"My observation is that acquisitions have the best chance of creating value where the business purchased as nearly as possible meets the following criteria: it is paid for by internally generated cash, it is a replica of one already owned by the purchaser, it is a bolt-on or quasi-bolt-on, and it is likely to improve the quality as well as the quantity of earnings. This is effectively the route we have followed in Halma." - David Barber, Halma Cofounder and Former CEO, 1998
Halma is a UK-listed serial acquirer of highly-specialized industrial and medical companies that dominate small, niche markets. In 1972, David Barber founded the company with a simple set of principles to create long-term shareholder value: use retained earnings to acquire profitable, niche businesses that management understands. Over the last 20 years, Halma has compounded FCF at 15% and the dividend per share has compounded 5% for over 33 years. Barber’s philosophy underpins Halma’s success and forms the foundation of the company’s strategy today.
There are many great write-ups by Scott Management here, Exploring Context here, and Demesne Investments here which explain and categorize types of serial acquirers. This note builds on these write-ups so if you haven't already, we'd advise reading each piece before this Weekly Analysis.
In December, we interviewed three former executives of Halma operating companies to explore a question that we believe is crucial for ‘platform-type’ roll-ups: how is the company organised to scale acquisitions.
Halma started as an ‘accumulator’ serial acquirer. Operating companies were decentralised with CEOs running individual P&L’s reporting to Barber. This structure is similar to Constellation Software (CSU) in VMS or Addtech in B2B distribution where the parent company aims to add value by sharing best practices across opcos rather than centralising services to cut costs.
This strategy worked well until the early 2000s when Halma’s revenue and FCF hit a wall. The business didn’t grow for 4 years. In 2005, Andrew Williams became CEO, sold off underperforming businesses and moved into the medical equipment business. Prior to 1997, over 70% of Halma’s acquisitions were small, niche businesses within existing health and safety sectors. We would argue that Williams started Halma’s transition from an ‘accumulator’ to a ‘platform’ acquirer. In other words, from a Constellation Software-like operating model closer towards a Danaher-like operating model.
In 2015, Halma deepened this transition by reorganising into four ‘Sectors’; Process Safety, Infrastructure Safety, Medical, and Environmental and Analysis.
Halma introduced ‘Sector Teams’ which added a layer of middle management between opcos and Sector CEO’s: the opco CEO would report to a Divisional CEO, who reports to a Sector CEO, who then reports to Andrew Williams, Halma Group CEO. The Divisional Sector CEO (DCE) would also have a CFO and a VP of M&A to help run the sub-sector. Halma effectively has two management layers above the operating company: the Divisional Sector CEO and the Sector CEO.
In some sub-sectors like trapped interlocking keys, Halma merged opcos and restructured the individual CEO’s to regional President roles:
"We did several things which we had spoken about for years. Halma did a nice job, perhaps to a fault, of letting companies run in a very decentralized environment. KIRK Key, Castell, Fortress and several others, all owned by Halma, were competing in the marketplace, but we were all still run as separate companies. During that time, we started to focus on how to create a trapped key interlock solution for North America. We created a North American and a European company to drive these different brands into the marketplace."
This means there is no CEO at the opco level with full P&L responsibility in Halma’s trapped key business. Typically, decentralised acquirers push as much responsibility down to those at the opco who are dealing with customers on a daily basis. In the long run, we believe Halma's reorg may create perverse incentives by reducing the agency of those running the operating company. However, it can generate significant revenue opportunities and was a key driver in Halma’s organic growth post-2015:
"We are in the switchgear and UPS marketplaces and needed to develop a product line in the machine guarding and logistics safety marketplaces. Prior to that consolidation, I would look at sister companies like Fortress who have a great product that sells well in this marketplace, and make something similar to them, to a point where we were incentivized to literally copy that product while utilizing our own cost to do that. Once you consolidate the businesses and drive that brand portfolio onto the marketplace, you are able to utilize some of the Fortress products and brand them as KIRK without reinventing the wheel and to drive growth that way. That was a big advantage to change the business structure to better manage the brands, instead of only a single business unit into the marketplace."
As with most serial acquirers, it's difficult to truly grasp how Halma organises each opco. Some are merged, some left standalone. Each sector is also different. For example, BEA Sensors, the second largest opco at Halma, was effectively unchanged in the reorg because it’s the only sensor business within the group and is large enough to operate alone.
Halma’s operating model resembles Danaher more than CSU in that it's platform-driven rather than truly decentralised. Opcos still have their own boards with the regional Presidents and the Divisional Sector CEO. We question whether the added layer of middle management risks distorting the incentives and reducing entrepreneurism at the opco level.
This can also lead to problems when dealing with more specialized opcos like Halma owns. For example, Crowcon, Halma’s gas detection business and one of the oldest in the group, lost 10 years of growth in the late 90s due to low-end disruption from an overseas competitor. We interviewed a Former Director at Crowcon who believes the biggest problem was the lack of gas detection expertise within Crowcon and higher up at Halma’s Divisional Sector level:
"Crowcon went wrong at that time by not having the market knowledge or expertise in their industry. The situation they found themselves in is one where whatever decision they make has to be right, because if they get it wrong, it would be disastrous. It is hard to understand the right decision. Many times, you end up with an inertia where people would rather do nothing than be wrong. That goes against the Halma entrepreneurial concept, but that is what happened to Crowcon…When I joined Crowcon for the second time, I was shocked at how little knowledge there was of the market we were operating in. That leads to the organization doing things any organization entering a new area would do, that is look at the markets to see what is going on. We did a lot of work around who used gas detection and why they used it and which were the most profitable areas for us; business development stuff."
It seems like Halma has regional Presidents running single opcos (BEA Sensors) or multiple product lines across multiple opcos (trapped interlocking keys). These individuals have deep expertise in the end market but it’s unclear exactly how much agency they have to run their business unit.
It’s interesting to compare Halma’s structure to Danaher, arguably the best performing "platform" serial acquirer. Danaher acquires larger, leading assets that form the beginning of a platform for bolt-on acquisitions in a new end market.
As platform acquirers scale, they typically choose to buy fewer but bigger businesses rather than increase the volume of small acquisitions. It's a much quicker and easier way to scale but typically leads ROIC to revert to the mean. Today, Danaher is 14x the size of Halma with under half the opcos. If Halma follows in Danaher's footsteps, the larger the company becomes, the more Williams' capital allocation skill matters to conduct fewer, large acquisitions.
There is also one big difference between the two: the Danaher Business System (DBS).