News
Are consensus forecasts and transparent reporting at odds?
An IR Society Survey into producing consensus forecasts
‘Great Company plc reported a 5% rise in pre-tax profits, disappointing the market and the shares fell 7% on the day.' How often have we read this headline and how galling is it when we know that the results actually came in bang in line with what the analysts were expecting, but the press had got the wrong figure! Even more importantly, with best practice transparent reporting, encouraging the inclusion in market updates (results, IMS, etc.) of some commentary about future expectations in relation to consensus, how should companies define and substantiate that consensus measure?
A hot topic
It is clear from the level of questioning we all receive that consensus forecasts and how to deal with them is one of the IR hot topics of the moment. In pursuit of some proper evidence of how top listed UK corporates have chosen to deal with this issue, the IRS Policy Committee asked 25 companies to respond anonymously about how they both collated and used consensus numbers. A response rate of 23 out of 25 shows the huge level of interest in this area.
Why worry about consensus at all?
Before reviewing the results, it is worth considering why companies and investors focus on consensus at all and who in fact uses these figures. The simple method of combining expectations arithmetically to form a consensus picture has been driven by two key factors - it provides a clean and transparent way for the whole market to gauge future expectations of financial performance, which arguably enables better share price formation and a more accurate reflection of a company's prospects in its share price; and secondly, it enables the company itself to monitor its own internal expectations of performance against the market's view and if necessary take action under its continuing obligations under the Listing Rules, to keep the market informed of any price sensitive information which may cause a dichotomy in expectations.
As a result it is common practice amongst at least the FTSE 350 companies to come to a view on what consensus analyst forecasts are and this is seen as a key role for the investor relations department.
How do issuers derive that consensus?
Our survey suggests that the most popular current method of collating these numbers seems to be from two main routes. Either internal compilation of a set of key figures from published (usually) analyst figures, or a similar process, but outsourced. Or secondly, companies send out a spreadsheet of those numbers they are interested in, for the analyst to fill in and return. A new web based model is also developing where companies set the spreadsheet, but the analysts input their figures and get back a consensus figure and range in return. This has the advantage of being fully auditable and significantly improves the response rate, although care needs to be taken not to lose the important regular contact with the analyst community that comes from compiling the data yourself.
In conjunction with the methods above, many companies also review third party aggregators' consensus figures in order to assess any major inconsistencies or anomalies. However, there is general widespread dissatisfaction with many of the third party numbers which are inherently reliant on published notes, for the following reasons: analyst figures are sometimes not sufficiently rapidly updated; broker internal models mean key figures can be differently derived and so inconsistent (this is a problem for any consensus collecting); old forecasts are left in consensus, which do not reflect announced new data; and keying errors sometimes result in wrong data being included.
Which numbers are the focus?
The set of numbers which companies use to review consensus forecasts are generally based around key P&L figures (revenue, EBITDA, PTP, EPS), but there are in addition clear sector derived relevant figures (eg Adjusted EBITDA in the gaming sector). Most companies also express interest in knowing analysts' forecasts of net debt and leverage. Most of our members surveyed were clear that where they could meaningfully ask, more detail was often sought, but the additional levels of complexity in marrying up definitions and exceptionals, etc. could lead to an exponential rise in time committed to the process, so a balance was generally found.
Additionally, the most common practice was to collect these numbers quarterly, around results or IMS reporting, although there were a minority who collected them more frequently. Time constraints again came in as a key factor here.
Who is included in the consensus?
Every effort was made by most companies to contact all the analysts who were published followers of their company. However, getting responses to requests for information was considered to be the most time consuming part of the process and after some pursing, if no response was forthcoming, those analysts were often omitted, rather than including the distortion of significantly out of date figures.
How is the consensus calculated?
One of the problems in calculating the consensus is how to treat ‘outliers'. These are not such a problem if the population of forecasts is reasonably large (say 10 or more). But if there are only a small number of forecasts (say less than 5) one outlier could artificially skew the arithmetic average, so some companies use the median for their consensus as a more representative number.
Who sees the resulting consensus numbers?
This is the area where there are perhaps the most interesting trends developing and where change is afoot. Historically, many companies kept any collected figures for internal use only or for sharing with either all analysts or with those who asked. As transparent reporting has risen in significance, so the validity of keeping these numbers inward facing has been questioned. In fact, as the vast majority of these figures are now taken from published research, there appears to be a growing view that these are therefore in effect public data, so could and should be widely shared.
Of the companies we spoke to, nearly half now publish this consensus data on their web sites or via a fairly widespread email. This is a significant increase over the past year. Although few included any external verification of their numbers, the numbers were all generally posted behind a disclaimer, making it clear that the figures are not in any way endorsed by the company ie are not guidance.
What can go wrong?
Some lessons to learn from the experience of those we surveyed made clear that with the slowest part of the process being the collection of the numbers, leaving it to the last minute is not helpful. This risks figures being either excluded or wrong and certainly increases the possibility of a lack of accuracy.
Another pitfall can be relying on analysts' models rather than their published notes. The latter are easily identifiable as the numbers in the market - using idiosyncratic and individual models can significantly increase misunderstandings or differing definitions. Equally, an emphasis on published numbers makes the resulting consensus more easily verifiable.
Lastly, it's important to include all the analyst followers who can possibly be contacted. A consensus is by definition a compilation of all the individual numbers.
What makes the process work well?
It's a given in a regulated market, that a proper "audit" trail should be kept of how published numbers are derived and consensus forecasts are no exception to that.
It was also a clear view from our survey, that companies which spell out both how many analysts have contributed to the numbers as well as the date the figures were compiled, are providing the most useful information to their financial audience.
Ensuring your company has a stated disclosure policy is an additional part of the process that makes dealing with published figures run smoothly.
And finally, using third party aggregator figures as a reference point in addition to in-house collated numbers is a good sense check and can only help improve the understanding of how the market is seeing your figures.
The use of consensus forecasts is a developing area. We would love to hear more views on how you address this currently at your company and what you think works well or badly in the process?
Please email consensus@irs.org.uk with your thoughts.
Emma Burdett
Ross Hawley
Mark Hynes
IR Society Policy Committee
Published: 14 March, 2011