How should financial intermediation services be taxed?


This research project asks fundamental questions about whether and, if so how, the returns to financial intermediation should be taxed. This has proved to be a difficult area of economic theory, in which economists have so far come to different conclusions. Part of the confusion appears to stem from there being two distinct types of financial intermediation services: savings intermediation, and payment services. Our approach distinguishes these two in a theoretical dynamic general equilibrium setting. The results indicate that, in general, there should be a tax on payment services. But when firms differ in the cost of intermediation services, then the tax on savings intermediation should be zero. Both of these results can be understood as applications of the well-known Diamond-Mirrlees production efficiency theorem.

Ben Lockwood



Financial intermediation services include services as intermediation between borrowers and lenders, insurance, and payment services. These services comprise a significant and growing part of the national economy – 7.9% in 2007. The question of whether, and how, financial intermediation services should be taxed is contentious. Within the EU, most financial services are exempt from VAT, and there is considerable debate about the possible benefits from bringing them into the VAT system.

In the debate on this topic, it is generally assumed that within a consumption tax system, such as a VAT, it is desirable to tax financial services supplied at the standard rate of VAT, and allow providers of intermediation services to claim back VAT they pay on inputs. However, this policy prescription is at variance with academic work on this issue which suggests that while payment services should be taxed at the same rate as consumption, intermediation between borrower and lender should not be taxed at all. This research takes a fresh theoretical look at these questions in a more general dynamic equilibrium setting. We study savings intermediation services and payment (or transaction) services and separately, in two different papers.

In the first paper we assume, realistically, that savings intermediation is not explicitly priced, but charged for via a spread – which can be taxed – between borrowing and lending rates set by competitive banks. The analysis suggests that the spread should in principle be subject to VAT. This is consistent with a widely accepted principle that intermediate goods should not be taxed; in this case, businesses that borrow from banks could set the input VAT on their borrowing against the VAT on their sales, as with any other input subject to VAT. But consumers would bear the cost of the VAT charged on borrowing. However, in most realistic cases, the rate of VAT would optimally be set equal to the tax rate on capital income, which is generally different to the optimal tax rate on  consumption.

In the second paper, we analyse payment services, such as the provision of credit and debit cards. All optimal tax structures distort the relative costs of payment media, by raising the cost of deposits relative to cash. In the simplest structure, cash should be untaxed and the rate of tax on transactions services could in principle be higher or lower than the consumption tax. But when the  model is calibrated to US data, simulations suggest that the transactions services tax should be considerably lower than this. This is because a transactions tax has a “double distortion”: it distorts the choice between payment media, and indirectly taxes consumption.

Ben Lockwood, CBT Working Paper 13/09; CBT Working Paper 14/23