Data Collection and Revision - The Bureau of Labor Statistics (BLS) releases monthly jobs reports, which include initial job numbers and revisions to the previous three months [1] - The Employment Situation report consists of two surveys: the household survey (calculating the unemployment rate) and the establishment survey (calculating nonfarm payroll employment) [3][4] - The establishment survey is sent to 650,000 workplaces, but only about 60% respond by the initial report release; this increases to over 90% after three months, leading to revisions [4][5] - BLS revises the numbers twice a year with unemployment insurance data and even 10 years later with census data [7] Factors Influencing Revisions - Large revisions are often attributed to economic shifts, such as coming out of or moving into a slow period [8] - Discrepancies between the current year and the previous year's employment patterns can also lead to significant revisions [9] - A specific example is the suspicion that schools are not hiring as much as they did last year due to the depletion of COVID-era funding [10] Data Integrity and Trust - Experts believe that political influence is not a factor in the revisions [10] - Eroding trust in the statistical system is detrimental to companies' decision-making processes and the country's infrastructure [12][13] - Improving the initial report completion rate is seen as the best way to reduce drastic revisions [11]

RT Chamath Palihapitiya (@chamath)Non Farm Payrolls are total garbage so I asked Grok:“Hey Grok, go look at the Bureau of Labor Statistics website for their non farm payrolls data. Tell me how many times their original forecasts have been revised since Jan 2020. And, of those revisions, how many times the data was revised up versus down. Categorize this during the Biden versus Trump presidencies.”Bottom line is that BLS isn’t so much conspiratorial as it is inadequate in its approach. They are all over the ...