01 Dec 2017
As automation extends its range of applications and productivity, more companies are profiting from using it. So where is best to invest? Simon Edelsten, manager of Artemis Global Select Fund, discusses …
In managing the Artemis Global Select Fund and Mid Wynd International Investment Trust, we are always looking for less recognised investment opportunities. We have been meeting and investing in companies in the automation sector for some years and have made good money, especially in automated warehouse companies benefitting from growing demand from e-commerce. Following this through to the components of an automated warehouse, we started meeting robotics companies, fork-lift makers, factory automation businesses and now artificial intelligence (AI) companies.
Talking to the large robot-making companies, it became clear that many industrial processes were already heavily automated and that robot prices were generally deflationary. Heavy engineering processes such as steel-making and automotive production have a fraction of the labour they employed a few decades ago.
Machines have, of course, been replacing labour since the Spinning Jenny in 1764 - and working people have feared losing their jobs to machines: the Luddites smashed these early machines. Over time, people find new jobs and, with automation making them more productive, these jobs are generally better paid.
So what has changed to make us want to invest more in automation now? There are three new dynamics. As the price of robots has fallen and their quality improved, they become cost- efficient for more tasks in the manufacturing line. There is an oft-cited Boston Consulting report which suggests that the fully amortised cost of a spot-welding robot fell below the average wage of a Chinese spot-welder last year. In the real world, it does appear that China, in particular, is suddenly buying a lot more robots for their new automotive, semiconductor and smartphone assembly plants, perhaps relying less on labour whose wages have risen some way over the last few years.
Secondly, robots are becoming more sensitive. The new applications for automation often require less heavy lifting and more delicate handling. Improvements in servo-motor quality and tuition gears - which make train doors slow down as they come together - allow a robot arm to swing fast and then pick up an object delicately. The current challenges for future robots (in Japan) are: sewing T-shirts - cloth is very difficult to handle - and packing bento boxes - which need to be packed fresh overnight for each following day's rush hour. There is also a national project to come up with automated rice paddy tractors to help the ageing farmers.
Thirdly, automated factories increasingly have fully networked robots, sensors and quality controllers. This combination of automation with computing allows automated factories to learn from their mistakes and to optimise quality and flexibility. In recent weeks, announcements by Canon and Nike, that their next plants will be in their home countries after years of off-shoring, have focused mainly on quality and flexibility, combined with simple supply chains. Although the competitiveness of these automated plants with emerging market labour seems a minor issue in these decisions, if companies such as Nike and Cannon no longer continue creating high quality work in emerging markets, some of these countries may struggle to create the well-paid jobs that their youthful workforce has been looking forward to.
The emerging market ‘demographic dividend’ may be waning. A report from McKinsey suggests that automation could eliminate over half of India’s jobs. This is based on India adopting levels of automation current in the rest of the world - especially in distribution and retailing. We see automation extending its range of applications and its productivity. Like the Luddites of the 18th century, some politicians will claim they can protect workers, but technological advances do not dissipate because they are challenging. At the end of the 19th century, London used three million horses for transport. The Times predicted the city would end up with piles of dung nine feet deep. Only a fraction of the jobs that were in horse-husbandry, carriage making, wheel-wrighting have been replaced by Uber drivers - but at least the streets are navigable.
Our conclusion is to focus on the manufacturers of key components for more advanced automation and to avoid the well established heavy materials robot-makers of old. The dominant global manufacturers of servo motors are Japanese and so are the dominant makers of reduction gears. The robots that can take on the new factory tasks tend to need the more precise servo motors, gears, sensors and a higher number of arm articulations. There are thus opportunities for us to focus our investments on the value-added elements of the automation wave and avoid the more deflationary products. We have been monitoring these stocks for some time and, when order books shot up in July, we rapidly invested in a range of these companies, this theme now making up over 10% of our fund’s assets.
Visiting these companies over the last few weeks to refresh our research, it became apparent that many have been co-operating for some years. Automation is something of a Japanese speciality with all the world’s reduction gear companies based there: Chinese companies have made little headway in trying to catch up here or in advanced robotics. The shrinking, ageing population of Japan has been recognized as a challenge for years and many companies have co-operated on their automation designs since the early 2000s, forming an industrial group and ensuring that components made by different companies can work together. All very Japanese. Few foreign companies chose to join this group. This may have allowed little known companies such as Daifuku to become the world leader in warehouse automation. This is now our largest investment in this area. While Japanese companies dominate our automation theme, Kion, which was spun out of Germany's Linde in 2013, has extended its strong position as a fork lift truck maker into warehouse automation.
Another brief diversion: some are saying that they can enhance factory automation by linking components through the ‘internet of things’. This will, no doubt, allow some components to be added to existing networks, but to avoid hacking plant owners generally want to isolate their core automation from the web. The Japanese approach, born in the age of Ethernet, has built a reputation for reliability over many years. Notably, many of the companies we have identified have rather higher profitability than is common in Japanese precision engineers and have rising repeat service revenues as clients rely on them to keep the plant in good running order.
The new theme in our fund has arisen around the same time as we have started to find that other investments have moved beyond our valuation limits. As the new automation theme in the fund has grown, the portfolio’s average multiple of cashflows has fallen and the correlation of investments has also fallen, without giving up either quality of earnings or much by way of potential revenue growth. Funding some of our automation investments from profits in online services investments reduces the fund's exposure to the most popular part of this theme, while increasing our exposure to the out-of-favour, modestly rated and lowly correlated Japanese market.
As an example, automation has become well over 60% of the operating profits of Mitsubishi Electric. This company has been seen as an over-diversified conglomerate, but as its automation and its strong air-conditioning divisions start to dominate profits, the stock should see its valuation rise from very modest levels.
Managing a global fund, we try to monitor a very broad range of growth trends. At any point in time some will be more heavily favoured than others. Our valuation discipline should generally lead us to move assets from highly rated to more lowly rated sectors and markets, asset allocation resulting from our valuation approach. This is how we think we can continue to invest in quality growth stocks on sensible valuations even after the strong performance global equities have achieved over the last eight years.
As long as new investment themes become available at reasonable valuations, we should be able to continue to grow the real wealth of our investors in the years to come.
It is not clear how far factory automation overlaps with AI, but new catchphrases such as 'the learning factory' point in this direction. AI is something of a catch-all for networks which process large amounts of data and try to improve functionality by iteration. This ‘data cracking’ allows computers to spot data correlations quickly, but computers are no good at working out causal links. This allows AI to improve on human outcomes in processes swamped with data, but not in areas where data is limited but meaningful.
Readers may be interested to note that the number of data points of an old company such as JPMorgan, in the Dow since it was launched in the 1920s, is roughly 97 years times 250 closing prices a year = 24,000 data points; in this context a small sample. Deep Learning algorithms tend to sniff at data sets smaller than millions. Taking minute by minute price moves increases the data set by 8x60 minutes = 320 fold = 7.7 million, so AI may be better at finding patterns for high frequency trading.
So AI may struggle to be good at stock selection unless pre-programmed with links to many other data factors which cause share prices to move - value, momentum etc. AI may be able to drive your car, but we will be sticking to fundamental data-aided human stock selection to manage our investment process.
Simon Edelsten manages the Artemis Global Select Fund alongside Alex Illingworth and Rosanna Burcheri; visit the fund page for further information about the fund, its performance and current positioning.
THIS INFORMATION IS FOR PROFESSIONAL ADVISERS ONLY and should not be relied upon by retail investors.
The fund may have a concentrated portfolio of investments. The fund may invest in emerging markets.
Third parties (including FTSE and Morningstar) whose data may be included in this document do not accept any liability for errors or omissions. For information, visit artemisfunds.com/third-party-data.
Any research and analysis in this communication has been obtained by Artemis for its own use. Although this communication is based on sources of information that Artemis believes to be reliable, no guarantee is given as to its accuracy or completeness.
Any forward-looking statements are based on Artemis’ current expectations and projections and are subject to change without notice.
Issued by Artemis Fund Managers Ltd which is authorised and regulated by the Financial Conduct Authority.