Data analysis

'Big data' is great to have, but many practical commercial and policy problems involve making the best of limited data to develop our understanding.

Businesses and policymakers can increasingly rely on a wealth of data to support their decision making. But having more and more data does not in itself lead to better choices. 

The challenge of turning limited real-world data into good decisions remains. Often the process starts with identifying the right question and making use of our prior understanding of the problem at hand. Black box models cannot substitute for knowing why conclusions are reached.

Regulatory and competition authorities are demanding in the evidence they require. Questions about market definition and market power are unavoidably empirical.

Understanding real-world data starts with extraction, collation, cleaning, transforming and checking.  Statistics and econometrics then provide powerful tools to extract robust answers from consumer data, business data and bespoke survey data.

Drawing useful conclusions from limited data may mean making assumptions and using theoretical models tailored to the task at hand. Complex business decisions can benefit from explicit modelling of the available choices and their unfolding consequences to understand trade-offs and risks.

DotEcon has undertaken extensive data analysis and modelling both as part of regulatory and competition proceedings and to assist critical business decision-making.  We are experienced in dealing with complex data sets and in gathering data from special-purpose market research.  We can build sophisticated decision support models.

Shopping for ways to pay

Complaints by merchants that they are unable to refuse card payments despite the higher cost they have to incur as a result of interchange fees paid to the bank of the cardholder have been at the heart of proceedings against payment card schemes. However, in many countries card acceptance is far from universal and merchants can (and in many cases do) try to steer their customers choice of payment method. How effective and how sustainable such discouragement practices are  depends on the response from customers.

To address this question, we designed and commissioned a large conjoint survey in which respondents were confronted with a series of hypothetical choices involving the choice of shope and the selection of payment method across a range of different retail environments. We then analysed the responses using a nested logit model that allowed for different degrees of substitutability between different forms of payment and between payment and shop choice.

We found evidence to suggest that customers are more likely to switch between payment methods than to go to another shop. Discouragement practices should therefore be very effective in changing the payment behaviour of customers while having little impact on the volume of business. We validated our model results against data on card use collected by a payments industry body. The  results were used as evidence in competition proceedings in the UK and in Europe.

Benchmarking the price of radio spectrum

What price resources such as radio spectrum might fetch in an auction is an important question for both regulators wanting to set realistic reserve prices and bidders wanting to get an idea of the the ‘going rate’ for the frequencies for which they plan to bid.

Benchmarking is one way of establishing the likely value of such resources, but care is needed to make sure that the right comparators are chosen. Various characteristics such as additional ongoing spectrum fees that have to be paid, obligations linked with the licences that may be costly and the level of competition will influence the prices established in other auctions. A reliable benchmarking exercise will need to correct for differences in these factors or be based on a selection of reasonably similar awards.

We maintain an internal database of spectrum allocations around the world, including a wide range of bands and award types together with other indicators to support our benchmarking analysis. Our dataset includes hundreds of thousands of licences and tracks information such as auction format, number of bidders, coverage obligations, additional spectrum fees etc. where available. This enables us to provide our clients with the most accurate and tailored benchmarks possible.