“In God we trust, all others bring data”
The high density Rosebank node – along with Sandton and Randburg – can be regarded as one of the “new” CBDs of Johannesburg. Rosebank is serviced by the Gautrain on its eastern border, and located in between the north-to-south road arterials of Jan Smuts Ave and Oxford Road (becoming Rivonia Road further north).
Rosebank is surrounded by high end residential suburbs: Saxonwold, Houghton Estate, Melrose and Melrose Estate, Dunkeld and Dunkeld West, Parktown North and Parkwood.
Rosebank as an office node is currently seeing one of the highest high density residential developments rates in the country. Whether through residential new builds, or apartment recycles, or office-to-resi conversions, its office workers are being serviced with options for a great address in a (generally) small home.
This article covers three topical items of interest:
Rosebank, for purpose of this exercise, is defined by the following shape (polygon) – image below.
If you click on the button below left, you can see Rosebank in Google Earth. (To view, you will need Google Earth on your machine. If you don’t have Google Earth, the button to download it is alongside / on right)
The above shape, roughly linked to the 4 suburbs detailed above, and encompassing, in total, 184.7 hectares in extent (1,846,567 sqm), of which 68.3 Ha is office, contains however only 194 office properties (Note: this is commercial real estate (CRE), and it is complex – one property can comprise many buildings, each building can comprise many units or “pockets” of space).
More detailed commentary on the Rosebank definition and image¹ is available in the Notes section below. (For interest’s sake we have added % office vacancies² per suburb to the graphic above.)
The much-hyped WeWork (estimated value $47B) IPO is happening in a couple weeks’ time in September. Rosebank is home to WeWork’s first location in Africa!
Thus it is fitting, in our analysis of Rosebank, to understand how WeWork has performed in its beachhead site.
First some context:
So how did site no. 1 in Africa perform? WeWork, which opened 1 Aug, started marketing space in February this year. Within 6 months all space had been signed up – with a waiting list. Thus WeWork is 100% full, with the overflow of potential JHB customers currently being referred to Sandton, and two additional sites in the vicinity of Rosebank rumoured to be under negotiation.
WeWork’s 12,800 sqm in recent leasing deals ranges from Naspers, Redhat, Platform 45, OpenVantage and a major ICT company, with the balance being occupied by local smaller businesses.
Before we get into the numbers, let’s dedicate some time to academic rigour. If you’re not into the detail, feel free to skip this second section, and move on to the third.
For starters, academically, no definition is right or wrong. However, for us, it is critical that data is stored in a way that allows various space definitions to be applied.
Imagine there was a landlord, let’s call her Landlord A, who owned a property in a large node, with one of her properties being the only defined vacancy in the node. This pocket, made up of say 3,000 sqm, sub-divisible down to 500 sqm, was available immediately. However, despite this, Landlord A was not able to let that space. Worse, inexplicably, in parallel, every quarter Landlord A heard that deals, ranging between 500sqm and 3,000 sqm, were being done in the node. But, not with her. Yet she is the only vacancy. How? And why?
Sound practical reasons for this exist. Those tenants in signed leasing deals were instead going to any of one of the following:
These 3 factors explain why the landlord’s space is not being leased. The landlord is actually operating in a more competitive market. While one definition has the node as only 3,000 sqm vacant, the vacancy is actually greater. A wider definition supports this more competitive reality.
Taking the above into account, the below table walks us from a wider definition, taking the above competing options into account, through to a more narrow definition.
Here is the same information, in a waterfall chart:
So, as is obvious from above, Gmaven’s vacancies definition is wider, expanding its filter for the following
(Gmaven also includes a tiny fraction of complex space where its definition / categorisation according to vacancy schedules is unclear (i.e. it could be deemed to be “rent-attracting supplementary space“, or conventional gross lettable area))
One point on the above vacancy rate… If two on-risk developments (totalling 46.1K sqm ) coming onstream shortly are ignored, total Rosebank nodal vacancies would plummet to a (currently) industry-beating 9.1%. See the waterfall chart above “>=1d to 3 months”.
Definitions for “Total rentable office space”³, “Total area available for leasing”⁴ and “Total space coming onstream”⁵ can be found in the notes section of this article. In simple terms, total area available for leasing can be understood as the answer to this question: assuming a theoretical spike in demand and frictionless occupation, how many square meters could incoming tenants sign to occupy within 3 months?
So how does this compare to other nodes? Comparative nodes to Rosebank (relatively established, central business district, high density, close access to quality residential property) score the following (more “narrow” definition) on their vacancy rates:
The high weighting of P grade (premium grade) vacancy is due to two properties, separately owned, shifting 35,161 sqm and 10,939 sqm of space onstream in the next 3 months. While deals on this space are under negotiation, nothing yet has been concluded. Excluding the above, P grade vacancies for the next 3 months are hovering at a 4.2% vacancy level.
The above term is exactly that – what the landlord is asking. In periods of higher demand and lower supply, rentals are concluded at asking levels. In softer environments such as these, landlords may look to structure leasing agreements, resulting in rentals below asking.
Gross rental⁶ is defined here, and at the foot of this article.
Comparative nodal information on rentals can be found below
Source: Rode’s Report 2019:2 Office rentals
The image that follows talks to new supply coming onstream in future. Only on risk / confirmed developments are displayed below. Tenant-driven developments are excluded.
From the chart below, while significant developable bulk exists Rosebank, property owners are not developing new properties on risk. Instead such land owners are exercising prudence and waiting for tenants to “kick start” developments.
The “now” date is 26 August 2019. Thus 0 to 3 months refers to all vacancies coming onstream before the end of November 2019. 3-6 months is all space becoming available from late November to late February 2020.
The following table provides all known significant leasing deals, concluded in Rosebank from 2019:Q4 onwards. To emphasise, this excludes the majority (by count) of leasing deals closed for space smaller than 1,000 sqm.
The above excludes certain significant deals underway, but not disclose-able.
For more information on the brokered deals above, you can contact the relevant industry professionals below:
Commercial real estate remains an information-opaque industry…
Thank you to Rosebank area experts as follows:
The valuable comparative data provided by MSCI / SAPOA and Rode is acknowledged and appreciated. As is the calibration of our term definitions and “acid testing” of our logic – thank you Kerry Millar of RealFMG.
We have done our very best to ensure complete accuracy. However, in the instance of omissions or errors, these are all our own.
Gmaven processes 95% to 100% of public domain vacancies (inventory) in South Africa – at present an operationally intimidating 6.7M square meters of vacant leasable area, covering 9.4K units. As is obvious above, this data is aggregated and anonymised for consumption by third parties and industry outsiders. However, for professionals familiar with the Gmaven system, all data is stored at appropriately high levels of granularity (from tenant installation allowances, to escalations, to complex spaces, to breakdown between expense types and obligations).
All vacancies are stored within properties, and all properties are assigned, among approximately 800 other fields, fields such as a property grade, total rentable area, primary property category and co-ordinates. Every vacancy is time-stamped, and all data is stored within and queried from a data warehouse. All data in this warehouse is subject to quality processes and audits – from rigorous data validation and exception management to reconciliations. Further, all vacancies data is tested for duplication property fund side. This necessary (but misleading) practice by property funds has arisen because certain broker vacant space search algorithms / data structures do not handle sub-divisible or combinable units.
For those technically inclined, we run a data vault warehouse which feeds into a Ralph Kimball warehouse. The data is queried through an OLAP service.
Polygons inevitably will cleave suburbs (i.e. will not always include suburbs in entirety). The “casualty” (cleaved suburb) in this definition is Dunkeld and Parktown North to the north and west, Melrose to the east, and Parkwood to the south.
Office vacancy % = office space available for leasing divided by built office space’s total rentable area. The time parameter is from now to 3 months inclusive.
Means the total area available for occupation, within 3 months, of all office properties. This would exclude those properties ear-marked for redevelopment to hotel or residential, or those sites where development will be user-driven.
This definition is important. It is defined as all space (currently built or in development) available or becoming available for rent in the next 3 months.
Leasing processes can take, on average, 6 months from commencement to conclusion. Consistent with this, it is our opinion that space, not available immediately, but falling within the now to 3 month availability window ought to be considered for inclusion into total area available for leasing. For purposes of this report, 50,124 sqm of this space is included in the 98.8K sqms.
The above includes:
It excludes:
There is a grey area between Rentable Area (offices, support area lobbies, circulation space common areas) and Supplementary Areas. While balconies, storerooms and parking bays can attract rental, technically none of these ought to be counted in the Rentable Area. For purposes of this exercise, where office space cannot be separately rented without these Supplementary Areas (i.e. “Complex Space”), these supplementary areas have been included in total area available for leasing. The total of such supplementary space: 1,834 sqm (various available time bands).
Means all space becoming available after 3 months. This space is either built (with tenants terminating their leases), in construction and not pre-let, or due to be constructed on risk. All tenant-driven developments are excluded.
All asking rentals provided above are gross, and exclusive of VAT. Gross rental (more info here) can be defined as those rental and related expenses payable directly by the tenant to the landlord. This generally excludes, for example, rental-related expenses such as utilities (lights, water and sewerage), or internal cleaning. To deal with the elephant in the commercial real estate room, gross rental between owners can vary. Sometimes other “grey area” property-related expenses are paid additionally by the tenant, other times these are bundled into the gross rental number. Example expenses are: aircon and internal maintenance, additional security, elective generator levies, additional cleaning and hygiene.
Further to above, with rental holidays, tenant installation allowances and other tenant incentives, gross asking rentals can be hard to compare (on an “apples for apples” basis) between buildings. A skilled CRE professional will be able to navigate you through the financial and legal complexity.
To keep Rosebank office vacancies assumptions down, units to let providing only net rentals (4.6% of total) have been ignored. Average rentals have been weighted by area available for leasing. In event of complex spaces (spaces that are not separately rent-able) – for example including terrace, storage, balconies – differing rentals across space sub-categories have been weighted based on area. Non-office space (e.g. storage) rentable separately has been excluded
This is a key number in South African commercial real estate (CRE). After a lot of hard work, we are able to deliver a firm, defensible value to SA – relevant to businesses (who almost all operate out of CRE), policy makers and the industry itself.
For a South African CRE industry buffeted by Covid, weak economic growth and public sector incompetence, is solar power too good to be true? Is solar indeed one of property’s unicorns – a positively geared investment with increasing cash flows – and what are the risks associated with it?
Apple, one of the most profitable businesses in the world, is the king of outsourcing. Its brains reside in California, while its grunt work is spread to specialists all over the world. Is CRE missing a trick? Like the strategic minds at Apple, can it farm back office services to cheaper, specialist hands – and drive greater profitability?
