There has been a certain amount of confusion concerning housing vacancy rates in Ireland over the past few weeks.  This is mainly a function of different data being conflated and how the results were reported and discussed (as noted by Pat McArdle in the Irish Times a couple of weeks ago).  Each study has produced three comparable pieces of data – vacancy including holiday homes, vacancy excluding holiday homes, and potential overhang (housing in excess of an expected base vacancy rate – there will always be some vacancy in a market).  The confusion has been caused, in part, by the comparison of the DKM/DEHLG overhang figure (122-147K), the NIRSA vacancy exc. holiday home figure (302K) and the UCD vacancy inc. holiday home figure (345K) as if they are estimating the same thing.  This has further been confused by CIF estimating that the overhang of brand new, unsold homes is c.40K.  Excluding the CIF estimate, there is general alignment between the four sets of estimates that have been put forward (the fourth set produced by Goodbody’s; there was also a global estimate put forward on propertypin that also has alignment).  All four organisations estimate that vacancy including holiday homes is over 300K, that vacancy excluding holiday homes is over 228K, and that the potential overhang is over 100K (and if the top rates are used for DKM/DEHLG and Goodbody then the alignment is relatively strong).

Table 1: Comparison of housing vacancy rates in recent studies

Table notes:

a)       Base vacancy refers to the expected underlying rate of vacancy normal in any housing market.

b)       Obsolescence refers to the expected number of houses falling out of the live stock because they are no longer habitable or change from residential use.

c)       Overhang is the number of units in excess of the base vacancy rate.  In Table 1, overhang (6) is calculated thus: (6)=(1)-(2)-(4)-(5)

d)       NIRSA has employed two different methods in its analysis.  The first method (not in brackets, see here and here) used house completion, address database and new mortgage data to estimate vacancy, but not potential overhang.  Their estimate of 302K included undercounted holiday homes but did not estimate how many (there were 49,789 in Census 2006).  In subsequent analysis to calculate potential overhang they used a projection of population household growth based on 96-06 household change, factoring in obsolescence and holiday home rates, so as to be able to estimate county rates (new mortgage data is not available at county level).  The two different methods lead to slightly different estimates of vacancy exc. holiday homes due to the estimation of their number (see note e), but overall vacancy inc. holiday homes is about the same.

e)       Using the second approach NIRSA estimated the total number of holiday homes at the end of 2009 as c.86K (using the DKM method of 5% of total stock in 2006 based on the 2005 Household Budget Survey [73K], and 5% of potentially available stock between Apr 06-09 [13K]).

f)        This is not the base vacancy, but the excess vacant stock in the 2006 Census (216,533) above an expected 6% base rate of total stock (106,177) minus 23,000 undercounted holiday homes (5% total 2006 stock would be 73,000 not 49,789).

The difference between the estimates is due to two factors: method and assumptions.  All four studies use the Census 2006 as their base.  The Census reported that there were 174,935 houses and 41,598 apartments vacant (not including temporarily vacant on census night), totalling 216,533 units, plus 49,798 holiday homes (an overall total of 266,322; a vacancy rate of 15%).  All four use housing completion data and new mortgages on new homes data to calculate vacancy exc. holiday homes to the end of 2009, with UCD also using population projections.  With respect to calculating potential overhang, DKM and Goodbody use new mortgage data on new properties along with an estimate of how many properties have been bought without a mortgage; NIRSA and UCD use population projections, holiday home demand, and obsolescence.  All four organisations use slightly different assumptions with respect to holiday home rates, obsolescence rates, how many houses were bought without a mortgage, how many newly bought houses are occupied, and expected underlying vacancy rate (e.g., with regards to the latter, UCD use 5%, DKM and NIRSA 6%, and Goodbody’s 7.3% [the European average]).  For comparison, as noted in the UCD report, the base vacancy rate in the UK is 3.4%.

There are a number of conclusions to be drawn from this.  First, there is general agreement across different bodies (government, private sector, academia) with respect to vacancy and overhang estimates (even if that isn’t clear from the media coverage).  Second, we need better data on housing vacancy and availability to the market.  Third, in the absence of such housing data there is a need to establish, as much as possible, some agreement on method and assumptions.  Fourth, in the reporting of vacancy data much more care needs to be excised to make sure like is being compared to like in order to minimise confusion.  To some extent we have been surprised by the amount of media coverage in recent weeks because the NIRSA and UCD figures are not too dissimilar to those reported by the Department of Environment, as prepared by DKM.