Alternative data is big business. An analysis by Alternativedata.org found that investment firms spent US$373 million in 2017 on such data, double what they spent in 2016. The figure is expected to reach US$1 billion by 2020.
The catch-all term ‘alternative data’ covers a huge range of information from satellite images to consumer feedback collected by web-scraping. Those paying for such data clearly believe they have real value. Satellite images of a retail carpark may give you a valuable insight into trading conditions. Collating consumer opinion from the web may provide a heads-up on the success of a new product launch or perhaps a looming product recall.
Some may be sceptical about the value of such information, but clearly some market participants have no doubts, as the substantial increase in spending on such data indicates.
Alternative data are inherently different from the traditional sources of data such as an issuer’s financial and trading statements in their content and source. Analysis of alternative data provides firms with the opportunity to know things about a company that others in the market don’t know. It may provide information that even the company itself does not know.
This has led to questions. Is trading on the basis of alternative data “unfair” and a risk to market integrity? Or is it just the natural evolution of good old-fashioned research?
Traditional vs Alternative Data
Traditional data sources differ from alternative data in two key ways – their source and their accessibility.
Traditional data, for example financial results, come from inside the issuer and are made available by the issuer making a disclosure through a regulatory information service. Typically, this information is available to anyone through a simple internet search although in some cases (such as investment news or analyst forecasts) subscription fees may apply. This process, including ensuring a wide availability of the information, is a regulatory obligation on the issuer.
In contrast, alternative data obtained from various sources outside the issuing company are typically gathered by intermediaries, which are not connected to the issuer in any way, and which then sell on the repackaged data to others, including financial services companies.
A lot of this information is what is known as ‘exhaust data’, that is to say it is a by-product of some other business process, for example credit card records from retail activity. The intermediaries process the data and provide them in a form that is digestible to the financial firm.
Whilst such data are often “scrubbed” of personal information in line with General Data Protection Regulation or similar legislation in other jurisdictions, the end user obtaining the information via an intermediary may have difficulty confirming that the data were gathered without infringing rights and obligations. For instance, was explicit consent given by the individual for the data to be processed in this way? Were any duties of confidentiality breached?
Often the data are provided only to those who pay. There is a strong commercial motivation to maximise the value of the information. That may be best achieved by restricting access to a narrow audience.
But isn’t this just ‘good research’?
There are some schools of thought that would say there is no fundamental difference between the recent advances in alternative data and traditional research and analysis. After all, the concept of taking input data, analysing it to determine an insight that others do not have and then profiting from that insight, are the same. But this, others say, ignores that these data and their analysis require access and resources, which mean it is not publicly available to an ordinary person or even to all financial services firms.
As we have noted above, the access to the data is potentially different – much key traditional data are freely publicly available whereas access to alternative data may be restricted or very expensive.
The skill set to undertake analysis of the data can be different – alternative data can require the analyst to process massive data sets. This requires resources and skills not typically available to the retail investor or even some financial firms.
Taken together this means while traditional research is potentially available to a large number of market participants and their clients, the gathering and analysis of alternative data is mainly the preserve of a small and specialised group of firms whose product is sold only to an equally small clientele of market participants who can afford to pay for this data.
This results in information asymmetry in the market, with some market participants having an informational advantage over others. Though some people have chosen to pay for the data whilst others have not, but, if they could afford to, potentially could have.
Is information asymmetry new? Is it a problem?
A degree of information asymmetry between market participants is not new. This is a result of different firms/individuals having different resources and skills available to undertake research and analysis.
Where good research is undertaken, there may be a short-term information advantage which is profitable to the researcher. This allows the firm to recoup its research costs and make a profit. This is a key incentive for all firms to pursue innovative research approaches.
What is more, the value of that innovative research is not entirely restricted to the innovator. The superior insight generated by the research leads to trading activity and so feeds into the process of price formation. The market value should, in theory, then adjust.
In other words, sound trading decisions, including those based on alternative data, should lead to more accurate pricing of securities. That is the theory, though opinions differ on how this plays out in practice and whether this public good significantly balances the ‘unfair’ advantage gained by the alternative data trader.
Does alternative data add to market efficiency?
Two recent papers have been published in the US looking at the impact of alternative data on trading in financial markets, particularly in relation to the price formation process. One argues that alternative data have not facilitated better price formation, but have increased information asymmetry. The other argues that alternative data have indeed increased price efficiency and decreased personal trades by managers within companies about which alternative data are available as information is reflected in prices sooner.
The first paper is a study by Katona et al. analysing satellite imagery data and concluding that the data provide market participants with an information advantage and increase information asymmetry, but do not facilitate price formation. Although they also note that alternative data have decreased information asymmetry between firm insiders and outsiders.
The second paper by Zhu looks at two alternative data sources, consumer browsing data and car counts in parking lots of retailers from satellite imagery. Zhu finds that there is improvement in long-run price efficiency using alternative data. She also finds that alternative data reduce personal trades by managers within companies where alternative data on the company’s future earnings are available. The paper hypothesises that this is because prices are able to reflect future earnings more effectively.
While the two papers have opposing views on whether alternative data drive better price efficiency, both find evidence that the value of inside information has decreased as a result of alternative data. In a way, this means that the information asymmetry between insiders and outsiders has been transferred to an information asymmetry between market participants – those with access to alternative data and those without different access.
What are the potential market integrity risks and is there a role for regulation?
Whether or not there are benefits to price formation from the growth in alterative data, there is still widespread concern about its potential effects from a market integrity perspective. There are, of course, other risks arising from using alternative data.
Some fear the market may bifurcate into persistent winners and losers. On the winning side would be those who can access and have the ability to process the alternative data and the losers would be those who lack access and skills.
Is this a valid fear? Or will market participants come to accept that, as with traditional research, there is always an informational asymmetry? Is research using alternative data really so different?
In the context of market integrity, the essential characteristic of insider dealing consists of an unfair advantage being obtained from inside information to the detriment of third parties who are unaware of such information. This is a clear line for the regime but is there a shared view between the market and regulators on what constitutes an “unfair advantage” given any informational asymmetry can provide some sort of advantage? An insider trading on information obtained by virtue of their role in the issuer is an example of an unfair advantage. But are we all agreed on what constitutes a “unfair advantage” in other circumstances which generate an informational asymmetry?
Is it appropriate to draw a line between what research can and can’t be done or what research must be made public, or given a minimum distribution? On the face of it, that would amount to a disincentive to innovation or investment in research.
But even this, some argue, is likely to be a temporary state of affairs that normal market processes will solve. What we are witnessing, so this argument goes, is simply the first-mover advantage often seen in the early stages of any innovation. It is natural and in fact allows the cost of investment in that innovation to be recouped. However, over time, as seen with other new research techniques, the innovation and technology trickles down to the rest of the market and the information gap narrows. Therefore, is the answer simply to wait for the equilibrium to be reset or is action required now to avoid a longer-term bifurcation?
At present, there is no consensus on the questions described in this article, except perhaps, that we need more data and more analysis and a debate including industry and regulators.