Policy rate changes may affect banking system deposit and loan volumes. This matters for how the monetary policy cycle impacts bank profitability, as well as monetary policy pass-through. Today’s post summarises how South African bank-level deposit and loan balances have changed with policy rate changes.
The first chart shows that nominal domestic deposit and private sector lending growth has been mostly positive in South Africa across all tightening and easing episodes since 2008, even if one accounts for the impact of inflation.
There has, however, been a lot of heterogeneity in individual bank deposit and loan betas, as well as large differences in loan and deposit growth across tightening and easing cycles. The second and third charts show that betas have generally been larger during tightening cycles than during easing cycles. Capitec’s deposit and loan betas have been the highest, reflecting the bank’s rising market share over the period, not that its deposit and loan growth have been more cyclically sensitive. Nedbank and Investec have had higher betas for loans than for deposits across both easing and tightening cycles, while ABSA’s betas were negative on average over easing cycles and positive during tightening cycles.
What do these estimates mean for monetary policy pass-through? The estimates presented are unconditional, in the sense that they do not control for the macroeconomic and bank-specific factors that affect credit supply and demand. Econometric evidence around the pass-through of policy rate changes to deposit and lending volumes suggest that pass-through to credit extension weakened post-pandemic as weaker credit demand and offsetting impact of increases in market risk and liquidity have dominated the impact of policy changes.
In a previous post, I also summarised evidence of a weakening of pass-through of monetary policy tightening to borrowing costs in South Africa. Part of the reason for this has been that bank funding costs did not match policy rate cuts following the onset of the COVID pandemic, which meant that financial conditions did not loosen as much as the lower policy rate would otherwise have implied.
This has implications for the conduct of monetary policy. As we showed in this paper, estimates of monetary policy pass-through are mis-specified unless one controls for bank-level funding costs and other drivers of bank-specific rate settings. It is becoming common for central banks to incorporate the possibility of time varying monetary policy transmission and funding spreads in their modelling. In the case of the South African Reserve Bank, its policy model does not separately account for bank funding spreads and lending spreads.
Footnote
Note that our estimates allow for a three month lagged pass-through to deposit and loan volumes.
Unlike the US banking system, South Africa’s net interest margins for the banking sector and the slope of the yield curve is weakly correlated. It is worth pointing out that assessing the implications of the monetary policy cycle for bank profitability requires looking beyond just net interest margins. Our models suggest that impairment and cost structures data is particularly important. Unfortunately, the bank regulator in South Africa chooses not to publish the bank-level income statement and prudential regulatory compliance data that it collects. However, our EconData platform makes it easy to update models that use public domain banking data.
IMF estimates suggest that among South Africa’s five largest banks, all except FirstRand have higher interest income betas than expense betas, with Standard bank having the highest positive differential. As today’s post suggests, that finding likely depends on the definition used for income and expense betas.
