As discussed in the body of this paper, landlords such as Kushner Companies have historically under-reported the amount of rent stabilized units in their buildings on building permit applications. This practice allows landlords to complete construction jobs without the oversight from DOB required for rent stabilized apartments to ensure that construction is not being used to harass tenants out of their apartments. In the longer term, it may also allow landlords to illegally deregulate units simply by lying about the rent stabilized status of a unit. This would allow landlords to raise rents and remove tenants more easily.(Allegations that Kush...)
In order to identify buildings where illegal deregulation and the misreporting of rent stabilized units occured, our team employed a simple pattern recognition algorithm on the list of all buildings that at any point in time had rent stabilized units, according to nyc-db's count of rent stabilized units. The theory underlying this analysis was that the practice of misreporting follows a similar pattern in the data: a non-zero rent stabilized unit count for one or multiple years, followed by a rent stabilized unit count of zero for one or multiple year, then followed by a non-zero rent stabilized unit count for one or multiple years. This pattern is distinct from that of standard deregulation (both resulting from natural flows in and out of units and from landlord harassment) in that there is a decline to zero followed by an increase to a non-zero count of units. While there is a chance that this is because the building received some subsidy in exchange for increasing the amount of rent stabilized units, it is more likely that the count of zero is the result of misreporting.
To find buildings following this pattern, the rent stabilized unit counts for all rent stabilized buildings were converted from separate yearly columns into one string per building. This string was in chronological order (e.g. the first character was the unit count from 2007 and the last was the unit count from 2016), and each character was either an "N" (non-zero) or "0" (zero). A list of strings denoting the nonzero-zero-nonzero pattern was generated (e.g. [N0N, N00N, N000N, ...]) and each building's full string was checked against this list to see whether it contained one of the strings in the list. Those that did were flagged as potentially having misreported their unit counts.
There are two limitations of note with this approach. First, as mentioned above it may be the case that the increase from zero to some non-zero amount of rent stabilized units could be the result of something other than the illegal deregulation of a unit: the terms of some subsidy received may demand rent stabilized units, for example. To determine whether or not this may be the case, our team merged a list of properties receiving housing subsidies from the Furman Center(CoreData.nyc - Furman...) and the list of misreporting buildings generated through our analysis. Only five percent of buildings suspected of misreporting the number of rent stabilized units receive subsidies, meaning that subsidies can be ruled out as a reason for a zero to non-zero change in at least ninety five percent of buildings suspected of misreporting.
The second limitation is that there may be instances where the number of rent stabilized units could be misreported but not misreported as zero rent stabilized units. The approach we adopt will not capture these records. However, it is reasonable to assume that misreporting will almost always occur as zero units, given that a key motivation for misreporting is to not trigger a building inspection. If a landlord misreported the stabilized unit count above zero, then the building inspection would still be triggered. This means that, while hard to verify, it seems likely that most misreporting of rent stabilized unit counts will misreport the number as zero.
While this list is of value to regulators and housing rights advocates in isolation, we also use this list as a portfolio-scale indicator in our broader analytic approach. The indicator is the percent of buildings in a portfolio that is suspected of misreporting the number of rent stabilized units, and the baseline for comparison is the percent of buildings in NYC with rent stabilized units that appear on the list.