Constructing a merger which balances the interests of Alliance & Leicester and Bank of Ireland will involve some delicate financial engineering.

The banks yesterday said they planned to implement their merger through a "dual listed company" structure, with shareholders retaining their existing holdings in the respective companies.

This means the operating units would be pooled, with Bank of Ireland receiving 55 per cent and A&L 45 per cent of future dividends and capital distributions.

A joint statement issued yesterday said the DLC structure would keep A&L's listing in London, and Bank of Ireland's twin listings in Dublin and London. This would maintain Bank of Ireland's membership of the ISEQ index and A&L's of the FTSE 100.

"Alliance & Leicester's shares are predominantly owned by UK investors, of whom slightly under half in value are retail holders, while Bank of Ireland's shareholder base is more international and institutional. The use of the DLC structure is intended to preserve the continuity of the respective shareholder bases," the two banks said.

The announcement is opaque: there is no reference to the creation of a holding company under the two listed companies.

This kind of dual listing structure has been tried before, in several variations, for example in the merger between Merita of Finland and Nordbanken of Sweden.

Merita has maintained its Helsinki listing and Nordbanken its quotation in Stockholm, each with its own shareholders. They jointly own a holding company - based in Finland for tax advantages - which controls all operating companies.

Because Nordbanken was bigger, its shareholders have a 60 per cent economic interest, but part of their holding is in non-voting preference shares until 2001, to allow the group to be split 50-50 in corporate governance terms. This is only a transitional phase, recognising the cultural and political sensitivities of such a cross-border deal.

The Fortis group in Belgium and the Netherlands also offers a parallel. So too, outside the financial sector, do structures such as Reed Elsevier, Unilever and Royal Dutch/Shell.

But these precedents cut both ways. Fortis has shown that the structure has to evolve if it is to take advantage of strategic opportunities - the acquisition of Generale de Banque in Belgium has upset the original balance between Belgian and Dutch interests. Reed Elsevier's difficulties in finding a chief executive have highlighted some of the governance issues.

In fact, the very reasons which argue for a dual structure at the outset can, in the longer run, turn into negatives.

In the case of A&L/Bank of Ireland, the structure serves two specific purposes: it handles the domestic issues involved in merging a national flagship such as Bank of Ireland into a larger group; and, cunningly, it manages to retain the takeover protection granted to A&L when it demutualised under the terms of UK building societies law.

That protection makes it very difficult for a third party to force its way in and upset the agreed deal.

But these factors also make it difficult to bid for the combined company in the future, removing any kind of bid premium from the long run valuation. There are usually persistent discrepancies between the two share prices, often for currency reasons, which are not arbitraged away. Any offer would have to be at premium to the more highly valued share.