The Revenue Commissioners have published a technical paper setting out the method used to calculate the estimated property tax values to guide home owners in the self-evaluation of the tax due on their property. The model is a hedonic econometric regression model, which is the standard method of estimating and tracking property values.
In many ways, the problem for the Revenue has not been the method to use, but assembling a set of data that can provide robust estimates. To this end, they have bought together property characteristics from a range of sources:
• Valuation data from the Revenue’s electronic stamp duty system, NAMA, and valuations commissioned by Revenue from professional valuers.
• Geo-Directory (the national address database): a list of all properties in Ireland, their type and location;
• Spatially derived data that indicate relative distances of all residential properties from a series of key amenities and services that add value to property;
• Geographically linked data from sources such as the CSO’s 2011 Census that provide characteristics about areas.
Sources of data used in estimating property tax
Having excluded a number of transactional data for various reasons (where the return indicates the transaction is a part of a larger series of transactions; where the return indicates a fractional interest in the property is being transferred; where the return indicates shared ownership; values under €25,000) the result is a dataset of 17,400, 15,000 and 19,200 transactions for 2010, 2011 and 2012 respectively. This set is relatively small as it based only on transactions since 2010.
34,400 (67 per cent) of these 51,600 properties were successfully matched to a Geo-Directory address point, thus providing property values for units at known locations. These are spread:
• Dublin (Dublin City Council, Dun Laoghaire-Rathdown County Council, Fingal County Council and South Dublin County Council) – 15,693 properties;
• Other cities (Cork Corporation, Limerick Corporation, Galway City Council and Waterford Corporation) – 3,039 properties;
• Rest of the country (all remaining county councils) – 18,014 properties.
These properties with known values provided a basis on which to estimate the value of near-by property with similar characteristics in each Electoral District in the country. Ideally, it would have been preferable to provide estimates for Small Areas given the variation in stock in an ED, but the relatively small data set precludes doing this robustly and EDs are the best that can presently be achieved.
Given that the value of property varies geographically, rates differ across the country, with the highest mean values in Dublin and the Leinster region.
The vast majority of properties are valued in the lower bands, reflecting the 50% fall in residential prices since 2008. Indeed, 91.3% of residential units are estimated to be worth less than 300K, with 60.6% less than 150K. I can imagine that the low estimated value of many properties might come as a shock to many homeowners and reveal the extent to which they might be in negative equity.
What this means is that the property tax value for 60.6% of residential property owners is estimated to be €225 or less per annum. For another 22.7% it will be €315, 8% it will be €405, and 3.1% it will be €495. Only 8.3% of residential property owners will pay €585 or more.
As with any model, the one developed by Revenue is not going to be 100 percent accurate and they provide some estimate as to the probability that the valuation is in the correct band. 91% are either in the right band, or one band above or below.
This suggests that there might be some horse-trading between property owners and Revenue as to whether they are in the right band or one very close to it, but in the main their estimates will be quite near to the actual value and the haggling will be over c.80-90 euro difference in cost between adjacent bands (with I suspect all challengers looking to move downwards).
To be fair to the Revenue Commissioners the approach they have taken is the industry standard and they have used all the possible data at their disposal. Their problem has been the small number of known valuations to work from and a lack of information about every property (type, number of rooms, etc). As such, the model seems to be as robust as it can be given the data constraints, though it is not without its issues such as having to provide estimates for EDs rather than Small Areas.