Asymmetric Directional Information Flow Between Output and Labour in Malaysia: An Information-Theoretic Analysis

Abstract

Empirical studies of output-labour relationships typically rely on regression-based causality models that assume linearity within a specified parametric form. These methods may obscure variations in the direction and strength of causal influence in non-linear, heterogeneous adjustment processes. This paper examines the directional information flow between output and labour in Malaysia using an information-theoretic approach that operationalises causality as the reduction of directional uncertainty. Pairwise causal interactions are analysed using Shannon transfer entropy on quarterly data series from 2011 to 2022 for real gross domestic product (GDP) growth, employment growth, and unemployment growth. The findings reveal a structure of asymmetric and hierarchical information flow. Unemployment growth exerts the strongest directional influence on real GDP growth, while employment growth demonstrates a modest unidirectional information flow to output, significant at the ten per cent level. Within the labour market, employment growth unidirectionally influences unemployment growth with no statistically significant reverse flow. The interaction between unemployment and output is bidirectional but asymmetric, with unemployment playing the dominant role. These results suggest that output-labour dynamics in Malaysia are characterised by selective directional dominance rather than uniform bidirectional feedback. Such asymmetric causal structures are difficult to isolate using parametric causality methods due to their reliance on symmetry assumptions and model specification. This study demonstrates the value of information-theoretic inference in analysing complex macroeconomic systems with heterogeneous adjustment dynamics.