Boku (BKU) – Forecast reliability

I have previously written how Boku have some work to do to meet FY 2019 forecasts. I thought I would go back and look at their history in meeting forecasts.

This article was extensively revised on 12/9/2019 to take into account Edison’s forecasts available on Research Tree.

Importance of reputation

I have observed that companies who are overly optimistic with their guidance and miss forecasts once tend to repeat this bias and miss over and over again. I also believe there are good reasons why this happens.

The market will naturally discount a company with a reputation for being overly optimistic, perhaps by adjusting downwards the forecasts from non-independent research, and in any case valuing it with cheaper observed Price to Sales, PE ratio and other metrics than more “reliable” companies.

This traps the company in a difficult situation. If it switches to more conservative guidance then at first the above discount will almost certainly be applied to the new, lower, more realistic guidance resulting in an immediate share price fall. On the other hand, the benefits to investor sentiment of hitting / exceeding guidance may then take years to come through.

Sources of forecasts

Broadly speaking, forecasts come from three places:

  • Direct guidance, generally via RNS statements. Many companies will only provide this close to (or after) their period end, and figures are generally limited to revenue and perhaps EBITDA later in the cycle.
  • Indirect guidance through their house broker or commissioned research provider. Here the guidance is wrapped in regulatory cover (and the research provider might even provide useful input / calculations) and therefore real profit and EPS forecasts are provided.
  • Independent research, which could be from a research provider, a journalist, or from investors doing their own research or sharing research. This is not necessarily more accurate than guidance from the company, however it can be more timely / reactive and less biased.

The most common way that investors can view forecasts is via a “broker consensus”. These also feed into analytics such as forecast P/E ratios. The consensus takes published research from reputable research providers (the first two categories above and a subset of the third) and applies some secret sauce in terms of averaging, rejecting old forecasts etc. This broker consensus has many potential problems, including:

  • Out of date forecasts. Generally very old forecasts will be excluded from the figures, but things can change all the time. A brokers note could be issued one day and a profits warning the next, making a day-old forecast obsolete. A new brokers note could be issued implying they are not happy with their current forecasts but they are retaining them for the moment pending new information. Often you see a false forecast trend e.g. where one piece of good news looks like a trend of multiple pieces of good news as different brokers update over a period of months.
  • Added / removed forecasts. The addition of a new, more conservative broker will reduce the consensus, making it look like the prospects for the company have turned down, even though nothing has changed. When a broker ceases coverage then their old (potentially out of date) forecasts generally take several months to drop out of the average.
  • Occasionally there are mistakes, such as erroneous data entry or currency confusion.

Boku’s broker consensus

Back to Boku. According to Research Tree (who get their data from Thomson Reuters), there are currently three brokers making up the consensus: Peel Hunt (house broker), Edison (commissioned research) and Macquarie (unknown). I have collated the consensus forecast for Boku for three different year-ends across 16 snapshots, plus actual figures for two of the years. Note: Boku’s year end is 31st December and these figures are in $m:

FY 2017: Mar 2018: 27.0. Actual: 24.4
FY 2018: Mar 2018: 36.1, May: 34.7, Aug: 34.1, Sep: 35.2, Oct: 37.5, Nov 2018: 37.3, Dec: 36.6, Jan 2019: 36.7, Feb: 37.7, Mar 2019: 37.9. Actual: 35.3
FY 2019: May 2018: 43.0, Aug: 42.3, Sep: 43.4, Oct: 45.8, Nov: 45.6, Dec: 44.8, Jan 2019: 50.0, Feb: 57.4, Mar: 57.8, Apr: 56.1, May: 55.9, Jun: 56.1, Jul: 55.7, Aug: 53.0, Sep: 52.1

On this basis, FY 2017 significantly missed forecasts still in force well after the year-end. The same also happened for FY 2018, plus they also missed the original forecasts. Remember that these forecasts feed into the forward P/E ratio and other metrics.

For FY 2019 there was a significant acquisition of a fast-growing identity company with year-earlier turnover of $5,1m, and a minimum revenue target of $10.0m for 2019, which contributed to an upgrade of $12.6m (some brokers taking over a month to react). Adjusting initial forecasts, even by the more conservative $10m figure, means that they are likely to miss every forecast made, with the possible exception of the latest one.

On EPS, the consistently forecast profit for FY 2018 did not materialise at the last moment (i.e. well after the period end). This could be due to differences in adjustment methodology, but historically adjustments have been minimal and the missed revenue forecast suggests (from the outside) that this was a genuine miss.

Therefore, going by the consensus figures, you would conclude that the company has a track-record of missing forecasts and therefore deserves to be priced at a significant valuation discount to reflect the lack of credibility of the current forecasts.

Edison’s forecasts

Of the three brokers listed above, only Edison is available on Research Tree (get a 10% discount with this link) and therefore available to non-institutional investors. When collated in a similar way to the above a completely different picture emerges. FY 2017 was a small beat in revenue and adjusted EBITDA, and a large beat on normalised EPS. FY 2018 shows progressive increases in forecasts throughout the year which were then beaten further. For FY 2019 you can clearly see $10m of extra forecast revenue from the acquisition, albeit with a negative effect on profitability.

Overall, Edison’s figures have forecast revenues much more accurately than the consensus and indeed have usually been the the conservative side. As they are paid by Boku themselves to provide this research, these forecasts will closely reflect guidance they have received from Boku.

That is not to say that independent research can’t do better – I commented in July that their FY forecasts looked difficult to achieve after their H1 Update. It was only after the actual H1 results and despite an upside revenue surprise that Edison took action and trimmed their forecasts.

Conclusion

The primary conclusion is not to rely on consensus forecasts.

Regarding Boku, ideally I would create a full spreadsheet model for the company complete with my own detailed projections. I only have time to do this for the most promising investments and my largest holdings, and have not yet done so in this case.

With this caveat and considering the forecast history, I now view the current consensus of 4.8c (3.9p) Normalised EPS for FY2020 as realistic. This puts the company on a forward PE of 25x at the 100p where I recently bought. Perhaps more importantly, the broker forecasts positive free cashflow of $26m during FY 2021 which is a cashflow yield of over 8%, again at 100p.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s