P&R’s new transparency is refreshing… but questions still remain over GST modelling

For months, Express has been asking P&R questions about its tax package and the modelling behind them.

Who built it? What data does it use? What assumptions have been made? Who has seen it?

And, perhaps most importantly, can anyone outside the small group working on it actually check the numbers?

For much of that time, the answers have been incomplete or significantly delayed.

P&R – and the States more generally – have had a habit of sending us statements that answer the parts of our questions it wants to, while ignoring the rest.

So credit where credit is due: P&R has now answered most of our latest set of questions, and Deputy Lindsay de Sausmarez deserves credit for that.

We did not get a detailed answer to absolutely everything, but we got substantially more information than we’ve had before – that is a welcome change.

It is also worth saying why transparency matters here, because this should not be reduced to another argument about whether GST – and the rest of P&R’s tax reform package – is good or bad.

Reasonable people can disagree about that.

They can disagree about whether Guernsey needs to raise tens of millions of pounds through a consumption tax, whether spending should be cut instead, whether income tax should rise, or whether there are better alternatives.

That is exactly what political debate is for.

But whatever your view on GST and the rest of the package, you should want the evidence behind the decision to be open to proper scrutiny.

Transparency is not something that should only matter when you agree with the policy – it is fundamental to a healthy democracy.

So, what have we actually learned?

The answers we’ve now received have actually made the situation clearer.

We know that not a single deputy has seen the underlying model – even P&R.

In several ways P&R has painted itself in a better light being more transparent.

Previously, it may have appeared to many that there was “one rule for us and another rule for them”, with P&R and the civil service allowed to see the modelling, while journalists and even other deputies weren’t allowed to.

We now know that isn’t true. That’s a good thing.

There was even a point when it was conceivable that a jailed former P&R member had seen the modelling, while other deputies hadn’t.

Again, we now know that wasn’t the case.

We know just five civil servants have had access to it, while Deloitte was given detailed outputs for its economic analysis.

We also know Dorey Financial Modelling has been brought in to validate the civil servants’ work and has access to the relevant census data because its officers are sworn census officers.

It’s reasonable to ask why a private firm can be trusted to see it when our elected deputies can’t.

It’s also reasonable to ask whether the process of picking Dorey was fair, as our previous questions about that haven’t been fully answered.

But P&R has also hinted at some potential reasons why Dorey may have been picked, given the tight timescales and their previous clearance.

It doesn’t mean there aren’t still questions, but it does answer some of them.

We know more about what is actually being fed into the model, including information on income, employment, benefits, pensions, property and housing.

And we know some fairly significant assumptions are being made, including that wage inflation is not incorporated, bad debt is within the margin of error and the effects of GST and reductions in income tax and social security on consumption are assumed to be neutral.

All of that is useful, but it also brings us back to the central question.

Why can’t deputies see it?

P&R’s answer is that the model contains detailed “pseudonymised” census information which could, in some circumstances, allow individuals or households to be identified.

That is a legitimate concern and nobody is suggesting confidential personal information should be made public.

But other jurisdictions have systems which allow accredited people to access sensitive data for legitimate purposes under strict safeguards.

The UK’s Office for National Statistics allows accredited researchers to access detailed census microdata for approved projects through secure systems, with strict controls over who can see it and what can leave the system. 

So protecting sensitive data clearly does not have to mean locking everyone except the model’s creators out.

So why can’t Guernsey do something similar?

If our elected deputies can be trusted to vote on tax policy affecting every household and business in Guernsey, why can’t they be trusted to inspect the model under appropriate confidentiality arrangements?

If there is a legal reason why they cannot be sworn in as census officers, or another mechanism cannot be found, P&R should explain it.

At the moment, the public is being asked to trust the work of just five civil servants – about 0.0075% of our population – and an external firm which has worked with the States before.

That puts a lot of pressure on those five human beings – and the private firm – to be right, because this is the biggest change to our tax system in a generation.

Who checks the checkers?

There are questions around the external validation too.

P&R says Dorey Financial Modelling is checking the modelling, rather than advising on whether GST is the right policy. Fine.

But we still do not know enough about how the firm was selected, why it was considered the appropriate choice, whether other firms were considered, what the procurement process looked like or how much the work is costing.

Those are reasonable questions, particularly when the person behind the firm has previously expressed support for GST and has worked with the States before.

Again, that is not an allegation of wrongdoing. It is a question of public confidence.

If the States wants people to accept that an external validation is genuinely independent, the obvious answer is to explain how that independence was protected and why this particular firm was chosen.

There are other gaps too.

We still cannot see the model, its formulae or its full methodology.

We’ve asked if the States could share the equations behind the model, even if they can’t share the data – they haven’t answered.

It is completely impossible that an equation or Excel or Power BI formula would allow a deputy or a journalist to identify anyone, so why the secrecy?

We have not been given a formal assessment of whether a version using dummy data could be released. We don’t even know if one was undertaken.

We don’t know whether an independent third party could reproduce the headline GST revenue estimates from the information already published – but it seems doubtful.

And, crucially, no deputy or truly independent organisation has been able to independently verify the calculations behind the projected GST revenue.

None of this means the GST figures are wrong.

It means we haven’t yet been given enough information to independently test them.

Honest people make mistakes

Nobody reasonable is suggesting the civil servants who built the model are trying to ‘rig’ the numbers.

Honest, competent people make mistakes, make flawed assumptions and can develop unconscious biases without ever realising it.

That’s precisely why independent checking and peer review exist.

We don’t know the qualifications or previous experience of the five civil servants who have had access to the model.

They may be extremely well qualified and have extensive experience of this kind of modelling in different countries. But we don’t know.

We do know they work for the same organisation and presumably know and work with one another.

That doesn’t mean the maths is wrong, but it is a good reason to have somebody genuinely independent try to challenge it.

Dorey may be perfectly capable of doing that, but we have not seen evidence that it has previously carried out this particular kind of tax-revenue modelling for another country or jurisdiction.

Again, that doesn’t mean the firm isn’t qualified – it means we should be told why P&R considers it the right independent checker.

Deloitte, on the other hand, has established international expertise in economic analysis, so why couldn’t appropriately qualified people from Deloitte be sworn in as census officers and given controlled access to the model?

The point isn’t that anyone is dishonest, it’s that trust is not a substitute for independent scrutiny.

Transparency is a start, not the finish line

P&R has taken a step towards greater transparency, and it should be recognised.

It would be easy to focus only on what remains unanswered and ignore the fact that we have now got substantially more information than we did before.

But transparency cannot ultimately mean asking people to trust the government because it has provided some of the answers.

The whole point of scrutiny is that we should be able to test them.

If the model is robust and its assumptions are reasonable, proper scrutiny should strengthen the case for it.

If there are weaknesses, deputies and the public deserve to know before a decision is made – and P&R should welcome it.

So yes, credit to P&R for finally showing us some of the “warts and all” transparency the head of the civil service has called for.

Now it needs to let that transparency stand up to scrutiny.