The fatality rate of COVID-19 is dramatically different in young people versus old people. I use two primary sources for this inference. The first is the CDC (https://www.cdc.gov/nchs/nvss/vsrr/covid19/index.htm) and the second is an article in the journal Lancet (see https://www.thelancet.com/action/showPdf?pii=S1473-3099%2820%2930243-7). The Lancet articles estimates a death rate of a little under 1% overall. The CDC gives no estimate but I use 1% based on a series of studies, including the Lancet study (see this blog: https://salthillstatistics.com/posts/70). In other words, the CDC data is used only to figure out the relative death rate by age and not the overall death rate, while the Lancet data provides both.
Here is a summary of the results.
While the Lancet and CDC are my main sources, these rates are also supported by data in Iceland and the Diamond Princess Cruise data. The Iceland data can be found here: https://www.covid.is/data. You might note that the age ranges provided by the CDC and Lancet differ slightly but the conclusion about the death rates are very similar, indicating the very different data they use (from China versus the US) bring us to the same conclusions.
A direct comparison to other death rates in general is difficult, because those rates are as a percentage of the population whereas these COVID-19 rates are as a percent of those infected. Not everyone would get infected even if we let the disease run wild, so the COVID-19 rates are over-stated in comparison.
Even so, for people under 25, rate of 1 in about 28,000 deaths for those infected is less than the chance of dying any given year in a car accident of 1 in 9,000. The chance of dying in any accident is roughly 1 in 2,000, more than 10 times higher than the chance of dying with COVID-19 if infected and under 25 (see the National Vital Statistics death report for these and other statistics if you really want to get morbid).
The risk of death from COVID-19 grows with age, but still, people in their 30s have less than 1/20th the risk of people in their 60s (1 in 1,185 versus 1 in 52). By the way, the seasonal flu is also dramatically more deadly for older people. The main difference for COVID-19, is that very young people (under the age of 5) have higher risks of dying from the flu than average but this is not true of COVID-19 (look again the the National Vital Statistics death report for these numbers).
The low death rates amount young people should make us feel a lot more comfortable about getting back to work, especially if we continue to have restrictions that reduce infections. However, we should not get too comfortable. COVID-19 is almost certainly more deadly than the seasonal flu if left unchecked, even for young people. The 2017 flu, for example (and 2017 was a typical year) killed about 1 in 100,000 people aged 25-34. If left unchecked, many more people would get COVID-19 than the seasonal flu, and the death rate is several times as high. This means that the unchecked COVID-19 death rate among people 25-34 could be easily be 20 times that of the seasonal flu this year (assuming half of people get infected in an unchecked population).
What if everyone below age 45 went back to work and school with no restrictions, and we somehow protected everyone else? Assuming a 50% infection rate, there would be about 45,000 deaths among the 190 million people aged 25-45. If we included those up to 55 years old, it would mean an additional 85,000 deaths on top of the 45,000 for including 40 million more people. Want to include those up to 65? That would be an additional 200,000 deaths (on top of the first 130,000).
I'm not suggesting these options, but they put bounds on how bad it could be if, based on age, we completely ignored all restrictions. An obvious middle ground would be to remove some restrictions by age group but continue others, so that, even for the younger population, most people are never infected. How can we open up as much as possible but still keep infections down? I don't think anyone has the answer to this yet.