Market noise is hard to ignore, unless you've got the facts (and the tools to unleash them)
A few weeks ago, US President Donald Trump tweeted that companies should not be forced to have to report quarterly earnings but should, perhaps, move to a six-monthly reporting period. There was no doubt backlash from both sides of the argument, Those in favor of the move argue that the quarterly cycle deters management teams from focusing on the company whilst their detractors argue that it’s the transparency that comes with the quarterly reporting period that has led to the US$23 trillion bull market since 2009.
An AFR article published last week goes into a good amount of detail on this topic highlighting the issues in their piece ‘The $31t argument against dismantling US earnings rules’. In fact, one of the ironies is that the plan to reduce earnings report frequency (which is meant to help companies) could actually be a hindrance. A six-month reporting cycle would provide investors with less timely information than they already get and would push earnings estimates further into the realm of ‘guesstimates’ territory. The other problem with a move like this is if there are events that effect that company in the current period, investors have to wait an extra couple of months to see what the consequences of those events will have on company earnings and performance.
Whether the regulators push for a change or keep things the same, there are still inherent problems with the way earnings have traditionally been monitored and understood. A quarterly reporting period still leaves a gap in which investors and analysts apply their best models to estimate where the company trends are heading. Additionally, company’s themselves also guide investors as to where they predict their performance is heading. This 2-pronged approach can be useful in the majority of cases but where an analyst model is wrong or where a company low-balls their earnings projection (so that they look wonderful when they magically beat estimates), this can be problematic.
One way in which we looked to tackle this problem was with a data driven and systematic approach to understanding corporate earnings. What we found was that such datasets exist and these were collected every month by the various national statistics agencies across the globe. These datasets, when combined and modeled in the right way, could be quite insightful as to the earnings trends at various sector and industry levels across multiple regions. With the knowledge that this data comes from the aggregation of pricing and volume and cost information of the very companies we were monitoring, the model, we saw, could be quite powerful. With this QMG dataset we were able to utilise our econometric modelling skills combined with powerful business intelligence software to map out corporate earnings at a sub-sector and industry level.
Because companies were reporting these figures to the statistics agencies we could trust the data and because it was coming out every month it was much more timely than waiting on analyst estimates. This model worked well in markets where statistics are published regularly but a little less useful in places where the data was less frequent. In any case, it was a way of looking at the markets through a factual lens and we felt this approach and data model should be an important part of any investors arsenal of analysis tools. I thought it was also quite innovative in how it pieced together existing information in a way that others had not seen yet could prove quite valuable in seeing what parts of a market were rising or declining (much like a hospital in which you have sick patients and those getting better).
Underneath the hood we used various pieces of business intelligence technology to analyse and enhance the datasets so that clients (internal and external) could more easily view and interact with the monthly results. From time series analysis to correlation relationships, we were able to make various contrarian predictions about company earnings. Some of these were quite memorable such as correctly calling the underperformance of the US company AutoNation’s earnings in 2017 where analysts were seeing positive y/y growth across a number of quarters yet our models correctly showed that sales would decline (see below where blue line is the data we modeled from the stats agencies for US Auto Retailers, black is AutoNation's actual earnings and red is the consensus estimates from Wall Street). There was also the decline of cheese and butter producers which led to our downgrade of Dairy Crest in the UK which stemmed from our model seeing a rise in the price of milk & cream, a key ingredient for Dairy Crest products.
Factual and timely data like what we had at QMG/Canaccord, coupled with powerful yet easy to implement tools like Qlik Sense and Microsoft Power BI enabled us to uncover insights that no one else was paying attention to. So whilst much attention is focused on market noise (such as which direction reporting frequencies will move towards), sometimes its good to take a step back and just see if there is an even better way to measure and analyse the markets. Using facts and the right tools to uncover them, you're heading in the right direction at least.