Empty Homes Part 3

Empty Homes Part 3


In this final entry of the Empty Homes series, we look at Cornwall, Coastal Suffolk and rural Cumbria using the same methods we as in the case study of Kensington and Chelsea. We find that, as in the central London case, low-use homes are significantly more expensive; that there is a correlation (although less strong) between house price and percentage low-use homes; that low-use homes skew towards the expensive end of the market. This strengthens the argument that the problems caused by low-use homes are the same in both rural and urban situations.


As was discussed over the last two blog posts (Empty homes part 1, Empty Homes Part 2) empty and low-use homes are a hot topic being discussed in media through out the UK and indeed internationally with the UN discussing how residential property is being used for investment purposes. For a refresher we are calling low-use homes  a home that is not a primary residence, this means it is generally either empty or a second home. For more back ground see the previous article.

This article tries to answer the same questions as the previous article:

  1. Is there a relationship between house price and low-use homes?
  2. Are there patterns in the buying of low-use homes?
  3. What does it mean for the housing market?


The method isn't necessary to read, if you want a detailed understanding of how this was done read the previous blog post "Empty Homes Part 2". Otherwise skip to the results

The method in this article is almost the same as in the previous case study on Kensington and Chelsea in London. However the bootstrapping method has been slightly improved. Now the mean house price is calculated in each bootstrap sample and the difference between the bootstrapped means is compared. 

This article uses the same housing value classes as the previous article roughly coming from Savills descriptions

Class Lower Mid Upper Prime Super Prime
Price Range 0-490k 490k-750k 750k-2M 2M-12M 12M+




We look at 4 different areas in this article, Cornwall, Coastal Suffolk, Cumbria and the City of Hull. The first three areas are popular tourist destinations are often discussed as having many holiday homes, (Cornwall, Suffolk, Cumbria), Hull is less of a tourist hot spot but is taken as a counter balance to the other areas.


Coastal Suffolk

Cumbria and the Lake District

The City of Hull

The results are clear from the maps, that areas in famous tourist spots unsurprisingly have high numbers of second homes.

The percentage difference from zero tell us how many low-use homes there are above or below the expected amount in each region


The highest concentration of low-use homes in Cornwall is at St Minerva and Endellion with 44% low-use

The areas of St Ives, Padstow and Roseland all have levels of low-use homes over 25%. The highest concentration of low-use homes in Cornwall is at St Minerva and Endellion with 44% low-use. The bootstrapped boxplots tell us that Cornwall has about twice as many mid upper and prime low-use homes than would be expected considering the the value of homes in the County. Their is a correlation of 66% showing a strong link between low-use and price. Inferring the average price of a low-use property and a regular home we find that Low-use homes are about 20% more expensive than normal properties which translates to about £50,000 and is statistically significant.


On average low-use homes in Cumbria are about 20% more expensive which is close to £35,000

In Cumbria the highest levels of second homes are concentrated in the central Lakedistrict around Lake Windemere, the highest levels are in Ambleside and Grasmere.  Looking at the boxplot we see that Cumbria has more than twice as many mid and upper priced low-use homes than we would expect given the property values in the county, the amount of prime property (Values over £2m) is what would be expected. On average low-use homes are about 20% more expensive which is close to £35,000. The correlation between low-use and price is about 47%, so whilst positive is not as strong as Cornwall.

Coastal Suffolk

LSOA in Aldeburgh and Southwold have levels of LSOA reaching 41% and 42% respectively.

Coastal Suffolk doesn't have such a wide spread of low use homes, however it has very intense levels in parts of Aldeburgh and Southwold reaching 41% and 42% respectively.  although there appears to be less prime property than expected, meaning that luxury property is more likely to be a full time residence. Similar to the previous cases the property prices of low-use homes are about 20% more expensive than regular homes which is approximately £50,000. The correlation is 45% slightly lower than Cumbria and significantly lower than Cornwall.

Coastal Suffolk is also a good area to look at when analysing the weaknesses of aggregation, the area of Orford and Eyke, just south of Aldeburgh, has 24% low-use, compares to Aldeburgh's 41%,  however Orford and Eyke is 24 times larger. So although Orford and Eyke appears to have a lower percentage of low use homes, it would be reasonable to presume that the majority of the low use homes are concentrated in the coastal area near Aldeburgh, which is unfortunately something we are unable to see.

Kingston upon Hull

Compared to the other regions Kingston upon Hull has much lower levels of low-use homes maxing out at 12% in Myton, this is not surprising because as a port town, Hull is neither going to have a large number of long distance commuters or have large numbers of holiday makers. The distribution of homes in Hull suggests that there are quite a lot more low-use mid, upper and prime homes than would be expected, there appears to be slightly more, however as the absolute number is very low (less than 5 for prime) any deviation appears large. The small but significant deviation also shows up in the t-test which finds the difference between the mean price of low-use property and regular homes as significant, but that the difference is very small, only about £4000, meaning that there doesn't appear to be much of a second homes affect in the city. The corellation between prices and low-use is essentially zero which is expected given such a small difference in the means.


The results of the previous article and this one are very similar, low use homes are statistically significantly more expensive. What's interesting is that there doesn't need to be a strong correlation between house price and ward emptiness to observe this effect. Cumbria only had a correlation of 0.47 but still had a very clear price difference between low-use property and actual homes, (although explaining close to 50% of price variance from low-use property alone still seems pretty good). In a way this makes a lot of sense, if you have a region where some areas are very touristy you may not expect to see a large number of areas with lots of low-use homes, maybe just a few areas such as St Ives and Padstow in Cornwall or around Lake Windemere in the Lake district. If only a few areas within a region have high concentrations of low-use homes and these areas have higher prices, then low-use homes will have higher prices even if there isn't a strong correlation between price and low-use homes overall.

What does this mean?

Most likely low-use homes are more desirable and are pushing up local house prices by increasing demand

Well it means that buyers of low-use homes are either buying the biggest/most desirable, homes pushing local residents out of the area or into smaller/cheaper homes. An alternative is that purchasers of low-use homes are increasing the prices directly through increased demand, pushing local residents out of the area or into smaller/cheaper homes. In either case the results are the same. From the tests we have done we can't say which of the alternatives the low-use market is causing but it is most likely a mixture of both, that is better houses are being bought and they are pushing up local house prices by increasing demand.

We ask how much before the negative effects on the economy are outweighed by the benefits? This point goes well beyond the scope of this article but it is definitely something that warrants further investigation.

Increased prices are a problem in both rural and urban areas but rural areas can be uniquely effected due to the combination of low salaries and the resulting seasonal tourist economy. The low salaries mean that local residents are more effected by the increased prices making housing relatively less affordable. In addition high-levels of low-use tourist property in an area will make the economy strongly cyclical with large amounts of activity during the peak tourist season and a low the rest of the year. This can make running local businesses challenging, which then puts further pressure on local employment and and community infrastructure. In the case of London we asked how much is too much before the market becomes vulnerable to confidence shocks and house price instability, in the rural case we ask how much before the negative effects on the economy are outweighed by the benefits? This point goes well beyond the scope of this article but it is definitely something that warrants further investigation.

Overall it seems that the housing market is affected by low-use homes which have consistently shown to be more expensive than regular homes. Whether low-use homes push up prices or take the best and most desirable homes we can't say, although it is most likely a mixture of both. 

Final thoughts

When this project started we never thought the findings would be so interesting and at times worrying (we certainly didn't think it would take so long). The commodification or financialisation of housing is becoming an increasingly hot topic and looking at these results it is not surprising why. Although some believe it's a human right to buy as many homes as you want, others feel having a home should be a human right. If second/third/fourth homes increase as a thing, we can expect to see higher prices and local residents pushed further out or downsizing to stay within budget.  We don't propose any solutions on how to change the market but hope that the work done quantifying the effects of low-use homes on these last few blog posts can help shift the conversation away from strictly empty homes of which there are considerably fewer, and look at the concept of low-use homes more broadly as a way to understanding why house prices are behaving as they currently do.

What Next

An academic paper will be written up using the data and findings and submitted to a journal for peer review, if accepted it will be a big bonus to what started out as a side hobby. Pre-print and any eventual publications will be linked to from here as soon as possible.

Data obtained from the Government by FOI cannot be given away or "reused" with out a licence, this is why we have not been able to provide the data we have used for this work. The standard procudure for FOI datasets is to give an Open Government Licence or OGL. However due to a loophole OGL's are very seldom given and not without a fight, luckily there is a second loophole that can be exploited making the OGL essentially irrelevant. We will be demonstrating this loophole with some of the data we have been given, basically trolling the government for making such inconsistent legislation.

Doing it yourself

As mentioned above we can't give away the data, however we can at least give away what we did with it, go to the github account and marvel at a big bowl of code spaghetti

Reading Palms

Reading Palms

Empty Homes Part 2

Empty Homes Part 2