Source: SF Fed; Christensen (2019)

• After the election of Prime Minister Shinzo Abe in December 2012, Japan embarked on a program of extensive policy reforms, popularly known as “Abenomics.” This program contained three components, designated by the Prime Minister as the “three arrows,” consisting of monetary, fiscal, and structural reforms. The monetary reforms included the adoption of an explicit two-percent inflation target by the Bank of Japan (BoJ) and the launch of an asset-purchase program that was planned to double the monetary base, commonly referred to as ”quantitative and qualitative monetary easing” (QQE). Fiscal reforms would include a ”flexible” policy consisting of short-run fiscal expansion followed by fiscal consolidation over the medium term, see Hoshi (2013). Finally, the structural reforms were broadly characterized as including labor market reforms and reduced subsidies in a variety of industries, particularly agriculture, as stressed by Hausman and Wieland (2014).
• The Japanese government bond market is large by international standards. As of December 2018, the total outstanding notional amount of marketable bonds issued by the government of Japan was 1,100.5 trillion yen, of which close to 1 percent represented inflation-indexed bonds.11 In total, Japanese government debt equaled 238% of Japanese nominal national GDP at the end of April 2019, far above the level of any other major industrialized country.
• Figure 3 shows the variation over time for four of our nominal yields. All maturities display a persistent drop in yields since the mid-1990s. There is also a persistent decline in the yield spreads, particularly in the neighborhood of the zero lower bound. The yield spread between the ten- and one-year yield was larger than 200 basis points at the start of the sample and less than 25 basis points at the end of the sample. Kim and Singleton (2012) find that a two-factor model is adequate to fit their data and we therefore choose to use a two-factor model for the nominal yields to capture both of these stylized facts.
• The Japanese government has issued inflation-indexed bonds—known as JGBi—since the spring of 2004. These are all ten-year bonds, which were issued in two separate periods. From March 2004 until June 2008, a total of 16 bonds were issued on a nearly quarterly frequency. The program was then temporarily halted in the aftermath of the global financial crisis. However, shortly after Shinzo Abe reassumed power, the program was resumed. New inflation-indexed bonds have been issued roughly once a year since then. These are government bonds whose principal amount fluctuates in proportion with the consumer price index (CPI) excluding fresh food.
• Our sample starts with three bonds and increases to sixteen bonds by 2008. The number of bonds available then gradually declines beginning in 2011, as bonds from the first wave of issuances start to mature. At the end of our sample there are seven bonds. The number of inflation-indexed bonds nR(t) combined with the time variation in the cross-sectional dispersion in the maturity dimension observed in Figure 4(a) provides the identification of the real factors in our model.

Estimation of Model excluded for post simplicity

• We consider six key policy announcements, which are listed in Table 5. These include the introduction of an explicit inflation target and open-ended expansion of the asset purchase program on January 22, 2013; the introduction of quantitative and qualitative easing (QQE) policy on April 4, 2013; the expansion of the QQE program on October 31, 2014; the movement by the BoJ into negative policy rates on January 29, 2016; the introduction of “yield curve control” by the BoJ on September 21, 2016 in addition to a commitment to overshoot its two-percent inflation target; and the strengthening of the framework for continuous powerful monetary easing announced on July 31, 2018.
• Our results demonstrate that the value of the deflation protection option is generally decreasing with policy announcements signaling enhancement or implementation of the Abenomics program. As a result, estimated changes in inflation expectations on these announcement dates are smaller than would be obtained without adjusting for changes in deflation expectations. For example, fitting our benchmark model without adjusting for changes in deflation protection yields estimates of changes in the five-year and ten-year BEI rates of 11.1 and 7.7 basis points, respectively, over our event window for the announcement of the BoJ adoption of an explicit two-percent inflation target (event I). However, after adjusting for the deflation protection option, the changes are more modest, at 9.1 and 6.5 basis points, respectively. Similarly, without adjusting for the change in the value of the deflation protection option, we would conclude that the adoption of yield curve control had pushed up the five- and ten-year yields by 0.5 and 1.1 basis points, respectively. However, after adjusting for the deflation protection option we estimate that both yields actually fell.
• Other events yielded similar results. The lone exception is the April 3, 2013 event, which announced the launch of QQE. For that event, we obtain a surprising estimate of a 4.9 basis point decline over our event window without the inflation protection option adjustment. This estimated change is attenuated to a decline of 2.9 basis points with the deflation protection adjustment included. Nevertheless, five out of our six events (and all of the ones with an estimated positive change in the ten-year yield without the deflation protection option adjustment) find a lower change in the ten-year yield after controlling for deflation protection.
• The result of these daily decompositions are reported in Table 8. As with the nominal yield decompositions, it is again not too surprising that the model indicates that most of the real yield response came about through changes in the real term premium, rather than through expectations about future real short rates.
• As for the real yield reaction overall, we note that the option adjustment tends to temper the estimated reaction. For the January 29, 2016 announcement, this effect is so large that the negative real yield response estimated without option adjustment turns positive after its inclusion. Hence, the introduction of negative nominal short rates pushed up real yields.
• More importantly, adjusting for the deflation option turns out to be critical for the assessment of the financial market reaction to these announcements. The most notable case is the January 29, 2016 introduction of negative interest rates. When we exclude the deflation protection option in the valuation of JGBi, the GJ (4) model indicates that the announcement resulted in a slight firming in the ten-year BEI, driven by an increase in the inflation risk premium. However, once we account for the option value, the model shows a large drop in the ten-year option-adjusted BEI of 7 basis points driven by declines in both the ten-year expected inflation rate and the ten-year inflation risk premium.
• This paper uses an arbitrage-free term structure model of nominal and real yields on Japanese government bonds to evaluate the impact of news associated with policy announcements under “Abenomics,” the extensive reform program adopted under Japanese Prime Minister Shinzo Abe. To our knowledge, our analysis is the first to assess the impact of these announcements with proper adjustment for the deflation protection enhancements embedded in the inflationindexed bonds issued under Abe. We demonstrate that due to Japan’s long experience with low inflation and pessimistic outlook, the value of these enhancements were typically large, ranging from 50-100 basis points since they were included in 2013. Moreover, they are volatile, suggesting that their incorporation would also be influential in the determination of high-frequency analyses of the impacts of policy announcements under Abenomics.
• We confirm this conjecture in an event study of six important policy announcements under Abenomics from January 2013 through July 31, 2018. Our results demonstrate that changes in inflation expectations on our announcement dates were generally smaller and less optimistic than one would obtain without the deflation protection adjustment. As such, one might conclude that the Abenomics program was not as “disappointing” as early analysis indicated. Our results suggest that market participants were skeptical about the prospects for Abenomics to engineer an escape from Japan’s low inflation environment from the beginning. However, one caveat is that the market’s skepticism may have been driven by pessimism about the government’s implementation of its announcer policy reform.