Economic tides and merger waves: insights from a long-run perspective

Abstract

Purpose

This study aims to examine the long-run relationship between macroeconomic and financial conditions and the aggregate number of mergers and acquisitions (M&As) in the USA, drawing on data spanning from 1928 to 2019.

Design/methodology/approach

The study estimated a Vector Error Correction Model (VECM) encompassing four variables: the aggregate number of M&As, industrial production, the rates on three-month U.S. treasury bills and the closing price of the Dow Jones Industrial Average.

Findings

There exists a long-run relationship among the four variables. An increase in industrial production is associated with a fall in M&A transactions, reflecting a tendency for M&A waves to start during economic downturns. Similarly, contractionary monetary policy, which often happens during good economic and financial times, leads to a decline in M&A activity. When the equilibrium among the four variables is disrupted, the aggregate number of M&As, along with financial conditions, works to restore the equilibrium.

Originality/value

To the best of the authors’ knowledge, this is the first study to examine the long-run relationship between macroeconomic and financial conditions using data spanning nearly a century.

Citation

Emiru, T. and Weisblatt, S. (2025), "Economic tides and merger waves: insights from a long-run perspective", Studies in Economics and Finance, Vol. 42 No. 1, pp. 75-88. https://doi.org/10.1108/SEF-09-2023-0566

1. Introduction

Mergers and acquisitions (M&As) have long captivated the attention of scholars, practitioners and policymakers, as they represent pivotal corporate strategies for accessing new markets, enhancing market share and optimizing operational efficiency. The intricacies of M&A transactions are multifaceted, influenced by a myriad of factors that range from cultural disparities and political uncertainties to social connections and the pursuit of new markets.

Ahern et al. (2015) shed light on the detrimental repercussions of cultural differences in cross-border mergers, demonstrating how such disparities can erect communication barriers and erode employee morale, ultimately culminating in the failure of the merger. Conversely, Ishii and Xuan (2014) have explored the impact of social ties on M&A outcomes, revealing how close connections between acquirer and target firms can yield suboptimal returns and financial performance. Furthermore, Bonaime et al. (2018) discuss the profound influence of policy uncertainty on M&A decision-making, where hesitation prevails amidst the unpredictable terrain of regulatory fluctuations. Tang et al. (2024) discuss the effect of chief executive officer (CEO) education on participation in merger waves and how the timing of such participation could be impacted by cash flows.

Earlier research has probed the essence of M&A activity, examining whether these transactions follow a deterministic economic path or an unpredictable trajectory. Shughart and Tollison (1984) found the presence of a first-order autoregressive process with high degree of persistence, meaning that past merger activity significantly predicts the likelihood of future mergers. On the contrary, Melicher et al. (1983) ventured into business and capital market conditions, discovering that while business conditions exhibit a modest explanatory power for M&A activity, capital market conditions wield a strong correlation with aggregate merger activity. Golbe and White (1993) took a different path, unraveling the rhythmic patterns embedded in merger activities and suggesting the existence of merger waves.

In recent years, Didier et al. (2019) have explored the network effects on cross-border M&A decisions, emphasizing the role of interconnectedness among firms in shaping the propensity for such transactions. These investigations illuminate the multifaceted nature of M&As and underscore their profound impact not only on corporate performance but also on research and development endeavors, as exemplified by the findings of Phillips and Zhdanov (2013). Cong and Zhou (2024) explore sequential mergers under incomplete information where mergers could benefit leader and follower firms but reduce the welfare of consumers.

This study contributes to the existing literature in two primary dimensions. First, it scrutinizes the historical trajectory of merger activities, determining whether they adhere to a random walk process or reveal long-run relationships with macroeconomic variables. By using a rich historical data set spanning from 1928 to 2019, we elucidate the fundamental nature of mergers in the corporate landscape. Second, the study explores the long-run interplay between neoclassical explanations of mergers and the clustering of merger activities in some periods.

Our data encompass the annual aggregate number of M&As, industrial production, the rates on three-month U.S. treasury bills and the closing price of the Dow Jones Industrial Average. The empirical analysis estimates a Vector Error Correction Model (VECM) using these variables. Stationarity tests reveal that all four variables are nonstationary I(1), and cointegration analysis subsequently validates the presence of a long-run relationship. Contingent upon the presence of a cointegrating relationship, the VECM framework enables us to investigate the long-run associations among the variables under examination. Standard regression analysis applied to differenced variables emphasizes short-term relationships and eliminates the long-term relationships that this study seeks to study. Our results indicate that industrial production, monetary policy and financial conditions, as reflected in the Dow Jones index, all exert their influence, albeit in different directions, on the number of mergers. Merger waves often initiate during economic downturns, and a contractionary monetary policy is correlated with a reduced number of M&A transactions. While our results suggest that neoclassical determinants of merger waves are still significant, they do not rule out behavioral explanations of merger waves.

The remainder of this paper is organized as follows. In Section 2, we provide a review of the pertinent literature. Section 3 is dedicated to discussing the data and the econometric model employed in this study. Moving on to Section 4, we present and analyze and discuss the outcomes of stationarity tests, lag-order selection, cointegration tests and the VECM estimation. Finally, Section 5 discusses the main results and Section 6 concludes.

2. Related literature

M&As have become a common corporate strategy to gain access to new markets, increase market share and improve operational efficiency. However, M&A transactions are complex and can be influenced by various factors. The literature on M&A identifies many factors that affect M&As.

The earlier literature on M&As revolved around the question whether merger activity is explained by economic and financial conditions, or it is a random walk process. For instance, Shughart and Tollison (1984) used an econometric model to study the merger process and found that M&A transactions are characterized by a first-order autoregressive process, meaning that the likelihood of a merger occurring is influenced by past merger activity. Melicher et al. (1983), on the other hand, hypothesized that merger activity is explained by business conditions (captured by industrial production) and capital market conditions (captured by stock prices). However, their empirical analysis found that business conditions weakly explain merger activities while capital market conditions are highly correlated with aggregate merger activity. Golbe and White (1993) fit a set of sine function on the merger data and find that the implied peaks and troughs are close to the actual peaks and troughs in the data. Thus, they conclude that merger activities occur in waves.

Research on M&As have also focused on identifying the reasons behind merger waves. These factors can be broadly classified into four categories. First, a significant portion of merger activity tends to coincide with or revolve around periods of notable economic events. The expansion and contraction of aggregate demand often correlate with the rapid growth of firms, which contrasts with the concept of “organic growth” (Gaughan, 2010). Second, regulatory changes can also serve as catalysts for waves of merger activity. A prime illustration of regulatory influence on merger activities is the Glass-Steagall Act and its repeal in 1999 which impacted M&A trends in the banking sector in the USA. Studies have also indicated the endogeneity of deregulation where deregulation is preceded by poor performance of firms and these firms with substantial excess capacity often become merger partners following deregulation (Ovtchinnikov, 2013). Third, another factor that has been identified as a driver of M&A waves is technological change. While technology shocks alone may not suffice to trigger a wave, they can coincide with other factors, such as macro-level liquidity, to propel industry-level waves (Harford, 2005, 2024). Finally, “defensive acquisitions” and “positioning acquisitions” have been mentioned as reasons for merger waves (Gorton et al., 2009). These occur when a target firm proactively takes measures to acquire the firm that is targeting it or positions itself as an attractive target.

The more recent literature on M&As examines the effect of specific factors on aggregate and cross-border merger activity. These factors include cultural differences, social ties, policy uncertainty, network ties, entry into a new market and so on. Ahern et al. (2015) find that cultural differences can have a negative impact on cross-border mergers. Cultural differences can create communication barriers and affect employee morale, leading to decreased organizational performance and ultimately resulting in the failure of the merger. Similarly, Ishii and Xuan (2014) studied the impact of social ties on mergers and found that social connections between the acquirer and target firm can negatively affect the returns to the acquirer and the combined entity. Bonaime et al. (2018) found that uncertainty about future regulations and government policies can lead to hesitation among firms contemplating mergers, as they may be unsure about the long-term impact of the regulatory environment on the merged entity. In another study on M&A, Ahern and Harford (2014) investigated the effect of inter-industry relations on the timing and incidence of merger waves. They found that firms in related industries are more likely to merge during a merger wave than firms in unrelated industries. Didier et al. (2019) studied the network effects of cross-border M&A decisions and found that the decision to engage in cross-border M&A is influenced by network ties between firms. Firms are more likely to engage in cross-border M&A if they have connections to other firms that have engaged in such transactions.

Not surprisingly, M&As influence firm performance during and after the takeover. Renneboog and Vansteenkiste (2019) investigated the effect of takeovers on firm performance and found that takeovers have a positive effect on long-term firm performance, although the effect may not be immediate. M&As also have an effect on firm R&D. Phillips and Zhdanov (2013) studied the effect of active acquisitions on firm R&D and found that firms that engage in active acquisitions tend to increase their R&D spending. This may be because M&As provides access to new technology, knowledge and resources, which can enhance a firm’s innovative capabilities.

Ohrn and Seegert (2019) discuss how tax preferences for capital gains in contrast to dividends distort the merger decisions of investors, incentivizing them to make lower quality acquisitions. Tang et al. (2024) discuss the effect of CEO education on participation in merger waves and how the timing of such participation could be impacted by cash flows. Cong and Zhou (2024) explore sequential mergers under incomplete information where mergers could benefit leader and follower firms but reduce the welfare of consumers. Liao et al. (2021) study the effect of media sentiment on M&As and find that firms are more likely to become an acquirer and pay a higher premium when media sentiment is high.

A segment of the M&A literature delves into domestic acquisitions, aiming to reveal the underlying factors driving such transactions. Kinateder et al. (2017) conducted a study on domestic acquisitions in BRICS countries, using a manually compiled data set to assess whether such acquisitions generate value for both acquirers and targets, while also identifying the factors influencing this value creation. Their findings indicate a significantly higher average return on announcement for target companies, although it is statistically insignificant for acquirer companies. Additionally, they observe a positive association between returns and gross domestic product growth rates. In related literature reviews, Yaghoubi et al. (2016a) and Yaghoubi et al. (2016b) extensively explore M&A waves, their causes and consequences. Notably, our study does not differentiate between domestic and cross-border acquisitions.

To conclude, M&A transactions are complex and subject to diverse influences. Cultural disparities, policy uncertainty, social connections, autoregressive processes, inter-industry relationships, cross-border M&A decisions, market structure, firm performance, diversification and firm R&D have all significant effect in shaping M&A activities. This study seeks to enhance the current literature by leveraging a comprehensive historical data set to identify long-term relationships among the number of mergers, aggregate economic activity and financial market conditions. Furthermore, it scrutinizes and discusses the long-run relationships between macroeconomic variables and aggregate merger activity that it uncovers. 

3. Data and econometric model

3.1 Data

This study uses time series data for the USA spanning from 1928 to 2019 to conduct an a cointegration analysis [1]. In particular, we estimate a four-variable VECM. The variables used in the analysis are the gross annual number of M&As, industrial production, the rates on three-months U.S. treasury bills and the closing price of the Dow Jones Industrial Average.

M&As data were sourced from Thomson Financial and the Institute for Mergers, Acquisitions and Alliances (IMAA). The historical data set spans the number of M&As, reaching as far back as circa 1890, as illustrated in Figure 1. The data identify seven distinct waves of merger activity: 1896–1904, 1922–1934, 1963–1974, 1985–1991, 1994–2001, 2003–2009, and from 2012 to the present.

The data for M&As were obtained from Thomson Financial and the IMAA. The historical data include the number of M&As dating back to circa 1890, as depicted in Figure 1. The data identify seven waves of merger activity: 1897–1904, 1916–1929, 1965–1969, 1984–1989, 1992–2001, 2003–2009 and 2012 to now.

Yaghoubi et al. (2016a, 2016b), Gaughan (2010) and Cho and Chung (2022) meticulously document the details of the first six waves, providing insights into the industries involved, their characteristics, causes and consequences. The initial wave (1897–1904), following the depression of 1883, notably impacted mining and manufacturing industries, with a prevalence of horizontal consolidations increasing monopolistic power. The second wave (1922–1934) saw industry consolidation, forming oligopolistic market structures because of antitrust provisions enacted by Congress (the Clayton Act of 1914). The third wave (1963–1974), labeled the conglomerate merger period, coincided with stringent antitrust enforcement against horizontal mergers, often using stock-financed acquisitions to leverage lower taxes. The fourth wave (1985–1991) featured hostile mergers and larger targets. The fifth wave (1994–2001) was characterized by mergers driven, in part, by robust U.S. aggregate demand, financed predominantly with equity. The sixth wave (2003–2009) was marked by a low-interest-rate policy by the Federal Reserve, witnessing the rise of private equity and leveraged, speculative acquisitions. The seventh, ongoing wave (2012-now), initiated with extensive government stimulus packages post-Great Recession. Noteworthy for its occurrence in tech-enabling industries and technology’s pivotal role, this wave also reflects a regulatory retreat. Hsu et al. (2017) associate this last merger wave with merger activity in the oil and gas industry in the USA.

Summing up, these waves were driven by changes in technology, changes in market structure, diversification, hostile takeovers and the emergence of new financing methods, issues pertaining to the governance of U.S. corporations, equity overvaluations and the rise of cross-border acquisitions, among others.

The rates on three-month Treasury bills were obtained from the data archives by Aswath Damodaran of New York University [2]. As depicted in Figure 2, Panel C, these rates reached their peak during the Volcker disinflation of the early 1980s, rising from their lows during the Second World War, and later dropping back to the zero lower bound during the Great Recession. Industrial production and the Dow Jones index in Panels B and D were obtained from Federal Reserve Economic Data and macrotrends.net, respectively. Both series exhibit familiar upward trends with setbacks occurring during periods of economic downturn.

We want to emphasize that our study investigates the neoclassical explanations for merger waves, as indicated by the choice of variables. Neoclassical theories attribute merger waves to economic, technological and regulatory shocks. While additional behavioral factors, including stock market misvaluations, managerial decisions regarding merger timing, overconfidence and hubris and agency theories, would potentially provide more insights into merger waves, they were not incorporated into our analysis because of data limitations. The econometric methodology we use, suitable for identifying long-term relationships, requires long-run historical data, which proves challenging to obtain for behavioral variables. Furthermore, it may also prove difficult to identify behavioral variables such as the extent of acquirer and target misvaluations in aggregate, long-term merger data and in a corporate environment where corporate governance is likely to be strong to allow them to thrive and drive the clustering of aggregate mergers.

3.2 Econometric model

The empirical analysis in this study estimates a VECM with four variables. This analysis uses annual historical data on the number of mergers, U.S. industrial production, rates on the three-month treasury bill and the Dow Jones index. This model serves to identify the short- and long-run relationships among the number of mergers, economic activity and financial conditions. By doing so, it elucidates whether merger activity in the USA is driven by business activity and/or financial conditions as in Melicher et al. (1983) or is better explained by a random walk process as in Shughart and Tollison (1984). This analysis will use data from 1934 to 2019, with the time period selection boiling down to the availability of data.

The VAR process with p lags is specified as follows:

Yt=c+A1Yt−1+…+ApYt−p+ϵt

Where ϵ_t is a vector of independently, identically distributed normal errors with variance-covariance matrix Σ. Y is a column vector of k I(1) variables which are stationary in first differences. c is a k × 1 vector of coefficients, and the A_i are p × k coefficient matrices. In our case, k = 4 and p = 2, as we will show shortly.

We rewrite the VAR-p process as VECM as follows:

ΔYt=c+ΠYt−1+∑i=1p−1ΓiΔYt−i+ϵt

Where Π=∑j=1pAj−I4 and Γi=−∑j=i+1pAj.

If there exist cointegrating relationships, we can write Π matrix as Π = αβ′. We can also incorporate trends into the cointegrating relationship, and the equation as follows:

ΔYt=α(β′Yt−1+μ+ρt)+∑i=1p−1ΓiΔYt−1+γ+τt+ϵt

In our estimation, we allow for a trend in the cointegrating relationship (ρ ≠ 0) but rule out trends in the differenced equations (τ = 0). Having τ ≠ 0 implies the variables have quadratic trends which we do not observe in Figure 2 as the data are in natural logarithms.

4. Estimation and results

4.1 Stationarity tests

In Table 1, we report the values of the test statistic for Augmented Dickey Fuller test for the first 10 lags. We also report the MacKinnon approximate p-values in parenthesis.

A higher absolute value for the test statistic relative to the critical value indicates the rejection of the null hypothesis of a random walk without drift, suggesting that the process is stationary. It is important to note that in Dickey–Fuller regressions, where we do not explicitly incorporate a drift parameter, if the null hypothesis of a unit root cannot be rejected, the reason could be attributed to a deterministic trend, a stochastic trend or both.

All four variables are nonstationary in levels. To account for this nonstationarity, a time trend is necessary if the data exhibit only a deterministic trend. If the data also possess stochastic trends, differencing is required to eliminate the random walk component. In either case, we will conduct a cointegration analysis to assess whether these three variables exhibit long-run relationships. Moreover, vector error correction analysis requires the data to be difference stationary. As can be seen from Table 1, all variables are stationary in first differences.

4.2 Lag order selection

Before estimating the VECM model, it is essential to determine the optimal lag length for the short-run component of the model using information criteria. The likelihood ratio test, Final Prediction Error, Akaike’s Information Criteria and the Hannan and Quinn Information Criteria all suggest that the optimal lag length is two, whereas Schwarz’s Bayesian Information Criteria suggests a lag length of one. As only one of these tests indicates a lag order of one and omitting the second lag could introduce biases related to omitted variables, we opt for a lag order of two.

4.3 Testing for cointegration

Next, we will assess whether there exists a long-run relationship among merger activity, industrial production, monetary policy (as measured by the rates on three-month T-bills) and macroeconomic financial conditions (as measured by the annual closing price of the Dow Jones Industrial Average).

Visual inspection of the time-series nature of the four variables in Figure 2 indicates that the number of mergers, industrial production and the Dow Jones closing prices all exhibit trends, whereas the rate on three-month T-bills potentially contains a stochastic trend. Additionally, it is worth noting that all of these variables are stationary in their first differences, as shown in the results in Table 1.

Before estimating the VECM model, we performed a test for the cointegration rank. Johansen’s test for cointegration reveals one cointegrating relationship, with a trace test statistic value of 31.6363. Given that the model involves four variables and only one cointegrating relationship, it is appropriate to use a VECM model that captures both short- and long-run dynamics.

It is also important to note that we constrained the trend to the cointegrating relationship and the variables in levels only. The differenced equations are assumed not to have a deterministic trend; in other words, we imposed the assumption that τ = 0. This decision is made because an unrestricted trend would imply a quadratic trend in the variables in levels, and that does not seem to be the case based on Figure 2.

4.4 Vector Error Correction Model estimation results

The results from the estimation of the model with one cointegrating relationship and two lags are presented below. It is important to note that our data vector is represented as ΔY_t = [Δ log(m&a) ΔIP Δi₃ Δlog(DJ)]′. Consequently, the vectors of cointegrating parameters (β), speeds of adjustment (α) and constant terms (γ) are all 4 × 1 vectors, while the coefficient matrices of the short-run dynamics Γ₁ and Γ₂ are 4 × 4 matrices. Below, we will focus on reporting the cointegrating vector and the speed of adjustment parameters, as they carry interesting economic interpretations.

The cointegrating vector, presented in Table 2, is given by β′ = [1 0.178 49.24 4.64] with a constant term of 29.33. The coefficient on industrial production is significant at 5% level while the coefficients on the three-month T-bill and the Dow Jones closing price are significant at 1% level. The time trend is also significant at 1% level.

This implies a long-run relationship of:

log⁡(m&a)=29.33−0.178 IP−49.24 i3−4.658log⁡(D&J).

The results are intuitive. An increase of one point in the industrial production index, our proxy for overall economic production, is associated with a decrease in merger transactions of approximately 17.8%. It seems that merger waves start during periods when aggregate production is lower. Likewise, a 1 percentage-point increase in the three-month T-bill rate is linked to a decline in the number of mergers by 0.49 percentage points. This finding complements the industrial production result, as it suggests that monetary policy tends to be contractionary during economic expansion, leading to a reduced financial market liquidity, and thereby the number of M&A transactions. Furthermore, improvements in financial conditions, as indicated by increases in the Dow Jones index, are associated with a smaller number of mergers. A 1 percentage-point increase in the Dow Jones index results in a decrease in the number of mergers by 4.68%.

The speeds of adjustment toward the long-run relationship are represented by the vector α. The estimation results are reported below:

α=[−0.028 −0.053 −0.0007 −0.038]

Before we provide interpretations for these coefficients, let’s make two observations. First, the dot product of α and β contains elements that are all negative. This confirms that the system is not explosive. If the cointegrating relationship is above equilibrium in the previous period, one or more of the variables will work to bring the system back to equilibrium.

Second, it’s worth noting that only the first and last elements of the vector are statistically significant (both exhibiting p-values close to zero). Therefore, while all four coefficients demonstrate the expected signs, it’s primarily the first and fourth variables, namely, the number of mergers and the Dow Jones index, that play a significant role in returning the system to equilibrium. When the cointegrating equation is out of equilibrium, the number of mergers decreases by approximately 2.8% and the Dow Jones index experiences a decline of about 3.8% to bring it back to the long-run equilibrium.

4.5 Diagnostic tests

Up to this point, the analysis has imposed a cointegrating rank of one based on Johansen’s trace test statistic. As the determination of the cointegrating rank is a crucial step in VECM analysis, we conduct a diagnostic test to assess the appropriateness of this determination. Furthermore, although the trace test suggested a rank of one, we also need to verify the stationarity of the identified cointegrating relationship.

We achieve this objective by calculating the eigenvalues of the companion matrix and assessing their proximity to unity. As the VECM analysis involves four variables, and we have established a cointegration rank of one, the companion matrix should exhibit three eigenvalues that are equal to one. The fourth eigenvalue should be relatively small and not close to one in absolute value. We visualize these eigenvalues in Figure 3. As depicted in Figure 3, three of the eigenvalues are exactly one, whereas the fourth highest eigenvalue is 0.537, which is significantly smaller than one. Thus, our choice of a cointegration rank of one and the stationarity of the cointegrating relationship are confirmed.

5. Discussion

The findings of this study, which reveal an inclination toward increased M&A activity during economic downturns, as well as a correlation between contractionary monetary policy and improved financial conditions with a decline in M&A activity, provide insights into the dynamics of corporate transactions in different economic scenarios. To better understand and contextualize these results, it is essential to link these results with existing literature and theoretical frameworks that shed light on the relationship between economic conditions and M&A waves.

One key aspect to consider is the observed phenomenon that returns tend to be higher than normal during recessions, as evidenced by studies such as Wann and Lamb (2016). This elevated return environment during economic downturns may be a crucial factor motivating the initiation of merger waves in economic downturns. Moreover, behavioral finance literature underscores that investor decisions are influenced not only by actual returns but also by perceptions of returns (for instance, see Cyert and March, 1963; Hoffmann et al., 2015; Khan et al., 2017). This implies that “good news in bad times is worth more than good news in good times” (Wann and Lamb, 2016). Similarly, good news about the state of the economy in periods of expansion tend to be perceived negatively while good news during a recession tend to be viewed more favorably McQueen and Roley (1993). The asymmetry in actual and perceived returns across the business cycle could offer a primary explanation for the observed surge in M&A activity during economic downturns.

Furthermore, the analysis of monetary policy’s role in M&A trends provides additional insight into understanding these dynamics. In normal economic times, monetary policy tends to be more contractionary, potentially leading to a shortage of liquidity in financial markets. This scarcity of liquidity could significantly impact the feasibility and cost-effectiveness of large-scale acquisitions. As larger acquisitions require substantial financial resources, the decreased liquidity resulting from contractionary monetary policy may impose constraints on companies looking to undertake such transactions. The effect of market liquidity on acquisitions may have been more evident in the sixth and seventh merger waves when the Federal Reserve maintained low interest rates. This finding is consistent with the findings of Harford (2005) which highlights the importance of market liquidity in propagating industry-level shocks to merger waves.

As we discussed in Section 3.1, the primary focus of this study has been to examine the neoclassical explanations of merger waves and assess their robustness through econometric analyses using long-term historical data. In contrast to the behavioral explanations, which rely on factors such as agency issues, hubris and misvaluations, neoclassical explanations rely on economic fundamentals to justify the clustering of rational merger activity. For instance, Harford (2005) argues that mergers are primarily driven by neoclassical explanations and their clustering in some periods is the result of the existence of sufficient market liquidity. Ahern and Harford (2014) find that industry-level shocks propagate to merger waves because of vertical and horizontal networks among firms. Many other studies also find support for neoclassical explanations of merger waves (Gort, 1969; Mitchell and Mulherin, 1996; Jovanovic and Rousseau, 2002).

Other studies focus on behavioral explanations to merger waves. Shleifer and Vishny (2003) suggest that markets can be irrational and that bidders may exploit an overvalued equity to acquire targets before the misvaluation is corrected. Similar situations may occur even in rational environments where information frictions exist regarding whether the overvaluations stem from market wide or firm-specific factors (Rhodes‐Kropf and Viswanathan, 2004). Rhodes–Kropf et al. (2005) explain how merger waves could manifest when market values are high relative to true values. Behavioral explanations of merger waves have also been supported by many other studies (Komlenovic et al., 2011; Gugler et al., 2012; Duchin and Schmidt, 2013; Malmendier and Tate, 2015).

It is important to note that the two explanations for merger waves are not mutually exclusive. The significance of each explanation may vary from one wave to another. Several studies supporting behavioral explanations, such as Rhodes–Kropf et al. (2005), Ang and Cheng (2006), Dong et al. (2006), Hsu et al. (2017), Sonenshine (2020) and Andriuškevičius and Štreimikienė (2021), are also consistent with neoclassical explanations. Behavioral explanations may dominate in some merger waves and neoclassical in others. Likewise, our study does not rule out the behavioral explanations; rather, it contributes additional evidence highlighting the importance of neoclassical explanations in driving merger waves.

6. Conclusion

This study investigated whether the aggregate number of mergers exhibits long-run relationships with macroeconomic variables. To achieve this objective, we utilized a comprehensive data set spanning nearly a century and estimated a VECM that includes the aggregate number of mergers, industrial production, three-month Treasury bill rates and the Dow Jones Industrial Average. The trace test statistic indicated the presence of one long-run relationship among these variables. In this long-run relationship, industrial production, serving as a proxy for overall economic activity, demonstrates an inverse relationship with the number of mergers, highlighting the tendency for merger waves to begin during economic downturns. Similarly, monetary policy, as measured by the rates on three-month Treasury bills, exhibits a negative relationship with the aggregate number of mergers. When economic activity is in an expansion, the Federal Reserve tends to raise rates, coinciding with a decrease in merger activities. Financial conditions, assessed through the Dow Jones index, show a similar long-run influence. These results suggest that merger waves begin during economic downturns when interest rates are lower. Furthermore, we demonstrate that the aggregate number of mergers and the Dow Jones index play significant roles in restoring the system to the long-run equilibrium when it drifts out of it. While these results emphasize that neoclassical determinants still play an important role in driving aggregate merger waves, they do not rule out behavioral explanations of merger activity.

Figures

Figure 1.

Merger waves, number of M&A transactions

Figure 2.

Mergers, industrial production, three-month T-bill rate, Dow Jones closing price

Figure 3.

Roots of the companion matrix

Table 1.

Augmented Dickey–Fuller test results for variables in first differences

Lags	Log mergeractivity	Log industrial production	Three months T-bill rate	Log Dow Jones closing price
1	−5.614 (0.0000)	−7.729 (0.0000)	−8.182 (0.0000)	−7.426 (0.0000)
2	−4.691 (0.0001)	−6.040 (0.0000)	−7.011 (0.0000)	−7.280 (0.0000)
3	−5.083 (0.0000)	−5.270 (0.0000)	−6.579 (0.0000)	−7.699 (0.0000)
4	−5.483 (0.0000)	−4.938 (0.0000)	−5.077 (0.0000)	−5.552 (0.0000)
5	−4.868 (0.0000)	−4.345 (0.0004)	−4.818 (0.0001)	−4.683 (0.0001)
6	−3.725 (0.0038)	−3.113 (0.0256)	−4.126 (0.0009)	−3.265 (0.0165)
7	−3.515 (0.0076)	−2.734 (0.0683)	−3.813 (0.0028)	−2.656 (0.0819)
8	−3.113 (0.0256)	−3.032 (0.0320)	−3.232 (0.0182)	−2.872 (0.0487)
9	−3.543 (0.0070)	−2.773 (0.0623)	−2.716 (0.0714)	−2.109 (0.2410)
10	−4.191 (0.0007)	−2.103 (0.2432)	−2.317 (0.1666)	−2.126 (0.2342)

Notes:

The 1, 5 and 10% Dickey–Fuller critical values are −3.520, −2.896 and −2.583; MacKinnon approximate p-values in parenthesis; the data span 1928 to 2019; “Log” refers to natural logarithms; the regressions do not have drift parameter. We have also estimated the Dickey–Fuller regressions with a drift parameter – the results do not change

Source: Authors’ own

Table 2.

Estimated cointegrating equation

Variable	Coeff.	Standard error	Test statistic	p-value	95% conf. interval
Log merger activity	1	–	–	–	–	–
Industrial production	0.178**	0.090	1.97	0.049	0.000	0.354
Three-month T-bill rate	49.24***	16.403	3.00	0.003	17.092	81.391
Log Dow Jones price	4.644***	1.524	3.05	0.002	1.662	7.635
Trend	−0.601***	0.111	−5.41	0.000	−0.818	−0.383
Constant	−29.332	–	–	–	–	–

Notes:

***p< 0.01; **p< 0.05; *p< 0.1; the data span 1928 to 2019; “Log” refers to natural logarithms. Johansen normalization restriction imposed. The value of the trace test statistic is 31.6363 with a 5% critical value of 42.44

Source: Authors’ own

Notes

1. While we were able to obtain data on M&As dating back to 1880s, data on industrial production and the rates on three-months T-Bill were obtained only from 1928.

2. https://pages.stern.nyu.edu/∼adamodar/New_Home_Page/datafile/histretSP.html