House prices in Amsterdam rose 46% between March 2015 and 2017, research by UvA economist Martijn Dröes and others shows. A new analysis tool gives insight into the different constituents of this price development. It is not just the rise that is surprising, so are the its underlying components.
‘It is not just the rise in prices that is unusual, but also the impact of new, non-fundamental factors that move Amsterdam home prices, such as Airbnb, Brexit and the popularity of Amsterdam among students’, says Martijn Dröes, assistant professor real estate at the UvA and the Amsterdam School of Real Estate, in his office at the UvA.
Amsterdam is not alone in this. In cities like Toronto, Berlin and London as well, factors that were non-existent or having little impact until recently are also driving up prices.
To explain the price development in the Amsterdam housing market, Dröes and his former classmate Ryan van Lamoen developed a test for breaking the price development into four different parts. With the test one can find out to what extent the price development diverges from what would seem reasonable given the traditional underlying factors. And – by extension – whether a price rise can be qualified as ‘exorbitant'.
To this end, Dröes and Van Lamoen, currently supervision specialist at the Dutch Central Bank (DNB), analysed transaction data of homes sold between January 1990 and March 2017, found in the database of Dutch realtors association NVM. Aerdt Houben, director of Financial Markets at DNB and professor of Financial Policies, Institutions and Markets at the UvA, helped fine-tune their ideas.
Subsequently, they applied the test to homes sold in Amsterdam between March 2015 and March 2017. In the Netherlands as a whole, median home prices rose 5% during that timeframe, whereas prices in the capital surged 46%. Dröes: ‘We’ve never seen such high prices and such large price jumps. Our analysis shows the market is overheated.’
By gradually parsing the price development, it can be corrected for the impact of traditional, fundamental factors. First for the impact of observable features, such as living space and home type. Then, for standard macro-economic factors such as interest rates. Third, the price development is corrected for the impact of non-observable features such as new neighbourhood facilities.
A fourth component remains: the so-called unexplained part of the price development.
Although based on logical thinking one could say that factors like Airbnb and Brexit are propelling home prices, the mechanisms behind it are insufficiently clear to economists to frame this part of the price rise in a formula. ‘It is important to know that, even if we have a test and all available data, we are unable to interpret everything. There’s still a black box.’
Nevertheless, Dröes considers the test a major leap forward: ‘So far, the regional effects of macro-economic factors in aggregated studies are averaged out. You get a graph with one line for the Netherlands and another line above that for Amsterdam. We partly explain why the graph looks like this.’
Research published in May by DNB showed that the price rise in the Amsterdam housing market, and in other markets as well, was not fuelled by credit.
Using their test, Dröes and his associates illustrated that part of the price development in Amsterdam can be explained by factors that can be framed in formulas; the first three components. Another part of the price development is more difficult to fathom: ‘Regardless the approach taken, the rise during the latest timeframe is striking, and can no longer be explained using normal formulas.’
Although the Amsterdam market is clearly overheated, Dröes does not want to speak of a bubble. The price rise is exorbitant – it accelerates quarter after quarter – but he feels there are still too many unknown factors to be able to call this a bubble. Further research is required to quantify the unexplained part of the price development. ‘But it’s clear that the premium paid over the asking price – adjusted for all the corrections – has risen twice as fast as it declined after the crisis, and that the premium paid has never been this large.’
Dröes is satisfied with the research findings, and emphasises that economics is not an exact science. ‘One can come a little closer. We’ve come closer, we’ve come to a level of detail we didn’t think was possible.’
He sees many opportunities for further research. Who is buying a house or apartment in Amsterdam? Is it meant as an investment or as a place of residence? And what is the role of Airbnb and its impact on the housing and rental market?
The test could be adapted with local variables to provide insight into the components that drive home prices in other cities or regions inside the Netherlands and abroad.
The larger the area involved, the more effort it will take to gather and process all the relevant data. In cases of wider use, the test could therefore be divided in two parts for practical reasons, Dröes suggests: first a simplified version, then – only for areas that seem particularly interesting – the entire test. ‘The first part will be a little less strong but applicable on a large scale. The results can be used as pointers for local further research.’
The test can give realtors better insight into the market, and consumers can get a better overview. Dröes: ‘Macroeconomic factors on realtors such as interest levels and income growth are not available’ websites. All one sees are the features of the home for sale and something about the neighbourhood. The test can help pull everything together.’
Since it signals overheating, the test can provide valuable information for policy-makers. Policy can stimulate supply in those areas that are most overheated and dampen demand by reducing subsidies and by improving the accessibility of surrounding areas, Dröes and his co-authors emphasise. Variants of the test can also be developed to gain more insight into other markets such as the bond market or the rental market.
Dröes is working on a scientific paper about the test: ‘It could become a benchmark.’