The FBI maintains a database of missing people. By changing the year on the URL above, you can look at other years too.
"As of December 31, 2014, NCIC contained 84,924 active missing person records. Juveniles under the age of 18 account for 33,677 (39.7 %) of the records and 43,289 (51.0 %) records when juveniles are defined as under 21 years of age.*"
"During 2014, 635,155 missing person records were entered into NCIC, an increase of 1.2% from the 627,911 records entered in 2013. Missing Person records cleared or canceled during the same period totaled 634,367. Reasons for these removals include: a law enforcement agency located the subject, the individual returned home, or the record had to be removed by the entering agency due to a determination that the record is invalid."
What I see in these two paragraphs: statements about 800K kids going missing every year are either (a) out of date (if you look at the NCIC report, there's a table a ways down that shows activity in the database over a couple decades, and the total entries have been declining for some years now) (b) exaggerated for programmatic purposes or (c) using a different reporting definition than the NCIC. Using the tables further down in the report, we see that 466,949 of the total entries are for under 18 and 498,872 are for under 21.
Almost a half million children or young adults go missing long enough to generate a report to NCIC every year.
And that, My Dear Readers, is somewhat higher than the same 1 child per 1000 total population that NY DOJ reported to the Observer (about 1.4 or thereabouts). http://observer.com/2015/03/can-we-microchip-our-kids-to-prevent-kidnapping/ (BTW, that was a really silly headline. Kidnapping is _not_ the risk that chipping would address.)
So I believe it. Let's go back to the original numbers, tho, and instead of thinking of this in terms of missing kid per total population and start thinking about it in terms of missing kid per population of kids. Otherwise, even tho there are _more people_ in the US under 18 than ever before, the _share_ of people in the US under 18 is about the lowest it has ever been, and that trend is likely to continue/increase (aging population, longer lifespan, later child rearing, fewer total kids per mother/woman of reproductive age, yada yada yada). If we look only in terms of missing kid per _total_ population, that will continue to improve (look, "fewer" kids missing. Yay!) while the parents freak out more and more (because the fraction of kids missing vs. total kids didn't move, or maybe even got worse, conceivably -- or got better, but compared to other kid risks isn't getting better fast enough).
Almost a half million are under 18, and a half million are under 21. 75 million people in the US are under 18. 0.63% of the population under 18 is reported missing to NCIC annually. More or less. It's a little tricky finding the under 21 fraction of the US population, but it looks like it doesn't move the meter by much. [If you find a math error, please let me know! I feel real shaky on some of this stuff. Also, I am assuming no repeaters, which MAY NOT BE JUSTIFIED.]
Here's another way to think about this. Here's the number for under 18 in the database on December 31: 33,677.
Here's the number for traffic fatalities in 2012 in the United States (adults and kids): 33,561.
We live in a world in which children have to sit in booster seats until they hit some height/weight limits, _so the seat belts work better_. Before that, they have to be in specially designed car seats which are expensive and have expiration dates and are frequently redesigned because they are discovered to have Some Issues. We just went through a bunch of car redesign because of a new crash test (frontal offset, IIRC). We have airbags. We did all that to get traffic fatalities down to the same number of kids per year as are currently reported _missing_ and unaccounted for in the NCIC database on an arbitrarily chosen date.
Policy is driven by body count. And we've driven total fatalities due to car crashes _below_ the number of kids in NCIC. Expect this issue to gain some policy attention over time. In the meantime, don't be surprised to find that parents who have already babyproofed their house, bought a Volvo, and taken other measures to ensure the safety of their child come looking for better technology to find a child that has wandered off (autism) or otherwise been misplaced (got/was put on the wrong bus, misplaced on a field trip, missed child care or custodial parent handoff, walked out the wrong bathroom door at the amusement park, etc.)
I've been attempting to match the racial breakdown numbers in the NCIC database to under 18 and under 21 by race for the population. The handling of Hispanic is different and I don't know how to reconcile things. I _do_ think that whites are underrepresented in the numbers, so I would urge any white person who is thinking that this isn't actually a serious problem to maybe do a lot of listening, and be willing to engage in respectful collaboration with groups who find that this is more of a problem, and, in the interests of all children staying safe and parents not being excessively stressed out, be willing to consider devoting resources to the problem.