Statistics (3) Census

I see that Donald Trump has instructed the Census Bureau to undertake a new population census. According to the US Post-Enumeration Survey (PES), eight states were overcounted while six were undercounted in the 2020 population census. Trump is concerned that these discrepancies distorted representation in Congress, which is based on census numbers, but there are a couple of issues with his concern.

The biggest threat to the fairness of electoral district boundaries is gerrymandering. Successive governments from both parties have gerrymandered congressional districts to the extent that many never change hands. Introducing a fairer way to set boundaries would contribute more to fair elections than better population estimates.

The other problem is that conducting a census of the population is almost impossible in the modern world. I have explained the difficulties faced in New Zealand in a previous article.

The problems are worse in the United States, as two save money, the Census Bureau operates a postal collection for the population census. That only works well if the agency has up-to-date address lists for all citizens. However, that is not the case. When people move frequently, it is very difficult to maintain accurate address lists. Consequently, many census questionnaires are not delivered to the correct people. And if people don’t bother to post their census form back, getting a response from them is very expensive.

For these reasons, I would be surprised if Donald Trump will get a more accurate census than was achieved in 2020.

Statistics (2) Employment

Trump was upset by the July 2025 payroll report (a less-than-expected monthly job add of 73,000 and a very large negative two-month revision of 258,000). He has also expressed concern that the Payroll Reports for the period leading up to the 2024 election were revised down following the election. He is correct in saying that the statistics for the last year of the Biden presidency overstated the state of the economy, but it probably did not have a huge effect on the outcome of the election (although it might have in different circumstances).

The problem is that early estimates of employment/jobs are very difficult to measure. In the United States, the payroll report is based on a survey of business establishments. This presents several difficulties. The robustness of a sample survey depends on the quality of the population frame from which the sample is selected.

The United States does not have an up-to-date register of businesses, so the sample results have to be adjusted for births and deaths of businesses. Businesses that have closed down will not respond to the survey, although they could still be in the survey. New businesses will not be included in the establishment survey because they will not be on the business register when the sample is selected. The adjustment for births and deaths can be quite large, yet they are difficult to estimate because the relative numbers depend on the stage in the business cycle.

The other problem is that funding for statistics has been cut. I read that funding for CPI surveys has been reduced, so I presume the same applies to the payroll survey as well. Getting a satisfactory response to the surveys of businesses can be difficult. There are methods that can be used to improve the response rate, but they all cost money.

In light of these difficulties, I am not surprised that the payrolls report is not as robust as commentators would like.

In New Zealand, the earliest estimates of monthly job numbers are derived from data supplied to the tax department as part of the PAYE system. There are a few issues with small businesses that have less frequent reporting requirements, but the appropriate adjustments can be estimated. But overall, the Monthly Employment Indicator is a robust statistic. It is supplemented by a quarterly survey of business establishments. The sample is selected from the business register that is kept up to date using GST registrations (value added tax), so there is less of a problem with births and deaths.

Statistics (1) Independence

I have always considered that government statisticians have a prophetic role. They gather data and produce information that demonstrates how the government is performing. For example, if the government is manipulating or debasing its currency, the consumers price index will expose it. If a nation, or its government, has lost its blessing, it should be exposed by GDP statistics.

If statisticians are to undertake this role, they must be independent of the government that they are monitoring. While they will be dependent on the government for funding, they must have the freedom to design and produce statistical information without any political interference. They must be free to use the best possible statistical methods and techniques to produce information about the performance of their nation’s economy.

In New Zealand, the Statistics Act recognises the independence of the Government Statistician. The statistical agency has control over the statistical methods and techniques adopted for the production of national statistics. Down through the years, Statisticians have guarded this authority carefully. Most citizens assume that statistics are not subject to political interference.

Statisticians around the world have always wondered if the statistical agencies in the United States that produce economic statistics are fully independent. The directors of the Bureau of Labour Statistics (BLS), which produces employment measures, the Bureau of Economic Analysis, which produces GDP measures and the National Accounts, and the Census Bureau, which operates business surveys that feed into the measurement of GDP, are all political appointees. This makes it harder for these statistical agencies to claim independence from political interference.

Now, Donald Trump has blown this issue up by firing the Director of the Bureau of Labour Statistics, claiming the jobs statistics she produced were biased, because she was a Biden appointee. He promised to appoint a director who would publish statistics that supported his view of the “beautiful economy”. This will make it very difficult for his appointee to produce independent statistics if they record bad news for the economy.

Picking Piketty Apart (6) Statistics

Thomas Piketty deals with four broad themes in his book called Capital.

1. Statistics
2. Explanatory theory
3. Forecast for the future.
4. Policy recommendations.

Statistics
Piketty puts considerable effort into proving that wealth is unequally distributed. His statistics show that inequality is increasing. Everyone assumes that this is the case, but it is good to have statistical proof.

Piketty presents a lot of statistics. As someone who has dabbled in statistics a bit, I like the way that he does not just do a data dump, but uses his graphs and tables to tell an economic and social story. Piketty and his colleagues have collected a massive database of statistical information. He presents the information in a way that is east to understand. Statisticians often dull their readers with numbers. Piketty uses statistics to illuminate his story.

I have also done enough with statistics to know that measuring income is hard, and measuring wealth is even more difficult, so I expect there will continue to be arguments about his statistics, but I doubt that these will change the underlying story.

That said, Piketty tends to gets a bit exuberant in some of his interpretations. He consistently see a U pattern, a series declines between 1914 and 1950, but is now returning to the level that prevailed in the 19th century. I see quite a few reverse Js. Many of the measures he describes as a U are a long way from returning to the level of the 19th century.

Top Twelve

According to MandM, Blessed Economist is just outside the Top Ten Christian Blogs on Public Discourse in New Zealand. I think it comes in as number 12. Wow!

I am not sure what that means. Most readers of Blessed Economist live outside New Zealand, so I presume that the statistics refer to blogs published in New Zealand, rather than those that are read here.

The statistics are based on measures of unique visitors and links as published by Alexa. I am surprised that this site does so well, because I have never encouraged links from other sites. One way to increase the number of links is to develop a huge blogroll. Following the principle of “You scratched my back, so I will scratch yours”, most of the bloggers on the blogroll will place a link back to the blog on theirs.

I have never got into this game, because I am uneasy about linking to a blog unless I have confidence in the material that will appear on it. I prefer to link to particular posts or groups of posts that express views that I can endorse. This probably reduces my scores, but I care more about integrity than publicity.