This
paper mainly investigates the threshold cointegration and Granger-causality
relationships between the CPI and PPI series in the selected countries
for policymakers to effectively control inflation. We first applied the
unit root test to ensure the integration order of
all the series,
and then both the linear Engle-Granger (E-G) and
the nonlinear Enders-Siklos (E-S) cointegration tests
for comparative analysis. Lastly, Granger causality tests are adopted in the
momentum threshold vector error correction model (M-TVECM), which is used to
estimate the different speeds of adjustment and explore
the causal relationship between CPI and PPI in the selected countries. While the E-G test cannot
detect cointegration in almost all countries, the
E-S test with
higher power when there is asymmetric adjustment, supports the
cointegration relationship in Canada, Denmark, Indonesia, Japan, Pakistan,
Spain, and Uruguay. The evidence also supports the existence of asymmetric
threshold adjustment in all cointegrated systems. In
addition, the empirical results indicate that Granger causality in the M-TVECM
can be classified into two categories. One kind is about CPI leading to PPI,
including Spain
only while another kind is about bidirectional causality between CPI and PPI
for other countries in the M-TVECM.
