.png)
.png)
Key points * Lack of holistic training in tax for HMRC Officers * 95% accuracy in processing shown by NAO report conceals an uneven distribution of errors * 'Lean' processing has not yet produced measurable improvements * 'Coding assistant' has significantly reduced errors
I WILL REPORT in a future issue on the Wyman Symposium last Thursday, organised by the ICAEW's Tax Faculty, but one issue raised there chimed with the thoughts I had already had planning this article. Dave Hartnett complained, amongst other things, that tax advisers do not seem to understand how the Revenue works anymore. That's probably true; many advisers got lost somewhere around the last reorganisation-but-one. But why has that happened? There are several reasons, some of which were addressed at the symposium such as the loss of Working Together at a local level, now thankfully resurrected. But it seems to me that one key way that advisers used to be kept up to date with the way the Revenue worked was that they recruited staff at TO or TO(HG) level (thinking in old money) to come and work as tax seniors. From all the discussions I have with advisers, they rarely recruit the equivalent Officer grade staff from HMRC now, and the reason why they don't gets us into the National Audit Office report, Accuracy in processing income tax.
Doing the knowledge
There was a time when staff at these grades would get a rounded training. They might not have been completely ready to work in the private sector, but they had a good thorough grounding in tax.
Now, the NAO report discloses that HMRC are simply not currently recruiting permanent staff. In the large processing offices one in four staff are on fixed term appointments, and only get the training that they need to do the specific part of the processing that they handle. As a result, staff in the focus groups organised by the NAO said that they were demoralised and that the ad hoc training on the job or computer based training that they received was insufficient.
At the Symposium, Rebecca Benneyworth pointed out that advisers used to have a working relationship with local Revenue staff. This is no longer the case, and most advisers (those not dealing with large business clients) have to deal with call centres. Advisers are increasingly dissatisfied with the service they get from these centres, with slow response times and also a lack of knowledge of the tax system and therefore an inability to get proper answers to technical questions.
Errors
The explanations in the report were in the context of mistakes made in processing tax returns and tax codes. It should be said at the outset that the total errors made are not large in the context of the amount of tax being collected. £157 million was overpaid due to errors, and £125 million underpaid. In total some one million taxpayers were affected, which is rather more significant a figure, but it is still less than 5% of income tax cases.
However, analysing the errors in more detail shows that they are far from evenly distributed. Errors in dealing with self-assessment cases have reduced from previous years, with accuracy reaching just under the target rate of 97%. However, accuracy on PAYE cases where no self-assessment return was sent out reduced to 95.1%. One reason for this, the report speculates, is the move to take cases out of self assessment, which led to an increase in more complex PAYE cases. The accuracy on the more complex PAYE cases that needed to be manually processed was significantly lower; 82%. Finally, if the cases where an error is made on (for example) a PAYE code which is then corrected by the final self-assessment are also included, the accuracy rate falls to 78%. Although HMRC might argue that this is unfair, because the right amount of tax is collected at the end of the year, it means either that the taxpayers had to find money to meet unexpected tax liabilities, or that they were deprived of money that was rightfully theirs for at least several months.
Reasons
Accuracy rates varied significantly between offices, particularly in processing PAYE where some offices were down at 66% accuracy and others were up at 93%. The report says that:
'Higher accuracy rates are associated with experienced, well trained staff, lower staff turnover and workloads that require less processing by hand. Some offices have achieved substantial improvements by targeting workloads more closely to the skills and experience of staff, increasing management focus on accuracy, and promoting the sharing of good practice and new ideas among staff.'
Well, it's not exactly rocket science. But in that case, how can accuracy be expected to improve in those offices where one in four staff are on temporary contracts, trained only for the task they are undertaking? And when Lean processing (see below) is leading to increased standardisation, how is it possible for new ideas to be generated in a workforce which feels that they are just the human cogs in a gigantic machine?
Similar issues come out of the more detailed surveys of what works for self assessment and PAYE respectively. Rural or mixed location offices produce better accuracy for both compared to purely urban offices, but again the trend is for concentrating work in larger offices which inevitably tend to be more urban. For self assessment, the report finds specifically that smaller offices with fewer employees produce more accurate processing.
For both types of work, offices reporting fewer sick days report greater accuracy. Sick days are normally an indicator of dissatisfaction, as employees who can't face going into work phone in sick, so the high level of demoralisation that most advisers report, and which the NAO report finds in the focus groups, is not good news for accurate processing. Using fewer temporary staff produces fewer errors in PAYE offices, so one in four temporary staff is not going to help. Counter-intuitively, though, more leavers results in slightly higher accuracy rates.
Slimmer and Lean
So if so many of the indicators of greater accuracy are pointing towards the old-style local offices spread throughout the country, why is the department moving in the opposite direction? Presumably because of the need to make efficiency savings. The department is aiming to lose nearly 7,000 full time equivalent posts by 2011. One of the main ways they intend to do so is by implementing 'Lean processing'. Lean is defined as a system of looking at business processing from the customer's point of view, reducing duplication and streamlining systems. HMRC have high hopes for the Lean programme:
'The Department carried out three Lean pilot studies in three large processing offices. These demonstrated significant improvements in quality and efficiency: a 45% improvement in productivity for processing Self Assessment returns, 35% improvement in quality and a reduction from 36 to six days average lead time for processing information received.'
However, the results as the system has been rolled out over the past few years have been at best patchy. It has certainly caused problems with Revenue staff (see the article from Ian Lawrence, 'Lean Times' Taxation, 4 January 2007, page 8). The NAO report says this:
'The introduction of Lean in individual offices was accompanied by a slight or no improvement in monthly accuracy rates in Self Assessment. On PAYE it was accompanied by a significant reduction in monthly accuracy rates. This could be due to it taking time for Lean to become effective and an initial deterioration in performance while the new ways of working are being implemented. Allowing for a six month time lag in the analysis shows that the implementation of Lean ceases to be a significant factor in affecting PAYE accuracy rates, suggesting that the reduction in accuracy might be a short run issue during implementation. Thus the econometric analysis provides no firm conclusions about the impact of Lean at this stage of its implementation.'
Certainly the detailed figures for various measures under Lean processing do not seem to show significant improvements in most areas. There is perhaps a trend to greater accuracy in self-assessment processing, though it is not particularly marked. There is no obvious pattern to productivity or volume of items processed, except the ones which relate to the tax calendar — spikes in January and April.
Delays
There is, however, one key measure which shows a marked increase, although it does not get much of a comment in the report. Lead time, the number of days between receiving information and processing it, rose for self-assessment cases in Lean offices from 20 days in April 2006 to 50 days a year later in April 2007. The lead time for dealing with changes in employment rose from three days to nearly 14 in the same time period.
The lack of comment on this in the report is presumably because the increase in lead time is not an immediate cause of inaccuracy. However, when there are problems with accuracy, the lead time in dealing with post becomes very significant, as taxpayers and their agents spend increasing amounts of time in frustrating delays trying to correct the errors. It also casts a great shadow over the proposals to continue cutting staff; if post handling delays have already increased to such unacceptable levels, there should be a freeze on further staff cuts until the Revenue can at least handle its incoming post effectively.
Computerisation
Let's look at something that might offer better news — computerisation. When they surveyed agents to get some feedback on the problems experienced with HMRC processing, the NAO found that 18% of errors on PAYE case processing and 30% on self-assessment processing came from HMRC staff wrongly inputting information. Online filing may have had its teething problems, but the one obvious problem it entirely eliminates is the opportunity to transpose figures or miss entries out when information is rekeyed by HMRC staff from a printed return into their computer system. Increased take-up of online filing is bound to make significant gains in processing accuracy, because it removes the human element from the cases where it is not needed — straightforward processing of routine information.
By far the biggest cause of error reported by agents was errors in coding. Here HMRC have also come up with a computer-based solution called 'Coding Assistant'. This reads the information necessary to calculate a tax code from HMRC databases and produces a code automatically. Although it started to be implemented in 2005, it only became fully operational for 2006/07, yet the department estimate that it has cut errors in that year by a staggering 700,000. Yet it cost a negligible amount to create and implement, both in pure PAYE cases and also when it was extended to handling the collection of tax through PAYE on self assessment cases.
'Coding Assistant … cost in the region of £80,000 to develop. It also improved the way the PAYE system handles tax due from Self Assessment returns, so that more cases can be dealt with automatically. This initiative cost £1,800 and reduced the number of related tax effect errors by 57 per cent between 2004-05 and 2005-06 and by a further 5 per cent between 2005-06 and 2006-07. Such initiatives have also contributed to efficiency targets by reducing the workload on staff.'
It has to be said that those figures for cost look startlingly low, and one wonders on what costing basis they were calculated, but the benefit of reductions in errors of that magnitude would be worthwhile even if the cost were significantly higher.
Conclusions
So there is a real prospect of reductions in errors by increasing computerisation. When those gains are achieved, HMRC may also be able to reduce its staff numbers, because computers will be doing the work that humans used to have to do.
But those gains should not be anticipated, and staff should not have been cut before the new systems can cope. One of the key indicators of that is lead times; the steady increase in lead times shown by the NAO report, and only too well-known and commented on by practitioners, should have been a warning that something was wrong.
And finally, when the routine processing work has been automated, so the human cogs in the Lean production line are freed from the drudgery of rekeying information, what sort of workforce does HMRC need? Not, surely, one where staff are employed on temporary contracts with training that is limited to the particular task they are performing, but well-trained, permanent staff with a holistic knowledge of tax who can deal with tax advisers in practice on an equal basis.
