Banking Output & Price Indicators from Quarterly Reporting Data

Abstract

The measurement of banking output (and therefore productivity) has long been controversial. This article applies the user cost approach in Fixler and Zieschang (1999) to quarterly reporting data from Luxembourg's banking sector. This requires associating the flows in the profit-and-loss account to different assets and liabilities in the balance sheet. The user cost of each asset/liability is then calculated as the difference between the rate at which it generates revenues/costs and a "reference rate" representing the opportunity cost of funds. A negative user cost then identifies an asset or liability as an output and a positive user cost identifies it as an input in the production process. In theory, this datadriven approach is capable of combining elements of both the two traditional approaches to measuring banking output (the production and intermediation approaches) since these classify inputs and outputs on an a priori basis. In practice, our results suggest that neither of these conventional approaches is wholly consistent with the data for Luxembourg. We then use multilateral Törnqvist indices to aggregate outputs and inputs separately and show that the resulting series are robust to alternative measures of the reference rate. The difference between the output and input index provides a measure of Total Factor Productivity (TFP) and an implicit price index is also derived. Results suggest that productivity growth in Luxembourg's banking sector has been high since the mid-1990s, displaying volatile but persistent dynamics and moving pro-cyclically. Productivity varies widely across banks but larger banks (in terms of total assets) tend to be more productive.

Assessing output and productivity growth in the banking industry

“…12 9 This treatment of bank fees and securities income is in accordance with the profit and loss account approach followed by Drake et al (2006) and Asimakopoulos et al (2008). See also Guarda and Rouabah (2007) and Fixler and Zieschang (1992).…”

“…This approach was elaborated by Hancock (1985), who developed a production theory for financial firms, whose inputs and outputs are determined empirically. For a bank asset, the user cost of money is defined as the difference between a benchmark rate (representing the opportunity cost of the bank) and the interest rate (rate of return) associated with holding this asset (Guarda and Rouabah, 2007). For a bank liability, the user cost of money is defined as the difference between the interest rate associated with this liability and the benchmark rate.…”

“…6 The empirical results showed that during 1985-1988 the output of large-sized US commercial banks recorded a considerable increase. Guarda and Rouabah (2007) follow a similar empirical approach, classifying bank products as inputs or outputs according to the sign of the respective user cost.…”

“…33 See Guarda and Rouabah (2007) and Morttinen (2002) for ΤFP measurement using a Tornqvist index. 34 For more details on labor quality see the next section.…”

“…13 Guarda and Rouabah (2007) and Fixler and Zieschang (1992) used, alternatively, more than one benchmark rates; without, however, reaching significantly different results. 14 More specifically, and according to the analysis of the user cost of money in section 2.1, loans and demand deposits contributed to the increase of bank revenues all the years under examination and time deposits in 14 out of 17 years.…”