Are Grocery and Restaurant Sales Friends or Enemies?

When speaking with clients and colleagues, one of the common themes I hear is grocery and restaurant sales are inversely related. When I talk about grocery sales going down, I inevitably hear, restaurants are killing grocery sales. A colleague once said, "they have to be negatively correlated because you can only eat three meals a day." If you are eating at a restaurant, you are not eating food from a grocery store.

This idea is correct and reflects consumers substituting one good or meal for the other. There are a finite number of meals in a day, so we can expect there to be a negative relationship between the two. Case closed? Are we missing anything? Let's explore the data and see if it uncovers any insights that will help us answer the question.

Lets start be looking at grocery sales and restaurant sales over time. We are going to remove Covid because it was an outlier and only muddies the water.

Looking at the monthly data from 2000 to 2019, the two appear to move together. You can see during the Great Recession they both dipped and both dipped again during 2015 and 2016 as lower oil prices helped to drive down prices and sales for both the grocery stores and restaurants. Overall these two lines appear positively correlated. How can that be? Are they both simply growing over time?

Lets remove time and see what happens. Below we have the YOY view which takes time out of the equation. These two lines also appear to move up and down together, i.e., they are positively correlated. The correlation coefficient is 0.29 which is mildly positively correlated. How can this be?

Is inflation causing the two to move together over time? We can divide both by their inflation rate to get the inflation free trends, which are a proxy for units. They still appear to be going up and down together. The correlation coefficient does fall to 0.18 but is still positive.

Still scratching your head? Why are they not be inversely correlated. Population? Yes, population growth could be causing both to move together. Nope, I ran the per capita sales and units for both and it is no different.

So WTF dude?

Grocery and Restaurant sales are not friends or enemies. The reality is they are mildly positively correlated at best. The reason is because people fail to consider the income effect. There is a substituion effect as we discussed above, but the income effect is much stronger. The bottom line is when people have more money they buy more groceries AND eat out more. In other words, income plays a bigger role in explaining the rate of growth for both grocery and restaurant sales than does subsituting between the two. So next time you are at a cocktail party and someone says restaurant sales are killing supermarket sales, which I hear at most gathering, you can bust out the income effect.