The numbers game, by Tony Travers

22 Nov 07
From public sector productivity to migration, official statistics have rarely been so hotly contested. Tony Travers analyses the reasons for the party-political punch-ups

23 November 2007

From public sector productivity to migration, official statistics have rarely been so hotly contested. Tony Travers analyses the reasons for the party-political punch-ups

Does public sector productivity matter? That is, does it matter if we can't measure it – if we cannot accurately compare what goes in against what goes out. Recent developments involving the Office for National Statistics point to a particular problem with this and with many politically contested numbers.

The debate about productivity has rumbled on for some time. Back in 2004, the ONS published what turned out to be controversial NHS productivity figures. These suggested that 'output' per unit of 'input' had fallen while Labour was pumping real-terms spending increases into the service. The then health secretary, John Reid, aggressively attacked the ONS and its research. Suitably chastened, the national statistician appointed Sir Tony Atkinson to review public sector productivity. His voluminous report in turn led to the creation of the UK Centre for the Measurement of Government Activity.

As a result of all this work, the ONS has published improved productivity measures for adult social care and education. Even adjusting for improvements in quality, productivity appears to be falling.

Public Finance reported education experts who 'poured scorn' on the figures. One union leader described the ONS study as 'complete nonsense', another as 'simplistic economics'. It was argued that the new measures failed to take account of intangible improvements to the classroom experience or to children's lives.

Moreover, the traditional way productivity was measured means that putting a teaching assistant into a classroom would, other things being equal, produce a 'fall' in productivity. Clunky measures of productivity that might be appropriate in a widget factory were not, it was argued, helpful in education.

There is a powerful alliance of opponents to public sector productivity measures – certainly to any that suggest there has been a worsening of performance. Ministers, head teachers, trades unions, companies supplying services and other parts of the 'education empire' all want to avoid any suggestion that real-terms spending increases do not lead to pound-for-pound improvements in output. The same is true for other services. There was little enthusiasm for Sir Derek Wanless's recent observations about falling productivity in the NHS.

The exception, of course, is on the Opposition benches. The Conservatives have gleefully jumped on any evidence about falling productivity as proof that the government is wasteful and incompetent. Anti-taxation campaigners, similarly, will point to 'falling productivity' as an element in their case for tax cuts. In short, ONS productivity measures are part of the weaponry of the adversarial struggle between government and Opposition. 'Falling' productivity in education or the NHS becomes part of Conservative leader David Cameron's case against Prime Minister Gordon Brown. The statisticians find themselves providing ammunition for a party political dogfight.

The ONS is trapped in the middle of another, separate, struggle over migration and the measurement of population. The sudden rise in international in-migration since 1997 has turned the issue into the public's Number One opinion poll concern. Coincidentally, the arrival of significant numbers of new people in many council areas has damaged the ONS's capacity to pick up year-to-year changes. As the funding of local government and the NHS depends on population figures, there is a powerful reason for councils with increasing populations to complain of ONS undercounts. The huge political salience of migration has the additional effect of leaving the statisticians in the position of supplying fuel to a political conflagration.

The Home Office's recent admission that it had under-estimated the number of working migrants produced a classic 'just how dodgy are these numbers?' response from the Opposition. Within hours of the original admission that the numbers were wrong, Whitehall also had to revise the total number of new jobs, of which the migrant workers were a sub-set. Suddenly, the migrants represented a half rather than a third of the total of all new jobs. The government managed to look both shifty and incompetent. Trust in official figures plunged still further.

Much the same is true of crime statistics – also produced by the ill-starred Home Office – and of the number of people living in poverty. Home Secretary Jacqui Smith's recent statement about the employment of 'illegals' within government hardly helps. Because the political stakes are high, official statistics become the basis of an all-against-all struggle. For those who do not like the numbers, generally the government, there will always be a temptation to shoot – or at least undermine the reputation of – the messenger.

This politicisation of statistics is undoubtedly worse in Britain than elsewhere. By setting so many quantifiable targets, the government accidentally turned political attention to the precision of official statistics. Sets of data that gave a broad view of how the country was changing – such as number of teen pregnancies or particles of pollution – have become measures by which the government is judged. If the figures have suggested slippage or failure, the government has felt tempted to attack the accuracy of the measurement – as John Reid did with NHS productivity. If a target is hit, the Opposition accuses Whitehall, and thus statisticians, of massaging the numbers.

Once it became clear that some official statistics were not particularly robust, the Opposition could make a more generalised case that poor-quality data were further evidence of the government's perfidy. We have now reached a position where all Home Office statistics are viewed with suspicion. The Opposition uses that department's many failures – numerical or otherwise – to imply failure elsewhere in the government's statistical output.

So we have arrived at a point where both government and Opposition can have an interest in attacking official statistics. This is a very bad place to be. In a democracy, it is essential that electors have independent and trusted data to help them assess the performance of the government of the day. If official statistics are seen as untrustworthy, people will be left to make judgements on the basis of little more than personal experience and prejudice.

Thus, if we return to the productivity issue, it is now the case that no-one can reliably assess the relationship between inputs and outputs in any of the public services. While the government presses the private sector to improve productivity towards the levels achieved in, say, France or the US, it will not be possible to assess the performance of public services.

The comparison with the private sector is revealing. Because the government has not set dozens of precise targets for private companies to hit, there is little interest in the quality of measures of, say, profitability or productivity.

Yet if all measures of public sector productivity are seen as flawed, there is no methodologically sound way of convincing the electorate that any extra money they pay in taxes is being wisely used. Additional spending on schools, hospitals, care for the elderly or the police might be seen as intrinsically good, but there can be no way of showing that extra money produces pro rata improvements.

The challenge to such productivity figures also puts the ONS, already under pressure because of its move from London to Newport in Wales, in a spot. If most other parties believe that it and the UK Centre for the Measurement of Government Activity are churning out measures that are 'simplistic' or 'nonsense', it will further undermine the credibility of official statistics.

This would be a second-order consequence of the politicisation of contested targets set by the government. Not only are the ONS's earlier estimates of productivity seen as inadequate, so are those produced after a vast expert inquiry into the subject.

But no one has put up an alternative set. Indeed, it is hard to imagine any other British institution devoting the huge amount of resources so far used by official statisticians to produce more robust figures. The time might have come to abandon any effort to develop agreed and trusted measures of public service productivity in the UK.

Is there any way to rescue official statistics from their declining status? In July 2006, the Treasury Committee observed 'levels of public confidence in statistics in the UK are worryingly low'. A new Statistics Board has been set up to assess and approve ONS's output. It will report to Parliament, which will also approve the appointment of its chair. By giving MPs the opportunity to scrutinise the appointment of the person who heads this new 'independent' body and also making Parliament the reporting body for the board, it is presumably hoped that the ONS will in future benefit from the kind of status accorded to the National Audit Office.

This reform might help, but it probably won't. The existing 'confidence' problem does not arise because ministers have been responsible for the ONS and its funding. Rather, the mistrust of official statistics has come about as a result of two distinct changes. The first, discussed above, is the politicisation of official statistics brought about by their use in measuring the government's performance against targets.

Second, trust in politicians – whether government or Parliament – has collapsed. The Audit Commission, which is appointed by the government, and the NAO do not derive their perceived independence from their method of appointment or accountability. They are seen as free of Whitehall control because they publish reports that produce 'watchdog bites master' headlines. This aspect of their output guarantees them a degree of trust and reverence.

If the ONS and its output are to be seen as independent once again, government and Opposition would have to stop using official statistics as ammunition in an endless public service war. There is very little sign of such a ceasefire, apart from some reduction in the use of targets. There will still be plenty of important statistics to be used by politicians to beat each other up with.

Decentralising political decision-making might help, because the 'national' ONS would be providing data by which 'regional' or 'local' bodies would be judged. Such a separation of data production at one level of government from their use at other levels would certainly reduce the pressure on the ONS.

But the short-term prognosis is less optimistic. Battles will continue over measures of productivity, crime, population, migration and dozens of other contested spheres of government activity. The ONS will operate within an accountability framework that is slightly more distant from Whitehall than in the past. But unless ministers and their press offices can stop themselves from manipulating and spinning statistics, the media will continue to treat the production of official data as 'politicised'. And, if they do, it will be a threat to the ONS.

Like so many of the problems that beset the British state, there is no simple fix. It was far easier to corrode the credibility of official statistics than it would be to reconvince a sceptical media and public that they can once again trust such numbers. The impossibility of measuring public sector productivity is itself a longer leading indicator of a growing difficulty for all governments in convincing the public that they can ever again believe that a 'fall in crime' means less crime. A tragedy indeed – and one with profound consequences for democracy.

Tony Travers is the director of the Greater London Group at the London School of Economics


Did you enjoy this article?