Fitting a dynamic yield macro model in a frontier market: evidence from Kenya

Abstract

We examine the dynamic interaction between macroeconomic factors and the unobserved yield curve factors in Kenya between 1 March 2011 and 31 March 2014. First we parametarise the yields only Term Structure Model using a Dynamic Nelson Siegel approach. We then characterise the dynamic interaction between the extracted unobserved yield curve latent factors and the observed macroeconomic factors, (inflation and CBR) using a Vector Autoregressive approach. We find that yield curve own dynamics account for the greatest percentage variation in the forecast error variance of the latent factors with the level factor accounting for a highly significant percentage of the one month forecast variance in its own dynamics and that of the slope and curvature factors, (100, 84 and 87 per cent respectively). The macro factors chosen on the other hand do not account for any of the variation in latent factors earlier during the forecast horizon though increases in their significance as the forecast horizon increases. We conclude that a large amount of idiosyncratic variation relate to the yield curve that is unrelated to observed macroeconomic factors in the short term with the observed macro factors becoming more influential as the forecast horizon increases.