Nearly half of the country is under shelter-in-place orders, as of this writing. Fifteen states have orders, but they account for 43% of the U.S. population and 47% of the personal income, with updated information on orders on Wikipedia and income data from the Bureau of Economic Analysis. (Update on 3/25/2020: 18 states, 45% of population, 50% of income.) Executive orders are generally available online; see, for example, California, Illinois and New York.

The orders allow people to go to work in essential sectors of the economy, but what is essential? It’s complicated.

Each state’s order is unique, so anyone trying to understand whether to keep a business open should look at the specific order and consult an attorney if necessary.

To an economist, essential industries are a challenging concept when the timeline is uncertain. In a one or two-day emergency, like a hurricane coming, essential services would mean healthcare, fire and police. Not car repairs or a chemical plant.

In a three-to-five day emergency, though, grocery stores and pharmacies are essential.

As the length of the emergency stretches out, the range of essential services increase. Car repairs are needed by nurses headed to hospitals and by many other essential employees. Construction work becomes essential for repair of electric, plumbing and roofing systems. And that chemical plant might be making the ingredients for plastic bags, which include the trash can liners needed to safely dispose of medical and household waste. With just-in-time inventory management, we cannot shut down factories very long before shortages plague us.

What does it take to keep all of these essential businesses running? We need the financial system. The executive orders usually mention banking, but insurance and investment brokerage are also vital. Tech support is necessary. Maintaining the hardware of essential businesses requires repair and replacement parts, which means manufacturing, wholesale and retail companies, and transportation and warehousing.

As an economics professor, I used broad categories: labor and capital, guns and butter. But working with businesses I learned about specificity. Wheat, in our textbooks, is a homogeneous commodity—until a mill owner explains hard and soft, red and white, as well as protein content. And wheat is pretty simple. Steel presents more varieties, and consider the many types of electronic components in use today.

In the real economy, decisions are not about the broad categories—wheat and healthcare—but particular products for particular purposes. Generalizations fail.

The concept of “essential goods and services” makes sense, but drawing lines accurately is impossible in practice, especially given that we don’t know when the shelter-in-place orders will end.

Looking at employment by sector in the U.S. economy, one could argue that about two-thirds of activity is essential. In my tabulation, for example, broadcasting of radio and television is essential, but production of new movies is not. Manufacturing paper products is essential, but apparel is not.

Businesses in gray areas might want to think about whether they are serving essential businesses. The supply chain, counting services as well as goods, is long and winding. Downstream from many companies is a hospital, fire station or grocery store.

Policy-makers should err on the humble side. The economy is a very complex web in which goods and services pass many intersections on their way to the final user. A seemingly non-essential good may be critical to the operation of an obviously essential service.