Over the weekend, Revenue Commissioners launched their guidance for the Local Property Tax, in the form of an interactive map. With just a couple of pieces of information (location, property type and whether it was built before or after 2000), it takes you to a map of the country where when you click on an area, it gives a guidance for the likely band for properties of that type in that area. So for example, detached homes in Dublin 4 are estimated to be in the “greater than €1m” band.
The calculator has generated much discussion over the last few days, with many people claiming that for their properties, the valuations are “way off”. Revenue Commissioners economist, Keith Walsh, was on a range of media outlets on Monday trying to explain that this is the starting point for a valuation, and not the definitive say-so on what your home is now worth.
Some people have phrased the question in terms of “how did Revenue Commissioners get it so wrong?”. To me, this is looking at it the wrong way. As soon as the Government had decided that it was not going to ask people to return in Year 1 the information needed by Revenue Commissioners to audit the system, Revenue were stuck. While there is a register of who owns what plot of land, there is no such register of what is on the land. There is a system – Geodirectory – that can tell Revenue what type of structure is at each address (apartment, semi-d, etc). But it can’t say anything about the size of the property in any detail.
And that is why the Revenue system is only a start point. No-one, certainly not Revenue, are saying that a two-up two-down terraced in Stoneybatter is worth the same as a five-bed Victorian house with a garden around the corner on North Circular Road. And if of course the owner of the latter tries to pretend that they are in the same band as the former, ultimately that will catch up with them.
But what impact does bedroom number and bathroom number have on the price? Working with the guys at Daft, I’ve been trying to help people out on that one. The result is the Daft.ie Local Property Tax calculator. You choose where you live, the number of bedrooms and bathrooms, whether you have a garden, and the property’s type, and it produces an estimate not just of your tax bill but also of the value of the property itself, as of Q1 2013.
How does it work?
The valuation you are given uses all information from the daft.ie archives since the start of 2011 – over 150,000 properties in total. For each property, information on location, size and type is known, meaning that the effect of these can be estimated. It is also possible – as is done every Daft Report – to capture how prices change over time, so that we can estimate relative prices (of say Cobh relative to Cork city centre) and not worry that this relative price is affected by when properties were listed.
For those who like the detail, the model is a hedonic price regression that – like the CSO’s index – uses a filter called Cooks Distance to exclude unusual properties which have a disproportionate effect on the results. Each property is assigned into one of five regions (Dublin, Other cities, Leinster, Munster and Connacht-Ulster) and one of 365 “micro-markets” around the country. (Meath, for example, has the following micro markets: Navan, Ashbourne, Dunboyne-Clonee, Trim, Enfield-Kilcock, West/North Meath, Kells, Laytown-Bettystown, Gormanstown+, Mornington-Drogheda, Dunshaughlin, Ratoath, Duleek+, Tara+, Meath (other), where a “+” denotes areas close by.)
Each property is also categorised by type, number of bedrooms, number of bathrooms relative to bedrooms and whether it has a garden. As mentioned above, it is also classified by month (actually by month and region, recognising that prices trends have varied across the country). The model then takes all the observations and through the magic of matrix algebra and modern computing gives a series of coefficients.
Those coefficients are actually factors, such as how the price of a 5-bed is relative to that of a 3-bed, everything else being equal. So, even with the same location, property type, and number of bathrooms, the model is telling us that 5-beds in Dublin relative to 3-beds are twice as expensive on average. This has obvious implications for the Property Tax.
Technically, the prices the model produces are asking prices, not transaction prices. Evidence from 2012 is that on average transaction prices were 10% below asking prices, so 10% has been deducted from the model output, to reflect market conditions.
And now the “buts”…
Of course, this is not the be-all-and-end-all either. While accounting for location, type and size will get you about three-quarters of the way in explaining variation in house prices, there is still a quarter to go. This is made up with factors that are not available across all properties, from things that hopefully soon will be measured, like size of the property and site in square metres and the building’s BER, to things are always going to be tough to capture, like the quality of the structure and of the finishing.
As with the Revenue system, this is meant to a step on the way, not the final destination. My advice for those whose properties are coming out with wildly different prices across the two tools is to first check the Property Price Register and if all those three sources don’t leave you relatively clear on your property’s value, you may need to get a professional valuation if you want to challenge the Revenue’s initial guidance.