Laurence Carr on Using Behavioral Forensics in the Investment Process

We had the pleasure of speaking with Laurence Carr, formerly Head of AKO Capital’s Behavioral Assessment Unit, with extensive experiences as Chief Detective at Merseyside Police and HMIC.

Laurence is a practitioner of research-based interview techniques adopted by UK law enforcement and has played a pivotal role in introducing these methods into a comprehensive investment analysis framework at AKO Capital. Additionally, he has supported the largest sovereign wealth fund, Norges Bank Investment Management, in shaping their behavioral assessment team.

In this exclusive interview, Laurence discusses the pitfalls and biases involved in traditional Q&A-style interviews, how these methods evolved through collaboration with academia, and useful frameworks to apply. Moreover, Laurence provides guidance on the role of Forensic Linguistics and Behavioral Assessments in law enforcement and applicability for investment analysis.

Offering insights from his time at AKO Capital, Laurence highlights how these methods have been or could be applied in investment applications, the relevance of the methods when analyzing companies with seemingly shareholder-friendly incentive structures like founder-led or owned companies, and areas where the methods have yielded success or have been ineffective. We explore the enhancement of these processes using technologies like Phonetics and AI and potential pitfalls from relying on technology-based tools.

Finally, Laurence provides advice to interested fund managers or those who have not formally adopted these techniques or processes into their framework.

Topics and themes:

• Laurence Carr’s Background
• Pitfalls of interview techniques used by Police Detectives
• Breaking down AKO Capital’s Behavior Assessment Unit (BAU)
• Assessments; assessing ‘credibility versus reliability’
• Applying a useful research-based interview framework
• The role of Forensic Linguistics and Behavior Assessments
• A case study from Laurence’s time at AKO Capital
• Relevance of BAU when investing in founder-led businesses
• False Positives/ Negatives from applying BAU tools
• Indicators used to improve BAU processes
• Effectiveness of AI or other tools in analyzing communication

This interview was conducted by MOI Global contributor Tarek Andari.

The following transcript has been edited for space and clarity.

Tarek: It’s a great pleasure to have Laurence Carr with us today as we explore how police detective frameworks and tools can be applied to investment analysis. Following approximately three decades of detective experiences at Merseyside Police and Her Majesty’s Inspectorate of Constabulary, Laurence transitioned to the multi-billion dollar fund management company, AKO Capital, where he established and led the Behavioral Assessment Unit for seven years.

Laurence has also supported the world’s largest sovereign wealth fund, Norges Bank Investment Management, and we’re very excited to have you with us today, Laurence, and dive into some of these experiences. So, thank you for joining us. It’s a pleasure to meet you as well.

Laurence: Yeah, thank you, Tarek. It’s good to be here. Happy to have the conversation.

Tarek: Thank you. Could you expand on the introduction to tell us why you chose to be a detective and how the field evolved over the course of your career?

Laurence: I joined law enforcement, as you do in the UK, and you become a generalist at the start, really. It’s a great job in as much as you get to do a lot of different things, really exciting things and interesting things. It’s not paid very well, but it’s a great job for other reasons. During the first few years, I was able to sample detective work for about a six to nine month period as a young officer. It was that really that piqued my interest. And of all the things I did in my police career, investigation for me was the most rewarding, the most interesting.

A little bit later on, before I joined the detective branch fully, I was also seconded onto a homicide investigation. Ultimately, it was that that interested me as well, being a senior investigative officer, leading a team of detectives and solving some of the homicides that took place. So that was it. That was how I ended up a detective.

Tarek: Has the field evolved, since you first started, with any new investigative techniques or technology enhancements?

Laurence: Yeah, I mean, I joined in what I call the life on Mars years. When I say that I’m referring to a TV program that was on in the UK in the noughties that received awards, BAFTA awards. It was about a young detective in the northern city, who could have been me, who joined the police in the seventies and eighties. Sorry, he joined the police in the noughties. During his duty, he got assaulted or he ended up in hospital in a coma, and the TV show is the dreams that he was having in that coma. In his dreams, he was transported back to the 1980s and late seventies, early eighties. Yet with today’s sensibilities. So, that program was called Life on Mars.

I joined the police during those life on Mars times in the early eighties when the interview process in particular was pretty biased. The detective would make up his or her mind whether the individual that they were about to speak to was guilty or innocent, and then tailor the interview accordingly. The objective was to get a confession from somebody rather than to get information from them. So, when I saw that evolve, if you look at what I did at AKO Capital ultimately, and one of the things that I was able to transfer across to my time in fund management, that was a big part of it. Equity analysts interview CEOs of the companies they’re invested in or that they’re doing due diligence on all the time.

They’re really, really clever people, the analysts, but they don’t get a lot of training or guidance on how to conduct the conversation with somebody to get the most information from them. So in the police, going back to my police time, we moved in the early nineties in the UK and then worldwide really from that really confirmation bias or bias approach to interviewing somebody and having a conversation with an individual, whether it be a witness or a suspect towards more one that was to get information from them. That happened because of a lot of miscarriages of justice that happened in the early nineties in the Guildford four, the Birmingham six and other cases, which were really bad in a number of ways. They were bad because innocent people were in prison for a long time based upon confessions that were false, and they were bad because, the offenders were out there walking around on the street doing more harm and bad because there was a lot of reputational damage to UK law enforcement that arose from that high publicity when the people appealed and won their appeals years later, and also bad because the victims didn’t get justice.

So that led to a royal commission, which led to academics becoming involved in UK policing’s approach to investigative interview or to conversation management and changed the whole thing, root and branch. I was lucky enough to be immersed in that as a practitioner in the early nineties, and then later as a strategist, as I got more and more higher up the ranks in UK policing. So that sort of led on for my region anyway and had some involvement nationally on interview techniques. That’s one thing that changed, lots of other things changed but the job itself, policing detective work never really changes: you’re investigating something. What does change is the tools that you’ve got to help you do that really, and that did change a lot in my time.

Tarek: I believe AKO Capital recommend an interview techniques book that touched on the transition from traditional Q&A to research-based interviews or enhanced cognitive interview techniques. What would, in your opinion, now that you’ve done detective work and you’ve worked in fund management, what do you think the fund management industry would be really missing if they don’t focus on adopting these enhanced interview capabilities and techniques? What would be the key takeaway?

Laurence: Well, I always say that, you know, people’s decisions are based upon information. So, all our decisions are based upon information in fund management and everywhere else in life, but really crucial in fund management. There’s a lot of risk around those decisions and you make better decisions if you’ve got better information. I don’t think anybody would argue with that premise.

As I alluded to earlier, it always sort of fascinated me how analysts and people in the industry, in fund management become or in industry are becoming, fundamental analysts, perhaps forensic accounting analysts as well, market analysts, are all good at forming models, but actually awful at asking questions, awful at managing the process of the interview from the start to the finish, to give themselves the best chance to obtain as much accurate, reliable and relevant information from their interlocutors as they possibly can. That’s the thing at AKO Capital that I sort of really helped in that regard. If you increase the amount of information you’re getting from a source, then you’re going to make better decisions ultimately at the end of the day; so, that was one big win.

Tarek: Is the big topic then ‘credibility versus reliability’? Would you phrase it in that sense?

Laurence: Well, I would say that you can’t put the cart before the horse. Going on to credibility analysis, and that’s another whole topic really in terms of, okay, so your first job is to conduct the interview and ask the questions in a manner, which is going to give you the best chance to get the best information from the individual, and it’s that information upon which you apply your judgments as to the credibility of what is being said to you. That’s not always just about lie detection. In fact, it’s mostly not just about lie detection, because lie detection is difficult. It’s also about the levels of confidence in the language that they use. It could be that, or it could be, how confident are they in what they’re actually saying.

You’ve got to apply a whole host of other judgments around that as well. For example, does the person actually know the answer to the question that you’re asking them? Is it the fact that they don’t know and they’re too embarrassed to say so? And all those judgments that you make along the way. It could be what’s the level of confidence that you can look at things like how often they use qualifiers and such like that in terms of credibility analysis. So, they were the two big things I did in fund management really, firstly, the interview training, the conversation management training, and then looking at what’s actually been said to the investment analysts and to the investment community in terms of quarterly earning calls, Q&A, etc, and how credible is it, or what the CEO, and CFO, is telling you.

Tarek: Would it be fair to split the work you do in three parts, starting with interview techniques, then linguistics is that the right term?

Laurence: Well, it’s funny. Yes, to a large degree. When I started at AKO Capital in 2012, and I started with Nicolai Tangen; he was very broad minded, and for me, a farsighted individual who employed me, because he was watching his analysts and thinking, how can we make the interviews better? Where do we go for somebody who might know about this? And he thought, oh, well, the police interview a lot of people, so he advertised in a police publication actually, and the advertisement said head of interrogation wanted for a hedge fund, and interrogation isn’t a word I would use, because it’s a sort of, it’s more of a, it’s too confrontational, that word, really but it’s part of my interest. So, what I was getting at, I started on day one with a blank piece of paper and a blank desk and just me to sort of try and prove the concept really. After about six or nine months, Nicolai said, yeah, yeah, yeah, let’s do more of this. Why don’t you employ somebody to help you?

That’s where the linguistics came in, because I sat there and thought, I don’t really want to fill this office full of retired old cops like myself. I’m not sure that’s the right thing to do. What discipline can I look to that actually sort of fits what it is we’re going to do, particularly around the language of the CEOs and CFOs and the written language predominantly, really, although we would watch webcasts and go in on occasions to live interviews as well, but mostly it was documentary analysis, and I hadn’t had knowledge of the discipline of forensic linguistics. Through my time in the police, I’d never actually used a forensic linguist in the police in my career, but I knew about the discipline, and really, that seemed to be the one that fitted best. When I left seven years later, there was a team, I think, about five of us, there was me and four forensic linguists who were carrying out the work.

Tarek: I don’t think I fully understand it. Would you say someone who’s very good at interview techniques wouldn’t necessarily need someone good at linguistics to conduct a good interview and then interpret all the information they get, relatively accurately? Or would they still need some form of linguistics training to be able to do good interviews?

Laurence: Well, I didn’t get any linguistics training. I had sort of a lot of experience over 30 years being lied to, really, by people on the streets of Liverpool (humor). So, I wouldn’t say it’s totally necessary that you’re a forensic linguist. A lot or some of it is common sense. Some of it can be learned. I learned a lot, I had those instincts, really, is what I would call them, because I’ve experienced it for a long time, as I described in my police career, but I have no formal training, and I couldn’t put a label on most of it, really, at that stage.

Over the time at AKO Capital, working with some really bright forensic linguists, we also formed some partnerships with academics from Stanford University, from Lancaster University in the UK, and ran some projects and collaborations with them. So, I became a sort of honorary forensic linguist, really, without the qualification. So, if that’s your question, if that answers your question, then great.

What I would say is you do need, as I said earlier, to be able to ask the questions correctly and in a manner, that’s going to get the information. As I said, we did two approaches, we did the documentary analysis, which was a large part of our work, really, on quarterly earnings call transcripts and we also would watch webcasts. I mean, the analysis of the earnings call transcripts was properly forensic, and by that, I mean, really, really, really detailed. By the time we would publish something to the analyst team, it would have been read once, read again and annotated, read again and summarized, quality assured by another member of the team who read it again. It was amazing what we picked up on the second and third read that we’d missed on the first read, it is really important to be detailed.

Tarek: I did have a question later in the session on internal biases for the person actually conducting the interview. And I know you touched on how the interview techniques evolved to avoid that, but it does feel like there is still some form of bias per person that you need to have sort of third-party verification, etc. So, is the third bucket then behavioral assessments, if we say interview skills, linguist skills and then a bit of behavioral assessment skills as well?

Laurence: By that, do you mean nonverbal behavior, body language?

Tarek: Body language, potentially as well. For example, if you have a CEO who is pretty aggressive with analysts asking questions then that’s a form of behavior, but also communication. Would you sum it up to say these three things are sort of under your remit?

Laurence: Yes

Tarek: The way I’d like to structure this is to focus on the interview techniques first. Are there any useful frameworks or tools that you found that worked perfectly for you over time or have you refined them to become even better? Is there anything you’d like to educate us on in that sense?

Laurence: Yeah, well, I spoke about the sort of genesis of how things developed in my police career in terms of conversation management and how academics were collaborated with in the early 90s to develop and carry out research into what had happened; just prior to that was UK policing started to take record interviews with people. So, there was lots of data available for the academics to listen to, tear apart and research.

Based upon that and other research they did, they came up with a model called the PEACE model, that’s P.E.A.C.E. What that sort of, that mnemonic stands for is Planning and Preparation, Engagement and Explanation, Account, Challenging, Close and Evaluate. That’s how the process was broken down into those five key constituent parts of a meeting really, or of an interview or of a conversation. As the name suggests, it started, the process, the model started well before the interview would take place with the planning and preparation phase. I spent a week training on that PEACE model as an early, as a detective sergeant in the early 90s. I could spend a day talking about each of the five facets. So, we’re not going to do that here today, we clearly haven’t got the time. Suffice to say, that model was ground-breaking really, and it sort of changed the whole mindset of the interview process at the time.

Now, that same model still exists and still is used today. There have been improvements along the timeline, as you can imagine, but in effect, the actual basis of the model starts with planning and preparation, and it’s how you go about best doing that. And then it goes into the engagement, how you greet the person, how you meet them, what is your first line going to be? How are you going to build rapport with them? Because it is a rapport-based model, and then the account, which is the actual eyes on, whites of their eyes’ conversation, how are you going to conduct that? What are the best things, way to phrase your questions, manage the process, manage the room, manage the people your side of the table, and manage the people the other side of the table, if you can. How are you going to close the meeting? And then what evaluation do you need to do afterwards? Now, that is still more or less the same all these years later. It’s just, as I said earlier, the nuances and maybe little bits of things that have been improved along the way, yeah.

Tarek: Two or three questions down the line I will ask for a case study, if you’re happy to share one with us, where maybe you’ve used the model just to bring it to light in terms of an investment application per se.

I’ve been following Dr. Rob Leonard from Hofstra University. He basically says, forensic linguistics maximize intelligence yield from linguistic evidence. Could you unpack that sentence with a simple example?

Laurence: Well, Rob Leonard is somebody I’ve had a couple of conversations with, and he’s a learned professor of forensic linguistics over in New York. In fact, the first person I recruited when Nicolai said recruit somebody, she’s now a PhD and works at Norges Bank, something I’m very proud of, the people who have started careers as a result of starting at AKO Capital and all the rest of it. She actually worked for Rob Leonard for a while, and the community is quite small.

So, to unpick what he said, I think it’s about, you could sum it up as reading between the lines, really, couldn’t you? I mean, what’s being said in the start words is often, there’s often other meanings within it. And of course, forensic linguists in criminal justice, sort of with that hat on, they will look at things like authorship analysis, which looks into, has the person who purports to have written this thing actually written it or typed it? Use of certain phrases on how people write and speak can tell you things about them. There are some other great academics.

In my time at AKO Capital, we did some collaboration with Stanford, a guy called Jeff Hancock, and his boss, I think at the time was Jamie Pennebaker, who wrote a great book called ‘The Secret World of Pronouns’ or something like that. It sounds really boring, but it’s not boring at all. Mr. Pennebaker, he’s a psychologist rather than a linguist, because there’s lots of psychology in the topic as well, and he’s got a great book that that tells you really what you might say about someone who speaks in a certain way or uses pronouns in a certain way, really. So yeah, fascinating, fascinating stuff.

Tarek: What about behavioral assessment? Is it something you do whilst conducting an interview or listening to an earnings call just by nature or is there more to it?

Laurence: On that topic, just to interject some of the things that we used to see in interviews, so you talk about case studies, there were more than one or two times when you’ve got a panel of, you might have an Investor Relations Manager, CFO and CEO. If I would go into a live meeting with people, I would be observing mostly.

I used to ask the odd question, but I’d be mostly observing, and it’s amazing, you can pick up the sort of glances between people, or one might nod his or her head and the other one might shake their head in answer to the same question. Or when someone’s saying yes to the interviewer, while someone alongside them is saying no with their head. That was a really important part of it as well.

Tarek: How would you exercise/ assess linguistic features like contraction patterns, discourse markers, lexical choice? Would you assess the CEO in one meeting and then assess the CFO in another meeting on relatively similar questions, and then see the patterns and what they say? Would that be an exercise, for example?

Laurence: Well, I think what was really important was baseline behavior, which sort of what you highlighted there in your question, it shows how important that is, because there are two things, I think, to your question. One is consistency, consistency in behavior, but also consistency in what they say to you, the answer itself. And people often ask me, what are the big indicators of something going on in the background, and consistency is obviously a huge one.

But also, if you’re trying to judge somebody’s behavior, you should really know as much about that, what that baseline behavior is in the first place. For example, do we always use this particular word just as part of a lexical choice? Or is it significant? Is it a move away from what the baseline behavior normally is? Yeah, so that was a big part of what we did as well. And it’s amazing, you know, it’s amazing how, even just reading transcripts of someone’s conversation, where they use certain phrases, how you can pick up baseline behavior, having never met the individual, or some parts of their baseline behavior.

Tarek: Circling back on something you said, would you then recommend investors to always have a silent observer with them in the room, given the role you were playing? Would you say that’s really important to have?

Laurence: I think it’s, well, two things. First of all, it’s quite dangerous to, or I believe that assessing body language is really dangerous. So, I think you need to have some experience in it and to be really careful before you make decisions based on something like that, really. I would also say where it’s got power is alongside the words. So, if you get the two things together, then the wrong words and the wrong body language, then that’s where the power lies.

To your question on whether to have someone present all the time, I mean it would be a nice luxury to have. I suppose it depends on who it would be more than anything, as I just alluded to, really.

As an interviewer, one of the great things or one of the great learnings for me was about proper listening. Especially in today’s world where everything’s so fast and we’ve got phones and everything else going on around us, very few people, and of course, the Zoom world and the sort, I could talk for some time on the training that I do about how to do notes. So, if you’re not recording an interview, you have to record it in some form or another. The whole process of who does that, how it’s done, whether you use a laptop, whether you use a pen and paper is a whole topic in itself, really. It’s those whole things are really fascinating.

If you’ve got your lead interviewer, they should be actively listening, and you don’t just listen with your ears, you listen with everything you’ve got, including your eyes. They should be the ones who might pick up a lot of this stuff anyway, you know, what’s being said to them, the way the person’s behaving. I would always appoint a separate note taker and not have the main interviewer taking notes. Having said that, I always have a pen and paper. If you are answering a question that I ask you, and you say something that I might want to probe you on later, that I don’t want to interrupt you because you might be going to say something really crucial next. I might write down one word to remind me what that probe is later on in the conversation but the main note taking task I give to somebody else.

So that’s two people. And there’s a way to manage it if you’re on your own as well, which we haven’t really got time to go into and still have a good conversation. So, yeah, it’s like the whole thing about numbers in the room, have you got the staff and all the rest of it play into that whether you can have someone that I wouldn’t say it’s necessary, necessarily crucial.

However, we did it, we risk managed it. So I wouldn’t be asked into all meetings with CEOs because it’s a waste of time but if we had a case that we were worried about, if it was maybe a big investment case of the long book, or if there was a new company we were looking at, or the CEO changed or there was some risk factor, really, that the equity analyst thought it might be worth having me cast an eye over things, that’s when it was done.

Tarek: Good to have the resource, I think why AKO Capital shines, I guess. Is there a key case study that you can share with us where something was either caught out early and you sort of saved yourself a lot of trouble or maybe a risk was cooking up and you were seeing it play through and eventually exited.

Laurence: Right before formally starting at AKO Capital, I was called in by Nicolai to listen in on an interview with the CEO of a multinational supermarket company. He said: “we’ve got a huge position in the long book in this company and we’re not too sure about them. We think he might have lied to us in the past”. So, I offered to listen in to the questions because they were really nervous about the position, and I was also asked to look at, in fact this fits both of your questions, to look at their interview techniques and give feedback on that.

So, I checked out the interview room; they did have a plan which I was pleased with, an interview plan. The interview started and Nicolai asked some questions, then all 3 of the AKO Capital team started writing down the answers that were being given. Diligently taking notes. The interview went on. Nicolai carried on, answering, asking questions, more notes. I started relaxing then because I thought there is some work, some damages I call them, some work I can do here and make this better for them. Just then, somebody cut in across Nicolai to ask the CEO a different question, and the interview went off in another direction.

At the end of the interview, we had a debrief, and I was asked to give them feedback on how the CEO would answer the questions, and I said, well I wasn’t too enamored with one or two of his answers. He was asked how and when he was going to turn things around, and he didn’t apply any timescales to it. I think it’s very easy to say you’re going to think, make things better. What’s less easy is to be accountable for when, and things like that during the during the debrief. I also was asked about the interview techniques. I said: “Yeah, who was actually listening?”, and they responded: “All of us”, then I asked: “Well, what do you listen with?”. They responded: “Your ears?”. I said “No, no, no, no”. And so, they were all taking content, all 3 of them. I just simply talked about the points on notetaking, and to them that was sort of mind blowing. Really, have the interviewer actually watch and look at the person who’s answering the questions.

So that’s one study in terms of what happened with the business case. I mean, it wasn’t just me, obviously, and I would never advocate making decision just based on what I said. They’d also had some indications from forensic accounting that there were some issues. They’d had some difficulties with the company earlier and I just put the icing on the cake by giving them the feedback about the way the CEO had answered and not answered some of the questions, and they exited, and sure enough, saved the investors a fortune, really, because the stock price plummeted. I think the profit warned a quarter later, or something like that and but by then we’d exited.

Tarek: How relevant does this work become when screening for companies that are majority owned by a family for example and where the governance/ incentive structure is shareholder-friendly?

Laurence: I smile because there were one or two companies that we’ve invested at the time where they were family-owned companies. And I think it’s great because you’ve got a bigger stake in in the whole thing if you built it yourself, and you feel you love it and all the rest of it, and so I think there’s a lot of power to that in terms of the investment case. But also, I’ve seen in one particular company, and in more than one company where you might have somebody who founded the company and then it’s sort of run the company with their family for many years and then become the chair of the Governance Board. Then, a poor CEO is trying to manage this whole situation where the chairperson is actually the company and is interfering with and getting involved in day-to-day operations, perhaps more than they should. So I smile for that reason, really, that I can sort of sympathize and empathize with that whole position and you might not have the full power that you feel you should have in the running of the company, because of the power of an individual or of a family who have developed, and you know, and brought up and grown a business really. So, there’s a lot of stuff you can deduce from just the things that happen in those companies, and then, when you interview them as well. That particular company merged with another company around the same time. It was a massive merger and who was in charge and who was running the place was anybody’s guess for the first 12 months, and I was looking up to be in one or two face to face interviews with the joint team of both CEOs, one from one side, one from the other, and let’s just say it took a long time to sort out

Tarek: So, it helps to still get organizational clarity, probably power of decision making and who it sits with in the end. So, we have the framework on interviews and the models that you mentioned and the skills, etc. Have there been times where you had, a false positive, or a false negative, applying your methods? For example, you’ve done your analysis, and it’s raised a few alarms but actually, the case turned out to be okay.

Laurence: Yeah, that happened a lot. I mean, as I said, I’m sold on the process, maybe I could use the one thing that we haven’t touched upon yet is the computational approach to language analysis. I found myself sat there after about a year or 2 as a small team, then I’m thinking, we’re quite reactive. We’re good for risk management. My team we troll through the earnings call transcripts of everything in the book, by the time we had a full team we did pick up things that are going to be risk for really big positions for the company, and enable the analysts to probe into them further, to hopefully discover whether there’s anything behind them or not but we’re not proactive. We’re not suggesting stocks even to the short team and so there must be a machine in the sky which we can build, and into one end you feed all the transcripts from the universe that we are interested in and out the other end come 10 bad ones and 10 really good ones. I always jokingly say, and I’m the big thinker because I couldn’t do that myself, but I thought it might be a good idea to start to try.

In around 2013, we started looking at natural language processing and using computational linguists and data scientists and stuff. So it’s like that, really before AI and before, really, I think you know there’s other firms out there now that have been in that sort of world and selling it to the financial industry for many years now, but probably before that, or at the outset of that, anyway. So we did a lot of work with some academics I mentioned earlier, Stanford and we looked at some dictionaries that are publicly available, that were sort of contextualized for finance dictionary by people called Loughran-Mcdonald, which were produced in about 2011 to 2013, something like that, and they were basically buckets of words that you could apply to analysis, it might be positive words, negative, no words, pronouns, whatever it might be, and funny enough, we were doing some work with Stanford around all our data because over the 7 years, we built up a massive library of data, analyzed transcripts that we looked at as a team all graded at a risk matrix, like from 1 to 5, being really good or bad in our opinion, and we were looking at positive and negative words. Funny enough, I came across the word ‘execution’ in the financial world, which has a different meaning to the word execution in another context. So that was one funny story, really, where we were getting hits on negative hits on companies. Then, looking at what’s behind this or that and analyzing it deeper. We put a human on it to actually look at it, and it was all about the fact that the CEO would use the word execution more times than they normally do in a particular call. However, it meant nothing really. So really important to follow up on that, but yeah, we weren’t always right. I would never claim that we were always right at all.

Tarek: How do you evolve the process? I picked up somewhere on your Linkedin, where you mentioned that you refined the process by qualitative and quantitative indicators, as well. What are the quantitative indicators in behavioral analysis, or how would you do that?

Laurence: Well, the quantitative approach is more about the language really than anything else. What I said to my team when we started doing this, my team got really nervous because we were just doing qualitative work, human analysis at the time, and they got really nervous. They asked: “Is this my job?”, I suppose, like people do perhaps now with AI, is it going to take my job? And I was always trying to reassure them, and of course it hasn’t, and it didn’t, because there’s real value in both.

The value I used to say, for example with the computational approach, what would now be the AI approach, is the helicopter view that looks across big data and is strategic and can look at individual companies for over a longer period of time, and can tell you things about how they might change the language, or how they might be using some red flag language at certain times but the human analysis can do things that machines don’t, for example machines don’t do irony. So, I talked about the little relationship things you might have even in a transcript where you’ve not met the CEO. You might have a CEO who’s made a joke at the expense of somebody, and that tells you something about the relationship between those two managers, really, and I don’t think a machine, I might be wrong, but I don’t think even now machines can pick up stuff like that. Also, part of our role as human analysts was to flag areas that the analysts might want to approach in their next meeting with the managers. There was real value in both qualitative and quantitative, really and the quantity to answer your question purely relates to words and how they use what.

Tarek: I was in a class where we had a senior executive guest speaker tell us how they’ve received a lot of training on public speaking to avoid AI models capturing any insights from what he/ she says on prepared remarks, and it sound like that’s where you’re trying to allude to the human element where the machine is still not capable of understanding.

Laurence: Well, yeah, it’s an interesting point you make. Because I’ve been challenged many times over the years by people who said there’s a lot of validity in the question. I’ve said, you know CEOs must train themselves to be able to speak even before AI. If we were looking at ‘tone’, positive words versus negative words to use more of those, or to answer questions in a way that you’re not going to be able to judge anything, well, I always go back to it, depending how you ask the question really. If somebody doesn’t answer the question, they don’t answer the question. It doesn’t really matter what they say, if the questions are asked correctly, to give them the best opportunity to answer, and they answer in a certain way, or they don’t, how they fail to answer. Then that’s an indicator in itself. It’s often most times you’ll never get to the bottom of why, but the indicators add up. I think it’s true that CEOs will undertake such trainings. I had media training in the cops, and I used to go on the TV and on the radio quite a lot, appealing for witnesses. So, we had an investment in doing those interviews for our cases, and to protect the public as well. I was trained how not to answer questions of the reporters, as politicians are trained in that way as well, but someone analyzing my interviews with the reporters would have recognized straight away, he hasn’t answered that question and that tells you something about me. My motive was to get my information across as quickly as possible, because the media interviews are over like that, and I need to get it across, that’s what I was taught. CEOs, I’m sure, are taught how to manage conversations with equity analysts. The key is asking the right questions.

Tarek: How open are you to an approach including asking the right question but also incorporating new technology to deliver an improved job. For example, what is your view on the use of ‘Phonetics’ to assist the analysis?

Laurence: Yeah, we did explore phonetics when I was at AKO Capital, actually, and we didn’t really come up with anything that it was telling us. We did about a 6 or 12 month study into one of our team had done some work in phonetics, and so I don’t understand phonetics. It’s about as you say, the tone of voice really more than anything, whether there’s any value in or whether there’s any academic rigor behind people’s tone change, and no doubt there is. The difficulty becomes with all these things that you’ve got to manage the relationships with the individuals that you’re having conversations with. It’s like recording, Zoom is recorded in that these days, isn’t it? But we have many long arguments at AKO Capital, and then it was discussed again at Norges Bank when I was at the sovereign wealth fund about whether, unless you’ve asked the CEO if they minded them recording on tape. The interview that stops you having to write things down enables a better conversation, but the feeling was that CEOs would be more guarded in the way they speak about things. I found in the police that it didn’t actually happen; people forgot about it usually and they forgot the tape was on and would have a normal conversation. It’s an interesting topic technology, so what I always used to say when we were having those discussions is, you can never do this covertly. You have to tell people because you have a relationship to maintain with individuals, and it’s a professional business relationship, and you’ve got a reputation to uphold as well. I find it difficult how you would, unless I feel I should always tell these people that we’re looking at the top we’re like, we’re recording this, I just think it’s in that regard it creates some difficulties with phonetics.

Tarek: For fund managers that haven’t adopted behavioral assessments formally or with this much rigor, how would you advise them to start?

Laurence: Give me a job, I suppose (humor). Yeah, it’s difficult, really, because, as I said at the start, I’m amazed how little out there, there is for people, particularly around the interview techniques thing because that’s the thing they can just they can buy in training, for having said that training is forgotten, usually in about 2 weeks, I think the research tells us. I say, when I’m delivering my training, if you remember one or 2 things from this that you use in the future, I’ll be really over the moon. The reason it worked better, I can always think, is because they had someone on the job follow up, and it was embedded. Now, whether people are going to do that I don’t know, but that’s what I’d say would have the biggest effect.

In fact, the interview type side of things became, I’d say 5% my role, and 95% of it became the forensic linguist approach to reading and judging what the answers were to questions not just on individual meetings, but across the earnings calls transcripts. There’s real value in text. I mean, there’s Alpha in text, I firmly believe that. It depends on the firm, the size of the firm. What you know, what the ability is to employ these people. I would just be buying in the best I could find in data science and in AI, and saying, sit there and play. Just let them play for a while really. Because I truly believe there is proper value, proper offering in the way people answer questions, or don’t answer them.

Tarek: Thanks so much Laurence, pleasure to have you with us today.

Book Suggestions:

• The Secret Life of Pronouns: What Our Words Say About Us – James W. Pennebaker
• Investigative Interviewing: The Conversation Management Approach – Eric Shepherd.
• A Guide to the Professional Interview: A Research-based Interview Methodology for People Who Ask Questions – Asbjørn Rachlew, Geir-Egil Løken, Svein Tore Bergestuen

About the interviewee:

Laurence Carr is a UK-based specialist consultant in credibility assessments and conversation management. Laurence was formerly the Head of AKO Capital’s Behavioral Assessment Unit and has experiences as Chief Detective at Merseyside Police and HMIC. Laurence is a practitioner of research-based interview techniques adopted by UK law enforcement and has played a pivotal role in introducing these methods into an investment analysis framework at AKO Capital. Additionally, he has supported the largest sovereign wealth fund, Norges Bank Investment Management, in shaping their behavioral assessment team.

The content of this website is not an offer to sell or the solicitation of an offer to buy any security. The content is distributed for informational purposes only and should not be construed as investment advice or a recommendation to sell or buy any security or other investment, or undertake any investment strategy. There are no warranties, expressed or implied, as to the accuracy, completeness, or results obtained from any information set forth on this website. BeyondProxy’s officers, directors, employees, and/or contributing authors may have positions in and may, from time to time, make purchases or sales of the securities or other investments discussed or evaluated herein.