Column A crash course in the economics of flooding

THIS WEEK, THE east of Ireland – Dublin in particular – saw unprecedented flooding.

For a city that normally sees about 70 centimetres of rain over the course of an entire year, yesterday was like something out of the Bible: seven centimetres fell in just six hours. Yes, you read that right: 10 per cent of the entire annual rainfall in 0.07% of the year!

The previous record for a full 24 hours for Ireland was less than 5cm. Unsurprising, then, that the floods in Ireland was the most read story on BBC News as Monday drew to a close.

The Dublin Flood

It was also unsurprising, given the freakish volumes of rain, that the city was ill-equipped to deal with this, particularly as it happened at a time of year when drains are most likely to be blocked by leaves. A major emergency plan was announced after a number of rivers, including the Dodder, the Poddle and the Camac, all burst their banks.

Extreme weather events and major natural disasters are, clearly, hugely important events for those involved. Thus, they must be important for economists to understand in and of themselves. However, they are also important because help reveal to economists how humans actually interact with each other and with the world around them.

For example, economists typically rush to assume – as a simplification to make life easier – that humans have a reasonable understanding of risk (ie,  attaching the correct probability to an event) and uncertainty (ie,  assigning the correct range of probabilities to an event). This may sound very arcane but it has implications both for economics as a subject and for everyday life.

For economics as a subject, if how humans react to extreme events suggests major deviations from the rational homo economicus, then much of the larger scaffold of economics – including macroeconomics – will certainly have to be rebuilt, if not torn down. For the real world, there are significant policy issues in relation to, for example, insurance against flooding or wildfire risk. Insurance markets may be incomplete or governments – in acting with good intentions – may actually subsidise bad choices by households.

What economic research says

There is evidence – from an article in the 2009 Journal of Real Estate Finance and Economics called ‘Do repeated wildfires change homebuyers demand for homes in high-risk areas?’-  that risk is mis-perceived by households. The authors use repeated occurrence of forest fires in a small geographical area to analyse the impact of natural disasters on house prices and to better understand perceptions of the risk.

In their model, which includes about 2,500 homes from the Los Angeles area over the period 1989-2002, they find that houses that have been the victim of one fire are, everything else equal, 10 per cent cheaper than other homes. But as this discount only happens after the fact, wildfire risk does not seem to be priced in until it happens: households do not properly understand risk. Worse than that, houses affected by a second fire had their prices further revised down, by 23 per cent. So when households did price in the risk of fires, they did it incompletely. After all, the underlying risk to the house hadn’t changed, just the market’s understanding of it.

Of course, one paper cannot be definitive. In relation to flooding, it seems that households do factor in the risks… but incompletely. A paper from 1989 does not find buyer myopia in relation to flood risk. Looking at a town in the US that suffered from flooding more than a decade previous to the study, the author finds a floodplain adjustment of just over 12 percent on average.

A more thorough study from 2004 by Bin & Polasky in the Land Economics journal, on 8,000 homes in North Carolina, found that houses located within a floodplain have lower prices than houses located outside a floodplain, a negative differential that increased after a major hurricane caused severe flooding. This suggests household are aware of the risk but mis-perceive it.

Evidence from Ireland

In relation to flooding, and to what happened in Ireland on Monday, it’s worth noting that it is rivers, rather than lakes or the sea, that pose the greatest risk. While major arteries such as the M50 ring road were affected, many of the worst flooded streets were close to Dublin’s smaller rivers (its largest river, the Liffey, having relatively solid defences).

Part of my doctoral research at Oxford focuses on what determines the cost of accommodation. The factors range from access to good labour markets, schools and retail facilities to other amenities that may be typically regarded as too touchy-feely for economics: the value of living near the coast or the social capital that a Gaeltacht area brings. In an analysis of about half a million ads from the daft.ie website over the period 2006-2011, a number of environmental amenities were included, such as proximity to the coastline, lakes and rivers.

The graph below highlights the effect of those three amenities on advertised house prices and rents. The model is set up to strip out other measurable factors, such as house size and type, local unemployment, the maturity of an area, etc. So, to the best extent possible, these are the estimated effects of living near the coast, a lake or a river.

The most obvious finding is the difference between the positive effect of coastline and lakes and the negative effect of rivers. As these effects seem to taper off (ie they are strongest for properties closest and weakest for properties further away), it does seem as though this model is capturing the effect of natural amenities.
The effect of coastline, lakes and rivers on house prices in Ireland, 2006-2010

What the Irish housing market is telling us, then, is that while people will pay to live on the coast (especially in the cities) or on a lake, to live near a river the price needs to be discounted by about five per cent. Thus, this five per cent is a crude estimate of the costs of flooding factored in to people’s decisions on where to live.

What is particularly interesting for future research is the difference between the effect on house prices and the effect on rents: owner-occupiers appear to face more costs than those of the landlord and tenant combined.

The next step for this research is use the database that exists of flooding events in Ireland and see what kind of memory Irish households have in relation to floods. This might help disentangle the effect of remembering past floods and fearing future ones. Indeed, if the risk of floods is correctly included, the presence of a river close to a property should be a positive amenity like the coast or like a lake.

Ultimately, though, to understand properly flood-risk and the effect it has on people’s homes, and whether Irish people themselves are factoring the risk in correctly, maps of the Irish flood-plains need to be made available. I understand this is due to happen but, for obvious reasons, the timing is a commercially sensitive decision.

Ronan Lyons is an Irish economist based at Oxford University, and runs the Economic Research unit at Daft.ie. You can read more articles on his blog, where this originally appeared.

Read also: So what DID cause Monday night's floods?