Professor Jane Hutton, a former non-executive director at USS, talks to Mona Dohle about the problems with forecasting pension scheme deficits and the risks of shifting from defined benefit to defined contribution.
How has your background in medical statistics influenced your views on the way pensions funds forecast their funding position?
I became involved in accounting standards for financial institutions because I started seeing things that weren’t accurate. I have been involved in medical statistics for quite a long time and the standards of data management and analysis there have steadily improved.
As a board member of Newton Gateway, an outreach branch of the Isaac Newton Institute for Mathematical Sciences in Cambridge, I suggested that we ought to look at the comparison between regulatory approaches in medicine and finance, of which I’ve seen a few. That doesn’t just include pensions.
This will also be the topic of debate at an event organised in collaboration with the Royal Statistical Society on the 28th of April where we will discuss how Mathematics and Statistics can improve regulation in the financial sector.
In my role as chair of the Statistics and Law Committee, I have dealt with several cases, including requests relating to the FCA investigation into RBS Global Restructuring Group. Because of that experience I concluded that the standards of data management and statistics in the financial sector could be improved. So these comments are about what I am seeing in terms of the quality of regulation in the financial sector in general.
This also reflects the conclusions from the government-sponsored Morris Review into the actuarial profession after the Equitable Life Scandal. One of the recommendations of the report was that the actuarial profession should work with universities, and take account of technical developments. It also recommended independent expert scrutiny of scheme actuary advice.
My experience so far is that there are some big differences in the quality of data, data management and documentation. In medical research, every bit of data must be accounted for. Any discrepancies and corrections must be dated and signed off. This is in contrast to the practices of a number of actuaries and accountants in the financial industry. The accountants just say happily, “Oh we all know the excel spreadsheets have got mistakes in them”. If you took that attitude in the pharmaceutical industry you’d be out. But dealing with inadequate software is treated as a joke by accountants.
The quality of documentation is also a serious issue. If you look at, for example, the profit and loss estimates for next year, I expect there to be a complete log which is immediately available. Instead we are getting responses that it would take at least five days to report on what basis profit and loss forecasts have been made. Accountants and actuaries across a range of pension funds aren’t providing technical reports but power point slides with minimal information, and forecasts on Excel spreadsheets rather than the appropriate software. The slides often have mistakes on them that we would penalise students for. These companies can charge the best part of a million pounds and I am horrified by the standards. Maybe the pension funds that I have seen are particularly bad, but given that one of those is very big, it is a worry.
We are talking about the building blocks, the quality of data available, not even about specific formulae to calculate the deficit. A bit like baking a cake but one of the ingredients has gone off, in which case even a good recipe wouldn’t work
Exactly, but that is crucial, it is a mathematical fact that errors propagate.
The gilts plus model includes assumptions about future investment returns when calculating a scheme’s funding position. Is it possible to make precise forecasts on that basis?
As a medical statistician, I have talked to many parents of children with cancer or cerebral palsy about uncertainties in life expectancy. Yes, it is difficult, but it can be done.
A key point of statistics is to address the question, how do we get good estimates? We need to think about bias and variance. There are methods, there is extensive mathematical theory but among pension schemes and their actuaries there seems to be a happy willingness to ignore it.
Do you have any other issues with how pension deficits are calculated?
One of the problems, for example, at local government pension schemes, is that there is a tendency to present an estimated deficit as if it were a single accurate number. Obviously, to make informed decisions you first need the best estimate without “a let’s play safe” fudge and then consider several scenarios.
You want the best estimate of your current position and your cash-flows going into the future. Then, to explore “prudence”, a word which The Pensions Regulator (TPR) likes to use, you could present the different scenarios as a table, based on different factors. Say, if life expectancy changes by two years, the deficit could change to X.
Instead, there are strikes at universities because of a blunt statement that “the deficit” is a set number. That is a false statement. Instead an estimate of the deficit, with an indication of uncertainty, and explicit statements of all assumptions should be presented. In the financial sector, where they claim to be numerate, they should not be saying that this is a concept that is too difficult to explain to the general public.
The issue is that even with different forecasts; the scheme needs to come up with a specific value for the deficit contributions.
Yes, but decisions about contributions should be fully informed, and take into account the long-term nature of pension cash-flows.
Unfortunately, I don’t think TPR wants that quality of information. My impression is that TPR doesn’t interact effectively with the Government Actuary’s Department (GAD). That department produced advice on setting the personal injury discount rate for the insurance industry. In the current market environment, a portfolio invested in gilts is not risk-free, and GAD notes this is pertinent to pensions. But TPR told me that this advice on investment in gilts doesn’t apply to them. Ultimately it is about power and control. As statisticians we had similarly hostile responses when we started getting involved in medical research and legal practice. It takes time. A lot of people may also be acting in good faith. They may be getting things wrong but not necessarily deliberately so. Often they don’t have the training in statistics to get things right. So with some collaboration they could do a lot better.
USS’ investment returns have been quite stable, so why did the deficit fall from £7.5bn to £3.6bn between 2017 and 2018 and then double again to £6.6.bn in 2019?
The methodology hasn’t quite stayed the same. The 2014 and 2017 valuations had changes in methods. The most significant were probably the stochastic model, which is something that ought to be fully documented before any valuation is signed off. And then the estimates of life expectancy are important as well.
But the main point is that what you should be presenting is not a single number. You should be presenting something that gives different scenarios; given different life expectancies, for example. When the Office for National Statistics presents data on life expectancy, they have a principal table and a low and a high table with different possibilities. It should be routine to give different scenarios. It would also be a good idea to engage with pension schemes and actuaries in other countries and ask how they are approaching the valuation of pension funds.
You raised your concerns about a lack of access to USS’ data in 2017 and reported it to the regulator in March 2018. What support did the regulator give you?
Neither the Pensions Regulator nor the Financial Reporting Council has yet completed their investigations. But they haven’t come back to me. I only went to the press three years after first expressing concerns. I gave the regulator over a year.
How can the presence of member-nominated trustees help?
Trustees should come from a diverse background. That is the thinking behind a paragraph in the 2006 Companies Act which talks about the importance of independent opinions.
In the Carillion scandal, there seemed to be a mismatch of power and information between member-nominated trustees and the board. What could be done to empower trustees and help them hold the board to account?
I would have liked to train trustees but access to data is crucial. The case of Carillion is a good example. It would be interesting to know what access to information their trustees had. What the regulator needs to do is to support trustees to access information.
Would you say the gilts plus approach is overly cautious?
Yes, it is important to see pensions as part of the wider economy. If you only think about pension contributions but are bankrupting companies which offer defined benefit pensions, then that is not a good policy.
One of the problems with the gilts plus model is that if you try to base forecasts of investment returns on gilts, when schemes are actually invested in a lot of other assets, you can get serious errors. And if salary and mortality rate estimates have rounding errors, there is a cumulative effect which can be misleading.
There are alternative methods to forecast investment returns that do not involve just assuming pension schemes are invested in negatively yielding gilts.
The worst thing the regulator does is describing this approach as de-risking. It is not de-risking; it is moving risk around. This is shifting all the risk onto the individual and effectively saying that they will not have any assured pension.
If you are going to say, as TPR seems to, that your main priority is that no one ever uses the Pension Protection Fund, then the flipside of that is that people will have no promised benefits. Because if there are no promises, you have no responsibility.
So you are talking about a shift to DC?
Yes, there is an interesting paper from Canada, when it looked at the possibility of public funds shifting to DC from DB. The paper looked at several different stakeholders and concluded that every single group is worse off.
The only reason there is a benefit to the private sector is because you are offloading your responsibilities onto the state. Closing BHS and British Steel increases unemployment, demand on social services rises. We know that unemployment is bad for your health. I am not suggesting you prop up completely failing institutions but if you do what the pensions regulator wants to do, which is to focus only on having enough money for the pensions, that creates problems. The 2014 Pensions Act attempted to mitigate some of these issues.
The Pensions Regulator is operating a regime in which they say: “We want you to have an insurance policy for your pensions in case you go bankrupt, and by the way, we are going to set the cost of the insurance policy so high that we’re going to guarantee that you go bankrupt.” Other authors, including Ian Clatcher, Con Keating and Anna Tilba have also pointed this out.
Whereas, the research on Canada is saying if you are going to do this, nobody will benefit. Therefore, going to DC is clearly not a good solution. The report also mentions two States in the US, Nebraska and West Virginia, who went from DB to DC. The US is hardly a socialist system. They went back to DB because DC was such poor value.
With DC you have zero guarantees except for the state being expected to pick up the pieces. There are options in between, such as a collective defined contribution (CDC) system like the Royal Mail has proposed. But if you went back 40 years ago, funds like USS were set up similar to that. In 1974, pretty much all pensions were final salary schemes but increases with inflation came with the disclaimer: “provided it can be afforded.”
Inflation was a lot higher back then, of course.
Exactly, and what has gone wrong is that the regulations have been one sided. One of the things that happened is that the law changed to demand that you must give pension increases, regardless of the state of funding. Then regulators demanded such high deficit contributions that it risked bankrupting the company.
Another problem is that in the 1990s, when things were going well, instead of saying, “we can build up our reserves,” the actuaries advised either making no contributions or reducing contributions. So the problem that we have now is thanks to the actuaries, which they won’t admit.
On the other hand, we are now swinging in the opposite direction. TPR is working on the assumption that the covenant for USS is not strong, so you’re hearing: “We have to sort out the deficit in seven years.” But there have been reports that universities, particularly those in the USS, are strong for at least another 30 years.
So how could we get a more accurate understanding of a pension scheme’s deficit at a given time?
One problem is that accountants talk about a valuation. It’s not a valuation, it is a budgeting exercise and as such it is long term, but it’s treated as if it must be paid at once. It is important to keep in mind that changes to the pension scheme’s financial position don’t happen instantly. Valuations used to happen every five years, now they happen every three years, and TPR is trying to require reactions within less than a year. This is nonsense. Pension are long term. What you should be doing is looking at the cash-flow in the long term. If you think it is looking a bit worrying, then you make an adjustment. Instead we are seeing massive swings in the estimated deficits which results in positive feedback. The more you panic, the worse everything gets. But the deficit is a fiction – a conservative estimate. What matters is, do we have enough money to pay members? It doesn’t take a statistician to see that the system is not working. The method is wrong because the deficit figures are so unstable.
This sounds a bit like you are arguing in favour of a more cash-flow aware or cashflow driven approach?
Looking at cash-flows makes sense, but we need several different approaches to looking at the deficit. That is also one of the points the Joint Expert Panel [set to provide independent advice on the USS dispute] makes.
In terms of investment strategy, different bits of the government seem to want different things. Local government schemes are being encouraged to invest in infrastructure not gilts. But the regulator is trying to suggest everything should go into gilts, but there are not enough gilts to go around. Quantitative easing is a policy designed to discourage people from investing in gilts, so TPR and the Treasury are contradicting one another. Again, moving into gilts is not de-risking, it is attempting to guarantee a deficit rather than allowing for the possibility of a massive surplus.
The USS pension fund has a surplus of around 9%. How much bigger a surplus do they want?
Can you explain this in more detail?
If you look at their best estimate funding position, in the 2017 valuation [page 82, Annual Report 2019], you get a surplus of £5.2bn against liabilities of £54.8bn, even with some ‘prudence’, and they’re still coming up with a sufficiently pessimistic scenario to say they need to close the fund or have much higher contributions.
From the point of view of an employer, the safest way to manage DB deficits getting out of control is to close the schemes and switch to a DC system.
It is convenient for the regulator or employers, but a disaster for individuals, a disaster for the health service and a disaster for social services. They face the consequences of much lower pensions. You cannot shop for annuities in the same way you can for a box of eggs.
So a shift to DC is fine if your primary interest is to make as much money as possible for the financial sector. The insurance companies involved in selling the annuities for DC pots (de-risking for employers) will get much higher fees than they would otherwise. That is one of the reasons why DC is so much less effective. But if you have a responsible government, you should consider the effects on the wider economy.
In response to the interview, TPR said that they value reports from whistleblowers and had met with Jane Hutton, but do not provide updates during an ongoing investigation. TPR also said that it does not assess the appropriateness of schemes technical provisions or discount rates based on predetermined relationships to gilt yields but judges their suitability on the risks in their funding and investment strategies and how trustees manage them.
USS declined to comment.