ONE reason why the American equity market has rallied since the election of Donald Trump is the hope that taxes on corporate profits will be cut. But that measure has to be paid for, and analysts are only just starting to figure out where the burden might fall.

The initial focus has been on the idea of border-adjustment taxes. But another way of raising revenue is to remove companies’ right to deduct their interest expenses from their taxable income. That proviso has been in place since 1918, when it was introduced to help firms struggling with the impact of the first world war—evidence that tax breaks, once granted, are hard to remove.

Allowing interest payments, but not dividends, to be deducted from corporate profits before tax is paid is a huge distortion to the system. It is a perk worth around 11% of the value of corporate assets. It has tended to encourage companies to take on more debt. By doing so, it may make the economy more risky at the margin: in a recession, highly-indebted companies are likely to go bust more quickly, whereas companies with lots of equity capital can ride out the storm. As a result, this newspaper has favoured the abolition of the deductibility rule.

The revenue gains to the American government would be large—around 1.6% of GDP as of 2013 (in 2007, when interest rates were higher, it would have been 4.3%). Robert Pozen of Harvard Business School has calculated that removing interest deductibility would allow the corporate-tax rate to be cut to 15%, from 35%. A cut that large seems unlikely, however, given another proposal that companies should be immediately allowed to write off their capital investment against tax, a measure that would reduce tax revenue.

The effect on the corporate world would not be uniform. Some companies have a lot more debt, and thus a lot more to lose, than others. Matt King at Citigroup has done some calculations on the winners and losers, on the assumption that the corporate-tax rate is cut to 20%. On that basis, companies with an interest coverage ratio (pre-tax profits divided by interest) of more than 2.4 would be better off.

But few companies actually pay the full 35% tax rate; the average effective tax rate is around 27%, Citigroup says. On this basis, Mr King reckons that only companies with interest cover of more than four times would gain. The effect would be good news for the strongest firms (those judged investment-grade by the rating agencies) and bad news for companies in the high-yield, or “junk”, sector (see chart). This would include many companies financed or owned by private-equity firms.

Whether this shift would make the system safer in the short term also depends on the broader aims of the Trump agenda. “If we are in fact heading into an environment of better growth and less regulation, this should drive more debt issuance in the near term, not less,” say analysts at Morgan Stanley, a bank. “Even if the tax shield goes away, the cost of debt is still relatively low versus historical levels.”

Another issue is that America would be the only country eliminating the interest subsidy.

Companies would still have the incentive to borrow in other countries that allowed them to deduct interest payments against tax. So companies could issue bonds in locations with high corporate-tax rates, and then use the proceeds to buy back bonds issued in America. This could result in even lower bond yields on investment-grade bonds in the American markets as investors chase a dwindling supply of them.

The upshot could be a sharp divide in the bond market, with investment-grade yields falling and junk-bond yields rising, as investors worry about the ability of the latter to service their debts without the tax benefit. The gap, or spread, between the two would rise as a result (at the moment, spreads are four percentage points, quite low by historical standards). If this took place in an atmosphere of generally rising bond yields, because inflation and economic growth are picking up, then it would be hard for riskier companies to refinance their debts.

The number of defaults has already been rising, largely because of the impact of lower oil prices on the energy industry. The default rate on junk bonds was 5% in the 12 months to January 2017, having been under 2% at the start of 2015. In 2016, 2.6 times as many bonds were downgraded by S&P Global Ratings as were upgraded. Besides pointing to the high credit valuations, Morgan Stanley also notes that “uncertainty has rarely been higher in this cycle”. It could be a dangerous time to fiddle with the tax code.