Archive for ‘Research’

August 16, 2011

The one bedroom London property product


This is our first attempt at aggregating some of our quality & investment data from our ratings for two neighbourhoods (Bayswater & Fulham). The sample size is very small (just 174 properties) and all prices are based on asking prices.

For our customers who are interested in our 4000+ properties data sample, we will be bringing this live to our business plan accounts in early 2012. Expect the following levels of information:

1. The impact of a number of property features (e.g number of bedrooms, garden, porter, parking, amenities) on the price per square foot at the neighbourhood level.

2. Time on Market & Asking to Sold price ratio for specific type of properties. Useful in determining the type of properties that are selling fast.

3. The impact of a unit increase in Rankdesk’s quality score on the price per square foot (Rankdesk will be providing 4 values for price per square foot depending on the quality rating for all London neighbourhoods). Read more about our quality scoring algorithm in our methodology.

Here’s a flavour of what to expect:

When we look at some of these results according to bedroom size, we get some interesting patterns (hence the fancy title). The one bedroom product refers to a type of standardisation in one bedroom properties irrespective of the property location. Check out the smaller deviation in both rental, asking prices and property sizes for one bedroom properties compared to two or three bedroom ones. What’s the reason?  We would love to hear your comments below!

February 10, 2011

Should Rankdesk be downgrading or upgrading properties above shops?


Jane Jacobs would ask us to improve our methodology

 

A couple of months ago, during one of our weekly meetings, we came across an interesting dilemma. Should we downgrade or upgrade properties that are above shops? Put in simple terms, should our analysts be biased towards market opinion by assigning brownie points only to attributes that will have a positive impact on the  future property price of the unit in question and the welfare of the buyer?

When we interviewed home buyers and investors in 2009 (80 out of the 92 respondents) told us that they would prefer not to have retail below their property. This is not very surprising when one thinks that:

1. The majority of investors or home buyers would not want an empty shop below their investment.

2. Some mortgage lenders would not consider properties above retail units.

3. The uncertainty of future use is a major liability. What if a change of use attracts retail businesses that may negatively impact the local character of the area, or be a nuisance?

Our meeting became a bit more heated when we brought in the ideas of the great American urbanist, Jane Jacobs to the table. (We actually have a photo of Jane stuck above our ranking desk, so we are big fans).

In her The Life and Death of Great American Cities Jacobs argued that cities and neighbourhoods are vibrant when there is a good mix of use. Retail, residential and commercial space should be mixed together rather than being zoned apart. A number of positive outcomes emerge from this mix of use. Streets are busier for longer periods of the day, there is informal policing of the street with shopkeepers keeping an eye on the sidewalk and more importantly residential areas are served by local shops. One does not need to walk 10 minutes for a loaf of bread.

Jane Jacobs would argue that rather than just benefiting the occupiers of the property in question, allowing retail use at ground level could help uplift the area as a whole.

Somewhere between Jane Jacobs and the worried buyer who dreads to see the opening of the next late night kebab shop below their living room, is the Howard de Walden Estate’s model of regeneration. Here, a good mix of retail and residential can help increase the quality of life of the neighbourhood as well as positively impact the property prices in the area, which will benefit landlord, leaseholder and the freeholder alike.

 

The Marylebone High Street model of retail to residential mix

 

The Marylebone High Street example of carefully controlling the types of retailers on a street can be succesful. Indeed the Howard de Walden Estate has benefited from an increased value of residential leases in the area, which cross-subsidise the unique ‘village like’ retail outlets (whose ‘place making’ capacity drove residential prices up in the first place).

The central component in a succesful residential-to-retail mix is the presence of a long-term stakeholder who vets the types of units that would be allowed in the area in both the short and long-term future. This creates confidence both to those who use the street on a daily basis and to future buyers who would trust the retail selection process.

We are currently mapping a series of streets in London, where there is a clear synergy between retail and residential use controlled by such stakeholders. As soon as we have the results, we will reconsider our uniform downgrading of properties above shops.

Anyone interested to attend our next  ‘methodology review’ meeting, please email us on support at rankdesk dot com

We would particularly like to hear from Mortgage brokers.

February 10, 2011

How do we test properties? PART I: Testing Property Layout


Before appearing on our weekly rankings, our analysts answer over 60 questions on the quality and investment performance of each property. Here is a list of questions related to the layout of the property.

Size of living room

Space for dining in Kitchen

Storage in kitchen

Separate dining room

Long or badly planned corridors

Space for baby cot in master bedroom

Fitted wardrobes in master bedroom

General storage facilities

En-suite bathroom(s)

Bathroom(s) with shower only

Separate WC

Ease of access in at least one bathroom

Living room and kitchen close to each other

Living room and entrance close to each other

Entrance lobby prior to living room

Living room / Bedroom(s) adjacent to each other

The Housing Quality Indicators, Guidelines and Standards and Quality in Design (2008) form the basis of our layout analysis. These specify the minimum area requirements for a set number of bedrooms and bed spaces. This criterion looks at the amount of space required to accommodate furniture and the space required to perform activities typical of each room.
We also assess the layout of the main bathroom and of a second WC using the Wheelchair Housing Design Guide (2006). The guidance states that an effective bathroom containing a shower or bath, WC and basin should provide appropriate space for general manoeuvre to approach and use specific fittings. Other factors investigated in the analysis are fitted wardrobes, which are valued by buyers and are a requirement in The Housing Quality Indicators (2008); storage in kitchen and other storage within the apartment, which has been listed according to the number of bedrooms in the Housing Quality Indicators document. Properties with disproportionately sized rooms and long corridors have been downgraded because such characteristics increase the overall internal area of the apartment without improving the quality of space.

February 7, 2011

Winter 2011: How is Rankdesk improving its algorithm?


Rankdesk has developed the RQI Method to analyse the quality and investment performance of residential property in central London. Read our latest methodology here.

We have recently started applying 5 year capital depreciation rates on properties with leases below 100 years and are constantly working on improving our ranking methodology

These are the latest tasks keeping us busy this winter:

1. Assign 5 year capital appreciation rates at neighbourhood level. We currently use an average 5 year capital growth rate of around 20% which is applied uniformly to all central London properties. We are currently reviewing research from some of London’s leading residential market analysts to help us develop neighbourhood specific rates.

2. We are continuing our regression analysis of some 4,000 central London properties advertised in 2005 to determine how different property parameters such as floor area and block level services are impacting asking prices. The results will be used to verify the current weights used in our ranking algorithm, which were derived from investor interviews.

February 7, 2011

How scalable is Rankdesk?


Investors often ask us how scalable our business model is and if we have got plans to open further ranking desks in other cities.  The short honest answer is that we are ‘relatively scalable’!

This is often surprising , for a company that bases most of its day to day work analysing properties remotely (click here for our blog entry on the remote analysis of properties and the HITs model). Surely, we are told, you can start analysing properties in Shanghai, Melbourne & Toronto and provide global weekly rankings from some of the world’s leading residential markets from the comfort of your London desk.

There are four main tests that would determine Rankdesk’s analysis of properties in a new city.

1. The quality of the property data provided by the leading estate agents in a city. Our analysts rely on a minimum level of information in order to provide a preliminary assessment of the quality and investment performance of the property. Interestingly, the London market is characterisied by a culture of rich property data with at least 65% of the Agents in the market meeting our minimum property data requirements. These include:

  • A floor plan with a clear demarcation of the North orientation
  • Photos of each room and the exterior of the building
  • A location map
  • Tenure details
  • Property specific expenses such as service charges
  • A property description

The quality of property specific data is particularly relevant in markets where agents are answerable only to vendors.

2. Secondly, we  rely on rich information at the city or regional level. We would typically consider residential markets where there is a history of property price movements (based on actual transactions) in at least 2 property cycles. This data would also show us if the residential market in question is ‘built on solid foundations’ or on actual demand (growing city economy, labour market dynamics & household formation)  rather than just speculation. This makes Rankdesk particularly suited to more mature residential markets with their equivalent growth rates and yield expectations.

3. Local Knowledge. Because of the very ‘local knowledge’ required to understand the quality and investment performance of a property in a specific neighbourhood, analysing a property from a foreign based ranking desk will provide abstract results. The  HITs model would technically allow us to carry out this analysis, but having a centralised hub of analysts who are not fed ‘local’ data from on site observation of neighbourhoods and comparable properties would make Rankdesk less diligent. We are even going a step further and exploring the idea of a premium service where we would carry out an actual site visit of a property if requested by a client.

4. The openess of the market to foreign investment and to foreign ownership of property. Bringing a property on our rankings that would have equal relevance to home purchasers as well as to non-domiciled owners increases our user base exponentially.

We would love to hear of any cities that our users would find relevant in our lists and that meet the criteria above!

February 4, 2011

Remote ranking of properties, Human Intelligence Tasks and the noisy neighbours


Rankdesk begun life with a simple observation: The level of online property data on estate agents’ websites in mature residential markets is so developed that third party companies (like Rankdesk) can provide advanced property assessments remotely.

We knew that there were a number of Automatic Valuation Systems in the UK, the US and Australia, which relied on this type of information but which only used about 20% of the data available. So how do you make use of the remaining 80% of rich property data (photos, floor plans, location maps, environmental performance certificates, video tours and the written property descriptions provided by estate agents)? We looked at the growing sector of so-called HITs – Human Intelligence Tasks – revolutionised by Amazon’s Mechanical Turk. HITs are individual tasks that are nearly impossible for computers to perform efficiently (reading a floor plan, comparing the quality of two different properties etc.) but which can be performed remotely by humans on a computer.

Before packing our suitcases to travel to the World’s leading HITs hubs, we knew that property (which is heavily reliant on Local Knowledge) could only be performed locally. In 2008, we set up our first ranking desk in Baker Street in London and begun a 2 year development process on how the HIT model could be applied to residential property.

The colourful chart below shows our ranking desk in action.

There are 3 full time members of Rankdesk and 3 part time members. Our Head of Research (that’s Priya) collects macro level statistics of the central London property market (these include, capital appreciation rates, rental price growths, vacancy rates and even depreciation rates due to low leases). Priya is also responsible for interviewing property buyers to determine the importance of different property parameters (quality, net yield, accessibility) & their weight of importance when making purchase decisions. All of Priya’s research is embodied in our very own ranking algorithm that drives our online platform. Our 3 part time analysts now have the power of a web based property ranking application on which to carry out their analysis. Each analyst has two computer screens on their ranking desk (the left screen displays the property particulars as they appear on the estate agents’ website. The right screen displays our web application). Each analyst answers around 60 questions on the quality and investment performance of each property. Before the properties are published on our Wednesday rankings, our Head Analyst (who is normall busy walking around central London collecting neighbourhood intelligence) checks through all the answers and assigns rental values based on comparable properties on the same street.

Of course, there are limits to the HITs model. Since we don’t visit the actual asset, we will never be able to determine how noisy your future neighbours will be! We all rely on the expert ‘on-site’ knowledge of estate agents and surveyors, to help us with this very tangible parameters.

There is an audio version of this presentation!

Follow

Get every new post delivered to your Inbox.