Does Central Bank transparency impact financial markets? A cross-country econometric analysis.

1. Introduction

Does the degree of information that a central bank releases to the public have any effect on the functioning and efficiency of financial markets? Is there a significant difference between the Federal Reserve announcing policy decisions at the time of Federal Open Market Committee (FOMC) meetings and the Federal Reserve forcing the public to ascertain policy decisions through subsequent movements of the federal funds rate? Some authors, including Blinder et al. (2001), Poole, Rasche, and Thornton (2002), and Chortareas, Stasavage, and Sterne (2002), argue that transparency both helps establish monetary policy credibility in the public's eyes and transfers clearer information to financial markets. Many prominent central banks, including the Bank of Canada, Bank of Japan, and Bank of England, moved towards greater transparency in the 1990s, with the U.S. Federal Reserve System following suit in 1994. However, the European Central Bank (ECB) has resisted implementing openness to the same degree, citing the need for speaking with a single, clear policy voice in contrast to a mix of arguments and opinions from the members of its monetary policy authority. (1)

In fact, is it possible that central banks have perhaps become too open in their discussions with the public, and that a partial return to the days of the Fed temple (2) is warranted? Consider the U.S. experience in 2003. Following a protracted U.S. economic slowdown and a nominal federal funds rate dropping to 1.25%, on November 21, 2002, Federal Reserve officials first suggested using long-term Treasury bond buybacks as a way to help lower long-term market interest rates. However, on July 15, 2003, in his semiannual report to the U.S. Congress, Federal Reserve Chairman Alan Greenspan recanted the hypothesized proposal, announcing that the action was highly unlikely. Bond markets responded emphatically, with the 10-year Treasury bond yield rising 20 basis points that day to 3.94%. Thus, public statements by the central bank moved U.S. financial markets, not because of changing economic conditions, but because of the revision of a publicly declared proposal, which in turn adjusted agents' future expectations. The corresponding volatility in bond markets would have been avoided under a system of reduced openness. Are the occasional public missteps by central bankers, then, worth the increase in public information gained through transparency? Fed officials are so concerned about this issue that in September 2003, the FOMC for the first time held a special meeting to discuss how to communicate effectively with the public.

When the U.S. Federal Reserve System made a move towards greater transparency in the conduct of monetary policy in February 1994, it began announcing its targets for the federal funds rate on the afternoon of FOMC meetings. Previously, the Fed did not announce its policy decisions until six weeks after the meeting. This left the public to guess at the Fed's actions, either by studying economic indicators or by watching the federal funds rate in the days and weeks following FOMC meetings, which led some economists and the media to label U.S. monetary policy as being conducted under "a veil of secrecy." Gaining accurate predictions of Fed policy was so important that an entire industry of "Fed watchers" developed. After 1994, then, was there a noticeable change in the dynamics of U.S. financial markets? Did the degree of uncertainty in interest rate movements drop after 1994 in response to the additional information released by the Federal Reserve System? Starting with the Reserve Bank of New Zealand, other central banks underwent similar institutional reforms throughout the 1990s (Table 1). How did their financial markets react, if at all, to the reduction in informational asymmetries between central banks and the public in these countries? Finally, let us consider countries in which the central banks resisted the trend towards greater openness. Did financial markets react in a similar manner to those in other countries, implying that other forces were at work in changing the nature of financial markets worldwide, or are there inherent differences between financial markets in countries that made the move to greater central bank openness compared to those that kept the same levels of transparency?

Focusing on a set of seven industrialized countries, we study whether selected central banks' move toward more open disclosure during the 1990s improved or worsened the predictability of the corresponding national financial markets. Using both threshold ARCH and vector autoregression frameworks, we find that for all countries except Germany, the forecasting error has decreased for interest rates on the respective government bonds across most maturity lengths, and that the expectations hypothesis has performed better at the short end of the yield curve. For central banks that made the move to greater disclosure, this effect was slightly stronger than those banks that resisted increasing the public's information set. Furthermore, both conditional and unconditional volatility dropped for both groups of central banks. These results are consistent with Tabellini's (1987) view that increased central bank openness removes an extra source of uncertainty, helping the smooth functioning of financial markets.

This study adds to the existing literature in the following ways. First, in contrast to many previous empirical studies involving transparency, such as Chortareas, Stasavage, and Sterne (2002) and Cecchetti and Krause (2002), we examine how the move to greater transparency has influenced financial market factors rather than macroeconomic factors. More specifically, we focus on how increased central bank openness may impact the expectations hypothesis, and thus the term structure of interest rates. Understanding the term structure is of prime importance to central banks, since they can most directly influence short-term rates, yet aggregate demand depends chiefly on long-term interest rates. The effectiveness of the monetary transmission mechanism, then, may be linked to the degree of transparency chosen by a central bank. Second, virtually all other studies, including Thornton (1996), Muller and Zellmer (1999), Rafferty and Tomljanovich (2002), and Coppel and Connolly (2003) analyze a single central bank. Instead, we directly compare central bank policies and financial effects in seven developed countries as a means of controlling for changing national and global conditions across the time span. This cross-country methodology also allows for a comparison between central banks that have moved towards greater disclosure and those central banks that have remained opaque. Third, we attempt to disentangle the effects of changing transparency from other central bank procedural changes, such as independence from the national government or implementing an inflation targeting regime.

The paper proceeds as follows. Section 2 lays out the definition of central bank transparency and the economic and political issues surrounding transparency, and summarizes the literature. Section 3 describes the data and models used. Section 4 outlines the main results. Finally, section 5 offers policy recommendations and concludes.

2. Background and Related Literature

Defining Transparency

Complete transparency simply means that the central bank and the public have access to the same information, incomplete or uncertain though it may be, when making economic decisions. (3) Since the central bank generally has access to information first, transparency then suggests that the central bank "accurately" passes information on to other agents. (4) "Accurately" is difficult to define in practical terms. Though many economists believe in the "more is better" creed, a distinction needs to be made here between accuracy and clarity due to the sheer amount of quantitative and qualitative information a central bank collects in the process of policy implementation. One possibility open to a central bank in a move toward greater transparency is to flood the public with all such internally gathered information (such as, for example, on the central bank's website). The obvious drawback to this approach is that the time and expertise needed by members of the public to sift through the information renders timely decision-making nearly impossible. Therefore, greater transparency by this benchmark may in fact hinder economic decisions and thus reduce social welfare.

Winkler (2000), in fact, argues that central bankers face an inherent tradeoff between honesty, clarity, and information efficiency. Any two can be achieved, but not all three. Central banks can be honest (that is, do what they say they are going to do) and clear (in their objectives or policies), thus subscribing to a simple set of policy rules, such as inflation targeting, that can easily be explained to the public and easily followed, but (internal) informational efficiency is sacrificed. Alternately, central banks could clearly state simple decision rules to the public, yet perform their own actions based on internal information, thus sacrificing honesty. Finally, central banks could focus on informational efficiency and eschew passing information to the public and any sort of rules-based policy, thus being honest yet opaque. In recent years, banks in democratic societies have moved away from the second alternative, as political accountability has become a larger issue for policymakers and the public. Thus one significant choice faced by central banks today is between clarity and informational efficiency. According to Winkler (2000), the move from the third scenario to the first scenario, and the corresponding loss of informational efficiency, is one potential downside of greater transparency.

Transparency and Financial Markets

Though openness appears justifiable on democratic grounds, the evidence is mixed concerning its impact on the economy. In particular, the release of information by the central bank and its subsequent dissemination by the public would appear to influence financial markets the most. It is an open question whether financial markets understand a central bank's decision-making process sufficiently to anticipate correctly any rate changes that are made, such that additional information potentially revealed by the bank is redundant. Until the 1990s, central bankers preferred to keep internal information from the public as much as possible, as it was feared more information could potentially destabilize national financial markets as agents attempted to forecast future central bank moves.

The U.S. Federal Reserve System was forced to defend publicly its refusal to reveal internal information during the Merrill vs. FOMC legal case in 1979. The Fed argued that secrecy was needed for the following five reasons. First, FOMC secrecy prevents unfair speculation. Second, if policy prescriptions were made public, inappropriate market reaction or market overreaction may occur if the public incorrectly anticipates the Fed's reaction to newly released information. Third, contemporaneous disclosure of policy and goals may harm the government's commercial interests by raising the cost of borrowing. Fourth, the Fed would rather not take a stand, a priori, on policy prescriptions, but would rather have the flexibility to deal with events on a case-by-case basis. Fifth, disclosure would make it harder for the Fed to smooth interest rates, as the public would react immediately to key economic indicator announcements. Thus, all of the arguments for secrecy translated into how more information could disrupt or distort financial markets.

Several theoretical papers examine conditions under which increased central bank transparency may not lead to socially optimal outcomes. Cukierman and Metzler (1986) argue that changing objectives between fostering economic growth and preventing inflation on the part of the central bank can lead to situations in which informational opacity is preferred. They establish a multi-period model that demonstrates ambiguity is preferred to transparency by a central authority with frequently changing political objectives and a high enough rate of time preference. Ambiguity provides the central authority with greater control over the timing of monetary shocks, and thus over its objectives, though the presence of more frequent shocks may lead to less policy credibility with citizens. Note that the central bank having multiple objectives is essential to the welfare-enhancing scenario that emerges from refusing to divulge information to the public. Hence, for central banks that pursue a single policy goal, in particular inflation targeting, there is no economic rationale not to be transparent. Furthermore, central banks have moved away from trying to catch markets off-guard to boost output temporarily (see Issing 1999 and Blinder et al. 2001). Faust and Svensson (2002) build upon their framework by allowing the central bank to choose separately the optimal degrees of control and transparency. They show that sufficiently patient central banks with a low enough inflation bias will always choose to have the minimum transparency possible. Dotsey (1987) and Rudin (1988) both argue that transparency should increase unconditional volatility, since clearer, more frequent information signals will cause asset prices to fluctuate more frequently to "news." And Geraats (2001) explains that economic transparency (5) may be undesirable in the presence of a conservative central banker that is subject to political pressures.

However, other authors have argued for greater transparency by central banks. Goodfriend (1986) systemically dismantles each of the Fed's five principle arguments for maintaining policy secrecy, and comes to the conclusion that there are very few circumstances under which opacity is desirable for a central bank. Blinder (1998) argues that a nation's central bank should explain its actions to the people, so as to remove the mystery behind the decisionmaking process. Though it would make bank officials and the entire process more accountable to the public, greater openness is a fundamental part of a democracy. If the bank cannot provide a clear explanation of a decision, then the decision may not be a good one. Blinder (1998) and Blinder et al. (2001) assert that more open public disclosure of central bank policies may enhance the efficiency of financial markets. First, greater information about how a central bank makes policy decisions would curtail excessive speculation. Second, clearer decision rules on the part of the central bank would help to reduce the volatility of markets, and thus enhance the predictability of future movements of financial assets.

Recent empirical studies involving central bank transparency find associations between increased public information issuance and improved national macroeconomic performance. Chortareas, Stasavage, and Sterne (2002) use the central bank data set compiled by Fry et al. (2000) to analyze cross-sectional relations between transparency and macroeconomic factors. Though they find evidence suggesting countries with transparent central banks also have lower inflation rates, their methodology is suspect due to the potential reversal causality issue that emerges, as countries with lower inflation rates are more likely to adopt openness measures. Using daily data, Poole, Rasche, and Thornton (2002) consider how predictable the timing and magnitude of FOMC rate moves were pre- and post-1994. They find that though the magnitude was largely known by financial markets, the precise timing of the policy action was almost never correctly ascertained before 1994. After the shift to openness, markets anticipated the direction and size of federal funds rate movements much more precisely, implying that the transparency shift did help financial market participants form more accurate future expectations. Thornton (1996) econometrically examines the consequences of the Fed's policy shift towards immediate disclosure on the federal funds rate. After 1994, much of the Fed's supposed secrecy was removed. Thornton (1996) argues that the new policy has reduced financial market uncertainty, which should result in a lower forecast error (represented by a lower Mean Squared Error) for all interest rates since 1994. He finds this is the case, though at least part of the reduced volatility may be due to relatively quiet financial markets, rather than the change in Fed policy. Recent studies that use similar methodologies include Muller and Zellmer (1999) for Canada, and Coppel and Connolly (2003) for Australia.

It is important to determine whether changes in Federal Reserve openness influence the Federal Funds market. However, it is also important to determine whether increases in efficiency in the Federal Funds market improve the efficiency of the markets for securities of longer maturities. We would like to know if changes in uncertainty regarding the Federal Funds rate influence the uncertainty regarding other interest rates. If the arguments by Cukierman and Metzler (1986) and Faust and Svensson (2002) are correct, then one would expect that increased uncertainty and reduced efficiency of the Federal Funds market may increase uncertainty and reduce efficiency in the markets which determine other interest rates. This is in large part due to the expectations hypothesis of the term-structure of interest rates, which states that other interest rates should be the appropriate average of the expected Federal Funds rate.

3. Data and Methodology

Data Description and Sources

The countries used in this analysis include Australia, Canada, Germany, Japan, New Zealand, the United Kingdom, and the United States. New Zealand and Germany were added as control countries, as no changes in central bank disclosure policies occurred in either country for the time span studied. Thus, by studying how financial markets in these two countries changed throughout the period, we may be able to determine whether the changes, if found, in the other countries were caused by the moves to openness or by other factors such as global economic conditions or faster information flows. The interest rates employed are rates on national government securities reported on the last business day of each month. (6) The monthly data were collected from both central bank websites and Bloomberg, and run from 1990:01 to 2003:05. (7) The data have been log transformed into continuously compounded interest rates. (8) Table 2 reports all descriptive statistics, including sub-period lengths. We investigate the spreads using 3-month, 6-month, 1-year, 2-year, and 3-year government bills and notes. We prefer to use common samples for all spreads within a given country; sample lengths for each country are reported in Table 2. Finally, we focus on the short end of the yield curve, since it is likely that significant market segmentation exists between short-term and long-term bond markets.

Model and Hypothesis

In this section, we clarify the theoretical link between transparency and the expectations hypothesis. For simplicity, let us start by considering the relation between 3-month and 6month rates. If the simple expectations theory of the term structure holds, then the no arbitrage condition is

[R.sup.6.sub.t] = [theta] + 1/2 ([R.sup.3.sub.t] + [E.sub.t][R.sup.3.sub.t+1]). (1)

This equation states that the yield on a 6-month bond ([R.sup.6.sub.t]) is equal to the average of the yield on a current 3-month bond ([R.sup.3.sub.t]) and the expected yield on a 3-month bond purchased when the original bond matures ([E.sub.t][R.sup.3.sub.t+1]), plus a term premium denoted as [theta]. Equation 1 can be rewritten as

([E.sub.t][R.sup.3.sub.t+1] - [R.sup.3.sub.t]] = [R.sup.3.sub.t]) = -2[theta] + 2([R.sup.6.sub.t] - [R.sup.3.sub.t]). (2)

If one assumes that expectations are rational, such that the expected future 3-month rate is equal to the ex-post realized short-term rate plus a forecast error term, then Equation 2 transforms into

([R.sup.3.sub.t+1] - [R.sup.3.sub.t]) = [alpha] + [beta]([R.sup.6.sub.t] - [R.sup.3.sub.t]) + [v.sub.t+1] (3)

and can be estimated via ordinary least squares. For the expectations hypothesis to hold accordingly, the null hypothesis that [beta] must equal 2 must not be rejected.

Various versions of the expectations theory of the term structure have performed remarkably poorly when applied to interest rate data in the 20th century. See Shiller, Campbell, and Schoenholtz (1983), Campbell and Shiller (1991), and Bekaert, Hodrick, and Marshall (1997) for a sample of the literature. There are two crucial assumptions in the expectations hypothesis, the breakdown of either of which may help explain the rejection of the theory. First, the term premium is time invariant. Second, market participants are rational. (9) Studies have shown that the breakdown of either of these two assumptions can downwardly bias the coefficient on [beta] in Equation 3. We focus on the first assumption and follow Mankiw and Miron (1986), who explicitly include a time-varying term premium in a model of the expectations theory of interest rates. (10)

Mankiw and Miron (1986) show that variations in the term premium can bias downwards the estimate of [beta] in Equation 3. Upon incorporating a time-varying term premium ([[theta].sub.t]) in the expectations theory, they find that the estimate of the slope coefficient in Equation 3 converges to

[MATHEMATICAL EXPRESSION NOT CONVERTIBLE IN ASCII], (4)

where [rho] denotes the correlation between [E.sub.t][DELTA][r.sub.t+1] and [[theta].sub.t], [[sigma].sub.2](x) denotes the variance, and [sigma](x) denotes the standard deviation of the variable in question. If the short-rate is not at all predictable, such that [E.sub.t][DELTA][r.sub.t+1] = 0, then plim [beta] = 0. Also, if [rho] = 0, as is the case when the term premium is constant, then plim [beta] = 2 as before. However, if [rho] > 0, then plim [beta] is a monotonically increasing function of [[sigma].sup.2]([E.sub.t][DELTA][r.sub.t+1]), with a lower bound of 0 and an upper bound of 2.11 Thus, with a time-varying term premium, higher predictability of future short-term rates results in a larger slope coefficient. (12) Mankiw and Miron (1986) argue that increased predictability of short-term rates directly translates into larger values of [[sigma].sup.2]([E.sub.t][DELTA][r.sub.t+1]), which results in higher values of plim [beta]. They also explain that with increased predictability of short-term rates comes a better fit of Equation 3, with changes in long-term rates more likely to mirror future short-term rate changes, and hence correspondingly higher [R.sup.2]. In fact, the authors find that the expectations hypothesis performed better in the United States during the 1890-1914 period, with positive and statistically significant [beta] estimates and higher [R.sup.2]. They attribute the poor performance that studies found for the United States in later periods to lower [[sigma].sup.2]([E.sub.t][DELTA][r.sub.t+1]), which comes about due to interest rate smoothing, a point raised by McCallum (1994) as well. Rejection of the expectations hypothesis in these studies comes about because plim [beta] is strictly less than 2 for the latter time periods studied, (13) not due to a failure of the expectations theory.

What effect does increased central bank openness have on the predictability of short-term rates? Tabellini (1987) models secrecy as generating parameter uncertainty. Through a Bayesian rule, agents have the opportunity to learn about the central bank's secret policy target. However, this learning process leads to greater volatility in asset prices, as the market initially overreacts to changes in monetary policy changes. Dismantling Dotsey's (1987) and Rudin's (1988) claims to the contrary, Tabellini (1987) proposes that secrecy tends to increase the unconditional variance of the Federal Funds rate. In addition, opacity increases the conditional variance of the forecast error of the funds rate. Finally, Tabellini (1987) explains that this volatility carries over to market interest rates of longer maturities.

Taken together, Mankiw and Miron (1986) and Tabellini (1987) show how increased central bank transparency leads to less market uncertainty about parameter estimates, helping to form more accurate interest rate forecasts and hence greater market predictability. These more precise forecasts increase plim [beta], making it more likely that the expectations hypothesis holds. In addition, if their assumptions hold, the expectations theory in general and Equation 3 in particular should more closely fit the data after the move to greater openness.

Empirical Model

We now turn to a more general version of the expectations theory of the term structure. According to the simple expectations hypothesis, [beta] = 1 in the following regression equation 

[S.sup.(n,m)*.sub.t] = [alpha] + [S.sup.(n,m).sub.t] + [[epsilon].sub.t], (5)

where [S.sup.(n,m).sub.t] = [R.sup.n.sub.t] - [R.sup.m.sub.t], [[epsilon].sub.t] is an error term, n is a constant, n is the duration of the long-term bond, and m is the duration of the short-term bond. Accordingly, [R.sup.n.sub.t] is the interest rate on the long-term bond at time t and [R.sup.m.sub.t] is the interest rate on the short-term bond. [S.sup.(n,m)*.sub.t] is the perfect foresight spread, which is defined as

[S.sup.(n,m)*.sub.t] = [k-1.summation over (i=1)] (1 - i/k) [[DELTA].sup.m][R.sup.m.sub.t+im], (6)

where k = n/m and [[DELTA].sup.m] means the change is calculated over m periods. (14)

Estimating Equation 5 with data before and after the policy change allows us to determine if there has been any improvement in the performance of the expectations hypothesis since the policy change. If increased openness improves the efficiency and reduces the volatility of financial markets, as claimed in the previous subsection, then one would expect the expectations hypothesis to perform better after a central bank's move to greater transparency. An improvement of the model may take the form of a larger slope coefficient and an increase in the explanatory power of the expectations hypothesis.

To test this hypothesis, we calculate the Root Mean Squared Error (RMSE) of the forecast for each interest rate pair both before and after the central bank's policy change. If increased openness increases efficiency and decreases volatility, then one would expect the RMSE to decrease after the policy change. If the increased transparency provided new and relevant information to markets, then one would expect that the errors caused by incorrect guesses regarding central bank monetary policy changes would become less frequent. This would eliminate one source of error in the market and reduce the RMSE of forecasts based upon market information, ceteris parabis.

We also consider the effect of transparency shifts on the conditional volatility of the associated interest rates by estimating Equation 5 using a threshold Autoregressive Conditional Heteroskedasticity (TARCH) framework. Both Kugler (2002) and Engle, Lilien, and Robins (1987) find evidence to suggest that the expectations hypothesis is rejected in previous studies in part due to the failure to allow for volatility clustering. We thus consider such a framework, but also allow for asymmetric responses to policy changes depending on the direction of the rate movement. In its simplest form, the TARCH(1,1) model, the error term in Equation 5 is distributed with mean zero and variance [[sigma].sup.2.sub.t], where

[[sigma].sup.2.sub.t] = [omega] + [alpha][[epsilon].sup.2.sub.t-1] + [delta][[epsilon].sup.2.sub.t-1][D.sub.t-1] + [gamma][[sigma].sup.2.sub.t-1]. (7)

The conditional variance at time t is a function of four parameters: the mean of the variance process, [omega]; news about volatility from last period (the ARCH term), [[epsilon].sup.2.sub.t-1]; a bad news dummy variable [D.sub.t] that takes on the value 1 if [[epsilon].sub.t-1] is negative and 0 otherwise; and the one-period lagged forecast variance (the GARCH term), [[sigma].sup.2.sub.t-1]. Thus, we are exploring the possibility that a consequence of central banks' move to openness may be not only a change in the RMSE of market forecasts, but also in the nature of the variance process. In particular, openness may increase or reduce volatility persistence of the corresponding errors as well as the effect of last period's news on the conditional variance. The asymmetry term [delta] allows for bad news to have different effects on volatility than good news, a frequent finding when examining financial data. Finally, we can test through this framework whether the mean of conditional volatility, as opposed to unconditional volatility, falls after the transparency shifts.

VAR Methodology

Equation 5 is straightforward to estimate via OLS or TARCH, but since the observations are overlapping the number of independent observations is relatively small. Campbell and Shiller (1991) also point out that tests of Equation 5 do not tell us whether the movements in the actual interest-rate spreads are similar to the movements implied by the expectations hypothesis. They establish the following VAR technique to determine if the expectations hypothesis can explain the shape of the term-structure of interest rates. The general approach is to use a VAR to predict interest rate spreads and to compare these forecasts to the actual values. If the lagged values of interest rate spreads and changes in interest rates summarize all of the available market information, then the forecasted values of these variables will represent the theoretically efficient values (i.e., the values one would expect if the expectations theory were driving the movement of various interest rates).

Let the information set available to market participants be represented by the history of [x.sub.t] [equivalent to] [DELTA][R.sup.m.sub.t],[S.sup.(n,m).sub.t]]. If [x.sub.t] can be represented as a pth-order VAR (where p is finite) then this system can also be written as a first-order VAR in companion form [z.sub.t] = A[z.sub.t-1] + [u.sub.t], where [z.sub.t] is comprised of [DELTA][R.sup.m.sub.t] and its p-1 lags (for 2p terms in total). Next, define a (2p x 1) vector g such that g'[z.sub.t] = [S.sup.(n,m).sub.t], where all the elements of g equal zero except for the p + 1st element, which equals one. We also define a (2p x 1) vector h such that h'[x.sub.t] = [DELTA][R.sup.m.sub.t], where all the elements of h equal zero except the first element which equals one. Finally, using Equation 6 we can compute the VAR forecast of the perfect-forecast spread, defined as [S.sup.T(n,m).sub.t]; this is the spread that would hold if expectations are rational and the expectations hypothesis holds. Campbell and Shiller (1991) show that the "theoretical spread" is

[S.sup.T(n,m).sub.t] = h'A[I - m/n)(I - [A.sup.n])[(I - [A.sup.m]).sup.-1]][(I - A).sup.-1][z.sub.t]. (8)

If the expectations hypothesis is valid, then the following equation should hold, on average:

[S.sup.(n,m).sub.t] = g'[z.sup.t] = [S.sup.T(n,m).sub.t]. (9)

That is, the expected yield spread (at time t) should equal the actual yield spread (observed upon maturity of all relevant bonds that were issued at time t) if the yield spread contains all the relevant information for market participants.

We use the VAR in Equation 9 to calculate the theoretical spread and then compare the actual and theoretical spreads over time. In particular, we look at (i) the correlation between the actual and theoretical spreads, and (ii) the ratio of the standard deviations between the actual and theoretical spreads. We compute these statistics for VARs estimated using data before and after the policy change. (15) If the expectations hypothesis holds exactly then the correlation and ratio of the standard deviations should both equal one. Therefore, if increased openness increases efficiency of financial markets then one would expect that both the correlations and ratios of the standard deviations should move closer to +1.0 after the policy change.

4. Results

Expectations Hypothesis

Table 3 shows the results, for all countries, obtained by estimating Equation 5 using OLS for bonds for all possible maturity spreads, with the p-values for the null hypothesis that the slope coefficient equals +1.0 in brackets, for the pre- and post-transparency sub-periods. Whether the expectations hypothesis holds depends on the country. For example, in Canada and the United States, the null hypothesis is rejected for one (two, United States) of the nine bond maturity pairs in the pre-break sample, but is not rejected for any pairs in the post-break sample for the United States (one, Canada). Furthermore, the coefficients are almost all statistically significant. Thus, for these two countries, the expectations hypothesis performs better following the move to greater transparency. However, for Japan and the United Kingdom the null hypothesis is rejected for 4 and 6 bond pairs (9 and 8 bond pairs) respectively in the pre-break (post-break) sample. Thus, there does not seem to be strong support for the expectations theory in the post-break sample for either country. These latter results are similar to what Campbell and Shiller (1991) and others have found. It is noteworthy that the estimates of [beta] get closer to 1 during the sample period after the change in central bank policy for all countries except Australia, Germany, and Japan. The standard errors of the coefficient estimates also decline after the changes in central bank policy for all countries except Germany.

Table 3 also reports the Root Mean Squared Error for the actual spreads derived from Equation 5 for the periods before and after the policy change respectively. It immediately becomes apparent that the RMSE decreases after 1994 for virtually all combinations of m and n for all countries, with the noteworthy exception being Germany. As Tabellini (1987) asserted, it seems that financial markets became more predictable after the policy change of the corresponding central banks.

Table 3 reports the estimation of Equation 5 using the TARCH(1,1) model. The qualitative results are roughly the same for all countries. However, use of the TARCH model has made rejection of the null hypothesis more likely for all countries. For example, in the United Kingdom, the null hypothesis is rejected for 6 (9) of the bond maturity pairs in the pre-break (post-break) sample. Estimates of the GARCH and ARCH terms are roughly the same for all countries across most bond maturity pairs, indicating that the volatility process itself was largely unaffected by the central banking procedural shifts. (16) For all countries, the asymmetry term [delta] is statistically insignificant, suggesting that good and bad market information have identical effects on bond market volatility. Finally, it is worth noting the estimated conditional variance has consistently dropped across all countries for almost all bond maturity pairs.

As a whole, the results from Table 3 support Tabellini (1987), and appear inconsistent with the theoretical arguments made by Dotsey (1987) and Rudin (1988) that increased transparency makes interest rates more unstable. This point is underscored by considering that the results for New Zealand are not markedly different than for the central banks that underwent a transparency shift, while for Germany increased instability actually appears to have transpired in the latter part of the time span. (17) However, these findings do not prove that transparency was necessarily the cause of improved efficiency of markets, as for the United Kingdom the move to inflation targeting and for the case of Japan the move to central bank independence occurred simultaneously.

To understand further the impact of transparency on bond markets, we conduct Chow Breakpoint tests to ascertain whether a structural break occurred within our data span. First, assuming the known break dates reported in Table 2 for each country, the month that central bank policies became more transparent, we find that the null hypothesis of no structural change is rejected at the 5% significance level for most bond maturity pairs. (18) Second, we let the data choose the break date, analogous to Andrews (1993), under the premise that events other than increased central bank transparency may have changed the predictability of financial markets during our span.

We estimate Equation 5 for break dates in the range [T.sub.b.sup.*], [T.sub.b.sup.*] + 1, ..., T - [T.sub.b.sup.*], where [T.sub.b.sup.*] = 0.1 T. Therefore, we do not consider break dates near the end points of the sample. This is called trimming. We choose the break date maximizing the Wald Statistic that tests the null hypothesis of no structural change. Results can be found in Table 4. We find that for almost every country, the break dates are relatively close to the transparency break dates chosen in this analysis. The exceptions seem to be Australia, in which the estimated break dates occur around 1992 (as opposed to a hypothesized break date of 1996:8), and Japan, in which the estimated break dates vary substantially across bond maturity pairs between 1992:9 and 1995:8 (as opposed to a hypothesized break date of 1998:4).

It is worth repeating that Japan has undergone very few rate changes (five) since April 1998, thus results may be suspect. (19) It also may be the case that Australian policy is similar enough to New Zealand policy such that the increased transparency by the Reserve Bank of New Zealand (RBNZ) in 1989 gave information to Australian market participants about how the Reserve Bank of Australia conducted its policies. For Canada, there are three periods for which the Wald test statistics are large and comparable in magnitude: 1991, 1992, and 1996. These dates coincide with the changes in monetary policy operating procedures outlined in Table 1. The U.K. data have the most striking results, as for all bond maturity pairs the break date chosen is within three months of our hypothesized break of October 1992. New Zealand was once again included as a control country, and it can be seen that late 1992 to early 1993 was chosen as a break date. One possibility is that the inflation targeting framework enacted by the RBNZ may have gained substantial credibility around this time in the eyes of the market once the United Kingdom adopted an inflation targeting policy. Germany was included as well, with the latter months of 1995 largely chosen as the break date. For the United States, for all bond maturity pairs, the break date chosen is within 26 months of our hypothesized break of February 1994. The evidence, coupled with the change in Fed policy, suggests that a structural break occurs in the early 1990s in the United States.

Finally, it is worthwhile to note that for all U.S. bond maturity length pairs, the breakpoints occur at least one year before the formal move to transparency in February 1994. This suggests that by this point the public may have learned the policies of the Fed under Alan Greenspan, helping to reduce uncertainty in the markets.

Vector Autoregressions

Campbell and Shiller (1991) report that regression tests of the expectations hypothesis are limited due to the issue of overlapping observations. In addition, the vector autoregression tests will give us insight as to the shape of the term structure, as well as the relation between the actual and theoretical spread between interest rates of varying maturities. What we would like to know is whether the actual spread behaves in a manner consistent with the theoretical spread implied by the expectations hypothesis. If so, we would expect to see a correlation close to +1.0. Table 5 reports the correlation between the actual and theoretical spread for the two subperiods. It can be seen that in New Zealand, Canada, and the United States, for most maturity lengths m and n, the correlation has moved closer to 1.0 since the policy shift. This result appears inconsistent with the view that increased openness at the central bank creates less efficient and more volatile financial markets. Conversely, for Australia, Germany, and Japan, the correlation has largely moved away from 1.0 since the policy shift. Even after excluding Japan due to its protracted recession and abnormally low rates in the post-break sample, it remains a puzzle why bond markets in Australia become less correlated following the central bank's move to greater transparency using the VAR framework.

Finally, Table 6 reports the ratio of the standard deviations between the theoretical and actual spreads. If the standard deviation of the actual spread closely approximates the theoretical spread, we would expect to see the ratio of the standard deviations close to +1.0. In contrast to the aforementioned results, this ratio consistently moves closer to 1.0 for most maturity lengths m and n for all seven countries, thus adding support to the contention that the magnitudes (though not necessarily the direction) of the bond movements have become more predictable following the increased release of information by the respective central banks. The standard deviations of the theoretical and actual spreads both decline in the post-1994 period, implying that relative tranquility of financial markets may explain the results. Overall then, the results are inconsistent with the notion that the increased transparency of FOMC policy decisions will decrease financial market efficiency.

When taken together, the results from Tables 5 and 6 show that the actual spread is more correlated with the theoretical spread after the move to greater openness (with the obvious exception of Japan), implying that the introduction of more transparent monetary policy decisions has not destabilized markets. It is worth noting, as can be seen from Table 6, that the estimated standard deviations in the VARs are not systematically biased, as there is no clear pattern across bond maturity length pairs or countries as to whether the actual spread is less or more volatile than the standard deviation of the theoretical spreads.

5. Conclusion

This study shows that a more open dialogue between central bankers and the general public is consistent with increased efficiency and reduced volatility in financial markets. Our results build on the existing literature in several important ways. First, by directly comparing market changes across a set of developing nations, it documents that the improvements occur for most bond maturities for most countries. For the majority of interest rate maturities, especially in Canada and the United States (and to a lesser extent, the United Kingdom), we have found that the sub-periods with greater information disclosure coincide with lower interest rate volatility and greater market predictability. Germany and Japan appear to be the only countries in which the opposite empirical results hold. Furthermore, both unconditional and conditional volatility drops across all countries. At any rate, it does not appear that the proponents of central bank secrecy have a strong argument--our findings show that the destabilization that many warned about ex-ante does not seem to have occurred. In fact, recalling the results for Germany, one of the control countries in the study, it appears as if central banks that became more transparent in the 1990s had relatively more predictable bond markets as compared with at least one central bank that remained opaque. This finding suggests that central banks that keep substantial information from the public may in fact be hindering the efficient functioning of financial markets, and the belief that too much information may be destabilizing is at odds with the public's need for the information as a way of improving the formation of expectations concerning future market and economic activity.

One important caveat to our results emerges from the estimated breakpoints, which for several countries, most notably Australia and Japan, occur before the formal moves to transparency. This suggests that it is difficult to separate the effects of learning and the movements in other central bank procedural policies from the effects of increased transparency. One useful extension would be to see how the bond markets in the inflation targeting countries of Australia, Canada, and New Zealand reacted to the United Kingdom's adoption of inflation targeting. An additional extension involves ascertaining the extent to which the United States' move to greater transparency affected markets in the other countries in this study, to add insight as to how much of an influence the U.S. Federal Reserve system wields in foreign markets. Though much more research has yet to be done, including analyzing how equities markets reacted to the change in transparency regimes and incorporating more quantitative measures of policy openness, the data suggest that greater information helps make for a more efficient market.

I would like to thank Matthew Rafferty, Nicole Simpson, Roisin O'Sullivan, the participants of the 2003 Southern Economics Association Annual Conference, and two anonymous referees for their helpful comments and suggestions. Any errors are clearly my own.

References

Akaike, Hirotugu. 1981. Likelihood of a model and information criteria. Journal of Econometrics 16:3-14.

Andrews, David W. K. 1993. Tests for parameter instability and structural change with unknown change point. Econometrica 61:821-56.

Bekaert, Geert, Robert J. Hodrick, and David Marshall. 1997. On biases in tests of the expectations hypothesis of the term structure of interest rates. Journal of Financial Economics 44:309-48.

Blinder, Alan, Charles Goodhart, Philipp Hildebrand, David Lipton, and Charles Wyplosz. 2001. How do central banks talk? Geneva Reports on the World Economy 3. Center for Economic Policy Research.

Blinder, Alan S. 1998. Central bank independence in theory and practice. Cambridge, MA: MIT Press.

Buiter, Willem H. 1999. Alice in euroland. Journal of Common Market Studies 37:181-210.

Campbell, John Y., and Robert J. Shiller. 1991. Yield spreads and interest rate movements: A bird's eye view. Review of Economic Studies 58:495-514.

Cecchetti, Stephen G., and Stefan Krause. 2002. Central bank structure, policy efficiency, and macroeconomic performance: Exploring empirical relationships. Federal Reserve Bank of St. Louis Review 84:47-60.

Chortareas, Georgios, David Stasavage, and Gabriel Sterne. 2002. Does it pay to be transparent? International evidence from central bank forecasts. Federal Reserve Bank of St. Louis Review 84:99-118.

Cook, Timothy, and Thomas Hahn. 1990. Interest rate expectations and the slope of the money market yield curve. Federal Bank of Richmond. Economic Review 76:3-26.

Coppel, Jonathan, and Ellis Connolly. 2003. What do financial market data tell us about monetary policy transparency? Research Discussion Paper, Reserve Bank of Australia.

Cukierman, Alex, and Allan H. Metzler. 1986. A theory of ambiguity, credibility, and inflation under discretion and asymmetric information. Econometrica 54:1099-1128.

Dotsey, Michael. 1987. Monetary policy, secrecy, and federal funds rate behavior. Journal of Monetary Economics 20:463-74.

Engle, Robert F., David M. Lilien, and Russell P. Robins. 1987. Estimating time varying risk premia in the term structure: The Arch-M Model. Econometrica 55:391-407.

Faust, Jon, and Lars E. O. Svensson. 2002. The equilibrium degree of transparency and control in monetary policy. Journal of Money, Credit and Banking 34:520-39.

Froot, Kenneth A. 1989. New hope for the expectations hypothesis of the term structure of interest rates. Journal of Finance 44:283-305.

Fry, Maxwell, DeAnne Julius, Sandra Roger, Lavan Mahadeva, and Gabriel Sterne. 2000. Key issues in the choice of monetary policy framework. In Monetary policy frameworks in a global context, edited by Mahadeva, Lavan and Gabriel Sterne. New York: Routledge, pp. 1-216.

Geraats, Petra M. 2001. Transparency of monetary policy: Does the institutional framework matter? Unpublished paper, University of Cambridge.

Geraats, Petra M. 2002. Central bank transparency. The Economic Journal 112:F532-F565.

Goodfriend, Marvin. 1986. Monetary mystique: Secrecy and central banking. Journal of Monetary Economics 17:63-92.

Greider, William. 1987. Secrets of the temple: How the federal reserve runs the country. New York: Simon & Schuster.

Hansen, Lars P., and Robert J. Hodrick. 1980. Forward exchange rates as optimal predictors of future spot rates: An econometric analysis. Journal of Political Economy 88:829-53.

Hardouvelis, Gikas A. 1988. The predictive power of the term structure during recent monetary regimes. Journal of Finance 43:339-56.

Issing, Otmar. 1999. The eurosystem: Transparent and accountable or 'Willem in Euroland'. Journal of Common Market Studies 37:503-20.

King, Mervyn. 1997. Changes in UK monetary policy: Rules and discretion in practice. Journal of Monetary Economics 39:81-97.

Kozicki, Sharon, and P. A. Tinsley. 2001. Term structure views of monetary policy under alternative models of agent expectations. Journal of Economic Dynamics and Control 25:149-84.

Kugler, Peter. 2002. The term premium, time varying interest rate volatility and central bank policy reaction. Economic Letters 76:311-6.

Mankiw, Gregory N., and Jeffrey A. Miron. 1986. The changing behavior of the term structure of interest rates. The Quarterly Journal of Economics 101:211-28.

McCallum, Bennett T. 1994. Monetary policy and the term structure of interest rates. National Bureau of Economic Research Working Paper #4938.

Muller, Philippe, and Mark Zelmer. 1999. Greater transparency in monetary policy: Impact on financial markets. Technical Report 86. Bank of Canada.

Poole, William, Robert H. Rasche, and Daniel L. Thornton. 2002. Market anticipations of monetary policy actions. Federal Reserve Bank of St. Louis Review 84:65-94.

Rafferty, Matthew C., and Marc Tomljanovich. 2002. Central bank transparency and market efficiency: An econometric analysis. Journal of Eeonomics and Finance 26:150-61.

Rudin, Jeremy R. 1988. Central bank secrecy: 'Fed watching', and the predictability of interest rates. Journal of Monetary Economics 22:317-34.

Shiller, Robert J., John Y. Campbell, and Kermit L. Schoenholtz. 1983. Forward rates and future policy: Interpreting the term structure of interest rates. Brookings Papers on Economic Activity 1:173-217.

Tabellini, Guido. 1987. Secrecy of monetary policy and the variability of interest rates. Journal of Money, Credit and Banking 19:425-36.

Thornton, Daniel. 1996. Does the Fed's new policy of immediate disclosure affect the market? Federal Reserve Bank of St. Louis Review, November/December:77-86.

Tzavalis, Elias. 2004. The term premium and the puzzles of the expectations hypothesis of the term structure. Economic Modeling 21:73-93.

Tzavalis, Elias, and Michael R. Wickens. 1997. Explaining the failures of the term spread models of the rational expectations hypothesis of the term structure. Journal of Money, Credit, and Banking 29:364-80.

Winkler, Bernhard. 2000. Which kind of transparency? On the need for clarity in monetary policy-making. European Central Bank Working Paper No. 26.

Received February 2005; accepted January 2006.

(1) See Buiter (1999) and Issing (1999) for a thorough and lively discussion of the transparency differences between the Bank of England and the ECB.

(2) A reference to Greider (1987).

(3) Thus, using the nomenclature from game theory, complete transparency refers to common knowledge but not complete knowledge.

(4) Agents include the media, financial markets, and households. Note that this information flow is reversed when the issue of bank supervision arises. However, many countries use government agencies rather than the central bank to regulate the banking system (Canada, Japan, and Sweden being notable examples). Furthermore, this study focuses on transparency as it relates solely to macroeconomic conditions and monetary policy.

(5) Geraats (2002) defines five types of transparency: i) openness about policy objectives, including explicit inflation targets and central bank independence; ii) disclosure of data, models, and forecasts (defined as economic transparency); iii) information about policy procedures, through minutes and voting records; iv) prompt announcement about decisions and future actions, as well as future policy leanings; and v) openness about policy actions and implementations.

(6) Daily interest rate data were also employed for the United States and Canada regressions in this section, with qualitatively similar results. Though daily rates are arguably preferable to use, the difficulty of obtaining daily rates for many of these countries and the similar findings across frequencies allow us to draw meaningful conclusions using monthly data.

(7) The data for Germany ends in 1998:12, the month before the creation of the Euro.

(8) Continuously compounded interest rates are defined as log[1 + (i/100)], where i is the annualized nominal interest rate for the government security in question.

(9) Cook and Hahn (1990) provides a thorough discussion of these two assumptions.

(10) Using survey data instead of assuming rational expectations, several studies including Froot (1989) and Kozicki and Tinsley (2001) have estimated Equation 3 and found evidence that supports the expectations hypothesis. Given that it is more likely that transparency affects predictability of markets rather than the rationality of agents, and given the problem inherent in gathering survey data for the countries in our study, we leave a critique of the rational expectations assumption for a future paper.

(11) If [rho] < 0, then again plim [beta] approaches 2 as [[sigma].sup.2]([E.sub.t], [DELTA][r.sub.t+1]) gets large. However, there are several small ranges for which the function is downward sloping. See Mankiw and Miron (1986) for more details.

(12) Though the model in this section considers 3-month and 6-month yields, the same conclusions hold for bonds of different maturity lengths. The associated calculations are left to the reader as an exercise.

(13) Other studies that also find evidence of term premium volatility include Tzavalis (2004), Kugler (2002), Tzavalis and Wickens (1997), and Hardouvelis (1988).

(14) It is straightforward to show that Equation 5 is equivalent to Equation 3 for these two bond rates.

(15) The VARs for each interest rate pair use lag lengths as determined by Akaike (1981) information criterion. Lag lengths are the same for the pre- and post-policy change samples.

(16) Formal hypothesis tests on the equality of the GARCH and ARCH coefficients across the sub-samples are available upon request. We also estimated a GARCH(1,1) model, which yielded almost identical results.

(17) The sample period for Germany ends in 1998:12, as the implementation of the Euro in January 1999 presumably had a considerable impact on German financial markets. Thus, the results are not as clear as would be the case if the time span extended to 2003 paralleling the other six countries.

(18) Results available upon request.

(19) Coppel and Connolly (2003) omit Japan from their analysis for precisely this reason.

Marc Tomljanovich, Department of Economics, Drew University, Madison, NJ 08809 USA; E-mail [e-mail].

Table 1. Timetable of Selected Countries and Central Bank
Transparency Shifts

Country               Date               Nature of Change

Australia        August 1996      Formalization of inflation-
                                    targeting framework
Canada           February 1991    Inflation targets first
                                    announced
                 July 1994        50-basis point operating band
                                    for overnight rate announced
                 February 1996    Announcement of official
                                    overnight rate target; press
                                    release whenever change in
                                    band occurs, including
                                    explanation for change
Germany                --         None for the time-period
                                    studied
Japan            April 1998       Central bank granted
                                    operational independence
                                    through Bank of Japan Act;
                                    this was accompanied by
                                    major shift in transparency
New Zealand      January 1989     Adoption of inflation
                                    targeting, via Reserve Act
                                    of 1989; operational
                                    independence also granted
                                    through act
United           October 1992     Explicit inflation target
  Kingdom                           announced; also, minutes of
                                    policy meetings released
                                    within six weeks
                                    (previously, 30-year delay)
                 May 1997         Bank of England Act,
                                    establishing operational
                                    independence; immediate
                                    disclosure of policy
                                    decisions
                 June 1998        Minutes of policy meetings
                                    released within 15 days
United States    February 1994    Target announcements made on
                                    day of FOMC meeting;
                                    release of minutes (6-week
                                    delay), transcripts (5-year
                                    delay)
                 January 2000     Slight revision to language
                                    used in public statement
                                    following FOMC meetings; no
                                    more neutral bias
                 December 2004    Release of minutes accelerated
                                    to 3-week delay following
                                    FOMC meetings

Sources: Germany, New Zealand, Japan, United States: central bank
websites. Australia: Coppel and Connolly (2003). Canada: Muller and
Zellmer (1999). United Kingdom: King (1997).

Table 2. Descriptive Statistics for 3-Month Government Securities,
Pre- and Post-Transpa-rency Break

Country                  Break Date          N           [mu]

Australia      1996:8       Pre-break        80          8.03
                            Post-break       79          5.08
Canada         1996:2       Pre-break        74          7.50
                            Post-break       87          4.02
Germany        None         Pre-break        87          5.24
                            Post-break       NA           NA
Japan          1998:4       Pre-break        85          2.44
                            Post-break       61          0.08
NZ             1989:1       Pre-break        48         19.79
                            Post-break      173          7.98
UK             1992:10      Pre-break        19         10.01
                            Post-break      127          5.45
US             1994:2       Pre-break        50          4.89
                            Post-break      111          4.44

Country         [sigma]      Maximum      Minimum

Australia         3.14        16.00         4.65
                  0.63         6.76         4.14
Canada            2.79        13.67         3.63
                  1.07         5.75         1.97
Germany           2.30         9.90         2.99
                   NA           NA           NA
Japan             2.33         6.80         0.12
                  0.11         0.39        0.002
NZ                4.31        31.55        14.22
                  2.87        14.39         3.98
UK                0.84        11.00         7.69
                  1.02         7.38         3.36
US                1.91         8.17         2.93
                  1.49         6.36         1.09

Data range starts in 1990:01 for all countries except for New Zealand
(1985:01) and ends in 2003:05 for most countries. Data for Germany end
in December 1998, as the ECB took control of monetary policy for the
EMU in January 1999. Bonds of different maturity lengths follow same
basic pattern, and are not reported here (available upon request). N
denotes number of observations, [mu] the mean, and [sigma] the standard
deviation for the government security in question.

Table 3. Estimation of [beta] from [S.sup.(n,m)*.sub.t] = [alpha] +
[beta] [S.sup.(n,m).sub.t] + [[epsilon].sub.t] using OLS and
TARCH(1,1)

                                                 OLS

                                                     Standard
                                                      Error
Bond Maturity Pairs                       [??]        ([??])

Australia
  3 mo-6 mo           Pre-[T.sub.b]    0.859 **      0.315
                      Post-[T.sub.b]   0.797 **      0.170
  3 mo-12 mo          Pre-[T.sub.b]    0.882 *       0.446
                      Post-[T.sub.b]   0.554         0.363
  3 mo-24 mo          Pre-[T.sub.b]    1.461 **      0.223
                      Post-[T.sub.b]   0.296         0.284
  3 mo-36 mo          Pre-[T.sub.b]    1.656 **      0.254
                      Post-[T.sub.b]  -0.322         0.378
  6 mo-12 mo          Pre-[T.sub.b]    0.771         0.553
                      Post-[T.sub.b]   0.135         0.683
  6 mo-24 mo          Pre-[T.sub.b]    1.537 **      0.250
                      Post-[T.sub.b]   0.716         0.521
  6 mo-36 mo          Pre-[T.sub.b]    1.794 **      0.278
                      Post-[T.sub.b]  -0.558 *       0.303
  12 mo-24 mo         Pre-[T.sub.b]    1.297 **      0.382
                      Post-[T.sub.b]   0.316 *       0.178
  12 mo-36 mo         Pre-[T.sub.b]    0.952 **      0.437
                      Post-[T.sub.b]   0.707 **      0.337

Canada
  3 mo-6 mo           Pre-[T.sub.b]    0.608 *       0.362
                      Post-[T.sub.b]   0.994 **      0.298
  3 mo-12 mo          Pre-[T.sub.b]    1.177 **      0.170
                      Post-[T.sub.b]   1.095 **      0.392
  3 mo-24 mo          Pre-[T.sub.b]    1.181 **      0.173
                      Post-[T.sub.b]   0.866 *       0.528
  3 mo-36 mo          Pre-[T.sub.b]    1.154 **      0.213
                      Post-[T.sub.b]   0.690 *       0.392
  6 mo-12 mo          Pre-[T.sub.b]    1.362 **      0.200
                      Post-[T.sub.b]   0.832         0.596
  6 mo-24 mo          Pre-[T.sub.b]    1.229 *       0.208
                      Post-[T.sub.b]   0.708         0.638
  6 mo-36 mo          Pre-[T.sub.b]    1.239 **      0.227
                      Post-[T.sub.b]   0.814 *       0.519
  12 mo-24 mo         Pre-[T.sub.b]    1.117 **      0.277
                      Post-[T.sub.b]   0.555         0.486
  12 mo-36 mo         Pre-[T.sub.b]    0.992 **      0.261
                      Post-[T.sub.b]   0.418 *       0.262

Germany
  3 mo-6 mo           Pre-[T.sub.b]    0.641         0.399
                      Post-[T.sub.b]   0.718 *       0.396
  3 mo-12 mo          Pre-[T.sub.b]    0.946 **      0.215
                      Post-[T.sub.b]   0.336         0.260
  3 mo-24 mo          Pre-[T.sub.b]    1.081 **      0.124
                      Post-[T.sub.b]  -0.384         0.251
  3 mo-36 mo          Pre-[T.sub.b]    0.957 **      0.121
                      Post-[T.sub.b]  -0.569 *       0.303
  6 mo-12 mo          Pre-[T.sub.b]    0.834 **      0.232
                      Post-[T.sub.b]   0.347 **      0.159
  6 mo-24 mo          Pre-[T.sub.b]    1.245 **      0.271
                      Post-[T.sub.b]  -0.527 **      0.261
  6 mo-36 mo          Pre-[T.sub.b]    1.099 **      0.185
                      Post-[T.sub.b]  -0.725 **      0.351
  12 mo-24 mo         Pre-[T.sub.b]    1.373 *       0.757
                      Post-[T.sub.b]  -0.795         0.493
  12 mo-36 mo         Pre-[T.sub.b]    1.017         0.791
                      Post-[T.sub.b]   0.009         0.249

Japan
  3 mo-6 mo           Pre-[T.sub.b]    0.474 **      0.103
                      Post-[T.sub.b]   0.134         0.190
  3 mo-12 mo          Pre-[T.sub.b]    0.668 **      0.226
                      Post-[T.sub.b]  -0.013         0.192
  3 mo-24 mo          Pre-[T.sub.b]    1.126 **      0.527
                      Post-[T.sub.b]   0.098         0.172
  3 mo-36 mo          Pre-[T.sub.b]    0.383         0.276
                      Post-[T.sub.b]   0.047         0.206
  6 mo-12 mo          Pre-[T.sub.b]    0.663 **      0.339
                      Post-[T.sub.b]   0.315         0.389
  6 mo-24 mo          Pre-[T.sub.b]    1.146         0.836
                      Post-[T.sub.b]   0.218         0.247
  6 mo-36 mo          Pre-[T.sub.b]    0.335         0.285
                      Post-[T.sub.b]   0.169         0.256
  12 mo-24 mo         Pre-[T.sub.b]    0.964         1.160
                      Post-[T.sub.b]   0.280         0.186
  12 mo-36 mo         Pre-[T.sub.b]    0.114         0.132
                      Post-[T.sub.b]   0.158         0.170

New Zealand
  3 mo-6 mo           Pre-[T.sub.b]    1.458 *       0.827
                      Post-[T.sub.b]   1.103 **      0.243
  3 mo-12 mo          Pre-[T.sub.b]    1.949 **      0.769
                      Post-[T.sub.b]   1.126 **      0.283
  3 mo-24 mo          Pre-[T.sub.b]    1.551 *       0.960
                      Post-[T.sub.b]   1.239 **      0.354

United Kingdom
  3 mo-6 mo           Pre-[T.sub.b]    0.925 **      0.409
                      Post-[T.sub.b]   0.527 **      0.172
  3 mo-12 mo          Pre-[T.sub.b]    2.318 **      0.342
                      Post-[T.sub.b]   0.527 **      0.172
  3 mo-24 mo          Pre-[T.sub.b]    1.696 **      0.260
                      Post-[T.sub.b]   0.713 **      0.142
  3 mo-36 mo          Pre-[T.sub.b]    1.133 **      0.197
                      Post-[T.sub.b]   0.568 **      0.102
  6 mo-12 mo          Pre-[T.sub.b]    2.794 **      0.343
                      Post-[T.sub.b]   0.597 **      0.138
  6 mo-24 mo          Pre-[T.sub.b]    2.407 **      0.194
                      Post-[T.sub.b]   0.653 **      0.203
  6 mo-36 mo          Pre-[T.sub.b]    1.682 **      0.302
                      Post-[T.sub.b]   0.575 **      0.153
  12 mo-24 mo         Pre-[T.sub.b]    0.155         1.301
                      Post-[T.sub.b]   0.536 **      0.199
  12 mo-36 mo         Pre-[T.sub.b]    0.493         0.918
                      Post-[T.sub.b]   0.536 **      0.086

United States
  3 mo-6 mo           Pre-[T.sub.b]    1.228         0.869
                      Post-[T.sub.b]   0.968 **      0.258
  3 mo-12 mo          Pre-[T.sub.b]    1.679         1.056
                      Post-[T.sub.b]   1.112 **      0.353
  3 mo-24 mo          Pre-[T.sub.b]    1.791 **      0.691
                      Post-[T.sub.b]   1.026 **      0.445
  3 mo-36 mo          Pre-[T.sub.b]    2.346 **      0.513
                      Post-[T.sub.b]   0.451         0.362
  6 mo-12 mo          Pre-[T.sub.b]    1.446         1.089
                      Post-[T.sub.b]   1.029 **      0.449
  6 mo-24 mo          Pre-[T.sub.b]    1.599 **      0.695
                      Post-[T.sub.b]   1.089 *       0.638
  6 mo-36 mo          Pre-[T.sub.b]    2.312 **      0.514
                      Post-[T.sub.b]   0.853         0.695
  12 mo-24 mo         Pre-[T.sub.b]    0.934         0.807
                      Post-[T.sub.b]   0.416         0.308
  12 mo-36 mo         Pre-[T.sub.b]    0.646 *       0.361
                      Post-[T.sub.b]   0.597         0.520

                                                 OLS

                                       p-Value      Root Mean
                                       [H.sub.0]     Squared
Bond Maturity Pairs                    [beta] = 1     Error

Australia
  3 mo-6 mo           Pre-[T.sub.b]    0.654         0.002958
                      Post-[T.sub.b]   0.234         0.001797
  3 mo-12 mo          Pre-[T.sub.b]    0.791         0.007177
                      Post-[T.sub.b]   0.219         0.003991
  3 mo-24 mo          Pre-[T.sub.b]    0.039 **      0.011619
                      Post-[T.sub.b]   0.013 **      0.006256
  3 mo-36 mo          Pre-[T.sub.b]    0.010 **      0.013964
                      Post-[T.sub.b]   0.001 **      0.005767
  6 mo-12 mo          Pre-[T.sub.b]    0.679         0.005942
                      Post-[T.sub.b]   0.205         0.003346
  6 mo-24 mo          Pre-[T.sub.b]    0.032 **      0.010976
                      Post-[T.sub.b]   0.585         0.005986
  6 mo-36 mo          Pre-[T.sub.b]    0.004 **      0.01337
                      Post-[T.sub.b]   0.000 **      0.005385
  12 mo-24 mo         Pre-[T.sub.b]    0.437         0.009092
                      Post-[T.sub.b]   0.001 *       0.004734
  12 mo-36 mo         Pre-[T.sub.b]    0.913         0.008691
                      Post-[T.sub.b]   0.385         0.004624

Canada
  3 mo-6 mo           Pre-[T.sub.b]    0.279         0.00531
                      Post-[T.sub.b]   0.983         0.00226
  3 mo-12 mo          Pre-[T.sub.b]    0.299         0.00727
                      Post-[T.sub.b]   0.808         0.00500
  3 mo-24 mo          Pre-[T.sub.b]    0.298         0.01011
                      Post-[T.sub.b]   0.800         0.00862
  3 mo-36 mo          Pre-[T.sub.b]    0.469         0.01137
                      Post-[T.sub.b]   0.430         0.00848
  6 mo-12 mo          Pre-[T.sub.b]    0.071 *       0.00607
                      Post-[T.sub.b]   0.778         0.00401
  6 mo-24 mo          Pre-[T.sub.b]    0.271         0.00942
                      Post-[T.sub.b]   0.647         0.00793
  6 mo-36 mo          Pre-[T.sub.b]    0.292         0.01112
                      Post-[T.sub.b]   0.719         0.00861
  12 mo-24 mo         Pre-[T.sub.b]    0.671         0.00803
                      Post-[T.sub.b]   0.359         0.00568
  12 mo-36 mo         Pre-[T.sub.b]    0.976         0.00809
                      Post-[T.sub.b]   0.026 **      0.00657

Germany
  3 mo-6 mo           Pre-[T.sub.b]    0.367         0.00203
                      Post-[T.sub.b]   0.477         0.00131
  3 mo-12 mo          Pre-[T.sub.b]    0.803         0.00280
                      Post-[T.sub.b]   0.011 **      0.00269
  3 mo-24 mo          Pre-[T.sub.b]    0.515         0.00388
                      Post-[T.sub.b]   0.000 **      0.00419
  3 mo-36 mo          Pre-[T.sub.b]    0.726         0.00412
                      Post-[T.sub.b]   0.000 **      0.00499
  6 mo-12 mo          Pre-[T.sub.b]    0.474         0.00235
                      Post-[T.sub.b]   0.000 **      0.00216
  6 mo-24 mo          Pre-[T.sub.b]    0.064 *       0.00382
                      Post-[T.sub.b]   0.000 **      0.00397
  6 mo-36 mo          Pre-[T.sub.b]    0.592         0.00423
                      Post-[T.sub.b]   0.000 **      0.00483
  12 mo-24 mo         Pre-[T.sub.b]    0.551         0.00502
                      Post-[T.sub.b]   0.000 **      0.00365
  12 mo-36 mo         Pre-[T.sub.b]    0.571         0.00515
                      Post-[T.sub.b]   0.000 **      0.00342

Japan
  3 mo-6 mo           Pre-[T.sub.b]    0.000 **      0.00266
                      Post-[T.sub.b]   0.000 **      0.00047
  3 mo-12 mo          Pre-[T.sub.b]    0.141         0.00518
                      Post-[T.sub.b]   0.000 **      0.00081
  3 mo-24 mo          Pre-[T.sub.b]    0.811         0.00743
                      Post-[T.sub.b]   0.000 **      0.00107
  3 mo-36 mo          Pre-[T.sub.b]    0.025 **      0.00970
                      Post-[T.sub.b]   0.000 **      0.00094
  6 mo-12 mo          Pre-[T.sub.b]    0.321         0.00405
                      Post-[T.sub.b]   0.078 *       0.00076
  6 mo-24 mo          Pre-[T.sub.b]    0.861         0.00691
                      Post-[T.sub.b]   0.002 **      0.00117
  6 mo-36 mo          Pre-[T.sub.b]    0.019 **      0.00910
                      Post-[T.sub.b]   0.001 **      0.00119
  12 mo-24 mo         Pre-[T.sub.b]    0.975         0.00560
                      Post-[T.sub.b]   0.001 **      0.00106
  12 mo-36 mo         Pre-[T.sub.b]    0.000 **      0.00599
                      Post-[T.sub.b]   0.000 **      0.00104

New Zealand
  3 mo-6 mo           Pre-[T.sub.b]    0.580         0.00944
                      Post-[T.sub.b]   0.672         0.00749
  3 mo-12 mo          Pre-[T.sub.b]    0.217         0.01465
                      Post-[T.sub.b]   0.656         0.00695
  3 mo-24 mo          Pre-[T.sub.b]    0.566         0.01040
                      Post-[T.sub.b]   0.499         0.00548

United Kingdom
  3 mo-6 mo           Pre-[T.sub.b]    0.855         0.00607
                      Post-[T.sub.b]   0.006 **      0.00180
  3 mo-12 mo          Pre-[T.sub.b]    0.000 **      0.00704
                      Post-[T.sub.b]   0.006         0.00180
  3 mo-24 mo          Pre-[T.sub.b]    0.007 **      0.00714
                      Post-[T.sub.b]   0.044 **      0.00344
  3 mo-36 mo          Pre-[T.sub.b]    0.498 **      0.00533
                      Post-[T.sub.b]   0.000 **      0.00461
  6 mo-12 mo          Pre-[T.sub.b]    0.000 **      0.00467
                      Post-[T.sub.b]   0.003 **      0.00289
  6 mo-24 mo          Pre-[T.sub.b]    0.000 **      0.00571
                      Post-[T.sub.b]   0.087 *       0.00538
  6 mo-36 mo          Pre-[T.sub.b]    0.024 **      0.00455
                      Post-[T.sub.b]   0.006 **      0.00551
  12 mo-24 mo         Pre-[T.sub.b]    0.517         0.00671
                      Post-[T.sub.b]   0.019 **      0.00485
  12 mo-36 mo         Pre-[T.sub.b]    0.354         0.01076
                      Post-[T.sub.b]   0.000 **      0.00446

United States
  3 mo-6 mo           Pre-[T.sub.b]    0.792         0.00205
                      Post-[T.sub.b]   0.901         0.00174
  3 mo-12 mo          Pre-[T.sub.b]    0.519         0.00497
                      Post-[T.sub.b]   0.752         0.00422
  3 mo-24 mo          Pre-[T.sub.b]    0.252         0.01779
                      Post-[T.sub.b]   0.953         0.00836
  3 mo-36 mo          Pre-[T.sub.b]    0.008 **      0.01277
                      Post-[T.sub.b]   0.129         0.00826
  6 mo-12 mo          Pre-[T.sub.b]    0.682         0.00383
                      Post-[T.sub.b]   0.948         0.00344
  6 mo-24 mo          Pre-[T.sub.b]    0.388         0.01049
                      Post-[T.sub.b]   0.889         0.00771
  6 mo-36 mo          Pre-[T.sub.b]    0.011**       0.01242
                      Post-[T.sub.b]   0.834         0.00921
  12 mo-24 mo         Pre-[T.sub.b]    0.935         0.00845
                      Post-[T.sub.b]   0.394         0.00582
  12 mo-36 mo         Pre-[T.sub.b]    0.328         0.01046
                      Post-[T.sub.b]   0.438         0.00640

                                             TARCH

                                                     Standard
                                                      Error
Bond Maturity Pairs                    [??]           ([??])

Australia
  3 mo-6 mo           Pre-[T.sub.b]    0.654         0.155
                      Post-[T.sub.b]   0.797 **      0.163
  3 mo-12 mo          Pre-[T.sub.b]   -0.192 **      0.042
                      Post-[T.sub.b]   0.619         0.253
  3 mo-24 mo          Pre-[T.sub.b]    1.396 **      0.223
                      Post-[T.sub.b]   0.083         0.142
  3 mo-36 mo          Pre-[T.sub.b]    1.589 **      0.271
                      Post-[T.sub.b]  -0.425 **      0.026
  6 mo-12 mo          Pre-[T.sub.b]   -0.435 **      0.206
                      Post-[T.sub.b]   0.205         0.127
  6 mo-24 mo          Pre-[T.sub.b]    1.613 **      0.294
                      Post-[T.sub.b]   0.063         0.118
  6 mo-36 mo          Pre-[T.sub.b]    1.853 **      0.268
                      Post-[T.sub.b]  -0.426 **      0.072
  12 mo-24 mo         Pre-[T.sub.b]    0.869 **      0.263
                      Post-[T.sub.b]   0.349 **      0.082
  12 mo-36 mo         Pre-[T.sub.b]   -0.138         0.161
                      Post-[T.sub.b]   0.618 **      0.118

Canada
  3 mo-6 mo           Pre-[T.sub.b]    0.981 **      0.323
                      Post-[T.sub.b]   0.961 **      0.094
  3 mo-12 mo          Pre-[T.sub.b]    1.248 **      0.087
                      Post-[T.sub.b]   0.792 **      0.143
  3 mo-24 mo          Pre-[T.sub.b]    1.057 **      0.079
                      Post-[T.sub.b]   0.619 **      0.093
  3 mo-36 mo          Pre-[T.sub.b]    1.349 **      0.116
                      Post-[T.sub.b]   0.715 **      0.055
  6 mo-12 mo          Pre-[T.sub.b]    1.721 **      0.090
                      Post-[T.sub.b]   0.768 **      0.161
  6 mo-24 mo          Pre-[T.sub.b]    1.063 **      0.096
                      Post-[T.sub.b]   0.582 **      0.085
  6 mo-36 mo          Pre-[T.sub.b]    1.228 **      0.075
                      Post-[T.sub.b]   0.676 **      0.045
  12 mo-24 mo         Pre-[T.sub.b]    0.887 **      0.155
                      Post-[T.sub.b]   0.395 **      0.126
  12 mo-36 mo         Pre-[T.sub.b]    1.059 **      0.148
                      Post-[T.sub.b]   0.421 **      0.104

Germany
  3 mo-6 mo           Pre-[T.sub.b]    0.001         0.014
                      Post-[T.sub.b]   0.718 **      0.226
  3 mo-12 mo          Pre-[T.sub.b]    0.534 **      0.137
                      Post-[T.sub.b]   0.335         0.271
  3 mo-24 mo          Pre-[T.sub.b]    1.145 **      0.069
                      Post-[T.sub.b]  -0.255 **      0.127
  3 mo-36 mo          Pre-[T.sub.b]    0.974 **      0.056
                      Post-[T.sub.b]  -0.377 **      0.047
  6 mo-12 mo          Pre-[T.sub.b]    0.469 **      0.006
                      Post-[T.sub.b]   0.347         0.222
  6 mo-24 mo          Pre-[T.sub.b]    1.146 **      0.085
                      Post-[T.sub.b]  -0.174 **      0.082
  6 mo-36 mo          Pre-[T.sub.b]    0.986 **      0.091
                      Post-[T.sub.b]  -0.291 **      0.049
  12 mo-24 mo         Pre-[T.sub.b]    0.620         0.516
                      Post-[T.sub.b]  -0.795         0.514
  12 mo-36 mo         Pre-[T.sub.b]    2.074 **      0.865
                      Post-[T.sub.b]   0.191 **      0.057

Japan
  3 mo-6 mo           Pre-[T.sub.b]    0.078         0.086
                      Post-[T.sub.b]   0.134         0.126
  3 mo-12 mo          Pre-[T.sub.b]    0.069         0.155
                      Post-[T.sub.b]  -0.012 **      0.124
  3 mo-24 mo          Pre-[T.sub.b]    0.202 **      0.052
                      Post-[T.sub.b]   0.098         0.198
  3 mo-36 mo          Pre-[T.sub.b]    0.007 **      0.014
                      Post-[T.sub.b]   0.048         0.166
  6 mo-12 mo          Pre-[T.sub.b]    0.452 **      0.189
                      Post-[T.sub.b]   0.315         0.164
  6 mo-24 mo          Pre-[T.sub.b]    0.089 **      0.137
                      Post-[T.sub.b]   0.218 **      0.233
  6 mo-36 mo          Pre-[T.sub.b]    0.155         0.172
                      Post-[T.sub.b]   0.169         0.246
  12 mo-24 mo         Pre-[T.sub.b]    0.265 **      0.116
                      Post-[T.sub.b]   0.280         0.282
  12 mo-36 mo         Pre-[T.sub.b]    0.007         0.016
                      Post-[T.sub.b]   0.158         0.163

New Zealand
  3 mo-6 mo           Pre-[T.sub.b]    1.475 **      0.278
                      Post-[T.sub.b]   1.105 **      0.238
  3 mo-12 mo          Pre-[T.sub.b]    2.071 **      0.305
                      Post-[T.sub.b]   0.965 **      0.094
  3 mo-24 mo          Pre-[T.sub.b]   -0.033         0.190
                      Post-[T.sub.b]   1.484 **      0.126

United Kingdom
  3 mo-6 mo           Pre-[T.sub.b]    1.114 **      0.485
                      Post-[T.sub.b]   0.321 **      0.082
  3 mo-12 mo          Pre-[T.sub.b]    2.882 **      0.414
                      Post-[T.sub.b]   0.521 **      0.084
  3 mo-24 mo          Pre-[T.sub.b]    1.881 **      0.292
                      Post-[T.sub.b]   0.639 **      0.103
  3 mo-36 mo          Pre-[T.sub.b]    1.352 **      0.299
                      Post-[T.sub.b]   0.431 **      0.066
  6 mo-12 mo          Pre-[T.sub.b]    2.786 **      0.664
                      Post-[T.sub.b]   0.459 **      0.091
  6 mo-24 mo          Pre-[T.sub.b]    2.203 **      0.266
                      Post-[T.sub.b]   0.854 **      0.053
  6 mo-36 mo          Pre-[T.sub.b]    1.687 **      0.181
                      Post-[T.sub.b]   0.757 **      0.077
  12 mo-24 mo         Pre-[T.sub.b]    2.866 **      0.712
                      Post-[T.sub.b]   0.783 **      0.080
  12 mo-36 mo         Pre-[T.sub.b]    2.054         2.987
                      Post-[T.sub.b]   0.370 **      0.057

United States
  3 mo-6 mo           Pre-[T.sub.b]    1.431**       0.646
                      Post-[T.sub.b]   0.967**       0.125
  3 mo-12 mo          Pre-[T.sub.b]    1.894**       0.419
                      Post-[T.sub.b]   0.855**       0.157
  3 mo-24 mo          Pre-[T.sub.b]    1.618**       0.310
                      Post-[T.sub.b]   0.258**       0.065
  3 mo-36 mo          Pre-[T.sub.b]   -0.174 **      0.217
                      Post-[T.sub.b]  -0.174 **      0.071
  6 mo-12 mo          Pre-[T.sub.b]    2.211 **      0.760
                      Post-[T.sub.b]   0.408 **      0.163
  6 mo-24 mo          Pre-[T.sub.b]    1.785 **      0.202
                      Post-[T.sub.b]   0.221 **      0.112
  6 mo-36 mo          Pre-[T.sub.b]    2.528 **      0.355
                      Post-[T.sub.b]  -0.237 **      0.096
  12 mo-24 mo         Pre-[T.sub.b]    1.032         2.246
                      Post-[T.sub.b]  -0.616 **      0.106
  12 mo-36 mo         Pre-[T.sub.b]    0.876         0.636
                      Post-[T.sub.b]   0.009         0.122

                                                 TARCH

                                        p-Value
                                       [H.sub.0]
Bond Maturity Pairs                    [beta] = 1      [??]

Australia
  3 mo-6 mo           Pre-[T.sub.b]    0.026 **      1.500 **
                      Post-[T.sub.b]   0.215         0.000
  3 mo-12 mo          Pre-[T.sub.b]    0.000 **      0.951
                      Post-[T.sub.b]   0.132         0.022
  3 mo-24 mo          Pre-[T.sub.b]    0.076 *       1.348
                      Post-[T.sub.b]   0.000 **      1.048
  3 mo-36 mo          Pre-[T.sub.b]    0.029 **      1.146
                      Post-[T.sub.b]   0.000 **      1.272
  6 mo-12 mo          Pre-[T.sub.b]    0.000 **      1.690
                      Post-[T.sub.b]   0.000 **     -0.275
  6 mo-24 mo          Pre-[T.sub.b]    0.037 **      1.699
                      Post-[T.sub.b]   0.000 **      1.156
  6 mo-36 mo          Pre-[T.sub.b]    0.001 **      1.468
                      Post-[T.sub.b]   0.000 **      0.293
  12 mo-24 mo         Pre-[T.sub.b]    0.621         1.768
                      Post-[T.sub.b]   0.000 **      0.967
  12 mo-36 mo         Pre-[T.sub.b]    0.000 **      1.366
                      Post-[T.sub.b]   0.001 **      1.317

Canada
  3 mo-6 mo           Pre-[T.sub.b]    0.951         0.561
                      Post-[T.sub.b]   0.677         0.842 *
  3 mo-12 mo          Pre-[T.sub.b]    0.004 **      1.438
                      Post-[T.sub.b]   0.148         0.633
  3 mo-24 mo          Pre-[T.sub.b]    0.471         1.582
                      Post-[T.sub.b]   0.000 **     -0.013
  3 mo-36 mo          Pre-[T.sub.b]    0.003 **      0.815
                      Post-[T.sub.b]   0.000 **      0.189
  6 mo-12 mo          Pre-[T.sub.b]    0.000 **      0.995
                      Post-[T.sub.b]   0.148         0.623
  6 mo-24 mo          Pre-[T.sub.b]    0.515         1.462
                      Post-[T.sub.b]   0.000 **     -0.245
  6 mo-36 mo          Pre-[T.sub.b]    0.003 **      1.563
                      Post-[T.sub.b]   0.000 **      0.207
  12 mo-24 mo         Pre-[T.sub.b]    0.466         1.325
                      Post-[T.sub.b]   0.000 **      1.236
  12 mo-36 mo         Pre-[T.sub.b]    0.688         1.556
                      Post-[T.sub.b]   0.000 **     -0.282

Germany
  3 mo-6 mo           Pre-[T.sub.b]    0.000 **     -0.214
                      Post-[T.sub.b]   0.213         0.000
  3 mo-12 mo          Pre-[T.sub.b]    0.010 **     -1.077
                      Post-[T.sub.b]   0.000 **      0.030
  3 mo-24 mo          Pre-[T.sub.b]    0.037 **     -0.194
                      Post-[T.sub.b]   0.000 **      0.415
  3 mo-36 mo          Pre-[T.sub.b]    0.656         0.321
                      Post-[T.sub.b]   0.000 **     -0.386
  6 mo-12 mo          Pre-[T.sub.b]    0.000 **     -0.464
                      Post-[T.sub.b]   0.003 **      0.060
  6 mo-24 mo          Pre-[T.sub.b]    0.086 **     -0.326
                      Post-[T.sub.b]   0.000 **      0.931
  6 mo-36 mo          Pre-[T.sub.b]    0.876        -0.447
                      Post-[T.sub.b]   0.000 **      1.263
  12 mo-24 mo         Pre-[T.sub.b]    0.176         2.051
                      Post-[T.sub.b]   0.000 **      0.100
  12 mo-36 mo         Pre-[T.sub.b]    0.270        -0.607
                      Post-[T.sub.b]   0.000 **      1.859

Japan
  3 mo-6 mo           Pre-[T.sub.b]    0.000 **     -1.867 **
                      Post-[T.sub.b]   0.000 **      0.010
  3 mo-12 mo          Pre-[T.sub.b]    0.000 **     -0.747
                      Post-[T.sub.b]   0.000 **      0.412
  3 mo-24 mo          Pre-[T.sub.b]    0.000 **      1.391
                      Post-[T.sub.b]   0.000 **      0.001
  3 mo-36 mo          Pre-[T.sub.b]    0.000 **      1.372
                      Post-[T.sub.b]   0.000 **      0.564
  6 mo-12 mo          Pre-[T.sub.b]    0.004 **      2.664
                      Post-[T.sub.b]   0.000 **      0.124
  6 mo-24 mo          Pre-[T.sub.b]    0.000 **      0.382
                      Post-[T.sub.b]   0.001 **      0.215
  6 mo-36 mo          Pre-[T.sub.b]    0.000 **      1.284
                      Post-[T.sub.b]   0.001 **      0.350
  12 mo-24 mo         Pre-[T.sub.b]    0.000 **      3.685
                      Post-[T.sub.b]   0.000 **      0.411
  12 mo-36 mo         Pre-[T.sub.b]    0.000 **      2.365
                      Post-[T.sub.b]   0.000 **      0.000

New Zealand
  3 mo-6 mo           Pre-[T.sub.b]    0.088 *       1.286
                      Post-[T.sub.b]   0.658         1.037
  3 mo-12 mo          Pre-[T.sub.b]    0.001 **      1.276
                      Post-[T.sub.b]   0.709         1.203
  3 mo-24 mo          Pre-[T.sub.b]    0.000 **      0.931
                      Post-[T.sub.b]   0.000 **      2.036

United Kingdom
  3 mo-6 mo           Pre-[T.sub.b]    0.813        -0.839
                      Post-[T.sub.b]   0.000 **      0.435
  3 mo-12 mo          Pre-[T.sub.b]    0.000 **     -3.399
                      Post-[T.sub.b]   0.000 **      1.055
  3 mo-24 mo          Pre-[T.sub.b]    0.003 **     -1.311
                      Post-[T.sub.b]   0.000 **      0.001
  3 mo-36 mo          Pre-[T.sub.b]    0.241        -3.331
                      Post-[T.sub.b]   0.000 **      1.771
  6 mo-12 mo          Pre-[T.sub.b]    0.007 **     -1.742
                      Post-[T.sub.b]   0.000 **      1.043
  6 mo-24 mo          Pre-[T.sub.b]    0.000 **     -0.754
                      Post-[T.sub.b]   0.007 **      1.486
  6 mo-36 mo          Pre-[T.sub.b]    0.000 **      4.967
                      Post-[T.sub.b]   0.002 **      0.641
  12 mo-24 mo         Pre-[T.sub.b]    0.009 **     -0.485
                      Post-[T.sub.b]   0.007 **      1.141
  12 mo-36 mo         Pre-[T.sub.b]    0.724         1.063
                      Post-[T.sub.b]   0.000 **      1.607 *

United States
  3 mo-6 mo           Pre-[T.sub.b]    0.505         0.044
                      Post-[T.sub.b]   0.797         0.033
  3 mo-12 mo          Pre-[T.sub.b]    0.032**       2.485
                      Post-[T.sub.b]   0.361         0.152
  3 mo-24 mo          Pre-[T.sub.b]    0.046**       1.988
                      Post-[T.sub.b]   0.000 **      1.086
  3 mo-36 mo          Pre-[T.sub.b]    0.000 **      1.979
                      Post-[T.sub.b]   0.000 **      0.783
  6 mo-12 mo          Pre-[T.sub.b]    0.111         1.323
                      Post-[T.sub.b]   0.000 **      1.244
  6 mo-24 mo          Pre-[T.sub.b]    0.000 **      0.904
                      Post-[T.sub.b]   0.000 **      1.310
  6 mo-36 mo          Pre-[T.sub.b]    0.000 **      1.869
                      Post-[T.sub.b]   0.000 **      1.171
  12 mo-24 mo         Pre-[T.sub.b]    0.988         0.054
                      Post-[T.sub.b]   0.000 **      1.092
  12 mo-36 mo         Pre-[T.sub.b]    0.845         0.011
                      Post-[T.sub.b]   0.000 **      1.534

                                             TARCH

Bond Maturity Pairs                    [??]         [??]

Australia
  3 mo-6 mo           Pre-[T.sub.b]    0.162         0.681 **
                      Post-[T.sub.b]   0.050         0.060
  3 mo-12 mo          Pre-[T.sub.b]   -0.013         1.145 *
                      Post-[T.sub.b]  -0.126         1.178 **
  3 mo-24 mo          Pre-[T.sub.b]   -0.205         0.651
                      Post-[T.sub.b]  -0.175         1.067
  3 mo-36 mo          Pre-[T.sub.b]   -0.184         0.669
                      Post-[T.sub.b]  -0.15          1.249
  6 mo-12 mo          Pre-[T.sub.b]    0.050         0.678 **
                      Post-[T.sub.b]  -0.925 **      1.086 **
  6 mo-24 mo          Pre-[T.sub.b]   -0.153         0.345
                      Post-[T.sub.b]  -0.259         1.188
  6 mo-36 mo          Pre-[T.sub.b]   -0.164         0.348
                      Post-[T.sub.b]  -0.122         1.029 **
  12 mo-24 mo         Pre-[T.sub.b]   -0.276         0.655
                      Post-[T.sub.b]  -0.043         1.221 **
  12 mo-36 mo         Pre-[T.sub.b]    0.107         0.739
                      Post-[T.sub.b]  -0.271 **      1.264 **

Canada
  3 mo-6 mo           Pre-[T.sub.b]    0.382**       0.764 **
                      Post-[T.sub.b]  -0.188**       1.009 **
  3 mo-12 mo          Pre-[T.sub.b]    0.155         0.521
                      Post-[T.sub.b]  -0.127**       0.852
  3 mo-24 mo          Pre-[T.sub.b]    0.036         0.444
                      Post-[T.sub.b]   0.039         0.906
  3 mo-36 mo          Pre-[T.sub.b]   -0.549         0.872
                      Post-[T.sub.b]   0.031         1.007
  6 mo-12 mo          Pre-[T.sub.b]    0.083         1.187 **
                      Post-[T.sub.b]   0.129         0.764
  6 mo-24 mo          Pre-[T.sub.b]    0.040         0.411
                      Post-[T.sub.b]  -0.131         1.037
  6 mo-36 mo          Pre-[T.sub.b]   -0.572 **      1.096 *
                      Post-[T.sub.b]   0.103         0.968
  12 mo-24 mo         Pre-[T.sub.b]   -0.238         0.707
                      Post-[T.sub.b]   0.016         0.830
  12 mo-36 mo         Pre-[T.sub.b]   -0.529 **      0.310 *
                      Post-[T.sub.b]  -0.554 **      1.292

Germany
  3 mo-6 mo           Pre-[T.sub.b]   -0.061         1.607
                      Post-[T.sub.b]   0.050         0.606
  3 mo-12 mo          Pre-[T.sub.b]    0.070         2.106 *
                      Post-[T.sub.b]   0.050         0.560
  3 mo-24 mo          Pre-[T.sub.b]   -0.644 **      1.354
                      Post-[T.sub.b]  -0.187         0.873
  3 mo-36 mo          Pre-[T.sub.b]    0.515 **      1.899
                      Post-[T.sub.b]  -0.043         1.044
  6 mo-12 mo          Pre-[T.sub.b]   -0.059 **      1.984 **
                      Post-[T.sub.b]   0.071         0.664
  6 mo-24 mo          Pre-[T.sub.b]   -0.711 **      1.322
                      Post-[T.sub.b]  -0.173         1.145
  6 mo-36 mo          Pre-[T.sub.b]   -0.509 **      1.605
                      Post-[T.sub.b]  -0.064         0.950 **
  12 mo-24 mo         Pre-[T.sub.b]    0.091         0.906 **
                      Post-[T.sub.b]   0.040         0.700
  12 mo-36 mo         Pre-[T.sub.b]   -0.011         2.088 **
                      Post-[T.sub.b]  -0.202         0.867

Japan
  3 mo-6 mo           Pre-[T.sub.b]    0.511 **      0.427 **
                      Post-[T.sub.b]  -0.05          0.711
  3 mo-12 mo          Pre-[T.sub.b]    0.194         0.869
                      Post-[T.sub.b]  -0.244         0.715
  3 mo-24 mo          Pre-[T.sub.b]    0.237 **      0.761
                      Post-[T.sub.b]  -0.032         0.946
  3 mo-36 mo          Pre-[T.sub.b]   -0.071         1.024
                      Post-[T.sub.b]  -0.156         0.887
  6 mo-12 mo          Pre-[T.sub.b]   -0.019         0.949
                      Post-[T.sub.b]   0.023         0.866
  6 mo-24 mo          Pre-[T.sub.b]    0.288 **      0.803
                      Post-[T.sub.b]   0.411         0.564
  6 mo-36 mo          Pre-[T.sub.b]   -0.018         0.887
                      Post-[T.sub.b]   0.144         0.864
  12 mo-24 mo         Pre-[T.sub.b]   -0.062         1.085 **
                      Post-[T.sub.b]   0.061         0.947
  12 mo-36 mo         Pre-[T.sub.b]   -0.024         1.111 **
                      Post-[T.sub.b]   0.051         0.899

New Zealand
  3 mo-6 mo           Pre-[T.sub.b]   -0.395         0.929
                      Post-[T.sub.b]   0.028         0.687
  3 mo-12 mo          Pre-[T.sub.b]   -0.058 *       1.019
                      Post-[T.sub.b]  -0.239         0.911
  3 mo-24 mo          Pre-[T.sub.b]    0.063         0.856
                      Post-[T.sub.b]  -0.419 **      0.878 **

United Kingdom
  3 mo-6 mo           Pre-[T.sub.b]   -0.265         2.219 **
                      Post-[T.sub.b]   0.079         0.524 **
  3 mo-12 mo          Pre-[T.sub.b]    1.275         0.463
                      Post-[T.sub.b]  -0.262         0.858 **
  3 mo-24 mo          Pre-[T.sub.b]    1.593 **      0.564
                      Post-[T.sub.b]  -0.155         1.073 **
  3 mo-36 mo          Pre-[T.sub.b]    1.467         0.525
                      Post-[T.sub.b]  -0.298         0.731
  6 mo-12 mo          Pre-[T.sub.b]    1.480 **     -0.375
                      Post-[T.sub.b]  -0.137         0.578 **
  6 mo-24 mo          Pre-[T.sub.b]    1.440 **      0.588
                      Post-[T.sub.b]  -0.144         0.619
  6 mo-36 mo          Pre-[T.sub.b]   -1.252         1.478
                      Post-[T.sub.b]   0.151         0.712
  12 mo-24 mo         Pre-[T.sub.b]   -0.162         2.335
                      Post-[T.sub.b]  -0.141         0.925 **
  12 mo-36 mo         Pre-[T.sub.b]   -0.081         1.552
                      Post-[T.sub.b]  -0.146 *       0.883 **

United States
  3 mo-6 mo           Pre-[T.sub.b]    0.062         0.945
                      Post-[T.sub.b]   0.057         0.913
  3 mo-12 mo          Pre-[T.sub.b]   -0.335 **      0.845
                      Post-[T.sub.b]  -0.156         1.199 **
  3 mo-24 mo          Pre-[T.sub.b]   -0.139         0.591 **
                      Post-[T.sub.b]  -0.213 **      1.364
  3 mo-36 mo          Pre-[T.sub.b]   -0.201         0.973
                      Post-[T.sub.b]  -0.056         1.227
  6 mo-12 mo          Pre-[T.sub.b]   -0.265         1.256 **
                      Post-[T.sub.b]  -0.239         1.152
  6 mo-24 mo          Pre-[T.sub.b]    0.179         1.289 **
                      Post-[T.sub.b]  -0.367 **      1.427 **
  6 mo-36 mo          Pre-[T.sub.b]   -0.111         1.072
                      Post-[T.sub.b]  -0.319 **      1.300 **
  12 mo-24 mo         Pre-[T.sub.b]    1.119        -0.049
                      Post-[T.sub.b]   0.122         0.891 **
  12 mo-36 mo         Pre-[T.sub.b]   -0.365         0.791
                      Post-[T.sub.b]  -0.357         1.266

Hodrick-Hansen (1980) standard errors used for all the regressions.
Estimates of [??] are not reported (for all countries). Data for
Germany only extend to 1998:12 because the EMU began in early 1999.
Trend break of 1994:12 was chosen as midpoint of the Germany sample.
The 6-month and 36-month data are unavailable for New Zealand due to
illiquid markets for those debt instruments. Break date of 1995:12
was chosen as midpoint of the New Zealand sample. Start date of
1990:01 chosen to match other country start dates.

* Statistically significant coefficients at the 10% level.

** Statistically significant coefficients at the 5% level.

Table 4. Estimated Break Dates

Bond Maturity Pairs   Australia   Canada      Germany     Japan

3 month-6 month       1992:5      1994:8      1996:1      1995:8
3 month-12 month      1992:3      1996:10     1996:1      1995:4
3 month-24 month      1992:4      1992:5      1995:10     1992:9
3 month-36 month      1991:9      1991:7      1994:6      1995:5
6 month-12 month      1992:4      1992:3      1994:8      1995:4
6 month-24 month      1992:5      1992:7      1995:9      1992:9
6 month-36 month      1991:11     1991:9      1995:2      1995:4
12 month-24 month     1993:2      1991:10     1993:8      1993:7
12 month-36 month     1993:3      1996:10     1995:12     1993:7

                      New         United      United
Bond Maturity Pairs   Zealand     Kingdom     States

3 month-6 month          --       1992:12     1992:7
3 month-12 month      1993:8      1992:9      1992:6
3 month-24 month      1992:7      1992:10     1992:6
3 month-36 month         --       1992:9      1991:8
6 month-12 month         --       1992:10     1992:6
6 month-24 month         --       1992:9      1992:7
6 month-36 month         --       1992:9      1991:9
12 month-24 month     1993:6      1993:1      1992:6
12 month-36 month        --       1992:12     1993:1

The 6-month and 36-month debt data are unavailable for New Zealand
due to illiquid markets. First and last 10% of sample trimmed to
allow for sufficient observations when estimating break date.

Table 5. Correlation between Actual and Theoretical Spread

Bond Maturity Pairs                  Australia   Canada   Germany

3 month-6 month     Pre-[T.sub.b]        0.875   -0.019     0.797
                    Post-[T.sub.b]       0.877    0.394     0.700

3 month-12 month    Pre-[T.sub.b]       -0.408    0.390     0.142
                    Post-[T.sub.b]      -0.351    0.740    -0.551

3 month-24 month    Pre- [T.sub.b]       0.888    0.744     0.039
                    Post-[T.sub.b]       0.641    0.717    -0.430

3 month-36 month    Pre-[T.sub.b]        0.776    0.786     0.033
                    Post-[T.sub.b]       0.520    0.731    -0.238

6 month-12 month    Pre-[T.sub.b]       -0.551    0.302     0.110
                    Post-[T.sub.b]      -0.304    0.613    -0.380

6 month-24 month    Pre- [T.sub.b]       0.831    0.714     0.069
                    Post-[T.sub.b]       0.581    0.549    -0.321

6 month-36 month    Pre- [T.sub.b]       0.660    0.797     0.897
                    Post-[T.sub.b]       0.577    0.803     0.426

12 month-24 month   Pre- [T.sub.b]      -0.309    0.817    -0.205
                    Post-[T.sub.b]      -0.164    0.594    -0.065

12 month-36 month   Pre-[T.sub.b]       -0.219    0.813     0.057
                    Post-[T.sub.b]      -0.551    0.523     0.018

                                               New     United    United
Bond Maturity Pairs                 Japan    Zealand   Kingdom   States

3 month-6 month     Pre-[T.sub.b]   -0.026     --       -0.101    0.594
                    Post-[T.sub.b]  -0.516     --       -0.280    0.745

3 month-12 month    Pre-[T.sub.b]    0.067    0.628     -0.667    0.431
                    Post-[T.sub.b]  -0.094    0.730      0.258    0.804

3 month-24 month    Pre- [T.sub.b]   0.372    0.908     -0.474   -0.059
                    Post-[T.sub.b]   0.119    0.945      0.164    0.666

3 month-36 month    Pre-[T.sub.b]   -0.037     --       -0.473   -0.324
                    Post-[T.sub.b]  -0.452     --        0.106    0.419

6 month-12 month    Pre-[T.sub.b]    0.102     --       -0.725    0.250
                    Post-[T.sub.b]   0.037     --        0.198    0.667

6 month-24 month    Pre- [T.sub.b]   0.142     --       -0.721   -0.166
                    Post-[T.sub.b]  -0.211     --        0.190    0.395

6 month-36 month    Pre- [T.sub.b]  -0.019     --        0.268   -0.513
                    Post-[T.sub.b]  -0.448     --        0.172   -0.180

12 month-24 month   Pre- [T.sub.b]   0.104    0.909     -0.154   -0.356
                    Post-[T.sub.b]  -0.134    0.932     -0.064    0.293

12 month-36 month   Pre-[T.sub.b]   -0.029     --       -0.310   -0.576
                    Post-[T.sub.b]  -0.539     --       -0.062   -0.120

The lag length for each VAR was determined by using Akaike and
Schwartz Information Criterions. A separate VAR is estimated for
each interest rate spread during each subsample, but the lag
length does not change.

Table 6. Ratio of Theoretical to Actual Spread's Standard Deviation

Bond Maturity Pairs                  Australia   Canada   Germany

3 month-6 month     Pre-[T.sub.b]      0.806     0.986     0.775
                    Post-[T.sub.b]     0.898     0.699     0.764

3 month-12 month    Pre-[T.sub.b]      0.871     0.946     0.581
                    Post-[T.sub.b]     0.976     0.855     1.208

3 month-24 month    Pre-[T.sub.b]      1.196     0.720     0.652
                    Post-[T.sub.b]     1.073     0.757     1.684

3 month-36 month    Pre-[T.sub.b]      1.250     0.635     0.633
                    Post-[T.sub.b]     1.162     0.689     0.534

6 month-12 month    Pre-[T.sub.b]      0.972     0.940     0.488
                    Post-[T.sub.b]     0.923     0.884     1.493

6 month-24 month    Pre-[T.sub.b]      1.192     0.690     0.813
                    Post-[T.sub.b]     0.921     0.768     1.469

6 month-36 month    Pre-[T.sub.b]      0.405     0.262     0.601
                    Post-[T.sub.b]     0.410     0.305     0.639

12 month-24 month   Pre-[T.sub.b]      0.713     0.727     0.436
                    Post-[T.sub.b]     0.769     0.610     1.250

12 month-36 month   Pre-[T.sub.b]      0.683     0.750     0.461
                    Post-[T.sub.b]     0.873     0.613     1.275

                                               New     United    United
Bond Maturity Pairs                  Japan   Zealand   Kingdom   States

3 month-6 month     Pre-[T.sub.b]    0.688     --       1.043    1.552
                    Post-[T.sub.b]   0.607     --       0.588    0.825

3 month-12 month    Pre-[T.sub.b]    0.863    1.868     1.607    1.454
                    Post-[T.sub.b]   0.863    1.627     1.071    1.091

3 month-24 month    Pre-[T.sub.b]    1.000    1.647     2.018    0.904
                    Post-[T.sub.b]   0.822    1.563     0.731    0.896

3 month-36 month    Pre-[T.sub.b]    0.260     --       2.215    0.764
                    Post-[T.sub.b]   0.428     --       0.568    0.778

6 month-12 month    Pre-[T.sub.b]    0.975     --       1.556    1.509
                    Post-[T.sub.b]   0.672     --       1.041    1.142

6 month-24 month    Pre-[T.sub.b]    0.821     --       2.009    0.887
                    Post-[T.sub.b]   0.734     --       0.745    0.794

6 month-36 month    Pre-[T.sub.b]    0.225     --       1.892    0.547
                    Post-[T.sub.b]   0.293     --       0.802    0.468

12 month-24 month   Pre-[T.sub.b]    0.725    1.646     2.225    0.673
                    Post-[T.sub.b]   0.682    1.411     0.565    0.787

12 month-36 month   Pre-[T.sub.b]    0.187     --       2.296    0.643
                    Post-[T.sub.b]   0.441     --       0.467    0.670

The lag length for each VAR was determined by using Akaike and
Schwartz Information Criterions. A separate VAR is estimated for
each interest rate spread during each subsample, but the lag
length does not change.