PhD candidate in Automated Valuation Models for Real Estate
Amsterdam Business School in cooperation with ORTEC Finance
- 20 november 2017
- Master's degree
- €2,222 to €2,840 gross per month
- 15 januari 2018
- 38 hours per week
The Amsterdam Business School (ABS) is a partner of Amsterdam Data Science, a network consisting of the academic knowledge institutes in the Amsterdam Metropolitan Area, and worldwide industry partners that focus on stimulating research and education in Data Science. The ABS is part of the Faculty of Economics and Business (FEB).The FEB provides academic programmes for more than 5,500 students and employs about 400 people. The Faculty conducts research in many specialist areas and participates in the Tinbergen Institute, one of Europe's leading graduate schools in economics, finance and econometrics.
In many applications, property valuation plays an important role: local government need periodic valuations for tax purposes, Main Financial Institutions must determine the collateral value behind a specific mortgage to price the risk of default, and (institutional) investors need property valuations to set reservation prices when acquiring and selling properties, and to track the performance of their portfolio.
The basis valuation method is the direct comparison method; the assessed value is based on transaction prices of `comparable’ properties, where prices need to be adjusted for differences in characteristics between the subject and comparable properties.
Property prices depend on market conditions, location and property characteristics. Advanced econometrical models including temporal, spatial and cross-sectional components can help us greatly with pricing these shadow prices and mass value properties.
Automated Valuation Models (AVMs) are widely used to determine the market value of owner-occupied housing based on the aforementioned models. Examples are Collateral Analytics and Zillow in the US and Ortec Finance in the Netherlands.
However, for commercial real estate (successful) AVMs are not available, not in industry and not in academia. There is a number of reasons for this lack. The first is that the number of commercial properties is relatively low, and so is the number of transactions. Whereas the US has more than a 100M houses, which resale on average every 8 years, there are only a few 100 thousand commercial properties. The second is that the commercial property characteristics are not centrally collected, in contrast to housing where reliable and complete administrative data is typically available. In commercial real estate, data collection depends on private companies, which aggregate their data from multiple sources (mostly brokers and real estate newspapers). Data is therefore incomplete and more affected by entry errors. Data on maintenance, overall `architectural feel’ and structure quality are famous examples of data that is usually not observed in commercial real estate data at all. Finally, commercial properties are more heterogeneous compared to housing. In combination with the lack of collected characteristics this is a challenging combination. As a result econometrical models tend to have a low fit and out-of-sample performance.
We seek a highly motivated student with:
- an MSc degree in Computer Science, Artificial Intelligence, Mathematics, Business Analytics, or a closely related field;
- a sound theoretical background in mathematics, data mining, and machine learning;
- strong programming skills in Matlab, R or Python.
Supervisors of the PhD project will be Prof. Marc Francke and Dr Alex van de Minne (MIT).
For further information, please contact:
The appointment will be for a period of 4 years, with an intermediate evaluation after 18 months, resulting in a PhD thesis. An educational plan will be drafted that includes attendance of courses and (international) conferences. The PhD candidate is also expected to assist in undergraduate teaching.
The gross monthly salary will range from €2,222 in the first year to €2,840 in the last year. The Collective Labour Agreement (cao) for Dutch Universities is applicable.
What do we offer you?
Some of the things we have to offer:
- a unique community of Data Science researchers;
- a friendly and informal working environment;
- a high-level of interaction;
- a location in the city centre;
- an international environment (10+ nationalities in the group);
- access to high-end computing facilities (cluster with CPU 4,000+ cores and 50+ GPUs).
Since Amsterdam is a very international city where almost everybody speaks and understands English.
Applications should be made via email@example.com. Please quote vacancy number 17-585 in the subject line.
All applications should include a:
- curriculum vitae;
- list of university courses taken with grades;
- single page maximum statement of motivation and research interests.
An interview and a scientific presentation will be part of the selection process. #LI-DNP
No agencies please