Economist Clement Bellet is an assistant professor and researcher at the Erasmus University in Rotterdam. His research topic is happiness economics and consumer and worker behaviors. He is the co-author of a recent study on workers’ happiness and their performance.
This article is also available as a Podcast.
Could you describe the research you did with Oxford University and MIT?
‘We know that many companies invest in employee wellbeing and happiness at work. But we know little about whether those investments make sense, so if happy workers are more productive. There might be many reasons why firms may want to invest in workers’ happiness, such as reputation and social responsibility. But we were interested in the productivity question: is it true that happy workers are indeed more productive?
For this study, we collaborated with a particular industry and company that is very interested in workers’ happiness: the call center industry. It is interested in happiness because the work that is done here has to do with interactions with costumers who tend to be unhappy themselves, and finding solutions for them. Another important characteristic of call centers is that productivity is measured in a clear and systematic way. For economists who want to study the link between happiness and productivity, it is a good location to do so.
Our study focuses on call center workers at British Telecom, one of the larger employers in the UK. We studied about 1800 sales workers in the call centers of British Telecom, which has more than 20 call centers across the nation. Sales workers work in 11 of those, so we focused on those. For six months, we measured the happiness of those workers, and we linked their reported happiness to their performance in terms of different metrics, the most important one being the number of sales. How many more sales would a happy worker make week to week? That was the objective of the paper.’
Ideally, this result would motivate firms to assess a number of policies not only on the classical metrics of performance but also on the basis of how happy their workers are and whether these classical metrics or targets may hurt happiness.
How did you measure happiness at work?
‘The firm was already measuring workers’ engagement, but in a non-systematic way, once a year. The kind of questions varied, and it wasn’t anonymous. So we couldn’t use those surveys for research purposes. We got everyone around the table – directors, management, the unions – and we agreed that we would anonymously ask workers for six months to report how happy they were during the past week. Every Thursday, they received an e-mail that asked how happy they felt this week, overall. The same question is asked by the Office of National Statistics in the UK, so it’s a well-established measure of happiness. We asked for their happiness on a 1 to 5 scale. They only had to click on one of the smiley faces in the e-mail, so it costed them very little time. They idea of this simple question was also to have as many people as possible responding, because one of the problems in this kind of research is the low response rate. We were pretty happy in terms of the participation rate of workers: 80% of the workers participated, and conditional on participating in the study, workers responded to a mean of 10 waves.
What were the most important results of the study?
‘We could go beyond mere correlation between happiness and productivity for the first time in a real workplace setting. If I’m happier in a given week and also make more sales, that relation is not necessarily a causal relation. It could well be that because I’m more productive, that makes me happier, for example because I get rewarded for making more sales. The other possibility is that when I’m more productive, I’m less happy, for example because I’m emotionally exhausted. Because of these possibilities, the correlational estimate we get looking at the link between happiness and sales within workers is likely to be biased. This correlation is an interesting result in itself though. It tells that if you get from the least happy to the happiest (which rarely happens), your sales increase by 13%. That’s the correlational relationship within workers. The most important result however is the causal effect, which is much bigger: workers moving from just one point on the happiness scale (which is already a big happiness boost considering our 1-5 points scale), would increase their sales by 24%. This big effect is of a similar magnitude than what was found in prior studies outside of real workplace settings, typically in laboratory experiments where happiness is manipulated, and subjects are asked to solve simple tasks.’
The most important result however is the causal effect: workers moving from just one point on the happiness scale, would increase their sales by 24%.
How did you get to this causal relationship?
‘Ideally, you want to have a positive or negative random shock on their happiness that is unrelated to their performance, and see how that affects their productivity. So, you want to look at the direct effect of happiness. One thing you could do is to ask the firm to show happy movies to their workers and look at the effect. We couldn’t do that, so we relied on what economist call instrumental variables. This is looking at a change in the context that workers face locally, that would only affect their happiness (generate a quasi-random shock) but not their performance. We used the local weather for that. We know from past research that bad weather – rain, fog, no sun – tends to depress people. Within a given week, across the 11 different locations, in some places it suddenly rained a lot, in other places it was not raining or it was even sunny. You would expect that in sunny locations people are slightly happier. This is unrelated to the actual workload or demand, because the demand is national, and they randomly allocate the calls to different call centers. We checked that there was no direct link between the weather and the workers’ performance, for instance commuting costs preventing them from going to work, or a desire to favor leisure activities over work activities, and we didn’t find any evidence for that. So we could look at the variation in happiness that is only explained by weather, and how much of that specific variation explains sales, and here we found a much higher effect.’
How did you come up with the weather as a variable?
‘The idea was based on prior research. In the happiness economics literature, there are a number of studies on weather and climate that found a consistent effect of weather on mood. Also, there are several papers that looked at weather and performance, but those were not able to measure mood or happiness directly. For instance, there is a US study that looks at how much weather impacts judges’ decisions in court. We said: now we have a measure of happiness, we want to have a shock in the UK across the country, that would quasi experimentally affect their happiness. Weather is the best measure for that, because the data can be collected, it varies across locations and it varies over time. In a six-month period, you have a lot of variation of good and bad weather. It’s very random, it’s hard to predict how weather is going to be from one week to the next and local weather is not related to local demand or local customer satisfaction in the first place.’
What we found, is that happiness in call center workers makes them work better, not more.
What were the other important results?
‘Beyond the causal link between happiness and sales, we were interested in what drives that link: is it about labor supply effects, do people work more hours, do they do overtime or take shorter breaks which is why they sell more? Or is it about labor productivity itself, so given the time you’re at work, you convert more calls into sales, or you adhere more to your schedule, or you work faster. So, working more versus working better. What we found, is that happiness in call center workers makes them work better, not more. When we looked at the explanation, our results are consistent with the literature on emotional labor. If you’re happier, you’re more likely to connect to people and better at managing your emotions, and people are more likely to trust you, and that leads to higher sales.’
For people who are not into research, what is the difference between a causal relation and a correlation?
‘For example, think of sales of ice-cream and sales of sunglasses. There is a correlation between these two: the more ice-creams you sell, the more sunglasses are sold. That doesn’t explain why the two are related: the fact that you sell more ice-creams doesn’t cause people to buy sunglasses. The reason here is what we call the omitted variable: that is temperature. On sunny days, people buy more ice-cream and sunglasses. In the case of happiness and performance, the omitted variable could be customer satisfaction. It could be that happy customers are more likely to buy and they also make workers happier, so higher sales can be explained by high customer satisfaction, not worker happiness. That’s why we needed a way to capture only workers happiness. Customers call from all over the UK and are quasi-randomly allocated based on worker availability, so local weather does not affect customer satisfaction. Another issue with correlation and causality, is what we call reverse causality. You could have a correlation between happiness and higher performance, but it’s hard to say whether happiness leads to higher performance, or whether higher performance leads to happiness. Again, to make a causal claim, what you need is to have a variation in happiness that cannot be directly explained by a variation in performance, which is the weather.’
In a sense, the link between happiness and performance reunites the interests of managers and workers.
How extraordinary was it that you found a causal relation between happiness and labor productivity? In other words: what is the impact of the study?
‘We believe that this is the first study to look at the causal relationship between workers’ happiness and their performance in a real workplace setting. When performance can be easily measured, managers know that peoples’ attitude and intrinsic motivation really matter. In call centers, which are not necessarily happy places, a lot of attempts are made to increase workers’ happiness, not always successfully. It was important to us that we first backed up the idea that creating workplaces where workers experience the least unhappy moments, is not only good for workers but also for employers. In a sense, the link between happiness and performance reunites the interests of managers and workers. However, there could still be a trade-off between how much it costs the firm to increase workers’ happiness and how much it benefits. We were not able to look at the cost of increasing workers happiness in the context of this paper. What you could do, is to test a number of management practices aimed at increasing workers’ happiness and measure their impact on productivity. There’s been a couple of studies that looked at things like working from home, giving workers more autonomy, or pay equality, on performance and happiness. In these studies, you don’t know what causes what as these practices affect performance and happiness at the same time. But by evaluating management measures focused on workers’ mood, one could make the case for the design of work environments where people both thrive psychologically and are efficient and successful. Ideally, this result would motivate firms to assess a number of policies not only on the classical metrics of performance but also on the basis of how happy their workers are and whether these classical metrics or targets may hurt happiness.’
Usually when you involve workers, especially when they feel strongly about the firm and motivated in their work, what they want is the company to thrive.
Do you think that the results benefit both employers and employees?
‘Yes, but assuming that workers are involved in the setup of management policies. Especially if you have the workers’ happiness focus in mind. The idea is to both increase performance but also affect the happiness of workers. If the discussion is about designing programs that benefit workers and firms through employee happiness, I think asking workers what makes them unhappy is important. It’s not what people might think, that obviously workers are going to ask for higher pay and less work. Usually when you involve workers, especially when they feel strongly about the firm and motivated in their work, what they want is the company to thrive and they know well what prevents them from doing good work. In a call center setting, when you feel so bad that you can’t answer customers in a positive way, you know that’s not helping the firm. So I think it can benefit both sides.’
So, employers shouldn’t see happiness as a productivity tool, as in let’s get happy so we get more productive?
‘Exactly, such a simple view could definitely backfire. What we show in this study, is if you manage to increase workers happiness, that would affect the overall productivity. How do you do it, is a much more complex issue, but a very important one. In many countries, the least happy places are still workplaces. There is a lot of margin for improvement. Managers shouldn’t see this as a productivity increasing tool, but as a call for internalizing worker’s happiness in their general strategies.’
If people have a disconnect between their work and their values, it’s not good. Let them know why it’s good to work for the firm, and have them associated with some of the decisions.
What are your tips to increase workers happiness?
‘That would depend on the industry and location of the firm. The important elements are first of all workers’ autonomy. In call centers, there is a lot of monitoring and people are asked to do things in a precise way. It’s probably too much. It might work to relax the monitoring, so let them say things the way they want to say them rather than scripting everything, or relax the targets on how many calls they need to answer. If you spend slightly more time on the phone, customers tend to be happier and are more likely to buy.
Secondly, there is the idea of meaning: why would you work for the firm. If people have a disconnect between their work and their values, it’s not good. Let them know why it’s good to work for the firm, and have them associated with some of the decisions. The next tip is the idea of voice and expression. Let workers say what’s not working. There are a lot of places where this is not possible. And last but not least: work life balance. Working hours, time schedule, working from home: there are a lot of improvements possible here.’
One of the recommendations of the study is to do more research. What would be the topic of a follow up study?
‘The first thing would be to have more field experiments. So, convince a company to generate an experiment on the basis of management policies. For instance, allowing workers to express themselves, and do this in some locations and not in others, and see if this has an effect on workers’ wellbeing and hence on their performance.
Another topic would be the consumers. Do happier workplaces also make happier consumers? Do socially responsible firms lead to consumers making better decisions? We know from several studies on fair trade labels, that some customers want to have a special relationship with the brand and the people who work there. They want to know them and have their stories told. The idea that you buy from people you can relate to, would be an interesting research topic.’
If you have a good manager, you’ll be happy, if you have a bad manager, you’ll be miserable. Did you find anything that relates to leadership and happiness?
‘We did measure managers’ happiness. They have an important role in terms of monitoring. A follow up study could be to look deeper into managers’ happiness and performance and team performance, and what could make them aligned. It could be that when workers are unhappy about their managers, that would weigh negatively on their performance. At the same time, the top management could see these managers as good managers, so there may be a disconnect in what workers and top managers think. Possibly the metrics of happiness could be an interesting way to assess this. At a correlational level, we could look at teams with different managers, and the link between the average team happiness and their average performance. To find a causal link, ideally you would like managers to move from one team to the next, and see how that impacts peoples’ happiness and productivity. Unfortunately, that is quite difficult to put in place in a real workplace setting as you need to convince the firm to experimentally move people around.’