Errors in data supplied by Land Registry - implications for our overseas property report
We’ve become aware of an error made by HM Land Registry (HMLR) in the data that underpins our research report on property owned by overseas individuals.
Their error has not affected the primary findings or conclusions of the report:
There are more properties in England & Wales owned by overseas individuals than anyone previously knew (the corrected total is now 181,701, not 247,016 as HMLR originally reported)
There continue to be substantially more properties owned by overseas individuals than overseas companies
The numbers have still risen fast, with the total number of titles in this category still more than doubling between 2010 and 2021 (in fact, nearly trebling, as it rose from 60,366 to 181,701)
We still believe that HMLR should publish official statistics on this topic - if anything, this error only makes that more pressing.
However, HMLR’s error does have implications for some of the detailed findings, and may affect anyone who used the data in their own research.
We’ve updated all our research findings, and we’re writing this post to let people know about the details, what it means, and what we’re doing about it.
Details
In September 2021, we published a report on property ownership by individuals based overseas. This report was based on data from HM Land Registry that we now know to have contained errors.
During 2020 and 2021, we sent four FOI requests to HM Land Registry (1, 2, 3, 4), all requesting that HMLR supply data on:
“the number of Land Registry titles owned by individuals with an overseas correspondence address”.
We received this data across the dates requested, analysed it and published a report on the findings. The number of titles reported by HMLR to be in this category was around 250,000.
Subsequently, we requested more recent data from HMLR. But when we reviewed the new data we received, we spotted that the total titles had apparently fallen to around 180,000 - nearly a 40% fall in a series that was previously trending gently upwards.
Our instincts warned us such a big drop might suggest a data quality problem. So we asked HMLR to conduct an FOI ‘internal review’.
In response, HMLR’s reviews found an error in the data that they’d supplied for the previous requests. Specifically, HMLR wrote that the data they originally supplied:
“included both private individuals and organisations, when this should have been limited to private individuals”.
In other words, HMLR’s original response had in error included titles owned by overseas companies (a dataset that HMLR already publishes elsewhere), as well as overseas individuals.
Unfortunately, as all the figures were aggregated, there was no way that we could have spotted the error at the time - even though we did considerable due diligence.
What this means
This means that the raw numbers in our original report were too high, through no fault of our own. We have since requested and received corrected data for the period covering our original report.
The corrected data does not affect the key findings of our report, which was that:
There are more properties in England & Wales owned by overseas individuals than anyone previously knew, and more properties owned by overseas individuals than overseas companies
The total number of titles in this category has trebled since 2010.
Nor does it affect its main policy recommendation, which was that HMLR should improve the data available on the topic, or the findings on the geographical spread of titles, which are still concentrated in London and in northern cities like Liverpool and Manchester.
But it does affect some of the detailed findings. In particular, HMLR’s new data shows fewer owners are based in tax havens, like Jersey and British Virgin Islands, than their erroneous data did.
What we’re doing
Some of this data was widely reported, so we’re taking steps to let people know about the implications of the error.
We’re aiming to communicate the changes transparently and in full:
We’ve flagged the issue and linked to this post from our original blog post, briefing, report, GitHub repo and the four original FOI requests (1, 2, 3, 4)
We asked HMLR to supply corrected backdated data from the original requests, from 2010 to 2021, and used it to publish an updated version of our original report, clearly flagged as such and with all changes marked inline
We’re proactively contacting journalists and researchers who requested access to the raw data to let them know about the issue
We’re publishing this post to let people know about the problem.
In summary
HMLR’s error is unfortunate, to say the least. People are entitled to rely on data provided under Freedom of Information requests.
But we’re glad we returned to the issue and spotted the discrepancy. And we’re grateful that HMLR has been open with us about its methodological error.
We also think this shows why it’s important for the Government to publish official statistics on topics of significant public interest, as our report argued. Data this important shouldn’t be supplied via an ad hoc process; it should be properly quality-controlled via official statistics channels. As our report argued, Scotland provides official statistics about property ownership by overseas individuals - England should too.
We’ll carry on working to fix the problem, and we’re very happy to talk to anyone who has been affected - please get in touch if that’s you.