Bank taxes in Euro­pe and the impact on bank stock mar­ket valu­a­ti­ons, divi­dend expecta­ti­ons and CDS spreads

The results of this study suggest that European publicly listed banks were not much affected by new tax introductions in the 2022-2024 period. Although there was an average negative reaction on the stock market, no clear reaction appears in either dividend expectations or CDS spreads. This indicates that there was no increased expected risk of failure to uphold sufficient capital positions. The stock market reaction can thus be attributed to general market uncertainty.

Setting the scene

Following the 2022 surge in inflation and the subsequent interest rate hikes by the ECB, the profitability of European banks has significantly improved. The current period of higher profitability marks a stark contrast with the prolonged phase of weak performance that characterized the European banking sector in the years after the Great Financial Crisis. This improvement in financial performance has not gone unnoticed by policymakers. Several European countries have responded to the rising bank profits by introducing or elevating taxes specifically targeting the banking sector. These taxes, often implemented as windfall taxes, aimed at redistributing unexpected gains. This study investigates whether such fiscal measures have had a measurable impact on banks’ stock valuations – expected risk and return, dividend expectations – expected shareholder return – and CDS spreads – expected default risk.

The methods used are largely inspired by two previous studies on the topic. The first, by De Vito et al. (2023), examined the stock market reaction to the Italian tax announcement and found that Italian bank stocks responded negatively. The second, by Martins (2024), extended this analysis in two ways. First, by also considering the Spanish tax announcement and second, by examining the reaction of foreign banks. His findings confirmed the negative stock market reaction, with the effects being particularly strong for domestic banks. In addition to these two studies, Maneely & Ratnovski (2024) was an important starting point for the literature review.

Data and method

An event study was conducted to assess the impact of the taxes. Conducting such an analysis requires event dates. To identify these, news articles were consulted to identify precise dates of the tax announcements. The primary sources, amongst others, included Bloomberg and Reuters. Based on those news articles, Table 1 presents an overview of the tax introductions between May 2022 and October 2024. The table shows that most of the taxes introduced were temporary. It is also clear from this table that most of the temporary introduced taxes are windfall taxes. As will become clear in the subsequent paragraph, the analysis was conducted based on four countries that implemented windfall taxes, and four that did not.

After the definition of the events, there is the need to determine the sample of banks to be analyzed. For the stock market analysis, the metric used are return indices retrieved from Refinitiv. Banks with insufficient trading liquidity, as well as the Caisses régionales from Crédit Agricole and the banks from Liechtenstein were excluded. This way, the sample ended up including 83 banks of which 28 are located in countries where a tax was announced – hereafter referred to as domestic banks. The tax announcements included were, in chronological order, Hungary, Spain, Czechia, Italy, the Netherlands, Romania, Belgium and Slovenia.

The 28 domestic banks served as the sample for the analysis of the dividend expectations. To ensure comparability across banks, this part of the research uses the expected dividend pay-out ratio (DPO). Professional forecaster make their expectations with regard to bank dividends, the average value of those forecasts is retrieved from I/B/E/S via Refinitiv. To construct the DPO, these were multiplied by the number of outstanding shares and divided by net income. For these two variables yearly observations were used, they were obtained via S&P Capital IQ.

The last metric that analyzed are the CDS spreads, with IHS Markit used as a data source. The banks covered in this source differ from the ones in the sources for the previously discussed variables. Banks with insufficient CDS data liquidity and duplicate variants of the same institution were excluded from the sample. This resulted in a sample of 48 banks of which 15 were domestic banks. The tax announcements subsequently examined were those of Spain, Italy, the Netherlands, and Belgium.

A look at the different markets

This section briefly summarize the main findings of the study. The results presented in this section are the ones for the average reaction of a bank to a tax announcement. This is the unweighted, averaged reaction of the domestic banks to the tax announcement in their respective country. Following De Vito et al. (2023) and Martins (2024) a negative reaction on the stock market is to be expected. Since taxes deteriorate earnings after tax, this can affect the capabilities of banks’ to distribute dividends to the shareholders in a negative way. Most of the banks in the sample were affected by a windfall tax, when excess profits are taxed no threats should arise for a bank’s creditworthiness and thus CDS spreads.

Uncertainty on the stock market

Figure 1 presents the cumulative average abnormal return (CAAR) in the stock market. This variable reflects the deviation from expected market behavior. The calculation of this variable exists of three steps. First, the expected behavior if there were no tax announcement is estimated. Second, the estimated value is subtracted from the actual return to calculate the abnormal return. Finally, the abnormal returns are cumulated over a specified period and averaged across the different banks to obtain the CAAR. Negative CAAR values indicate that a tax announcement continues to exert downward pressure on stock prices. The confidence bounds illustrate whether the observed values are statistically different from zero. These are included because any estimation is subject to some degree of error.

On the stock market, the reaction was consistent with expectations. Figure 1 shows an average abnormal loss of 2,2 percent on the day of the announcement. A modest recovery is visible within the depicted window, with the cumulative abnormal loss narrowing to -1,7 percent three days after the announcement.

To check whether the tax announcements had a lasting impact on stock prices, Table 2 was created. This table shows the CAAR values over different time periods. The last row also reports the p-values. Those tell how likely it would be to see the results if, in reality, the announcements had no effect at all.

In other words, the starting assumption is that the CAAR value should be zero – meaning no impact from the tax announcement. The p-value shows how unusual the actual outcome is under that assumption. For example, 1 day post-announcement, the observed CAAR was -2,16. If there really were no reaction, the chance of seeing a value of that magnitude is essentially zero.

A common rule is that if the p-value is below 0,05, the result is considered statistically significant – that is, unlikely to be just random noise. Since the CAAR values are consistently below zero, this suggests that tax announcements tend to push stock prices downward. Table 2 therefore indicates that, on average, a tax announcement creates uncertainty in the stock market that lasts up to 20 trading days after the event.

Robust dividend expectations

Figure 2 visualizes the DPO data from the domestic banks in two boxplots. For this variable, monthly data were used. The left boxplot displays the DPOs one month before the announcement. The right boxplot shows the values one month post-announcement. Important to notice is that the number of shares of a bank, as well as the net income do not change after an announcement. Changes in the DPO are thus completely due to changes in the dividend expectations made by the professional forecasters.

This was perhaps the most surprising finding of the conducted research: post-announcement, the average expected DPO did not really change. On Figure 2, this is indicated by the little cross that almost did not move from the first to the second boxplot. Moreover, when changes occurred at the individual bank level, they were predominantly upward. This suggests that professional forecasters expected banks to have sufficient earnings. In their view, they would remain able to pay additional taxes while maintaining dividend levels.

No increased default risk

Figure 3 gives the cumulative abnormal average spread change (CAA∆S), as in Figure 1 this is the deviation from expected market behavior. The procedure for calculating the CAA∆Ss is the same as that used for computing the CAAR values. Contrary to Figure 1, in this figure a positive value means a negative impact. As this indicates that a bank’s default risk has gone up.

On the day of an announcement there was a significant average rise of 0,76 percent. However, already the next day the cumulative effect was no longer distinguishable from zero. This can be derived from Figure 3 because of the confidence bounds around the CAA∆S estimates. When confidence bounds include zero, based on the data, zero is a plausible value for the true effect. An effect of zero means that the effective and estimated CDS behavior – the CDS behavior if there was no tax announcement – are the same. From the second day post-announcement, there was thus no longer a certain deviation from expected CDS behavior. Markets did not believe that the average bank would be subject to a higher default risk.

What to take from this?

When interpreting these results, it is essential to consider the specific dimensions captured by each metric. Stock prices incorporate both risk and expected return, offering a broad view of market sentiment. Dividend expectations, on the other hand, only reflect anticipated returns to shareholders. CDS spreads serve as an indicator of a bank’s perceived default risk, capturing the market's assessment of creditworthiness.

The short-lived reaction in CDS spreads indicates that, after an initially sharp response, financial markets quickly reassessed the situation. Within a relatively short period, they appeared to conclude that the proposed taxes would not pose a substantial threat to the affected financial institutions. This suggests that initial concerns were either overstated or effectively mitigated by subsequent clarifications.

The almost missing reaction on the front of the dividend expectations indicates that forecasters did not expect banks to pass the tax levy on to their shareholders. It may suggest that analysts assumed banks would absorb the cost themselves. Forecasters might believe that bank profits would be sufficiently high to cope with the additional tax burden without the need to reduce dividends.

When one now thinks about the findings with regard to stock prices, it may seem contradictory that the CAAR stayed beneath zero for a prolonged period. While default risk and shareholder returns did not showed strong responses to the announcements, stock markets reacted strongly. This could reflect a large amount of uncertainty amongst investors on the stock market.

These results suggest that publicly listed banks were not expected to be severely affected by the announced tax measures. However, given that market reactions fundamentally reflect expectations rather than realized outcomes, further research is necessary to assess the actual long-term effects of these taxes on bank performance and stability.

References

De Vito, A., Pancotto, L., Perdichizzi, S., & Reghezza, A. (2023). Don’t go on holiday in August! Market reaction to an unexpected windfall tax on banks. Economics Letters, 233. https://doi.org/10.1016/j.econlet.2023.111407

Maneely, M., & Ratnovski, L. (2024). Bank Profits and Bank Taxes in the EU, WP/24/143, July 2024.

Martins, A. M. (2024). Banks stock market reaction to the Italian and Spanish windfall tax announcement: an event study. Journal of Banking Regulation. https://doi.org/10.1057/s41261-024-00246-x