On the UK’s Fiscal Black Hole

This Thursday, the UK government will introduce a budget with significant tax increases and spending cuts. The UK press are largely reporting this as necessary to close a “black hole” in the UK’s public finances. It is frequently suggested that much of this black hole is a direct result of the fiscal profligacy proposed by the short-lived Truss government. The reporting generally suggests that a contractionary budget is a matter of common sense—books need to be balanced and after September’s chaotic events, the new government needs to demonstrate to financial markets that it is serious about fiscal discipline.

I disagree with this assessment. In fact, with the UK economy clearly tipping into recession, a large fiscal contraction is not common sense. It is certainly not what textbook macroeconomics would call for. Here, I discuss a few different aspects of the UK’s fiscal black hole, question whether austerity is necessary and point to one concrete alternative to spending cuts and tax increases as a way to reduce much of the projected budget deficit over the next few years.

It’s a long post so let me tell you how it is structured.

First, I describe what the UK’s fiscal “black hole” actually is and how it results from a series of essentially arbitrary fiscal targets rather than from a genuine crisis that must be urgently addressed.

Second, I argue against the widely promoted idea that the Truss-Kwarteng interlude showed that financial markets need to see a big fiscal adjustment now and that there would be further disruptions if we don’t see a harsh austerity programme on Thursday.

Third, I discuss two technical issues relating to the Bank of England which have had a significant influence on the current fiscal policy situation. The first is a change in the treatment of the Bank of England’s balance sheet in the government’s fiscal targets which has had a big influence on why there has been such a swift move towards austerity. The second is the Bank’s policy on paying interest on reserves to commercial banks which provides an example of an alternative policy available to the tax increases and spending cuts being implemented.

The UK's Public Finances?

The UK’s Public Finances?

What is the UK’s Fiscal Black Hole?

The so-called black hole stems from set of fiscal targets that the UK government has set itself and which are measured and monitored by the Office for Budget Responsibility (OBR).  As of now, the principal targets are that the current budget (excluding capital spending) be balanced by the third year of the OBR’s forecast and that a measure of net debt as a share of GDP be declining in the third year.

Of these two targets, the net debt target is likely to be the more binding now because it includes debt generated by capital spending and because the OBR’s forecast is going to include a decline in GDP due to the upcoming recession, thus raising the debt ratio. The Resolution Foundation have a detailed pre-budget presentation which points to the net debt rule as the key likely driver of the size of the consolidation.

The most recent forecasts from the OBR for the net debt to GDP ratio are from March. Their March forecasts for various measures of the net debt to GDP ratios are shown below. Since late 2021, the measure chosen by the government for its target has been “public sector net debt excluding Bank of England”, the green line below. You can see that as of March, the projection for this series was that it would be flat to slightly declining in the coming years, just about in compliance with the rule.

The OBR's Forecasts for Net Debt to GDP Ratios

The OBR’s Forecasts for Net Debt to GDP Ratios

Since March, several factors have worsened the fiscal outlook. The Bank of England has raised interest rates, increasing the cost of servicing public debt. The projected recession will reduce forecasted tax revenues and raise welfare spending and the government’s energy package will be expensive. Given all this, it is expected that without adjustments in the upcoming budget, the target measure of net debt to GDP will move upwards over the coming three years. This is the “black hole”: It is the amount of spending cuts or tax increases required to get this net debt to GDP measure falling in the third year of the forecast instead of increasing.

To many, the idea of having rules to ensure sound fiscal policy may seem sensible. However, there is little agreement among economists about the form such rules should take or indeed whether fiscal rules are a good idea at all. The UK has had various fiscal rules since 1997 and it is not clear they have been particularly helpful. A 2021 survey of macroeconomists by the LSE Centre for Macroeconomics showed about half believed fiscal rules had harmed the UK’s economic performance and nearly one-third favoured scrapping fiscal rules altogether.

One problem is that the specific rules being followed are essentially arbitrary, either relying on special “magic numbers” for debt or deficit levels (3% and 60% ring a bell for anyone?) or specific horizons over which certain objectives should be met. The UK’s current fiscal rules are new, having been announced in the 2021 budget. These new rules changed the debt target from total public sector net debt (the blue line above) to the measure excluding the Bank of England, which showed a lower net debt level. The rules were also changed from requiring a reduction in the net debt as a share of GDP over the next year to having it fall by the third year of the OBR’s forecasts. On their own, these were both sensible changes but (as I will discuss below) the change in net debt definition turned out to have unintended consequences.

So that’s the black hole. It is not in any way a pressing matter in which the UK is on the brink of “running out of money”. It is a consequence of an arbitrary rule the government has imposed upon itself with its specific form having been set just over a year ago. Had the UK government set a longer time horizon for the net debt ratio to be falling or used a different definition of the net debt ratio, the black hole would be smaller or perhaps non-existent.

Rather than sensible macroeconomics, deciding to impose a large fiscal contraction right as the economy is heading into recession runs counter to conventional macroeconomic wisdom. While the Truss-Kwarteng “let it rip now” approach to fiscal policy was clearly badly timed given current inflationary conditions, the alternative of swinging towards a big fiscal contraction to be implemented as the economy enters recession is also inappropriate. UK fiscal policy should be neutral for now but the normal approach of allowing “automatic stabilisers” to accommodate increases in welfare spending and reductions in tax revenues should operate once the economy goes into recession.

As a sign of how this austerity is badly timed, it is being widely reported that the OBR will calculate the contractionary effects on GDP of the upcoming fiscal austerity, which will worsen the projected recession and further increase the net debt to GDP ratio thus widening the so-called black hole. A chunk of this hole is being dug by the austerity itself.


Bond Market Vigilantes?

But What About Markets?

What about the idea that austerity is needed to satisfy financial markets, the so-called bond market vigilantes? For many, the moral of the Truss-Kwarteng experiment was that markets require fiscal discipline and a failure to balance budgets raises the cost of borrowing and further worsens the public finances. I think this is a misinterpretation of what happened in September.

Without doubt, September’s financial market events in the UK were extraordinary but much of the turmoil related to forced selling of gilts by pension funds. While dramatic, these events were due to short-term “market plumbing” issues and were resolved fairly quickly.  Beyond these short-term stresses, the key substantive macroeconomic problem was that Truss and Kwarteng introduced a set of fiscal policies that would have stimulated the economy right at the time when the Bank of England was acting to cool the economy to bring inflation down.

For some, there was also a concern that the Truss government would undermine the independence of the Bank of England and thus that higher inflation would be implicitly tolerated over the medium-term. These concerns will have required additional inflation-related compensation for investors in UK government bonds.

With Truss deposed and most (but not all) of Kwarteng’s tax cuts reversed, investors are no longer concerned that UK fiscal policy will exacerbate the current inflation problem or that the government will undermine the Bank of England’s independence. UK bond yields are up since the summer but not by much more than in other countries, all of which are also facing a persistent inflation problem and going through a monetary tightening.

Are the markets now baying for a return of austerity? I don’t think so. There was no sense prior to the Kwarteng fiscal measures that financial markets were requiring a large fiscal contraction to be willing to hold the UK’s debt and those measures have now been reversed. Also, there is nothing especially high or worrisome about current or projected levels of UK public debt. The UK gross government debt to GDP ratio for this year is about 100%, close to the average for the euro area and well below the 112% in countries like France, where nobody is claiming that financial markets are demanding austerity.

Of course, if Jeremy Hunt were to announce a much smaller austerity package than expected on Thursday, perhaps by adopting a new looser set of fiscal rules, it is possible that markets could react badly and raise long-term bond yields. However, I think a neutral budget that neither adds to the current inflation nor worsens the upcoming recession would be accepted by markets as a reasonable fiscal Ying to the Bank of England’s contractionary Yang.

The Strange Tale of the Bank of England’s Phantom Debt

One of the stranger aspects of the current turn towards austerity is the influence of a technical change in the chosen debt target. In the EU, debt targets are framed in terms of gross government debt and assets owned by governments are not considered. The EU rules also do not incorporate central banks, keeping their assets and liabilities “off balance sheet.”  Until last year, the UK government’s chosen debt target differed from the EU’s approach both by subtracting off some assets and thus being a measure of “net debt” and also by incorporating the balance sheet of Bank of England.

This approach could have produced sensible net debt measures but the actual implementation did not. The Bank of England’s balance sheet shows its assets and liabilities. Its principal liabilities are the reserve accounts that commercial banks maintain with the Bank. The Bank acquired its QE portfolio and made loans to commercial banks via making credits to these accounts. These reserve accounts are liabilities for the Bank because it has a policy of paying interest on reserves. The Bank publishes its balance sheet weekly but a more accessible version is available in its 2022 annual report, which shows its February 2022 balance sheet.

The balance sheet shows that the Bank’s assets slightly exceed its liabilities. In fact, its true net position is better than this. Its balance sheet includes £86 billion in sterling bank notes in circulation that both the Bank and OBR count as liabilities. Since these are perpetual zero-interest notes, they have no cost to the Bank and so the Bank effectively has assets that well exceed its genuine liabilities. The key point here is The Bank of England has no net debt.

Despite this, the OBR made two decisions when constructing its net debt measure that meant incorporating the Bank of England’s balance sheet produced higher rather than lower net debt figures. First, the OBR decided to not include the assets the Bank acquired under its Term Funding Scheme (TFS), which loaned money to banks, while at the same time OBR decided to include the reserves that were created by the TFS loans as liabilities. OBR did this because they decided to only include liquid assets when “netting off” assets against liabilities and they decided the TFS loans were not liquid.

I don’t know why OBR considered the TFS loans to not be liquid. The Bank would never need to sell these assets so it’s an “angels on pins” discussion anyway but the loans are fully collateralised, so I’m sure the Bank could sell them fairly easily if it ever decided to do so. Anyway, in reality, the TFS scheme did not have any negative effect on the net debt of the Bank of England or the wider public sector but the OBR decided to measure it as though it did.

Second, the OBR measures the Bank’s gilt holdings acquired via QE at their face value while measuring the reserve liabilities created at the market value of the bonds purchased. Bond yields and prices move inversely, so the decline in yields during the Bank’s QE programme meant they bought bonds for market prices that were above the face value. For almost all of the post-QE purchases, the yields on the gilts purchased were higher than the interest on reserves, meaning it made profits on these investments but the OBR decided to make these transactions look like they made the Bank on net worse off.

In terms of the mechanics of how QE affected the public finances, the Bank passed most of the interest payments on its gilts back to the government, so effectively the QE programme refinanced UK government debt at the low interest rate paid on reserves and this saved the public sector money. Again, however, the OBR accounting made it look like a programme that increased UK net indebtedness.

Ultimately, central banks that print their own currency and can create money at the touch of a button are a source of revenue for governments. If your accounting says this part of the public sector has a negative net asset value, you’re doing something wrong.

Taken together, the OBR’s two dubious measurement decisions meant the incorporation of the Bank of England added a large amount of “phantom debt” to the OBR’s net debt measure. For 2021/22, the total net debt is estimated to be 95.5% of GDP while the net debt excluding the Bank of England is 82.5% of GDP. But (and this bit is important ….) the two key drivers of the phantom net debt—the TFS and the QE holdings—are projected to wind down in the coming years, so the phantom net debt declines from 13% of GDP in 2021/22 to only 3% in 2026/27.

Given how bad the OBR’s methods for incorporating the Bank of England are, the Treasury were clearly correct to switch from “net debt” to “net debt excluding the Bank of England”.  The OBR itself has effectively admitted the new measure is better, saying in their March report that this change “removes the uneven effects across years caused in particular by the TFS”.

Obligatory Blog Post Meme

Obligatory Blog Post Meme

Now one might imagine that changing to this new measure would have caused there to be less pressure to introduce austerity. The new measure both makes more sense and shows lower net debt levels. However, as Rob Calvert Jump and Jo Michell have noted, this is not how things work in the world of UK fiscal rules. The new measures have increased the need for austerity.

Imagine the scenes at the Treasury after the new, better, net debt measure has been adopted.

Normal Person: “Hey look we’ve got a new measure of net debt that excludes all the phantom debt previously included by the OBR. Surely this means we don’t have to worry quite so much about debt now?”

Fiscal Rules Guy: “Well no, actually we need to worry more about debt than before. We need more spending cuts and tax increases

Normal Person: “Why?

Fiscal Rules Guy: “Well the new measure of net debt is lower but it doesn’t fall anymore over the next few years like the old measure did, so we need austerity.

Normal Person: “But the old measure only fell because the non-existent phantom debt declined. Are we really introducing austerity because we stopped counting the non-existent debt?

Fiscal Rules Guy: “Rules are rules dear fellow.

Suffice to say 10% of GDP is a lot of spending to cut or taxes to raise over the next few years and this measurement change as a rationale for a bunch of extra austerity doesn’t fit well with common sense.

Interest on Reserves

The other way the Bank of England impacts the budget is that the Bank pays interest on reserves at its “Bank rate.”  With the Bank raising interest rates, these additional payments to commercial banks will reduce the profits of the Bank of England. As a result, the Bank will go from passing profits on to the Exchequer to not doing so and, if the Bank makes losses, it may be years before it returns to passing on a dividend to the Treasury. Effectively, the Bank’s payments of interest on reserves represent a one-for-one fiscal cost for the public sector.

The payment of interest on reserves is a modern phenomenon. The Bank of England started paying interest on reserves in 2006, so there is no long tradition of banks receiving compensation on their reserves. This changed with the introduction of QE by the world’s central banks.

Prior to QE, central banks had raised interest rates by engineering a shortage of reserves, so that banks were willing to pay high interest rates to borrow funds from each other. After QE, there was no shortage of reserves, so central banks had to find another way to raise interest rates, if this was what monetary policy called for. They did this by paying interest on reserves. A bank that received a 4% interest rate just for keeping money with the central bank would not be willing to make loans at lower rates than this, so the interest rate paid on reserves became the floor for market interest rates.

When interest on reserves was first introduced, there weren’t many discussions of its fiscal impact these are now significant. For example, as of November 2nd, UK banks had £950 billion in reserve balances.  If the Bank of England were to compensate those reserves at a 4% interest rate, this would imply interest payments of £38 billion per year.

Are these payments required to implement the higher interest rates the Bank wants for its monetary policy? No. It is widely understood in central banking that what is required is only that the marginal reserves in the system are paid the interest on reserves. The government doesn’t need to compensate all the reserves a bank holds. It just requires that the marginal pound of reserves that a bank holds would receive this compensation.  This “tiering” approach has been used by both the Bank of Japan and the ECB, albeit in the context of negative interest rates, so that only some of the reserves that banks held with them were subject to the negative rates. But the logic that it was the marginal reserves that mattered for market rates worked perfectly.

The idea that central banks do not need to compensate all reserves may not have been discussed much in the context of the “black hole” but it is hardly a radical and dangerous idea. Former Bank of England Deputy Governor, Sir Paul Tucker, a pillar of the international central banking establishment, has proposed tiering in a recent paper, arguing the fiscal savings could be £30 billion and £45 billion over the next two budgetary years. These are large figures and would account for a big fraction of the tax increases and spending cuts being considered.

Sir Paul Tucker: Dangerous Radical?

Sir Paul Tucker: Dangerous Radical?

Some argue that failing to pay interest on reserves amounts to a tax on the banking sector. I don’t think this is correct. Taxes are deducted from income earned or wealth held and this policy does neither of these things. Banks holding an asset that doesn’t receive any compensation may reduce their profits but central banks have long adopted policies that either boost or reduce bank profits and these have never been labelled fiscal policies.  At the end of the day, however, if the failure to compensate all reserves means bank shareholders lose money or depositors earn a lower interest rate on deposits, then there are strong arguments that these outcomes would represent a fairer way to stabilise public finances than the other options being considered.

Rather than argue that introducing tiering would be the Bank of England stepping into fiscal policy, I would describe this issue in a different way. Interest on reserves is a monetary policy. It was introduced to allow central banks to set specific levels of market interest rates and thus meet their inflation target goals. However, central banks are public bodies and there is no public policy case for central banks to spend public money if it does not actually help the central bank meet its specified objectives. Because compensating only reserves above a specific level works to allow central banks to meet their monetary policy objectives, there is no monetary policy rationale for compensating all reserves. In other words, deciding to pay interest on all reserves is the Bank of England stepping into fiscal policy, rather than sticking to its monetary policy mandate.

There Are Alternatives

It has been a whirlwind few weeks and there are several compelling moral narratives that have convinced many that the UK needs to implement an extremely harsh budget. Covid and energy packages surely must be paid for. The sins of Truss and Kwarteng’s profligacy surely must be atoned for. Sensible fiscal rules surely must be obeyed. The costs of Brexit must surely be finally faced up to. And the Bank of England’s independence surely rules out any discussion of its policies being altered because of their fiscal implications.

The reality, however, is that the UK’s public debt position is in the middle of the international pack and there are almost no discussions in other countries with similar debt levels of the need for a new round of austerity. And if the UK government wants to reduce its fiscal deficit, it is reasonable to ask whether payments that benefit bank shareholders and depositors should be protected while cuts are made to public services that are already threadbare.


Coda: Since I wrote this yesterday, I have read these (much shorter) pieces by Giles Wilkes and Duncan Weldon that cover some similar ground. (Hat tip to Stephen Bush’s excellent Inside Politics newsletter.) I think I agree more with Duncan than Giles but both are worth a read if you’ve made it this far.

Are Central Banks Storing Up a Future Fiscal Problem?

In the past week, I’ve come across two different pieces (one by the BIS and one by the UK’s Office for Budget Responsibility) warning that the maturity of public debt in advanced economies has been shortening and this could have fiscal implications during a recovery.

This warning might seem surprising since the Covid crisis and the low long-term interest rates of recent years have seen governments issue lots of long-term debt. Indeed, conventional measures of average public debt maturity are increasing. In the euro area, the average maturity of public debt has increased from about six years a decade ago to about 8 years now.  In the UK, the average debt maturity is a whopping 15 years.


So what are the BIS and OBR talking about? They come up with different numbers by calculating an average maturity that includes reserve balances held by commercial banks at central banks. These reserve balances are generally compensated by central banks using their short-term policy rate and BIS and OBR are treating them as overnight debt when doing their average maturity calculations.

The BIS say about central bank asset purchases

Considering the consolidated public sector balance sheet, these operations retire long-term government debt from the market and replace it with overnight debt – interest bearing central bank reserves. Indeed, despite the general tendency for governments to issue at longer maturities, central bank purchases have shortened the effective maturity of public debt. Where central banks have used such purchases more extensively, some 15–45% of public debt in the large AE jurisdictions is in effect overnight.

Similarly, the OBR’s calculations show the median maturity of UK debt falls from 11 years to 4 years if you include reserve balances as an overnight liability. As an aside, I’m not sure median maturity is particularly interesting. If half my debt is due in 2 years, then that’s the median maturity. But it does matter whether the other half is due in 3 years or 100 years. I suspect median maturity has been chosen here for dramatic effect.

There are two reasons to worry about a shortening maturity for government debt. First, short maturities raise the probability of some kind of funding crisis. If the debt is all due relatively soon and investors decide not to roll over their funding, then there could be a debt crisis requiring a sovereign default and\or a large fiscal adjustment.

Second, and less dramatically, if the economy recovers and central banks raise interest rates, then the short-term debt will re-price at these higher interest rates and raise interest costs for the government.

Should we be worried? I don’t think so.

Rollover Concerns

I’ve always been uncomfortable with the labelling of central bank reserves as liabilities. Yes, in recent years, central banks have chosen to pay interest on them but they get to choose that interest rate and it can be zero if they want it to be. These are not debts that look like a normal person’s debts. But whatever about the labelling, I don’t think you can legitimately include central bank reserves in an average maturity calculation designed to measure potential debt rollover pressures.

These reserves are created by central banks to pay for asset purchases and there is nothing the banking system can do to get rid of them. An individual bank might be unhappy with having so much on reserve with the central bank but if they try to get rid of them by buying a security or extending a loan, their reserves just end up with another bank. As Ben Bernanke said, they’re like a hot potato.

Given this, there is no direct comparison between a sovereign bond maturing and an overnight reserve balance “maturing”.  The owner of the maturing sovereign bond has to be paid and the money must be found from somewhere to pay it off – either the government’s existing cash balances or via new borrowing. In contrast, the banking system cannot demand that its reserves be “paid up.”  In any case, the reserves already essentially are money and they can be turned into cash on demand.

So the risk of funding crises for most advanced economies is minimal and the real story of recent years is that governments around the world, but particularly in the UK, have locked in lots of very long-term low-interest debt.

Rising Interest Costs

This is, on the face of it, a more legitimate concern. Central banks used to raise interest rates by generating an artificial shortage of reserve balances, thus increasing the market interest rate paid to borrow these balances. In an age where reserves are in huge excess supply, this method doesn’t work anymore. Instead, central banks raise market interest rates by increasing the interest rate they pay on reserve balances to commercial banks and this rate then acts as a floor on market rates.

If central banks end up paying high interest rates on their reserve balances while holding a portfolio of long-term fixed rate bonds, then in theory, the central bank could start to make losses as interest paid exceeds interest earned. This could reduce their remittances back to central government and thus have fiscal effects. Taken to an extreme, there could be a scenario where there needs to be payments in the other direction because the central bank has so little money left it can’t fulfil its interest on reserves policy.

Thankfully, there is no need to worry about this stuff. Crucially, we know now that central banks can affect market interest rates without compensating all reserve balances. The Bank of Japan and the ECB operate a tiering system in which the interest rate on reserve balances is zero up until a bank’s reserves reach a certain level and then an interest rate is applied on those remaining balances. This keeps the marginal cost of funds in the economy at the policy rate while detaching the policy rate from the average cost of remunerating reserves.  Currently, these central banks have a negative policy rate but tiering could be operated in future to raise interest rates without having much impact on the central bank’s profits.

Even if central banks decided to continue paying the policy rate on all reserve balances, there are reasons not to be too concerned. While reduced central bank profits during a monetary tightening are possible, outright losses are less likely. A 2015 Federal Reserve study simulated a sharp monetary tightening and concluded there may need to be a few years when the Fed reduced remittances to the Treasury to zero but the overall impact was relatively small. There are also other assets held by central banks that will provide an increased return as interest rates rise. For example, the interest payments on reserves created by the Eurosystem to make loans to banks are always more than offset by the higher interest payments charged on these loans.

So you could argue that technically neither the BIS or OBR are outright wrong about the maturity of “consolidated public liabilities” but there are good reasons to not be concerned about the points they raise. And claims about “ticking debt bombs” from the usual austerity cheerleaders should be taken with a massive dollop of salt.

The UK’s £39 Billion Brexit Bill

A few points on the £39 billion payment which the UK government has agreed to pay the EU as part of the withdrawal agreement.

This payment is regularly raised by Brexiters as a key negotiating issue. It is often claimed the £39 billion saved by refusing to make this payment will allow the UK to spent lots of money on important priorities e.g. Priti Patel sayswe would also not need to hand £39 billion over to the EU, giving the Government resources to support economic growth, job creation and new trade opportunities.” It is also claimed the EU are very scared the payment won’t be made and will thus make last concessions to secure the money e.g. Daniel Moylan at BrexitCentral reckonsIt is possible that the EU could give in, agreeing the wholesale removal of the Irish backstop from the text in exchange for the money on which they have so desperately counted.”

The reality, however, is that both sides of this argument – the benefit to the UK and the costs to the EU – tend to be wildly over-stated by Brexiteers. Here is a useful table from a House of Commons Library explainer on the topic, breaking down the different elements of the £39 billion.


The table shows that the £38.7 billion bill includes £16.3 billion in contributions to the EU’s multi-annual budget during the transition period. During this period, the UK will continue to receive funding from EU programmes.The House of Commons Library report estimates the UK will receive about £8 billion of this funding back during 2019-20 in the form of payments to farmers, structural funds and funding for research (see page 8). These payments would not occur if there is no deal. This means the financial “benefit” to walking away from the withdrawal agreement would really be £31 billion, albeit at a cost of the UK being seen to have reneged on financial commitments it made to its EU partners.

Second, this net payment of £31 billion would be spread over time.  Scrapping the additional net contributions to the EU budget would amount to saving of £4 billion per year this year and next.  The rest of the payment is spread over time, with the bulk of it, £19.8 billion, being paid over 2021-28 and a smaller amount of £2.6 billion being paid over 2021-64. That works out to be a benefit of £2.5 billion per year over 2021-28 and £56 million per year thereafter until 2064.

How much will the UK government be able to do with this money? UK GDP was £2 trillion pounds in 2017. A trillion is one thousand billion, meaning the £4 billion savings this year and next would be one fifth of one percent of UK GDP and the £2.5 billion per year savings over 2021-28 would be about one eighth of one percent of UK GDP.

So while £39 billion may seem like a huge figure when quoted without context, the reality is that this will do very little to boost the spending power of the UK government. In fact, these numbers are well below the kinds of figure by which UK budgets often overshoot or undershoot doing to various random events without anyone paying much attention e.g. the OBR’s March 2017 forecast of the budget deficit over 2017/18 was £17 billion too high (see page 35). The dividend from not paying the Brexit bill will effectively be rounding error in the public finances.

And what about the EU27? The £2.5 billion net payment per year the EU would be losing over 2021-28 equates to €2.8 billion euro per year. EU27 GDP is estimated to be €14.538 trillion last year. That’s €14,538 billion. So this loss would equate to 0.02 percent of EU27 GDP per year for eight years. That’s one fiftieth of one percent of GDP. The idea that this tiny amount is being “desperately counted on” by the EU is laughable and anyone relying on the EU to fold because the UK won’t pay its “Brexit bill” will be sorely disappointed.

The bottom line on this issue is that, viewed from a macroeconomic perspective, the Brexit bill figures are tiny and people really should not believe that this bill represents an important part of the costs or benefits of Brexit for either the UK or the EU.

There is a wider issue here in relation to explaining macroeconomic issues to the public. Any figure larger than a few million seems to ordinary people to be “a lot of money”. So “£350 million pounds per week for the NHS” or “£39 billion to spend on whatever we want” seem like huge windfalls to a large fraction of the public who, let’s be fair, can’t be expected to know the actual totals for GDP or public expenditure. It’s thus important for those contributing to public debate to put raw macroeconomic numbers in an appropriate context.

The EU’s Backstop is a Great Opportunity for Northern Ireland

The past week has been the most fraught yet in the Brexit negotiations.  The EU and UK have not agreed on how the “Irish backstop” proposed in December should operate. The UK government and the DUP are unhappy that the EU believes the “backstop” arrangement should only apply to Northern Ireland.  The EU backstop would essentially keep Northern Ireland (but explicitly not the rest of the UK) in the EU customs union and single market unless other arrangements are agreed that would also rule out the need for a hard Irish border.

In Westminster and Northern Ireland, there is a lot of concern about the EU’s proposal, with many viewing it as implying a “border in the Irish sea” and Theresa May arguing that it would “threaten the constitutional integrity of the UK” and that “no UK prime minster could ever agree to it”.  In Northern Ireland, unionists have argued this approach is inconsistent with UK’s commitment in the December agreement that there would be “unfettered access for Northern Ireland’s businesses to the whole of the United Kingdom internal market”.  Arlene Foster has repeatedly insisted that this arrangement would be “catastrophic” for the Northern Ireland economy and, in a notable upping of the ante, said on Friday that the EU’s approach would amount to Northern Ireland being “annexed” from the UK.

I believe these concerns are fundamentally misplaced. Rather than being threatened economically, Northern Ireland would gain from the implementation of the EU’s backstop. To understand why, let’s look at how the backstop would work in practice.

Unfettered Access

Let’s start with the most obvious objection to the EU’s backstop: the idea that a “sea border” would make it harder for Northern Irish firms to sell into Great Britain. The DUP’s Arlene Foster has emphasised this repeatedly in recent months. Consider, for example, this statement from Foster in March

“What was there, in the [EU draft agreement] legal text, is around creating a border down the Irish Sea,” she said.

“And of course that’s not something we in Northern Ireland could allow from a constitutional point of view, but also in terms of economics, it would be catastrophic to have a border down the Irish Sea.”

“56% of our goods from Northern Ireland go to Great Britain, so it is incredibly important that that border does not exist.

This probably sounds like a reasonable concern but it is without foundation.  Firms in Northern Ireland will remain within the UK, so there will not be and cannot legally be customs checks for goods produced in Northern Ireland when travelling to the rest of the UK.  In addition, the UK itself promised in the December agreement that unfettered access would be maintained for Northern Irish firms and that “no new regulatory barriers” would affect these firms. Together, these points mean Northern Irish firms will have unfettered access to the rest of the UK under the EU backstop.

If you’re surprised by this and perhaps think I’m making it up, you might not know the UK civil service has already worked out how this backstop would work and they have said there would be unfettered access for Northern Ireland firms for the rest of the UK market. The document describing how this would work was part of a series of UK civil service documents leaked to politicians in the European Parliament and was discussed in various press stories in May.

This “customs channel” proposal implies no land border checks on the island of Ireland but some checks at the small number of ports in Northern Ireland that transport goods to Great Britain. Goods would either go through a “green channel” with no checks or a “red channel” which has checks. Crucially, goods from firms in Northern Ireland would go through the green channel.   Goods coming from the Republic of Ireland to the Great Britain via Northern Ireland’s ports would probably have to go through the red channel if the UK required customs or regulatory checks on goods from the EU.

Viewed this way, the “border in the Irish sea” terminology is misleading because it will not affect firms from Northern Ireland. A better terminology would be “enforcing the land border at Northern Ireland’s ports.”  With a small number of ports in Northern Ireland, all of whom are already checking shipping documentation of some sort, the implications for trade frictions for Irish firms would be far less severe than the alternative of enforcing customs and regulatory checks on a meandering 310 mile border with about 200 different crossing points.

Over time, various issues would need to be sorted out about how this arrangement would evolve. These issues are not trivial but they would be would not be too difficult to work out between the EU and UK.  A few examples.

  • Most likely, Northern Ireland would be given a “special economic zone” status, so that the free movement of goods from its firms to both the rest of the UK and EU is agreed and accepted internationally by all members of the WTO.
  • Assurances may be required that the UK will discourage tariff-hopping via “name plate” firms setting up in Northern Ireland just to label products as a way to avoid tariffs that the UK may charge on EU goods (though these tariffs should be minimal if the EU and UK agree a sensible free trade deal).
  • Some goods moving from Great Britain to Northern Ireland may need to be checked to make sure they meet EU regulatory requirements. Goods that are heading on towards the Republic of Ireland would also need to have EU customs procedures applied.
  • The UK has promised there would be no new regulatory barriers affecting Northern Irish firms accessing the market in Great Britain. If the UK starts passing new product regulations that deviate from the EU’s, any new regulatory bill passing parliament could include a clause stating that Northern Ireland’s products (produced according to EU rules) are also allowed to be sold throughout the UK, i.e. that there is a form of regulatory equivalence. Ultimately, it is up to the UK government to honour its promise to allow Northern Irish firms unfettered access to markets in Great Britain and there should be little difficulty in implementing this.

Economic Impact on Northern Ireland

So the EU backstop isn’t going to turn Northern Ireland into an economic dystopia.  It is far more likely to have a positive economic effect. Northern Ireland would become the only place where firms could export freely to both the EU and the UK.  One could easily imagine Northern Ireland obtaining new foreign direct investment because of this unique selling point. Business leaders in other regions of the UK (for example Scotland) would view this kind of special status as a great opportunity if they could attain it but the EU has made clear that this can only apply to Northern Ireland.

Northern Ireland’s agriculture and food sector is also likely to do better under this regime than under any alternative approach.  If the UK decides to pursue trade deals that allow cheap food from outside the EU into the UK, then Northern Irish farmers would have to compete with these imports in the Great Britain market (most likely in an environment with less generous state-funded agricultural subsidies).  However, regulatory alignment with the EU would mean produce like chlorinated chicken and hormone-injected beef could not be sold in Northern Ireland, providing some protection for local producers.  Northern Irish farmers would still also have full access to the European Union market without facing tariffs or quotas, which would provide some opportunity to diversify their sales.

Most importantly, the EU’s backstop keeps the land border open and allows the all-island Irish economy to continue to operate freely. Contrary to Boris Johnson’s  uninformed sneering, the reality is that more firms in Northern Ireland sell goods to the Republic than to Great Britain and integrated all-Ireland supply chains are of huge importance to many firms, North and South.  Northern Ireland would keep this important element of its economy, without losing anything.

Constitutional Integrity? Annexation?

But what about the constitutional integrity of the United Kingdom which is supposedly a great concern to Theresa May? Well the UK’s constitutional integrity is a complicated thing, not least because it doesn’t actually have a constitution.

The reality is the UK has a patchwork of different governance arrangement across its regions.  Northern Ireland, in particular, already differs sharply from the rest of the UK in lots of ways, including its form of government, rules on gay marriage and abortion and the plethora of ways in which the Good Friday agreement has introduced North-South co-operation.  In truth, the DUP’s desire for Northern Ireland to have a different corporate tax rate from the rest of the UK is probably a more substantive difference in economic policy than anything new that would emerge from the proposed backstop.

Is this Northern Ireland getting annexed by the EU or Republic? Clearly not. Northern Ireland would still send MPs to Westminster. Its people would still pay UK tax, hold British passports (if they wish), have access to the NHS and the UK social welfare system, and be subject to UK laws in most areas. The people of Northern Ireland would probably barely notice their new status. By contrast, they would certainly notice the return of a hard border, which is the most likely alternative option if the EU’s backstop offer is rejected.

The Politics: Still Time (Just About) for a Broader Discussion

It now looks like nothing will be settled between the UK and EU until the Autumn. This still leaves some time for an informed debate in Northern Ireland and Westminster about the consequences of the EU backstop proposal. For a number of reasons, this debate has not taken place so far and there is a general lack of understanding about how the EU backstop would work. These reasons include:

  • Northern Ireland does not have an assembly operating. The assembly would certainly have held extensive committee meetings to discuss the various options and this would have helped to debunk the scaremongering about the so-called “border in the sea”.
  • Many in the UK government probably know the EU backstop would work well for Northern Ireland. Under other circumstances, they could give assurances to everyone in region that the backstop would work well for them – it is not the EU’s job to explain how unfettered access to the UK market could be maintained. However, the Conservatives are reliant on a hard-line unionist party for their majority at Westminster and do not want to upset the party keeping them in power by asking them to consider a proposal they are so uncomfortable with.
  • The last UK general election meant the DUP and Sinn Fein were the only parties in Nothern Ireland with MPs and Sinn Fein do not attend parliament. This has left the Westminster debate without nationalist or non-sectarian voices (such as the Alliance Party) who could have argued for the EU’s backstop. It is also seems likely that the current DUP leadership has handled these issues in a more aggressive manner than, for example, Peter Robinson would have (see for example Robinson’s speech last week imploring leaders on both sides of the community to compromise to get the Assembly up and running again.)
  • Finally, the Irish government appears to have outsourced most of the communication on Brexit to Michel Barnier (who has a lot on his hands) rather than working hard to publicly explain the benefits of the EU backstop to all sides of Northern Ireland’s community.

Northern Ireland may not have a functioning assembly but the MLAs that were elected last year must surely understand they have a political responsibility for what happens next.  Even if the assembly does not formally convene again this year, it must be possible for the MLAs to meet to debate the options and perhaps hold a “consultative vote” on the EU backstop. The DUP do not speak for all of Northern Ireland (they received 29% of the vote in the 2017 assembly elections) and they should not be the only party with an influence on the final outcome. Even if it had little legal standing, a vote by a majority of assembly members registering approval for a form of the EU backstop would have a profound political impact.

A final word on the long-term political implications. The DUP clearly believe that Northern Ireland remaining in the EU’s customs union and single market is a step towards a united Ireland. I doubt if this outcome would actually change perceptions of a united Ireland by much since it would mainly involve a continuation of the status quo for most people in Northern Ireland.  Indeed, a special economic zone status within the UK but with full access to the EU could be popular even with some nationalists. However, the alternative outcome – the return of a hard land border with Northern Ireland firms having poorer access to the EU – may convince many nationalists that they are better off to re-join the EU via unification. With a border poll sufficient to trigger unification, a rejection of the backstop may turn out to be a crucial stop on the road towards (rather than away from) a united Ireland.

Brexit and the Irish Border: Let Northern Ireland Decide?

NI map

Seventeen months on from the Brexit vote, the UK government has largely avoided setting out clear and realistic positions on key issues. Their stance on almost everything continues be a form of cake-and-eat-it.

Nowhere is this more clear than on the question of the Irish border. The UK’s position is that there will be no hard Irish border even though they plan to take the UK out of the EU’s single market and customs union.  When pressed on this, they use the phrases “flexible and imaginative” and “technologies” but don’t put forward much by way of specifics.  (If you think I’m exaggerating, read this).

In the absence of anything concrete from the UK government, the EU has put forward its own flexible and imaginative suggestion that Northern Ireland could remain part of the EU’s single market and customs union. This proposal has been received negatively by the Conservative Party and the DUP. As best I can tell, there has been little public discussion of the objections that have been put forward, so here I want to discuss the economic arguments for and against the EU’s proposal and to propose that Northern Ireland be allowed vote on it.

The Economic Arguments

If Northern Ireland were to remain part of the EU’s single market and customs union, then this would allow free movement of goods between the North and South of Ireland. However, once the UK leaves the EU’s single market and customs union, then there would need to be some form of checks on goods leaving Northern Ireland for Great Britain as well as those goods coming in the other direction. The extent of these checks would depend, over time, on the extent to which the UK departs from its current EU-consistent policies on tariffs and regulations.

One economic argument against this proposal is that Great Britain is a more important market for Northern Irish firms than the Republic of Ireland and the rest of the EU.  Figures show that in 2015, Northern Ireland had €14.4 billion in final sales to Great Britain while final sales to Ireland were €2.9 billion and sales to the rest of Europe were €2.1 billion.  In this sense, surely it makes sense from an economic perspective for Northern Ireland to rule out any trade barriers with Great Britain?

There are two counter-arguments to this position. The first is that the final sales figures under-state the significance of cross-border economic linkages to Northern Ireland. The second is that the inconvenience to firms of NI\GB trade costs will likely be much smaller than the costs associated with an Irish border for goods.


Importance of the All-Ireland Economy

The trade figures just described come from an appendix to the UK government’s position paper on Northern Ireland and Brexit.  However, the paper also acknowledges that

the sale of finished products to Great Britain relies upon cross-border trade in raw materials and components within integrated supply chains meaning trade with both Great Britain and Ireland are vital to Northern Ireland’s economy.

And that cross-border trade is a crucial part of the Northern Irish economy:

Over 5,000 businesses in Northern Ireland exported goods to Ireland in 2015, one and a half times as many as sold goods to Great Britain.

So cross-border trade clearly plays a very significant role in Northern Ireland’s economy. In addition to these substantial linkages, there are also a wide range of all-Ireland bodies devoted to supporting all-Ireland economic linkages in areas such as agriculture, the environment, energy and so on.  Northern Ireland leaving the EU’s single market would likely undermine the economic benefits that have been achieved in these areas.


A Seamless Irish Border?

Once the UK leaves the EU’s customs area and agrees new trade deals with third-party countries, then the EU will require an Irish customs border to protect the integrity of its customs union. There is no point, by the way, in presenting this as a “big undemocratic EU bullies little Ireland” story: There is no way Irish farmers will allow the UK to pursue cheap food deals with the US or Brazil and then have these products imported to the Republic without customs checks.

Brexiteers are currently saying the EU will be “to blame” for the subsequent border checks but the border will only be there because of the UK’s decision to leave the EU. Moreover, despite silly talk from various British politicians about the UK “not being bothered” to enforce a customs border in Ireland, the UK will be under legal obligations to do so via its WTO commitments.

So if Northern Ireland leaves the single market, the customs union and European Economic Association (EEA), there will be border-related checks on goods: Customs checks, rules-of-origin checks, regulatory compliance checks.  Will the much-discussed technological solutions make all of this seamless? No

Some in the British press have pointed to the Sweden-Norway border as an example of how well an Irish border could work. For example, this BBC report starts with stories about average wait times of eight minutes and plans for a future frictionless border.  Only later in the story do we find “there is still plenty of paper to be processed, first with the customs agents, then at the customs office” and that the eight-minute figure might have been over-hyped. “A Swedish trucker grumbles to me that it can take an hour and a half, and he is unimpressed with the level of customer service.”

For lots of reasons, an Irish customs border will be more complicated. For starters, unlike Norway, the UK will be leaving the EEA, meaning extra layers of trade and regulatory checks will be required that do not exist on the Norway-Sweden border.  Another important difference is the political sensitivities of border checks in Northern Ireland and the possibility that formal checkpoints could be a target for terrorist organisations.

The complexities of the physical Irish border are also daunting. Currently, there are an estimated 1.85 million cars crossing the border each month through about 200 crossing points as well as 208,000 light vans and 177,000 lorries. And the nature of the border they are crossing? Here’s Fintan O’Toole

It meanders for 310 miles, and it is not a natural boundary. It was never planned as a logical dividing line, still less as the outer edge of a vast twenty-seven-state union. It is simply composed of the squiggly boundaries of the six Irish counties that had, or could be adjusted to contain, Protestant majorities in 1921. And it cannot be securely policed. We know this because during the Troubles it was heavily militarized, studded with giant army watchtowers, overseen by helicopters, and saturated with troops—and it still proved to be highly porous. It is an impossible frontier.

Trade Costs Associated with an Irish Border

Even if physical customs borders managed to be relatively seamless, Northern Ireland leaving the customs union would still damage many of the businesses that rely on integrated cross-border supply systems.  To give one example, consider the example of dairy businesses in which milk from Northern Ireland is moved over the border for processing, then perhaps moved back to the North for further processing and then perhaps sold in the Republic. When Michael Lux, a German customs expert and former European Commission employee was asked about these kinds of businesses by a House of Commons committee, he responded

“I am always saying to companies, “You need at least two people doing the customs business, if you do it yourself, because one of them may be ill or on holiday and then you will have nobody to do it.”  Alternatively, you can use a service provider that is a logistics company. Depending on the complexity, they will charge you between €20 and €80 per declaration; so the cost will increase enormously just due to the fact that, each time you are doing something that involves a crossing of the border, it creates a cost. That will be part of the cost of the milk and, later, of the cheese, and I cannot imagine that anybody will continue these practices. It would just be too costly.”

Manufacturers in Northern Ireland are gradually learning how costly it would be for them to leave the customs union. Stephen Kelly, Chief Executive of Manufacturing Northern Ireland, told a Commons committee:

I have some evidence here for the Committee today, on just what that country-of-origin certification and the paperwork around that would actually mean in terms of cost to an individual business. Between the development and the time required to produce those certificates, plus the letters of credit from banks that are required to export alongside, the total is £478 per shipment. That is roughly the same price as shipping a container from Northern Ireland to GB or two-thirds of the price of shipping a full container from south-east Asia to Northern Ireland ….

the dangers that are staring our members directly in the face right now is a £478 charge every time they transfer anything across a border, and that is just the paperwork element of it, never mind any tariff elements

To summarise, even a sophisticated “light touch” implementation of an Irish customs border would involve delays and costs that would have a highly negative effect on businesses that have relied on integrated North-South supply chains.

Trade Costs Associated with Northern Ireland Remaining in the Customs Union

What about the alternative? Wouldn’t trade costs associated with Northern Irish firms moving goods to Great Britain also be a big problem? This is clearly not an ideal scenario but there are a number of mitigating points.

The first is that this is a much simpler “border” to monitor. Almost all of Northern Ireland’s trade with Great Britain is shipped via freight and two-thirds of this is shipped via Belfast port (see page 10 here). Goods are shipped already require various pieces of paperwork to be filled out, so it may be possible to add customs and regulatory forms to these and, yes, to use technology to ensure that all of the trucks arriving at Belfast port are ready to board with minimal delays.  It would certainly be a lot easier than monitoring the famously-complicated Irish border.

A second is that it may be possible for the EU and UK to agree to designate Northern Ireland as a special economic zone that would allow streamlined procedures to make moving goods to Great Britain as cheap as possible.  This could include exemptions from various regulatory or rule-of-origin checks for the vast majority of Northern Irish firms and a guarantee from the UK government that there would be no fees charged for any documentation required.  I would guess the Irish government would be willing to share in the costs of administering the checks required in moving goods from Northern Ireland to Great Britain.

This outcome—remaining in the customs union and single market but with simple low-cost procedures for moving goods into the UK—could make Northern Ireland a highly attractive option for international firms.  It would allow them to get direct access to EU markets while also getting lower-cost access to the British market.  Designed in a flexible and imaginative way, Northern Ireland could potentially prosper as a result of its special status.

UK border

Movement of People

One complication when discussing Brexit is that when borders get discussed, most people immediately think about passport control and delays in travel for people moving through airports.  Thus, the idea of Northern Ireland remaining in the EU customs union gets represented as a “border in the Irish sea” and people imagine that Northern Irish residents will have to go through passport control to get into Great Britain. In fact, Northern Ireland remaining part of the customs union and single market would likely have no implications for border controls for people. It would simply affect the movement of goods.

Both the Irish and British government seem committed to maintaining the “common travel area” which would allow Irish and British people to move freely between the two countries as well as maintaining other rights such as the right to work.  I don’t believe anyone is suggesting Irish border controls to monitor people coming from the EU going to Belfast to then enter the UK.

Indeed, the whole focus on border controls as a way of keeping EU citizens out of the UK is misplaced. Unless the UK is planning to eliminate tourism from the EU, there will presumably be a bilateral agreement between the UK and EU on a light-touch visa-free system allowing tourist visits of sixty to ninety days.  The UK will have to decide how to deal with those who overstay these visas but absent the right to work legally in the UK, I doubt if this is going to be a major problem.  And there would really be little point in coming to Ireland to travel to Belfast to enter the UK with plans of working illegally since you could just go directly.

So let’s leave aside issues relating to passport controls: If Northern Ireland remains in the EU customs union and single market, its people will be able to move freely back and forth to “the mainland”.

DUP conference 2017

The Politics

So that’s the economics of it.  Neither of the options on the table are ideal and none will match the current level of market access enjoyed by firms in either the North or South of Ireland. But, on balance, a plan to keep Northern Ireland as part of the EU’s single market and customs union probably has more economic benefits for Northern Ireland than costs.

The politics are infinitely more complicated. The largest party in Northern Ireland, the Democratic Unionist Party (DUP) define themselves by their commitment to maintaining Northern Ireland as part of the UK and they resist anything that looks like it is loosening these links.  Hence, the reaction of DUP’s leader to the EU’s proposal as “reckless” and all about the Irish government “getting the best deal for themselves”.   Similarly, the Conservative Party seem dumbfounded at this idea, with the Secretary of State for Northern Ireland, James Brokenshire commentingI find it difficult to imagine how Northern Ireland could somehow remain in while the rest of the country leaves. I find it impossible.”

DUP opposition to the EU proposal has also not been helped by implications that the proposal is part of an Irish plan to somehow fast-track reunification. I think this is misplaced. The current Varadkar\Coveney Irish leadership team strikes me as perhaps the least republican in Irish history.  And one can believe this is the best proposal available without being in any way focused on a united Ireland. Personally, I would vote against a united Ireland if it was put to a referendum in the Republic and would worry greatly about the economic, political and security capacity of the Republic to successfully absorb Northern Ireland.

So even if worries about “a border in the Irish sea” (which sounds like every unionist’s nightmare) could be clarified and the economic costs of a hard Irish border explained, it is likely that most DUP supporters will accept the economic costs of the new Irish border rather than accept anything that looks like a step towards integration with the Republic. And to be fair, unionists may wonder what their status as UK residents actually means if they started to depart from the regulatory framework of a post-Brexit UK (e.g. if Northern Ireland didn’t get to be part of the Brexiteer vision of the UK as the Singapore of Europe).

The reality, however, is that Northern Ireland already differs sharply from the UK in lots of ways, including its form of government, rules on gay marriage and abortion and the plethora of ways in which the Good Friday agreement has introduced North-South co-operation.  The proposal to stay in the customs union could be considered just an additional recognition that Northern Ireland is a very specific place with a special status and its impact may likely be far less obvious than the re-introduction of border controls on the island.

Also, DUP opposition to the EU proposal is not, on its own, reason to assume it should not be considered. The DUP only received 28 percent of the vote in the recent Assembly elections, with the anti-Brexit Ulster Unionist Party receiving 12 percent and the anti-Brexit anti-hard-border Nationalist parties receiving 40 percent. Indeed, many young people in Northern Ireland are moving past the simplistic Nationalist\Unionist identities and may consider voting for whichever option they believe will be good for their economic security.


A Proposal: Let Northern Ireland Decide

Northern Ireland voted against Brexit by a fairly comfortable 56 to 44 margin but that was a vote for all of the UK to remain in the EU.  We don’t know how Northern Ireland would vote on a proposal to remain in the EU’s customs union and single market while the rest of the UK does not. But I believe the people of the North deserve the opportunity to make this decision.

A future Northern Ireland economic model dictated by either London or Brussels could prove to be a long-standing source of resentment to large numbers of people in the North.  In contrast, a future economic model decided by a referendum would continue the approach of requiring democratic consent for major decisions that was established with the Good Friday agreement.

So my recommendation to the Irish government is as follows: Agree to allow the Brexit talks to move to Phase 2 in return for a commitment from the UK to hold referendum in Northern Ireland by May 2018 on the EU’s proposal for it to remain part of the single market and customs union.

This proposal gives both the UK and Irish government a way out of what currently seems to be an impasse: Ireland can claim to have found a route to maintain the all-Ireland economy, while the UK government can get on with what it really cares about (trade talks) without actually taking the decision to keep Northern Ireland in the single market and customs union.

Is it feasible for the UK government to implement this proposal given its parliamentary reliance on the DUP? I think so. The DUP will doubtless object to a referendum but a British parliamentary majority for the proposal should be easy to obtain with the support of the Labour Party.  After they’ve decided whether to bring the UK government down (thereby losing all the money Theresa May promised them and potentially letting Jeremy Corbyn take over) the DUP can then campaign again to get Northern Ireland out of the single market and customs union. If people of Northern Ireland decided to agree with them, then everyone in Ireland would need to prepare for the border that would be the consequence.