First of all, the number of confirmed lost bitcoin is far lower than the amount we are estimating as lost. So, you don't know whether people are not touching their bitcoin for long time or they have lost the access.
And the result can not be correct because you don't know which coins belong to which person/institution exactly unless they publicly announced it.
Are you sure? It really is a hard metric to estimate, because any data that we collect would likely be self submitted, and therefore could be fake data. For example, I must come across at least one person a month claiming they have lost big amounts of Bitcoin, and have simply "forgotten" where it is, or forgot the credentials. Now, these people could in fact be lying, or they could be misremembering, or indeed actually lost the amount of coins they lost. Also, we have to consider what people mean by "lost", and whether they've actually be lost, or simply they got scammed etc.
Now, all of the above would support your argument that we are over estimating the amount of Bitcoin that has truly been "lost". However, there's also the flipside where people inflate, and deflate the amount that they actually lost. So, there could well be people out there that's lost thousands, and are too embarrassed, or some people don't want to paint a target on their back.
I'd probably agree we are overestimating the amount, but I can't explicitly state that because we'll never really have great data to work with. The only really useful data we have is the amount of Bitcoin that is in circulation. That's really the only metric that's going to be reliable.