Lithium industry in the behavior of the mergers and acquisitions in the US oil and gas industry

ABSTRACT

Is lithium affecting the US oil and gas industry strategies? Lithium has an increasingly strategic role as clean technologies emerge, affecting the strategies of oil and gas companies in response to energy trends. This paper contributes to this literature, studying the dynamics of lithium industry and mergers and acquisitions in the US oil and gas industry in time–frequency domain. Methodologies based on continuous wavelet transform and vector autoregression models are used, and the results indicate that both time series are correlated in the long term, where mergers and acquisitions’ US oil and gas industry dependence on lithium industry has increased, starting in the early 2014 until the end of the sample. Evidence of causality is not found between both time series.

KEYWORDS:

• Causality
• lithium industry
• mergers and acquisitions
• oil and gas industry
• wavelet analysis

JEL CLASSIFICATION:

• COO
• C22
• E30
• Q42

Acknowledgment

Comments from the Editor and two anonymous reviewers are gratefully acknowledged.

Notes

¹ https://sites.google.com/site/aguiarconraria/joanasoares-wavelets.

² The seminal paper by Torrence and Compo (1998) is one of the first works to discuss significance testing for wavelet and cross-wavelet power. Based on a large number of Monte Carlo simulations, these authors concluded that the local wavelet power spectrum of a white noise or an AR(1) process, normalized by the variance of the time series, is well approximated by a chi-squared distribution. Torrence and Compo (1998) also derived empirical distributions for cross-wavelet power. On the other hand, Ge (2007, 2008) reconsidered the discussion of the significance testing for the wavelet, cross-wavelet power, and wavelet coherency.

Aguiar-Conraria and Soares (2014) concentrate on the use of a specific wavelet (the Morlet wavelet) and, assuming a Gaussian white noise process, analytically derive the corresponding sampling distributions. However, these sampling distributions were shown to be highly dependent on the local covariance structure of the wavelet, a fact that makes the significance levels intimately related to the specific wavelet family used, meaning that they cannot be generalized. Naturally, no work has been done on significance testing for the partial coherency, as this measure has not been introduced elsewhere. Maraun et al. (2007) argued that point-wise significance tests, like the ones described, generate too many false positive. They proposed an area-wise test, which aims at correcting false positives of point-wise tests, based on the area on shape of the significant regions. Lachowicz (2009), however, shows that some more work needs to be done in this area.

Following the examples and the toolbox provided by Aguiar-Conraria and Soares (2014), the tests of significance are either based on very simple Monte Carlo simulations or bootstrapping. They fit an ARMA (p, q) model and then construct new samples by bootstrap or by drawing errors from a Gaussian distribution. In the first option, they use the very basic bootstrap technique described in section 2.1 of Berkowitz and Kilian (2000). Related with the statistical test for the phase difference, Ge (2008) showed that under the null of no linear relation between two variables, the phase angle will be uniformly distributed. Hence, it will be dispersed between −π and π. Because of that, Ge (2008) argues that one should not use significance tests for the wavelet phase-difference. Instead, its analysis should be complemented by inspection of the coherence significance.

³ This study uses the daily number of mergers and acquisitions in the US oil and gas industry to form the aggregate monthly series from 2011 to 2017.

Additional information

Funding

The second-named author gratefully acknowledges financial support from the Ministerio de Economía y Competitividad: [grant number ECO2017-85503-R].