MONETARY POLICY REPORT | SUMMER 2024 (box 2)
(authors: Eva Hromádková, Ivana Kubicová, Branislav Saxa)
The Czech economy saw a rapid rise in the CNB’s key monetary policy interest rates in 2021–2022. This was followed by flat interest rates for roughly a year and a half and then a phase of rate cuts from the end of 2023 to the present. The current and expected future repo rate affects market rates, which in turn pass through to client lending and deposit rates. This box describes the changes in the completeness and speed of a part of the monetary transmission mechanism, specifically the pass-through of market rates to client rates on koruna-denominated loans. The results indicate that in recent years, the pass-through has been complete, taking place entirely within one month, in the case of firms and almost complete, but with a time lag, in the case of mortgages.
The analysis in this box[1] assumes that the client lending rate is composed of the cost of funding (approximated with a market rate of comparable maturity) plus a mark-up affected by, among other things, the risk premium. The choice of specific market rate relevant to a given type of loan (the reference rate) is based on information about bank pricing policy and the results of an analysis of the correlation of selected market rates with client rates on individual types of loans. As the reference rate, we used the 3M PRIBOR for loans to non-financial corporations and the 5-year interest rate swap (5Y IRS) rate[2] for housing loans to households. The aggregate risk mark-up is time-variable. Its fundamental value was approximated using the output gap for loans to non-financial corporations and the unemployment rate for housing loans.
The importance of compensating for credit risk can be illustrated during periods of negative economic shocks and expected weaker economic activity using the spread between the client rate and the reference rate. The spread between the 3M PRIBOR and the rate on koruna-denominated loans to non-financial corporations (see Chart 1) more than doubled after the outbreak of the Global Financial Crisis (GFC) in 2009 (from 1 pp to almost 2.5 pp). In subsequent years, it stabilised between 1.5 pp and 1.8 pp. In 2020, the shock related to the Covid pandemic caused the spread to increase again to slightly above 2 pp. This time, however, the high uncertainty did not affect the spread so dramatically, as unprecedented government help mitigated the risks either directly through credit guarantees or indirectly through broad financial assistance for firms. Currently, the spread is around 1.5 pp and client rates on koruna-denominated corporate loans are following the gradual decline in the 3M PRIBOR.[3]
Chart 1 – Client rates on koruna-denominated loans to non-financial corporations follow the 3M PRIBOR with no time lag; the spread between the rates is wider at times of negative economic shocks
%; spread in pp
A slightly different pattern can be observed for rates on housing loans (see Chart 2). The spread between the client rate and the reference rate in this market increased after the beginning of the GFC and remained elevated for years, despite a sizeable decline in both the reference rate and the client rate. A substantial increase in the IRS rate observed between 2016 and 2018 led to an increase in the client rate. However, strong competition and lagged transmission of the IRS rate to client rates resulted in the client rate persisting at lower levels than implied by the increase in the IRS rate, and the spread narrowed substantially. After the outbreak of the Covid pandemic, the spread widened again, partly because of increased uncertainty, but also due to an abrupt and substantial decline in the reference rate. The spread was negative for a time in 2022, as the IRS rate rose more quickly and more than the housing loan rate. The spread is currently close to 1 pp.
Chart 2 – The housing loan rate responds to changes in the IRS rate with a visible lag
%; spread in pp
Econometric analysis – loans to non-financial corporations with the 3M PRIBOR as reference rate
The coefficient of immediate pass-through of the 3M PRIBOR to client rates on koruna-denominated loans to non-financial corporations is statistically significant for almost the whole estimation period (see Chart 3). This means that changes in the reference rate are transmitted to the client rate in the same month. Until approximately 2015, the immediate pass-through of the 3M PRIBOR to the client rate hovered around one, but this fast part of the transmission subsequently weakened somewhat and the estimates that take the latest developments into account imply approximately 60% pass-through of a change in the 3M PRIBOR to client rates within one month. According to our estimates, the long-run pass-through of the 3M PRIBOR to client interest rates (see Chart 4) was complete until 2017, after which the coefficient dropped slightly to around 0.8. In the estimates covering the last two years, it has returned to around 1, i.e. full transmission.
Chart 3 – Almost 60% of the pass-through of changes in the 3M PRIBOR to rates on new loans to non-financial corporations takes place within one month
coefficient of short-term pass-through of 3M PRIBOR to client rate (within one month); confidence interval
Chart 4 – From the long-term perspective, the transmission to client rates for non-financial corporations is complete
coefficient of long-term pass-through of 3M PRIBOR to client rate; confidence interval
Note to Charts 3 and 4: The red areas denote periods in which the cointegration between the variables is not statistically significant, computed based on the F-statistics and t-statistics of the whole cointegration relationship. If both statistics are significant at least at the 10% level of significance, the long-term equilibrium is statistically significant. The grey band denotes a confidence interval of two standard deviations on either side of the solid line.
Econometric analysis – housing loans with the 5Y IRS rate as reference rate
The persistence of client rates on housing loans is substantial and determined mainly by standard commercial practices. Our estimates indicate that the typical lag between a change in the reference rate and the maximum impact on the client rate is 4 months. The estimated coefficient of immediate pass-through from the 5Y IRS rate to the housing loan rate is therefore around zero in the long run. Only recently has it moved to a statistically significant positive – though still low – level (see Chart 5). The key (for transmission) long-run pass-through of a change in the 5Y IRS rate (see Chart 6) is in the range of 0.6–1 throughout the estimation period. It was almost complete in the 8-year rolling windows ending in the period of 2015–2017 and also in 2023–2024. In the case of housing loans, the transmission can thus be considered complete at present.
Chart 5 – In the case of housing loans, the response of client rates to a change in the 5Y IRS rate is spread over time; this is reflected in a statistically insignificant, or very small, immediate (short-term) pass-through coefficient
coefficient of short-term pass-through of 5Y IRS rate to client rate (within one month); confidence interval
Chart 6 – The transmission in the housing loans segment can also be considered complete from the long-term perspective
coefficient of long-term pass-through of 5Y IRS rate to client rate; confidence interval
Note to Charts 5 and 6: The red areas denote periods in which the cointegration between the variables is not statistically significant, computed based on the F-statistics and t-statistics of the whole cointegration relationship. If both statistics are significant at least at the 10% level of significance, the long-term equilibrium is statistically significant. The grey band denotes a confidence interval of two standard deviations on either side of the solid line.
The above implies that the pass-through of changes in market rates to client rates is currently complete and, taking into account the specifics of each segment, fast as well. From the monetary policy perspective, we can thus say that the interest rate channel of transmission is working as normal and as expected.
[1] The analysis employs monthly data on market and client interest rates covering the period from January 2004 to April 2024. The econometric estimate was performed using an autoregressive distributed lag (ARDL) model, which, in its error-correction representation, distinguishes both the existence of a long-run relationship between the variables considered and the gradual return to the estimated equilibrium via the short-term dynamics of the dependent variable. In contrast to the standard approaches, which measure transmission using the entire available time period, the coefficients for both the long-run relationships and the short-term dynamics are estimated for rolling time windows of a constant length (6 years for non-financial corporations and 8 years for housing loans to households). This allows us to identify potential changes in the strength and speed of the pass-through. Detailed information on the data, the methodology and the interpretation of the results can be found in Eva Hromádková, Ivana Kubicová and Branislav Saxa (2023): How does interest rate pass-through change over time? Rolling windows and the role of the credit risk premium in the pricing of Czech loans.
[2] The average fixed interest rate period for mortgages (which make up the bulk of housing loans) increased until around 2020, peaking above 7 years. It then mostly decreased, falling below 5 years in 2023, partly because banks became concerned about offering long fixed-rate periods due to potential losses on early repayment.
[3] Koruna-denominated loans currently make up around 50% of loans to corporations. How euro-denominated loans to Czech corporations respond to changes in ECB rates is shown, for example, in the box How client interest rates on loans and deposits have changed over the last two years in the Summer 2023 MPR.