Since MiFID II’s research unbundling rules came into force in January 2018, with the regulatory intention of eliminating the sellside’s practice of offering investment research as inducements to utilise their execution services by mandating that explicit fees are charged for research, the ‘value’ of research has become a pressing question. Rather than passing these fees onto their clients, most European asset managers have chosen to absorb this cost. This cost is drawing increasing scrutiny from budget holders looking to cut costs in leaner times. To defend their research budget, managers must determine the value derived from the research they consume. However, this is fraught with difficulties and many asset managers are looking to technology providers to help ascertain its real value.
By: GreySpark Senior Specialist Jennie Brotherston

Asset managers utilise third-party research services from a variety of sources, however, the vast majority comes from sellside brokers, typically, from whom they also purchase execution services. Research services can include not only written material, phone calls, emails and meetings, but also other things which do not meet the technical definition of research, such as corporate access and market data. Under MiFID II, brokers providing both research and execution services are required to unbundle charges for research from charges for execution, where previously research fees were implicitly included within dealing commissions, spreads and execution costs. These explicit research costs should reflect the value of research services in a realistic way and substantive research can no longer be provided free of charge. MiFID II also introduced a stringent set of rules, including budgeting, reporting and maintenance of separate client accounts, known as research payment accounts (RPAs), for firms who opt to pay for research using client money, rather than paying directly from their own resources. As a consequence, the large majority of European asset managers opted to pay for research services directly from their own revenues, rather than incur the cost, time and regulatory headache of complying with the new client money rules.
Now that it is explicitly expressed, research spend is coming under increasing management scrutiny. Studies show that buyside firms have reduced their research budgets by as much as 20-30% on average since the reforms came into force. Indeed, FCA data suggests that £180m was saved on research spend by the buyside in the UK alone in the first year of MiFID II. Faced with a fresh assault on their bottom line by the COVID-19 economic downturn in 2020, asset managers are re-focussing on the correlation between their research spend and its intrinsic value to their business
Valuing Research Interactions
The value of research is in the insights it brings to investment professionals, the principle users of research services. In a real-world context, there is often no obvious causality between any piece of research consumed from the universe of available research and the performance of an investment professional. Quantifying the value that can be attributed to any one research provider is further confounded by the additional difficulties of assessing the quality and quantity of research consumed by each investment professional.
The process by which asset managers purchase, evaluate and pay for research services currently differs in the detail but is similar in the broad brushstrokes (see Figure 1). Research budgets are set by management at firm level and are also sometimes broken down by department, desk or investment professional. Given the budget, research services from third-party providers are consumed by investment professionals (on a solicited and unsolicited basis). Data on the type of services consumed, as well as the number of interactions, are collated by each research provider, and are often also captured by the asset manager using some kind of interaction capture software which, typically, also provides some functionality to try to normalise the data across different research providers.
Typically, evaluation of this data is done retrospectively and periodically by investment professionals – individuals or teams – when they attempt to validate or assign an attribution value to each research provider based on the normalised data. A weighting is sometimes applied to the output of this process and the result used to determine how much each of the research providers should be paid for the period.
All of this information is brought together to assign a monetary value to each research provider, using some or all of the budget, which is then paid for the preceding period
Quantifying the Number & Type of Research Interactions
Capturing interactions and classifying them for analysis is not a straightforward process to automate completely due to the various channels used to consume the research, and so asset managers rely on the research providers themselves initially to generate this data. Since the classification of research is an important indicator of value — and hence the size of the fee commanded – it is not unreasonable to expect that inherent bias is introduced both in terms of the quantity and how each interaction is classified. This bias is not easily extricated from the dataset because although investment professionals may attempt to validate the data, this is usually more of a glance than a thorough review due to the pressure on their time and the volume of data involved.
Assessing the Quality of Research Interactions
While validation or other forms of attribution (broker vote) provide a degree of quality evaluation, they rely on investment professionals making value judgements about research and interactions, often en masse and weeks or months after the events they are evaluating, no doubt challenging even the most reliable memory. Studies have shown that investment professionals tend to favour more recent interactions, as well as the more familiar or friendly providers. To confound any realistic view further, the more simplistic processes often use evaluations from previous periods as a starting point, which further exacerbates biases towards certain providers, regardless of the quality of the interactions. Many also rely on a simple ‘rate card’ structure (paying a given amount for each interaction type) to assign suggested monetary value to interactions as a starting point, which does not encourage objective qualitative evaluation.
It is easy to see how these issues can lead to an unreliable evaluation. While investment professionals naturally understand the importance of a good evaluation process, they often lack the time and tools to undertake comprehensive, timely and accurate analyses.
Solving the Research Evaluation Conundrum
There is a clearly a need to better capture and cleanse interaction data supplied by research providers. To eliminate any bias from the data on its value, research providers should not be completely relied on for interaction data. While there are existing software platforms that asset managers can use to attempt to address data capture and normalisation into uniform categories, most either do so using fairly simplistic mappings or they require a high degree of input from the research consumers.
Some platforms also seek to develop and improve the vote process with, for example, better inputs, more data analysis displayed to users and friendlier interfaces, but it remains difficult to implement a truly effective solution that eliminates the problems described above.
The Way of the Future: Data-centric Evaluation of your Research
Increasingly, other areas of the financial services industry are beginning to utilise behavioural analysis with a high degree of success to draw useful information from large raw data sets. The problem of quantifying and evaluating research interactions is ripe for behavioural analysis techniques. Correlation analysis between these cleaned metrics may extrapolate the benefit by enabling a more complex analysis that could provide a more complete evaluation.
All of this could be done with no intervention from investment professionals – analysis could be conducted using just the raw interaction data – which would present time and effort savings in addition to reducing or eliminating many of the problems and biases inherent in so many existing solutions.
As such, vendors like Quintain Analytics stand apart from their competitors with their ability to apply smart behavioural analytics to unstructured raw interaction data sets and return useful, informative results, which will ultimately save both time and money for asset managers, allowing investment professionals to focus on maximising returns, whilst at the same time ensuring that research spend is fully optimised. GreySpark believes that this kind of smart analysis can draw new insights from research interaction data and is an example of the next generation of research evaluation provision.
Harnessing Data to Power Performance
We provide award winning advanced data analytics packages & software to help investment management firms (asset managers, hedge funds) navigate through the massive amount of unstructured interaction data from their research providers, allowing them to make more informed decision on research spend.
Our behavioural factor analysis works to identify where the true value of research provider relationships are and when overlayed with budgets, can quickly identify if potential over or under payments are being made, helping manage research costs in a dynamic manner. We utilise the data available to you from a variety of sources, either your own data or from other external providers, allowing us to return value specific to your investment process.
Whether its analysis for cost management and broker relationships purposes or benchmarking interaction trends against peer groups, Quintain Analytics have data packages and software solutions to help solve pain points within the investment management community.
For more information visit: quintainanalytics.com
or email info@quintainanalytics.com