A Below Average Earnings Report
The latest earnings data for Jersey, released today, show a huge pay disparity between sectors. On average, those working in the finance industry in June earned £600-a-week more than workers in hotels, restaurants and bars. After inflation, average pay across the board rose by just 0.1%.
As always the statistics release concentrate on the arithmetic mean, a wage statistic largely discarded in the UK, in Europe, and even in Guernsey because wage distributions are skewed, meaning that a relatively small number of high earners distort the overall picture.
“average weekly earnings in Jersey were 2.6% higher than in June 2016”
They note that:
“median average weekly earnings of full-time equivalent (FTE) employees was £570 per week”
And
“ mean average weekly earnings of full-time equivalent employees was £730 per week”
It can be seen from this how the results are skewed quite dramatically – there is £160 difference per week between median pay and mean average pay.
The mean or average is very sensitive to outliers, or abnormally low or high values, while medians are much less affected by outliers.
The latest earnings data for Jersey, released today, show a huge pay disparity between sectors. On average, those working in the finance industry in June earned £600-a-week more than workers in hotels, restaurants and bars. After inflation, average pay across the board rose by just 0.1%.
As always the statistics release concentrate on the arithmetic mean, a wage statistic largely discarded in the UK, in Europe, and even in Guernsey because wage distributions are skewed, meaning that a relatively small number of high earners distort the overall picture.
“average weekly earnings in Jersey were 2.6% higher than in June 2016”
They note that:
“median average weekly earnings of full-time equivalent (FTE) employees was £570 per week”
And
“ mean average weekly earnings of full-time equivalent employees was £730 per week”
It can be seen from this how the results are skewed quite dramatically – there is £160 difference per week between median pay and mean average pay.
The mean or average is very sensitive to outliers, or abnormally low or high values, while medians are much less affected by outliers.
As Darrell Huff notes in "How to Lie from Statistics", it is important to used median for skewed distributions:
But Jersey statistics united stresses that
“A median average cannot be calculated from the company-level data collected for the Index of Average Earnings, since this requires earnings at an individual employee level.”
Jersey uses a sampling method, which cannot do this.
Guernsey, of course, simply uses the Social Security figures to get the median earnings at an individual employee level, because as long as the median is below the upper threshold for maximum social security, the figures will still get you the median. They have been doing this for years, and I only found out when I attended a presentation and was flabbergasted to see them using medians when the Jersey statistics unit assured me it was not possible.
Their reports note this:
“Earnings data is recorded by the Social Security Department each quarter and is used to calculate median earnings of employees. The median is the middle value when data are sorted into numerical order. It is a measure of earnings from primary employment, unadjusted for the number of hours worked i.e. the level can be impacted both by changes in the number of hours worked and rates of pay.”
“The data used in this bulletin are supplied by the Social Security Department and include all employed people in the Bailiwick (excluding Sark) earning over the lower earnings limit.”
So the Guernsey reports are full of details about median wages, which are much more informative that Jersey, all of which use the mean, and are therefore subject to distorting effects from hgh earners.
Guernsey also notes that:
“nominal median earnings increased by 2.3% between the year ending December 2015 and the year ending December 2016, from £30,953 to £31,656”
And they also look at quartile earnings, which provide a very useful measure of how wages are spread and how they vary:
“Using four quarter averages, the lower quartile earnings increased by 3.0% between the years ending 31st December 2015 and 31st December 2016, whilst the upper quartile earnings increased by 2.4%.
The lower quartile comes to £21,848 per annum, while the upper is £46,616 per annum.
They also have a table showing “the median, lower and upper quartile earnings of all employees by age group”
“The highest median earnings (£37,180) occurred in the 40-44 age group. The lowest median earnings were in the youngest and oldest age groups, at £15,990 and £18,850 respectively.”
Why can’t we have such detailed statistics rather than ones skewed by still using the average mean for wages?
Like standard deviation, mean is very sensitive to the most abnormal of values, particularly very high values. Why would one use a measure for what people “typically” earn, that is so strongly affected by atypical salaries?
Time for change?
"When you are told that something is an average you still don't know very much about it unless you can find out which of the common kinds of average it is - mean, median, or mode."
"The £10,000 figure I used when I wanted a big one is a mean, the arithmetic average of the incomes of all the families in the neighbourhood. You get it by adding up all the incomes and dividing by the number there are. The smaller figure is a median, and so it tells you that half the families in question have more than £2,000 a year and half have less."
"One kind of average is as good as another for describing the heights of men, but for describing their pocketbooks it is not. If you should list the annual incomes of all the families in a given city you might find that they ranged from not much to perhaps £20,000 or so, and you might find a few very large ones. More than nine-five per cent of the incomes would be under £5,000, putting them way over towards the left-hand side of the curve. Instead of being symmetrical, like a bell, it would be skewed. Its shape would be a little like that of a child's slide, the ladder rising sharply to a peak, the working part sloping gradually down. The mean would be quite a distance from the median. "
“A median average cannot be calculated from the company-level data collected for the Index of Average Earnings, since this requires earnings at an individual employee level.”
Jersey uses a sampling method, which cannot do this.
Guernsey, of course, simply uses the Social Security figures to get the median earnings at an individual employee level, because as long as the median is below the upper threshold for maximum social security, the figures will still get you the median. They have been doing this for years, and I only found out when I attended a presentation and was flabbergasted to see them using medians when the Jersey statistics unit assured me it was not possible.
Their reports note this:
“Earnings data is recorded by the Social Security Department each quarter and is used to calculate median earnings of employees. The median is the middle value when data are sorted into numerical order. It is a measure of earnings from primary employment, unadjusted for the number of hours worked i.e. the level can be impacted both by changes in the number of hours worked and rates of pay.”
“The data used in this bulletin are supplied by the Social Security Department and include all employed people in the Bailiwick (excluding Sark) earning over the lower earnings limit.”
So the Guernsey reports are full of details about median wages, which are much more informative that Jersey, all of which use the mean, and are therefore subject to distorting effects from hgh earners.
Guernsey also notes that:
“nominal median earnings increased by 2.3% between the year ending December 2015 and the year ending December 2016, from £30,953 to £31,656”
And they also look at quartile earnings, which provide a very useful measure of how wages are spread and how they vary:
“Using four quarter averages, the lower quartile earnings increased by 3.0% between the years ending 31st December 2015 and 31st December 2016, whilst the upper quartile earnings increased by 2.4%.
The lower quartile comes to £21,848 per annum, while the upper is £46,616 per annum.
They also have a table showing “the median, lower and upper quartile earnings of all employees by age group”
“The highest median earnings (£37,180) occurred in the 40-44 age group. The lowest median earnings were in the youngest and oldest age groups, at £15,990 and £18,850 respectively.”
Why can’t we have such detailed statistics rather than ones skewed by still using the average mean for wages?
Like standard deviation, mean is very sensitive to the most abnormal of values, particularly very high values. Why would one use a measure for what people “typically” earn, that is so strongly affected by atypical salaries?
Time for change?
Posts
1 comment:
Let's cut to the chase. With just a small amount of breaking down of silo walls, individual wages (and total family wages) could be very easily extracted from Income Tax's database and it would be a trivial Excel challenge to sort them to get a median figure. The number crunching should take less than a split second...
Post a Comment