Looking at the statistics for the number of public servants in the UK on can see that their numbers are relatively stable over the past 20 years. However it seems strange to me since huge improvements in technology have occurred since that time, which should have made possible to fire a huge number of government employees who were previous dealing with the inefficiencies of paper-based systems. Even services like the police could be reduced in numbers since you can now monitor an entire city from a single office, rather than relying on ever-present foot patrols.

So why aren't we seeing large scale reductions in government employment over the past 20 years? Why do we still need so many people managing things instead of computers?

  • 76
    Why do you assume that any government agency has an incentive to be efficient?
    – user5586
    Commented Aug 5, 2017 at 18:48
  • 9
    – Orhym
    Commented Aug 6, 2017 at 1:09
  • 26
    Why would you fire a whole lot of people instead of just using those people to do more useful things?
    – mrr
    Commented Aug 6, 2017 at 23:14
  • 11
    "The bureaucracy is expanding to meet the needs of the expanding bureaucracy." - Oscar Wilde Commented Aug 7, 2017 at 17:29
  • 14
    A common misconception is that computers make work easier. In reality the amount of work increases while computers make things possible that would otherwise be absolutely impossible.
    – sbecker
    Commented Aug 8, 2017 at 8:59

10 Answers 10


UK Civil Service staff numbers are as follows:

UK Civil Service staff numbers

As you can see, there were around 750k civil servants around when Thatcher took office and computers started to become mainstream. It has been falling down to ~400k since. The UK's population grew about 10% over the same time period.

Put another way, there actually has been large scale reductions in government employment and productivity increases since computers have started to become mainstream.

  • 10
    The reduction is only a correlation however. Too many factors are involved to pin it just on computers. Commented Aug 5, 2017 at 15:42
  • 36
    Confounding factor: privatisation. Huge numbers of staff were transferred to private companies doing effectively the same job on worse contracts with a percentage profit to shareholders taken off the top.
    – pjc50
    Commented Aug 5, 2017 at 16:10
  • 6
    @indigochild You are aware of which year Thatcher took office, right?
    – MJeffryes
    Commented Aug 5, 2017 at 23:53
  • 5
    @muru I agree with indi here; the salient point of the graph is the huge difference in staff numbers before and after WWII, from which they are still descending.
    – Taemyr
    Commented Aug 7, 2017 at 8:09
  • 5
    2. The post war period saw the creation of the welfare state, which is responsible for the majority of civil servants, as well the NHS, the Education act and the nationalisation of key industries.
    – tallus
    Commented Aug 7, 2017 at 23:53

There are two separate effects caused by making it easier to process paperwork. The one that you identified is that it takes less work to process the paperwork that had been required previously. The other effect is that it makes it easier to add more paperwork. If they add enough additional paperwork requirements, that offsets the reduction in the effort to process the old requirements.

Another problem is that an increased ability to produce paperwork works for both sides. It's easier for the criminal to produce fake paperwork. So the government needs more paperwork to check that the paperwork is valid.

Before computers, you would have to store all that paper as paper. Huge, giant file cabinets. With computers, you can store much more paperwork. Scan it in and one hard drive can hold a file cabinet's worth of paperwork. With eFile, you may not even have to scan the paperwork.

TL;DR: there's more paperwork now than there was before computers.

Even services like the police could be reduced in numbers since you can now monitor an entire city from a single office, rather than relying on ever-present foot patrols.

This is completely backwards. The point of foot patrols is not to "monitor" the city. We haven't needed that since phones became omnipresent. The point of foot patrols is to show presence. So criminals go, "Oh, I just saw a police officer. Maybe now isn't the best time to steal that lady's purse." Or deal drugs. Or whatever. We still don't have a technology that really replaces that (although there is some work on security robots).

That's another problem with viewing technology as simplifying things. A side effect of foot patrols was that people could report crimes directly to the foot patrols. But the real purpose was to signal to people that the police cared. If you entirely take away the monitoring role from the police, there is still a need to signal caring.

In addition, any response to monitoring requires police. The foot patrols mean that there are officers right there. What do officers do when there aren't calls? Hide in some remote location? Or openly patrol the city, on foot?

Most aspects of life that can be automated are just part of the process. Even if you take away the entirety of the portion that can be automated, we still have the other portions. And many times, the other portions are nearly as much work as before the automation.

Take water samples for an example. We can automate pouring the water into a testing machine. We can automate printing out and distributing the reports. But what we can't automate is having a trusted person drive out to the location that needs tested, collect the water, and send it to the testing location. Even if we could automate that, we can't automate checking that the automating equipment hasn't been compromised. Pull the equipment out of the stream and put it in a bucket instead. Then you can produce as much pollution as you want. The automation happily generates reams of fake paperwork.

To fix that, we send a person out to the location who checks that the sensor is in the stream and not a bucket. That person tests the sensor and puts it back. Then takes a water sample, labels it, and sends it to the testing location. Which of course is exactly the job that we were trying to automate away. Worse, a significant portion of that job is travel. And travel isn't helped much by skipping every other location.

Net effect of technology:

  • Expensive new equipment.
  • Increased reach (more samples taken).
  • Modest decrease in people doing the original job, as auditing the remote results takes a substantial portion of the work.
  • New work is required to maintain the automation.

We had a similar issue with gasoline. Gasoline efficiency (measured in gallons per mile) doubled. But gasoline usage stayed the same. People used the same amount of gasoline but increased their commute distance instead.

Increased efficiency doesn't necessarily lead to decreased usage. It only does so if we are satisfied with the same output.

  • 21
    I think the answer could emphasize output earlier; it’s kind of implicit throughout, but you don’t actually draw explicit attention to it until literally the last word. There’s not just more paperwork “because,” there’s more paperwork because we always wanted more (more thorough checking, more details and data available, and so on), it’s just that before we were limited in our ability to collect, store, and handle so much. More efficiency can mean the same number of people get more done, rather than getting the same amount done with fewer people.
    – KRyan
    Commented Aug 5, 2017 at 18:47
  • 6
    The problem with computers making police work more efficient is that lots more things effectively become illegal. That is, if dealing with serious crime X becomes more efficient, the police don't just take the rest of the day off. They now have time to harass you about minor offenses Y and Z, which they seldom had time to bother with before.
    – jamesqf
    Commented Aug 6, 2017 at 3:53
  • 10
    @jamesqf Why is this a problem?
    – kettlecrab
    Commented Aug 6, 2017 at 4:30
  • 5
    @Anon234_4521: 1) Because the taxpayers don't see the benefit of increased efficiency. 2) Because more people are victimized by the legal system.
    – jamesqf
    Commented Aug 7, 2017 at 4:17
  • 5
    I think harass and victimize are part of a stigma against police the media and public have helped push around recently. I think the efficiency of the police force increases when they can handle more at once, and that the only people "victimized" would be those committing minor crimes anyway. They have to be enforced to have meaning.
    – kettlecrab
    Commented Aug 7, 2017 at 4:49

Thirty-plus year state government employee here. Hired 1977, retired 2010. Was there from a time when "the computer" had it's own big room to the time when we had them on our desks. Much of our job before was collection of data, then when the machines came, it became collection and entry, not much of a time saver there!

At first the computer did save some time by making difficult things easier, but it also made things that had been impossible possible. And, over time, these previously impossible things were deemed "necessary". There goes more time.

Also any thing that makes it possible to do twice as much work will be sold as making it possible to do three times as much work and used to rationalize giving the staff four times as much work so every step forward comes with two steps back, if not more.


Solow Productivity Paradox : "You can see the computer age everywhere but in the productivity statistics"


This is true for the private sector and state sector, by the way. Computers make much less difference to productivity than you might expect - what you get instead is whole new industries.

  • 11
    Anecdotally, I spent some time typing faxes containing product sales into the computer for a tie company. I'm sure that in the 1950s, nobody would have or could have got weekly sales for every product from every shop that sold their line of ties, so computers instead of reducing the manpower just made it possible to track sales at a much finer level.
    – prosfilaes
    Commented Aug 7, 2017 at 2:46

tl;dr: because government agencies don't operate under those kinds of constraints

Take a step back: why should new technology require fewer staff? If technology make staff more efficient, do you fire the (now extra) staff or move them to some other kinds of tasks?

Firms are interested in earning profit. Efficiency gains lower costs and result in more profit. So for a firm it makes sense to replace staff with technology. They lower costs and earn more profits.

Governments don't work this way. A government agency has a set amount of money it can spend [ Young, 2006]. Efficient agencies don't have larger budgets than inefficient agencies - they are just able to do more with their resources. So when an agency becomes more efficient (such as by introducing new technologies) it will usually those resources. For staff, this generally means keeping roughly the same number of positions, even though positions may be moved between functions. Realistically, even if an agency did cut staff due to technological improvements, it would likely use that money to hire other staff for other programs.

Kernaghan and Gunraj (2004) discuss one particular way this occurs: increased efficiency due to technology also drives staff specialization. Agencies may retrain staff to allow them to specialize rather than fire them.

Finally, an agency in most cases must spend it's entire budget in a year [ also Young, 2006]. If an agency were to cut positions and save money, it would have a lower budget next year. This is also different from private firms: there is a disincentive to cut overall costs.

Is this bad? A commentor suggested this answer is prejudiced against government employees. However, this answer shows why a rational, intelligent manager of a public agency would not cut staff even when technology makes them more efficient. Basically, that would mean paying costs (in terms of a lower budget in future years) and receiving no benefit.

  • 4
    Do you have any source on your affirmations, or is it just prejudice against public administrations ? pjc50's answer shows that the private sector's productivity did not progress more that the public sector's one in the computer age. Denis de Bernardy's answer shows that the number of civil servants in UK did recrease over the period. Meanwhile, lowering costs of state agencies has been a widespread political motto in most of the western world since the eighties.
    – Evargalo
    Commented Aug 7, 2017 at 7:16
  • 2
    Its not prejudiced, there is nothing negative here. This is some basic public administration knowledge. I am sure I can drag it out of an intro book. Commented Aug 7, 2017 at 11:36
  • And of course the number of employees can decrease through the budgeting process, which is beyond thw agency's control. Commented Aug 7, 2017 at 11:38

One factor I do not see addressed in other answers is legislation.

The Civil Service is there to ensure that legislation is enacted. That means the more legislation there is, the more work required to do the job.

And the amount of legislation (statutes) on the legal books steadily increases all the time.

Legislation is constantly being revised, revoked, updated and replaced. And all of that generates enormous amounts of work which the Civil Service is obligated to do. It's not an optional extra, it's what they're there for.

So in some ways it's a small miracle that the Civil Service numbers ever go down.

To change this would require a radical reform not simply of the existing statutes, but the entire framework of law making. You would need to change the court system, the way laws are drafted and developed and interpreted. And you would need to stop politicians voting in new laws just to look like they're doing something.

And that's unlikely to happen.

So computers have probably helped by allowing a lot of the underlying tasks to be automated to some extent or made more efficient. But in the meantime we've simply kept adding to the number of things (statutes) that the Civil Service and Local government have a legal obligation to do.

I've found it hard to find good sources for this. Lots of news stories, lots of opinions, but not many stats. But this site at Columbia college of Arts and Sciences will give you a flavor of the extent of the problem. Dig around and see just how much work the activity of politicians and lawyers forces onto the Civil Service. This of course refers to the US, but it's almost certainly reflected in every other country on the globe.


Public services aren't sold, so its hard to work out how much benefit they provide. However, national statistics are taken. https://www.ons.gov.uk/economy/economicoutputandproductivity/publicservicesproductivity/datasets/growthratesandindicesfortotalpublicserviceoutputinputsandproductivitytable1

    Output  Inputs  Productivity 

1997    100.0   100.0   100.0
1998    102.7   101.5   101.2
1999    105.5   104.7   100.8
2000    109.0   109.2   99.9
2001    113.4   112.7   100.6
2002    118.6   119.8   99.0
2003    124.8   126.8   98.4
2004    130.0   132.2   98.4
2005    134.2   136.4   98.4
2006    137.8   139.1   99.1
2007    139.5   141.1   98.9
2008    143.7   145.8   98.6
2009    146.5   149.0   98.4
2010    148.1   150.0   98.7
2011    148.4   148.1   100.2
2012    150.1   149.3   100.5
2013    151.4   149.5   101.2

As you can see, productivity hasn't risen much in public services. The main effect is larger inputs and outputs. So the public sector has grown but it isn't any more efficient than the late 90s. The public sector grew by 51% since 1997, the economy grew by 30% and population by 10%.

Public services haven't benefited from productivity growth. This may be because these sectors don't benefit from much from computers. Alternatively these sectors may have used new technologies to improve quality, but at a higher cost, this is particularly true in healthcare.


Computer automation doesn't magically happen only on the supply side. Automated demand also happens -- more people use their computers to create work for the government. If automation amplifies both groups equally, you would expect no change in the number of government employees necessary to do the demanded work.

For one example, the European Patent office first allowed electronic filing (via epoline) in December 2003. Between 2000 and 2006, a (roughly, by eyeball, from figure A.1.2 of the WIPO 2008 World Patent Report) 33% increase in filings occurred at the EPO. From the same source, the number of filing to the US PTO is approximately flat from 1933-1983 (although visibly concave), increases in slope shortly after 1983 and then increases in slope again around 2000. Relevantly, the US PTO made eFiling software available in October 2000. Although some other countries show similar second derivatives of filing rates (and some don't) over the 1973 to 2003 time span, for instance, Japan inflects upward in 1953 and about 1980 while China has inflections only after 1993, none are synchronized with the US PTO's software release. In other words, although there is a trend at some patent offices towards higher rates of filing, only the PTO's rates increased in synchrony with the availability of eFiling with the PTO.

As another example, in 1999, the (U.S.) FCC implemented the ECFS system for electronic filing of rulemaking requests. (It was originally planned (63 Fed. Reg. 24,121 (1998)) to be available in July 1998, but the difficulty of delivering large-scale software interfered. Read-only ECFS came up in October 1998 and submissions became possible by September 1999.) The paper "When Do Interest Groups Use Electronic Rulemaking?" investigates reasons for increases in filing (relative to the background rate). The implementation of an automated filing system would be classified as an exogenous event in this paper. The paper finds that exogenous events cause short term increase in filings.

Comparing before ECFS versus after seems to be less than easy. However, for the narrow category of spectrum allocation requests, there is some data obtainable from the FCC Allocation History File. (Note that these are records of successful rulemaking. Many filings whether electronic or paper do not result in eventual rulemaking. We proceed under a (null) hypothesis of matched rates of rulemaking per filing.)

  • 2 amendments to 01 October 1992 edition
  • 2 amendments to 01 October 1993 edition
  • 5 amendments to 01 October 1994 edition
  • 3 amendments to 01 October 1995 edition
  • 8 amendments to 01 October 1996 edition
  • 4 amendments to 01 October 1997 edition
  • 2 amendments to 01 October 1998 edition
  • 6 amendments to 01 October 1999 edition
  • 9 amendments to 01 October 2000 edition, the first period in which ECFS filing were possible
  • 9 amendments to 01 October 2001 edition
  • 8 amendments to 01 October 2002 edition
  • 10 amendments to 01 October 2003 edition
  • 8 amendments to 01 October 2004 edition
  • 5 amendments to 01 October 2005 edition
  • 4 amendments to 01 October 2006 edition ("Staff Update"s and "Staff Correction"s are omitted in this list of counts.)
  • ... and 10 more years of 2 to 5 amendments per year

We see that immediately after the implementation of ECFS, filings appear to increase. (A (one-sided) t-test applied to the amendments to the 1999 and earlier editions and the amendments to the 2000 through 2004 editions testing that the mean has increased gives a p-value of 0.00038..., indicating very strong evidence for the claim of increased rate of filing.) However, unlike the exogenous events described in the above paper, this increase persists for half a decade rather than a month. Consequently, this increase is not of the same nature as the short term exogenous events described in the paper. In other words, automation created a measurable increase in filing over a much larger time span than is explained by typical exogenous events.


In the UK, labour productivity has increased very slowly and then essentially flattened in the last two decades. What that means is that during the period covered by your plot, private businesses haven't been any better at doing more with fewer people, the trend you pointed out applies to the British economy as a whole (and has in fact been identified as a problem by some economists).

Over a longer period, there would be a bunch of things to say about population trends, changing missions, the fact that the number of civil servants did actually decrease, budget restrictions (which can be counter-productive as you don't invest and reorganise without spending money first) and whether there is still a difference to explain after taking all that into account but over the period considered, and as counter-intuitive as it might seem, there simply isn't any kind of IT revolution or huge improvements in technology the public sector would be missing out on.

  • Hasn't the productivity paradox been resolved by now? I've always thought it was only a matter of false assumptions during the 90's, which has been ruled as invalid by now. Commented Aug 7, 2017 at 8:32
  • @JonathanReez US productivity did pick up a bit, I just provided this link as another example of a period where technology wasn't producing the economic effects many people expected it would (calling it a “false assumption” with hindsight doesn't resolve anything: the perception at the time was still exactly the one you have now).
    – Relaxed
    Commented Aug 7, 2017 at 9:17
  • Now, regarding the UK specifically, productivity has stalled in the last decade and that's a problem specific to the UK, as the rest of the EU and OECD has seen some moderate increase over the same time period (the UK curve is close to that of Italy, with only Greece clearly below it).
    – Relaxed
    Commented Aug 7, 2017 at 9:20

The introduction of computers and internet don't necissarily mean that the government decided to go hi-tech. For all we know the computers of all government employees are running Windows XP and only use email, word, excel and maybe a bit of database entry. It's a miracle the NHS is even on Windows 7 (remember that they have a fixed budget and aren't well known for having good computer systems).

Besides which, how would having a database instead of a filing cabinet mean less staff? The forms still have to be filled in by humans. They might get filled and filed quicker, but there's still a neverending supply of them, especially with the massive increase in population.

Can you honestly think of an example of a technological advancement that has rendered a particular job completely obsolete?

What about the additional jobs it has created - someone has to maintain that technology, someone had to design and build the government website, someone has to keep it up to date. If there have been losses, those created jobs would somewhat offset those losses.

Then there's the typical ingrained bureaucracy of "we can process forms faster now, let's make more forms so we can document things in even finer details". Public sector managers are still managers - their minds orient towards 'business'. They still want employess to do GAP analysis and make Gantt charts and have a nice pile of reports on their desk (paper never truly goes away). They want a ton of fancy charts to read so they can pretend they know how to herd cattle and pat themselves on the back for their micromanaging ability.

As for police foot patrols, computers don't erradicate the need for those. Imagine this: foot patrols are gone, instead we havea a person sat at a desk watching CCTV feeds. Suddenly, this person spots a crime in progress right in the centre of town. Now ponder this: who does the person get to respond to the crime? That's right, the nearest officer on foot patr- oh wait, we got rid of those.

Police foot patrols aren't just there to make police presence felt, they're there as a patrol. If a crime happens nearby (whether they are the ones to spot it or not) and the foot patrol can get there quickest (e.g. because it's an area inaccesible to cars) then the foot patrol will be dispatched instead of a patrol car.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .