Impacts of investor's sentiment, uncertainty indexes, and macroeconomic factors on the dynamic efficiency of G7 stock markets

Abstract

This paper examines the impact of macroeconomic factors, microstructure factors, uncertainty indexes, the investor sentiment, and global shock factors on the dynamic efficiency of G7 stock markets. We use a non-Bayesian Generalized least squares-based time-varying model by Ito et al. (Appl Econ 46(23):2744–2754, 2014; Appl Econ 48(7):621–635, 2016) and the time-varying adjusted market efficiency method. The results show using the augmented mean group estimator and heterogeneous panel causality method a strong relationship between stock market efficiency and oil prices. In addition, all stock markets became more inefficient during COVID-19 crisis and upward trend in oil prices. Furthermore, by means of the heterogeneous panel causality test, we find evidence of unidirectional from all the considered factors, except for the consumer confidence index variable, to stock market efficiency. Moreover, we show a significant bidirectional causality between the time-varying market efficiency and both interest rates, exchange rates, market volatility, economic policy uncertainty, and the composite leading indicator. The implications of our findings for investors and policymakers are discussed.

Appendix

1.1 Stock market efficiency factors

To examine the driving forces of market efficiency, we use an onboard set of strongly related to stock markets. These variables are divided into five major groups: (1) Macroeconomic, (2) Microstructure, (3) Uncertainty, (4) Sentiment, and (5) Global shock factors.

(1) Macroeconomic Factors The macroeconomic variables included in our study are chosen based on their theoretical relationship with stock prices and their potential effect on stock market efficiency. Changes in some macroeconomic factors may affect the evolving degree of efficiency in stock markets. To test H1, we use the exchange rate (H1a), interest rate (H1b), and crude oil price (H1c) as major macroeconomic variables that can be used as reliable indicators of the time-varying degree of stock market efficiency.

• The Exchange Rate: We use the real effective exchange rate as an indicator of the value of a currency according to its trading partners. The BIS EER allows for time-varying weights and accounts for the mainland’s indirect trade with the rest of the world via Hong Kong (Klau and Fung 2006). Then, the BIS broad indices for the dollar, euro, and yen closely track the corresponding official series of the US Federal Reserve, the ECB, and the Bank of Japan, respectively, while the narrow and old indices seem to show more divergence (Klau and Fung 2006). The revised weights better depict trade flows and should increase the BIS effective exchange rate indices' utility as reliable indicators of exchange rate fluctuations and their effects.

• The Interest Rate: We use the 3-month treasury bill rate.

• The Crude Oil Price: We use the Brent crude oil price as a benchmark of crude oil prices.

(2) Microstructure Factors We expect that the degree of market efficiency increases in more liquid markets (H2a) and decreases with stock market volatility (H2b).

• Market Liquidity Instead of traditional proxies of market liquidity such as Aminhud's liquidity measure and Hui–Heubel liquidity ratio, we use the liquidity measure proposed by Danyliv et al. (2014) which has two advantages: (1) It eliminates the currency values from the calculations and instruments while exploring different international stock markets, and (2) its calculation requires only instantaneous measurement over time. The liquidity measure is given by the following formula:

  $$LIX_{t} = log_{10} \left( {\frac{{Vol_{t} P_{Close,t} }}{{P_{Hight,t} - P_{Low,t} }}} \right)$$

  (17)

  where $Vol_{t}$: Transaction volume at time t; $P_{Closet}$: the closing price at time t; $P_{Hight,t}$: the highest price; $P_{Low,t}$ the lowest price.

• Market volatility: The variance of the market index is modeled as a function of a constant term, information on fluctuations in the previous period error term, and information on fluctuations in the previous period volatilities, captured by a GARCH (1,1) model proposed by Engel (1982) and Bollerslev (1986) as follows:

  $$\sigma_{t}^{2} = \omega + \alpha \varepsilon_{t - 1}^{2} + \beta \sigma_{t - 1}^{2}$$

  (18)

  The GARCH (1,1) model is the appropriate representation of conditional variance (Husain and Uppal (1999)) and is consistent with the typical stylized facts noticed in financial data (Leptokurtic financial returns; Volatility clustering, and leverage effects tendency).

(3) Uncertainty factors We use two mains proxies that reflect the economic uncertainty in the G7 countries: the EPU (H3a) and CLI (H3b)

• The EPU index: The economic policy uncertainty index

• The CLI: The composite leading indicator of the OECD

(3) Sentiment Factors Despite a variety of investor sentiment proxies suggested in the literature, many of these sentiment measures are short-lived and unreliable (He 2012). Therefore, they may not be fully reflected in closing prices. In our study, we choose two different proxies for investor sentiment: the sentiment endurance index (H4a) and the consumer confidence index (H4b). It's essential to select reliable proxies for investor sentiment, and in this case, the sentiment endurance index and the consumer confidence index provide valuable insights into market sentiment. By utilizing these proxies, we aim to capture the impact of investor sentiment on stock market efficiency more accurately.

• Sentiment Endurance index (SE): The sentiment endurance index proposed by He (2012) is calculated as follows:

  $$SE_{t} = \left( {\frac{{P_{Close,t} - P_{Low,t} }}{{P_{Hight,t} - P_{Low,t} }}} \right) - 0.5$$

  (19)

  where positive $SE_{t}$ reflects a bullish sentiment toward the closing price at time t, while a negative $SE_{t}$ reflects a bearish sentiment.

• Consumer confidence index (CCI) The CCI is an alternative sentiment proxy measured in the majority of countries and it represents the only constant method for generating a sentiment proxy that allows for cross-country comparisons.

(5) Global shock factors We include two main global shocks that has been shown an effect on the worldwide economies and on financial markets: the financial crisis of 2008 (H5a) and the COVID-19 pandemic (H5b)

• Global financial crisis A dummy variable that indicates the 2007–2008 financial crisis. It takes 1 during the period from 01/02/2007 to 01/12/2008, 0 otherwise.

• COVID-19 pandemic A dummy variable that indicates the COVID-19 period. It takes 1 during the period from 01/03/2020 to 01/09/2020,¹ 0 otherwise (Table 17).

Table 17 The appropriate lag order selection based on the information criteria