Minus nought point one percent. That was the price, and it is worth pausing on what a negative number in that position actually asserts: that a bank leaving money at the central bank overnight would get back less than it put in.
On Tuesday the Bank of Japan moved that price to a range of zero to nought point one, its first increase since February 2007. It also abandoned the policy of pinning the ten year government bond yield, which it had run since 2016.
The date to look back at is 29 January 2016, when the negative rate was adopted, and the question is what eight years of it turned out to mean.
What the policy was trying to do
The theory was straightforward and it was not unreasonable.
Japan had spent two decades with prices flat or falling. Falling prices are a trap because they reward waiting: a household that expects something to be cheaper next year defers the purchase, and an economy of deferred purchases produces the conditions that make things cheaper next year.
A negative policy rate attacks that by making holding cash expensive for banks, on the theory that a bank charged to park money will instead lend it. Yield curve control was added to hold long term borrowing costs down too, so that the effect reached mortgages and corporate debt rather than stopping at the overnight market.
What it turned out to mean, in three parts
The first thing eight years produced was an enormous position. Holding a bond yield at a target means buying whatever quantity is required to hold it, and the quantity required is not chosen by the central bank. It is chosen by everyone selling.
The result is a central bank that owns a very large share of its own government’s debt, which is not a policy so much as a structural fact that subsequent policy has to be conducted around. Unwinding it is a separate problem from raising a rate, and the rate was the easy half.
The second thing is what happened to the banks the policy was working through. Lending is funded by deposits and earns a spread, and a spread compressed toward zero for eight years is eight years of a business model producing less than it costs. Regional banks in particular spent that period consolidating, cutting, or reaching for yield in places they would not otherwise have gone.
The third is the currency. A country holding rates at the floor while others raised theirs will see money leave for the higher rate, and the yen spent the last two years weakening substantially. That made imports dearer, which delivered some of the inflation the policy had spent a decade failing to produce, by a route nobody would have chosen.
A fourth effect is the one that is hardest to see and may matter most. A price held at a floor for eight years stops being a price. Borrowing costs are the mechanism by which capital is sorted between projects worth doing and projects not worth doing, and a rate at zero performs very little sorting. Firms that would not survive a normal cost of funds survive; capital that would have been reallocated is not.
Nobody can put a figure on that, which is the problem with it. The cost of a policy that suppresses a signal is invisible in exactly the way a suppressed signal is.
Why the exit came now rather than a year ago
Inflation above the target had been present for a while, and on its own it was never going to be enough.
The condition the bank had said it wanted was inflation driven by wages rather than by import prices, and the evidence arrived in March, when the annual round of union wage negotiations produced increases at the strongest rate in decades.
That distinction is the whole of the timing. Prices rising because a currency fell is a squeeze on households and reverses when the currency does. Prices rising because pay is rising is the self-sustaining thing the policy had been chasing since the 1990s, and the bank moved within days of seeing it.
What the move is not
It is not tightening in any sense a borrower will feel. A policy rate of zero to nought point one is still, by any historical standard, free money, and the bank said explicitly that it expects conditions to remain accommodative.
The distance covered on Tuesday is a tenth of a percentage point, and the bank was at pains to say so. The significance is entirely in the sign changing, which matters for what it signals about the next decision rather than for anything it does to this quarter’s lending.
Why the rest of the world was watching a tenth of a point
Because a persistent gap between Japanese and foreign rates has been one of the reliable features of global finance, and a great deal of money is arranged around it.
Borrowing where money is nearly free and deploying it where returns are higher is a trade that works as long as the cheap end stays cheap and the currency does not move against you. Eight years of a guaranteed floor made that unusually safe, and a lot of capital took the offer.
None of that unwinds on a tenth of a point. What changes is that the floor is no longer a promise, and a trade that depended on a promise now depends on a forecast.
What to watch
Not the next rate decision, which will be small and slow and thoroughly signposted. Three other things.
The bond position, and whether the bank does anything about the size of it, because that is the genuinely difficult problem and nothing was said about it on Tuesday.
The yen, which is the cleanest live reading on whether markets believe more increases are coming. A currency that fails to strengthen after a rate rise is a currency whose holders have concluded the rise was the end rather than the beginning.
And next spring’s wage round. Tuesday’s move rests on one year of settlements. If the second year does not follow, the bank will have exited on the strength of a single data point, and the interesting question will be whether it can go back.
Eight years is long enough that a lending officer who started in 2016 has never once priced a loan against a policy rate that could move. It can move now, by a tenth of a point, which is not enough to change any spreadsheet in Japan and is enough to change what the spreadsheet is for.
