Here's the methods protocol [0] for designing version 6 of the Human Mortality Database being visualized at that link.
Notice how I explicity said the word _designing_ and not something like 'calculating' which would misleadingly characterize such a scheme as something that is objectively derivable from verifiable information with valid uncertainty bounds and without the potential for advocating policy with dubious constructed data.
If you've ever thought p-hacking was a riot in academia's social sciences bullshitting, wait until you read through stuff like this with think-tanks coming up with whatever numbers are necessary to justify a policy to secure more funding to shroud misinformation in a thin veil of perceived scientific validity to people who feel comforted if you just put on a white coat. Not only does it go through "careful rexaminations" of methodology with new versions every year tweaking subtle but fundamental things like "we used to assume uniform distributions of births across the entire year on v.5 and as of v.6 now we decided to assume a proprietary non-uniform distribution that flucuates per year (and our policy objectives...) because there were wars that changed reproductive strategies that we conveniently forgot to account for that last year and the error cascades through to every successive year's death risk. Stay tuned for v.7!"
On top of that and pretty much a direct consequence of this so called 'research' being complete bollocks, is that none of this junk is ever peer-reviewed. Couldn't even clear that hurdle.
Sorry to ruin your parade but the statistics accessible via think-tanks and public governments are a complete fabrication. Want to see actual life expectancy tables? You need to buyout a multi-national insurance conglomerate that has survived the test of time with financial crises and political cycles, peeking into actuary tables without the gimmicks of literal data engineering. I highly doubt you, nor anyone here, would ever get close to these secrets. And even then, the usual business model of such corporations that can attain that large of a risk pool and _could_ identify these stats are usually incentivized to profit from regulatory capture rather than the public-face tagline of accurately predicting the likelihood of risk to perfectly balance premiums against claims and extract profit from investing on the float.
Think tanks and political organizations (including government bodies, the UN and such) are more in the business of selling numbers, than reflecting reality accurately.
The problem is that most others don't have access to the infrastructure and raw data to get accurate numbers themselves (or redo the data collection with better methodology).
I don't have the link right now, but one of my favorite tales is a group of traders who called the bluff of both a pervasive incumbent government spouting all kinds of GDP fudging nonsense around the election cycle and a sprawling company who tends to work closely with and aggressively court rusted towns down on their luck as a source of cheap labor with dependent customers who are forced to rely on conditional subsidies that serve to keep labor in line during a growth hacking period to monopolize and export the company-town model requiring more negotiating leverage by leading shareholders with over-optimistic revenue guidance.
Of course, the story is more of a feel-good rationalization of fickle lottery tickets, due to the realism of "the market can be irrational longer than you can remain solvent" even if you're correct and especially due to the incentives involved to keep the insanity going as long as possible, bail out insiders, and conveniently backtrack on contracts due to concerns that actually losing money as a counter-party is not acceptable for the entrenched interests because the boogeyman of contagion would spread as the house of cards comes collapsing down with highly leveraged banks and pension funds unwinding the horrible bets they made that were rationalized by a feeling of untouchability when cozzying up with politicians.
But the details of the story are more interesting in that the methods used, such as realtime satellite based photography to more accurately estimate sales are definitely out of the current realm of accessible infrastructure and data collection methods for mere mortals who know in the back of their mind the stats are a ruse but can't possibly prove it.
It suggests that life expectancy has gone up sinificantly for all age groups, not just life expectancy at birth.