by J. Christina Wang, Susanto Basu, and John G. Fernald
No. 2, September 2004 - December 2004

Motivation for the Research
In many service industries, measuring real output is a challenge because it is difficult to measure quality-adjusted prices. The financial services industry lacks even an agreed-upon conceptual basis for measuring nominal, let alone real, output. Conceptually, the most vexing measurement issue arises because banks and other financial service providers often do not charge explicit fees for services, but rather incorporate the charges into an interest rate margin—the spread between the interest rates they charge and pay.

In this paper, the authors address the lack of an agreed-upon conceptual basis for measuring bank output and propose resolutions of some major long-standing debates on this issue.

Research Approach
The authors develop and analyze an optimizing model with financial intermediaries that provide financial services to resolve asymmetric information between borrowers and lenders. These intermediaries are embedded in a dynamic, stochastic, general-equilibrium model in which assets are priced competitively according to their systematic risk, as in the standard consumption capitalasset- pricing model.

Key Findings

• The model demonstrates the conceptual shortcoming in the existing national accounting measure of bank output: By counting the risk premium as part of nominal bank output, the current national income accounting measures treat economically identical alternative funding institutions differently and, as a result, also alter the output of the borrowing firm depending on its source of funding.
  
  
• The correct reference rate for measuring nominal bank lending services must incorporate the borrower’s risk premium; that is, the borrower’s risk premium is not part of bank output, and one should use an ex post, rather than an ex ante, measure of the risk premium on bank funds in the reference rate.
  
  
• The logic of the model also applies to loans to households (for example, mortgages and credit cards). The implication is that, to avoid overstating GDP, one should not count the risk premium in such loans as part of either bank output or the consumption of financial services by the household sector.
  
  
• The model highlights the conceptual problem in the bank output measure employed by the micro banking literature, which uses the deflated size of banks’ portfolio of interest-bearing assets (loans plus market securities).
  
  
• The appropriate price deflator for financial services is not generally the overall price level; financial services are a kind of information product, similar to other information processing services, such as consulting.
  
  
• Capital gains should be counted as part of financial services output only if the return is an anticipated implicit compensation for actual services rendered.

Implications
The model yields one overarching principle for measuring bank output: Focus on the flow of actual services provided by banks. This principle applies equally to measuring nominal and real banking output—and, by implication, to measuring the implicit price deflator for financial services.

The model, and its implied measure of bank output, can be readily applied to valuing implicit services by financial institutions other than banks, such as insurance companies. The general principle is the same: Apply asset pricing theories to price the financial instrument by itself; the difference between that value and the security’s actual value yields the nominal value of the implicit services.

More generally, the authors advocate a model-based approach to measurement for conceptually challenging areas of financial services.

Full text of Working Paper 04-7