Replication Package
Replication Using Public Data
Identifying lease extensions using publicly available data only:
The analysis in the our paper uses newly available data on lease extensions. However this lease extension data is not public: the publicly available Land Registry lease data does not record lease extensions before September 2021. Here, we present a method for identifying lease extensions that can be applied to public Land Registry data, meaning that our results can be replicated using public data only. We then present results using only these publicly available lease extensions and show similar findings to the main text.
In the publicly available Land Registry lease data, when leaseholders extend, the original lease is overwritten in the Land Registry with a new, longer lease term starting at the same date as the original lease. For instance, consider an original lease for "99 years from January 1, 1980". Suppose this lease was extended for 90 years in 2010. Then, the original lease would be removed from the land registry, and replaced with a new, longer lease that read "189 years from January 1, 1980", where 189 years is the sum of the initial lease term and the extension amount and the start date of the lease remains in 1980. As such, the public land registry only records a "snapshot", which does not single out leases that have been extended in the past.
Algorithm to identify lease extensions using publicly available data only:
To identify lease extensions in public Land Registry data, we can utilize three facts:
(1) Two of the most common initial lease terms are for 99 and 125,
(2) The official and most common extension length is for 90 years, and
(3) When a lease is extended, the new lease has the same recorded length as the original lease, plus the amount of lease extension.
These three pieces of information suggest there should be a mass of lease terms for 99, 125, 189(=99+90) years and 215(=125+90) years, where the 189 and 215 year leases almost certainly refer to lease extensions. This is exactly what we observe in the data. The most common initial lease terms are 99 and 125 (with a smaller peak at 150), but we also see masses at 189 and 215, which are very uncommon initial lease terms.
In Appendix A.3 of our paper, we provide additional information indicating that a vast majority of leases with recorded terms of 189 and 215 years are in fact lease extensions.
Therefore, we can create a data set of lease extensions from publicly available data by following these steps:
Download the Land Registry Transaction Data and Land Registry Lease Data and merge using the fuzzy merge procedure described in the code.
Identify all leaseholds with lease terms of 189 and 215 years in the Land Registry Lease Data and flag them as extensions.
A. Set the extension date as the lease registration date.
B. Set the extension amount to 90 years.
C. Adjust the implied duration before extension to be 90 years less.
Follow the steps described in the paper to identify controls for these extensions.