The faulty towers of Solvency II

After more than 15 years’ gestation, the Solvency II regulation is finally entering into force on January 1, 2016. The recent preparatory reporting gives room for reflection on what new world European “insurance and reinsurance undertakings” (ie, insurers) and their “administrative, management and supervisory bodies” (ie, boards) are facing.

The first years of the Solvency II process were promising. The Solvency I regulation was obsolete at inception and the ideas of a more holistic, technically advanced and principle-based regulation were inspiring. An early milestone of the process was the 2002 Sharma report investigating the root causes of insurance failures from 1996 to 2001. Poor management was found to be the root cause of most insurance failures—and regulatory cost, complexity and distraction were still unknown threats of the future. So where did it all go wrong?

Let’s start with the following analogy. Your regulator tells you to make an exact replica of the Mona Lisa. You receive no illustrations of the painting, but several 300-page documents, written by a legal minded third party, full of irrelevant details regarding Mona Lisa’s shoes and underwear.

Instead of sending the painting to the regulator, colours and coordinates should be entered into an Enigma machine for coding and validation. The validation doesn’t check whether the painting depicts a woman, but rather if the ‘golden ratio’ has been applied with two decimal places when and where applicable. Upon request the “male or female artist or artists” should produce an exact description of how the painting was made, especially specifying time, date and cause of any changes.

Three pillars

The Mona Lisa analogy is inspired by the following examples from the three pillars of Solvency II.

Pillar one: capital requirements

Instead of providing a well-designed spreadsheet able to document and calculate the standard model capital requirement, the EU’s regulatory agency, the European Insurance and Occupational Pensions Authority (EIOPA), publishes 300-page documents with errata and Q&As in the same magnitude describing the finer points of risk models by references to EU law rather than the laws of probability.

So instead of having one model that could easily by validated by thousands of users, the thousands of users have had to develop their own standard models in parallel or, more likely, buy third party black-box solutions.

Pillar two: governance

Instead of providing templates of the jungle of policies that have to be written and board-approved, EIOPA developed a new series of 300-page documents containing detailed requirements regarding the content of these policies.

Pillar three: reporting

Instead of providing reporting templates that could actually be used for reporting, EIOPA provided spreadsheets describing what formulae to be put in what cells, but without formulae.

Second, instead of designing a data flow securing minimum repetition of the same data, EIOPA provided hundreds of consistency checks, on the assumption that data entered twice (or more) is consistent.

Third, spreadsheet data has to be converted to eXtensible Business Reporting Language (XBRL) files. Unlike commercial software, EIOPA’s T4U (the Enigma machine) requires input on a spreadsheet format even more cryptic than the XBRL format it produces. As this tool is useless, the same algorithms had to be developed in parallel by a large number of companies or black-box service providers.

In the case of both Solvency II and the Mona Lisa analogy, the task, the approach and above all, the combination of the two, give rise to concern. If you wanted a collection of nice looking paintings, that is exactly what you should ask for. If you wanted a collection of Mona Lisa replicas, a photo of the original is the necessary and sufficient piece of information that should be given.

As for Solvency II, the initial idea of principle-based regulation built around calculably defined capital requirements would have been well suited to be described in a directive. The detailed and incalculable regime it has developed into is better suited to be described and supported by spreadsheets, policy document templates and reporting software.

The Solvency II evolution is nicely illustrated by Pillar I. In the beginning the suggested capital requirement was a 99 percent tail value at risk—ie, the average one-in-100-year event. It was, however, acknowledged early that even though this risk measure has nice theoretical properties, it entails the practical challenge of envisaging and averaging all the types of extreme events that might not have been observed for the past 100 years.

It was therefore replaced by the median one-in-100-year event, which is equal to a one-in-200-year event (99.5 percent value at risk). Meteors hitting major cities, alien invasions and other events less probable than the prescribed 0.5 percent can thus be ignored.

The probability and consequences of hyperinflation, extreme sunspot activity, World War III, rapid climate change, internet breakdown and all other events more probable than 0.5 percent should, however, still be estimated.

In parts of the standard model, this calculation was simplified in three steps. In the first phase a log normal distribution was applied to calculate the one-in-200 event, in the second phase three times the standard deviation was applied and in the last (current) phase three times a politically decided standard deviation is applied.

A politically decided multiplier would have been slightly better (as a correct standard deviation is quite useful), but both alternatives have their merits, as the idea of calculating a one-in-200-year event is ridiculous.

Expanding documentation

Even though the calculation of the incalculable developed pragmatically, the documentation surrounding the calculations exploded. In the first Quantitative Impact Study (QIS) the technical specifications were eight pages long. The second QIS introduced the first version of the Solvency II standard formula. The spreadsheet and documentation were still manageable, at 66 pages.

QISs 3 to 5 and the eight years thereafter serve as a horrid illustration of how the devil, actuaries and most of all, the legal minded consultants, are in the detail. The spreadsheets and supporting documentation grew exponentially, as the project moved away from the principles-based system that was originally intended.

Another absurd fact is that this development concerns the “simple” standard model. Larger companies and other companies with lack of capital are taking on the alchemic challenges of calculating the “real” one-in-200-year event in internal models, that need to be documented, validated(!) and audited.

The burden of Solvency II depends greatly on how it will be implemented by local regulators, auditors and companies. The preparatory reporting increased these concerns, as focus was put more on consistency and technical issues than content.

It is probably too late to change the Solvency II regulation, but it is not too late for European regulators to start providing useful guidance, software and a pragmatic view on compliance. Otherwise, the insurance industry is at risk of losing focus on the more important aspects of day-to-day risk and profitability—topics that have absolutely no focus under the Solvency II regulation.

Manuel the Scandinavian Actuary with support from equally worried colleagues.

What our online readers said

“Alas, Solvency II was not even drafted by someone ‘legally-minded’! It is only since the delegated regulations were published earlier this year that we lawyers have been able to give any ‘hard’ guidance on capital weighting and investment structures, essential to the whole concept of Solvency II. The Lamfalussy process of enactment has been a failure in this regard, as have EIOPA’s attempts at clarification and guidance. Insurers will need every last day of their transition period.”

“This is a useful reminder that insurance undertakings actually exist to take risk and make money from doing so and that that the marginal benefit of ever more compliance should be weighed against real benefits.”

“We are creating a huge data-graveyard, and some believe that this enormous amount of data will hinder poor management. Fortunately, we will be able to call on model uncertainty to justify all misses.”

“Brilliant! I’m sure some software providers will claim reporting and capital calculations can’t be handled by Excel or other easily distributable tools. Both the standard model and all reporting templates were, however, originally developed in Excel—and would never have been envisaged without Excel. Errors attributed to spreadsheets are usually caused by bad use or impossible tasks. A brilliant example of a well-designed spreadsheet is the Solvency II calculator offered by SecondFloor.”

“Reducing ambiguities and reducing complexity in implementation would indeed achieve more clarity on the validity (and limitation) of models in shorter and more meaningful timeframes. Such steps could greatly help insurers, reinsurers and regulators to be as agile as they will need to be in the future of interconnected financial markets. Any such simplification would need to have the end in mind: a simplified, but valid view of risk(s). Some of these views might equally be interpreted in isolation without requiring an entire information management framework.”

“Well written, and I couldn’t agree more. It is my honest opinion that the time for polite talk has passed. Solvency II has become a travesty of actuarial thinking and a nightmare to all who are affected by it. It is a stupid and monstrous regulation.”

“I fully agree! With the current Fawlty Towers-inspired service and communication from EIOPA, Solvency II implementation is dealt with in silos. Reporting requires IT, capital requirements require actuaries, governance requires compliance experts—and it is impossible to get ownership and overview of the overall regulation and implementation. This also seems to be the case with local regulators. There is also a risk that employees in business-critical functions are drowned in Solvency II work and/or leave the industry.”

“For small firms, Solvency II is overly complex.”

“When politics and non-insurance people, however well intended, try to write insurance regulation, the end result is something that does nothing except add cost to an already well-regulated industry.”