBS’s RWAs are just 25 per cent of its total assets, while Goldman Sachs are 67 per cent of total assets, according to banks’ published numbers adjusted for different accounting treatments.

The differences matter because RWAs, rather than total assets, are used to calculated common equity capital ratios, the most important measure of a bank’s safety and soundness.

Investment banks with extensive internal model capabilities could spend anything from €50m to €150m a year maintaining them, according to Donal Gallagher, head of Quaternion Risk Management, a risk analytics firm working on behalf of large banks. That figure captures the headcount and IT costs, but not one-off costs like new regulatory initiatives.

No one thinks that models will be eliminated altogether, since they still carry a real economic benefit, and regulators want them preserved for bank safety reasons.

But bank executives, traders and modellers all believe that the bar to clear for investment in future models will be higher.

A senior executive at a large European bank said it was “inevitable” that banks would do less modelling over time “to the extent that internal models don’t give you a (capital) advantage”.

Mr Gallagher estimates that a bank that now spends €100m annually could save about €40m a year.

“That’s a conservative estimate, the true savings could be higher,” he adds. “Then there is the additional reduced cost of a more straightforward implementation of … regulatory initiatives in the future.”

Total operating costs for a midsized investment bank are about €4bn a year.

Large US and European banks declined to comment on their plans. But cutbacks are particularly attractive in a market where investment banks face plunging trading revenues and a slow market for new floats and dealmaking.

The European bank executive argues that regulators “haven’t thought through” the longer term consequences of their decisions which he believes will encourage banks to “spend less time identifying the risk and more time thinking about the capital”.

Regulators argue that banks still have to manage their risks, so they will invest in modelling even without the capital benefits. One stresses that what they are trying to eliminate is “aggressive modelling” not modelling altogether.

The IIF’s Mr Portilla says a key question is whether banks need financial incentives to build risk models. “The answer is yes and no,” he says. “’No’ in the sense all banks perfectly know that they need to assess risks in the correct way.

“But in today’s world regulatory capital incentives within organisations do drive behaviour, do drive pricing, do drive portfolio composition,” he adds. “We can pretend that doesn’t happen but in reality it is a key factor.”
 

vertigo

Active Member
you can tell the authors of the article are clueless after reading the first line, i think they are very confused. complicated risk software? that must be the biggest joke in the quant industry. it is well known that all banks have simplistic and crap risk software - limited functionality, bugs, etc. i would call them 'bloated', not 'complex'. there is nothing complicated about a risk model.

the authors then replace the phrase 'complex risk models' with 'complex internal models'. risk models are a subset of internal models, but the converse is not true: pricing models, for example, models that calculate the sensitivities (Delta, Vega, etc) internal models but not risk models. whilst this is a minor point, anyone who has worked in a bank would understand the difference between a risk model and an internal model. there are not even the same category.

nor do they define which risk models (market risk, credit risk) will be impacted by new regulation. yes, moves in interest rates will be quantified as market risk, but loans could be credit risk too.

it is very difficult to run a profitable business under a standardised approach. that is why banks apply for internal model approach. if the authors are trying to convince us that banks will just switch to a standardised approach and cut the budget for internal models, they are living in a fantasy land, where Elvis and Tupac are alive, because that will not happen - banks will push aggressively to keep permission for the internal model approach. there are big budgets allocated to enhance the internal models & standardised models to be consistent with new regulation.

i dont think this will detract any PhD students or MFE students from joining banks to work as quants - nor will it lower the demand for banks to employ quantitative focused students. it simply means more work on regulatory aspects, which is a shame because the regulators are clueless.
 
The article seems paid only, so cant dissect through the details. But agree completely with Vertigo's response here.

Standardisation of Models has been a prominent theme, and it would continue to be so. Lots of regulators have now switched to a kind of feedback mode - i,e they look for first set of numbers, and then work on top of it to make further tweaks - with an intention to only make the regulations tougher. As such, they dont seem to have a strict set of policies to be just implemented and be done with. From their perspective, it might be cos they dont want banks to see through and adapt to rigid rules. So they keep the framework flexible and evolving. Which only means, banks for past few yrs have increased emphasis on Reg - Strats.

Its a different arguement, if the net $ spent is reduced since they can always have larger teams sitting in Low cost locations and have smaller.specific teams facing respective regulators. May be the article was hinting at this phenomenon.
 
Top