My two main areas of research are economic history and economic geography. In other words, both time and location matter when we want to understand economic outcomes, such as why certain cities or countries have higher living standards or faster growth than others. My D.Phil. at Balliol College, Oxford, which was awarded in July 2014, was on the economics of Ireland’s property market bubble and crash. Much of my research focus follows on from this and therefore currently I am working mainly on housing markets.
One strand of my research attempts to understand the impact of various factors – in particular conditions in the credit market and in the planning system – on the housing market, both in terms of house prices and units built. This involves developing measures of both credit conditions and planning conditions in the Irish context, as well as understanding the many interrelationships between house prices and other factors.
With the help of the team at daft.ie, I am also involved in measuring household expectations in relation to the housing market. What we think will happen house prices over the coming number of years has a huge influence on whether or not we want to buy. But what do we think will happen? And what shapes those expectations? I hope to find out.
Another strand of research I am actively working on is extending our frame of reference for the Irish housing market back, beyond the current ‘start date’ of 1980 or thereabouts all the way back to the mid-1800s. The first wave of globalization – roughly speaking from the mid-1800s to 1914 – was full of features we think unique to our age, such as sophisticated international capital markets, rapid technological progress and political backlashes to international trade. Thus, extending our frame of reference back can teach us a lot about what makes the current system different and whether it is stable or not.
This ties in with other cliometric (i.e. quantitative economic history) research I’m involved with. Together with Richard Grossman and Kevin O’Rourke, I have constructed a monthly share price index for Irish equities, going back to 1825. This lays the foundations for a range of studies, comparing trends in Irish wealth to similar trends elsewhere and to internal Irish conditions.
A final strand of research that I’ll mention here is amenity valuation. It is often said that markets have little to do with public services, such as schools, prisons and landfill sites, and almost nothing to contribute on what are termed ‘non-market services’, such as a walk in the park or summer sunshine. This is far from true, however. We pay for access to these amenities in part through the rent or mortgage we pay each month. I am analysing the relationship between house prices and a range of amenities, including proximity to schools, rail and road infrastructure, flood risk, and green space. A better understanding of the benefit these amenities give allows us to spend public moneys better.
A list of publications, working papers and projects is available here.
Recently, I published a working paper, entitled East, West, Boom & Bust: The Spread of House Prices & Rents in Ireland, 2007-2012. The abstract is below, while the full paper is available here.
Property prices in the Republic of Ireland fell by half between 2007 and 2012 on average, but little is known yet about how price falls differed across property type and location. This paper examines this issue, using a dataset of over one million property listings to calculate 2007 and 2012 prices for a set of standardised properties for over 1,100 sales regions and 312 lettings regions. It finds four stylised facts about the distribution of house prices and rents in Ireland during this period. Firstly, the spread of prices across different property sizes increased significantly in the crash. This is consistent with a “property ladder” effect during the bubble temporarily pushing up the relative price of smaller properties. Secondly and conversely, the spread of rents from largest to smallest property sizes fell between bubble and crash. Thirdly, there was at most a small fall in the spread of both prices and rents across space. Lastly, in both bubble and crash periods, the spread of rents was constrained relative to the spread of prices, particularly in the upper tail, a finding suggestive of renter search thresholds.
The heatmap tool is up and running on its own homepage. There are basically nine maps in there: house prices, rents, and yields (the ratio between the two), for 2007q2 (the peak), 2012q2 (now) and the change between the two periods. If you click on the image below it should open up in a new window and give you an overview of how to use the tool, which to be honest is pretty intuitive anyway. (The bookmarks are very handy – e.g. the shortcut to zoom in to Dublin – plus you can also set your own shortcuts.)
The tool is meant to be a public service – anyone can go on and find out typical prices and rents for the area they’re interested in, and how those have changed since the peak. The prices shown are like for like, weighted averages over a basket of five standardised property types, so the maps also show the contours of land value in the country (assuming for simplicity no variation in construction costs). (This should of course be incredibly useful to Government, if it is interested in implementing a site value tax… hint hint.)
Careful guys, here comes the science bit
The analysis that underpins all these figures is described in detail in a working paper of mine, which also outlines the main stylised facts of the structural changes in Ireland’s property market over the last five years. The paper sits in a literature, back in vogue after what’s happened in OECD housing markets in recent years, that tries to understand what happens in a bubble and crash.
In terms of methodology, all daft.ie listings, sales and rent, were taken for the period 2006Q1-2012Q2. Those whose location is known poorly (i.e. not to townland level) were removed, as were observations for 2008, to give discrete bubble and crash periods (2006-2007 and 2009-2012 respectively). Each listing was mapped, and then the greatest number of zones consistent with reasonable sample sizes in both bubble and crash periods was created. These zones are typically collections of CSO Electoral Divisions (outside cities) and Enumerator Areas (within cities). In the sales market, there are 1,117 zones while in the lettings market, there are 312 (mostly due to thin rural lettings markets in the bubble period).
In addition to its zone, each property was assigned to a regional market (Dublin, other cities, Leinster, Munster or Connacht-Ulster) and had rich information on size and type. These were interacted in a number of combinations (for the full specifications see the Working Paper), to allow not just the map of property values to vary between bubble and crash periods but also to allow differentials associated with particular property types and sizes to vary, not only between bubble and crash periods but also across regional markets.
That was the boring science bit. The conclusions are, happily, more interesting.
In the sales segment, the marginal price of space rose substantially in the crash – put another way, the fall in price of a five-bedroom detached house (48% on average) was significantly smaller than that of a one-bedroom apartment (62%). The differential between these two properties increased from 118% to 164% nationally. An overview of the price differentials and how they changed between bubble and crash periods is given in the graph below.
In the lettings segment, the opposite was the case: the marginal price of space fell in the crash. The differential between a five-bedroom property and a one-bedroom property narrowed from 97% to 82%.
The geographic spread of both prices and rents was largely preserved across bubble and crash periods, falling only slightly in the models presented. The Gini coefficient of prices fell slightly (in Model (4), from 25% to 24%), similar to what happened that for rents (20% to 19%).
As suggested by these Gini coefficients, it is clear that the spread of rents was significantly less than the spread in house prices in both bubble and crash periods. Most of this appears to be concentrated in high-amenity areas: while the 50:1 percentile ratios of prices and rents were similar (1.89 compared to 1.93), the 99:50 ratio was substantially greater in prices (3.02 compared to 1.86).
What does all this mean? Is this useful at all, besides having pretty maps? Well, from a research point of view, the contrast between the first and second findings above suggests a model where income and substitution effects apply in different situations. Income effects appear to dominate where there is no outside option (demand for accommodation is income-inelastic if you are a renter): space in the crash is a luxury for tenants. Where there is an outside option – many would-be first-time buyers held out during this period – substitution effects kicked in: if you are going to buy, you might as well buy big. An idea for future research is modelling tenure choice with the real estate cycle in mind.
Location is ultimately a short-hand for a bundle of amenities, ranging from labour and consumer markets to social and natural capital amenities. The final finding – that the geographic spread of rents is in some sense constrained at the upper end of the distribution – is consistent with either renters under-valuing certain amenities, for example due to search costs, or with buyers over-valuing those amenities, for example due to fear about future access to an amenity that is in fixed supply (e.g. access to schools).
Buyer over-valuation due to a desire to lock in access to amenities would reasonably be at its most acute in a bubble and be seen in pro-cyclical pricing of attributes and amenities. This is an issue that Ed Glaeser, one of the leading urban economists, has been writing about using U.S. data for the period 1996-2006. He and his co-authors find evidence of pro-cyclical pricing of amenities. The evidence here, however, certainly in relation to attributes, is that pricing was counter-cyclical, suggesting that it is renter under-valuation at work. (I explore this is more detail in another paper.)
Lastly, the price and rent series constructed in the course of this research extend naturally to a price-to-rent ratio for each of Ireland’s 4,500 Census districts. Up next for me is the description and explanation of the significant variation over time and space in this fundamental barometer of the property market.
Over the past month, I’ve given a couple of talks on the Irish economy and the Irish property market in particular. While not exactly following the model of the “Single Transferable Speech” adopted by some, there was understandably – given the similar topics – a good deal of overlap between talks given to the public at the Central Dublin Library and an-EU sponsored conference on the Irish Economy in NUI Galway.
Both talks build on not only some of the academic research I’ve been doing recently but also material that only exists thanks to this blog and the feedback from readers, such as this post “Are we nearly there yet?“, comparing house prices now to their long-run level and to incomes and rents in Ireland since the 1970s.
A video of my talk in Galway is up on Vimeo here, while the slides are available both on Slideshare and on Scribd: both are embedded below also. All five sessions from the Galway conference are up on the Digital Revolutionaries Vimeo page. John McHale’s presentation contained a really neat graph with revisions to Ireland’s growth expectations by various bodies over the past 18 months, while Aidan Kane’s talk contains lots of fascinating information on Ireland’s historical debt issues, going waaaay back into the 1600s!
My proposal – full report here – was relatively straightforward: use the best information we have currently (1.3 million sales and lettings ads posted on daft.ie between 2006 and 2011), and the best methods available for establishing the components of house prices (hedonic price regressions) to implement the best known form of taxation (Site Value Tax) on an interim basis, in an area where Ireland desperately needs new revenues: residential property. And when better information becomes available – in particular the Revenue Commissioners register of transactions – then that can be used for a full Site Value Tax. My map outlining relative land values in 4,500 districts across the country is reproduced below.
SVT: the sales pitch
A Site Value Tax (SVT) is an annual tax that is paid on the value of the land that you own. If you own a four-bed semi-detached in suburban Dublin worth €400,000, you can think of that €400,000 as being the value of the building (say €300,000) added to the value of the land (say €100,000). Your tax bill would be something like 2% of the €100,000.
Why do I say SVT is the best known form of taxation? Ultimately, because it’s fair and efficient. It’s that rarest of taxes, popular with not only both left- and right-wings but also with environmentalists. Left-wingers like Site Value Tax because ultimately real estate is the single biggest form of wealth – and what left-winger worth their salt doesn’t like a wealth tax? Right-wingers like Site Value Tax because it does not distort economic outcomes: land can’t go anywhere, unlike pretty much every other input you can think of, so just because it’s taxed doesn’t mean that rents have to go up or that business has to move elsewhere. And environmentalists like Site Value Tax because it encourages the best use of land, which is a scarce resource. Why would you keep a site derelict if you’re getting taxed as much as the same the guy next door making that land work?
A Site Value Tax is particularly appealing when viewed in the context of local government. Think back to why any land you own is worth more than agricultural land. It’s because of a range of amenities to which your land offers access – from public services like education, health and public safety, through environmental amenities like urban green space, coastline or lakes, to more intangible services, such as access to thick labour or consumer markets. Unlike, say, the value associated with shares in a firm, no one person or group of people creates the value associated with land – society does and thus SVT is the return that society gets for creating the amenities we enjoy.
SVT: glitches and hitches
The Irish property-owner, however, may not be as impressed with such lofty talk. What about those who bought in the boom and who are now stuck in negative equity? What about those, such as elderly couples, who are land-rich but cash-poor? What about those who live on rural sites that might be ten, fifty or even a hundred times the size of those in urban homes?
A Site Value Tax fundamentalist would say that none of this matters. Those who bought during the boom are not going to un-buy because this tax is brought in and besides the SVT reflects current land values, not bubble-era values. They would argue that those who are land-rich but cash-poor should be encouraged to move on, as a country where every set of parents who refuse to downsize on retirement push their own children’s homes further out. And a country of large rural sites imposes greater costs on urban dwellers subsidising their scattered neighbours – thus a site value tax should – and would – reflect this, they would argue.
A Site Value Tax realist knows that these things matter. Hence I prepared a series of FAQs in the full report prepared for Smart Taxes and the Department of the Environment, available here. Bubble-era buyers, for example, could be given a graduated tax credit from introduction of SVT for a five-year period (similar to mortgage interest relief). After this, 2004-2008 buyers would be liable for the same amount of tax as their neighbours on similar plots.
Those with large plots of land but little income – in particular pensioners – could easily be accommodated with the use of lien on the property, where the tax bill is postponed until the property is ultimately sold. And rural dwellers will almost certainly pay less than their urban counterparts anyway, with such a large differential between urban land values and residential ones. Allowing rural landowners to decide once and for all which of their land is residential and which agricultural would also assuage fears that rural life would be irrevocably destroyed.
In truth, any lobby group can be accommodated – we just need to be clear that this is what we’re doing. We’re shifting the burden from one group on to the rest of society. We may have good reasons for doing this but we shouldn’t fall for emotive arguments that try and disguise that.
One lobby group I think we should pander to is people, as opposed to empty land. What do I mean by this? Suppose we have two adjacent 1-acre city centre site. One owner leaves theirs empty (Case A) while the other builds 100 apartments, each worth €200,000, and rents them out (Case B). In case A, the site is worth €5m and in case B the block is worth €20m, of which the site is worth €5m.
Under a full property value tax, the empty site has a liability of €25,000 while the apartment block pays €100,000. It seems very unfair, though, that the site being used productively, from society’s point of view, has to pay the burden of the tax. Under a 1% SVT, both sites would pay a tax of €50,000 – already the person leaving their site empty is being encouraged to use their site to generate a return.
Now, suppose there were a 2% site value tax but with a per-person “green space” tax credit of €250 (roughly 1% of an acre per person, at a nationwide average per-acre value of €25,000). If the 100 apartments housed 250 people, that would mean the apartment block receives tax credits of €62,500, off their bill of €100,000, while the empty site is hit with a full tax bill €100,000. This modified SVT would shift the burden of taxation on to zoned-but-undeveloped land.
SVT: making us rich?
How much could SVT raise? The tax I proposed worked off an assumption that the Goverment would like to emulate best practice in this area and generate about €2bn in residential property tax annually – this is where the €625 per household figure from the press came from. If applied to commercial property also, a full SVT which replaced commercial rates, stamp duties and the 80% windfall tax, would constitute about €1bn in new revenue streams.
The natural response of anyone to the suggestion of a new taxation averaging €625 a year is “No thanks” (or possibly worse). To argue this, though, is effectively an argument for higher income tax and higher VAT. This is because everyone agrees that Ireland needs to raise about €4.5bn in new tax revenues over the coming years (€4.65bn by 2015, according to the 2011 Medium-Term Fiscal Statement) and even if organic growth delivers, as the Department of Finance expects, €1.4bn, that’s €3.25bn needed through fresh taxation measures.
Ultimately, there are only three types of taxes: those on incomes (which hurt competitiveness), those on consumption (which are bad for equity) and those on wealth, including property. So those who argue out of hand against a property tax such as SVT are arguing for the €3.3bn in new revenue streams to come entirely from some combination of income taxes or consumption taxes, both of which hurt jobs. And with Ireland’s VAT rate the highest in the world outside the Nordic countries and Ireland’s income tax rates among the highest in the world, the scope in these areas is limited.
For me, the key point is that when a full property tax is proposed, it needs to be done as part of an array of alternatives. The Minister cannot simply say, in the context of needing to raise €2bn: “Here’s our idea for a property tax. Do you like it?” No matter how nice the tax is on paper, the answer is going to be overwhelmingly “NO!”. Any property tax needs to be proposed as one of two (or three) options: “We can either introduce this property tax, or else we will need to raise the lower rate of income tax from 20% to 25%. Both harm disposable incomes but only one harms the creation of jobs. Which one do you want?”
A bold postcript
Note that the €625 average figure above was an input of the research (we need this to raise €2bn), not an outcome. The tax could be introduced at any level. At €100 a household, it could merely replace the household charge – but that’s not going to fund too many local amenities.
Alternatively, and much more boldly, the introduction of a 10% SVT on residential and commercial property – which would raise perhaps close to €10bn – could be done in conjunction with a reduction in VAT and a reduction in income taxes. This would help close the deficit, boost Ireland’s competitiveness, improve the fairness of the tax system and encourage efficient use of land. What’s not to like?
Yesterday saw some of the worst flooding in Dublin in recent memory. Today’s post examines the economics behind the floods, in particular what economics can tell us about these events and what they can tell us about economics. Building on doctoral research, estimates are presented of the effect that flood-risk, living near a river, has on house prices and on rents. The baseline appears to be a 5% cost for those living near rivers. Read more
This week sees the launch of “Next Generation Ireland”, a book bringing together ten emerging Irish researchers across a range of fields. In this post, I go through what I have learnt from my experience editing the book, about challenges as diverse as engaging our Diaspora, climate change and Ireland’s foreign policy. Read more
Last week, this blog won its second award, “Best in Blogging” at the 2010 Digital Media Awards. It’s an apposite time, therefore, to do a little bit of stock-taking. This post thanks those who’ve made the blog what is – especially the readership! – before launching version 2.0, which features research. The first research item presented is the Impact of World War On Labour Market Inequality: Insights from the Building Industry. Read more