Keywords

2.1 Introduction

During the coronavirus pandemic, central banks have been entrenched as the first and main line of defence against economic and financial crises (Stiglitz et al., 2020). Cochrane (2020) indicates that there are some arguments for a sharp interest rate cut when serious economic disruption hits, but first of all governments need detailed, pandemic-induced financial crisis plans. The pandemic period constitutes a powerful reminder that the basic political and economic unit is the nation-state, and the borders suddenly do matter (Stiglitz et al., 2020). In March and April 2020, 21 central banks in advanced and emerging economies made quantitative easing announcements on government or government-backed securities. Some of these interventions were coupled with the substantial interest rates cuts, for instance, in Poland (Rebucci et al., 2020).

Changes in interest rates are crucial for investment decisions because a one-time increase (decrease) in rates decreases (increases) the market value of assets. Bae (1990) distinguishes current (anticipated) from unanticipated changes in interest rates. Ehrmann and Fratzscher (2004) document that equity returns react more strongly to monetary policy shocks when changes in interest rates are unexpected. Nissim and Penman (2003) point out that unanticipated shifts in interest rates should revise expectations of current and near-future revenues, expenses and earnings according to the direction of the interest rate change.

Unexpected changes in the term structure of interest rates unknown a priori to investors arise particularly often in economic downturns (Cieslak, 2018). According to the literature, investors have a greater appetite for risk-taking when interest rates are low, and it is not confined to institutions. In the recent paper, Lian et al. (2019) demonstrate that low-interest rates lead to significantly higher allocations to risky assets among diverse populations of institutional and individual investors. As a consequence, even rather conservative investors that usually exhibit a medium level of risk aversion seek more risky alternative investment opportunities. Several studies provide empirical evidence that banks, pension funds and mutual funds invest more in riskier assets when spot rates are low (e.g. Maddaloni & Peydró, 2011; Hanson & Stein, 2015; Andonov et al., 2017; Di Maggio & Kacperczyk, 2017; Daniel et al., 2018). In this context, a substantial rise of investors’ activity on a stock market is not surprising when interest rates are low. Therefore, the proposed research question is as follows:

  • Did declining interest rates during the pandemic period affect investors’ activity on the Polish stock market?

To answer the research question, daily trading volume is used to investigate investors’ activity on the Warsaw Stock Exchange (WSE). Trading volume is treated as a proxy of stock liquidity, and it is defined as a number of shares traded over a particular period of time (Olbryś & Oleszczak, 2020). The event window method is used to assess whether the mean results of daily trading volume during the pre-event, event and post-event time windows significantly differ between each other in the case of all event windows. An event means a particular considerable decline in spot rates. However, a particular event window includes consecutive declines in interest rates. As a robustness check, we compare the empirical findings within the whole sample period and the event windows.

It is well documented in the literature that volume displays significant differences across trading hours of the day and across days of the week. Quite extensive studies of this topic have been conducted on the WSE (e.g. Będowska-Sójka, 2014; Nowak & Olbryś, 2015; Miłobędzki & Nowak, 2018; Olbryś & Oleszczak, 2020). However, to the best of the author’s knowledge, the empirical findings concerning investors’ activity measured by daily trading volume on the WSE during the COVID-19 pandemic period are novel and have not been reported in the literature thus far.

The remainder of this study is organized as follows: Section 2.2 presents the problem of unexpected changes in interest rates during the COVID-19 pandemic period in Poland. Section 2.3 describes the methodological background of the event window approach. Section 2.4 contains data description and discusses the empirical results of investors’ activity on the WSE. The last section recalls the main findings, concludes and indicates further directions of the research.

2.2 Changes in Interest Rates During the Pandemic Period in Poland

During the pandemic period in Poland, interest rates have been substantially reduced by National Bank of Poland (NBP) to support economy and to enable the entrepreneurs to take out low-interest loans. As a consequence, within three months, the WIBOR 1Y (Warsaw Interbank Offered Rate 1Y) fell from 1.84% (27 February 2020) to 0.30% (4 June 2020). Table 2.1 reports details concerning daily percentage changes in WIBOR 1Y. The shifts that arose on 16 March 2020, 17 March 2020, 18 March 2020, 9 April 2020 and 29 May 2020 were the most critical.

Table 2.1 Daily percentage changes in WIBOR 1Y during the period from 20 February 2020 to 12 June 2020
Full size table

Figures 2.1 and 2.2 illustrate the term structure of the WIBOR 1Y rates. For instance, Fig. 2.1 shows the almost flat interest rate structure in Poland during the one year period from 12 June 2019 to 12 June 2020. Figure 2.2 displays results reported in Table 2.1. Percentage changes in WIBOR 1Y within the period from 20 February 2020 to 5 June 2020 are presented.

Fig. 2.1
figure 1

WIBOR 1Y during the one year period from 12 June 2019 to 12 June 2020. (Source: Olbryś, 2020)

Full size image
Fig. 2.2
figure 2

Daily changes in WIBOR 1Y within the period 20 February 2020–5 June 2020. (Source: Olbryś, 2020)

Full size image

It is important to note that after the quite long time period of almost flat interest rate structure in Poland, the topic regarding financial market risk caused by changes in interest rates has become significant again in 2020.

2.3 The Event Window Approach

The event window framework is typical for event studies in economics and finance (MacKinlay, 1997). The event window approach requires constructing reasonable windows around the event dates (Di Maggio & Kacperczyk, 2017).

The aim of our event window study is to deeply investigate and compare daily trading volume behaviour around events related to the most critical changes in interest rates reported in Table 2.1. Defining [T 1, T 2] as the event window that contains k-trading days, we base on MacKinlay (1997) and propose the following procedure for constructing the non-overlapping pre-event, event and post-event windows of equal length for daily data:

  • Pre-event window [T 1 − k, T 1 − 1] (k-trading days),

  • Event window [T 1, T 2] (k-trading days),

  • Post-event window [T 2 + 1, T 2 + k] (k-trading days).

Given daily percentage changes in WIBOR 1Y presented in Table 2.1, we consider the following four main event windows EW1, EW2, EW3 and EW4, with the corresponding pre-event and post-event time windows of equal length:

  1. 1.

    Event window 1 (EW1) from 16 March 2020 to 23 March 2020 (6 trading days)

    • Pre-event window 1 (Pre-EW1) from 6 March 2020 to 13 March 2020,

    • Post-event window 1 (Post-EW1) from 24 March 2020 to 31 March 2020.

  2. 2.

    Event window 2 (EW2) from 8 April 2020 to 14 April 2020 (3 trading days)

    • Pre-event window 2 (Pre-EW2) from 3 April 2020 to 7 April 2020,

    • Post-event window 2 (Post-EW2) from 15 April 2020 to 17 April 2020.

  3. 3.

    Event window 3 (EW3) from 29 May 2020 to 4 June 2020 (5 trading days)

    • Pre-event window 3 (Pre-EW3) from 22 May 2020 to 28 May 2020,

    • Post-event window 3 (Post-EW3) from 5 June 2020 to 12 June 2020.

  4. 4.

    Event window 4 (EW4) from 16 March 2020 to 4 June 2020 (55 trading days)

    • Pre-event window 4 (Pre-EW4) from 2 January 2020 to 13 March 2020,

    • Post-event window 4 (Post-EW4) from 5 June 2020 to 17 August 2020.

2.4 Data Description and Empirical Results on the WSE

This section describes the data set and presents empirical findings concerning comprehensive investigation of investors’ activity on the WSE during the whole sample period from 2 January 2020 to 30 December 2020 (252 trading days), including the period of considerable declines in interest rates on Polish financial market, i.e. from 16 March 2020 to 4 June 2020.

2.4.1 Data Description

The data set consists of daily data and contains the opening, high, low and closing prices and volume for each security. The free historical market data come from the Stooq.pl (available at https://stooq.com/db/h/). The 18 major and most actively traded WSE companies and the WIG20 index as a benchmark are investigated. The equities were incessantly included in the WIG20 index within the whole sample period from 2 January 2020 to 30 December 2020. This information is based on the historical index portfolios (available at https://gpwbenchmark.pl/en-historyczne-portfele). Table 2.2 reports the basic information about all selected companies.

Table 2.2 Basic information about the 18 major and most liquid WSE-listed companies included in the WIG20 index within the whole sample period
Full size table

The whole sample period can be divided into the following two sub-periods: (1) before the COVID-19 pandemic lockdown from 2 January 2020 to 12 March 2020 and (2) during and after the lockdown from 13 March 2020 to 30 December 2020. Daily values of the WIG20 index are presented in Fig. 2.3. It is pertinent to note that after the quite short period when the WIG20 index moved down substantially (with the lowest value equal to 1305.73 on 12 March 2020—the first day of the lockdown), the market index moved up within the subsequent sub-period.

Fig. 2.3
figure 3

Daily values of the WIG20 index within the whole sample period from 2 January 2020 to 30 December 2020

Full size image

2.4.2 Empirical Results on the WSE Based on the Event Window Approach

The main idea of the event window approach is to investigate a particular stock market characteristic’s behaviour around events. Therefore, the goal of this study is to assess whether the mean results of daily trading volume during the pre-event, event and post-event time windows significantly differ compared to each other in the case of all event windows. To address this issue, the t-statistic for sample means is utilized:

$$ t=\frac{\overline{x_1}-\overline{x_2}}{\sqrt{\frac{s_1^2\left({n}_1-1\right)+{s}_2^2\left({n}_2-1\right)}{n_1+{n}_2-2}\cdot \left(\frac{1}{n_1}+\frac{1}{n_2}\right)}}, $$
(2.1)

where \( \overline{x_1}\ \mathrm{and}\kern0.5em \overline{x_2} \) are sample means, \( {s}_1^2\ and\ {s}_2^2 \) are sample variances, and n 1 and n 2 denote sample size, respectively. The following two-tailed hypothesis is tested:

$$ {\displaystyle \begin{array}{c}{H}_0:{\mu}_1={\mu}_2\\ {}{H}_1:{\mu}_1\ne {\mu}_2\end{array}}, $$
(2.2)

where μ 1 and μ 2 are the expected values of daily trading volume during the compared periods, and the null hypothesis states that two expected values are equal. Tables 2.3, 2.4, 2.5, and 2.6 contain the summarized findings of the significance test for differences between two means for four event windows EW1, EW2, EW3 and EW4 in the chronological order based on Table 2.1. Standard deviations are given in parentheses. Calculations of the t-statistic (1) values are based on the sample empirical results presented in Tables 2.3, 2.4, 2.5, and 2.6. The null hypothesis is rejected when |t| > t ∗, where the critical value of t-statistic at 10% significance level is equal to \( {t}^{\ast }={t}_{0,10;{n}_1+{n}_2-2} \).

Table 2.3 Summarized results of the average daily trading volume, standard deviation and the significance test for differences between two means for the first event window (EW1)
Full size table
Table 2.4 Summarized results of the average daily trading volume, standard deviation and the significance test for differences between two means for the second event window (EW2)
Full size table
Table 2.5 Summarized results of the average daily trading volume, standard deviation and the significance test for differences between two means for the third event window (EW3)
Full size table
Table 2.6 Summarized results of the average daily trading volume, standard deviation and the significance test for differences between two means for the fourth event window (EW4)
Full size table

The empirical findings reported in Tables 2.3, 2.4, 2.5, and 2.6 are not homogeneous. As would be anticipated, the investors’ activity measured by daily trading volume was significantly higher during the first event window (Table 2.3). For instance, the null hypothesis of the equality of expected values was rejected for 14 out of 18 companies and the WIG20 index in the case of the pair post-EW1/EW1 sub-periods. The results suggest that the interest rate changes were unexpected, investors were rather surprised, and generally, they reacted violently to substantial interest rates cut. Tests for differences between two means for the second (EW2) and third (EW3) event windows are presented in Tables 2.4 and 2.5, and they show that these differences are not significant in the most cases. It is not surprising as the second pre-event, event and post-event time periods included the Easter holiday in Poland. The third event was the last one, and, less formally speaking, it seems that investors have accepted interest rate cuts.

Table 2.6 presents summarized results of the significance test for differences between two means for the fourth event window (EW4), which includes the whole period of considerable declines in interest rates on Polish financial market, i.e. from 16 March 2020 to 4 June 2020. The empirical findings are interesting and worth special notice as they confirm that daily trading volume was significantly higher during this period for 14 out of 18 companies, namely KGH, PKO, PZU, PKN, DNP, PEO, LPP, PGN, CPS, SPL, PGE, LTS, CCC and ALR. This evidence is consistent with the hypothesis that declining interest rates during the pandemic period affected investors’ activity on the Polish stock market.

2.4.3 Robustness Analyses

The additional goal of the study is to conduct robustness tests and assess whether the mean results of daily trading volume during the whole sample period and the event windows (EW1, EW2, EW3 and EW4) significantly differ compared to each other. To address this issue, the t-statistic (1) is utilized, and the two-tailed hypothesis (2) is tested. Summarized findings are reported in Table 2.7, and they require some comments. The null hypothesis H0 is outweighed by the hypothesis H1 in 10 out of 19 cases for the pair WS/EW1 and in 15 out of 19 cases for the pair WS/EW4, including the WIG20 index. It indicates that the mean results of daily trading volume during these periods significantly differ compared to each other. The evidence is that the influence of the first interest rate changes was the most crucial. The further changes were less important. Moreover, taking into consideration the whole sample period (WS) and the period including all interest rate cuts (EW4), the results confirm that trading volume was significantly higher during the EW4 time window. All these observations are in accordance with expectations.

Table 2.7 Summarized results of the significance test for differences between two means of the average daily trading volume for the group of companies
Full size table

2.5 Concluding Discussion

This paper is geared towards practical implications of changes in the term structure of interest rates unknown a priori to investors. Lian et al. (2019) investigate a phenomenon referred to as ‘reaching for yield’, which means that individuals invest more in risky assets when risk-free rate is low. The authors propose to explain this phenomenon by mechanisms related to investor’s preferences and psychology rather than by conventional portfolio choice theory. They document that preferences and psychology may affect financially well-educated individuals as well as professional investors. The authors conclude that increased risk-taking may help stimulate the economy, but may also pose challenges for financial stability.

The low-interest-rate monetary policy is crucial for investors’ decisions. Daniel et al. (2018) study the impact of monetary policy on investors’ portfolio choices and asset prices. Their findings suggest that low-interest-rate policy may affect the risk premium of income-generating assets, lead to under-diversification of portfolios and cause redistributive effects across companies that differ in dividend policy. Di Maggio and Kacperczyk (2017) stress that monetary authorities worldwide introduce a policy of keeping short-term interest rates at record low level.

Taking the above into consideration, the goal of this research was to assess whether considerable declines in interest rates increased the Polish stock market activity during the pandemic period. The empirical findings based on the event window approach confirm that daily trading volume within the period of the rate cuts was significantly higher for the most analysed companies. It means that low-interest rates led to significantly higher allocations to risky assets on the Warsaw Stock Exchange. This evidence is consistent with the literature. For instance, Brzeszczyński and Kutan (2015) suggest that the Polish Central Bank actions increase investors’ activity. Gurgul and Majdosz (2005) examine whether base rate announcements released by the National Bank of Poland have a significant impact on stock returns and trading volume, but their findings are not such homogenous.

Di Maggio and Kacperczyk (2017) assess the influence of low-interest-rate policy on the U.S. money fund industry, and their empirical findings come from an event study analysis of monetary authority announcements. Therefore, one of the possible directions for further research could be to conduct a full event study to identify the impact of low-interest-rate policy during the pandemic period on asset returns on the WSE. To the best of the author’s knowledge, no such investigation has been conducted on the Polish equity market thus far.