By David Renn, Principal Consultant, Citisoft
The size of the data spend in financial services is astounding, but even more incredible is the lack of transparency and knowledge around these costs. Citisoft believes that the true cost of ownership of data is often unknown, inaccurate or poorly calculated. At a time when regulatory, client due diligence and P&L considerations are exerting increasing pressure upon the data management function, Investment management firms need a structured method for auditing their data costs and calculating their full cost of data ownership.
According to recent research by Accenture, global broker/dealers spend more than $50 million a year maintaining reference data, with the largest spending $75 million plus. Asset managers’ data spend can run into tens of millions annually, yet these figures only refer to the direct cost of purchasing data.
According to the Vizulate Reporting survey of 42 European asset managers at SUPRA 2011, 52% of respondents said that their most pressing system issue was concerned with data. Quality and timeliness of internally and externally supplied data was identified as a key challenge.
Meanwhile, a SimCorp survey that polled nearly 100 executives from 50 buy-side firms across North America revealed that over 40% are not confident that the data they are receiving from disparate systems is consistent and of high quality. Released in January, the poll also revealed that 67% of respondents believe that there is significant effort (and hence cost) involved in reconciling data between disparate systems and sources.
The evidence is clear: despite the immense sums of money being thrown at the problem, the data question won’t go away. Citisoft’s clients estimate their full cost of data at between three and five times the initial purchase cost.
This statistic alone is concerning enough, but in reality most firms have little idea of exactly how to measure the total cost of ownership (TCO) for data. It’s therefore hardly surprising then that Citisoft estimates that half of its asset manager clients see measuring their total cost of ownership of data as a current priority.
An essential input to a TCO analysis is auditing existing usage with the goal of identifying duplicate and redundant data and, where possible, reducing overall data spend. This first step requires checking invoices from data vendors, consolidators, managed service providers, custodians, brokers, and indeed any source of purchased data.
The pricing models of vendors are rarely transparent however, so a full audit should also involve checking vendor’s calculations and renegotiating where appropriate. Particular care needs to be taken that any data distributed outside of a firm is covered by licences from suppliers. This includes benchmarks, constituents and Bloomberg data.
Often, identical data is received via multiple channels (i.e., the same data being purchased both directly from say a benchmark consolidator service as well as from the benchmark provider itself). This is not necessarily bad practice as data consolidators can be cost effective as they also provide cleansing and validation of data, but it should be reviewed and monitored regularly.
TCO is a term that’s been discussed in IT circles for many years. But in relation to asset managers’ data costs, there is no clear definition of which costs should be included in this calculation. The direct cost element of TCO should not simply be the cost of data purchased from the market, there should be an element of cost apportioned from less transparent activities too.
For example, measuring how many touch points there are on data produces surprising results. If each fund manager spends 30 minutes carrying out daily manual checks of his fund’s positions because he does not trust the data, there is a very significant operational overhead which may not have been apparent when calculating data costs.
Is it any wonder some firms believe the true cost of ownership of the data they purchase from the market is so much higher than the easily metered direct license costs? Some other direct costs will also need to be apportioned into an annual TCO, such as data management project costs and data management software which may require significant capital spend in year one and then move to maintenance costs thereafter. There is also an ongoing recurring cost for data that sometimes seems to increase at the whim of the vendors.
Indirect costs – the hidden multiplier In any TCO analysis, there will also be indirect costs that should be considered for inclusion in the TCO calculation. Poor or incomplete data will have an enormous effect and will be a subsequent drag on a firm’s profitability – but how this cost is measured is open to interpretation.
Such costs include:
- Increased risk exposure to counterparties – risks need to be constantly managed and the more manual the process, the more costly management becomes and the more error prone the process.
- Inefficient trade management – errors and corrections need to be processed and closed out, potentially at a trading loss; trading errors can also result in fines from regulators and may also pose reputational risks.
- Extraordinary staffing costs – for example having to re-run a data feed at 3am due to a system failure incurs additional costs.
Some of the costs that result from data errors might be accounted for outside of a data cost of ownership analysis, but ideally at least part of these costs should be included within the calculation.
Such indirect costs will always be hard to measure and may need to be estimated based on, say, a three-year average. It is imperative that they are factored in wherever possible, even though there may be internal political aversion to highlighting such costs.
Data is at the centre of everything that asset managers do. Timely, consistent and good quality data drives investment edge and alleviates operational and regulatory risk. In these uncertain times where cost reduction is equally as important as revenue generation, it seems a logical step for firms to take a closer, more structured look at the total cost of ownership of their data.