Tax and valuation fantasies

For reasons that are not important here, I am a member of a professional institution of property valuers. That gives a perspective on the proposals for new taxes that is unusual among the usual collection of journalists, economists and think-tankers who propose new or modified taxes, especially on property and wealth, with minimal understanding of the problems involved in assessing and collecting such taxes. The so-called “mansion tax” proposed in the Chancellor’s recent budget is an example.

I have no idea what the Valuation Office Agency (VOA) – the organisation that will be responsible for preparing the list of residential properties above the threshold of £2 million – said when presented with the proposal, but I doubt that its reaction was anything other than to express serious reservations about its implications. There is a myth among non-valuers that assessing property values is a straightforward and (reasonably) reliable exercise. It is not, which is why UK governments have tended to be reluctant to revise the valuations used for business rates – the UK’s version of a business property tax – even at 5-year intervals.

There is, of course, the mirage of computer-assisted valuations. The broad idea is that the property tax agency compiles a large database consisting of information on each taxable property – land and floor area, number and type of rooms, construction materials, etc. This is combined with data on registered transactions – property sales, new rentals. Turn the statistical wheel and out pops a formula for valuing all the properties in the database.

If only things were that easy! Property databases tend to be very out-of-date and expensive to update. If you send a survey to property owners asking them to complete a questionnaire, a good proportion won’t reply and how are you to verify the information without a large effort sending people round to check the answers. Think of the problems experienced by National Statistical Offices in carrying out censuses each decade. The costs are enormous, and the results are increasingly unreliable. And the stakes are entirely different. Asking about someone’s age may be a sensitive matter but it will not affect how much tax is paid in future.

Even if respondents are entirely honest, the scope for confusion and misreporting is very large. Do you have any idea about the floor area of your residence? Should it include garages or other attached structures? What about the number of rooms of different types? Does that include the small toilet under the stairs or an attic space used for storage or …? In addition, property values may vary substantially by location over relatively small distances due to undocumented features. Should the valuer take account of differences between properties with sea, mountain or lake views and those without such views in the same locality? And what about catchment areas for schools or noisy roads or …?

UK governments are criticised for not updating their council tax registers at regular intervals. There are, of course, political reasons for a reluctance to alter assessments, but in administrative terms the prospect of a full revision to the council tax register is a nightmare and provides ample cover for not acting. Remember too that the assessment is only whether the property falls in one of 8 bands of property values, so that it is much coarser than would be required for any tax on property values or wealth.

Computing the initial set of property values is only the start of the process. Any valuation can be challenged via a hierarchy of appeals which may take a long period to resolve. Appeals of valuations for business rates may take 5 or 10 years when large sums are in dispute. For individual property-owners the costs of appealing assessments may not be worth the potential savings but group actions to challenge the valuation formulae can pay off. Inevitably, there are administrative pressures on valuation agencies to lower valuations that may be contentious and easy to dispute.

There is a legal fiction – or Platonic ideal – that there is a “correct” valuation for each property. No careful property valuer makes such an assumption. Most will quote a range of values which, in statistical terms, might represent the lower and upper quartiles of the distribution of possible values – i.e. there is a 25% chance that the value will be less than the lower end of the range and a 25% chance that the value will be more than the upper end of the range.

That is a problem for tax valuations. A valuation agency might adopt the lower quartile to minimise the probability of appeals, but consequently it may be accused of going easy on properties that outsiders feel are undervalued. Any attempt to capture in legislation the idea that valuations are uncertain with a distribution of possible values rather than a single value is likely to run into intractable difficulties in both drafting and dealing with the ignorance of politicians and bureaucrats concerning probability and statistics.

Turning to the mansion tax announced by the Chancellor, the scope for things to go wrong is very large. The stated intention is to introduce the tax in April 2028, allowing a little over 2 years after the legislation is passed to carry out the valuations. That is tight time scale with a high likelihood that things will go wrong. Certainly, the potentially large number of appeals won’t be dealt with by April 2028. The Treasury appears to believe that about 150,000 properties will have to pay the tax. That means that at least 250,000 properties will have to be valued to include a reasonable range of properties valued at less than £2 million.

One initial question is whether there is a reasonable database of properties that have sold for more than, say, £1.5 million in the last 5 years? Of course, the VOA can use returns for Stamp Duty and Land Tax (SLDT) but these contain very few details of property characteristics and the same is true for Land Registry entries. To supplement such information the VOA would probably need to contract with either property agents or online marketplaces such as Zoopla, but both may fear the reaction of their customers if they are seen to cooperate too closely with the VOA. It is very likely that any statistical analysis will be based on small sample sizes with high levels of uncertainty about the resulting estimates.

Collecting information on at least 250,000 properties will be a huge task. There is little reason to expect much cooperation from potential taxpayers, so designing a questionnaire and following up on missing or incomplete returns will require a large increase in either employed or contracted staff for the VOA. Most are likely to be temporary or contracted hires, which presents significant problems of motivation and pay levels. The information that can be collected must be simplified to the minimum necessary to produce reasonable levels of uncertainty for the valuations, so the timing requires an almost complete database for past transactions before the data questionnaire is finalised.

And so on. It is very likely that those who thought up the proposal for a mansion tax and those who approved it along with the proposed time scale haven’t any real idea of the complexity of the task of constructing adequate database from scratch and the resources required to do this.

Equally, the practical costs of raising a relatively small amount of revenue – the estimate is about £400 million per year – are likely to be large. While some of this will be the once-off cost of constructing the database and initial valuations, the cost of updating the database won’t be small even if assessments are frozen for a period of 5-years. The VOA has a staff of about 3,800. Employing an additional 1,000 employed and contracted staff is likely to cost at least £80 million per year (including all overheads) plus the costs of handling appeals. This is a very expensive way of raising tax revenue.

This is only the beginning of the fantasies proposed by people who have minimal understanding of the complexity of valuing property and other assets. One idea popular with journalists and some economists is a tax on site land values, following the advocacy of Henry George in completely different circumstances 150 years ago. In theory, the tax would be imposed on the unimproved value of land, notwithstanding the fact that no one has any idea of what that means in the middle of London.

Since there are almost no observable transactions for unimproved land, those who propose a land tax usually favour a different approach to valuation. They want to start from the current market value of a property and then deduct the estimated cost of improvements which they envisage as the cost of buildings. This neglects the cost of providing infrastructure and utilities which contribute to the property value.

More important, the method ignores the fundamental problem of calculating a valuation as the difference between two large and very uncertain numbers. Any physicist or statistician is taught very early that the variance of the difference between two numbers is equal to the sum of the variances of each number. What this means is that the uncertainty of an estimate of land value is increased substantially by constructing it in the manner assumed. Since unimproved site values are likely to be small relative to estimates of current property values and development costs, any assessment of land values constructed in this way are likely to be highly contentious and very unreliable.

Any tax on the value of unimproved land is likely to be almost random in its incidence. It is likely to both very unpopular and subject to endless appeals. The people who advocate it are either ignorant of the practical difficulties of valuation or put dogmatic fantasies above any considerations of reason.

Proposals to raise revenue by imposing a wealth tax fall equally within the realms of fantasy. For example, the IFS Wealth Tax Commission (WTC) argued that all private wealth above £500,000 per person should be subject to a one-off wealth tax of 1% per year for 5 years. Of course, no-one in their right mind would believe that this was really a one-off exercise for a government dealing with an ongoing fiscal crisis. Still, as a one-off tax this proposal is almost completely daft.

The WTC suggests that there would be approximately 8.2 million taxpayers and that properties should be valued by the VOA. Even if most properties are owned by couples, the VOA would have to value over 5 million properties. It is inconceivable that this could be done in less than 5 years and the number of appeals would swamp the system in practice. They estimate administrative costs of £400 million per year over 5 years to implement the tax. This is equivalent to roughly 5,000 staff.

For context, HMRC has a staff of 67,000 employees plus an unknown number of contracted staff. The notion that it could implement a complex and completely new – and very unpopular – tax with an increase of less than 10% in the number of HMRC staff is ridiculous. Put aside the issue of property valuations, the problems of dealing with unquoted companies and pension funds are almost insuperable, unless the tax is treated as nothing more than a levy on nominal values of either balance sheet assets or fund values.

As an alternative, the WTC suggests that an annual wealth tax of about 0.6% on wealth of over £2 million would only affect about 625,000 taxpayers and would raise nearly £10 billion per year after administrative costs. Putting aside obvious mistakes in the reported estimates, they seem to believe that the administrative costs would be about £100 million per year on a regular basis. This number is probably an order of magnitude too low once the practical difficulties and the large amount of avoidance activity are considered.

The larger point here is not to attach much weight to claims about potential revenues and administration costs for complex new taxes, even when backed by academic “evidence”. There is a reason why tax agencies rely on familiar taxes. They have been developed over decades, which leads to the large volume of the tax codes required for implementation. Even in jurisdictions that have relied on property taxes for a long period, the legal and administrative requirements of regular revaluations are onerous.

The mansion tax is just a political gimmick, served up to appease factions in the current governing party. It will almost certainly collect less revenue and incur larger administration costs than expected. While it may be welcomed by the intended audience, there will be a different but ultimately more important audience for whom the real message is the Biblical one from Psalm 146: put not your trust in princes!