The Valuation Office Agency already have records of all commercial land and buildings for Business Rates purposes, so that requires minor tweaks only, in this section I will outline how residential land could/should be valued.
• What do we mean by "annual site-only rental value assuming optimum permitted use" or "site premium"? (skip to article)
• Using recent selling prices as a guide to rental values (skip to article)
• Indexing up recent selling prices (skip to article)
• Factors/relative values (skip to article)
• Banding (skip to article)
• Appeals (skip to article)
• Revaluations (skip to article)
• Estimating the total tax base for UK residential land (skip to article)
1. The site premium.
One stupid argument trotted out by the Faux Libertarians is as follows: Rothbard asks: How, then, will the government be able to separate site value from improvement value?
Compare very similar buildings in different areas or on different sized plots and subtract the difference!
For sake of argument, let's initially assume that all UK residential plots are being put to their optimum permitted use (i.e. that for which they have planning permission, which is probably true for 95% of all plots). The other 5% are plots with derelict buildings on them; plots with useable buildings on them which are unoccupied; and there are around half a million plots which have planning permission but which have not been developed yet. Of the 95% which are actually being used (which have buildings on them up to the current permitted limit), there are only a few per cent at the top end (small building, big plot) where it would be economically/commercially worthwhile replacing an existing building with some smaller houses or flats (with or without LVT). So for valuation purposes, the working assumption is that they will be taxed on their current, actual use. That's that caveat dealt with.
Proper LVT is based on the "site premium" element of annual rental values, i.e. if two physically identical houses in two different areas rent for different amounts, the difference between the rents relates to the "location, location, location" and that is subject to LVT. Here are five more or less identical three-bed semi-detached houses from the first, fourth, fifth and tenth deciles, with one from the top one per cent for luck (downloaded February 2013):
If we add likely Council Tax/TV licence fee to each of those rents and deduct £4,000 annual maintenance and repair costs, we get the following site premiums:
TS4 - £2,940
LS28 - £4,800
LS19 - £6,600
IG9 - £16,800
TW1 - £39,900
For various reasons (explained below) it is best to split up the UK into smaller defined areas (there are about ten thousand local council wards or postcode sectors, so there are about 3,000 homes in each sector which seems like the ideal unit to me) and use the average three-bed semi detached houses (26% of all homes) in each sector as our reference point for that sector. The people from HM Land Registry and the Valuation Office Agency have already done all this in a far more sophisticated fashion (and would fall about laughing if they read this), it's just that the politicians have told them to keep quiet about it.
2. Using recent selling prices as a guide to rental values.
a) The required mass of data on current rental values is, for an outsider, almost impossible to find (you can look up current asking prices on Rightmove or Zoopla one postcode at a time) and there is not the same official 'history' as there is for selling prices, but there is a clear enough link between the two.
For example, the average selling price for the period 2012 of a three-bed semi-detached in the four areas above are as follows:
TS4 - £120,384
LS28 - £165,513
LS19 - £207,728
IG9 - £418,577
TW1 - £834,461
The site premium from 1. divided by recent selling prices is as follows:
TS4 - 2.4%
LS28 - 2.9%
LS19 - 3.2%
IG 9 - 4.0%
TW1 - 4.7%
The margin of error gets bigger the higher the values are, but 3% of recent typical selling prices appears to be the bare minimum.
b) HM Land Registry now make all their price paid data for England and Wales since 2009 available online, split into selling prices for flats, terraced houses, semi-detached and detached houses.
Please note: "You will need to add the following attribution statement if you use or publish our price paid data: Data produced by Land Registry © Crown copyright 2013."
I downloaded price paid data for the year 2012 and using the magic of pivot tables and so on, I then worked out the weighted average selling price of a standard 3-bed semi in each of 2,200 postcode district. I ranked them in ascending order and typed in the average rents from www.home.co.uk for each 22nd district up to £500,000 and then one for each £50,000 increment up to £2 million (134 data points) and ended up with the following chart:
There is a distinct kink in that curve - it's steep at first and then flattens off. We get a better fit if we split it into two - houses up to £700,000 and houses from £700,000 upwards (the top two or three percent by selling prices):
The trendline for homes currently worth up to £700,000 is steep, it works out at selling price x 6%. We can add on typical Council Tax of £1,000 and deduct £4,000 annual maintenance costs to arrive at the site premium.
The trendline for homes currently worth more than that is much flatter, it works out at selling price x 1.5% + £30,000. We can deduct one-fifth of the resulting figure for annual maintenance and insurance costs.
But if we have to simplify this down to one single figure, the site premium is on average about 3.4% of current selling prices in 2012.
3. Indexing up recent selling prices
Establishing the notional 2012 selling price of an individual home/plot (and hence its likely site premium) which was actually bought and sold in the past year or two is simple and uncontroversial. These recent sales can be used to establish an average selling price for each type/size of home/plot in each area and hence the site premium for each. HM Land Registry publish actual selling prices for all sales from 1995 onwards. Obviously, the more sales data you have the more reliable the average, but to get more data you have to go further back in time and the historic selling prices are slightly less reliable as a guide to current selling prices. It is just a question of having a sensible trade-off:
a) According to the Nationwide's Regional Quarterly Indices, average selling prices across the UK are little changed since 2006, having risen until the end of 2008, then fallen again and have been bobbing around the same level since 2009.
b) According to HM Revenue & Customs' Table 16.5, 7.7 million homes were sold in the years 2006 to 2012, which is around a third of all privately owned homes.
c) So for all homes sold within the past six years, we can establish a reasonably reliable notional 2012 selling price, and this covers a third of all homes. So we have enough data points in nearly all areas to be able to place (relative) values on the other two-thirds of homes, we then multiply them by 3% and allocate them to the most appropriate LVT Band.
4. Factors/relative values
Ronan Lyons showed that most homes in Ireland can be valued to within a reasonable degree of accuracy by using just two "factors"; the area in which the home is and the size/type of house, see here. Of course, the more factors you use, and the more finely graded the options are (such as dividing the country into a thousand rather than a hundred districts) the better, the more accurate the final result, although this can lead to spurious accuracy and leads to more appeals.
a) According to the DCLG's Housing stock report 2008, England's housing stock consists of the following (figures for the whole of the UK will not be wildly different):
converted flat - 3.7%
purpose built flat: high rise - 1.5%
purpose built flat: low rise - 13.4%
small terraced house - 9.8%
medium/large terraced house - 18.8%
semi-detached house - 26.0%
bungalow - 9.4%
detached house - 17.4%
These are not the categories I would have chosen because HM Land Registry's price paid data uses only four categories. The average relative selling prices of the four categories, taking a semi-detached house as 100%, within any postcode sector are as follows:
Flats - 60%
Terraced house - 80%
Semi-detached house - 100%
Detached house - 160%.
b) We can do a similar like-for-like comparison as in 1. above for each of the main categories to establish the site premium for each type in each area. This will give us the typical relative values for the site premiums in most areas. Remember: it's only relative and not absolute values which matter. For example, if you are in a room with a dozen people all milling around and you have to guess how tall each one is in feet and inches, you'd struggle, but getting them to line up tallest on the left, shortest on the right is easy enough. If you are then told that somebody in the middle of the row is 5'6" tall, you can easily guess how tall the others are.
c) Let's say for sake of example, it turns out that relative values are as follows (if you multiply up the two sets of figures and take the weighted average, you end up with a nice round 1.0):
Small/high rise flat - 55%
Large/low rise flat - 65%
Small terraced house - 75%
Medium/large terraced house - 85%
Semi-detached house - 100%
Bungalow - 130%
Detached house - 160%
d) So to get the ball rolling, you can automate the initial valuation for each home/plot on the basis of i. where it is and ii. what it is just using the first part of the postcode (the "postcode sector") and choosing from the list. Clearly, this is just my very rough approximation. It would be better to calibrate the results according to actual rental values, the size of the plot and the overall size of the permitted building etc, then go back and adjust the factors accordingly, but this is all perfectly do-able. It's barely more difficult than the original Council Tax revaluation banding exercise, which was largely done "by hand".
There are of course different ways of calculating the site premium:
- taking rental values and deducting an amount to cover the "bricks and mortar" cost;
- working out an overall average rental value per square yard in each area and multiplying that by plot sizes
- using recent selling prices as a proxy for rental values
- we can do individual per home/plot valuations or use average figures for each category of home/plot
- we can choose smaller or larger valuation areas
... each with their advantages and disadvantages, so there is no point pretending that any method gives us some scientifically verifiable figure accurate to three or four significant figures, so it is then best to put all homes into Bands (like Council Tax bands) and all homes/plots in the same Band in each area pay the same LVT.
a) I personally prefer putting all homes/plots into bands which are twenty per cent "wide" because we can then have twenty-six bands (A to Z) and the ratio between the site premium/LVT on homes in Band Z is about one hundred times as much as that for homes at the top of Band A (the same ratio as we had with the old Domestic Rates).
b) Using the average value of semi-destched homes in each postcode sector and applying the two trendlines from 2c above, and assuming the same number of homes in each sector, we end up with the following number of homes on each band and total revenues of £185 billion a year for privately owned homes in England and Wales:
Detailed information for Scotland and Northern Ireland is not so easily available. The average current selling prices for homes in those countries is £150,000 and £130,000 respectively, much lower than the average for England and Wales of £250,000.
Assuming there are two million privately owned homes in each country, we can pro rate the £185 billion, which gives us revenues of £12 billion for Scotland and £10 billion for Northern Ireland, total revenues £207 billion.
I have left social housing out of these calculations. The average rent plus Council Tax demanded for social housing is about £100 a week, £20 billion a year in total, but this nets off with £20 billion Housing Benefit being paid out, so it's easiest to ignore both sides.
c) The system is not unduly reliant on the number of homes in the upper bands. If we re-allocate all homes in Bands T and above to Band S, so that the maximum tax bill on any home (regardless of its true rental value) is £20,000 a year (assuming two adults claiming the personal allowance), there would be 523,000 homes in Band S and total revenues drop to £190 billion.
According to HMRC's Table 3.4 there are currently 560,000 individuals who earn £100,000 a year or moreSeventy per cent of those earners are employees, who currently pay/suffer £38,000 a year in PAYE/NIC and (say) another £10,000 in VAT, Council Tax etc.
With a flat tax of 20% on incomes and a net LVT bill of £20,000, those top half a million earners would see modest tax reductions, even if they live in the top half a million homes by value.
d) Nine-tenths of all homes will end up in one of ten bands (highlighted in yellow).
So assessing the site premiums of all but the most extraordinary or unusual homes/plots and allocating them into one of eight bands is a doddle. We just have to allocate unusual homes/plots (converted windmills, houses with unusually large or small gardens, flats over shops, oddly shaped plots squeezed in between or behind other buildings etc) into whichever Band seems most appropriate, bearing in mind the open market rental value, likely selling price, plot size, the size of the building itself and the permitted maximum size of building on that plot etc.
Yes, we have to accept that there will be a flurry of appeals against the initial assessments.
Tactically/psychologically is makes sense to remember the following: if a home/plot has a site premium of "about £10,000" (i.e. somewhere between £9,000 and £11,000, there are different ways of calculating it) then it is better to assess it at £8,000 and charge 100% LVT (i.e. £8,000 per annum) than it is to assess it at £10,000 and levy 80% LVT (also £8,000 per annum). There would be far fewer appeals against the former than the latter, in the same way as appeals against income tax assessments are about whether a certain item is taxable or allowable - they are not against the rate of income tax itself.
a) Banding also reduces the number of appeals. If the raw site premium of your home/plot comes out at £7,300 and you go into Band K, you will struggle to argue that the site premium is less than £6,310, and it makes the appeals procedure simpler. The appellant has to show that his home/plot is in the wrong Band - i.e. that there are homes/plots in a lower Band with a higher site premium, he might be right, but that might lead to those homes/plots being moved up a Band.
b) There will be a clear set of rules for assessments, even if the result is a bit arbitrary in some cases. The appellant will have to show that the rules have not been applied consistently (i.e. there was a mathematical mistake) and/or that the home/plot in question is in fact much smaller than assumed, or suffers from some particular factor, i.e. is next to an electricity sub-station etc, which means that it should be in a lower Band (i.e. that the correctly calculated value is up to twenty per cent lower than the originally assessed value).
c) So the valuation authority can be relatively relaxed about allowing appeals during the initial assessment stage; if half of people appeal and half of those manage to get moved to a lower band, then that only reduces the tax base by five per cent. Even if everybody appeals and every home/plot is moved down by one Band, then the tax base is only reduced by 20% (£40 billion).
d) The main thing is that future increases in rental value are captured. If we significantly reduce taxes on earned income (by £200 billion), disposable incomes of tenants and first-time buyers increases and a lot of that increase will flow through into higher rental values (let's say by twenty per cent in the first two years, it might be much more). So even if every initial appeal is allowed, there will only be a shortfall in the first year or two; by the third year, site premiums will have risen by more than twenty per cent, so by the third year, the tax levied on homes in each Band will go up by twenty per cent and everybody will end up paying what they were first assessed at anyway.
Using Eric Pickles' estimate, the initial valuation exercise might cost £10 per home/plot, or even a few hundred pounds if there is an appeal (which is less than people pay to have an accountant prepare their Self-Assessment tax return prepared year in, year out). But keeping the tax base up to date is a very minor issue, it will cost pennies per year per home/plot:
a) Once allocated to a Band, there is little need to ever move a home/plot from one Band to another. In subsequent years, all that we need to know is by how much rents have changed in each area (keeping an eye on selling prices) and then the tax for each Band in each area is shifted up or down accordingly.
b) So if the site premium for an average semi-detached house in an area was assessed at e.g. £7,000 at the first valuation date (and which went into national Band K, £6,310 to £7,585, LVT due £6,310) and rents for semi-detached houses in that area have gone up by £300 a year by the next valuation date, the tax on Band K homes/plots in that area is simply increased from £6,310 to £6,610 and the tax for all other Bands in that area is increased by 4.8% accordingly.
c) Just like with Council Tax, once allocated to a Band, most homes will stay in that Band. In very cheap areas a semi-detached house will be in Band C or D, and in very expensive areas it will in Band X, Y or Z. After a few years, the tax on a Band K home in one area might be quite different to the tax on a Band K home in another area. This is no different to Council Tax Band D costing £1,000 in some areas and £1,500 in others.
8. Estimating the total tax base for UK residential land
a) Having established that 2012 selling prices were around 30 times the annual site premium and that the site premium is around 3.5% of 2012 selling prices, to arrive at a reasonable estimate of our total tax base, all we need to know is the total value (at 2012 selling prices) of all housing in the UK and times it by 3.5%. Please note: I am not recommending LVT based on 3.5% of future capital values, I am saying that the site premium of most houses is approximately 3.5% of its current potential selling price.
b) The figures quoted by Nationwide and Halifax of around £160,000 are misleading, the mathematical average is far higher than that, as Acadametrics explain. HM Land Registry's price paid data referred to above also shows that the average price paid for all homes in 2012 was £250,000. The average price paid in Scotland was £150,000 and in Northern Ireland was £130,000.
c)In January 2013, Zoopla said that the total market value of privately owned UK housing was £5,960 billion, I believe this to be exaggerated.
On closer inspection:
19 million privately owned homes in England & Wales @ £250,000 = £4,750 billion
2 million in Scotland @ £150,000 = £300 billion
2 million in Northern Ireland @ £130,000 = £260 billion
4 million units of social housing @ £100,000 = £400 billion
= Total current value (selling prices) £5,710.
d) £5,710 x 3.5% = £200 billion total site-only rental value for UK residential land of £200 billion a year, which is what I forecast in 5. above.