Electricity Blackout and Its Ripple Effects: Examining Liquidity and Information Asymmetry in U.S. Financial Markets

Abstract

The massive blackout that occurred in August 2003 left over 50 million people in the northeastern and midwestern parts of the United States without electricity and caused billions of dollars in economic losses. This event highlighted the importance of sustainable and resilient energy infrastructure. Our study examines the impact of this blackout on the sustainability of financial markets by analyzing the liquidity and information asymmetry of U.S. stocks listed on major exchanges. Our results show that the blackout had a negative impact on the financial market’s liquidity, as evidenced by a significant widening of bid–ask spreads and a decrease in the market quality index. We also find an increase in information asymmetry during the blackout period, as measured by higher realized spreads. Furthermore, our study reveals that the blackout had a border impact on global financial markets and the negative effect on liquidity persisted even after two weeks.

1. Introduction

In recent years, there have been several instances of infrastructure failures that have raised concerns about their impact on sustainability issues in our society. Incidents such as the power outage during the winter storm in Texas in 2021 [1] and the blackouts in Bangladesh that affected 140 million people in 2022, as well as in Pakistan in 2023 [2], have brought renewed attention to the importance of having reliable and resilient power grids.

On 14 August 2003, at approximately 4:10 p.m. EST, a blackout affected the northeastern and midwestern parts of the United States, leaving them without electricity. More than 500 generating units from 265 power stations in the U.S. and Canada went offline, resulting in reduced generation of around 61,800 MW. In Ontario, Canada, over 10 million people experienced disruptions to their daily lives, resulting in nearly 19 million lost work hours. The manufacturing industry faced an estimated Current Account Deficit (CAD) 2.3 billion decline in deliveries, and there was a 0.7% drop in Canadian GDP in just one month due to a power outage, which is unpredictable in terms of timing [3].

The blackout that occurred in August 2003 led to significant economic impacts for various industries. Moreover, it has been observed that the blackout had a tangible effect on the values of firms within financial markets. Various attempts were made to examine the economic consequences of the blackout on the United States. In [4], the impact of the blackout on electrical firms’ security values was investigated using the event approach. Reference [5] summarizes many studies that evaluate the overall economic consequences associated with the blackout. Reference [6] describes the typical cascading processes of major blackout events and provides insights on how the economic impact of such blackout events can be evaluated.

Blackouts can result in significant financial losses for individuals, businesses, and governments. Furthermore, a blackout might have an influence on liquidity because the financial market relies on sufficient liquidity for normal operation. Liquidity is closely related to financial market efficiency [7], as the higher level of liquidity facilitates price discovery and lowers transaction costs, ultimately improving market quality. Recently, many studies have been carried out on the effect between liquidity and external shocks, such as the COVID pandemic [8,9,10], economic policy uncertainty [11,12], and the business cycle [13]. The uncertainty surrounding the future cash flows of firms may lead to investor concerns. The negative and positive effects on the firms can be attributed to increased information asymmetry and uncertainty surrounding their cash flows. Furthermore, information asymmetry and ambiguity in information quality can reduce market efficiency, lower liquidity, and increase information-based trading [14]. Market makers require higher compensation for the heightened risk and uncertainty during a blackout, which results in wider bid–ask spreads [15,16,17,18]. Additionally, the combination of asymmetric information and herding behavior among traders further amplifies the decline in market liquidity during blackout events [19]. However, no studies have explored the relationship between blackouts and their effect on liquidity and information asymmetry in financial markets.

This paper first examines the impact of the blackout on liquidity and information asymmetry within financial markets. It specifically examines the impact of the blackout that occurred in August 2003 on firms’ liquidity on the main stock exchange in U.S. The findings of this study provide valuable insights and contribute to the existing literature that has already identified exogenous shocks such as oil price or supply shock, COVID pandemics, economic policy uncertainty, and the business cycle as factors that negatively affect stock liquidity. Also, this paper examines how the blackout affected the information and information asymmetry of non-U.S. firms listed on the NYSE.

This paper is structured as follows: In Section 2, a comprehensive overview of the various metrics of liquidity and information asymmetry is provided. Section 3 presents the data employed in this study and provides a detailed description of the empirical analysis. Finally, Section 4 serves as the conclusion of the paper, where the key findings and implication are summarized.

2. Measures of Liquidity and Information Asymmetry

This section will provide various procedures for calculating the numerous liquidity- and information-based trading metrics.

The quoted spread for firm “i” at time “t” is calculated by taking the difference between the ask and bid prices in Equation (1).

Quoted Spread_i,t = (Ask_i,t − Bid_i,t)

(1)

where Ask_i,t represents firm i’s ask price at time t and Bid_i,t represents firm i’s bid price at time t. The time-weighted average quoted spread for each firm is then calculated during each day in August 2003. The quoted spread represents the discrepancy between the bid price and the ask price of a security. It is calculated as the difference between these two prices. It represents the implicit cost of trading market orders at the quoted price without any price improvement.

To calculate the effective spread of firm “i” at time “t”, which measures the cost of trading when transactions occur at prices within the bid and ask quotes, Equation (2) is used.

Effective Spread_i,t = 2D_i,t (P_i,t − M_i,t)

(2)

where D_i,t represents a binary variable which is assigned a value of one for buy orders and a value of negative one for sell orders. The value of D_i,t is estimated using the method given in reference [20]. P_i,t represents firm i’s transaction price at time t, and M_i,t represents the midpoint derived from firm i’s bid and ask quotes posted most recently. The average effective spread for each firm is calculated by considering the trade weight each day. The market quality index (MQI) [21] is used to measure the impact of ratings on liquidity, taking into account both the quoted spread and market depth. It is computed by dividing the quoted depth by the quoted spread, providing a direct assessment of liquidity, as shown in Equation (3).

$$\mathrm{Market}\hspace{0.33em}\mathrm{Quality}\hspace{0.33em}\mathrm{Inde}{\mathrm{x}}_{\mathrm{i},\mathrm{t}}=\frac{\left(1/2\right)\mathrm{Quoted}\hspace{0.33em}{\mathrm{Depth}}_{\mathrm{i},\mathrm{t}}}{\mathrm{Quoted}\hspace{0.33em}{\mathrm{Spread}}_{\mathrm{i},\mathrm{t}}}$$

(3)

It is important to note the market quality index cannot be accurately computed using the Trade and Quote (TAQ) data for NASDAQ, as the data only provide information on the size of the first inside dealer quote for NASDAQ. As a result, the market quality index is only reported for firms listed on NYSE/AMEX exchanges, where the necessary data are available to calculate the index.

The realized spread measures the profit earned by market makers through the difference between the selling and buying prices of securities. The realized spread considers the influence of trades executed by informed traders who may have more information about the security than the market maker, resulting in a price change that is not attributable to the market maker’s actions (manifested by the price impact of trades). The realized spread is formulated in Equation (4):

Realized Spread_i,t = 2D_i,t [(P_i,t − M_i,t+5)]

(4)

where i is the firm; t is the time interval; D_i,t represents the trade direction, with a value of 1 indicating a buy trade and −1 indicating a sell trade; P_i,t is the transaction price; and M_i,t+5 represents the mid-quote, which is obtained by computing the average of the bid and ask prices observed 5 min after the transaction. The trade-weighted average realized spread is then calculated for both the event and control periods for each firm during each 30 min interval. This allows for a comparison of the realized spread between the two periods and provides insight into the market maker’s revenue performance during each period.

It is important to recognize that bid–ask spreads primarily serve as indicators of market liquidity rather than direct measures of information asymmetry. However, it can be seen that information asymmetries can influence spreads to some extent [22]. Therefore, bid–ask spreads partially reflect such information asymmetries. While there are more direct measures available for assessing information asymmetries in capital markets, such as the PIN measure [23] that correlates with trading volume and the VPIN measure that provides insights into informed trading activity for a broader range of securities [24], these measures are not used due to data limitations in this paper.

3. Data and Empirical Results

3.1. Data

Information about liquidity variables for 5497 firms listed on major financial markets, including the National Association of Securities Dealers Automated Quotations (NASDAQ), American Stock Exchange (AMEX), and the New York Stock Exchange (NYSE), was used in this paper. Also, information about liquidity variables for non-U.S. stocks using the TAQ database, which is offered by the NYSE, was collected. Information on utilities and electrical manufacturing companies in the U.S. was collected based on the list [4]. This database contains extensive historical data about market prices, trading volume, bid–ask spreads, and other important liquidity measures. To ensure data accuracy, the study applies standard data filters [25] to remove errors and outliers. The following filters are used: (1) excluding quotes with a negative bid price, ask price, bid size, or ask size, as well as quotes with bid–ask spreads exceeding 4 USD(US dollar) or negative spreads; (2) the deletion of trades and quotes that are not in chronological order, occur before the market opens or after it closes, or if they exhibit a change exceeding 10% relative to the last tick; (3) the deletion of trades with negative prices or volumes; and (4) the deletion of firms that are missing in the Trading and Quote (TAQ) database. All of these filters helped to ensure the reliability of the data and confirmed that the data could be used for analysis.

The descriptive statistics for each variable utilized in this study are presented in

Table 1. Descriptive statistics.

			Percentile
Variable	Mean	Standard Deviation	Min	25	50	75	Max
Price (USD)	19.84	38.19	0.07	6.93	15.55	26.03	2502.37
Return volatility	0.0025	0.0047	0.0000	0.0005	0.0011	0.0027	0.1446
Dollar trading volume (in thousands)	12,895	81,021	24	2954	6193	11,915	5,436,000
Market value of equity (in millions)	2020	11,018	0	60	213	800,468	287,637,745
Quoted spread	0.1989	0.3608	0.0100	0.0476	0.0934	0.1975	4.0000
Effective spread	0.0989	0.1673	0.0000	0.0282	0.0506	0.1013	3.3400
Realized spread	0.0757	0.1530	−0.9101	0.0161	0.0339	0.0750	3.3000
Market quality index	1307.08	13,192.61	0.33	55.75	185.51	530.46	568,775.61

Note: The variable “Price” represents the average share price, “Return volatility” denotes the standard deviation of daily closing quote–midpoint returns, “Dollar trading volume” refers to the average daily dollar trading volume, “Market value of equity” represents the market value of equity, “Quoted spread” represents the average quoted percentage spread weighted by time, “Effective spread” represents the average effective percentage spread weighted by trade, “Market quality index” is computed as the ratio between the time-weighted average quoted depth and the time-weighted average quoted percentage spread, and “Price impact” refers to the average price impact.

. The average share price is 19.84, with a standard deviation of 38.19. Firms in this sample have an intraday return volatility of 0.0025 with a standard deviation of 0.0047. The average quoted spread is found to be 0.1989, while the average effective spread is 0.0989. Furthermore, the average dollar trading volume is 12,875 thousand.

3.2. Empirical Results

3.2.1. Change in Liquidity Surrounding the Blackout

This study examines the impact of the blackout, which occurred on 14 August 2003, on the liquidity of firms on the main stock exchange in U.S. The August 2003 blackout was used as our event day because the blackout took place on 14 August, approximately at 4:10 p.m. EST after the financial market closed. To compare liquidity, information asymmetry, and market quality changes, August 1 was selected, two weeks before the blackout, as our control day.

An overview of the changes in liquidity and information-based trading such as realized spread, market quality index, and return volatility is presented in

Table 2. Change in liquidity and information asymmetry.

	Pre-Event	Event (Blackout)	Difference
Quoted Spread	0.1696	0.2107	0.0411 (7.41 ***)
Effective Spread	0.0890	0.1070	0.0180 (6.06 ***)
Realized Spread	0.0646	0.0852	0.0206 (7.63 ***)
Market Quality Index	1747.7	854.4	−893.3 (−2.77 ***)
Return Volatility	0.0021	0.0028	0.0007 (9.49 ***)

Note: This table displays the liquidity, information asymmetry, market quality, and return volatility differences in stocks surrounding the blackout incident in 2003. The table includes metrics for quoted spread, effective spread, realized spread, market quality index, and return volatility. The date of the event was 15 August 2003, which was the largest blackout in US history. The pre-event day was 1 August. The statistical significance levels are denoted by the symbol *** which represents the 1% levels.

. Prior to the blackout on 14 August 2003, the quoted spread was 0.1696. However, following the incident, it significantly increased to 0.2107, representing a 24% increase from the pre-incident value. The observed change in the quoted spread was highly significant, as indicated by a t-value of 7.41. Similarly, the effective spread was 0.0890 in the pre-incident period, and it increased significantly to 0.1070 after the incident, indicating a decline in liquidity following the blackout. In addition to liquidity measures, Table 2 also reports the realized spread, an information asymmetry measure. The realized spread was 0.0646 during the pre-incident period. After the incident, it significantly increased to 0.0852, indicating a significant increase in information asymmetry. Furthermore, the market quality index and return volatility measures reveal a significant decline in market quality after the blackout incident. Overall, the results suggest that stocks had a significant adverse impact on liquidity and that there was an increase in information asymmetry following the incident.

3.2.2. Regression Results for Firms Listed on Major Financial Markets

A multivariate approach was further employed to investigate the blackout’s impact while controlling for variables related to trading activity. Using regression models, the study analyzed the impact of the blackout on liquidity, information asymmetry, and market quality measures, while incorporating an event dummy variable. This variable represented the day after the blackout on 15 August 2003 and was set to 1 for that day and 0 for the two weeks before the blackout. The blackout took place on 14 August at approximately 4:10 p.m. EST, after the financial market closed. Therefore, the event date was set as the first trading day after the blackout, which was 15 August.

In this paper, various control variables, including return volatility, price, trading volume (logged), and market value of equity (logged), were considered. A regression analysis was performed using pooled cross-sectional and time-series data to examine the impact of the blackout event on quoted, effective, and realized spreads and the market quality index. It was controlled for other variables and adjusted for heteroscedasticity using Huber–White estimators. By conducting a regression analysis using pooled cross-sectional and time-series data, the impact of the blackout event on liquidity, information asymmetry, and market quality measures could be statistically quantified. This approach would also help in identifying the specific effects of the blackout event on market dynamics.

The regression results for liquidity, information asymmetry, and market quality are presented in

Table 3. Regression results for the change in liquidity.

Dependent Variables	(1)	(2)	(3)	(4)
	(Quoted Spread)	(Effective Spread)	(Realized Spread)	(MQI)
Blackout dummy	0.0147 ***	0.0197 ***	0.0224 ***	−0.3220 ***
	(4.85)	(7.89)	(9.63)	(−10.06)
Price	0.0043 ***	0.0030 ***	0.0019 ***	−0.0248 ***
	(9.42)	(11.39)	(6.42)	(−5.54)
Volatility	9.5528 ***	7.1466 ***	5.3614 ***	41.4198 **
	(7.08)	(6.27)	(4.86)	(2.33)
Log(Volume)	0.0092 ***	0.0052 **	0.0053 **	0.4373 ***
	(2.59)	(2.05)	(2.09)	(9.76)
Log(Mcap)	−0.0406 ***	−0.0291 ***	−0.0224 ***	0.3284 ***
	(−22.05)	(−21.98)	(−16.95)	(18.42)
Constant	0.4465 ***	0.3297 ***	0.2446 ***	−2.0202 ***
	(11.79)	(12.94)	(9.06)	(−4.87)
Observations	10,994	10,897	10,868	5006
Adjusted R2	0.2460	0.1937	0.1259	0.3355

Note: The presented table displays the results of the ordinary least squares (OLS) regression model: ${\mathrm{Q}\mathrm{u}\mathrm{o}\mathrm{t}\mathrm{e}\mathrm{d}\mathrm{S}\mathrm{p}\mathrm{r}\mathrm{e}\mathrm{a}\mathrm{d}}_{\mathrm{i},\mathrm{t}}$ or ${\mathrm{E}\mathrm{f}\mathrm{f}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{e}\mathrm{S}\mathrm{p}\mathrm{r}\mathrm{e}\mathrm{a}\mathrm{d}}_{\mathrm{i},\mathrm{t}}$ = β0 + β1 ${\mathrm{B}\mathrm{l}\mathrm{a}\mathrm{c}\mathrm{k}\mathrm{o}\mathrm{u}\mathrm{t}\mathrm{D}\mathrm{u}\mathrm{m}\mathrm{m}\mathrm{y}}_{\mathrm{i},\mathrm{t}}$ + β2 (${\mathrm{P}\mathrm{r}\mathrm{i}\mathrm{c}\mathrm{e}}_{\mathrm{i},\mathrm{t}}$ or 1/${\mathrm{P}\mathrm{r}\mathrm{i}\mathrm{c}\mathrm{e}}_{\mathrm{i},\mathrm{t}}$) + β3 ${\mathrm{R}\mathrm{e}\mathrm{t}\mathrm{u}\mathrm{r}\mathrm{n}\mathrm{V}\mathrm{o}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{l}\mathrm{i}\mathrm{t}\mathrm{y}}_{\mathrm{i},\mathrm{t}}$ + β4 Log(${\mathrm{V}\mathrm{o}\mathrm{l}\mathrm{u}\mathrm{m}\mathrm{e}}_{\mathrm{i},\mathrm{t}}$) + β5 Log(${\mathrm{M}\mathrm{a}\mathrm{r}\mathrm{k}\mathrm{e}\mathrm{t}\mathrm{C}\mathrm{a}\mathrm{p}\mathrm{i}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{i}\mathrm{z}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}_{\mathrm{i},\mathrm{t}}$) + ${\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}$; where ${\mathrm{Q}\mathrm{u}\mathrm{o}\mathrm{t}\mathrm{e}\mathrm{d}\mathrm{S}\mathrm{p}\mathrm{r}\mathrm{e}\mathrm{a}\mathrm{d}}_{\mathrm{i},\mathrm{t}}$ represents the average quoted spread of firm i on day t, ${\mathrm{E}\mathrm{f}\mathrm{f}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{e}\mathrm{S}\mathrm{p}\mathrm{r}\mathrm{e}\mathrm{a}\mathrm{d}}_{\mathrm{i},\mathrm{t}}$ represents the trade-weighted average effective spread of firm i on day t, ${\mathrm{B}\mathrm{l}\mathrm{a}\mathrm{c}\mathrm{k}\mathrm{o}\mathrm{u}\mathrm{t}\mathrm{D}\mathrm{u}\mathrm{m}\mathrm{m}\mathrm{y}}_{\mathrm{i},\mathrm{t}}$ equals 1 for the day of the blackout and 0 two weeks prior to the blackout, ${\mathrm{P}\mathrm{r}\mathrm{i}\mathrm{c}\mathrm{e}}_{\mathrm{i},\mathrm{t}}$ represents the average stock price of firm i on day t, ${\mathrm{R}\mathrm{e}\mathrm{t}\mathrm{u}\mathrm{r}\mathrm{n}\mathrm{V}\mathrm{o}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{l}\mathrm{i}\mathrm{t}\mathrm{y}}_{\mathrm{i},\mathrm{t}}$ represents the standard deviation of intraday mid-quote returns of firm i on day t, ${\mathrm{V}\mathrm{o}\mathrm{l}\mathrm{u}\mathrm{m}\mathrm{e}}_{\mathrm{i},\mathrm{t}}$ represents the average daily dollar trading volume of firm i on day t, ${\mathrm{M}\mathrm{a}\mathrm{r}\mathrm{k}\mathrm{e}\mathrm{t}\mathrm{C}\mathrm{a}\mathrm{p}\mathrm{i}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{i}\mathrm{z}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}_{\mathrm{i},\mathrm{t}}$ represents the market value of equity of company i in day t and ${\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}$ represents the error term. The standard errors are adjusted using Huber–White estimators to address heteroscedasticity in the regression analysis. The statistical significance levels are denoted by the symbols ***, and ** which represent the 1%, and 5% levels, respectively.

. It demonstrates that the event dummy coefficients for the quoted, effective, and realized spreads are positive and statistically significant. This result indicates that the blackout had a negative effect on the liquidity of firms listed on major financial markets. In other words, the spreads between the bid and ask prices for these firms increased, resulting in higher costs for investors when buying and selling shares. This could be attributed to increased uncertainty and information asymmetry among market participants after the blackout.

To obtain a more accurate understanding of the relationship between the blackout and liquidity, a regression analysis was conducted on the Market Quality Index, incorporating both the blackout dummy and control variables. The findings suggest that the market quality of the firms listed on major financial markets was negatively affected by poor liquidity and increased information asymmetry during the blackout. Due to the unavailability of meaningful market depth data for NASDAQ market-listed firms, the calculation of the Market Quality Index was limited to NYSE/AMEX market-listed firms. The coefficient of the blackout dummy on the Market Quality Index was found to be negative and statistically significant, indicating that the blackout resulted in lower market quality, as presented in Table 3. This suggests that the blackout significantly impacted the quality of the market, as indicated by the Market Quality Index. The negative coefficients of the blackout dummy indicate that the blackout had a negative effect on market liquidity and increased information asymmetry, leading to a decline in market quality.

3.2.3. Regression Results for the Utility and Electrical Manufacturing Firms

In this section, the effects of the blackout on the liquidity and information asymmetry of the utility and electrical manufacturing firms in U.S. are investigated. The electric power industry includes electric utilities, which generate and distribute electricity to customers, and manufacturers of electric power equipment, which supply the industry with products such as transformers and switchgear. The blackout of 14 August 2003 significantly impacted both sectors.

The regression results for firms involved in power sector are presented in

Table 4. Regression results for the change in liquidity for the electric utility stocks and manufacturing stocks.

Dependent Variables	Electric Utility Firms				Manufacturing Firms
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	(Quoted Spread)	(Effective Spread)	(Realized Spread)	(MQI)	(Quoted Spread)	(Effective Spread)	(Realized Spread)	(MQI)
Blackout dummy	0.0142 *	0.0155 *	0.0163 **	−0.3630 *	0.0053	0.0065	0.0010	0.1092
	(1.87)	(1.85)	(2.11)	(−1.86)	(0.18)	(0.30)	(0.05)	(0.36)
Price	0.0015 ***	0.0012 **	0.0009 *	−0.0616 ***	0.0035 **	0.0027	0.0023	−0.0380 ***
	(3.41)	(2.19)	(1.90)	(−5.71)	(2.21)	(1.67)	(1.57)	(−3.25)
Volatility	−7.6943	−0.7971	0.7604	846.2004 ***	6.7057	7.2378	3.8940	−246.3738
	(−1.12)	(−0.09)	(0.10)	(9.20)	(0.50)	(0.86)	(0.59)	(−0.60)
Log(Volume)	−0.0013	−0.0033	−0.0033	0.5959 **	0.0054	0.0053	0.0031	0.5603
	(−0.12)	(−0.37)	(−0.40)	(2.17)	(0.18)	(0.25)	(0.16)	(0.86)
Log(Mcap)	−0.0155 ***	−0.0104 **	−0.0073 *	0.4275 ***	−0.0328 **	−0.0260	−0.0248	0.1212
	(−2.69)	(−2.56)	(−1.96)	(5.31)	(−2.59)	(−1.58)	(−1.62)	(0.43)
Constant	0.2435 ***	0.1765 ***	0.1285 **	−4.1638 **	0.3594	0.2701	0.2768	0.1154
	(3.68)	(2.88)	(2.27)	(−2.10)	(1.47)	(1.46)	(1.61)	(0.04)
Observations	66	66	66	64	35	33	33	17
Adjusted R2	0.2800	0.1314	0.0998	0.5996	0.3709	0.2214	0.1834	0.6229

Note: The contents of the note in Table 3 are the same. The statistical significance levels are denoted by the symbols ***, **, and * which represent the 1%, 5%, and 10% levels, respectively.

. It shows that the blackout dummy variable’s coefficients are positive (negative) and statistically significant at 10% for the quoted and effective spreads (MQI) and at 5% for the realized spread. Although the significance level is lower due to the smaller sample size, the results still suggest an increase in information asymmetry and a decrease in liquidity. After the blackout, these effects resulted in lower market quality for the electric utility firms. For electric equipment manufacturing firms, all coefficients of the blackout dummy on spread measures and the market quality index are positive but not statistically significant. This indicates no significant changes in liquidity, information asymmetry, or market quality surrounding the blackout. However, caution should be exercised when interpreting these results, as the lack of statistical significance could be due to the smaller sample size.

3.2.4. Regression Results for Non-U.S. Firms

Regression analyses were conducted to examine the effects of the blackout on liquidity, information asymmetry, and market quality for non-U.S. firms listed on the NYSE. Many non-U.S. firms listed on financial markets operate overseas, and disruptions in the U.S. economy can have spillover effects across global markets. Therefore, analyzing the changes in liquidity for non-U.S. firms can provide valuable insights into the impact of the blackout beyond U.S. borders.

The regression results for non-U.S. firms are presented in

Table 5. Regression results for the change in liquidity for non-U.S. firms.

Dependent Variables	(1)	(2)	(3)	(4)
	(Quoted Spread)	(Effective Spread)	(Realized Spread)	(MQI)
Blackout dummy	0.0121 **	0.0124 ***	0.0105 ***	−0.2495 ***
	(2.46)	(3.28)	(3.23)	(−3.23)
Price	0.0031 ***	0.0019 ***	0.0010 ***	−0.0426 ***
	(9.05)	(7.07)	(4.28)	(−12.99)
Volatility	5.4826 ***	4.0680 **	0.5755	−1.0805
	(3.06)	(2.02)	(0.38)	(−0.03)
Log(Volume)	−0.0144 ***	−0.0053 *	−0.0064 **	0.5687 ***
	(−3.65)	(−1.76)	(−2.45)	(7.17)
Log(Mcap)	−0.0236 ***	−0.0171 ***	−0.0114 ***	0.3319 ***
	(−13.13)	(−11.11)	(−8.84)	(13.55)
Constant	0.4561 ***	0.2823 ***	0.2238 ***	−2.9530 ***
	(13.05)	(10.02)	(8.98)	(−4.56)
Observations	760	749	746	760
Adjusted R2	0.4605	0.4095	0.2572	0.4725

Note: The contents of the note in Table 3 are the same. The statistical significance levels are denoted by the symbols ***, **, and * which represent the 1%, 5%, and 10% levels, respectively.

, indicating that all spread measures have positive and statistically significant coefficients of the blackout event dummy variable. This suggests that following the blackout, quoted and effective spreads increased substantially. In addition, the coefficient of the dummy variable on the realized spread is also positive and significant. This result indicates that after the blackout incident, information-based trading increased significantly, leading to higher information asymmetry. This poor liquidity and higher information asymmetry also led to lower market quality, as shown in the market quality regression results. These results suggest that the blackout had ripple effects beyond U.S. borders, underscoring the importance of disaster preparedness.

3.2.5. Regression Results for Liquidity Recovery

This section investigates the duration of the liquidity shock following the blackout. Two dummy variables are introduced to analyze this. The blackout took place on 14 August at approximately 4:10 p.m. EST, after the financial market closed. Therefore, the event date was set as the first trading day after the blackout, which was 15 August. The first dummy variable, called the week-1 dummy, takes the value of 1 for 15 August, the blackout event day, and 0 for the five trading days after the blackout, specifically 22 August. The second dummy variable, called the week-2 dummy, takes the value of 1 for 15 August and 0 for the ten trading days after the blackout, specifically 29 August. By introducing these dummy variables, the extent to which the liquidity shock persists over time can be examined.

The regression results on the U.S. firms’ recovery speed from the liquidity shock caused by the blackout are presented in

Table 6. Liquidity recovery after the blackout.

Dependent Variables	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	(Quoted Spread)	(Quoted Spread)	(Effective Spread)	(Effective Spread)	(Realized Spread)	(Realized Spread)	(MQI)	(MQI)
One week after the blackout	0.0184 ***		0.0213 ***		0.0216 ***		−0.2668 ***
	(6.16)		(8.64)		(9.22)		(−8.48)
Two weeks after the blackout		0.0187 ***		0.0222 ***		0.0230 ***		−0.2150 ***
		(6.28)		(8.93)		(10.17)		(−6.74)
Price	0.0044 ***	0.0040 ***	0.0035 ***	0.0030 ***	0.0022 ***	0.0020 ***	−0.0261 ***	−0.0262 ***
	(10.39)	(9.46)	(7.86)	(10.89)	(5.31)	(7.83)	(−5.70)	(−5.41)
Volatility	9.2703 ***	9.3484 ***	6.5050 ***	6.7508 ***	4.5572 ***	4.5459 ***	36.0142 **	33.5604 ***
	(6.88)	(6.38)	(5.81)	(5.71)	(4.09)	(4.10)	(2.33)	(2.66)
Log(Volume)	0.0108 ***	0.0109 ***	0.0027	0.0024	0.0028	0.0022	0.4541 ***	0.4922 ***
	(3.16)	(3.10)	(0.88)	(0.94)	(0.96)	(0.91)	(10.16)	(10.17)
Log(Mcap)	−0.0420 ***	−0.0394 ***	−0.0303 ***	−0.0274 ***	−0.0233 ***	−0.0215 ***	0.3288 ***	0.3209 ***
	(−20.78)	(−21.89)	(−16.90)	(−21.08)	(−14.24)	(−17.70)	(17.86)	(16.97)
Constant	0.4455 ***	0.4187 ***	0.3576 ***	0.3311 ***	0.2751 ***	0.2606 ***	−2.1966 ***	−2.4823 ***
	(12.31)	(11.01)	(10.19)	(12.44)	(8.21)	(10.41)	(−5.33)	(−5.54)
Observations	11,030	11,000	10,939	10,907	10,908	10,880	5041	5049
Adjusted R2	0.2730	0.2388	0.2132	0.1901	0.1297	0.1253	0.3457	0.3364

Note: The presented table displays the OLS regression results for the given model: ${\mathrm{Q}\mathrm{u}\mathrm{o}\mathrm{t}\mathrm{e}\mathrm{d}\mathrm{S}\mathrm{p}\mathrm{r}\mathrm{e}\mathrm{a}\mathrm{d}}_{\mathrm{i},\mathrm{t}}$, ${\mathrm{E}\mathrm{f}\mathrm{f}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{e}\mathrm{S}\mathrm{p}\mathrm{r}\mathrm{e}\mathrm{a}\mathrm{d}}_{\mathrm{i},\mathrm{t}}$, or ${\mathrm{M}\mathrm{a}\mathrm{r}\mathrm{k}\mathrm{e}\mathrm{t}\mathrm{Q}\mathrm{u}\mathrm{a}\mathrm{l}\mathrm{i}\mathrm{t}\mathrm{y}\mathrm{I}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{x}}_{\mathrm{i},\mathrm{t}}$ = β0 + β1 ${\mathrm{B}\mathrm{l}\mathrm{a}\mathrm{c}\mathrm{k}\mathrm{o}\mathrm{u}\mathrm{t}\mathrm{W}\mathrm{e}\mathrm{e}\mathrm{k}\mathrm{D}\mathrm{u}\mathrm{m}\mathrm{m}\mathrm{y}}_{\mathrm{i},\mathrm{t}}$ + β2 (${\mathrm{P}\mathrm{r}\mathrm{i}\mathrm{c}\mathrm{e}}_{\mathrm{i},\mathrm{t}}$ or 1/${\mathrm{P}\mathrm{r}\mathrm{i}\mathrm{c}\mathrm{e}}_{\mathrm{i},\mathrm{t}}$) + β3 ${\mathrm{R}\mathrm{e}\mathrm{t}\mathrm{u}\mathrm{r}\mathrm{n}\mathrm{V}\mathrm{o}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{l}\mathrm{i}\mathrm{t}\mathrm{y}}_{\mathrm{i},\mathrm{t}}$ + β4 Log(${\mathrm{V}\mathrm{o}\mathrm{l}\mathrm{u}\mathrm{m}\mathrm{e}}_{\mathrm{i},\mathrm{t}}$) β5 Log(${\mathrm{M}\mathrm{a}\mathrm{r}\mathrm{k}\mathrm{e}\mathrm{t}\mathrm{C}\mathrm{a}\mathrm{p}\mathrm{i}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{i}\mathrm{z}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}_{\mathrm{i},\mathrm{t}}$) + ${\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}$, where ${\mathrm{B}\mathrm{l}\mathrm{a}\mathrm{c}\mathrm{k}\mathrm{o}\mathrm{u}\mathrm{t}\mathrm{W}\mathrm{e}\mathrm{e}\mathrm{k}\mathrm{D}\mathrm{u}\mathrm{m}\mathrm{m}\mathrm{y}}_{\mathrm{i},\mathrm{t}}$ week-1 or week-2 dummy equals 1 for 15 August and 0 for the five (ten) trading days after the blackout, specifically 22 August (29 August). All variables in the above model are the same as those of the model used in Section 3.2.2, except for blackout dummy. The statistical significance levels are denoted by the symbols ***, and **, which represent the 1%, and 5% levels, respectively.

. The results indicate that the blackout’s effects on market liquidity persisted for at least two weeks after the event. The coefficients for spreads and market quality index in week 1 are positive and highly significant, indicating that one week after the incident, liquidity for U.S. firms was still deteriorating. Even after two weeks, the coefficients of the week-2 dummy for spreads and market quality are positive (negative) and significant, indicating that two weeks after the incident, liquidity and market quality remained poor. These results suggest that it took some time for the financial market to recover from the shock fully and that the blackout had a lasting impact on the financial market.

4. Conclusions

This paper examined the impact of the blackout on the liquidity and information asymmetry of firms listed in major financial markets. The results of this study show that the blackout had a negative impact on the financial market’s liquidity. It implies a significant decline in liquidity during the blackout, evidenced by a considerable widening of bid–ask spreads and a decrease in the market quality index. Also, the results revealed increased information asymmetry during the blackout, as indicated by wider realized spreads. This implies that the blackout caused uncertainty and made it more challenging for market participants to access accurate and timely information, exacerbating information asymmetry in the market.

Similarly to the results mentioned above, the results of this study on U.S. firms involved in the power sector are consistent with those of firms listed in major financial markets, although the significance level is lower due to the smaller sample size. However, the results of electric equipment manufacturing firms indicate little significant changes in liquidity, information asymmetry, and market quality surrounding the blackout.

Furthermore, the findings of this empirical study reveal that the blackout’s effects extended beyond U.S. borders, negatively impacting the liquidity of non-U.S. firms listed on the NYSE market, suggesting spillover effects. This suggests that the blackout had a broader and global impact on financial markets. Significantly, the findings of this empirical study reveal that the negative liquidity shock persisted even after two weeks, indicating that the blackout’s effects were not temporary but had a lasting impact on the financial market’s liquidity.

In this study, the findings have significant implications for various stakeholders, including policymakers, regulatory bodies, and market participants. The study highlights the critical need to invest in infrastructure, establish effective regulatory frameworks, and leverage technology to enhance collaboration and information sharing. By prioritizing these areas, stakeholders can mitigate the impact of blackouts on market liquidity and foster a more resilient and efficient market environment for all.