Investment, Tobin's Q, and Cash Flow Across Time and Frequencies
| Author | Fabio Verona |
| DOI | http://doi.org/10.1111/obes.12321 |
| Published date | 01 April 2020 |
| Date | 01 April 2020 |
331
©2019 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd.
OXFORD BULLETIN OF ECONOMICSAND STATISTICS, 82, 2 (2020) 0305–9049
doi: 10.1111/obes.12321
Investment, Tobin’s Q, and Cash Flow Across Time
and Frequencies*
Fabio Verona
Bank of Finland, Monetary Policy and Research Department; University of Porto, CEF.UP,
Porto, Portugal (e-mail: fabio.verona@bof.fi)
Abstract
The investment literature has long acknowledged the time- and frequency-varying dynam-
ics of the relationship between investment, Tobin’s Q and cash flow. In this paper, we use
continuous wavelet tools to estimate and assess the relationship between these variables
simultaneously at different frequencies and over time. We find that (i) Q and cash flow
are complementary sources of information for investment, the former being more impor-
tant for firms’ investment decisions in the medium-to-long run and the latter at business
cycles frequencies; and (ii) investment-Q sensitivity declines over time at all frequencies,
while investment–cash flow sensitivity declines at business cycles frequencies but remains
largely stable over the medium-to-long run.
I. Introduction
According to the original Q theory of investment proposed by Tobin (1969), corporate
investment is an increasing function of average Q, the market valuation of a firm divided
by the replacement cost of its capital stock. However, if Q fails to control for the entire
investment opportunity set (i.e. if Q is an imperfect measure of both short- and long-run
investment expectationsand decisions), then other variables could act as useful instruments
in investment regressions. Over the years, several papers, both empirical (e.g. Bond and
VanReenen, 2007) and theoretical (e.g. Abel and Eberly, 2011), have indeed demonstrated
JEL Classification numbers: C49, E22, G31
*This is a revised version of the paper that circulated as ‘Q, investment, and the financial cycle’. The views
expressed in this paper are those of the author and do not necessarily reflect the views of the Bank of Fin-
land. The results in this paper were obtained using the ASToolbox2018, a wavelet Matlab toolbox available at
https://sites.google.com/site/aguiarconraria/joanasoares-wavelets/the-astool-
box. We would like to thank Luis Aguiar-Conraria and Maria Joana Soares for sharing their toolbox; Andrea
Caggese, Hursit Celil (discussant), Olivier Dessaint (discussant), Martin Ellison, Michael Funke, Esa Jokivuolle,
Alberto Martin, Manuel M. F. Martins, Salvatore Nistic´o, Ricardo Reis, Toni Whited, Francesco Zanetti and
conference and seminar participants at the 2015 CFE Conference (London), the 2015 Helsinki Macro Research
Away Day, the 2016 FMA Conference (Helsinki), the 2016 Finance Forum (Madrid), the 2016 Meeting of the
Portuguese Economic Journal (Coimbra), the 2016 Annual Conference of the International Association forApplied
Econometrics (Milan), Aalto University, Bank of Finland, and Hamburg University for their useful comments and
suggestions. Finally, I thank the editor, Francesco Zanetti (in his editorial role in addition to discussion above) and
three anonymous referees for comments that helped guide the revision.
332 Bulletin
that a number of variables (mainlyrepresenting liquidity and finance constraints) are useful
in explaining aggregate investment. Furthermore, investment spending of firms is likely
to be sensitive not only to the availability of external equity finance (as proxied by Q),
but also of internal funds (cash flow) and external debt finance (bank loans and corporate
bonds). Hence, other financial variables beyond Q likely influence investment decisions.
Empirical evidence has shown that the investment sensitivities to Q and cash flow vary
dynamically and nonlinearly over time. For instance, McLean and Zhao (2014) find that
investment-Q (investment–cashflow) sensitivity varies overthe business cycle, with higher
sensitivity during expansions (recessions). Several authors (see e.g. Lettau and Ludvigson,
2002; Abel and Eberly, 2012) have also suggested that Q and liquidity variables may not
be related to investment in the same way at all frequencies and horizons. For instance,
Abel and Eberly (2012) argue that movements in Q could primarily affect investment over
long-horizons into the future, as Q (based on equity prices) is a forward-looking variable
that captures information about the value of long-term growth options available to the
firm. Hence, Q may not be very informative about near-term investment plans and thus
could perform poorly in explaining current investment.Likewise, liquidity variables mainly
reflect current technology, productivity and demand, so current cash flow, sales or profits
could predict short-run investment better than long-term investment.
In the light of this empirical evidence, three features should be taken into account when
estimating the equation linking investment, Q and cash flow. First, investment depends on
several variables. Even if a variable is potentially a sideshow for aggregate investment (as
Grullon, Hund and Weston, 2018 argue for Q), it is nevertheless important to control for
it as it could affect the investment sensitivity with regard to the other variable. Second,
the relationships between these variables are time-varying. Third, the relationships are
frequency dependent. Against this background, the main contribution of this paper is to
analyse the time-and-frequency-varying role that Q and cash flow have played as driversof
aggregate investment dynamics. We do so by using the continuous wavelet transform tools
developed by Aguiar-Conraria, Martins and Soares (2018). These tools allow us to esti-
mate investment-Q and investment–cashflow sensitivities over time and across frequencies
simultaneously in a multivariate setting.
Our first main result is that the information content of Q and cash flow are comple-
mentary rather than alternative to each other, since Q and cash flow relate to investment at
different frequencies. Namely, we find a positive, stable medium-to-long-run relationship
between investment and Q, while the positive and stable relationship between investment
and cash flow is largely confined at business cycles frequencies. The second main result
concerns the time variation of the investment sensitivities to Q and cash flow. We find
that investment-Q sensitivity declines over time at all frequencies, while investment–cash
flow sensitivity declines at business cycles frequencies but remains largely stable at lower
frequencies.
These findings are similar to those obtained using the financial accelerator model of
Bernanke, Gertler and Gilchrist (1999), whereas the availability of credit depends on the
firm’s leverage, which in turn fluctuates endogenously over the business cycle. Over the
long run, however,investment is driven by the fundamental valueof the fir m’sprojects, thus
firm’s investment becomes less sensitive to the availability of finance. These results also
highlight the importance of looking beyond business cycle fluctuations and support the
©2019 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd
To continue reading
Request your trial