See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/227621546

Macroeconomic factors and yield curve for the emerging Indian economy

Article  in  Macroeconomics and Finance in Emerging Market Economies · March 2011

DOI: 10.1080/17520843.2011.548612 · Source: RePEc

CITATIONS READS

9 142

1 author:

Kakali Kanjilal

27 PUBLICATIONS   432 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Dynamics of crude oil and gold price post 2008 global financial crisis – New evidence from threshold vector error-correction model View project

All content following this page was uploaded by Kakali Kanjilal on 01 April 2020.

The user has requested enhancement of the downloaded file.

Macroeconomics and Finance in Emerging Market Economies
Vol. 4, No. 1, March 2011, 57–83

Macroeconomic factors and yield curve for the emerging Indian economy
Kakali Kanjilal*

Indira Gandhi Institute of Development Research, Mumbai, India

(Received 30 December 2008; final version received 10 July 2009)

This article investigates the dynamic linkages between the estimated parameters
of a zero coupon yield curve and macroeconomic variables like inflation, gross
domestic product growth in the presence of a monetary policy indicator in India
for the period July 1997 to February 2004. The study finds that there exists
strong causality from financial factors, defined by three parameters of the yield
curves (‘Level’, ‘Slope’, ‘Curvature’) to macroeconomic factors; growth, inflation
and monetary policy indicators (changes in the call money rate). However, the
causality in the opposite direction is found to be weaker. It is found that
the yield and macro factors do not cause each other before the launch of a
liquidity adjustment facility, so the evidence of causality from financial to
macroeconomic factors can be attributed to the introduction of a liquidity
adjustment facility in June 2000. The causality from yield factors to macro
factors is primarily driven by the fact that the ‘changes in level’ of yield curve
brings an impact on inflation through the changes in monetary policy. This
finding suggests that monetary policy plays a key role in driving the causality.
This also implies that the indirect instrument of monetary policy mechanism is
becoming increasingly important to influence the aggregate demand in the
economy.
Keywords: term structure of interest rates; inflation, growth; monetary policy;
financial reforms; LAF; VAR modelling; Granger causality; Impulse response

1. Introduction
The article aims at exploring the dynamics of the linkage, if any, between the term
structure of interest rates and the macroeconomic factors in the Indian economy post
the financial reforms. There has been large gap in the yield curve models developed
by macroeconomists and financial economists. Macroeconomists focus more on
the role of fundamentals of economy like expectations of inflation and future
growth, unlike financial economists whose prime focus is interest rate forecasting or
bond pricing or market surveillance. Parsimonious models like Nelson-Siegel or
Svensson’s fail to offer insight into the nature of the underlying economic forces that
drive the movements of the yield curve. The changes in the yield curves or term
structure of interest rates are often interpreted as ‘Level’, ‘Slope’ and ‘Curvature’ in
the literature. However, a good economic interpretation of these factors causing the
movements in the yield curve has always remained a challenge. Movements in

*Email: kakalikanjilal@rediffmail.com

ISSN 1752-0843 print/ISSN 1752-0851 online

Ó 2011 Taylor & Francis
DOI: 10.1080/17520843.2011.548612
http://www.informaworld.com
58 K. Kanjilal

the yield curves is an outcome of the expectations of the market which capture the
changes in the key macroeconomic fundamentals; inflation, growth, monetary
policy. So the shape of the yield curve can be used as a good predictor or reflector to
effectively manage monetary policy, which would influence inflation and growth in
an economy.
There has been a growing consensus for the last two decades regarding the
relationship between the term structure of interest rates and the real activity in an
economy, although establishing a theoretical rational between the real sectors of the
economy and the financial sectors is a little complex as their connectivity cannot be
ascertained only by a pair of simple equations as can be done in the case of inflation
and the slope of the yield curve. Inflation has been a major concern of policymakers
for many years. In the recent years, the prime focus of central banks of many
countries has been the price stability in the economy. To meet this goal, central
banks require tracking the degree of inflationary movements or at least some
indicators of inflation in the country. Central banks pursuing the direct inflation
targeting strategy or an intermediate-target strategy need to monitor the indicators
of financial market expectations. These indicators could contribute to the central
bank’s own inflation forecasts. One of the common indicators used by most of the
central banks in developed countries is the term structure of interest rates. A
theoretical linkage of growth with term spread (defined as the difference between
long-term rate and short-term rate) is supported by Bernanke (1990), Bernanke and
Blinder (1992), and Lowe (1992).1
This study integrates macroeconomic factors and yield curves with two-fold
objectives; first, to investigate the dynamic linkages between macro and yield
factors, and second, to examine if the integrated macro-yield model improves the
prediction of yield curve. To explore the possibility of a central bank’s intervention
in this dynamics, an indicator for monetary policy is also introduced. The
dynamics of yield curve and macroeconomic factors cannot independently be
determined without a monetary policy indicator as in most cases it is found that
monetary policy plays a major role to these movements and the predictive content
of the yield curve is primarily due to the conduct of monetary policy. Estimated
parameters of the yield curve (representative of the financial factors), inflation,
growth (representation of the macroeconomic factors), along with policy variable
are combined in a vector auto regression (VAR) framework to study the linkage
between ‘yields-to-macro’ and ‘macro-to-yields’. The empirical analysis covers the
post reform period July 1997 to February 2004 of Indian financial markets. In
India the interest rates were not the true reflections of the market determined
interest rates until the mid1990s, as the financial markets were heavily regulated
until then. So one of the most important hurdles the study would face is the lack of
large sample set.
The rest of the article is organized as follows: section 2 gives a review of the
literature. Section 3 gives an overview of monetary policy in India with a
specific focus to its transmission mechanism through money market instruments.
Section 4 describes how the estimated parameters can be treated as Level, Slope
and Curvature. It also highlights the association of Level, Slope and Curvature
with macroeconomic factors. Section 5 describes the estimation methodology.
Section 6 talks about the data. Section 7 describes the estimation results.
Section 8 has concluding remarks with the interpretation of the results for the
Indian scenario.
 Macroeconomics and Finance in Emerging Market Economies 59

2. Literature survey
Factor models developed by Knez et al. (1994), Duffie and Kan (1996), and Dai and
Singleton (2000) are often interpreted as Level, Slope and Curvature and do not have
good macroeconomic interpretations. There has been a vast literature developed in
the area of term structure of interest rates which focuses only on the yield-to-macro
link or how yield curves can be used a predictor of fundamentals in the economy.
Mishkin (1990), Estrella and Hardouvelis (1991) and Estrella and Mishkin (1998)
have focused on the unidirectional predictive power of the yield curve for the key
macroeconomic variables like inflation and real activity. Estrella and Hardouvelis
(1991) formulated a probit model to predict the probabilities of US recessions one
year ahead. Estrella et al. (2003) have examined the association of term spread and
real activity for both US and German data using continuous and binary models.
They found that the binary models are more stable than the continuous models.
Kozicki and Tinsley (2001), Dewachter and Lyrio (2002) and Rudebusch and Wu
(2003) allowed feedback from an implicit inflation target derived from the yield curve
to help determine the dynamics of the yield curve. In the past few years, there has
been a lot of work which combine the macroeconomic factors in the yield curve
modelling. Ang and Piazzesi (2003), Hördahl et al. (2002), Wu (2002), and Evans and
Marshall (1998, 2001) incorporate macroeconomic variables into multi-factor yield
curve models. They considered unidirectional macro linkages as output and inflation
are assumed to be determined independently of the shape of the yield curve, but not
vice versa. Diebold et al. (2003, 2006) provide a macroeconomic interpretation of the
Nelson-Siegel representation of the yield curve by combining it with VAR dynamics
for the macroeconomy. In the dynamic interactions between macroeconomy and
yield curve, they have found weak evidence of the effects of yield curves on future
movements of the macroeconomic variables and a strong relationship is observed
from macroeconomic variables to the yield curves.
In India there have been some studies to investigate the characteristics of the
post-deregulated financial market. Joshi (1998) has shown that there is a linkage
between the conduct of monetary policy and the movements in the interest rates of
money market instruments in India. There have been similar studies by Bhoi and
Dhal (1998) that provide evidence of the integration of financial markets in India.
In the Indian scenario, the work by Kangasabapathy and Goyal (2002) closely
resembles the study by Estrella and Mishkin (1998). Kangasabapathy and Goyal
(2002) found yield spread, defined as the difference between long-term and short-
term, to be a good predictor of real activity. This study is different from any of the
previous studies in India as it incorporates both the macroeconomic fundamentals,
inflation, growth, monetary policy and yield-curves, explained by Level, Slope and
Curvature to understand their dynamics linkages.

3. Significance of money market in monetary policy transmission

Monetary policy in India aims at maintaining a judicious balance between price
stability and economic growth. In view of the increasing openness of the Indian
economy, macroeconomic and financial stability have attained great importance
these days.
In India the conduct of monetary policy until 1997–98 was based on an
intermediate target of broad money (M3). The aim was to regulate money supply in
60 K. Kanjilal

line with the expected growth in real income and a projected level of inflation. In the
phase of ongoing openness in financial sectors, the interest rate channel or exchange
rate channel has gained much importance in transmission mechanism of monetary
policy compared to quantity variables. In April 1998, the RBI has formally adopted
a multiple indicator approach where interest rates or rates of return in different
financial markets along with the data on currency, credit, trade, capital flows, fiscal
position, inflation and exchange rates are juxtaposed with the output data for policy
perspectives (Mohan 2007). In the increasing evidence of growing inter linkages in
the financial markets there has been a shift in focus of monetary policy approach
from direct to indirect instruments.
The success of a framework that relies on indirect instruments of monetary
management such as interest rates is contingent upon the extent and speed with
which changes in the central bank’s policy rate are transmitted to the spectrum of
market interest rates and exchange rate in the economy and to the real sector. The
money market has a pivotal role in the transmission mechanism of monetary policy.
The money market performs a crucial role in providing a channel to equilibrate
short-term demand for and supply of funds, thereby facilitating the conduct of
monetary policy. In India the money market mainly comprises of a call money
market, the activities in other money market segments, the commercial bills market
and the inter-corporate deposits market are limited. The call money market deals in
funds for two to four days. In India, like many other developing countries, the
evolution of the money market and its structure has been integrated into the overall
deregulation process of the financial sector which started in the late 1980s. Given
the importance of the money market as an indirect instrument in the monetary
transmission mechanism, several reform initiatives were taken since the late 1980s.
Among the several measures, most important was to turn the call money market
into a pure inter-bank market and to develop repo market towards improving
the monetary transmission mechanism to influence the aggregate demand in
the economy in 1999. Introduction of a liquidity adjustment facility (LAF) in June
2000 gave a different dimension in the monetary policy operating procedure in
India. In this approach, the Reserve Bank of India (RBI) decides its policy rates
(repo and reverse report rates)2 and executes repo and reverse repo operations
which provide a corridor for overnight money market rates. The LAF was
introduced primarily to develop a robust short-term operational model for better
understanding the transmission mechanism of the monetary policy. In line with this,
a liquidity assessment model is being structured, on the basis of available data, to
assess both the mechanism and inter-linkages among different segments of the
financial systems. The call money market is influenced by liquidity conditions in the
economy mainly driven by capital flows, Reserve Bank’s operations affecting banks’
reserve requirements on the supply side and tax outflows, the government
borrowing programme, seasonal fluctuations and non-food credit off-take on the
demand side. At the time of easy liquidity, call rates tend to hover around the
Reserve Bank’s repo rate which helps to absorb short-term surplus fund from
the system. During the periods of tight liquidity, call rates tend to move with the
reverse repo rate and bank rate.
In the post-LAF period, the call money rate remained more stable as the RBI has
been successful in managing the short-term liquidity under diverse market conditions
more effectively through repo and reverse repo operations. So the LAF combined
with other open market operations has emerged as an effective instrument to signal
 Macroeconomics and Finance in Emerging Market Economies 61

monetary policy stances in the economy. This also helped the central bank to adjust
the supply of money in response to the movement of key macroeconomic indicators
in a much faster way. The LAF avoids targeting a particular level of overnight
money market rates in case of exogenous influences (for example, volatile
government cash balances and unpredictable foreign exchange flows) impacting
liquidity at the shorter end of term structure of interest rates. This has enabled the
RBI to affect the demand for funds through policy rate channel. It enabled it to set
a corridor for call money rates consistent with the policy objectives of the
central bank.

4. Estimated parameters of yield curve: relationship with empirical proxies and

macroeconomic factors
The section demonstrates the reasons of choosing estimated parameters of the term
structure of interest rates as Level, Slope and Curvature. The term structure of
interest rates is first estimated by the Nelson-Siegel method for the period July 1997
to February 2004, minimizing the ‘yield errors’. The algorithm to estimate the spot
rates is provided in Figure 1.

4.1. Estimated parameters and factor loadings

The estimated parameters (b0, b1, b2) of the Nelson-Sigel spot rate function r(m)
(defined below):

rðmÞ ¼ b0 þ b1½f1  exp ðm= t1Þg= ðm= t1Þ

þ b2 ½f1  exp ðm= t1Þg= ðm= t1Þ  exp ðm= t1Þðb0; t1; t2 > 0Þ ð1Þ

where m is the maturity and r(m) is the spot rate, can be interpreted as the latent
dynamic factors. The loading on the first factor b0 is 1, a constant, so in the limit,
does not decay to 0. This is a long-term factor. The loading on the second factor b1 is
{1- exp (-m/t1)}/(m/t1), a function that starts at 1, but decays monotonically and
approaches to 0, hence this is considered as a short-term factor. The loading on the
third factor b2 is [{1 - exp (-m/t1)}/(m/t1) – exp(-m/t1)], starts at 0 and then
increases and approaches to 0 further. This factor is viewed as the medium term
factor. The long-term, short-term and medium-term factors can also be interpreted
as Level, Slope and Curvature of the yield curve. b0 governs the level of the yield
curve. An increase in the long-term factor (b0) will lead to an increase in all yields
equally as the loading is identical across all maturities. However, an increase in b1
will change the short-end of the yield curve more than its long-end, as the short rates
load heavily to b1, thereby changing the slope of the yield curve. Hence, it can be
interpreted that b1 governs the slope of the yield curve. The medium-term factor b2
is closely related to the curvature of the yield curve. Both the short rates and the long
rates load insignificantly to b2 as compared to the medium-term rates, which load
heavily to the third factor. Hence an increase or decrease in b2 will bring impact on
the medium term maturities.
Figures 2 to 5 demonstrate the factor loadings of the estimated parameters b0,
b1, and b2 for January 1998, January 1999, December 2002 and December 2003.
62 K. Kanjilal

Figure 1. Yield error minimization estimation algorithm.

These represent that at yield curve can be described as ‘Level’ (Lt(est)), ‘Slope’ (St(est)),
and ‘Curvature’ (Ct(est)), at any point in time. These four data points are randomly
chosen. The finding however is similar for all other data months.
 Macroeconomics and Finance in Emerging Market Economies 63

Figure 2. Nelson-Siegel factor loadings, January 1998.

Note: b0 ¼ b0, b1 ¼ b1 and b2 ¼ b2.

Figure 3. Nelson-Siegel factor loadings, January 1999.

Note: b0 ¼ b0, b1 ¼ b1 and b2 ¼ b2.

Figure 4. Nelson-Siegel factor loadings, December 2002.

Note: b0 ¼ b0, b1 ¼ b1 and b2 ¼ b2.
64 K. Kanjilal

4.2. Estimated parameters, empirical proxies and macroeconomic factors

The empirical proxies for the three factors are crucial in order to understand the gap
between the estimated parameters and the actual yields. Several empirical proxies for
the Level, Slope and Curvature factors have been tested for the Indian scenario with
no significant difference in the original findings. A common empirical proxy for level
is an average of short-term (3 months yield), medium-term (24 months yield) and
long-term (120 months) yields. Diebold (2003) has also used the same empirical
approximation for the Level factor. (yt (3) þ yt (24) þ yt (120))/3 is taken as the
proxy for level. A proxy for empirical estimation of slope is taken as (yt (5) – yt
(144)). A standard proxy for curvature is considered as (2* yt (24) – yt (5) – yt (144)).
Figures 6, 7 and 8 plot the empirical proxies along with the Lt(est), St(est) and Ct(est).

Figure 5. Nelson-Siegel factor loadings, December 2003.

Note: b0 ¼ b0, b1 ¼ b1 and b2 ¼ b2.

Figure 6. Correlation between empirical and estimated level.

 Macroeconomics and Finance in Emerging Market Economies 65

Figure 7. Correlation between empirical and estimated slope.

Figure 8. Correlation between empirical and estimated curvature.

The correlation between Lt(est) and empirical level factor (yt (3) þ yt (24) þ
yt (120))/3 is 80%. However, the degree of association between the empirical slope
factor and St(est) is 47%. The correlation between the empirical curvature (Ct(est)) and
is 8%. The degree of correlation varies between the pre-LAF and post-LAF periods.
For the period July 2000 through to February 2004, a high degree of correlation is
observed for the estimated Level (82%), Slope (60%) and Curvature (58%) with the
empirical proxies defined above. The high degree of correlation between the Lt(est),
St(est) and Ct(est), and their empirical proxies lend credibility to the interpretation of
b1, b2 and b3 as Level, Slope and Curvature factors in the post-LAF period. The
lower degree of association between the estimated slope and curvature with their
empirical proxies reflects the regulated nature of the market in the pre-LAF period.
In the initial period of the data span, very few debt instruments were regularly traded
in the market. The spread between the long-term and short-term instruments shows
a declining trend for India in the entire data span considered. This should indicate
that the Level of the yield curve almost explains the entire span of the yield curve.
This also implies a relatively flattening yield curve trend for India during the period
considered.
66 K. Kanjilal

Diebold et al. (2003) investigated the association between Lt(est) with inflation
and St(est) with capacity utilization for the US economy. The correlation between
Lt(est) with inflation and St(est) with capacity utilization were approximately 40%. In
this case, the degree of correlation between Lt(est), St(est) and Ct(est) with inflation or
growth is in the range of 10% to 15%. However, the correlation between Lt(est) and
the call money rate (CMR) is around 50%. The finding is consistent with the fact
that monetary policy in general will bring most impact to the level of the yield curve
as the transmission of monetary policy runs from short term interest rates to longer
term influencing aggregate demand in the economy.

5. Estimation methodology
This section attempts to give an overview of the interactions of yield curves with
inflation, growth and the role of central bank before explaining the estimation
methodology to understand the ‘macro-to-yield’ and ‘yield-to-macro’ link in an
economy.
In general, the transmission of monetary policy happens through short-term to
long-term rates. In the case where the central bank wants to adopt a tight monetary
policy, it will sell government bonds/instruments, reducing the liquidity in the
system. The central bank can infuse the liquidity in the economy buying back
government bonds/instruments. When the market expects the central bank to cut
rates the shorter term instruments become expensive as they continue to offer higher
interest or coupon rates. As a consequence, the yield declines, adjusting to the lower
interest rate environment and the yield curve becomes steeper. On the contrary,
when the market expectation is that the central bank will increase interest rates, the
price of the debt instruments fall causing the yield to increase, and the yield curve
flattens. The central bank’s decision to increase or decrease interest rates depend on
the economic scenario in the country. If there is growth prospect in the economy,
investment activity will be buoyant, resulting in increasing demand for money. This
might lead to constant inflationary pressures. In such cases the central bank adjusts
the fast rise in demand to the slower growth in supply, increasing the cost of money.
Economic growth and inflation affect the yield curve as the monetary policy is largely
influenced by the health of the economy.

5.1. Estimation methodology: VAR and Granger’s causality

The estimated Level (Lt(est)), Slope (St(est)), Curvature (Ct(est)), growth (changes in the
industrial index of production (IIP)), along with inflation (changes in the wholesale
price index (WPI)) and the call money rate (CMR) are used in a VAR framework to
understand the dynamics of financial and macroeconomic variables for the period
July 1997 to February 2004. The policy variable, CMR, helps to identify the
response of macroeconomic and financial factors to the changes in monetary policy.
Lt(est), St(est) and Ct(est) are taken as financial factors and growth (GWTt), inflation
(INFt) and CMRt are taken as macro variables.
As a first step, the existence of a non-stationary feature of each variable is
examined. In case the series is non-stationary in nature, the first difference of the
series is taken to make it stationary. In the next step the VAR model is estimated
incorporating the appropriate lagged terms of the variables. The estimated model is
 Macroeconomics and Finance in Emerging Market Economies 67

then used for an in-sample forecast of the yields. Next, the direction of individual and
block Granger causality among the variables is investigated. In case of significant
existence of structural break, a dummy or exogenous variable is incorporated in the
VAR estimation. Finally, sensitivity analysis or impulse response function is carried
out to identify the dynamics of macro variables with respect to financial variables and
vice versa. This analysis is repeated for the pre-LAF and post-LAF periods to see if
the difference in findings are due to the introduction of liquidity adjustment facility
by the central bank (LAF) in the month of June 2000.

5.1.1. VAR estimation methodology

Let us consider an unrestricted VAR of order p represented by

X
p
yt ¼ a0 þ a1 t þ Fi yti þ Cwt þ ut ð2Þ
i¼1

where yt is a (n 6 1) vector of endogenous I(0) variables, t is the linear time trend, a0

and a1 are (n 6 1) vectors, wt is a (q 6 1) vector of exogenous variables and ut is
a (n 6 1) vector of unobserved disturbances where ut * N (0, O), t ¼ 1 2 . . ...T.
VAR can be constructed either in terms of the level of the data or in terms of
their first differences, i.e. I(0) variables. If the series are non-stationary, the Granger
causality test may give misleading results unless the data are transformed to induce
stationarity.

5.1.2. Augmented Dickey Fuller Test (ADF) test for stationarity

ADF test is conducted with the following model:

X
p
DXt ¼ a0 þ ð1  kÞbt  ð1  kÞXt  1 þ gj DXtj þ et ð3Þ
1

where Xt is the underlying variable at time t, et- is the error term and a0, b, k and gj
are the parameters to be estimated. The lagged terms are introduced in order to
justify that errors are uncorrelated with them. For the above-specified model the
hypothesis, which would be of our interest, is:

H0 : ð1  kÞ ¼ 0

5.1.3. Block Granger causality test

X
m X
n
DXt ¼ a þ bi DXti þ gj DYtj þ ut ð4Þ
i¼1 j¼1

X
q X
r
DYt ¼ a þ bi DYti þ cj DXtj þ vt ð5Þ
i¼1 j¼1
68 K. Kanjilal

The optimum lag length m, n, q and r are determined on the basis of Akaike’s (AIC)
and/or Schwarz Bayesian (SBC) and/or log-likelihood ratio test (LR) criterion. DY
Granger causes DX if,

H0: g1 ¼ g2 ¼ . . . ¼ gn ¼ 0 is rejected.

Against HA: ¼ at least one gj 6¼ 0, j ¼ 1 . . . n,

and DX Granger causes DY if,
H0: c1 ¼ c2 ¼ . . . ¼ cn ¼ 0 is rejected.
Against HA: ¼ at least one c j 6¼ 0, j ¼ 1 . . . r.

5.1.4. Pair-wise Granger causality tests/block exogeneity Wald test

Pair-wise Granger causality tests whether an endogenous variable can be treated as
exogenous. For each equation in the VAR, CHSQ (Wald) statistics test for the joint
significance of each of the other lagged endogenous variables in that equation.

6. Data
The analysis covers the data from July 1997 through to February 2004. The time
series of estimated parameters of the yield curve Level, Slope and Curvature are
taken as financial factors. Annualized monthly changes in WPI and IIP are used as
inflation and real economic activity respectively. CMR is the main component of
the money market. The money market has always played a pivotal role in the
monetary policy transmission mechanism in the Indian context, like many other
developing economies. Hence, the CMR is considered as a proxy for monetary
policy indicator. Inflation, real activity and CMR represent macro factors. To
understand the regime-shift effect of LAF, which was introduced in June 2000 as
part of the monetary and credit policy, the analysis is also carried out for the pre-
LAF (July 1997 to May 2000) and post-LAF periods (July 2000 to February 2004)
separately.

7. Empirical results
As a first step the stationarity of the series is carried out using an ADF test.
The Table 1 below displays the ADF tests for all variables under consideration,
Level (Lt(est)), Slope (St(est)), Curvature (Ct(est)), GWTt, INFt, DCMRt. Lt(est) and
CMRt possess unit root at 5% level of significance. However, both the series become
stationary in its first difference. No significant structural break is observed in any
variables for the period under study July 1997 to February 2004.

7.1. Results: overall sample

We estimate VAR with six variables DLt(est), St(est), Ct(est), GWTt, INFt, DCMRt.
VAR of order one is selected based on Akaike and SBC criteria. In this case, the
estimated VAR model is,

Yt ¼ Constant þ AðestÞYt1 ð6Þ

 Macroeconomics and Finance in Emerging Market Economies 69

Table 1. Augmented Dicky Fuller (ADF) unit foot tests.

Variable t-Statistic Test critical values at 5% Probability

Lt(est) 72.940249 73.469235 0.1561
DLt(est) 75.162545 73.475305 0.0004
St(est) 75.973453 73.468459 0
Ct(est) 75.376175 73.468459 0.0001
GWTt 711.21877 73.478305 0
INFt 75.558771 73.470032 0.0001
CMRt 72.209611 73.478305 0.4764
DCMRt 75.505683 73.478305 0.0001

Notes: Null hypothesis: the variable has a unit root. Exogeneous: constant, linear trend.
Optimal lag length is decided based on AIC criteria.

where Yt ¼ (DLt(est), St(est), Ct(est), GWTt, INFt, DCMRt) is a (561) vector of

endogenously determined variable. In this case, Wt (in Equation 7) does not contain
any exogenous variables as no significant structural break is found in the data span
considered.
The estimated VAR model is shown in Table 2 where each of the variable at time
point ‘t’ is regressed on its own lagged value and the lagged values of other five
variables at time ‘t-1’. Inflation with a lag of one period is significant at the 5% level
in the equation of changes in level and changes in CMR. This estimated VAR model
is then used for in-sample forecasts of the parameters level (b0), slope (b1) and
curvature (b2) of the Nelson-Siegel model for the five months; October 2003 to
February 2004. The estimated yield parameters Level, Slope and Curvature obtained
from the VAR model of macro and yield factors are used to estimate the yield curve
for October 2003 to February 2004 using the Nelson-Siegel model. The objective is
to verify if the goodness of fit3 of the yield curve improves with the estimated yield
parameters obtained from the multivariate VAR model of macro and financial
variables. Table 3 shows the root mean square errors (RMSEs) of estimated spot
rates or term structure of interest rates by the Nelson-Siegel method (shown as
‘Actual’) and the multivariate VAR estimates of macro and financial variables
(shown as ‘Predicted’) for the period October 2003 to February 2004.
The actual and forecasted RMSEs do not differ at all. This suggests that in the
Indian context, inclusion of macro variables does not seem to improve ‘in-sample’
prediction of the yield curve significantly. This finding is different from Diebold et al.
(2003), who proved that for the US economy the yield curve model along with key
macro economic variables improve the prediction of spot rates. The liquidity
adjustment facility which was launched in June 2000 through repo and reverse repo
operations enabled the central bank to manage the liquidity conditions in response
to the movement of key macroeconomic indicators in a much faster way. After the
launch of the LAF the interest rates in the money market and government securities
market have become more consistent with policy objectives of central banks which
vary under diverse macroeconomic conditions. There has been a significant
development in the government securities market over the past five years in terms
of its participants, turnover, and liquidity following a series of reform measures
adopted by the RBI which would lead to better integration of the government
securities market with other financial sectors and macroeconomic factors in the
 70

Table 2. Vector autoregression estimates.

Variables DLt(est) St(est) Ct(est) GWTt INFt DCMRt

DLt(est)71 70.3732 0.024821 70.02311 70.050405 0.012289 0.132528
[73.83976] [0.89027] [70.96582] [71.36536] [3.21990] [0.57732]
St(est)71 1.828912 0.256048 70.05585 70.395878 0.023119 1.62653
[2.74108] [1.33776] [70.34007] [71.56208] [0.88240] [1.03213]
Ct(est)71 1.520739 70.22421 0.473581 70.500128 0.032319 71.6653
[1.88195] [70.96724] [2.38082] [71.62947] [1.01856] [70.87255]
GWTt71 70.53695 70.00709 0.003441 70.467099 70.00466 70.71797
[71.84825] [70.08509] [0.04811] [74.23296] [70.40876] [71.04634]
INFt71 72.21155 0.699156 70.86264 71.806333 0.314601 717.1228
K. Kanjilal

[71.00137] [1.10357] [71.58674] [72.15331] [3.62770] [73.28260]

DCMRt71 0.069506 70.0036 0.01043 70.000978 0.003548 70.46576
[2.26731] [70.40890] [1.38214] [70.08399] [2.94734] [76.43267]
Constant 0.059334 70.0349 0.006932 0.001239 0.002915 0.17874
[2.11630] [74.33961] [1.00439] [0.11635] [2.64767] [2.69922]

Notes: t-statistics in [ ]. VAR of order one is selected based on AIC criterion.

 Macroeconomics and Finance in Emerging Market Economies 71

Table 3. In-sample forecasts of term structure of interest rates, root mean square error
(RMSE).

Months Actual Predicted

October 2003 0.52071 0.52081
November 2003 0.75204 0.75165
December 2003 0.73701 0.73702
January 2004 1.79016 1.79012
February 2004 2.80090 2.80088

Table 4. Likelihood ratio (LR) test of block Granger causality.

Entire sample period Pre-LAF Post-LAF

July 1997–February July 2000–February
2004 July 1997–May 2000 2004
Block of
variables CHSQ (9) Probability CHSQ (9) Probability CHSQ (9) Probability
Financial 29.9956 0* 14.2364 0.114 19.5841 0.021*
to macro
Macro to 20.9976 0.103 10.8071 0.289 14.5437 0.104
financial

Notes: * implies hypothesis of non-causality is rejected. Financial factors: DL t (est) S t (est) C t (est). Macro
factors: GWTt, INFt and DCMRt.

economy. So the finding that the inclusion of macro factors (inflation, growth) do
not appear to improve the yield curve estimates could be attributed to the sample
period considered in this study.
As a next step, both block and individual Granger causality tests of the financial
and macro variables are carried out for the entire sample period. Table 4 shows the
block Granger causality both from financial factors (DLt(est), St(est), Ct(est)) to macro
factors (GWTt, INFt, DCMRt) and macro to financial factors.
The probability value of 0 associated with the LR test of block non-causality (in
Table 4) suggests that the financial factors represented by the estimated level, slope,
curvature do influence the movements of macroeconomic factors with a weaker
feedback effect from macroeconomic factors to financial factors, where the
associated probability value that macro factors Granger cause the financial factors
is 10%. One of the possible reasons could be that the yields in the government
securities markets for the sample period considered are market related. They do not
reflect the true market determined rates. Hence, changes in the macroeconomic
factors do not have significant reflections on the changes in the financial factors.
However, we could possibly find the reverse trend; macro factors driving the
financial factors in more recent data given the significant development in the
government securities market.
To obtain a clear picture of the dynamics of each variable and understand the
intervention of the central bank in the macro-to-yield and yield-to-macro link the
individual Granger causality among the variables is examined for the entire sample
period July 1997 to February 2004. The results are shown in Table 5.
The variables listed in each row are regressed on the variables across the columns.
For example, in Table 5, DL(est) is regressed on S(est), C(est), growth (GWT), inflation
72 K. Kanjilal

Table 5. VAR Granger causality/block exogeneity Wald tests, entire sample period (July
1997 to February 2004).

Dependent
variables DL(est) S(est) C(est) GWT INF DCMR
DL(est) 7.513518 3.541724 3.416019 1.002737 5.140688
[0.0061] [0.0598] [0.0646] [0.3166] [0.0234]
S(est) 0.792573 0.935558 0.00724 1.217865 0.167198
[0.3733] [0.3334] [0.9322] [0.2698] [0.6826]
C(est) 0.932807 0.115647 0.002315 2.51773 1.910304
[0.3341] [0.7338] [0.9616] [0.1126] [0.1669]
GWT 1.864208 2.440084 2.655168 4.636739 0.007055
[0.1721] [0.1183] [0.1032] [0.0313] [0.9331]
INF 10.36775 0.778633 1.037465 0.167083 8.686785
[0.0013] [0.3776] [0.3084] [0.6827] [0.0032]
DCMR 0.333299 1.065301 0.761347 1.09482 10.77545
[0.5637] [0.302] [0.3829] [0.2954] [0.001]

Notes: Probability values are in [ ]. Highlighted cells imply existence of causality.

(INF) and changes in CMR (DCMR). The probability values in excess of 5%

associated with the null hypothesis of non-causality will not be considered to have
any strong impact on the dependent variable. We observe that the dynamics of yield
and macro factors start with the inflation change. For example, in the equation of
changes in CMR, inflation is the significant variable. So inflation in the first place
brings an impact on CMR. The changes in CMR have an impact on inflation. In the
equation of inflation, we see that changes in CMR is statistically significant, so as the
level of yield curve. So changes in CMR influences the movements (increase/
decrease) of inflation and also brings an impact on the level of yield curve.4 Thus, the
changes in CMR influences the estimated level of the yield curve causing an impact
on the overall movements of yield curve. Changes in level, in turn, brings an impact
on inflation; finally inflation causes an impact to the overall growth of the economy.
And growth has a weaker impact on the changes in the level of the yield curve. So it
is observed that the policy variable CMR has a very important role in bringing an
impact on inflation, growth and overall yield curve movements. This finding is
consistent with the fundamental macroeconomic movements in an economy. The
indirect instruments of monetary policy are becoming an important tool in
influencing the aggregate demand or growth in the economy. Hence, it is observed
that in the Indian scenario, the central bank does play a vital role in establishing the
linkage between macroeconomic and financial factors through the changes in
monetary policies. The result also suggests that the Level of the yield curve can be
used as indicator for inflation. The introduction of the LAF in June 2000 has
significant role in revamping the liquidity factor in government securities market in
India. So the section below divides the entire sample into two periods: pre- and post-
LAF and examines the causality for the two sample periods July 1997 to May 2000
and July 2000 to February 2004 separately.

7.2. Results: pre-LAF and post-LAF periods

The Reserve Bank of India has introduced a liquidity adjustment facility in June
2000 as part of the monetary and credit policy. The LAF was introduced primarily to
 Macroeconomics and Finance in Emerging Market Economies 73

develop a robust short-term operational model for better understanding of the

transmission mechanism of the monetary policy. In line with this, a liquidity
assessment model is being structured, on the basis of available data, to assess both
the mechanism and inter-linkages among different segments of the financial systems.
Repo auctions to absorb the liquidity and reverse repo-auctions to inject liquidity in
the economy started on a daily basis after June 2000. This helped the central bank to
adjust the supply of money in response to the movement of key macroeconomic
indicators in a much faster way. This is indeed one of the major achievements for the
sovereign market in India.
To understand whether introduction of LAF has changed the dynamic linkages
between the macro and the financial factors we separate out the pre- and post-LAF
period. Table 4 shows block Granger causality tests for the pre- and post-LAF
period separately. In the pre-LAF period no causality is found from any direction.
So neither financial factors nor macro factors has any linkages in the movements of
each other. But the result in the post-LAF period suggests that the causality exists in
both the directions. However, causality running from macro to financial factor is
found to be weaker compared to the financial factors to macro link. This is
consistent with the finding of the entire sample period. So the causality from the
financial to the macro factors in the Indian scenario evolves post the introduction of
liquidity adjustment facility.
A pair-wise Granger causality tests are also shown in Tables 6 and 7 for the pre-
and post-LAF periods respectively.
In the pre-LAF period none of the variables are found to have a strong causal
impact. It is found that inflation causes the changes in CMR with a feedback effect
to inflation. However, the evidence of causality is found to be weaker in this
period. This could be due to the fact that the use of indirect instruments of
monetary policy was in the process of making for the sample period under study.
The result for the post-LAF period as shown in Table 7 is similar to the overall
sample. Inflation causes the changes in CMR, which in turn influences the
movements in the level of the yield curve. And we find causality runs from changes
in level to inflation and growth. But the causality from the changes in level to
growth is weaker. Also, we see a weaker causality running from inflation to

Table 6. VAR Granger causality/block exogeneity Wald tests, pre-LAF (July 1997 to May
2000).

Dependent variables DL(est) S(est) C(est) GWT INF DCMR

DL(est) 0.26113 3.68E-06 0.344906 0.021596 1.024626
[0.6093] [0.9985] [0.557] [0.8832] [0.3114]
S(est) 0.491535 0.015293 0.537953 0.371391 0.005193
[0.4832] [0.9016] [0.4633] [0.5422] [0.9426]
C(est) 0.900265 0.023089 1.097081 2.061459 1.298125
[0.3427] [0.8792] [0.2949] [0.1511] [0.2546]
GWT 0.002108 0.278977 0.454625 1.638465 0.132496
[0.9634] [0.5974] [0.5001] [0.2005] [0.7159]
INF 0.728759 0.504135 2.189808 0.11106 3.279767
[0.3933] [0.4777] [0.1389] [0.7389] [0.0701]
DCMR 0.008784 2.597045 0.591171 3.809058 3.538602
[0.9253] [0.1071] [0.442] [0.051] [0.06]

Note: Probability values are in [ ].

74 K. Kanjilal

growth. This leads us to infer that the Level of the yield curve can be used as an
indicator for the movements of the inflation. So it is clear that the role of the
central bank (RBI) or the indirect instrument of monetary policy has become much
stronger in influencing the aggregate growth in India after the introduction of the
liquidity adjustment facility.

7.3. Sensitivity analysis

In order to see the responses of individual variables while there is a shock in the
system, impulse response functions are analysed for the entire sample and the post-
LAF period.5 The results should further validate the previous findings. Figures 9 to
14 show results of impulse response functions for the entire sample period.
Each of the graphs shows the response of all variables while there is one unit
standard error shock in one specific variable. For example, Figure 9 shows the

Table 7. VAR Granger causality/block exogeneity Wald tests, post-LAF (July 2000 to
February 2004).

Dependent variables DLt(est) St(est) Ct(est) GWTt INFt DCMRt

DLt(est) 8.036065 4.712088 1.902009 0.729316 4.215012
[0.0046] [0.03] [0.1679] [0.3931] [0.0401]
St(est) 0.016905 4.279774 0.00724 1.217865 0.167198
[0.8966] [0.0386] [0.9322] [0.2698] [0.6826]
Ct(est) 0.932807 0.115647 0.939653 0.001122 0.164592
[0.3341] [0.7338] [0.3324] [0.9733] [0.685]
GWTt 2.976206 1.326795 1.099969 2.355619 0.037513
[0.0845] [0.2494] [0.2943] [0.1248] [0.8464]
INFt 12.49285 1.81068 0.383457 0.061256 1.225752
[0.0004] [0.1784] [0.5358] [0.8045] [0.2682]
DCMRt 0.837949 0.374026 0.91438 0.010439 5.404851
[0.36] [0.5408] [0.339] [0.9186] [0.0201]

Notes: Probability values are in [ ]. Highlighted cells imply existence of causality.

Figure 9. Impulse response to one unit SE shock of Dlevel (entire sample period July 1997 to
February 2004).
 Macroeconomics and Finance in Emerging Market Economies 75

Figure 10. Impulse response to one unit SE shock of Slope (entire sample period July 1997 to
February 2004).

Figure 11. Impulse response to one unit SE shock of Curvature (entire sample period July
1997 to February 2004).

Figure 12. Impulse response to one unit SE shock of Growth (entire sample period July 1997
to February 2004).
76 K. Kanjilal

Figure 13. Impulse response to one unit SE shock of inflation (entire sample period July 1997
to February 2004).

Figure 14. Impulse response to one unit SE shock of DCMR (entire sample period July 1997
to February 2004).

response of slope (est), curvature (est), growth, inflation and changes in the CMR
(DCMR) when there is one unit standard error shock to changes in Level of yield
curve. The shock in changes in level (Dlevel) brings a significant impact to the
changes in the monetary policy variable, DCMR, with no significant deviations in
other variables in the system. Figures 9 to 14 suggest that while the system receives a
shock in any variables either inflation or growth or slope, it is the changes in CMR
that deviates most in the short-term. Figure 13 shows the responses of financial and
macro variables while there is an external shock in inflation. This needs further
discussion given the dynamic linkages observed between macro and financial factors
in the above section. The response or deviation of changes in CMR is the highest
followed by the changes in level of yield curve with one unit standard error shock in
inflation. Changes in CMR and changes in level take almost five to six months to
come back to the normal level after an inflation shock in the system. The shock in
 Macroeconomics and Finance in Emerging Market Economies 77

inflation also brings a deviation in growth. Thus, we observe that the response of
monetary policy is very prompt or immediate while there is an inflation shock in the
system. This also suggests that the indirect instrument of monetary policy is
becoming important to stabilize the economy while there is an external
macroeconomic shock.
Figures 15 to 20 show similar findings for the post-LAF period July 2000 to
February 2004.
In the post-LAF period it is observed that the changes in level of yield curve
reacts most followed by the changes in CMR while there is a shock in inflation.
The shock in inflation also brings significant impact in growth. The changes in

Figure 15. Impulse response to one unit SE shock of Dlevel (post-LAF period July 2000 to
February 2004).

Figure 16. Impulse response to one unit SE shock of Slope (post-LAF period July 2000 to
February 2004).
78 K. Kanjilal

Figure 17. Impulse response to one unit SE shock of Curvature (post-LAF period July 2000
to February 2004).

Figure 18. Impulse response to one unit SE shock of Growth (post-LAF period July 2000 to
February 2004).

level of yield curve takes three to four months, on average, to come back to the
normal state. This finding implies that the level of the yield curve responds most
with the external shock in the economy. The result also suggests that with the
constant endeavour of the RBI to develop the government securities market in
India, the yield rates are becoming more market related and the term structure of
interest rates can be used as one of the important indicators to predict future
behaviour of inflation or growth.
 Macroeconomics and Finance in Emerging Market Economies 79

Figure 19. Impulse response to one unit SE shock of inflation (post-LAF period July 2000 to
February 2004).

Figure 20. Impulse response to one unit SE shock of DCMR (post-LAF period July 2000 to
February 2004).

8. Conclusion
The study aims at understanding the dynamics of the linkage, if any, between the
term structure of interest rates and the macroeconomic factors of the deregulated
Indian economy. Movements in the yield curves is nothing but an outcome of the
changes in the key macroeconomic fundamentals; inflation, growth, and monetary
policy. So shape of the yield curve can be used as a good predictor or reflector to
effectively manage monetary policy, which would influence inflation and growth
in an economy. This study integrates macroeconomic factors and yield curves with
80 K. Kanjilal

two-fold objectives: first, to investigate the dynamic linkages between macro and
yield factors, and second to see if the integrated macro-yield model improves the
prediction of yield curve. To explore the possibility of the Reserve Bank of India’s
intervention in this dynamic relationship an indicator for monetary policy is also
introduced. The dynamics of yield curve and macroeconomic factors cannot
independently be determined without a monetary policy indicator as in most cases, it
is found that monetary policy plays a major role in these movements and the
predictive content of the yield curve is primarily due to the conduct of monetary
policy. The validity of the rational expectation hypothesis for the government
security markets in India provides a strong basis to test the bidirectional linkages
between the debt instruments and the key macroeconomic indicators in the
intervention of monetary policy.6
It is shown that the factor loadings of the parameters (b0, b1, b2) estimated by
the Nelson-Siegel model can be explained as Level, Slope and Curvature for a zero-
coupon yield curve at any point in time. A time-series of Level, Slope and Curvature
along with inflation (annualized monthly changes in WPI), GDP growth (annualized
monthly changes in IIP) and a policy variable call money rate (CMR) are used to
understand the long-run dynamics between financial and macroeconomic factors for
the period July 1997 to February 2004 in a multivariate VAR framework. The in-
sample predictions of term structure of interest rates show that the inclusion of
macro factors does not improve the prediction of yield curve. The result differs from
Diebold (2003) where he found that the inclusion of macro factors improved the
yield curve fit.
Block Granger causality between macro and financial factor shows that there is
strong causality from financial factors, defined by the three parameters of the yield
curves (level, slope and curvature) and to macroeconomic factors (growth, inflation
and changes in CMR) for the entire sample period. However, the causality in the
opposite direction is found to be much weaker. The study also finds that in the pre-
LAF period (July 1997 to May 2000), yield and macro factors do not cause each
other. The finding in the post-LAF period (Jul 2000 to February 2004) is similar to
the one for entire sample period. Pair-wise Granger causality shows the dynamics of
causality starts from inflation. Inflation causes the changes in CMR, which in turn
influences the movements in the level of the yield curve. The causality runs from
changes in level to inflation and growth. However, the causality from level to growth
is found to be weaker. Thus, CMR responds to inflation and changes in the CMR
drives the change in the level of the yield curve which in turn smoothens or tightens
inflation and finally brings an impact on the aggregate demand in the economy. The
finding is consistent with the fundamental macroeconomic movements in an
economy. The result implies that the indirect instruments of monetary policy are
becoming an important tool in influencing the aggregate demand or growth in the
economy. Hence, it is observed that in the Indian scenario, the RBI does play a vital
role in establishing the linkage between macroeconomic and financial factors
through the changes in monetary policies. The finding also suggests that the Level of
the yield curve can be used as indicator for inflation. The introduction of the LAF in
June 2000 has had a significant role in revamping the liquidity factor in the
government securities market in India.
Sensitivity analysis for the overall sample and for the post-LAF period show that
changes in CMR or monetary policy variable and changes in the level of the term
structure of interest rates deviate most once there is a shock in the system. This
 Macroeconomics and Finance in Emerging Market Economies 81

implies that the indirect instrument of monetary policies are becoming increasingly
important to stabilize the aggregate demand in the economy. Also, with the
continuous development of the government security market in India, yield rates are
becoming more and more market related and have started reflecting the market
behaviours. Thus, with the ongoing reform measures in the financial market in India,
and especially in the government securities market in India, term structure of interest
rates can be used as one of the main indicators to predict future behaviour of
inflation or growth in the near future.
The study falls in the initial phase of reforms and hence is constrained by the
smaller sample, so to prove that term structure of interest rate plays a leading role to
predict the future macroeconomic behaviours, this study can be extended or
replicated with more recent data starting from the post-LAF period, June 2000. The
finding in this case can be expected to be more robust for a couple of reasons. Firstly,
the sample size will be larger. Secondly, the developments in the government
securities market have been immense in terms of its liquidity and market turnover
post the launch of the liquidity adjustment facility in June 2000 to date. Government
securities markets are progressively becoming more integrated with financial
markets. Yield rates are becoming more market oriented which should reflect the
expectations of the economy.

Acknowledgements
This work is a part of the PhD thesis titled ‘Term Structure of Interest Rates: Indian
Macroeconomic Issues’ under the guidance of Professor Ashima Goyal at IGIDR. The
author would like to express sincere thanks and deep sense of gratitude to Professor Ashima
Goyal, whose moral support, indispensable guidance and constant encouragement made
the efforts possible. The author is also grateful to referees for valuable comments and
suggestions.

Notes
1. Lowe (1992) asserts that the predictive power of the slope of the yield curve is primarily
due to the liquidity effect in the economy, which is basically the negative relationship
between the interest rates and money supply. Movements in the short-term interest rates
as a result of the monetary policy changes underlie the predictive power of the yield curve
for future real activity as because real activity responds to the movements in short-term
interest rates with a lag. The fact that the transmission mechanism of monetary policies
affect the slope of the yield curve is established by Bernanke (1990), Bernanke and Blinder
(1992). The second view concerning the predictive capability of the term spread about
business cycle of the economy stems from the ‘consumption-based asset-pricing model’.
Harvey (1988, 1989) has shown a systematic relationship between the stage of the business
cycle and the shape of the yield curve. The theory says that the forecasting power of the
yield spread for future real activity originates from the economic agents’ expectations
about the future state of the economy and intertemporal utility maximizing behaviour.
This view has been supported by Hu (1993), Plosser and Rouwenhorst (1994), and Kim
(1998).
2. Repo auctions, which are to absorb the liquidity, and reverse repo-auctions, which are to
inject liquidity in the economy, started on a daily basis after June 2000.
3. Goodness of fit criterion is decided based on root mean square error of the yields.
4. In the equation of changes in level, changes in CMR is statistically significant.
5. The results for the pre-LAF period are not discussed as the linkage between macro-to-
yield and yield-to-macro is found only for the post-LAF period.
6. The underlying assumption of this framework is based on the fact that the rational
expectations theory holds true for the government securities market in India. The study is
a chapter of the PhD thesis ‘Term Structure of Interest Rates: Indian Macroeconomic
82 K. Kanjilal

Issues’. One of the chapters explores the validity of rational expectation hypothesis (REH)
for the government securities markets in India. The finding goes in favour of the theory for
the market.

Notes on contributor
Kakali Kanjilal has been working at American Express India (pvt) Ltd as Senior Manager in
Credit Risk Analysis Department for the last seven years. She received her PhD thesis titled
‘Term Structure of Interest Rates: Indian Macroeconomic Issues’ in IGIDR in February 2009.
She completed an MSc (Statistics) from Gauhati University, Assam, in 1994. Her areas of
interests are macro-finance and financial econometrics.

References
Ang, A., and Piazzesi, M. 2003. A no-arbitrage vector autoregression of term structure
dynamics with macroeconomic and latent variables. Journal of Monetary Economics 50:
745–87.
Bernanke, Ben S. 1990. On the predictive power of interest rates and interest rate spreads.
NBER Working Paper No. W3486, http://ssrn.com/abstract¼226685.
Bernanke, Ben S., and Blinder, Alan S. 1992. The federal funds rate and the channels of
monetary transmission. The American Economic Review 82: 901–21.
Bhoi, B.K., and Dhal, S.C. 1998. Integration of financial markets in India: An empirical
evaluation. RBI Occasional Papers 19, no. 4: 345–80.
Dai, Q., and Singleton, K. 2000. Specification analysis of affine term structure models. Journal
of Finance 55: 1943–78.
Dewachter, H., and Lyrio, M. 2002. Macroeconomic factors in the term structure of interest
rates. Manuscript, Catholic University Leuven and Erasmus University, Rotterdam.
Diebold, F.X., Rudebusch, G.D., and Aruoba, B. 2003. The macroeconomy and the yield
curve: A nonstructural analysis. Working papers in applied economic theory 2003-18,
Federal Reserve Bank of San Francisco.
Diebold, F.X., Rudebusch, G.D., and Aruoba, B. 2006. The macroeconomy and the yield
curve: A dynamic latent factor approach. Journal of Econometrics 131: 309–38.
Duffie, D., and Kan, R. 1996. A yield-factor model of interest rates. Mathematical Finance 6:
379–406.
Estrella, A., and Hardouvelis, G.A. 1991. The term structure as a predictor of real economic
activity. Journal of Finance XLVI no. 2: 555–76.
Estrella, A., and Mishkin, F.S. 1998. Predicting U.S. recessions: Financial variables as leading
indicators. Review of Economics and Statistics 80: 45–61.
Estrella, A., Rodrigues, A.P., and Schich, S. 2003. How stable is the predictive power of the
yield curve? Evidence from Germany and the United States. Review of Economics and
Statistics 85, no. 3: 629–44.
Evans, C.L., and Marshall, D. 1998. Monetary policy and the term structure of nominal
interest rates: Evidence and theory. Carnegie-Rochester Conference Series on Public Policy
49: 53–111.
Evans, C.L., and Marshall, D. 2001. Economic determinants of the nominal treasury yield
curve. Manuscript, Federal Reserve Bank of Chicago.
Harvey, C.R. 1988. The real term structure and consumption growth. Journal of Financial
Economics 22: 305–34.
Harvey, C.R. 1989. Forecasts of economic growth from the bond and stock markets. Financial
Analysis Journal 45: 38–45.
Hördahl, P., Tristani, O., and Vestin, D. 2002. A joint econometric model of macroeconomic
and term structure dynamics. Manuscript, European Central Bank.
Hu, Z. 1993. The yield curve and real activity. IMF staff papers 40, no. 4: 781–806.
Joshi, V. 1998. India’s economic reforms: Progress, problems, prospects. Oxford Development
Studies 26, no. 3: 333–50.
Kanagasabapathy, K., and Goyal, Rajan. 2002. Yield spread as a leading indicator of real
economic activity: An empirical exercise on the Indian economy. IMF Working Paper No.
02/91, May.
 Macroeconomics and Finance in Emerging Market Economies 83

Kim. 1998. The term structure of interest rates in a simple stochastic growth model: Evidence
from Australian data. Working Papers in Economics, No. 98-02.
Knez, P., Litterman, R., and Scheinkman, J. 1994. Exploration into factors explaining money
market returns. Journal of Finance 49: 1861–82.
Kozicki, S., and Tinsley, P.A. 2001. Shifting endpoints in the term structure of interest rates.
Journal of Monetary Economics 47: 613–52.
Lowe, P. 1992. The term structure of interest rates, real activity and inflation. Economic
Research Department, Reserve Bank of Australia, Research Discussion Paper, 9204: 1–38.
Mishkin, F.S. 1990. What does the term structure tell us about future inflation. Journal of
Monetary Economics 25: 77–95.
Mohan, Rakesh. 2007. Monetary policy transmission in India. RBI Monthly Bulletin, April:
421–72.
Plosser, C.I., and Rouwenhorst, K.G. 1994. International term structures and real economic
growth. Journal of Monetary Economics 33: 133–55.
Rudebusch, G.D., and Wu, T. 2003. A macro-finance model of the term structure, monetary
policy, and the economy. Manuscript, Federal Reserve Bank of San Francisco.
Wu, T. 2002. Monetary policy and the slope factors in empirical term structure estimations.
Federal Reserve Bank of San Francisco Working Paper 02-07.

View publication stats