Employability beyond skills: uneven conversion of capital for international graduates, and recognition regimes in host and home labour markets

International higher education is increasingly justified through an employability promise: that graduates acquire transferable skills that translate into labour-market success. Yet international graduates often experience uneven employment outcomes despite reporting strong employability capabilities. This paper argues that such disparities cannot be explained by skill possession alone. Instead, employability depends on how graduates’ capital is recognised, performed and converted into value within specific labour-market contexts. Drawing on a cross-national study of international Chinese business graduates, we advance the concept of employability recognition regimes: spatially situated systems of evaluation through which employers interpret and valorise graduate attributes. Empirically, the study combines Graduate Outcomes Survey data from business graduates at a major Australian university, textual analysis of 107 graduate recruitment advertisements in Australia and China, and semi-structured interviews with 20 employers and HR experts. The findings show that employability capital is differentially recognised across host- and home-country labour markets. In Australia, employability is evaluated through individualised signalling, narrative performance, learning agility and professional presence within a service-oriented economy. In China, evaluation is shaped by credentialist and relational logics emphasising institutional prestige, credential verification, organisational alignment, resilience and interdisciplinary capability linked to digital and industrial transformation. The paper contributes to employability scholarship by theorising recognition regimes as systems that mediate the cross-border conversion of graduate capital into labour-market value. It also informs transnational employability support that helps graduates translate their skills, credentials and experiences into locally credible signals.