South African Reserve Bank : Estimates of bank-level funding costs in South Africa

South African Reserve Bank

Working Paper Series

WP/21/05

Estimates of bank-level funding costs in South Africa

Tim Olds and Daan Steenkamp

Authorised for distribution by Konstantin Makrelov

12 April 2021

© South African Reserve Bank

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means without fully acknowledging the author(s) and this Working Paper as the source.

South African Reserve Bank Working Papers are written by staff members of the South African Reserve Bank and, on occasion, by consultants under the auspices of the South African Reserve Bank. The papers deal with topical issues and describe preliminary research findings, and develop new analytical or empirical approaches in their analyses. They are solely intended to elicit comments and stimulate debate.

The views expressed in this Working Paper are those of the author(s) and do not necessarily represent those of the South African Reserve Bank or South African Reserve Bank policy. While every precaution is taken to ensure the accuracy of information, the South African Reserve Bank shall not be liable to any person for inaccurate information, omissions or opinions contained herein.

South African Reserve Bank Working Papers are externally refereed.

Information on South African Reserve Bank Working Papers can be found at https://www.resbank.co.za/en/home/publications/Papers/working-papers

Enquiries relating to the Working Paper Series can be addressed to:

Head: Economic Research Department

South African Reserve Bank

P O Box 427

Pretoria 0001

Tel. [phone]

Estimates of bank-level funding costs in South Africa

Tim Olds* and Daan Steenkamp†

12 April 2021

Abstract

We develop a new dataset of bank-level balance sheets data to estimate bank-level funding costs. These estimates are useful for monitoring funding pressures and other risks to the banking sector as well as understanding the impact of prudential regulations and market conditions on the transmission of monetary policy. We show that bank funding cost spreads are materially higher now than before the Global Financial Crisis of 2008, in spite of lower interest rates. We show that during the COVID-19 crisis, aggregate funding costs have fallen in level (i.e. percentage) terms, but that funding costs have increased when expressed relative to reference rates. We show that the relative cost of raising deposits has increased, as deposit rates have not fallen by as much as the repurchase rate and other money market rates.

JEL classification: E40, E44, G21

Keywords: bank funding costs, composition of funding, financial market conditions

• South African Reserve Bank, PO Box 427, Pretoria, South Africa, 0001. Email: [e-mail].
  † SARB. Corresponding author. Email: [e-mail].

1

1. Introduction1

This paper measures the costs of bank funding using bank-level data on bank funding and the costs of different components of bank liabilities. The cost of bank funding affects the cost of lending to businesses and households and therefore has implications for monetary policy, by affecting growth and inflation. The extent to which banks pass on funding cost changes to specific lending rates may vary over time, as banks decide whether to absorb such changes and accept a change in the margin that they earn when making loans. This is likely to depend on factors such as the state of the economy (and therefore the demand for loans) or competition for bank lending. Given potential impacts of bank profitability, bank funding costs may also have implications for financial stability.

Even though bank funding costs matter for monetary policy and financial stability, there has been relatively little work done to develop methodologies for accurate measurement. Most papers that create proxies of funding costs tend to use market indicators of funding spreads. For example, Beau et al. (2014) proxy their funding spreads on long-term wholesale funding spreads or credit default swap (CDS) premia.2 Jondeau et al. (2020) construct forward funding spread indicators for the United States and euro area that approximate the cost of obtaining bank funding on money markets.3 Illes et al. (2019) construct the weighted average cost of bank liabilities for 11 European banking systems,4 while Kapuscinski and Stanislawska (2018) use the same approach as Illes et al. (2019) to measuring funding costs for the Polish banking sector.

For South Africa, Rapapali and Steenkamp (2020) construct a simplified measure of bank funding costs for the South African banking sector, and show that their measure is similar to a measure obtained by weighting together surveyed funding costs from major banks. This paper refines the methodology developed by Rapapali and Steenkamp (2020) and applies it to individual bank data. We scrape bank funding data from the BA900 survey data from the South African Reserve Bank (SARB) website for all 36 currently registered banks. We also use confidential funding rate data from the BA930 survey at bank-level. We create funding cost estimates for the aggregate banking sector and individual banks, as well as on a component-by-component basis at bank- and industry-level. Ours is the first study we know of to produce granular bank-level estimates based on bank balance sheet data and data on the actual costs of funding components. We only show aggregate results in this paper to preserve the confidentiality of the underlying bank-level data.

We argue that these estimates are useful for monitoring funding pressures and other risks to the banking sector, as well as understanding the impact of prudential regulations and market conditions on the transmission of monetary policy. The contribution of our paper is to summarise how banks have responded to the changes in market conditions and regulatory

• We are grateful for comments and suggestions from two anonymous referees and help with data set construc- tion from Lisa De Beer, Danie Meyer, Tabea Mokotong, Lesego Morope, Wessel Mostert, Pontso Ndobo and Myrtle Van Jaarsveld.

• Specifically, they proxy secondary market bond spreads using the constant-maturity unweighted average of secondary market spreads to swaps on senior unsecured bonds, spreads on retail bonds as spreads for fixed- rate retail bonds over equivalent-maturity swaps, or using CDS premia based on an unweighted average of large bank senior CDS premia.

• They proxy bank rollover risk associated with maturing short-term wholesale funding using the spread between three-month interest forward rate and the corresponding default-free overnight interest swap forward rate.

• They use data on the volume of deposits, short-term debt securities (under 1 year maturity), covered bonds and funding from central bank operations for each banking system, alongside market rates to proxy the new business rates applicable to each form of liability funding.

2

requirements post-Global Financial Crisis (GFC), and to assess the implications for the costs of individual bank funding. We also focus on describing developments in funding costs in the build-up to and during the COVID-19 crisis, and compare funding costs between large and small banks.

We show that bank funding costs are materially higher now than before the GFC of 2008, in spite of lower interest rates. We show that this reflects, in large part, an increase in the relative cost of raising deposits. We show that during the COVID-19 crisis, aggregate funding costs have fallen in level (i.e. percentage) terms, but that funding costs have increased when expressed relative to reference rates. We also assess the heterogeneity in funding costs for large and small banks, and show that smaller banks have lower funding costs for deposits, while being more dependent on deposit funding, meaning that they have lower liability-side funding costs overall.

2. What are funding costs?

To fund the loans that banks make to consumers and firms, banks obtain funding from retail and corporate wholesale deposits, issuing debt instruments, as well as from their capital base, which comprises the owners' equity in the bank and accumulated profits.5

Figure 1 provides a simplification of the balance sheet of a typical bank in South Africa. The sources of bank funding are recorded on the liability-side of their balance sheets and the uses of bank funding on the asset-side (see Beau et al. 2014 for a more detailed discussion). The asset and liability sides of bank balance sheets typically have maturity mismatches: many sources of funding, such as 'on-call' retail deposits, could be withdrawn from banks at any time, while a large proportion of their loans (i.e. mortgages) will have a much longer maturity. Banks therefore have to manage maturity mismatches, which they can do, for example, by ensuring they have sufficient stable funding and could replace withdrawn or maturing funding if needed. Government regulations also stipulate that banks have to maintain reasonable levels of liquid assets to be able to accommodate unexpected changes in funding conditions. Since such regulations reduce the amount of loans a bank could make for a given quantum of funding and channel funding to specific types of assets (usually government securities), they impose an implicit cost on banks. The methodology used to estimate bank funding focuses on the liability-side of bank balance sheets, given the availability of granular survey data from the BA900 survey conducted monthly by the SARB.6

• 'Retail' funding typically refers to funds obtained from individuals, households and small firms, usually in the form of deposits. 'Wholesale' funding typically refers to funding obtained from large corporates, banks, pension funds or the insurance industry or investors.

• A future working paper in this series by Diesel et al. (2021, forthcoming) will estimate the implied cost of regulatory liquid assets for South African banks, which allows for assessment of the additional costs stable funding and liquidity buffer requirements imply for South African banks.

3