Traveling through Colombia’s coffee region, my days have been spent drooling over roasted arabica beans on organic coffee fincas, or having religious experiences while sampling the remarkable brew at some of the region’s cafes. It all made me realize that I truly am addicted to the stuff. Without at least two cups of java in the morning, I am a morose, gelatinous, dreary-eyed, delirious blob. And that got me thinking: if coffee is such a crucial input into my own productivity, what about the world at large? Are countries that drink more coffee more productive?
I am not the first person to ask this question. There is a problem, however, when it comes to relating productivity with coffee consumption. On a country level, at least, productivity is generally measured as GDP per capita, i.e. the value of goods and services provided by a country divided by its population. That means that we’re comparing coffee consumption with productivity in terms of a country’s wealth—as opposed to something else, like number of widgets produced, or the number of snaps sent per day.
The issue with GDP, however, is that coffee consumption naturally grows when a country’s inhabitants are more wealthy. Thus, when we observe the positive correlation between coffee consumption per person and GDP per capita (see chart), it’s way more likely the arrow of causality is running in the other direction, i.e. wealth is driving coffee consumption, rather than the other way around.
So do we give up there? Not just yet. In my grand armchair theory about coffee, gains in productivity are (in part) reaped from the extra hours that the precious elixir enables us to pour into our livelihoods each day. It’s difficult to verify this theory empirically, given the issue re: comparing productivity and coffee consumption described above. Moreover, there is a separate debate over whether toiling away more hours adds or detracts from worker productivity. But setting that question aside for a moment, I wondered: do we at least observe that countries with higher coffee consumption also have workers who are more likely to burn the midnight oil at the office?
The answer is, surprisingly, not at all. There is, in fact, an unmistakably negative relationship between cups of coffee per day and the number of hours worked per person. So does this mean I need to totally flip my theory? Does coffee consumption actually make us lazier, because we’re so busy taking all those coffee breaks? Just look at the Netherlands way the eff out there on the bottom right. It makes total sense, since we know exactly what those Dutch are really up to during all those coffee breaks…
In reality, the story is not so simple. When you take a closer look at the countries that form the negative trend, something becomes quite apparent. The countries in the top-left are generally less wealthy than the countries in the bottom-right. Thus, my attempt to ignore country wealth by focusing instead on hours worked was all for naught, because it seems that country wealth is once again rearing its ugly head as a lurking variable.
Just as there is a very strong relationship between wealth and the consumption of coffee, there is also a strong relationship between a country’s wealth and the number of hours people work. It turns out that wealthier countries work fewer hours on average than less wealthy countries. This trend is pretty clear in the graph below—except for Singapore hanging out on the top right, slaving away but still making serious bank. You go, Singapore! Never change.There are a lot of articles out there that try to explain why more productive countries work fewer hours (here’s one). Some conclude that because workers in richer countries are more productive, they need to work less. I think this line of thinking can be potentially problematic, particularly if one equates productivity with efficiency. That could lead people to think that people in poorer countries are lazier on the job, or perhaps incompetent. But we have to remember what productivity actually means in this context. Recall from above that it is the value of goods and services a country makes divided by its population. And when we talk about value here, we are speaking in terms of how the market rewards these goods and services, not in terms of the sweat that goes into making them.
My own take is that workers in richer countries aren’t necessarily working more productively (i.e. more efficiently) than their counterparts in poorer countries, but rather that the types of jobs in richer countries on average tend to be more highly paid than in poorer countries. If you live in Vietnam and weren’t fortunate enough to have decent access to an education like your Oxford-educated friend in England, you probably won’t earn as much per hour as she will. And in order to make ends meet, you’re gonna need to put in more hours on the job.
Anyway, this is starting to veer quite a ways from coffee, and stray closer to another interest of mine, economics. The main thing to remember is that a country’s wealth has a positive influence on coffee consumption and a negative influence on the number of hours worked. Because of this complex tangle of relationships, it can be misleading to rely only on graphs that look at two variables at a time. Luckily, a statistician’s toolbox isn’t limited to scatterplots. By using linear regression, we can actually examine the relationship between coffee consumption and hours worked while controlling for the effect of a country’s wealth.
We can, in fact, control for a host of other variables we might think are important as well. For example, as a hot beverage, we might expect coffee to be less popular in countries with higher average temperatures. We might also control for region of the world as a proxy for culture, since guzzling coffee isn’t quite as big of a thing in say, India or China, as it is in the West. Those countries, for example, seem to prefer tea.
So what happens once we control for all of these variables? Well, it all depends on whether you include Singapore or not. In statistical jargon, Singapore is what is referred to as an influential observation. In other words, it’s an outlier that messes everything up if it is included in the analysis. Whatever is going on in Singapore is clearly very unique to Singapore. If we include it as part of our effort to describe a general trend, it will prove to be more of a distraction than anything else. Thus, we toss it out. Sorry Singapore. I know I said I loved you, but you gotta go. Stay golden.
Once Singapore is out of the picture—and we control for all the variables listed above—it turns out that coffee consumption has no statistically significant effect on the number of hours worked in a country. Thus, the answer to the title of this article is . . . Neither! On a country-level basis, coffee neither makes people work harder nor does it make them take more breaks out of the office. Sorry if that’s a boring conclusion, but don’t shoot the messenger. I’m just telling you what the numbers say.
Of course, this is not the final word on the subject. There may be more granular data out there, with consumption and productivity information recorded at a personal level (ideally as part of a randomized double-blind experiment using caffeine pills vs placebos). Such data would be much better suited to answering the question than the national-level data we’ve been looking at. But maybe you still managed to learn a thing or two about coffee, economics, or statistics in the process. Either way, it’s time for another cup of joe.
- Coffee Consumption Data (Thank you Atlantic magazine, for making me copy the data by hand from an image file)
- Economic Data
- Temperature Data