FRBSF Economic Letter
2000-04 | February 18, 2000
Volatility Spillovers in the U.S. Treasury Market
- The around-the-clock market for Treasury securities
- Modeling volatility spillovers
- Empirical results
The U.S. Treasury market is the largest and most active debt market in the world with about $3.6 trillion of tradable securities outstanding as of September 1999. Treasury securities are traded almost around the clock, starting in Tokyo, then moving to London, and then on to New York. Given this market structure, the prices and hence the yields on Treasury securities can readily incorporate economic announcements and other developments when they become known. For example, on August 27, 1998, Treasury yields fell sharply in overseas trading due to Russia’s suspension of trading in the ruble and reports of weak economic growth from South Korea and Malaysia. When financial markets opened later that day in New York, that information had already been incorporated into Treasury yields, although they continued to fall in light of other financial news.
This Economic Letter reports on research by Fleming and Lopez (1999) that examines the linkages between the three geographical segments of the Treasury market. Specifically, they focus on the volatility of Treasury yields, that is, the variance of the changes in Treasury yields observed in each market segment. Previous research indicates that yield volatility exhibits patterns that can be readily modeled. Using standard econometric models, the authors find that the volatility of the yield changes that occur during the New York trading hours directly affects (or “spills over” into) the yield volatility experienced during the overseas trading hours. The intuition for this result is that market developments during the New York trading hours contain important information that continues to affect yield volatilities in the subsequent trading hours.
The authors also find that the yield volatilities experienced during the Tokyo and London trading hours spill over into each other but do not have a detectable effect on the volatility experienced during the New York trading hours. Two major reasons for this asymmetry in volatility spillovers are the preponderance of major U.S. economic announcements made during the New York trading hours and the concentration of Treasury investors within the United States. These results suggest that the information generated during the New York trading hours generally has the greatest impact on Treasury yield volatility and on the Treasury market more generally. This result contrasts with other research on the foreign exchange market, also an around-the-clock financial market that finds volatility spillovers among all of the market segments.
Trading in Treasury securities is conducted over the counter and not on an established financial exchange. The major market makers are the 30 primary government securities dealers, who are financial firms that interact directly with the Federal Reserve Bank of New York in the course of the Fed’s open market operations. According to Dupont and Sack (1999), the primary dealers averaged $193 billion in daily transactions during the second quarter of 1999. The core of the Treasury market (and the main source of research data) is the interdealer broker market, which accounts for over 90% of trading between dealers and about 40% of total trading. Interdealer brokers, who are independent of the primary dealers, collect and post dealer quotes and execute trades between dealers, thereby facilitating information flows in the market and providing anonymity to the trading dealers. The brokered market is extremely liquid and is characterized by large minimum trade sizes, small bid-ask spreads, and modest brokerage fees.
Trading takes place 23 hours a day during the week in the three trading centers, as shown in the figure. (Note that there is no trading on weekends and that trading decreases to 22 hours a day when New York switches to eastern daylight saving time.) The global trading day begins in Tokyo at 8:30 a.m. local time (6:30 p.m. eastern standard time (EST)) and continues until roughly 5:00 p.m. (3:00 a.m. EST). Trading then passes to London, where it is 8:00 a.m. At about 12:30 p.m. London time, trading passes to New York, where it is 7:30 a.m. EST, and continues until roughly 5:30 p.m. Regardless of location, the trading process is the same; the same Treasury securities are traded by the same dealers through the same interdealer brokers with the same brokerage fees. Trades agreed upon during the Tokyo and London trading hours typically settle in the same way as New York trades, which is one business day later in New York through the Government Securities Clearance Corporation.
Trading volume in the Treasury market is highly concentrated with, on average, about 94% of trades taking place in New York, 4% in London and 2% in Tokyo. This concentration is due in part to the release of U.S. macroeconomic announcements during New York trading hours, which can often lead to sharp price revisions and surges in trading volume. Fleming and Remolona (1997) find that several announcements, such as the producer price index and consumer price index announcements, are associated with sharp changes in Treasury yields. The fixed hours of Treasury futures trading on the Chicago Board of Trade (8:20 a.m.-3:00 p.m. EST) and the predominance of U.S. institutions as owners of Treasury securities (approximately 60% of the amount outstanding) also contribute to this trading concentration. A related feature of the Treasury market is the higher yield volatility experienced during the New York trading hours. As indicated by Fleming (1997), the average absolute yield change, a common measure of yield volatility, for the five-year Treasury note is about 6.2 basis points for New York trading, but it is just about 1.4 basis points for both the Tokyo and London trading hours.
The data used by Fleming and Lopez to model the market segments’ yield volatilities are based on trading in the interdealer broker market from February 24, 1992, to August 19, 1994, a total of 635 trading days. The source of the data is GovPX, Inc., a joint venture by the primary dealers and interdealer brokers to improve the public’s access to Treasury prices and yields. GovPX consolidates and posts real-time quote and transactions data from five of the six major interdealer brokers, accounting for roughly two-thirds of the trading volume in the interdealer broker market. The study focuses on the yields of the five-year Treasury note as well as those of the two- and ten-year notes. The five-, two-, and ten-year notes were the most actively traded securities during the sample period, accounting for about 25%, 20%, and 15% of total trading volume, respectively. They were also the most actively traded in each of the three market segments in the same order.
For each market segment, the yield change is calculated as the difference between the last transaction yield of that segment’s trading hours and the last transaction yield of the previous segment’s trading hours. Missing yields due to holidays in individual trading centers are filled in with the last yield of the previous market segment, effectively imposing a zero yield change. Per-hour volatility for each market segment is then computed as the square of the segment’s yield change divided by the number of trading hours during that market segment. The length of the Tokyo trading hours is measured from the New York close and is typically 9.5 hours. The London and New York trading hours are defined to be 4.5 and 10 hours, respectively. (Note that since this is an over-the-counter market, the lengths of the segments’ trading hours are necessarily approximate.)
For this data set, the average yield changes for the five-year Treasury note are basically zero for the New York and Tokyo market segments, but slightly positive for the London market segment. The preponderance of large yield movements during New York trading hours is clearly demonstrated by the segments’ average per-hour volatilities, which are 0.3 for Tokyo, 0.6 for London, and 4.1 for New York. Similar quantitative results are found for the two- and ten-year Treasury notes.
In order to test for volatility spillovers in the Treasury market, the authors use a modified version of the standard GARCH model, which characterizes the dynamics in a market segment’s per-hour yield volatility as a function of the segment’s previously observed, squared yield changes. For their model, the authors introduce additional terms representing the squared yield changes from the other two segments’ trading hours. For example, the model for today’s Treasury yield volatility during the New York trading hours is a function of past squared New York yield changes and the squared yield changes from today’s Tokyo and London trading hours. If the coefficients on the two out-of-market yield terms are positive and statistically significant, then volatility from these market segments can be said to spill over into the third market segment. If the coefficients are not statistically significant, then the null hypothesis that volatility spillovers are present is rejected.
The empirical results for the five-year Treasury notes are reported here and are qualitatively similar to those of the two- and ten-year notes. For the Tokyo market segment, the coefficients on the squared New York and London yield changes were significant at the 10% level, while for the London market segment the coefficients on the squared Tokyo and New York yield changes were significant at the 5% level. These results indicate that volatility spillovers into these market segments are present; that is, information generated outside of these market segments has an impact on their within-market yield volatilities. However, for the New York market segment, the coefficients on both the squared Tokyo and London yield changes are not significant at the 10% level. Volatility spillovers are thus not indicated for the New York market segment; only information generated during its own trading hours is relevant for modeling its volatility dynamics.
These findings for the Treasury market contrast with those for the foreign exchange market, which is another around-the-clock financial market. For this market, Engle, Ito, and Lin (1990) found that each market segment’s volatility was affected by all the other segments’ squared price changes. The differences in information arrival in these two markets may explain the contrasting results. While the foreign exchange market has a basically constant level of volatility and information arrival across trading hours, Treasury yield volatility is highest during New York trading hours, largely due to macroeconomic announcements made during those hours. It seems reasonable that volatility spillovers emanating from New York would occur in the Treasury market and be readily detectable, while spillovers into New York do not occur or are not easily detectable due to the greater degree of information generated during those hours.
Since trading volume is a key difference between the Treasury market segments, the authors extend their analysis to include lagged, own-market trading volume into the model. In both the Tokyo and London market segments, volatility spillovers are still present for the five-year Treasury yield, even after adjusting for the effect of trading volume. For the New York market segment, the coefficient on the lagged trading volume is not significant, and the null hypothesis of volatility spillovers is again rejected. The results for the two- and ten-year Treasury yields are qualitatively similar.
The Treasury market presents researchers with an opportunity to examine how around-the-clock trading affects market volatility. Previous research on the foreign exchange market indicates that volatility spillovers are present for all market segments. However, the authors reach a different conclusion for the Treasury market. They find that volatility spillovers are present for the Tokyo and London market segments but do not detect such spillovers for the New York market segment. The main reason for these results is that information arrival in the Treasury market, as gauged by the release of U.S. macroeconomic announcements and trading volume, is highly concentrated during the New York trading hours. Thus, spillovers from New York into the overseas market segments seem more likely to occur and easier to detect than spillovers from overseas into the New York market segment. Additional research into the microstructure characteristics of this global financial market, such as the degree of market liquidity in these market segments, is needed to further explain the dynamics in Treasury yield volatility.
Jose A. Lopez
Dupont, D., and B. Sack. 1999. “The Treasury Securities Market: Overview and Recent Developments.” Federal Reserve Bulletin (December) pp. 785-806.
Engle, R.F., T. Ito, and W. Lin. 1990. “Meteor Showers or Heat Waves? Heteroskedastic Intra-Daily Volatility in the Foreign Exchange Market.” Econometrica 58, pp. 525-542.
Fleming, M.J. 1997. “The Round-the-Clock Market for U.S. Treasury Securities.” Federal Reserve Bank of New York Economic Policy Review (July) pp. 9-32.
Fleming, M.J., and J.A. Lopez. 1999. “Heat Waves, Meteor Showers and Trading Volume: An Analysis of Volatility Spillovers in the U.S. Treasury Market.” Federal Reserve Bank of San Francisco Working Papers in Applied Economic Theory 99-09.
Fleming, M.J., and E. Remolona. 1997. “What Moves the Bond Market?” Federal Reserve Bank of New York Economic Policy Review (December) pp. 31-50.
Opinions expressed in FRBSF Economic Letter do not necessarily reflect the views of the management of the Federal Reserve Bank of San Francisco or of the Board of Governors of the Federal Reserve System. This publication is edited by Sam Zuckerman and Anita Todd. Permission to reprint must be obtained in writing.
Please send editorial comments and requests for reprint permission to
Attn: Research publications, MS 1140
Federal Reserve Bank of San Francisco
P.O. Box 7702
San Francisco, CA 94120