Nonlinear behaviour in EMBI series from Eastern Europe: evidence of ‘window size effect’

Abstract

The  main objective of this article is to explore how the number of observations included in a window may impact the rejection rate of linear windows. We employ two methodologies proposed in the literature in order to study the linear and nonlinear behaviour. We use daily Emerging Markets Bond Index (EMBI) index spreads from six of the most important Eastern European countries (Bulgaria, Hungry, Poland, Russia, Serbia and Ukraine). The empirical results show what we call ‘window size effect’ because when we include more than 50 observations in each window, the rejection rate increases using the two different methodologies. Therefore, our findings support the idea that, even in this well informed and sophisticated market, the weak-form of the efficient market hypothesis cannot be confirmed.

Keywords:

• Hinich test
• nonlinearity
• EMBI

JEL Classification:

• C12
• C14
• G10
• G14
• G15

Notes

¹  Nonoverlapping windows mean that they do not overlap with each other (they are independent). The Hinich test can work with different lengths of windows avoiding overlapping.

²  Readers interested in a mathematical derivation of this statistic and its properties can refer to Hinich (1996).

³  Belonging to the disbanded Federal Republic of Yugoslavia (2003), Serbia and Montenegro separated in 2006; ever since then, Serbia has been a sovereign country.

⁴  Pb = bases points.