Stock Market Volatility, Monetary Stability and Risk-taking Behaviors

Transcription

1 Stock Market Volatility, Monetary Stability and Risk-taking Behaviors Jo-Chi Huang ANR: Academic supervisor: Prof. Dr. Olivier De Jonghe Second reader: Prof. Dr. Rik Frehen Thesis submitted in partial fulfillment of the requirements for the degree: Master of Science in Finance Tilburg School of Economics and Management, Tilburg University, The Netherlands Date: August 22, 2017

2 i

3 Stock Market Volatility, Monetary Stability and Risk-taking Behaviors Jo-Chi Huang * August 31, 2017 * I am indebted to Prof. Dr. Olivier De Jonghe for his valuable suggestions, infinite patience and guidance. I would also like to thank instructors and classmates at Tilburg University. Finally, I thank my family and friends for their love and support. ii

4 Contents 1. Introduction Background and Related Literature Methodology Model Construction Variable Construction Risk-taking Stock Market Volatility Monetary Stability Control Variables Alternative Measure of Explanatory and Dependent Variables Results Stock Market Volatility and Future Risk-taking Monetary Stability and Future Risk-taking Monetary Stability and Stock Market Volatility Deflation Conclusion References Appendixes iii

5 List of Tables Table 1: Equity Indices that are Used to Calculate the Stock Volatility Gap Table 2: Descriptive Statistics and Correlation Matrix-I Table 3: Descriptive Statistics and Correlation Matrix-II Table 4: Sampling Countries and Years Table 5: Stock Volatility Gap and Risk-taking Table 6: Monetary Stability (measured by dummy variable of low inflation) and Risk-taking Table 7: Monetary Stability (measured by the interaction term of low and stable inflation dummies) and Risk-taking Table 8: Monetary stability (measured by dummy variable of low inflation) and Stock Volatility Gap Table 9: Monetary Stability (measured by interaction term of low and stable inflation dummies) and Stock Volatility Gap iv

6 v

7 Abstract This thesis investigates the relationship between stock market volatility, monetary stability and risk-taking behaviors. The regression model of risk-taking on stock market volatility is generated to test the link between those two variables. In addition, I construct the dummy variable of monetary stability and estimate the impact of monetary stability on risk-taking and stock market volatility. It has been found that stock market volatility leads to an increase in risk-taking, supporting Minsky s financial instability hypothesis which suggests that the environment with low volatility leads to higher risk-taking. While the new environment hypothesis suggests that monetary stability may make market participants more optimistic about the future economy, there is no evidence for a positive relationship between monetary stability and risk-taking. The empirical results show that monetary stability leads to lower stock volatility, which could be explained by higher reliability of returns predictions, lower resource misallocations and market participants lesser sensitivity to risks. Finally, deflation is associated with lower risk-taking. This could be explained by lower companies earnings and higher credit rationing after deflation. 1

8 1. Introduction Minsky s financial instability hypothesis suggests that low risk environment leads to regulatory relaxation and rising optimism, which may cause an increase in risk-taking behaviors. An increase in risk-taking behaviors may transform the financial system from stable to unstable, and therefore increases the chance of a financial crisis. Jon Danielsson et al. (2016) show that low volatility in stock markets leads to greater risk-taking and a higher possibility of a financial crisis, supporting Minsky s financial instability hypothesis. Similar to Minsky s financial instability hypothesis, the new environment 1 hypothesis suggests that monetary stability could also make market participants more optimistic about the future economy and thus less sensitive to, or concerned about, the downsides of various economic problems which could lead to financial vulnerability. Before the new environment hypothesis, the conventional view was that monetary stability promoted financial stability by making the forecast of investment returns more reliable and reducing monetary misperceptions. More reliable predictions regarding investment returns and decreased monetary misperception would reduce resource misallocations, and thus lead to higher levels of financial stability. While the conventional view and the new environment hypothesis differ on whether monetary stability is beneficial or detrimental to financial stability, these two theories are not mutually exclusive as monetary stability can both increase the reliability of return predictions and cause rising optimism which may make market participants less sensitive to various risks. Volatility in the equity market can reflect both the volatility of fundamental and market participants ability to perceive risks. This means that a change in stock market volatility can reflect the effects of monetary stability according to the perspectives of the conventional view and the new environment hypothesis. This thesis investigates the relationship between stock market volatility, monetary stability and risk-taking behaviors. Monetary stability is generally considered equivalent to price stability, and 1 The terms conventional view and new environment hypothesis are derived from a speech by Otmar Issing (2003) to the Conference on Monetary Stability, Financial Stability and the Business Cycle in

9 is defined by low inflation or low and stable inflation. For example, the European Central Bank (ECB) defines price stability as the growth rate in the Harmonised Index of Consumer Prices (HICP) of below 2%. However, the range of inflation rates that is used to identify price stability can vary, depending on the different price indexes that are used to calculate the inflation. Also, the range of inflation rates for price stability may be different for different countries. For example, for developed countries, inflation at a range between 2% and 3% is generally considered as price stability. However, for emerging markets, the range should be higher, as emerging countries generally have higher inflations. To avoid the problem of what counts as a benchmark for low inflation for different indexes or countries, I use the inflation trend calculated by the Hodrick- Prescott filter as a benchmark of monetary stability. In this paper, I measure monetary stability using two proxies: a low inflation dummy and the interaction term of low and stable inflation dummies. Low inflation is the dummy variable that takes the value of one when inflation is in the range of zero and its trend. The interaction term of low and stable inflation dummies is the multiple of the low inflation dummy and the stable inflation dummy. The stable inflation dummy is the dummy variable that takes the value of one if the standard deviation of the monthly inflation is lower than its trend. This thesis consists of three empirical models. The first empirical model tests the relation between stock market volatility and risk-taking behaviors. I replicate the regression model generated by Jon Danielsson, Marcela Valenzuela and Ilknur Zer (2016), and use the stock volatility gap as the explanatory variable. Jon Danielsson et al. (2016) test the effect of low stock volatility on the probability of a financial crisis and the level of risk-taking. They use high and low stock volatilities as explanatory variables. To calculate high and low volatility, they take the standard deviation of stock monthly returns and use the Hodrick-Prescott filter to calculate its trend. Then they calculate the volatility gap, by subtracting the stock volatility trend from stock volatility and dividing these deviations into two variables: high and low volatility variables. The variable of high volatility takes the value of the volatility gap if the volatility gap is positive, and takes the value of 0 if the volatility gap is negative. The variable of low volatility takes the value of the absolute value of the volatility gap if the volatility gap is negative, and takes the value of 0 if the volatility gap is positive. However, since my intention is to measure the level of stock market volatility, I use the value of the stock volatility gap without separating it into high and low volatility gaps. 3

10 Following Jon Danielsson et al. (2016), investors risk-taking behavior is measured by the creditto-gdp gap. The credit in the variable of credit-to-gdp gap refers to domestic credit to the private sector. An increase in domestic credit to the private sector implies an increase in risk-taking behaviors in both the financial and private non-financial sectors. It is worth stressing that the increase in risk-taking, measured by the credit-to-gdp gap, is not equivalent to financial instability. Instead, the credit-to-gdp gap is only one of the early indicators of financial instability (Alessi & Detken, 2014; Borio & Lowe, 2002). There are several other financial variables, such as the house price gap and the debt service ratio, which can be considered as early indicators of financial instability (Alessi & Detken, 2014). The second model is constructed to investigate the overall relationship between monetary stability and future risk-taking. The new environment hypothesis suggests that low and stable inflation may lead to rising optimism, and rising optimism would make market participants less sensitive to, or concerned about, the source or signs of financial instability. If low and stable inflation can make market participants more optimistic about the future economy, it is likely that risk-taking behaviors will also increase. To investigate the relationship between monetary stability and risk-taking, I generate the same regression from the first empirical model and use monetary stability as the explanatory variable. Monetary stability is measured using two proxies: the low inflation dummy and the interaction term of low and stable dummies. Risk-taking is measured using the credit-to- GDP gap. The third model is constructed to test the relationship between monetary stability and stock market volatility. I regress stock market volatility, measured by the stock volatility gap, 2 on monetary stability, measured by two proxies: a low inflation dummy and the interaction term of low and stable inflation dummies, 3 and control lags of the dependent variable, the natural logarithm of GDP per capita, institutional quality, and the share of export in the GDP. 2 The dependent variable in the third model, the stock volatility gap, is the same as the explanatory variable of the first model. 3 The explanatory variable of the third model, monetary stability, is the same as the explanatory variable of the second model. 4

11 In addition, following Jon Danielsson et al. (2016), the sum of the absolute value of monthly stock returns is employed as an alternative proxy of stock market volatility, and credit-to-gdp growth is used as an alternative proxy of risk-taking. These two alternative proxies are used in every model to check the robustness of the results. The empirical results are presented as follows: First, the decrease in stock market volatility leads to the increase in risk-raking. Stock market volatility, measured using the stock volatility gap, is positively correlated with risk-taking. This supports Minsky s financial instability hypothesis, which suggests that a long period of low volatility would lead to rising optimism, financial deregulation and therefore an increase in risktaking behaviors. Second, there is no evidence for a positive relationship between monetary stability and risk-taking. In the second empirical model, some regressions results show that monetary stability does not have significant effects on risk-taking, and some regression results show that monetary stability leads to lower risk-taking. However, there is no regression result showing that monetary stability leads to higher risk-taking. Third, monetary stability leads to lower stock market volatility. The negative relationship between monetary stability and stock market volatility could be explained by both the conventional view and the new environment hypothesis as low stock volatility may indicate both less change in fundamentals and market participants lesser sensitivity to risks. The conventional view holds that monetary stability can make return forecasts more reliable. In addition, price stability can make it easier for companies to distinguish the change in relative price from the change in aggregate price, and therefore can reduce resource misallocations. The increase in the reliability of return forecasts and the decrease in resource misallocations lead to a smaller change in equity fundamental. In the perspective of the new environment hypothesis, monetary stability would make market participants more optimistic about the future economy and less sensitive to risks, and market participants lesser ability to perceive risk would be reflected in lower level of uncertainty and therefore lower volatility in the stock market. 5

12 Fourth, deflation is correlated with lower risk-taking. Deflation leads to lower aggregate demand and thus real returns from investment. The decrease in real returns would reduce companies incentives to invest. In addition, the decrease in earnings would lead to the increase in default rate and thus credit rationing, leading to less lending and investments. The rest of this thesis is structured as follows. Chapter 2 presents the research background and literature review. Chapter 3 explains the construction of the model and variables. Chapter 4 presents the results of the regressions and the interpretation of the empirical results. Chapter 5 is the conclusion. Chapter 7 contains the references. The summary statistics and tables of regression results are presented in the Appendixes. 6

13 2. Background and Related Literature Minsky (1992) analyzes the relationship between financial stability and financial crises. He proposes the financial instability hypothesis, suggesting that a long period of stability leads to rising optimism, regulatory relaxation and therefore a higher possibility of financial crisis. After the financial crisis of 2008, there has been an increase in the literature examining the relationship between volatility and crises. Stability is usually associated with low volatility, and the existing literature has extensively investigated the relationship between volatility and future economic states. For example, Fabio Fornari and Antonio Male (2009) find that various financial variables, such as stock volatility and term spread, can predict industrial production growth and business cycles, and John Y. Campbell et al. (2001) find that stock volatility can predict GDP growth. Noha Emara (2012) investigates the link between price stability and economic growth. She finds that while the level of inflation does not have a significant effect on GDP growth, inflation volatility does. Her research shows that higher inflation volatility can reduce economic growth. While most of the literature on volatility focuses on the relationship between volatility and economic growth, Jon Danielsson et al. (2016) investigate the relationship between low stock volatility and financial crises. They found that low stock volatility has a significant effect on banking and stock market crises, supporting Minsky s financial instability hypothesis. Brunnermeier and Sannikov (2014) use an economic model to illustrate that low aggregate risk can actually lead to higher leverage, and higher leverage leads to higher systemic risk. Minsky s financial instability hypothesis suggests that a long period of stability leads to rising optimism, financial deregulation and therefore an increase in risk-taking. Rising optimism is normally associated with higher risk-taking, especially during a period of regulatory relaxation. Many authors (Hellmann et al., 2000; Shijaku & Kalluci, 2013) suggest that financial deregulation also leads to higher risk-taking behaviors by financial institutions. Financial liberalization intensifies the competition. More intense competition increases the costs of funds and reduces lending profit margins. A decline in profit margins would therefore increase financial institutions incentive to take risks. From the perspective of Minsky s financial instability hypothesis, borrowers can be divided into three types: hedge borrowers, speculative borrowers and Ponzi borrowers. Hedge borrowers are 7

14 borrowers who can pay their interest and principal debt from their cash-flows. Speculative borrowers are borrowers who are able to use their cash-flows to cover their interest but need to refinance their principal loans. Ponzi borrowers are borrowers who do not have sufficient cashflow to repay the principal debt or the interest, and need to bet on an increase in the price of their assets to refinance their loans. Minsky (1992) suggests that a long period of stability may encourage risk-taking. Higher risk-taking indicates an increase in Ponzi borrowers, and an increase in Ponzi borrowers means that the financial system becomes more vulnerable to changes in asset prices. When the market price of underlying assets drops drastically, Ponzi borrowers have to go bankrupt, as the decrease in asset prices makes it more difficult for them to refinance their loans. This process is called debt deflation (Fisher, 1933). Debt deflation not only illustrates Minsky s financial instability hypothesis, but also highlights the importance of monetary stability for achieving sustained financial stability. The importance in maintaining monetary stability has been emphasized by the conventional view (Schwartz, 1988; Bordo et al., 1998). The conventional view suggests that high and volatile inflation, usually considered as monetary instability, may make the forecast of future returns less reliable. Also, a high inflation rate would make it more difficult for individuals to distinguish a shift in relative prices from a shift in aggregate prices, and therefore may increase resource misallocations. Moreover, the decrease in the predictability of future returns and the increase in resource misallocations would make borrowers more likely to default. Disinflation, the transition from high inflation to low inflation, is the example that is most frequently used to support the perspective of the conventional view. Anna Schwartz (1988) argues that unexpected disinflation may cause or promote financial instability, as the sudden decrease in the inflation rate increases the real burden of liabilities. High and sustained inflation creates an expectation of high inflation, and the expectation of a high inflation rate leads to a higher interest rate. A sudden decrease in the inflation rate increases the real value of interest and principal debt, making borrowers more likely to default. However, some authors hold that monetary stability also has some drawbacks. The new environment hypothesis suggests that the world has gradually entered into a new environment that is shaped by the following three factors: a low and stable inflation rate, financial liberalization, and an aggregate supply shock (Borio et al., 2003). Over the past few decades, financial 8

15 deregulation has taken place in various countries. Financial liberalization increases credit availability and lending. Higher lending means more investment projects, and higher investments may increase productivity and profits. An increase in a firm s profits may increase price levels. Higher price levels may lead to credit expansion, and encourage more investment. Higher investment may further increase a firm s profitability. However, this boom could overstretch. If the government cannot identify the threat in the early phase of the bubble, the financial instability may lead to catastrophic crises. Meanwhile, the development of technology has caused an increase in aggregate supply. However, the increase in aggregate supply may reduce inflation, making it more difficult to spot the bubble. In addition, the positive development in productivity may lead to an increase in corporate profits and companies having stronger balance sheets, providing a reason to believe in sustainable higher growth. In this situation, while the success in maintaining low and stable inflation may reduce resource misallocations, it may also create a false perception of security and optimism in the future economy, as low and stable inflation is typically associated with a smaller chance of bubbles. The sense of security and optimism may cause market participants to ignore or care less about various problems that might lead to financial instability. Some research papers have used historical data to analyze the relationship between price stability and financial stability. For example, Claudio Borio et al. (2002) find that financial instability can develop in an environment of disinflation or low inflation. Bordo et al. (2000) find that aggregate price shock contributes to financial instability, measured in terms of the financial condition index they have constructed. Existing literature holds different viewpoints in the relationship between inflation and bank lending. For example, on one hand, Boris (2001) found that the increase in price level leads to higher bank lending. He suggests that the increase in property price would lead to higher net worth of borrowers and therefore higher credit availability. Higher lending leads to more investments, and more investments could further boost property price. In addition, Bordo (2000) suggests that, apart from the increase in the net worth of borrowers, inflation would lead to the expectation of high inflation in the future, reducing the risk premium of lending. Therefore, inflation could lead to the increase in lending. On the other hand, Boyd et al. (2001) found that in economies with low to moderate inflations, the increase in inflation is associated with the increase in volatility of equity returns and the decrease in lending activities. He argues that the increase in inflation would lead 9

16 to the decrease in real returns. This would decrease the incentive to lend and increase the incentive to borrow. Moreover, lower real returns may draw borrowers with low quality into credit market, making credit rationing more extensive. At the end, market friction would have negative impacts on credit and stock markets (Huybens & Smith, 1999; Azariadas & Smith, 1996). Boyd et al. (2001) argues that when inflation increases, the marginal effect of inflation on credit and stock market reduces greatly, so the effect of the change in inflation rate on the performance of financial sector is more significant in economies with low inflation. This thesis replicates the empirical model designed by Jon Danielsson et al. (2016) and uses the stock volatility gap as an explanatory variable to test the relationship between equity market volatility and risk-taking. The empirical results are consistent with the study by Jon Danielsson et al. (2016), and support Minsky s financial instability hypothesis. The main contribution of this thesis is the investigation into the effects of monetary stability on stock market volatility. 10

17 3. Methodology 3.1. Model Construction This thesis consists of three empirical models. The first empirical model investigates the relationship between stock market volatility and risk-taking, the second empirical model tests the relationship between monetary stability and risk-taking, and the third model investigates the relationship between monetary stability and stock market volatility. To investigate the relationship between financial stability and future risk-taking, the first model is based on the paper Learning from History: Volatility and Financial Crises by Jon Danielsson, Marcela Valenzuela and Ilknur Zer (2016). Jon Danielsson, et al. (2016) focus on the relationship between periodic high and low volatility gaps with regard to future risk-taking, and the probability of financial crises. In the first model of this thesis, the explanatory variable is stock market volatility, measured using the stock volatility gap, and the dependent variable is risk-taking, measured using the credit-to-gdp gap. The control variables are lags of dependent variable, inflation, the natural logarithm of GDP per capita, the change in general government-debt-to-gdp ratio, institutional quality and the real interest rate. In addition, regional and decade fixed effects are controlled in the regression. All independent variables are constructed based on the average values over the previous five years. Following Jon Danielsson et al. (2016), standard errors are robust and two-way clustered at year and country levels. The regression below represents the first model. CR_GAP i,t = + αs_vol i,t 1 to t 5 + θcr_gap i.t 5 to t 1 + INF i.t 5 to t 1 + lngdp i.t 5 to t 1 + PD/GDP i,t 5 to t 1 + βpolcomp i,t 5 to t 1 + φrintr i,t 5 to t 1 + υ i + τ t (t 1)to(t 5) = i=1 C (t i) C = δ, CR_GAP, INF, lngdp, PD/GDP, POLCOMP, RINTR The dependent variable, risk-taking, measured using the credit-to-gdp gap (CR_GAP i,t ), is the difference between the credit-to-gdp ratio and its trend, that is estimated by the Hodrick-Prescott filter with a smoothing parameter of 100. The explanatory variable, financial stability, measured using the stock volatility gap (S_vol), is the difference between stock volatility and its trend, 11

18 estimated by the Hodrick-Prescott filter with a smoothing parameter of INF is the annual inflation rate, lngdp is the natural logarithm of the GDP per capita, PD/GDP is the change in the public-debt-to-gdp ratio, POLCOMP is the institutional quality, and RINTR is the real interest rate, while υ i and τ t denote the region and decade fixed effects. Minsky s financial instability hypothesis suggests that low volatility can lead to an increase in risktaking. Therefore, it is expected that α, the coefficient of the stock volatility gap, will be negative. The second empirical model tests the relationship between monetary stability and future risktaking behaviors. The structure of the second regression model is the same as the first one, except with regard to the explanatory variable. The explanatory variable used in this model is monetary stability, measured using a low inflation dummy or the interaction term of low and stable inflation dummies. The dependent variable is risk-taking, which is the same as for the first model. Since the explanatory variable is measured using low inflation, the control variable of the inflation rate is dropped. The regression below represents the second model. CR_GAP i,t = + αinf low i,t 5 to t 1 + θcr_gap i.t 5 to t 1 + lngdp i.t 5 to t 1 + PD/ GDP i,t 5 to t 1 + βpolcomp i,t 5 to t 1 + φrintr i,t 5 to t 1 + υ i + τ t The new environment hypothesis holds that monetary stability may lead to rising optimism. If monetary stability can lead to rising optimism, risk-taking behaviors may also increase. If monetary stability can increase risk-taking, it is expected that α, the coefficient of monetary stability, will be positive. The third empirical model examines the relationship between monetary stability and stock market volatility. I regress stock market volatility on monetary stability. In the third model, the explanatory variable is monetary stability, and is measured using a low inflation dummy or the interaction term of low and stable inflation dummies. The dependent variable is stock market volatility, and is measured using the stock volatility gap. The institutional quality is controlled, as the institutional quality may have an effect on economic stability (Acemoglu et al., 2003). In addition, following Acemoglu et al. (2003), I control the natural logarithm of GDP per capita, as income level may have an impact on the volatility of the business cycle (Kraay and Ventura, 2001). Last, following Daron Acemoglu et al. (2002), the export share in the GDP is also controlled, as 12

19 the level of volatility might be related to the exposure to shocks. The regression below represents the third model. S_vol i,t = + αinf low i,t 5 to t 1 + θs_vol i,t 1 to t lngdp i.t 5 to t 1 + βpolcomp i,t 5 to t 1 + Export/GDP i,t 5 to t 1 + +υ i + τ t From the perspective of the conventional view, monetary stability leads to greater reliability of returns prediction. In addition, monetary stability can reduce the difficulty for companies to distinguish relative change in price from aggregate change in price, so resource misallocations should be reduced. The increase in the reliability of returns prediction and the decrease in resource misallocations should lead to more stable fundamentals. Lower volatility in fundamentals should be reflected in lower stock volatility. Therefore, it is expected that α, the coefficient of monetary stability, will be negative. From the perspective of the new environment hypothesis, monetary stability may make market participants feel more secured and optimistic about the future economy, and therefore market participants may become less sensitive to, and cautious about, risks. If market participants have a worse ability to perceive risk, the level of uncertainty and volatility should be lower in the equity market. Therefore, if monetary stability would make market participants less sensitive to risk, it is expected that α, the coefficient of monetary stability, will be negative Variable Construction Risk-taking Following Jon Danielsson et al. (2016), risk-taking is measured using the credit-to-gdp gap, which is the credit-to-gdp deviation from its trend. The credit-to-gdp gap is generally considered one of the early indicators of a banking crisis. This thesis uses the credit-to-gdp gap as a measure of risk-taking, as a greater credit-to-gdp gap reflects higher corporate leverage and bank lending compared to the trend. The credit-to-gdp gap is measured using the following process. First, calculate the credit-to-gdp by dividing the total domestic credit to the private non-financial sector by the GDP. The private non-financial sector includes both households and non-financial companies. The data are mainly obtained from the Bank for International Settlements (BIS). The data for domestic-credit-to-gdp 13

20 from BIS are provided quarterly. In order to get the annual value, the average of the four quarterly values within each year is taken. However, the data from BIS do not cover the entire sample, as there are seven countries that are not covered by BIS. Therefore, data from Datastream were used to compensate for the countries that are not covered by the BIS database. Second, deconstruct the credit-to-gdp ratio into two parts: the trend and the deviation from the trend. The trend is estimated by the Hodrick-Prescott filter, with a smoothing parameter of 100. The credit-to-gdp gap is the difference between the credit-to-gdp ratio and the credit-to-gdp trend that is calculated by the Hodrick-Prescott filter Stock Market Volatility I measure the stock market volatility using the stock volatility gap. Following Jon Danielsson et al. (2016), stock volatility is deconstructed into its trend and the deviation from the trend. The difference between the volatility and the volatility trend is the volatility gap. Jon Danielsson et al. (2016) separate volatility gaps into high and low volatility gaps. In their paper, the variable of a high volatility means a positive deviation of volatility from the volatility trend, and the variable of a low volatility means a negative deviation of volatility from the volatility trend. They take the absolute values of both positive and negative volatility deviations from the trend to get the high and low volatility gaps. They use high and low volatility gaps as the main explanatory variables for their regressions. However, since my goal is to measure the level of stock market volatility, I use the stock volatility gap to measure stock market volatility, without separating the volatility gap into variables of high and low volatility gaps. The stock volatility gap is constructed as follows. First, get the monthly data of stock returns for the equity indices in the different countries from the Datastream database. 4 4 Datastream uses the following formula to calculate the daily return for the equity index and accumulates daily returns to the monthly level with the assumption that dividends are reinvested on the same index: RI t = RI t 1 PI t (1 + PI t 1 DY ) 100 n Where: RI t = return index on day t 14

21 Table 1 in Appendixes displays the 40 countries included in the sample, the equity indices that were used to calculate the stock return, and the code to transform the daily returns into monthly percentage returns for every country. Second, adjust the monthly return for inflation in each month to get the real monthly return for every country in the sample. The monthly inflation rate is measured using monthly CPI growth rate and is obtained from Datastream. The inflation-adjusted real return is calculated using the following formula: Inflation Adjusted Return = (1+Return) (1+inflation rate) 1 Third, the inflation-adjusted monthly returns are 1% winsorized for every country, meaning 1% of the highest and lowest observations are replaced by the upper bound and the lower bound in each country. The upper bound and lower bound for each county is the highest and lowest values within the 99% range of the observations in a country across sampling years. Fourth, calculate the annual standard deviation of monthly returns from January to December for each country. Fifth, use the Hodrick-Prescott filter to get the trend of the annual standard deviations of monthly stock returns. Generally speaking, the smoothing parameter of the Hodrick-Prescott filter for RI t 1 = return index on previous day PI t = price index on day t PI t 1 = price index on previous day DY = dividend yield of the price index n = number of days in financial year The data is obtained on a monthly basis, which is the value of the daily return accumulated to the end of every month. However, this monthly return data is not a percentage value. Therefore, by entering RI,PCH#(X,1M) in the datatype column when retrieving the data from Datastream, the monthly return is then transformed into a percentage value. 15

22 annual data should be 100. However, since the annual stock volatility is very volatile, I use 5000 as the smoothing parameter of the Hodrick-Prescott filter to calculate the trend of stock volatility. 5 Fifth, calculate the volatility gap by subtracting the volatility trend from the volatility Monetary Stability Monetary stability is generally defined as low inflation or low and stable inflation. For example, ECB identifies an HICP inflation rate below 2% as price stability. However, the definition of low and stable inflation depends on the type of price index. Also, the range of inflation rate for price stability may vary for different countries. For most developed countries, a sustainable inflation rate is between 2% and 3%, but the range might be different for emerging economies. To eliminate these problems, I use an inflation trend calculated by the Hodrick-Prescott filter with a smoothing parameter of 5000 as the reference level for monetary stability. In this thesis, monetary stability is measured by two proxies: a low inflation dummy and the interaction term of the low inflation dummy and the stable inflation dummy. Following the general definition of monetary stability, I use low inflation to measure monetary stability. Low inflation is the dummy variable that takes the value of one when the inflation rate is at the range between zero and its trend. I also create a dummy variable of high inflation, which is defined by an inflation rate that is larger than zero and its trend. Apart from high and low inflations, it is found that some countries in the sample have deflation in some years. The theoretical view about the effect of deflation on the economy is different from the effect of high and low inflations. Based on theoretical view, deflation may cause an increase in saving and reduce aggregate demand, and the decrease in aggregate demand may trigger a decrease in production and an increase in the unemployment rate. The decrease in production and the surge of the unemployment rate may increase the default rate in both the household and private sectors. Therefore, the variable of deflation is created as a separate group, apart from high and low inflations. 5 Jon Danielsson et al. (2016) also use 5000 as the smoothing parameter of the Hodrick-Prescott filter to calculate the volatility trend in their research. 16

23 In addition, I use the interaction term of low and stable inflation dummies to measure monetary stability. The interaction term of low and stable inflation dummies is generated using the multiple of the low inflation dummy and the stable inflation dummy. The stable inflation dummy is the dummy variable that takes the value of one when inflation volatility is larger than its trend, calculated by the Hodrick-Prescott filter with a smoothing parameter of Inflation volatility is measured by the annual standard deviation of monthly inflation rates Control Variables The dependent variable of the first and second empirical models is the credit-to-gdp gap, the proxy of risk-taking. Following Jon Danielsson et al. (2016), besides region and decade fixed effects, I control lags of the dependent variable, inflation, the natural logarithm of GDP per capita, changes in the general government-debt-to-gdp, institutional quality and the real interest rate. Inflation might have a negative effect on financial institutions credit to the private sector, since it reduces real returns (Boyd et al., 2001). Therefore, inflation is controlled in the regression, and the data are obtained from Datastream. Since it has been shown that there is a positive correlation between economic growth and credit expansion in many research papers (Angeles, 2015; Garcia- Escribano et al., 2015), I include the variable of GDP per capita, obtained from the World Bank Institution, in the regression. In addition, it has been found that government debt has a crowdingout effect on bank lending to the private sector (De Bonis & Stacchini, 2013), so I control general government debt, obtained from the International Monetary Fund (IMF), in the model. Since institutional quality can have an impact on economic stability (Acemoglu et al., 2003), it is controlled in the regression. The data for institutional quality are measured by the POLCOMP variable from the database of the Polity IV Project. The value of POLCOMP reflects the level of institutionalization and the extent of government s control on political participation. Low POLCOMP indicates unstable political situations and lack of opportunity for alternative parties or leadership roles to obtain power. Real interest rate levels may have an effect on credit (Hofmann, 2001). Therefore, the real interest rate, obtained from the Organisation for Economic Co-operation and Development (OECD) and World Bank databases, is controlled in the model as well. 17

24 The dependent variable of the third empirical model is the annual standard deviation gap of stock monthly returns, the proxy of stock market volatility. In the third model, I control the natural logarithm of GDP per capita, institutional quality and the export share of GDP, which reflects the exposure to shocks. The variables of the natural logarithm of GDP per capita and institutional quality have been introduced in the previous paragraph. The export share of GDP represents the exposure to shocks, 6 and the data are obtained from the World Bank database Alternative Measure of Explanatory and Dependent Variables Inspired by Fabio Fornari and Antonio Male (2009), as well as Jon Danielsson et al. (2016), the sum of the absolute value of stock returns is used as the alternative measure of volatility in the stock market. The gap in the alternative measure of stock market volatility is calculated by subtracting the stock volatility trend, calculated by the Hodrick-Prescott filter with a smoothing parameter of 5000, from the stock volatility. In addition, following Jon Danielsson et al. (2016), the credit-to-gdp change is employed as an alternative measure of risk-taking. 6 Daron Acemoglu et al. (2002) control the export share of GDP in their regression of volatility of GDP growth, which is very similar to my dependent variable, the stock volatility gap. 18

25 4. Results 4.1. Stock Market Volatility and Future Risk-taking To investigate the relationship between stock market volatility and risk-taking behaviors, I replicate the regression model created by Jon Danielsson et al. (2016). All variables, except the explanatory variable, are the same as for the model generated by Jon Danielsson et al. (2016). In my empirical model, the explanatory variable is stock market volatility and is measured using the stock volatility gap. The results are presented in Table 5. In Table 5, the negative coefficient of the stock volatility gap indicates that low volatility in stock market may encourage risk-taking behaviors. The negative relationship between stock market volatility and risk-taking could be explained by Minsky s financial instability hypothesis. According to Minsky s financial instability hypothesis, a long period of stability leads to rising optimism and financial deregulation. Rising optimism and financial liberalization would further lead to an increase in risk-taking behaviors, and this relationship has been supported by the empirical results presented in Table Monetary Stability and Future Risk-taking The new environment hypothesis suggests that monetary stability would create a sense of security and make market participants feel optimistic about the future economy. If monetary stability can really generate optimism and a sense of security, it may also increase risk-taking behaviors, just as financial stability does. To investigate the relationship between monetary stability and risktaking, I employ the same regression as for the first empirical model, with the low inflation dummy as explanatory variable. The results are presented in Table 6. Table 6 illustrates that the statistical significance of the effect of low inflation on risk-taking is very sensitive to the proxy of risk-taking. Columns II and V in Table 6 represent the regression of risk-taking on low inflation. It is found that low inflation has a significant effect on risk-taking only when risk-taking is measured using the credit-to-gdp gap. Columns I and III in Table 7 represent the regression of risk-taking on stable inflation. It is found that stable inflation has a significant effect on risk-taking only when risk-taking is measured using the credit-to-gdp gap. Columns II and IV in Table 7 show the regression of risk-taking on the interaction term of low inflation and stable inflation dummies. Again, it is found that low and stable inflation has significant effects only when risk-taking is 19

26 measured using the credit-to-gdp gap. Moreover, variables of monetary stability with statistical significance are all negatively correlated with risk-taking, meaning low inflation may reduce risktaking. Based on the regression results of Tables 6 and 7, there is no evidence that monetary stability would encourage risk-taking behaviors. It is worth stressing that these results do not indicate that the new environment hypothesis is wrong. The new environment hypothesis suggests that monetary stability makes market participants feel secure and confident about the future economy, so they are less sensitive to, or less concerned about, the risk of bubble. In other words, while the new environment hypothesis suggests that monetary stability may make the financial market more vulnerable to stress, it does not suggest that monetary stability would directly cause an increase in risk-taking behaviors. Hence, while these regression results indicate that monetary stability does not lead to an increase in risk-taking behaviors, it does not mean that the new environment hypothesis is disputed. There are two possible reasons for no evidence regarding to the relation between monetary stability and risktaking. First, monetary stability does not make market participants more optimistic to future economy, and therefore risk-taking behaviors do not increase. The second possible reason is that while monetary stability makes people more optimistic to the economy, the optimism can only cause market participants lesser sensitivity to stress, rather than the increase in risk-taking Monetary Stability and Stock Market Volatility To investigate the relationship between monetary stability and stock market volatility, I run the regression of the stock volatility gap on monetary stability. In Columns II and V of Table 8, low inflation is statistically significant and negatively correlated with the future stock market volatility gap, meaning that monetary stability may lead to lower stock market volatility. In Columns I and IV of Table 8, high inflation is statistically significant and is positively correlated with the future stock volatility gap, meaning that high inflation may lead to higher stock market volatility. Columns I and III in Table 9 show that stable inflation does not have a significant effect on stock market volatility. However, Columns II and IV show that low and stable inflation leads to lower stock market volatility. These results show that monetary stability leads to lower stock market volatility. 20

27 To sum up, based on the empirical results in Tables 8 and 9, it can be concluded that monetary stability leads to lower stock market volatility. The negative relationship between monetary stability and stock market volatility can be resulted from the following two channels. First, according to the conventional view, low monetary stability makes the forecast of returns more reliable. In addition, since inflation may render companies unable to distinguish the shift in the relative price level from the shift in the aggregate price level, lower inflation can decrease resource misallocations. The increase in the reliability of return predictions and the decrease in resource misallocations could further lead to less change in equity fundamentals. Stability in companies fundamentals leads to lower volatility in equity market. Second, apart from the change in fundamentals, shift in stock prices may also be driven by the reaction of market participants to new events or shocks. In this sense, low volatility can also indicate market participants lesser sensitivity to risks. According to the new environment hypothesis, monetary stability can make market participants more optimistic to the future economy and less sensitive to or concerned about the build-up of financial imbalances. The rising optimism and the decrease in the sensitivity to risks may be reflected on a relatively lower level of uncertainty, and therefore lower volatility, in the equity market. It is worth stressing that while the empirical results show that there is a negative relationship between monetary stability and stock volatility gap, it does not mean that the new environment hypothesis is supported as rising optimism, motivated by monetary stability, may have both direct and indirect effects on the volatility of equity market. The direct effect of rising optimism refers to market participants lesser sensitivity to risks, which can be reflected on the lower level of uncertainty and therefore lower volatility in the equity market. The indirect effect of rising optimism refers to higher possibility of financial instability due to the increase in financial vulnerability resulted from rising optimism. Once a financial crisis occurs, volatility in the equity market will increase extensively. The direct and indirect effects of rising optimism on the volatility of equity market imply that rising optimism, resulted from monetary stability, could have both positive and negative effects on the volatility of equity market. Since this thesis cannot distinguish the direct and indirect effects, these empirical results should not be considered as an assessment of the new environment hypothesis. 21

28 4.4. Deflation Columns III and VI of Table 6 indicate that deflation leads to lower risk-taking. There are various reasons for the negative relationship between deflation and risk-taking. Deflation increases the purchasing power of a currency. The increase in purchasing power of a currency encourages savings and discourages investments. In addition, while deflation reduces the costs of materials and equipment, it also reduces product prices and aggregate demand. The decline in product prices and aggregate demand reduces companies profits, and lower companies earnings would further lead to higher default rate and credit rationing. Also, deflation creates the expectation of deflation in the future, so banks would require higher risk premium in lending (Bordo & Wheelock, 1998). Therefore, deflation can lead to the decrease in investments, lending and risk-taking. 22

29 5. Conclusion In this thesis, I investigate the effects of monetary stability on risk-taking and stock market volatility in 40 countries for up to 30 years. Minsky s financial instability hypothesis suggests that a long period of low volatility leads to rising optimism and regulatory relaxation, which may lead to an increase in risk-taking. Higher risk-taking may further increase the possibility of financial crises. Empirical results in this thesis show that lower stock market volatility leads to higher risktaking. This is consistent with the empirical results of the research by Jon Danielsson et al. 7 (2016), and supports Minsky s financial instability hypothesis. Deflation is found to be related to lower risk-taking behaviors. It could be that deflation leads to lower aggregate demand and thus lower real returns. The decrease in real returns would reduce companies incentive to invest. Also, the decrease in company earnings, resulted from the decrease in aggregated demand, can increase the default rate, leading to credit rationing. The decrease in credit availability can lead to lower investment and risk-taking behaviors. Therefore, deflation leads to the decrease in risk-taking. It is also found that monetary stability cannot increase risk-taking. However, it does not mean the new environment hypothesis is wrong as the new environment hypothesis does not suggest that monetary stability would lead to higher risk-taking. It only suggests that monetary stability would make market participants more optimistic to the future economy and believe that the economic growth is permanent. This optimism would make people less sensitive to the risk of bubble. Therefore, the empirical results do not dispute the new environment hypothesis. The statistical insignificance of the effect of monetary stability on risk-taking can be explained by two possible scenarios. The first scenario is that monetary stability does not trigger rising optimism. The second scenario is that monetary stability can lead to rising optimism, but the rising optimism makes market participants less sensitive to risks but does not increase risk-taking. If the second scenario is true, it can be inferred that while both Minsky s financial instability and the new environment hypothesis suggest that rising optimism can lead to higher level of financial vulnerability, the psychological effects generated by stock market stability and monetary stability are different. 7 Jon Danielsson et al. (2016) find that low stock market volatility has significant effects on financial crises and risk-taking. 23