Lazard Asset Management Looks to Reinvent Asset Classifications

“Why do we have asset classes? It seems like a complete waste of time.”

This was a question put to Jai Jacob, managing director and portfolio manager at Lazard Asset Management, by one of the data scientists in his multi-asset investment team.

It was a question that made Jacob scratch his head, wondering the same thing.

And now, the asset manager is developing themed asset categories for investors by finding new correlations in alternative datasets or personalizing portfolios to fit their criteria.

Its multi-asset investment team is using techniques like natural language processing (NLP) and network theory to identify new associations between companies, looking at areas like their supply chains or location, or mentions in the same articles in the press.

To explain the firm’s thinking, let’s take the US dollar as an example.

If an analyst forecast that the US dollar would strengthen, typically, the asset allocation response would be to increase exposure to US small caps, as they would expect to benefit in this scenario.

But, Jacob says, this is an “imperfect mapping exercise”.

“It’s making two assumptions and bundling them into that one investment decision, whereas if I had an asset class called the ‘US dollar is going to strengthen’, I can rely purely on that investment insight …and whatever happens to be exposed in a positive way to the US dollar, will find its way into that asset class,” he says.

The idea is that the multi-asset management team can create themed portfolios on demand to more accurately fit the investment decisions of its investors and their changing appetites. This can include re-adjusting asset allocations due to gaps in exposure or recalibrated risk in parts of a portfolio.

“So, it’s not that we have new asset categories that we’re going to stick with for the next 20 years,” Jacob says. “It’s saying we need the ability to create asset classes on the fly, to be able to better implement investment views.”

However, getting to this point took some time, he admits.

The Lightbulb Moment

At the end of 2018, Lazard AM consolidated its portfolio management team and workflow onto a single multi-asset platform, which became its Dynamic Portfolio Solutions (DPS), bringing together both quantitative and fundamental investing, and data science.

This was the by-product of years of planning and discussions, but it wasn’t until 2014 when Jacob and Paul Moghtader, who headed up the quantitative equity business in Boston at the time, realized the growing importance of data science for developing bespoke portfolios.

“We felt that we needed to start bringing on a different kind of investment professional,” Jacob says. “It was not without its growing pains, but the basic idea was that the concepts that are embedded in income statements and balance sheets, those have been picked over for decades and we needed to figure out how to move beyond that with all the other information that is out there in the world.”

The problem was, they needed specialists.

So, they hired a team of data scientists, pulling in talent from various universities and sectors, including health and technology.

But that wasn’t enough.

The next step was to offer these data scientists a career path, and one that validated their importance in the investment management process. As opposed to throwing them in a room with terabytes of credit card data and waiting for the magic to happen, Jacob says.

Instead, they merged the quant team, portfolio managers and data scientists, bringing them under one roof. Jacob says that one of the main reasons the firm did this was to elevate the relevance of data science in the investment process to the same level as other roles.

Today the multi-asset group is made up of 32 people that work across the DPS platform.

“We really rethought the investment talent within the platform and said, rather than putting people in these investment team boxes, let’s try to orient and arrange the talent that we have at our disposal by discipline,” Jacob says.

Generations Asset Class

Generations is one of the progressive portfolio themes Lazard AM is working on today. The concept looks at trends such as the number of people leaving and entering the workforce, consumption habits, saving patterns, and purchasing power. This will also incorporate alternative data from news articles, social media platforms, and other forms.

The effort combines both fundamental and data science skills, where they can isolate these patterns of different cohorts to develop an asset classification.

“The fundamental insight shows the different consumption patterns and the broad trends, and then the data scientists will help map those ideas into specific companies because there isn’t a particular sector or sub-industry that those ideas have to obey,” Jacob says.

One example is the amount of money a particular age group spends on healthcare and the changes in life expectancy over the years, as more people tend to live longer.

Another popular trend looks at the way millennials or people just entering the workforce tend to accumulate fewer physical items and spend more of their disposable income on experiences, which comes in the form of events or travel.

However, in light of the recent coronavirus pandemic, Jacob notes that the latest results would show a different picture compared to a few months ago.  

And there is no predicting when those patterns will return to normal.