Stupid question but can you explain exactly what is meant by productivity? I assume it's something like (value of output)/(cost of input). So more output for the same input means higher productivity. If we are talking about say a factory making widgets productivity it seems straightforward (The value of all the widget's sold) divided by (the cost of running factory and all the labour costs etc). But for a service like the passport service I don't understand. You say
"For most public services the default method of estimating output is to calculate the net cost of providing the service after excluding purchased inputs. This is equal to the sum of (a) wages and salaries including employment benefits, plus (b) actual or estimated rents for buildings, plus (c) the annualised cost of equipment and other non-property assets."
Under that definition the wages and salaries of the service are part of the Output but shouldn't that be part of the Input? Wouldn't your definition mean that paying the staff higher wages would increase productivity?
A step back: the core purpose of measuring national income is to understand how the total resources available to pay (a) wages, etc for labour, (b) profits to the owners of capital, and (c) rents for land and natural resources are generated and change over time. The usual definition of aggregate productivity focuses on labour and is calculated as total national income divided by the total amount of labour used.
Not surprisingly there are many issues in how we measure both the numerator and the denominator, but one crucial point is that the numerator is not the total value of output, since that would lead to lots of double counting. Instead it is the total value of output minus the value of inputs other than labour, capital and land used to produce that output. That is referred to as gross value-added (GVA). By this accounting definition GVA is equal to the income paid to the providers of labour, capital, and land.
So aggregate or sector (labour) productivity is total or sector GVA divided by total or sector labour input (usually number of people employed in total or in the sector). Economists are also interested in capital productivity, which is total or sector GVA divided by total or sector estimates of the amount of capital employed in total or in the sector. There are other measures that I won't go into.
All of this works fine as long as we have estimates of the value of output in a sector that do not depend on how much we pay for labour, capital, etc. Unfortunately for public services we don't have such estimates, so instead the GVA of public services is measured as the total sum of payments to factors of production. As you say in your 3rd paragraph, the consequence is that paying workers in public services more (in real terms) or using more capital per worker increases their apparent productivity.
Statisticians know that this is silly but they are trapped by the combination of (a) the absence of independent measures of the output of public services, and (b) the requirement in national accounting that total incomes and total expenditures must add up to be the same - this is a variant of double-entry bookkeeping.
Thank you Gordon, A few years ago South Cambridgeshire District Council put its staff on a 4 day week. There was no reduction in staff numbers and no cut in salaries. The argument for it was that it would improve productivity. It seemed crazy to me however Google tells me that in that council 21 out of 24 service areas improved performance. Oddly enough despite that I don't think any other council has adopted the 4 day week. Perhaps they should try a 3 day week and improve productivity even more.
This is the fixed overhead phenomenon. Starting & stopping work absorbs, say, 15-30 mins of each work day, so there can be more useful worktime in a 37.5 hour work week over 4 days than 5 days. As you note the same would be true for a 3-day week. Output per hour tends to fall off over long working days as people get tired or lose concentration.
Many hospitals schedule medical staff to work 12 hour shifts because this is convenient administratively. How many have ever properly measured output per hour, errors, etc for different shift patterns? Management convenience wins out over any serious evaluation.
For workers 3 or 4 days a week work patterns reduce the costs of travelling or childcare, etc but there is little acknowledgement that they may be more stressful and fatigued leading to errors and less effective working.
As a final point, when I was at school my school operated five and a half days a week (with Wednesday afternoon for sport). For someone who had to spend nearly 45 mins getting to or from school, a half-day school session was a huge nuisance. I would have been only too happy to trade a longer school day for 5 days a week. On the other side, teachers were convinced that teenagers didn't concentrate for more than 5-6 hours of classes per day - and they were probably right. The trade-offs are complicated.
I went to Cardiff High School for Boys which went in on Saturday mornings and then we had either Tuesday or Thursday afternoon for sport with the other afternoon off.
Different sources agree that manufacturing accounts for 8-9% of GDP.
Reported estimates of the share of services in GDP vary a lot more because of differences in definitions. The World Bank databank reports a share of 72% in 2024 while the ONS reports a share of 79% in 2025 and the House of Commons Library puts the share at 81% for the same year. I don't know why the World Bank figures are such an outlier as usually they rely on national statistical agencies. Most other sources use a figure of about 80% for the UK.
The differences arise from how government activities are treated. In the US, the Bureau of Economic Affairs (which prepares National Accounts) reports the share of private services in GDP (about 72%) and the share of federal & other governments in GDP (about 11%), so one can finish up with a range of numbers depending on how the government share is divided.
Not really due to Mrs Thatcher. The share of manufacturing in GDP fell by about 7 percentage points from 1975 to 1990. The share fell by the same amount from 1995 to 2010 under the Blair-Brown governments. It has roughly stabilised since then. The driving forces were initially the growth of the oil & gas sector and later the growth of the financial sector.
Stupid question but can you explain exactly what is meant by productivity? I assume it's something like (value of output)/(cost of input). So more output for the same input means higher productivity. If we are talking about say a factory making widgets productivity it seems straightforward (The value of all the widget's sold) divided by (the cost of running factory and all the labour costs etc). But for a service like the passport service I don't understand. You say
"For most public services the default method of estimating output is to calculate the net cost of providing the service after excluding purchased inputs. This is equal to the sum of (a) wages and salaries including employment benefits, plus (b) actual or estimated rents for buildings, plus (c) the annualised cost of equipment and other non-property assets."
Under that definition the wages and salaries of the service are part of the Output but shouldn't that be part of the Input? Wouldn't your definition mean that paying the staff higher wages would increase productivity?
Sorry if I've misunderstood.
A step back: the core purpose of measuring national income is to understand how the total resources available to pay (a) wages, etc for labour, (b) profits to the owners of capital, and (c) rents for land and natural resources are generated and change over time. The usual definition of aggregate productivity focuses on labour and is calculated as total national income divided by the total amount of labour used.
Not surprisingly there are many issues in how we measure both the numerator and the denominator, but one crucial point is that the numerator is not the total value of output, since that would lead to lots of double counting. Instead it is the total value of output minus the value of inputs other than labour, capital and land used to produce that output. That is referred to as gross value-added (GVA). By this accounting definition GVA is equal to the income paid to the providers of labour, capital, and land.
So aggregate or sector (labour) productivity is total or sector GVA divided by total or sector labour input (usually number of people employed in total or in the sector). Economists are also interested in capital productivity, which is total or sector GVA divided by total or sector estimates of the amount of capital employed in total or in the sector. There are other measures that I won't go into.
All of this works fine as long as we have estimates of the value of output in a sector that do not depend on how much we pay for labour, capital, etc. Unfortunately for public services we don't have such estimates, so instead the GVA of public services is measured as the total sum of payments to factors of production. As you say in your 3rd paragraph, the consequence is that paying workers in public services more (in real terms) or using more capital per worker increases their apparent productivity.
Statisticians know that this is silly but they are trapped by the combination of (a) the absence of independent measures of the output of public services, and (b) the requirement in national accounting that total incomes and total expenditures must add up to be the same - this is a variant of double-entry bookkeeping.
Thank you Gordon, A few years ago South Cambridgeshire District Council put its staff on a 4 day week. There was no reduction in staff numbers and no cut in salaries. The argument for it was that it would improve productivity. It seemed crazy to me however Google tells me that in that council 21 out of 24 service areas improved performance. Oddly enough despite that I don't think any other council has adopted the 4 day week. Perhaps they should try a 3 day week and improve productivity even more.
This is the fixed overhead phenomenon. Starting & stopping work absorbs, say, 15-30 mins of each work day, so there can be more useful worktime in a 37.5 hour work week over 4 days than 5 days. As you note the same would be true for a 3-day week. Output per hour tends to fall off over long working days as people get tired or lose concentration.
Many hospitals schedule medical staff to work 12 hour shifts because this is convenient administratively. How many have ever properly measured output per hour, errors, etc for different shift patterns? Management convenience wins out over any serious evaluation.
For workers 3 or 4 days a week work patterns reduce the costs of travelling or childcare, etc but there is little acknowledgement that they may be more stressful and fatigued leading to errors and less effective working.
As a final point, when I was at school my school operated five and a half days a week (with Wednesday afternoon for sport). For someone who had to spend nearly 45 mins getting to or from school, a half-day school session was a huge nuisance. I would have been only too happy to trade a longer school day for 5 days a week. On the other side, teachers were convinced that teenagers didn't concentrate for more than 5-6 hours of classes per day - and they were probably right. The trade-offs are complicated.
I went to Cardiff High School for Boys which went in on Saturday mornings and then we had either Tuesday or Thursday afternoon for sport with the other afternoon off.
Thanks for your help. No more questions.
What are the U.K. figures for national income? On the 75-85% services, 10% manufacturing, where does the U.K. sit?
Different sources agree that manufacturing accounts for 8-9% of GDP.
Reported estimates of the share of services in GDP vary a lot more because of differences in definitions. The World Bank databank reports a share of 72% in 2024 while the ONS reports a share of 79% in 2025 and the House of Commons Library puts the share at 81% for the same year. I don't know why the World Bank figures are such an outlier as usually they rely on national statistical agencies. Most other sources use a figure of about 80% for the UK.
The differences arise from how government activities are treated. In the US, the Bureau of Economic Affairs (which prepares National Accounts) reports the share of private services in GDP (about 72%) and the share of federal & other governments in GDP (about 11%), so one can finish up with a range of numbers depending on how the government share is divided.
So keeping 80% and 10% in my head isn’t far wrong
I instinctively knew manufacturing wasn’t high, but didn’t realise it was so low. Maggie’s transition is complete!
Not really due to Mrs Thatcher. The share of manufacturing in GDP fell by about 7 percentage points from 1975 to 1990. The share fell by the same amount from 1995 to 2010 under the Blair-Brown governments. It has roughly stabilised since then. The driving forces were initially the growth of the oil & gas sector and later the growth of the financial sector.