Empirical research suggests that shocks to U.S. treasury term premia have had a persistent subsequent impact on term premia in other developed and emerging fixed income markets. Global financial integration and inflation seem to increase the sensitivity of non-U.S. markets. A 200bps rise in the U.S. premium from current compressed levels could boost the term premia in other countries between 50 and 175 basis points. Hence, a U.S. shift towards reflationary policies or greater net supply of long-term treasuries could greatly increase borrowing costs around the world, exposing weaknesses in overleveraged economies and sectors.

*The post ties in with the lectures on macro trends, particularly the case for a global perspective on monitoring such trends, and the lecture on non-conventional monetary policies, particularly the part on the risks of addiction to non-conventional policies.*

*The below are excerpts from the paper. Emphasis and cursive text have been added. Some formulas have been expressed verbally.*

### Measurement and importance of term premia

“__Long-term interest rates at any maturity can be decomposed into two key components according to the expectations hypothesis: (1) an expectation of future short-term rates; and (2) term premium__…that compensates investors for holding a long-term bond as opposed to rolling over a sequence of short-term bonds over the same period. Given that inflation erodes the nominal value of long-term bonds more than their short-term counterparts, a positive term premium can be interpreted [*partly*] as a compensation for inflation risk.”

“We follow previous studies and use an **affine term structure model** to extract term premium for each economy. The model assumes that the driving force of the yield curve are the first three principal components of the yield curve and imposes no arbitrage condition to derive the expectations components and term premium…__By construction, the yield on a sovereign bond is the sum of its short rate expectations component and the term premium, for any tenor__…Using the concept of no-arbitrage, modern financial theory postulates that the expectations component can be computed using the notion of a risk-neutral measure…Given any observed yield and estimated expectations component the term premium for any tenor can be obtained residually for any maturity of a bond.”

“We find that the share of term premium in explaining yield variation is maturity dependent. __For short-term bond yields such as the 1-year yield variation, the share of term premium is roughly 30%, as opposed to 80% in 10-year bond yields__. Taken at face value, these figures suggest while the co-movement of short-term yields is accounted for by the risk-neutral expectation of future short-term rates, the simultaneous decline in long-term yields is due more to declining long-term term premia.”

### Estimating the impact of a U.S. term premium shock

“This paper sheds additional insights on the dynamic interaction between term premia globally.”

“We obtain __weekly zero-coupon sovereign bond yields for 26 economies__ from Bloomberg. The earliest available data for some economies is from March 1989 but the sample period of yield-curve data for some EMEs is rather short. The sample ends in December 2016.”

“To address the possible impact arising from a change in US term premium, __we estimate a vector autoregression__ (VAR) and conduct an impulse response analysis based on the estimated term premia…Specifically, we run the following weekly frequency VAR: the changes in term premia in the current week are depend on the impact of changes in term premia in the previous week, the impact of the concurrent change in implied U.S. equity volatility, impact of the concurrent change in the broad dollar exchange rate index, a constant and an error term.”

“The __lagged spillover__ [*of a U.S. treasury term premium shock on term premia in other fixed come markets*] __is positive and statistically significant,__ indicating that the effect of US term premium shock is persistent with a significant estimate of the autocorrelation coefficient of 0.78.”

“We consider an interest rate shock of a 200-basis-point increase in the US term premium, which mimics a rise of the US term premium from the current level of almost zero percent to its long-run pre-crisis mean level of 2% between 1980 and 2008…The __figure below shows the cumulative 10-week impulse responses in term premium to the US shock__.”

### Who is hit hardest by U.S. term premium shocks?

“To shed light on why some economies are more responsive to the shock, we regress the estimated responses from the rolling window VAR on various country characteristics in a panel regression. We find that __trade and financial linkages and inflation are the key determinants__.”

“Key findings are summarized as follows…

- Financial linkages across countries can give rise to large swings in capital flows and asset prices. Previous studies have suggested that EMEs that are more financially open are prone to abrupt reversals in their current account positions and sudden stops…[
*Our empirical analysis*] suggests that__economies that are more financially integrated with the global economy are more vulnerable to shocks__from the US… - A significant portion of long-term interest rates reflects compensation for inflation risks. In responding to bond market shocks, it is conceivable that
__economies with higher inflation would be subjected to a larger surge in its sovereign bond yields__. The positive and significant coefficient for inflation [*in our analysis*] appears to confirm this intuition. - Trade integration is measured by the sum of a country’s export and import to its GDP. Previous studies have mixed findings on whether trade can enlarge or dampen foreign shocks… We find that the beneficial role of a trade channel outweighs its possible negative effects on financial stability as
__the coefficient of trade integration is significantly negative which reduces the pass through of the US term premium shock__.”