Journal of the Asia Pacific Economy

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/rjap20

Interest rate term structure and the Chinese fiscal

policy: a mixed frequency term structure approach

Yuhuang Shang , Xuyang Zhang & Qing Wang

To cite this article: Yuhuang Shang , Xuyang Zhang & Qing Wang (2020): Interest rate term
structure and the Chinese fiscal policy: a mixed frequency term structure approach, Journal of the
Asia Pacific Economy, DOI: 10.1080/13547860.2020.1862391

To link to this article: https://doi.org/10.1080/13547860.2020.1862391

Published online: 21 Dec 2020.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=rjap20
JOURNAL OF THE ASIA PACIFIC ECONOMY
https://doi.org/10.1080/13547860.2020.1862391

Interest rate term structure and the Chinese fiscal policy:

a mixed frequency term structure approach
Yuhuang Shanga,b, Xuyang Zhanga and Qing Wanga
a
Institute of Chinese Financial Studies, Southwestern University of Finance and Economics, Chengdu,
China; bFinancial Security Collaborative Innovation Center, Southwestern University of Finance and
Economics, Chengdu, China

ABSTRACT KEYWORDS
We specify a mixed-frequency Nelson–Siegel term structure (MF- Term structure; fiscal policy;
NS) model with fiscal policy information to investigate the influ- mixed frequency; forecast
ence of fiscal policy on interest rate term structure. This paper
finds the following results: First, fiscal policy information can sig-
nificantly improve both the in-sample fitting and out-of-sample
predictions of the Chinese yield curve. More importantly, com-
pared with short-term bonds, the fiscal variable has a greater con-
tribution to long-term bonds. Second, the response of the level
factor of the yield curve to monetary impulse is positive, while
the response to fiscal surprise is negative. Furthermore, we sug-
gest that monetary policy shows an inflation expectation effect
for the level factor. However, fiscal policy shocks on the level fac-
tor via the wealth effect. Finally, fiscal policy account for the
larger forecast variance of yield curve than monetary policy does
in the short forecast horizon. Meanwhile, this paper further shows
that fiscal policy has a greater contribution to forecast variance of
long-term yield than short-term yield.

1. Instruction
A classic and popular issue is to understand the dynamic relations between macro-
economics and the term structure of the yield curve in macrofinance. The joint mod-
eling of yield curves and macroeconomics has received much attention since Ang and
Piazzesi (2003). Recently, a number of theoretical and empirical studies have focused
on the relationship between the yield curve and monetary policy (Rudebusch and Wu
2008; Ang et al. 2011; Bikbov and Chernov 2013; Kung 2015; Rogers, Scotti, and
Wright 2018; Auclert 2019), because the nominal short rate is the monetary policy
instrument. Inflation (Ang, Bekaert, and Wei 2008; G€ urkaynak, Sack, and Wright
2010; Wright 2011, Ulrich 2013; Boragan Aruoba 2020) and real activity (Ang and
Piazzesi 2003; Diebold, Rudebusch, and Aruoba 2006; Jardet, Monfort, and Pegoraro
2013; Joslin, Priebsch, and Singleton 2014; Paccagnini 2016) are also introduced into
the term structure model in a growing amount of literature. Nevertheless, an attempt

CONTACT Yuhuang Shang [email protected]

ß 2020 Informa UK Limited, trading as Taylor & Francis Group
2 Y. SHANG ET AL.

to thoroughly uncover the dynamic relations between fiscal policy and the term struc-
ture of the yield curve seems to be lacking (Afonso and Martins 2012).
How does the term structure of interest rates respond to a rise in government
expenditure? Does the fiscal policy variable help forecast the yield curve? Is there a
different effect on term structure between fiscal and monetary policy behavior? There
are no clear answers for these issues in either empirical or theoretical research.
However, this is the key piece of information that policy-makers and financial market
participants need to make decisions.
Some early researchers focused on fiscal behavior and long-term interest rates
(Canzoneri, Cumby, and Diba 2002; Laubach 2009; Afonso 2010; Schuknecht, von
Hagen, and Wolswijk 2010; Hauner and Kumar 2011). They suggested that there
seems to be a significant impact of budget deficits and government debt on long-
term interest rates. Furthermore, some papers focus on the term spread of the yield
curve. Canzoneri, Cumby, and Diba (2002) studied the effect of projections of cumu-
lative budget deficits on the term structure spread. In addition, many studies have
focused on developed markets such as that of the U.S. (Dai and Philippon 2005;
Marsal, Kaszab, and Horvath 2016) and Europe (Haugh, Ollivaud, and Turner 2009;
Sgherri and Zoli 2009; Afonso and Rault 2010). Some other papers have compared
the U.S. and the European market (Bernoth, von Hagen, and Schuknecht 2006;
Nickel, Rother, and R€ ulke 2009; Afonso and Martins 2012). However, as far as we
know, fewer papers focus on the term structure of sovereign bonds in developing
countries. In fact, fiscal policy usually provides a greater contribution to macroeco-
nomic than monetary policy does in developing countries such as China. Yan and
Guo (2018) also suggest that the Chinese Government is very powerful, and fiscal
policy plays an important role in China’s economy.
Many researches advocate specifying the term structure model to investigate the
dynamic relations between fiscal policy and the term structure (Afonso and Martins
2012; Dai and Philippon 2005). It is well known from the finance literature that the
term structure may be parsimoniously represented by estimates of the level, slope and
curvature of the yield curve (Afonso and Martins 2012). This type of term structure
model is the dynamic Nelson–Siegel (NS) model (Nelson and Siegel 1987; Diebold,
Rudebusch, and Aruoba 2006). Specifically, Diebold, Rudebusch, and Aruoba (2006)
specify the dynamic-factor Nelson–Siegel model with macroeconomic indicators.
The latent risk factors that drive the term structure dynamics and the observed
macroeconomic variables characterizing the state of the economy are linked in this
framework. Afonso and Martins (2012) study the relationship between term struc-
ture and fiscal behavior in this approach. A sub-class of the macro-finance literature
has used affine arbitrage-free models of the yield curve. Dai and Philippon (2005)
provide empirical evidence of fiscal deficits driving nominal yield curve dynamics
with affine macro-finance model.
A shortcoming of the traditional dynamic Nelson–Siegel model is that it does not
make use of the available information. Because macro-level and yield curve data are
usually observed with different (mixed) frequencies, some crucial variables with dif-
ferent frequencies fail to be introduced into the same frequency model. Shang and
Zheng (2018) suggest that the specification of mixed-frequency term structure models
 JOURNAL OF THE ASIA PACIFIC ECONOMY 3

helps to provide more useful forecast information. However, the mixed-frequency

term structure models are still silent on the effects of fiscal policy on the term struc-
ture. In sum, the discussion of the relationship between fiscal policy and term struc-
ture in the modern term structure model framework is still limited. As a result, the
link between fiscal policy and term structure is not well established, especially in
developing countries.
In this paper, we introduce fiscal behavior into the mixed frequency Nelson–Siegel
(MF-NS) model and investigate the contribution of fiscal variables to the term struc-
ture factor, in-sample fit and out-of-sample forecast. Here, we pay more attention to
the yield curve of developing countries, and the mixed-frequency macroeconomic var-
iables of China, such as inflation rate, government expenditure, monetary supply and
GDP, are collected. The main works of this paper are as follows: First, we specify the
mixed-frequency Nelson–Siegel term structure model with fiscal information, which
is abbreviated as the MF-NS_F model. The Bayesian estimation method (the MCMC
algorithm) related to the mixed frequency model is also discussed. Second, by com-
paring the mixed frequency Nelson–Siegel (MF-NS) model without fiscal behavior,
we investigate the influence of fiscal policy on the in-sample fit of the yield curve.
Third, according to the impulse response and variance decomposition methods, this
paper also investigates the effects of fiscal behavior on term structure and bond
yields. Finally, we discuss out-of-sample forecasts of mixed-frequency models with fis-
cal information and test the power of our mixed-frequency model forecasts with
Diebold and Mariano statistics.
Our contribution to the literature is twofold. First, employing the mixed-frequency
term structure model, we shed new light on the relationships between fiscal policy
and term structure in an empirical study. We find that fiscal and monetary behaviors
can produce significant impulse-to-level factors in China. Fiscal policy will increase
the demand for government bonds and thus decrease the yield return via the wealth
effect. This paper also shows that fiscal policy always provides a greater contribu-
tion to the variance decomposition of the yield curve than monetary policy in the
short forecast horizon. This result relates to the research of Yan and Guo (2018).
They study fiscal behavior and the sovereign yield curve in China. They estimate
the level, slope and curvature factors of the yield curve and then analyze interac-
tions between fiscal behavior and the yield curve with VAR model. Beyond Yan and
Guo (2018), this paper employs the mixed frequency Nelson–Siegel (MF-NS) model
to investigates the effects of fiscal behavior on term structure. Other studies, such as
Marsal, Kaszab, and Horvath (2016), investigate fiscal policy and the term structure
of interest rates in a DSGE model, but they try to explore the asset pricing implica-
tions of fiscal policy.
Second, our results emphasize that fiscal policy contributes to both in-sample fit-
ting and out-of-sample forecasting of yield bonds with mixed frequency macroeco-
nomic information. More specifically, the fiscal variable shows great contributions to
long-term bonds for out-of-sample forecasting. Our empirical work is similar to the
literature on predictions of the yield curve. Favero, Niu, and Sala (2012) investigate
the forecasting performance of the term structure model with macroeconomic varia-
bles and find that macro-factors are very useful in forecasting the medium and long
4 Y. SHANG ET AL.

rates. In our research, we further find that fiscal policy has more contribution to
long-term bonds. Our empirical forecast result is also in line with the findings of
Huse (2011). He suggests that a term structure whose dynamics are driven uniquely
by observable macroeconomic state variables outperforms the (latent-variable) bench-
mark model in an out-of-sample study. Shang and Zheng (2018) also forecast yield
bonds with mixed frequency macroeconomic information. But they did not show the
rule of fiscal policy.
The remainder of this paper is organized as follows: In section 2, we introduce the
traditional term structure model and present the specification of mixed-frequency
term structure models. In section 3, we describe low-frequency macro-variables and
the data on government bonds in China. Section 4 provides and analyzes the param-
eter estimation and other empirical results. In section 5, we present conclud-
ing remarks.

2. The model
2.1. Dynamic Nelson–Siegel model with macroeconomic
There are many term structure models in the literature. Diebold and Li (2006) show
the dynamic Nelson–Siegel model with a state-space approach.

yt ¼ Kft þ et (1)

ft l1 ¼ A1 ðft1 l1 Þ þ gt (2)

 


et 0 H 0
N , (3)
gt 0 0 Q

where
 yt ¼ ½yt ðs1 Þ, :::, yt ðsN Þ0 is the N  1 vector of the bond yield. ft ¼
0
Lt St Ct is the 3  1 term structure factor, and Lt , St and Ct capture the level,
slope and curvature factor of the term structure, respectively. The error terms et and
gt are assumed to be independently and identically distributed white noise. l1 and A1
are the coefficient matrices of the state equation. K is the N  3 coefficient matrix of
the measurement equation. It can be expressed as the following equation.
0 1
1es1 k 1es1 k s1 k
B1 e C
B s1 k s1 k C
B C
B 1es2 k 1es2 k s2 k C
B1 e C
K¼B
B s2 k s2 k C
C (4)
B .. .. .. C
B. . . C
B C
@ 1esN k 1e sN k
A
1 esN k
sN k sN k

Diebold, Rudebusch, and Aruoba (2006) further specify the dynamic Nelson–Siegel
ðmÞ
model with macroeconomic information. Let xt be the macroeconomic observations
 JOURNAL OF THE ASIA PACIFIC ECONOMY 5

such as GDP and CPI. Diebold, Rudebusch, and Aruoba (2006) assume that the
macroeconomic indicators mainly effect on the latent factors of the yield curve. And
then, the model can be written as:

yt ¼ Kft þ et (5)

! ! !
ft lf ft1 lf gf , t
ðmÞ ¼ A2 ðmÞ þ ðmÞ (6)
xt lx xt1 lx gx, t

! 


et 0 H 0
ðmÞ N , (7)
gt 0 0 X

Equation (5) describes the links between yield vector the latent factor of yield
curves. The state equation (6) shows the dynamic relationship between the latent fac-
tor of yield curves and macroeconomic variables.

2.2. Mixed-frequency Nelson–Siegel model

According to Shang and Zheng (2018), we specify a mixed-frequency model to
improve the fitting and forecasting of the yield curve with more available macro-level
and financial observations,
ðmÞ
Let ft and xt represent unobservable term structure factors and observable high
ðqÞ
frequency macroeconomic indices, respectively. Here, zt denotes the observable low
ðmÞ
frequency macroeconomic variable and zt represents the related unobservable high
frequency information.
ðqÞ ðmÞ
We need to specify the frequency change equation between zt and zt : Let
ðqÞ ðmÞ
zt ¼ PðLÞzt , where L is the lag operator and PðÞ is a P th-order polynomial.1
Let us take quarterly GDP as an example.
ðqÞ ðqÞ ðqÞ ðqÞ
Setting Gt is the aggregate of quarterly GDP and zt ¼ ln ðGt Þ ln ðGt12 Þ, we
then have the following growth rate relationship between quarterly and monthly
GDP.

ðqÞ 1 ðmÞ 1 ðmÞ 1 ðmÞ

zt ¼ zt þ zt1 þ zt2 : (8)
3 3 3

Consistent with Mariano and Murasawa (2003), Schorfheide and Song (2015),
Shang and Zheng (2018), we also assume that the aggregate of quarterly GDP is 3
times the geometric mean of unobservable monthly GDP. And then we can show
Equation (8).
Combining Equations (6)–(8), we can write the mixed-frequency term structure
model as the following state-space model. Equation (9) represents the measurement
equation, Equation (10) is the state equation, and Equation (11) contains the infor-
mation of the error terms.
6 Y. SHANG ET AL.

0 1
0 1 ft
B Pt
ðmÞ C 0 1
yt B C
B ðmÞ C 0 1B ðmÞ C B ey, t C
B Pt C KN3 0N3 0N3 B Mt C B 0 C
B ðmÞ C B C B
B Mt C ¼ @ 013 I33 013 AB
ðmÞ
Gt CþB 0 C C (9)
B C B C B
B GðmÞ C 013 013 K13 B ðmÞ C @ 0 C A
@ t A B Zt C
ðqÞ B ðmÞ C 0
Zt @ Zt1 A
ðmÞ
Zt2

0 1 0 1 0 1
ft 0 1 ft1 gf , t
B ðmÞ C lf B ðmÞ C B C
B Pt C B C B Pt1 C B ðmÞ
gp, t C
B ðmÞ C B lp C 0 1B
B ðmÞ
C B
C B
C
C
B Mt C B ls C U77 072 B Mt1 C B
ðmÞ
gs, t C
B C B C
B ðmÞ
Gt C¼B lg C þ @ 016 111 011 011 AB
B
ðmÞ
Gt1
C B
CþB ðmÞ C
C
B C B C B C B gg , t C
B ðmÞ C B lz C 016 011 111 011 B ðmÞ
C B ðmÞ C
B Zt C B C B Zt1 C B gz , t C
B ðmÞ C @ 0 A B ðmÞ C B C
@ Zt1 A @ Zt2 A @ 0 A
ðmÞ 0 ðmÞ
Zt2 Zt3 0
(10)

! 


ey, t 0 H 0
ðmÞ WN , , (11)
gt 0 0 X

 0 ðmÞ
where the state variable ft ¼ Lt St Ct is the 3  1 term structure factor. gt ¼
½ gf , t gðmÞ
x, t
ðmÞ 0
gz, t  is an error term vector of unobservable factors. The matrix K is
dependent on the parameter k in the measurement equation. K ¼ 1=3 1=3 1=3
ðmÞ
controls the relationship between the quarterly and monthly frequency dates. Pt is
ðmÞ ðmÞ
monthly CPI information. Mt is monthly monetary policy variable. Gt is monthly
fiscal policy observation.
According to Diebold, Rudebusch, and Aruoba (2006), we assume that the covari-
ance matrix of the error terms in the measurement equation is a diagonal matrix.
Therefore, we have H ¼ r2 I, where I is the identity matrix. The parameters k and r2
in the measurement equation need to be estimated. In the state equation, we estimate
the intercept term l ¼ lf lx lz , coefficient matrix U and covariance matrix X:
ðmÞ T
We also extract the unobservable factors fLt , St , Ct , zt gt¼1 from the mixed-frequency
data information.

3. Data
This paper focuses on the government bond, fiscal and other macroeconomic varia-
bles for China. The yields of government bonds are relatively high-frequency monthly
observations. Macroeconomic variable includes the inflation rate, government expend-
iture, monetary supply and gross domestic product (GDP). The inflation rate, govern-
ment expenditure and monetary supply are monthly observations. GDP is quarterly
 JOURNAL OF THE ASIA PACIFIC ECONOMY 7

Figure 1. The yield curve of Chinese Government bonds.

macro-information in China. Therefore, many mixed-frequency data values are col-

lected for empirical study.
For Chinese government bonds, the number of maturities that we choose is 13.
The maturities of the bonds are 3,691,224,364,860,728,496,108 and 120 months. The
sample period is from January 2002 to March 2019. We collect the yield data from
the WIND database. Figure 1 shows detailed information about the yield curve of
Chinese government bonds.
The inflation rate, government expenditure, monetary supply and GDP are col-
lected for macroeconomic variables. The data come from China Monthly Economic
Indicators and the National Bureau of Statistics of China (NBS). Figure 2 shows the
mixed-frequency information about the macroeconomic variables.
First, the inflation rate is sampled at a monthly frequency. It usually has an influ-
ence on the nominal yield. This paper considers the monthly year-on-year change
rate of the CPI as a proxy variable for the inflation rate of China.
Second, the monthly year-on-year change rate of M2 is computed as a proxy vari-
able of Chinese monetary policy. The quantitative monetary policy instrument plays a
more important role than the priced monetary policy instrument in China.
Third, government expenditure can be observed at monthly frequency. we consider
it as a proxy variable of Chinese fiscal policy. The year-on-year change rate of macro
variable is usually published by the Chinese official organization. The year-on-year
rate can avoid the seasonal adjustment of month on month rate. Therefore, the
monthly year-on-year change rate of Chinese government expenditures is computed.
We delete the samples from January in some years, since these observations are out-
liers. These outliers are driven by the Chinese Spring Festival effect. The outliers do
not reflect the economic rule that we focused on.
The sample period of the above three variables is from January 2002 to March
2019. Finally, Chinese quarterly GDP data are available. This is a crucial index to
8 Y. SHANG ET AL.

Figure 2. The mixed-frequency macroeconomic information.

describe the business cycle of the Chinese real economy. We use the quarterly year-
on-year growth rate of GDP as a proxy variable for economic growth. The sample
period is from the first quarter of 2002 to the first quarter of 2019.

4. Empirical results
In this section, we investigate the role of fiscal behavior in yield curve fitting with a
mixed-frequency term structure model. The effects of fiscal behavior on term struc-
ture are also studied with impulse response and variance decomposition in this paper.
In addition, we examine the importance of fiscal behavior for yield curve prediction.

4.1. Parameter estimation results

Table 1 shows estimates of parameters with MCMC method in Panel A and Panel B.
First, both fiscal and monetary behaviors have a significant effect on the level factor.
This implies that macroeconomic policy mainly has an influence on long-run yield in
China. Second, at a significance level of 5%, the level factor has a significant positive
influence on inflation. This means that the term structure factor contains the inflation
expectation information. In particular, the slope factor has a significant negative
impact on monetary policy. Additionally, the curvature factor has a significant nega-
tive impact on fiscal policy. This implies that the term of the yield curve plays an
important role in macroeconomic policy. Unlike monetary policy, fiscal policy is
mainly affected by the longer-term yield. In Panel B of Table 1, we also display the
 JOURNAL OF THE ASIA PACIFIC ECONOMY 9

Table 1. The parameter estimates of the mixed-frequency model with fiscal policy.
Panel A: Model parameter estimates
0.0704 r2 0.000001
ðmÞ ðmÞ ðmÞ ðmÞ
k Lt1 St1 Ct1 Pt1 Mt1 Gt1 Zt1 l
Lt 0.9107 0.0902 –0.0045 0.0136 0.0167 –0.0055 0.0035 0.2391
St –0.0299 0.6745 0.0753 0.0193 –0.0112 –0.0013 –0.0172 0.0671
Ct 0.1206 0.0841 0.7765 –0.0025 –0.0182 0.0002 –0.0007 –0.4009
ðmÞ
Pt 0.1842 –0.0031 0.1861 0.8765 0.0165 –0.0078 0.0908 –1.0946
ðmÞ
Mt –0.1392 –0.4302 –0.0654 –0.1219 0.9051 0.0537 –0.0250 0.9422
ðmÞ
Gt –0.7431 –0.1496 –0.5271 0.3396 –0.0771 0.8359 0.1594 3.3306
ðmÞ
Zt 0.1033 –0.2811 –0.0734 –0.0706 –0.0170 0.0316 0.8783 0.1194
Panel B: Covariance matrix estimates
ðmÞ ðmÞ ðmÞ ðmÞ
X Lt St Ct Pt Mt Gt Zt
Lt 0.0688
St –0.0710 0.1639
Ct –0.1035 0.1341 0.4298
ðmÞ
Pt 0.0095 0.0144 0.0232 0.3143
ðmÞ
Mt 0.0002 –0.0109 –0.0407 –0.0186 0.7951
ðmÞ
Gt –0.0138 0.0050 0.0102 –0.0287 0.6110 6.9984
ðmÞ
Zt –0.0364 0.0817 0.1629 0.0924 0.2244 –1.0044 0.8336
The significance of bold at the 5% level.

estimated covariance matrix X: Many results of the off-diagonal covariance appear to

be significant individually. We find that there are significant variance correlations
among the term structure factors. However, the macroeconomic variables such as fis-
cal and monetary policy have no significant variance correlations with term struc-
ture factors.
Note: k and r2 are parameters in the measurement equation., Lt , St , and Ct are
ðmÞ
the level, slope, and curvature factor, respectively; Pt represents the inflation rate,
ðmÞ ðmÞ ðmÞ
Mt is the monetary policy variable, Gt is the fiscal policy variable, Zt is the
monthly GDP latent factor. l is the intercept in the state equation, and X is the
covariance matrix of the state equation. The estimation results are the median of the
Bayesian posterior estimation. Bold entries denote parameter estimates significant at
the 5% level. The MCMC sampling for the posterior distribution is 10,000 times. The
initial 2,000 samples are discarded.

4.2. Fiscal policy and in-sample fitting

This paper investigates the role of fiscal behavior for in-sample fitting results. For the
benchmark model, we take CPI, monetary supply and GDP as macro-variables. For
the mixed-frequency model, CPI and quarterly GDP are chosen as macro-observa-
tions. This benchmark model is referred to as the MF_NS_OF model since it does
not include fiscal variables. Meanwhile, the model with fiscal variables is abbreviated
as the MF_NS_F model. This paper uses the MAE to evaluate the in-sample fitting
results. The MAE is expressed asMAEðsn Þ ¼ T 1 RTt¼1 j^y ðsn Þt yðsn Þt j, where ^y ðsn Þt
indicates the fitting value of the yield curve with maturity sn and yðsn Þt represents
the true value of the yield curve with maturity sn :
10 Y. SHANG ET AL.

Figure 3. In-sample fitting with and without fiscal information.

Figure 3 shows the in-sample fitting results of the bond yields with different
maturities. As shown in Figure 3, except for the 24-month bond, the MF_NS_F
model has better performance on the in-sample fitting than the MF_NS_OF model.
The mixed-frequency model with fiscal variables can improve the fitting of long-term
bond yields more than that of short-term bond yields. This means that the fiscal vari-
able contains more useful information for improving the fitting results of bond yields,
and this information contributes more to long-term bond yields.
This paper also examines the estimated results of term structure factors to further
evaluate the fitting performance of the term structure model with fiscal variables.
Referring to Diebold and Li (2006), the level factor Lt is considered a long-term fac-
tor, the slope factor St is considered a short-term factor, and the curvature factor Ct
is considered a medium-term factor. The dynamic term structure factor level, slope
and curvature have specific economic meaning, since they play important roles in
describing the features of the term structure. They are also important bridges to con-
nect macroeconomic variables such as monetary and fiscal indices. The term structure
factors are not observable. These factors are strongly dependent on the term structure
model. Therefore, the fitting effect of the term structure factors can help to determine
the model specification. Many papers aim to find and compute suitable empirical
proxies for these unobservable term structure factors. They use these proxies to evalu-
ate whether term structure factors make sense.
In Figure 4, we plot the respective term structure factors with various empirical
proxies. As shown in Figure 4, we find that the estimated level factor is closely linked
to a comparison series. This link is a common theme in recent term structure litera-
ture, including Rudebusch and Wu (2003), Diebold and Li (2006), and Shang and
Zheng (2018). We also present the estimated slope factor together with a standard
 JOURNAL OF THE ASIA PACIFIC ECONOMY 11

Figure 4. Estimated term structure factors and empirical proxies.

empirical proxy in Figure 4. Similar to the level factor, the slope factor is closely
linked to its corresponding empirical proxy. Finally, we show the estimated curvature
factor and a closely linked comparison series: a standard empirical proxy for the
curvature factor. In sum, the time series of the unobservable term structure factors
estimated with fiscal variables are strongly similar to those of the empirical proxies.
These results mean that the fiscal variable has better performance in describing the
features of the term structure. In summary, fiscal policy information contributes to
the in-sample fitting performance of both the yield curve and term structure factors.

4.3. Fiscal behavior and term structure factors

We now investigate the relationship between fiscal behavior and term structure fac-
tors via impulse response functions, which are shown in Figures 5 and 6. First, the
response dynamics of both fiscal and monetary behavior for the term structure factors
are examined in Figure 5.
As shown in Figure 5, the level factor shows a significant response to both fiscal
and monetary behavior. The response of level factor to monetary policy shock is
increasing and then falling after 10 months. The surprise increase in the fiscal shock
leads to a decreasing in the level factor that is last for about 15 months. However, the
slope and curvature factors show very little response to the macroeconomic impulse.
This implies that fiscal and monetary policies mainly contribute to long-run yields in
the Chinese bond markets. This is consistent with the view that the longer yields
tend to be associated with fiscal sustainability (Afonso and Martins 2012).
12 Y. SHANG ET AL.

Figure 5. The response of term structure factors to macroeconomic policy.

Figure 6. The response of macroeconomic policy to term structure factors.

Furthermore, the response of level factor to monetary impulse is positive, while

the response to fiscal surprise is negative in Figure 5. This means that monetary policy
shows an inflation expectation effect for level factors. An increase in monetary supply
will raise the inflation expectation and then increase the nominal return of financial
assets. Rudebusch and Wu (2003), Diebold, Rudebusch, and Aruoba (2006) also find
this long-inflation expectations mechanism for level factor. However, proactive fiscal
behavior usually produces an economic boom in China. This result will generate the
wealth effect that increases the demand for government bonds and thus decreases the
 JOURNAL OF THE ASIA PACIFIC ECONOMY 13

yield return. On the other hand, economic expansion will also reduce the risk premium
of financial assets, since a part of the risk comes from the business cycle. In sum, the
fiscal and monetary policy can produce significant impulse to the level of term struc-
ture. However, these behaviors show different mechanisms for the level factor.
Second, we consider the response of the yield curve to macroeconomic policy. As
shown in Figure 6, both fiscal and monetary policy respond to a negative level factor.
An increase in the level factor lead to tight monetary policy over the next 20 months.
Similarly, the response of fiscal policy to level factor shock also tends to be tight. But,
the time of duration of tight fiscal policy is less than that of monetary policy.
A surprise increase in the slope factor will also produce tight monetary policy.
And the shock of the slope factor to monetary policy is larger than that of level fac-
tor. It shows that slope factor contains more future information for monetary policy.
This result is consistent with the view that there is a close connection between the
slope factor and the instrument of monetary policy (Diebold, Rudebusch, and Aruoba
2006). However, Figure 6 shows that the shock of the slope factor to fiscal policy is
not significant. This implies that only short-term yield shows little connection with
fiscal policy in China.
In Figure 6, we further find that shocks to the curvature of the term structure also
affect the macroeconomic policy. A surprise decrease in the curvature factor raises
monetary policy. Specifically, the fiscal policy response to the curvature factor is sig-
nificantly negative. The shocks to the curvature factor are similar to that of level fac-
tor. But, effect of curvature factor is less than that of level factor. These responses
mean that fiscal policy is manly affected by medium- and long-term yields in China.

4.4. Fiscal behavior and the yield curve

In addition to the impulse response results, we also provide variance decompositions
for analyzing fiscal behavior and yield curve interactions. Table 2 reports variance
decompositions for the 3-, 6-, 24-, 60-, 84-, and 120-month yields at forecast horizons
of 51,530 and 60 months.
First, we discuss the contributions of the variance decomposition that come from
the term structure factors. Given any forecast horizon, the level factor shows a greater
contribution to the yield curve with an increase in maturity. In contrast, the slope
factor exhibits less influence when the time duration of maturity increases. The
curvature factor mainly contributes to the forecast variance of the medium-term
yield. The forecast variance of the level factor is mainly influenced on its own. These
results show the properties of classical term structure factors and are consistent with
the term structure literature.
Next, the contribution of macroeconomics to variance decomposition is investi-
gated. Given any maturity of the bond, we find that fiscal policy always provides a
greater contribution to the variance decomposition of the yield curve than monetary
policy in the short forecast horizon. However, monetary policy accounts for more
variance in the yield curve than fiscal policy in the long forecast horizon. In the short
forecast horizon, we find that fiscal policy has a greater contribution to long-term
yield than to short-term yield. With an increase in the forecast horizon, a greater
14 Y. SHANG ET AL.

Table 2. The variance decomposition of the yield curve.

ðmÞ ðmÞ ðmÞ ðmÞ
Term Horizon Lt St Ct Pt Mt Gt Zt
3-M 5 0.0019 0.8421 0.1297 0.0123 0.0001 0.0114 0.0024
10 0.0055 0.7141 0.2199 0.0224 0.0022 0.0325 0.0035
30 0.0425 0.5923 0.2538 0.0208 0.0451 0.0396 0.0059
60 0.0478 0.5685 0.2510 0.0203 0.0584 0.0478 0.0063
6-M 5 0.0028 0.7773 0.1943 0.0115 0.0001 0.0121 0.0019
10 0.0071 0.6646 0.2685 0.0211 0.0023 0.0335 0.0029
30 0.0457 0.5542 0.2880 0.0197 0.0464 0.0405 0.0055
60 0.0509 0.5318 0.2833 0.0193 0.0600 0.0488 0.0059
24-M 5 0.0179 0.5415 0.4142 0.0095 0.0003 0.0159 0.0006
10 0.0314 0.4978 0.4088 0.0174 0.0058 0.0376 0.0013
30 0.0802 0.4230 0.3703 0.0159 0.0650 0.0414 0.0042
60 0.0828 0.4063 0.3610 0.0156 0.0777 0.0519 0.0046
60-M 5 0.3198 0.3405 0.3030 0.0091 0.0052 0.0224 0.0001
10 0.2799 0.3465 0.2925 0.0146 0.0252 0.0407 0.0004
30 0.2889 0.2866 0.2491 0.0124 0.1188 0.0414 0.0028
60 0.2687 0.2813 0.2537 0.0121 0.1241 0.0564 0.0037
84-M 5 0.5373 0.2422 0.1817 0.0080 0.0085 0.0223 0.0000
10 0.4559 0.2668 0.1911 0.0125 0.0358 0.0377 0.0002
30 0.4218 0.2208 0.1661 0.0105 0.1381 0.0404 0.0022
60 0.3854 0.2237 0.1819 0.0103 0.1390 0.0564 0.0034
120-M 5 0.7112 0.1606 0.0902 0.0066 0.0111 0.0204 0.0000
10 0.6094 0.1943 0.1092 0.0101 0.0439 0.0330 0.0001
30 0.5373 0.1633 0.1014 0.0087 0.1486 0.0391 0.0016
60 0.4872 0.1738 0.1259 0.0087 0.1461 0.0551 0.0032

portion of the variance contributions of the yield curve come from fiscal behavior.
Finally, the inflation factor accounts for higher variance than the GDP factor. This
implies that the inflation expectation is important for the future volatility of the yield
curve in China.
Note: The term indicates the maturity of the yield curve. The horizon is the fore-
cast time of variance decomposition. Lt , St , and Ct are the level, slope, and curvature
ðmÞ ðmÞ
factors, respectively. Pt represents the inflation rate, Mt is the monetary policy
ðmÞ ðmÞ
variable, Gt is the fiscal policy variable, and Zt is the monthly GDP latent factor.

4.5. Fiscal behavior and out-of-sample forecasting

This paper will further investigate the out-of-sample forecasting performance of the
term structure model with fiscal behavior. Similar to the study of in-sample fitting,
this section compares the out-of-sample performance between the MF_NS_OF model
and MF_NS_F model. To mitigate the negative effect of the business cycle, this paper
will use the rolling-window approach to compute the results of the out-of-sample
forecast. The number of estimated samples is fixed in the rolling-window approach.
However, the sample itself is changed each time. The initial estimation period uses
only data available from January 2002 to December 2015. The out-of-sample forecast
period is from January 2016 to March 2019. We first estimate the parameters of the
term structure models with a Bayesian approach. Then, the forecast values of the
yield curve are computed at the posterior medians.
Based on Shang and Zheng (2018), this paper calculates the FMAE based on two
different dimensions (cross-section and time-series) to compare the out-of-sample
forecast results. At each forecast time, we get the FMAE of bonds with different
 JOURNAL OF THE ASIA PACIFIC ECONOMY 15

Figure 7. The forecasted results of yield curve.

maturities. The FMAE is computed by taking the average of the absolute forecast
error of the bond yield (FMAEsi ¼ T 1 RTt¼1 j^y ðsi Þt yðsi Þt j), where ^y ðsi Þt is the fore-
cast value of the bond yield with maturity si at out-of-sample time t: yðsi Þt is the
true value of the bond yield with maturity si at time t: j^y ðsi Þt yðsi Þt j is the absolute
forecast error of the bond yield with maturity si at out-of-sample time t: We take T
equal 39, which is the total number of the out-of-sample forecast time t:
This paper first provides the forecasted results in a 3-D graph. We can directly
and clearly observe the forecast performance. As shown in Figure 7, the 1-month-
ahead forecasting results of yield curve is from January 2016 to March 2019. The
forecasted yield curves are similar to the corresponding actual results as shown in
Figure 1. The dynamics of yield curve can be predicted with our mixed frequency
model. Based on these forecasted results, we further investigate the contribution of
fiscal policy to yield curve forecast.
Table 3 reports the out-of-sample forecast results for different maturities with the
rolling window approach. This paper mainly focuses on the forecast results for short
horizons, such as 1 month, 2 months and 3 months. We first investigate the 1-month-
ahead forecast results of the yield curves. Except for the bonds at maturities of 3 and
12 months, the forecast results of the bonds show that the forecasting performance of the
MF_NS_F model is better than that of the MF_NS_OF model. This suggests that the
fiscal variable can significantly improve the forecasting performance. Compared with the
short-term bond, the fiscal variable shows a greater contribution to the long-term bond.
Similar to the 1-month-ahead forecasting results, the 2- and 3-month-ahead forecast-
ing results suggest that the MF_NS_F model is superior to that of the benchmark model.
For the 2-month-ahead forecasting results, 11 of the 13 FMAE results of the MF_NS_F
model are superior to those of the MF_NS_OF model. For the 3-month-ahead forecast-
ing results, only 1 FMAE result of the MF_NS_F model is inferior to an FMAE result of
the MF_NS_OF model. This reveals that the forecasts of the term structure model with
the fiscal variable show robust results as the forecast horizon lengthens.
Additionally, we test the forecasting power of the above two models with Diebold
and Mariano (1995) statistics. The null hypothesis of Diebold and Mariano (1995)
statistics is that the two forecasts have the same mean-squared error. This paper
16 Y. SHANG ET AL.

Table 3. FMAE (FMAEsi ) results of the yield curve.

H¼1 H¼2 H¼3
Maturity (month) MF_NS_F MF_NS_OF MF_NS_F MF_NS_OF MF_NS_F MF_NS_OF
3 20.11 18.25 28.41 27.30 31.12 38.86
6 15.11 23.05 23.89 31.07 29.13 30.50
9 15.31 15.91 22.73 24.16 29.18 28.77
12 15.93 14.27 23.67 22.99 26.93 27.50
24 14.43 15.95 22.04 22.65 24.61 24.32
36 12.37 13.49 19.08 19.39 23.25 23.32
48 11.33 11.93 17.27 17.98 22.29 22.84
60 11.17 11.49 16.90 17.73 22.17 24.62
72 11.16 13.22 16.70 19.41 21.56 26.46
84 10.90 16.59 16.46 22.25 20.71 26.07
96 9.54 16.18 15.13 21.67 20.30 25.27
108 9.03 15.17 14.56 20.69 20.41 24.93
120 9.25 14.78 14.68 20.27 31.12 38.86
Note: Results in boldface indicate that the corresponding model has better performance in out-of-sample forecast-
ing. MF_NS_F is the mixed-frequency affine model with fiscal variables, and MF_NS_OF is the term structure model
without fiscal variables. H ¼ 1 indicates the 1-month-ahead forecast value, H ¼ 2 indicates the 2-month-ahead fore-
cast value, and H ¼ 3 indicates the 3-month-ahead forecast value.

shows Diebold–Mariano (D-M) tests of our MF_NS_F model forecasts against those
of the MF_NS_OF model. As shown in Table 4, we present the Diebold and Mariano sta-
tistics and corresponding P-values. The negative values of D-M shown in Table 4 indicate
the superiority of the MF_NS_F model forecasts. For any forecast horizon, we find that
the long-term forecast results of the MF_NS_F model are significantly superior to those
of the MF_NS_OF model at the 1% level. However, for the short-term and medium-term
yield curves, only the Diebold–Mariano statistic of the 6-month yield shows the significant
superiority of the MF_NS_F model forecasts. This result further shows that fiscal behavior
mainly contributes to the forecast of long-term bonds in China.
In addition, this paper investigates the features of forecasts from the perspective of
the time series. We also compute the FMAE of each forecast time by taking the aver-
age of the absolute forecast error of the bond yield with all maturities, FMAEt ¼
N 1 RNi¼1 j^y ðsi Þt yðsi Þt j, where N is the total number of bonds with different matur-
ities. We further define DFMAEt as the difference between the FMAEt values of the
above two models. If the value of DFMAEt is greater than 0, the term structure model
with the fiscal variable shows better out-of-sample forecast performance at time t.
Figure 8 plots the DFMAE results for different forecast horizons. As shown in
Figure 8, H ¼ 1 represents the DFMAE results for 1 month ahead. H ¼ 2 and H ¼ 3
show the DFMAE results for 2 months ahead and 3 months ahead, respectively. For
the DFMAE for 1 month ahead, we clearly see that the majority of the DFMAE results
are greater than 0. This implies that the predictive power of the term structure model
with fiscal variables outweighs that of the benchmark model in most forecast periods.
The DFMAE results for 2 months ahead and 3 months ahead add to the robust results
of the 1-month-ahead values. This further suggests that we can employ fiscal variables
to improve the out-of-sample forecast of the yield curve.

5. Summary and conclusions

What is the mechanism relating fiscal policy and term structure to mixed-frequency
macroeconomic information? Does the fiscal policy variable help forecast the yield
 JOURNAL OF THE ASIA PACIFIC ECONOMY 17

Table 4. The results of the Diebold–Mariano forecast accuracy tests.

H¼1 H¼2 H¼3
Maturity (month) DM P-value DM P-value DM P-value
3 2.1876 0.0349 0.8431 0.4046 –0.2096 0.8352
6 –3.7970 0.0005 –3.4725 0.0013 –3.5821 0.0010
9 –1.2607 0.2151 –1.7184 0.0941 –2.1364 0.0395
12 0.8271 0.4133 1.4470 0.1563 1.5091 0.1400
24 –0.9388 0.3538 0.6067 0.5477 1.2755 0.2103
36 –0.3798 0.7062 0.6540 0.5172 1.2976 0.2027
48 –0.5356 0.5954 0.1614 0.8726 0.6779 0.5022
60 –0.8390 0.4067 –0.8476 0.4021 –0.7684 0.4473
72 –3.7918 0.0005 –3.1740 0.0030 –2.7877 0.0084
84 –5.8801 0.0000 –4.8975 0.0000 –4.1508 0.0002
96 –5.5102 0.0000 –4.6481 0.0000 –3.9666 0.0003
108 –4.6142 0.0000 –4.0140 0.0003 –3.4928 0.0013
120 –3.9519 0.0003 –3.5483 0.0011 –3.1415 0.0034
Note: D-M is the Diebold–Mariano statistic.
denotes significance relative to the asymptotic null distribution at the 10% level.
denotes significance relative to the asymptotic null distribution at the 5% level.
denotes significance relative to the asymptotic null distribution at the 1% level.

Figure 8. DFMAE Results in different out-of-sample horizons.

curve? These issues are not well addressed in the literature. In this paper, we intro-
duce fiscal policy into the mixed frequency Nelson–Siegel (MF-NS) model and inves-
tigate the contribution of fiscal variables to the term structure factor, in-sample fit
and out-of-sample forecast. We first investigate the influence of fiscal policy on the
in-sample fit of the yield curve and describe the shape of the yield curve. Then, with
the impulse-response and variance-decomposition methods, this paper also analyzes
the effects of fiscal behavior on term structure and bond yields. Finally, we discuss
out-of-sample forecasting through mixed-frequency models with fiscal information,
18 Y. SHANG ET AL.

and we test the power of our mixed-frequency model forecasts with Diebold and
Mariano statistics.
The main conclusions that are drawn in this paper are as follows: first, the mixed-
frequency term structure model with fiscal policy information has better performance
on in-sample fitting than the benchmark model does. More importantly, the mixed-
frequency model with fiscal variables can improve the fitting of long-term bond yields
more than short-term bond yields. This means that the fiscal variable helps to
improve the fitting results of bond yields and contributes more to long-term
bond yields.
Second, fiscal and monetary policies mainly contribute to long-run yields in the
Chinese bond markets. Furthermore, the response of this factor to monetary impulse
is positive, while the response to fiscal surprise is negative. We find that monetary
policy shows an inflation expectation effect with level factors. However, fiscal policy
shocks influence the level factor via the wealth effect.
Third, given any bond maturity, we find that fiscal policy always contributes more
to the variance decomposition of the yield curve than monetary policy does in the
short forecast horizon. In the short forecast horizon, this paper shows that fiscal pol-
icy has a greater contribution to long-term yield than short-term yield.
Finally, we show that the fiscal variable can significantly improve the performance
of out-of-sample forecasting. Compared with the short-term bond, the fiscal variable
shows a greater contribution to the long-term bond. The D-M test proves that fiscal
variables can significantly improve the out-of-sample results of the long-term bond
yield. The DFMAE results imply that the predictive power of the term structure
model with fiscal variables outweighs that of the benchmark model in most fore-
cast periods.

Note
1. The value of P indicates the number of higher frequencies in one period of low frequency.
The value of P is affected by the type of data, such as flow data or stock data.

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
This study was funded by National Office for Philosophy and Social Sciences; the National
Natural Science Foundation of China (no.71701165, 71950010); and the National Social
Science Foundation of China (no.20BJY255, no.20&ZD081).

Notes on contributors
Yuhuang Shang is at Institute of Chinese Financial Studies, Southwestern University of
Finance and Economics, Chengdu, China. Research interest: term structure; mixed fre-
quency modeling.
 JOURNAL OF THE ASIA PACIFIC ECONOMY 19

Xuyang Zhang is at Institute of Chinese Financial Studies, Southwestern University of Finance

and Economics, Chengdu, China. Research interest: monetary policy.
Qing Wang is at Institute of Chinese Financial Studies, Southwestern University of Finance
and Economics, Chengdu, China. Research interest: macro-finance; monetary policy.

References
Afonso, A. 2010. “Long-Term Government Bond Yields and Economic Forecasts: Evidence for
the EU.” Applied Economics Letters 17 (2): 105–109. doi:10.1080/13504850903049627.
Afonso, A., and M. M. F. Martins. 2012. “Level, Slope, Curvature of the Sovereign Yield
Curve, and Fiscal Behaviour.” Journal of Banking & Finance 36 (6): 1789–1807. doi:10.1016/
j.jbankfin.2012.02.004.
Afonso, A., and C. Rault. 2010. Long-Run Determinants of Sovereign Yields. CESIfo Working
Paper 3155.
Ang, A., G. Bekaert, and M. Wei. 2008. “The Term Structure of Real Rates and Expected
Inflation.” The Journal of Finance 63 (2): 797–849. doi:10.1111/j.1540-6261.2008.01332.x.
Ang, A., J. Boivin, S. Dong, and R. Loo-Kung. 2011. “Monetary Policy Shifts and the Term
Structure.” The Review of Economic Studies 78 (2): 429–457. doi:10.1093/restud/rdq006.
Ang, A., and M. Piazzesi. 2003. “A No-Arbitrage Vector Autoregression of Term Structure
Dynamics with Macroeconomic and Latent Variables.” Journal of Monetary Economics 50
(4): 745–787. doi:10.1016/S0304-3932(03)00032-1.
Auclert, A. 2019. “Monetary Policy and the Redistribution Channel.” American Economic
Review 109 (6): 2333–2367. doi:10.1257/aer.20160137.
Bikbov, R., and M. Chernov. 2013. “Monetary Policy Regimes and the Term Structure of
Interest Rates.” Journal of Econometrics 174 (1): 27–43. doi:10.1016/j.jeconom.2013.01.002.
Boragan Aruoba, S. 2020. “Term Structures of Inflation Expectations and Real Interest Rates.”
Journal of Business & Economic Statistics 38 (3): 542–521. doi:10.1080/07350015.2018.
1529599.
Bernoth, K., J. von Hagen, and L. Schuknecht. 2006. Sovereign Risk Premiums in the European
Government Bond Market. GESY Discussion Paper No. 151.
Canzoneri, M. B., R. E. Cumby, and B. Diba. 2002. “Should the European Central Bank and
the Federal Reserve Be Concerned about Fiscal Policy?” Rethinking Stabilization Policy
29–31.
Dai, Q., and T. Philippon. 2005. Fiscal Policy and the Term Structure of Interest Rates. (No.
w11574) National Bureau of Economic Research.
Diebold, F. X., and C. Li. 2006. “Forecasting the Term Structure of Government Bond Yields.”
Journal of Econometrics 130 (2): 337–364. doi:10.1016/j.jeconom.2005.03.005.
Diebold, F. X., and R. S. Mariano. 1995. “Comparing Predictive Accuracy.” Journal of Business
& Economic Statistics 13 (3): 253–263. doi:10.2307/1392185.
Diebold, F. X., G. D. Rudebusch, and S. B. Aruoba. 2006. “The Macroeconomy and the Yield
Curve: A Dynamic Latent Factor Approach.” Journal of Econometrics 131 (1–2): 309–338.
doi:10.1016/j.jeconom.2005.01.011.
Favero, C. A., L. Niu, and L. Sala. 2012. “Term Structure Forecasting: No-Arbitrage
Restrictions versus Large Information Set.” Journal of Forecasting 31 (2): 124–156. doi:10.
1002/for.1181.
urkaynak, R. S., B. Sack, and J. H. Wright. 2010. “The TIPS Yield Curve and Inflation
G€
Compensation.” American Economic Journal: Macroeconomics 2 (1): 70–92. doi:10.1257/mac.
2.1.70.
Haugh, D., P. Ollivaud, and D. Turner. 2009. What Drives Sovereign Risk Premiums. OECD
Economics Department Working Paper 718.
Hauner, D., and M. Kumar. 2011. “Interest Rates and Budget Deficits Revisited-Evidence from
the G7 Countries.” Applied Economics 43 (12): 1463–1475. doi:10.1080/00036840802600574.
20 Y. SHANG ET AL.

Huse, C. 2011. “Term Structure Modelling with Observable State Variables.” Journal of
Banking & Finance 35 (12): 3240–3252. doi:10.1016/j.jbankfin.2011.05.004.
Jardet, C., A. Monfort, and F. Pegoraro. 2013. “No-Arbitrage Near-Cointegrated VAR (p)
Term Structure Models, Term Premia and GDP Growth.” Journal of Banking & Finance 37
(2): 389–402. doi:10.1016/j.jbankfin.2012.09.003.
Joslin, S., M. Priebsch, and K. J. Singleton. 2014. “Risk Premiums in Dynamic Term Structure
Models with Unspanned Macro Risks.” The Journal of Finance 69 (3): 1197–1233. doi:10.
1111/jofi.12131.
Kung, H. 2015. “Macroeconomic Linkages between Monetary Policy and the Term Structure
of Interest Rates.” Journal of Financial Economics 115 (1): 42–57. doi:10.1016/j.jfineco.2014.
09.006.
Laubach, T. 2009. “New Evidence on the Interest Rate Effects of Budget Deficits and Debt.”
Journal of the European Economic Association 7 (4): 858–885. doi:10.1162/JEEA.2009.7.4.858.
Marsal, A., L. Kaszab, and R. Horvath. 2016. Fiscal Policy and the Term Structure of Interest
Rates in a DSGE Model. FinMaP-Working Paper.
Mariano, R. S., and Y. Murasawa. 2003. “A New Coincident Index of Business Cycles Based
on Monthly and Quarterly Series.” Journal of Applied Econometrics 18 (4): 427–443. doi:10.
1002/jae.695.
Nelson, C., and A. Siegel. 1987. “Parsimonious Modelling of Yield Curve.” The Journal of
Business 60 (4): 473–489. doi:10.1086/296409.
Nickel, C., P. Rother, and J. R€ ulke. 2009. Fiscal Variables and Bond Spreads-Evidence from
Eastern European Countries and Turkey. European Central Bank Working Paper 1152.
Paccagnini, A. 2016. “The Macroeconomic Determinants of the US Term Structure during the
Great Moderation.” Economic Modelling 52: 216–225. doi:10.1016/j.econmod.2014.11.013.
Rogers, J. H., C. Scotti, and J. H. Wright. 2018. “Unconventional Monetary Policy and
International Risk Premia.” Journal of Money, Credit and Banking 50 (8): 1827–1850. doi:10.
1111/jmcb.12511.
Rudebusch, G., and T. Wu. 2003. “A No-Arbitrage Model of the Term Structure and the
Macroeconomy.” manuscript, Federal Reserve Bank of San Francisco.
Rudebusch, G. D., and T. Wu. 2008. “A Macro-Finance Model of the Term Structure,
Monetary Policy and the Economy.” The Economic Journal 118 (530): 906–926. doi:10.1111/
j.1468-0297.2008.02155.x.
Schorfheide, F., and D. Song. 2015. “Real-Time Forecasting with a Mixed-Frequency VAR.”
Journal of Business & Economic Statistics 33 (3): 366–380. doi:10.1080/07350015.2014.
954707.
Schuknecht, L., J. von Hagen, and G. Wolswijk. 2010. Government Bond Risk Premiumsin the
EU Revisited –The Impact of the Financial Crisis. European Central Bank Working Paper
1152.
Sgherri, S., and E. Zoli. 2009. Euro Area Sovereign Risk During the Crisis. International
Monetary Fund Working Paper 09/222.
Shang, Y., and T. Zheng. 2018. “Fitting and Forecasting Yield Curves with a Mixed-Frequency
Affine Model: Evidence from China.” Economic Modelling 68: 145–154. doi:10.1016/j.econ-
mod.2017.07.002.
Ulrich, M. 2013. “Inflation Ambiguity and the Term Structure of US Government Bonds.”
Journal of Monetary Economics 60 (2): 295–309. doi:10.1016/j.jmoneco.2012.10.015.
Wright, J. H. 2011. “Term Premia and Inflation Uncertainty: Empirical Evidence from an
International Panel Dataset.” American Economic Review 101 (4): 1514–1534. doi:10.1257/
aer.101.4.1514.
Yan, Y., and J. Guo. 2018. “The Sovereign Yield Curve and Fiscal Behaviour in China.” Pacific
Economic Review 23 (2): 271–295. doi:10.1111/1468-0106.12159.