In the last couple of days I’ve been asked to comment on two issues around property data, both relating to vacancy (though we could easily have a similar discussion with regards to housing completions, homelessness, etc).  The first relates to housing vacancy and a report by Fingal County Council that contends that the vacancy levels in the local authority have been ‘grossly overstated’. The second about commercial vacancy and present rates. In both cases it’s difficult to provide strong answers because systematic data collection with respect to both is problematic and the state does not provide official data on either, except on housing vacancy every five years through the census which is a sub-optimal timeframe to be working from.

With respect to housing vacancy. I can’t find the report or press release from Fingal CC, but a story in the Irish Times reports that they believe vacancy levels are well below those reported in the census. It’s difficult to assess fully whether that’s the case without seeing the full methodology or data. What is reported in the IT is:

“The council initially conducted a desktop exercise on the 3,000 supposedly vacant properties. When commercial properties, as well as those in construction or in the planning process, were eliminated the figure fell to 361 properties. ”  They then visited 74 of the 361 homes to check on occupancy, though it’s not stated how those 74 were sampled.

Of those 74 visited, they discovered that only 13 were actually vacant. In other words, rather than having a vacancy rate of 5% (as reported in the 2016 census – 4,944 vacant units + 289 holiday homes), they have a rate of about 1% – far below what might be an expected base vacancy level of 6% (there are always some units vacant due to selling, gaps between renting, working temporarily elsewhere, people in healthcare, etc.). I have no doubt in the 18 months since the census in April 2016 properties that were vacant will have been occupied, however it seems unlikely that vacancy is so far below base vacancy, which is what the IT piece seems to be suggesting.

In terms of method it is unlikely that the CSO shared the individual addresses of vacant properties as identified in the census with Fingal. But if they were working from census data then it does not include commercial properties, nor properties under-construction, or in the planning process, or derelict. So removing those properties from census counts would make no sense – they were never counted by the CSO. Indeed, in a rebuttal story in the Irish Times, the CSO stand over their data and method – which is to send enumerators to every property in the country, to visit upwards of ten times if they fail to get an answer, and to talk to neighbours to try and ascertain the use status. I’m assuming that Fingal got their data instead from Geodirectory who source the information on occupancy from postal workers delivering or not mail. How accurate those data are I’m not sure and presumably the company would stand over their fidelity.

Regardless of the method, there is clearly a large discrepancy between what Fingal CC are finding on the ground in their small sample and what the census enumerators found 18 months ago, and presumably what An Post workers are finding. That discrepancy suggests we need a much more systematic and timely way of generating data on housing vacancy.  The government have set up a crowdsourcing means to generate vacancy information – vacanthomes.ie – where members of the public can log homes that they think are vacant, which can then be checked by local authority staff. There are well known problems with crowdsourcing such information, including coverage, representativeness and keeping the data up-to-date, and these data certainly could not be used as official statistics. Much more realistic would be a quarterly vacancy survey (much like the quarterly household survey) – probably carried out by the CSO who have no vested interest in local housing/planning data.

In terms of commercial vacancy, the state produces no statistics on the rates of vacancy for offices, retail units or industrial sites. It is a massive hole in our knowledge of the property sector. The only data that are produced are those by Geodirectory (which are limited in detail) or the property sector itself (hardly an unvested party, and the data are a product and can disappear from websites or go behind paywalls, and lack spatial granularity – usually just Dublin/rest of country or regions). In relation to commercial properties there is also a need to understand their characteristics, such as type, spec, condition, location, etc. as well as the size of space vacant, not just how many units. For example, imagine that there are ten units on a high street.  Nine of them are 1000 sqm in size and one is 5000 sqm.  If the larger unit is vacant then the vacancy rate per unit is 10 percent. However, the vacancy rate by floor area is 35 percent.  In other words, one cannot simply look at the absolute number of vacant units, rather we also need to consider the type and size of the units that are vacant. Trying to prepare local and county development plans with a fuzzy knowledge of existing development is a sub-optimal way of conducting planning and can lead to oversupply and property crashes (as per the last 20 years). Like housing, we therefore need good, reliable, timely data to understand the commercial property sector and we need the state to produce them.

In my view, there needs to be a branch-and-root review of property data in Ireland. This needs to start with asking the question: what data do we need to generate to best understand planning, housing, commercial property, infrastructure need, etc? Then to discover where the gaps are and to review the veracity and fidelity and fit-for-purpose of existing data generation and to fix as necessary. This includes assessing whether the data are being generated by the most appropriate generator. We then need to put in place the processes to produce those data.

With good quality data that people trust we might avoid different agencies producing wildly varying estimates of some element of housing or commercial property, such as vacancy rates, and we would greatly aid our planning and economic development. However, if we carry on as we are, we’re going to continue to fly half-blind and only have a partial or flawed understanding of present conditions and we are going to replicate mistakes of the past.

Rob Kitchin