Comment by lettergram | Hacker News Reader

Comment by lettergram | Hacker News Reader

Parent Full thread

My prediction is we hit (or in reality, have already hit) 48 million unemployed (~30%) in the U.S.

It’s just many aren’t filing for unemployment because they can’t get through OR they decided not to apply (due to receiving stimulus / severance).

Full predictions / data here:

https://austingwalters.com/covid-19-u-s-fatalities-economic-...

The gist is that retailers are going bankrupt (see J. Crew & Neiman Marcus) and places like Macy’s are furloughing most of the staff. Airlines are similarly hit, and I can go on (it’s linked above). If retail and hospitality (and similar sectors) shrink by 80% for 6 months that’s a massive number of unemployed - 33 million.

Further, recruiting and marketing has cut staff at most non-FAANG places (see Uber, Lyft, AirBnB, etc), This will start to impact the tech industry. Banking is going to see massive defaults in the next 3-6 months... the list goes on.

Easy to see unemployment hitting 30% in the next 2 months.

> It’s just many aren’t filing for unemployment because they can’t get through OR they decided not to apply (due to receiving stimulus / severance).

The BLS numbers have nothing to do with unemployment filings. They base their unemployment numbers off two surveys: https://www.bls.gov/news.release/empsit.tn.htm

People not looking could skew the numbers, but since the vast majority of the job losses in the report are considered temporary layoffs, and since the BLS always counts temporarily laid off people as "unemployed" regardless of other criteria, I don't expect it to be a significant factor.

They do however have to be seeking employment... many of whom are not at the moment. Figuring it better to wait, collect welfare, then apply when job postings pick up. I know several family members doing this

I was once one of these panelists. As one, I represented 0000’s of other people via their extrapolation. While my experiences at the time (2009) were common, they did not represent that of many of my fellow unemployed colleagues and friends. In other words: these surveys are guesses. Nothing more... take them as estimates only and know that the truth is always different.

Is there a better alternative? I understand your criticism, but this is not a dataset that can be reasonably obtained in its entirety, so sampling is basically required.

Their methods have been largely consistent over the years, so even if it isn't complete, it's easy to compare to historical trends, which is more important in this context.

Interesting that you feel this data isn’t available on a per case basis... it’s even more interesting that I end up with downvotes foe sharing my actual experiences. HN is tone deaf. Most of you folks work in modern tech! Modern in the sense that your company and job are likely based on big data... ugggh

I think you got downvoted for dismissing the whole concept of random sampling.

In a random survey you're not representing your friends. They will presumably be represented by randomly picking someone who happens to be more like them. (Also unumployed.)

I didn’t dismiss anything of the sort, that’s what’s funny. I shared my actual experience of being chosen as a panelist and have a vastly different outcome and outlook than many of my friends.

It’s not an uncommon assertion that the Fed data is not accurate. Number 1 it’s political. And perhaps more importantly, it’s designed to obfuscate the individual and that, while statistically accurate for static data is anything but for this type of situation. People and specific employment issues are anything but static...

You literally said it's just a guess. It's not. It's a scientific method of estimation which has known expectations of accuracy, and when done right, has a track-record of being able to provide reasonable approximations. Saying it's 'just a guess' implies something very different for which you are rightly downvoted.

Second, explaining the basics of random sampling by saying what applies to you, did not happen to apply to your friends/family, implies you have no understanding of the concept of random sampling either. There is absolutely no requirement for your situation to be the same as your friends/family to make this method useful, but by posting it as an argument, you imply that this does matter. For this, your post is also rightly downvoted.

As for 'static data', it's a sample to provide insight in a snapshot in time. It doesn't claim to be anything but static. It does exactly what it sets out to do.

Garbage in, garbage out.... of course the data over time will correct but the anomalies created by a few outliers are in fact a representation of bad practices. And those outliers are not measured because one cannot ever account for every potential situation when it comes to people. So they create a margin.... works for figures but NOT for policy!! My point,by the way. No wonder SV is still chasing ad optimization as it’s primary business. It’s impossible to ever become efficient, but that doesn’t stop people from chasing the paradigm. Because it’s the cracks and crevices that allow for exploitation, that’s why.... oh never mind, I’m talking psychology to people who think in binary and never the two shall twain

People who work in big data understand how statistics, extrapolation and time-series work. You might be more interested in the ADP data, as they use actual payroll data. Their payroll research shows 20.2 million jobs lost [1].

[1] https://adpemploymentreport.com/2020/April/NER/NER-April-202...

....what they're saying is that there's no feasible way of getting an exact count for unemployment and you have to use some extrapolation.

The plural of datum is data. If you are trying to correctly estimate a number like 14.7% to within 0.1%, you only need to survey maybe half a million people, making each person "represent" 600 others in your words. If you want it within just 1%, you only need about 5 thousand people, making each person "represent" about sixty thousand others. That doesn't make it any less correct, nor is it "extrapolation" in the usual sense. Closer to interpolation. It's definitely not just a guess.

I never mentioned the figures... I was referencing the panel. These were questions, not facts. I was asked a series of questions related to being unemployed and / or seeking employment. That data was extrapolated to millions of others and presented to the public. It was not accurate. My experiences were not common so their random use of my data meant the report was skewed. Even if only slightly. Math is objective. Words are not.

> these surveys are guesses.

Well yeah, that's what statistics is. What matters is your confidence level. Logistically I can't sample the entire population of a city, let alone a country. Even if I could I can't guarantee that people are answering honestly. Guesses are all you are ever going to get.

The risk of re-opening means there will no longer be cover for unemployment benefits. But since the pandemic is still happening, demand won't recover fast enough to put all these people back to work, and the furloughs become permanent layoffs.

Workforce participation dropped from 61% in the US to 51%. That means that 10% of the US are no longer working.

That comes out to 33 million people.

10/61 is 16.3% jobless, but there is "only" a 14.7% unemployment rate. Some of the balance is retiring, some are hospitalized.

Is it also not the case that if you're not looking for work, but do not have a job, that you're not counted as unemployed? You're simply not part of the labour force, despite perhaps being willing and able to work in more favourable economic conditions.

This is at least the international labour organisation (ILO) definition.[0]

This is always a bit tricky as it means that when people are discouraged from finding work (e.g. because it is hopeless and they cannot find a job) and stop searching, that they fall outside the unemployment figures. This undervalues true unemployment a bit.

This may help explain some of the discrepancies in the statistics.

[0] https://www.ilo.org/ilostat-files/Documents/description_UR_E...

This discussion comes up almost every time U-3 (official unemployment) is reported.

There's a reason why the USA keeps track of U1, U2, U3, U4, U5, and U6. Because everyone has their own opinion about who, or what, should count as unemployed.

It sounds like the number YOU are looking for is U4, which is U3 + Discouraged Workers.

--------

U3 is reported because U3 has been reported for the past 100 years. If we want to compare today's unemployment with the unemployment in 1930s, we use U3.

https://www.bls.gov/news.release/empsit.t15.htm

True, but at the same time, we cannot say that there is a 49% unemployment rate despite the fact that 49% of the populace are not employed.

>OR they decided not to apply (due to receiving stimulus / severance).

Does receiving either prevent you from getting unemployment? If not, why would you turn down free money?

Severance can definitely delay receiving unemployment. Receiving the $1200 stimulus does not, but receiving the PPP funds through your employer would. The latter consequence is causing some employers to not use that loan/grant.

For myself, I was furloughed March 23, had about a week of paid Vacation that I used up until March 30th. Filed for unemployment on the 31st. My health benefits continued until April 30th at which time I would be laid off.

Employer got PPP and I was un-furloughed on April 27th and they have enough runway for us until June 30th when the PPP runs out. Everyone got a sizable paycut, myself 37%.

I work in corporate events / trade-shows / experiential advertising and this whole industry won't survive.

Thanks for sharing this analysis.

Do you think this will be their downfall in the long run?

Who are "they"?

It's not good for the now unemployed people but hopefully they'll get back on their feet!

Reply on news.ycombinator.com