JP Morgan’s always interesting Flows & Liquidity team have weighed in on the great Japanese yield panic. Japanese government bond yields have jumped since the Bank of Japan launched QE on steroids at the start of April and volatility has risen with them — the 60-day standard deviation of the daily changes in the 10 year JGB yield jumping to 4bp per day, the highest since 2008 (that’s longer term yields on the right for a bit of context):
That has understandably scared people who remember the volatility-induced selloff shock of 2003. From JPM:
At the time, the 10y JGB yield tripled from 0.5% in June 2003 to 1.6% in September 2003. The 60-day standard deviation of the daily changes in the 10y JGB yield jumped from 2bp per day to more than 7bp per day over the same period.
As documented widely in the literature, the sharp rise in market volatility in the summer of 2003 induced Japanese banks to sell government bonds as the Value-at-Risk exceeded their limits. This volatility induced selloff became self-reinforcing until yields rose to a level that induced buying by VaR insensitive investors.
Looking at flow data from 2003, it seems that it was Japanese banks, broker/dealers and foreign investors who sold JGBs at the time. And it was VaR insensitive investors, postal savings and domestic pension funds and insurance companies who absorbed that selling.
How sensitive are Japanese banks to an interest rate volatility shock this time around?
According to JPM, while major Japanese banks are close to average in terms of their vulnerability to interest rate rises, regional and Shinkin (i.e. cooperative banks) are the most vulnerable they have ever been.
And while it is true that regional and Shinkin banks are smaller than major banks, they together hold a large Y50tr of JGBs vs. Y120tr of JGB holdings for major banks (our emphasis):
A theoretical 100bp interest rate shock, i.e. a parallel shift in the Japanese bond yield curve of 100bp, would cause a loss of ¥3tr for Major banks, ¥5tr for Regional banks and ¥2tr for Shinkin banks. As a % of Tier 1 capital, these theoretical losses are close to 35% for Regional and Shinkin banks vs. only 10% for Major banks. The maturity mismatch, the difference between the average remaining maturity of assets minus that of liabilities, has risen for all banks over the past few years. But it was the highest ever at the end of last year for Shinkin banks at 2.2 years, and the highest ever for Regional banks at1.8 years. Major banks had a much lower maturity mismatch of 0.8 years at the end of 2012.
This divergence between Major banks and Regional/Shinkin banks largely reflects differences in the maturity of their bond holdings. The average remaining maturity of bond investments has lengthened to around 4 years at Regional banks and nearly 5 years at Shinkin banks vs. 2.5 years for Major banks.
So in terms of their sensitivity to JGB interest shocks, Japanese banks appear to be more vulnerable than they were in 2003. For example in 2003, the expected theoretical loss from a 100bp interest rate shock was around ¥2tr for Major banks, ¥3tr for Regional banks and ¥1tr for Shinkin banks, significantly lower than they are currently. The maturity mismatch was around 0.8 years for Major banks, i.e. similar to the mismatch reported by the BoJ for the end of 2012. But the maturity mismatch was a lot lower at the time for Regional and Shinkin banks, at 1.2 and 1.5 years, respectively.
That said, the big banks have almost certainly gotten more sophisticated in their risk management making a 2003 repeat less likely. Unfortunately, the regional and Shinkin banks are probably still lagging.
Maybe the real point here though, is that this is all just another unintended consequence of QE.
If you want a stable VaR (which is the size of their positions times volatility) you take larger positions as volatility collapses and have to cut positions when hit by a shock. More simply put, QE increases the risk of VaR shocks. In Japan from JPM one last time:
These maturity mismatches and sensitivity to interest rate shocks have been intensified by QE because 1) of the mechanical rise in duration as yields decline and 2) because banks struggle to maintain their interest margins by extending the maturity of their bond portfolios so that they can capture extra yield. Indeed, the sharp lengthening of the maturity of the bond portfolios of Regional and Shinkin banks would appear to be a reflection of the pressure QE and a persistent low yield environment exert on banks to extend maturity. The average maturity of the bond portfolios of Regional banks was 3 years in 2007 vs. 4 years in 2012. The average maturity of the bond portfolios of Shinkin banks was 2.5 years in 2007 vs. 4.7 years in 2012.
Related links:
Japan’s unfinished policy revolution – Martin Wolf FT
The BoJ massive – FT Alphaville
Simpsons QOTD – Twitter