Over the last 20 years the border-effect literature repeatedly documented the trade-reducing effect of inter- and intra-national borders. Thereby, the sheer size and persistence of observed border effects from the beginning raised doubts on the genuine effect of underlying borders. However, when so-called "border effects" result either from statistical artefacts or from differences in fundamentals, why should their spatial dimension then inevitably coincide with the geography of present or past political borders? This paper identifies a discontinuous trade reduction along a geographic dimension that neither existing nor defunct political borders can explain. Trade between the East and the West of Japan is 23.1% - 51.3% lower than trade within both country parts. Including a rich set of explanatory variables, suggests that recent agglomeration trends, reflected by the contemporaneous structure of Japan's business and social networks, rather than cultural differences, shaped by long-lasting historical shocks, can explain the east-west bias in intra-Japanese trade.