Professor David Johns
Chairman, GAIC
Department of Health
Room 652C Skipton House
80 London Road
LONDON SE1 6LH.
28 September 2003
Dear Professor Johns,
I am writing to set out more clearly some of the points which I tried to make at the HGC/GAIC joint public meeting on 22 September. I am copying this letter to HGC, because in my view a crucial problem is that some fundamental questions are still not being addressed by any of GAIC, HGC or government.
I was pleased to have the opportunity of this meeting, although at the end of the day I think I felt more rather than less concerned about the direction of policy. There seemed to be an assumption, or perhaps a hope, on the part of many speakers (including some members of GAIC) that the current moratorium could quietly be allowed to wither away after 2006. I suspect a hidden agenda for this to happen by default.
It is easy for observers such as me to criticise without offering much in the way of practical solutions. To try to avoid this, I have included suggested “action points” for HGC or GAIC in respect of each of my comments. Of course I do not expect anyone to do anything merely because I say so!
1. Actuarial questions not addressed by GAIC or HGC
There appears to be a tendency for members of HGC to regard technical actuarial questions as mainly the province of GAIC. This does not work because (as far as I can see) GAIC’s terms of reference are not derived from a public policy paradigm at all. The GAIC terms of reference continue to be derived from the commercial perspective of insurance industry, essentially asking the question: are particular tests commercially useful to insurers? The public policy interest served by GAIC is limited to ensuring that insurer’s decisions about commercial usefulness are based on evidence.
In particular GAIC’s terms of reference exclude the materiality of adverse selection in the overall context of different insurance markets; and whether genetics will make much difference to this; and whether a moratorium for policies up to a ceiling is a workable solution; and similar questions.
When I made this remark at the meeting, Angus Macdonald immediately pointed out that there is some literature which does address these questions. This is of course correct. First of all there is his own work, where the approach is broadly to assume that individuals are able to withhold some information from insurers and that some adverse selection occurs, and then investigate whether a market under those assumptions is stable. Another approach, taken by several recent papers in the economics literature, is to investigate real-world insurance markets to see whether adverse selection can be detected, and to measure its extent. I have listed some of the papers which taken this second approach at the end of this letter. The literature certainly exists; but my point is that this the questions addressed in this literature are not being addressed by policymakers. This omission is very convenient for the insurance industry, but bad for public policy.
Suggested action: either GAIC’s remit should be explicitly extended to include the above questions, or HGC should explicitly take ownership of them.
2. Insurance pricing: theory v practice
At the meeting I made some observations which I felt cast doubt on the notion that competition between insurance companies ensures fair pricing; and also on the notion that pricing is a scientific, precise process which could easily be upset by small variations in adverse selection. I suggested that price changes were driven more by commercial dynamics. In particular I observed that
- Typical premium rates for critical illness insurance for identical lives have risen by up to 50% since the start of 2003.
- Typical premium rates for term insurance for identical lives fell by up to 40% in the latter part of the 1990s; conversely some companies have recently increased rates by up to 20%.
- The variation between the 6-8 cheapest rates in the market for term insurance for identical lives is up to 40%. The worst rates in the market are often more than 100% above the best rate for an identical life.
None of the insurance industry representatives at the meeting suggested that my figures were wrong, or offered any alternative interpretation of them. Yet elementary observations such as these on real world prices very seldom seem to be put forward by insurers in discussions about public policy.
Suggested action: HGC and GAIC to note the figures, and perhaps seek further evidence. There are some papers which give further evidence (for example, the last reference at the end of this letter.)
3. Structural bias inherent in GAIC
I see two structural biases inherent in the operation of GAIC (and to a lesser extent HGC, at least in relation to insurance).
The first bias is that in any consultation or meeting there is a small army of actuaries and others who are highly paid by the insurance industry to facilitate and promote genetic discrimination, whilst virtually no technical resources are available to support the interests of people who will be disadvantaged by genetic discrimination.
The second bias is inherent in the membership of GAIC itself. With the possible exception of Professor Wilkie, all members of GAIC who profess expertise in insurance also have commercial interests in promoting genetic discrimination. Thus whenever GAIC in its deliberations asks itself a question about insurance theory or practice, the answer available to GAIC is likely to be a selective one which paints the industry in the best possible light.
Such structural biases are not unique to GAIC: they also occur in other consultations by public bodies. For example, when the Financial Services Authority (FSA) consults on regulatory changes affecting consumers, it tends to receive an avalanche of responses carefully developed at great expense by actuaries and lawyers on behalf of insurance companies, and just one or two responses from or on behalf of consumers. However, the FSA does appear to be cognisant of this bias, and recognises the need to discount to some extent the weight of ‘industry’ responses. (The description “structural bias” for this problem in regulatory consultations is not my own invention: I have borrowed it from a speech by Sir Howard Davies, the chairman of the FSA.)
Suggested action: like the FSA, GAIC and HGC should be cognisant of the structural bias and discount to some extent the weight of ‘industry’ argument. GAIC should also consider commissioning actuaries or others to develop and elucidate the case against insurers’ access to genetic tests, as a counter-balance to the case which the industry will inevitably make for access.
4. Public health consequences of GAIC approvals
In some respects the applications currently being considered by GAIC are very much more controversial than the application already approved. In particular there is a great deal of anecdotal evidence, and more recently credible published evidence, that insurance concerns deter many at-risk women from undergoing BRCA testing, and hence the opportunity in appropriate cases of prophylactic treatment.
I think one could (with some heroic further assumptions) make a stab at estimating the expected number (in a mathematical sense) of “excess deaths” per annum which might be attributable to the approval of insurance companies’ access to the BRCA1/2 test. One might flex the assumptions to compare the effect under a moratorium with a high limit (where the “deterrent effect” might be negligible), and under conditions where the moratorium is discontinued (where the emerging evidence from the US suggests the “deterrent effect” might be much higher).
I think this type of calculation could be presented a quite credible way, and the implications could be hugely controversial. I was surprised that the Department of Health representative at the meeting did not seem to think there was a problem.
Suggested action: GAIC and HGC should consider the evidenced on the “deterrent effect,” and think very carefully about the public health consequences of any approval of insurers’ access to BRCA1/2 test results.
5. GAIC’s complaints procedure
I did not make this comment at the meeting, since I was conscious of having already used more than a fair share of the available time. I also have no direct experience of the GAIC complaints procedure. However, looking at the procedure as a detached observer, I am not surprised that GAIC hears few complaints, because the complaints procedure appears structured to prevent complaints from ever reaching GAIC. First the company has the opportunity to deal with the complaint; if that fails, the ABI has a second opportunity to delay and deter, and generally spin things out; and only then would an exceptionally persistent, well-informed and self-confident complainant ever reach GAIC.
This seems at variance with the suggestion in the Whose hands on your genes report that GAIC should be prepared to consider ‘anecdotal’ complaints, as a way of informing itself about consumers’ concerns. It also compares unfavourably with the procedure for other (ie non-genetic) complaints about insurance companies. If as a consumer you have a complaint against an insurance company, you go straight to the independent Financial Ombudsman (and the company immediately has to pay the Ombudsman a case fee to have the complaint considered, even if it considers the case is hopeless); there is no question of the ABI having an opportunity to confuse or fob off the complainant.
Suggested action: the complaints procedure should be changed so that consumers who have a problem with an insurance company relating to genetics make it known directly to GAIC, without the intermediate involvement of the ABI. (If in the context of the current moratorium this results in any significant caseload for GAIC, that would be a strong indication that something was going wrong.)
Other comments
I hope you will be pleased that my remaining comments are observations only, and do not lead to any suggested action points for GAIC!
Martin Richards who was sitting near to me made the very good point that ‘utmost good faith’ should be a mutual as opposed to one-sided obligation, and in particular that the reasons for increased premiums should be made available on request. I was disappointed that the ABI representative’s response was still the excuse that information would only be made available via the customer’s GP. The practical effect of this, as the ABI well knows, is that customers will usually find it very difficult to establish the fact that their premiums have been loaded, let alone the reasons for the loading. For people who already have a disability, the Disability Discrimination Act is rendered meaningless.
There was also a question (I think from one of the patient support groups) about the position of a person who decides to take a genetic test in (say) 2005 in the context of the moratorium, but then applies for insurance in (say) 2008. If I understood the ABI response correctly, they were unable to confirm that the existence of the moratorium when the test was taken in 2005 would protect the information in 2008. Effectively the ABI is saying that it intends to renege on the moratorium. It amuses me that this is contemplated in almost the same breath as bleating about “utmost good faith.”
Thank you for your time and patience at the meeting.
Yours sincerely
Guy Thomas
Cc Baroness Helena Kennedy, chair Human Genetics Commission
Appendix: papers reporting empirical investigations of adverse selection
Pauley, M.V. et al. (2003). Price elasticity of demand for term life insurance and adverse selection. National Bureau of Economic Research working paper, 9925. (Life insurance)
Cawley, J. & PHILIPSON, T. (1999). An empirical examination of information barriers to trade in insurance. American Economic Review, 89, 827-846. (Life insurance)
Cardon, J.H. & Hendel, I. (2001). Asymmetric information in health insurance: evidence from the National Medical Expenditure Survey. Rand Journal of Economics, 32, 408-427. (Health insurance).
Chiappori, P-A. & Salanie, B. (2000). Testing for asymmetric information in insurance markets. Journal of Political Economy, 108, 56-78. (Motor insurance.)
LangjÆr-ØhlenschlÆger, M. & Mcgaughey, K. (2001). A life term. Staple Inn Actuarial Society Paper, 22 May 2001. Available from Institute of Actuaries library.. (Not an investigation of adverse selection, but does contain information on the market variations in premium rates between companies and over time.)