Subscriber OnlyEconomy

Forecasters keep underestimating new home completions - we need to get it right

Opinion: Arguing between 27,000 and 30,000 housing completions may seem like splitting hairs but it has important policy and political implications

A London property investor once told me the secret of his success was to eschew consensus forecasts. If they were correct this was commodity information and his margins would be competed away. But, more often than one might expect, he found the consensus to be wrong. He attributed this to a commercial bias in property research. So instead of following the herd this client sought-out contrarian insights that, based on supporting data and logic, had a high probability of being right. That is how he made his money.

Speaking at a conference last December, economist Conall MacCoille observed that the property industry has consistently under-predicted housing output in Ireland. This is true even at short forecasting horizons. For example, member surveys by representative bodies late last year projected around 25,000 completions, whereas the actual outcome was substantially higher at 29,851.

No doubt the industry can live with underestimates of housing output. After all the greater the perceived supply deficit the stronger the outlook for rents and prices, and the more political leverage it has to lobby for government subsidies. However, before assuming commercial bias it is worth noting that State bodies also have an indifferent record in forecasting housing delivery. For example, one institution actually trimmed its prediction from 24,500 last April to 23,500 in July, before eventually aligning with the consensus forecast of 25,000 by October.

As usual in spring housing delivery forecasts for the year ahead are now emerging. And, typically enough, there is a strong consensus. The Central Bank, Goodbody Stockbrokers, the ESRI and the Banking and Payments Federation have all recently predicted that 27,000 units will be built – 10 per cent fewer than last year and 7 per cent below the Housing For All target of 29,000.


This seems like a reasonable shout given that housing completions exceeded commencements last year and that development margins are being squeezed by build costs rising more quickly than property values. Nonetheless I think we will do better. The relaxation of Covid restrictions triggered a surge in commencements between April and October 2021 which has only partially washed through as completed stock. Furthermore, the lagged relationship between starts and completions has become more complicated in recent years. Apartments generally take longer to build, and their share of national residential output has risen sharply. The net result is a swollen pipeline of properties that are currently on-site, under construction.

According to the Dublin Housing Supply Coordination Task Force, 18,600 dwellings were under construction in the capital last September. Netting-off Q4 completions and commencements suggests that approximately 18,100 units remained on-site in December – just shy of two years’ supply at 2022 levels.

Assuming dwellings get completed in their starting sequence we can estimate that 10 per cent of these properties were on-site for 15 months or more as this year began. A further one third were under construction for 12-15 months, while a slightly smaller proportion had been under way for 6-12 months. These timelines suggest that housing completions in Dublin will rise significantly again this year. Even allowing for the possibility that Dublin could account for its highest ever share of housing delivery in 2023, this scales-up to a national forecast of around 30,000 units.

The 2024 outlook is somewhat ambiguous. Intuitively we expect the dip in commencements between April and October last year to eventually have some bearing on completions. And although construction inflation is slowing, costs are still rising faster than property values. Nevertheless, if 2023 output materialises as envisaged, a sizeable residual of properties that are already under construction in Dublin will remain for completion in 2024.

Meanwhile, there has been something of a rebound in commencements between November and February, at least in Dublin where starts are up 37.2 per cent year-on-year. Consistent with this, forward-looking indicators in the BNPPRE Construction PMI, including new orders, input purchases, employment and future expectations, are all pointing towards increased activity.

Arguing between 27,000 and 30,000 housing completions may seem like splitting hairs. But there are two important implications. Firstly, if completions reach 30,000 output will have continued to rise in 2023 rather than contracting per the consensus view. Secondly the Government’s policy target will have been surpassed rather than missed. Anticipation of this will be a useful context for policymakers as they inevitably face pressure for further market interventions later in the year.

Dr John McCartney is director of research at BNP Paribas Real Estate Ireland