ECONOMICS – WORKING PAPERS 2024/04

The impact of the digital

and green transitions on
investment inefficiency
October 2024
The impact of the digital
and green transitions on
investment inefficiency

October 2024
The impact of the digital and green transitions on investment inefficiency

© European Investment Bank, 2024.

EIB Working Paper 2024/04
October 2024

All rights reserved.

All questions on rights and licensing should be addressed to publications@eib.org.

European Investment Bank

98 -100, boulevard Konrad Adenauer
L-2950 Luxembourg

Authors
Francesco Cimini (EIB)
Fotios Kalantzis (EIB)

This is a publication of the EIB Economics Department.

economics@eib.org
www.eib.org/economics

About the Economics Department

The mission of the EIB Economics Department is to provide economic analyses and studies to support the Bank in its operations and in
the definition of its positioning, strategy and policy. The department, a team of 40 economists, is headed by Director Debora Revoltella.

Disclaimer
The views expressed in this publication are those of the authors and do not necessarily reflect the position of the European Investment
Bank. EIB working papers are designed to facilitate the timely exchange of research findings. They are not subject to standard EIB
copyediting or proofreading.

For further information on the EIB’s activities, please consult our website, www.eib.org. You can also contact our Info Desk, info@eib.org.

Published by the European Investment Bank.

Printed on FSC® Paper.

pdf: QH-01-24-012-EN-N ISBN 978-92-861-5824-7 ISSN: 2599-736X doi: 10.2867/1050208

1 – INTRODUCTION ........................................................................................................................................ 4

2 – LITERATURE REVIEW AND TESTING HYPOTHESES..................................................................... 5

2.1 – GREEN INVESTMENTS AND INVESTMENT INEFFICIENCY ............................................................................................ 5
2.2 – DIGITALISATION AND INVESTMENT EFFICIENCY ...................................................................................................... 6
2.3 – DIGITALISATION, GREEN INVESTMENTS AND INVESTMENT EFFICIENCY ........................................................................ 7

3 – DATA AND METHODOLOGY ................................................................................................................ 7

3.1 – SAMPLE SELECTION ......................................................................................................................................... 7
3.2 – EXPLANATORY VARIABLES ................................................................................................................................. 8
3.3 – THE DEPENDENT VARIABLE: INVESTMENT INEFFICIENCY ........................................................................................... 9
3.4 – OTHER EXPLANATORY VARIABLES ....................................................................................................................... 9
3.5 – MODELS ..................................................................................................................................................... 11

4 – RESULTS ...................................................................................................................................................... 12

5 – ROBUSTNESS CHECKS........................................................................................................................... 14
5.1 – FIRST ALTERNATIVE SPECIFICATION OF THE INVESTMENT EFFICIENCY EQUATION ......................................................... 14

6 – CONCLUSIONS ......................................................................................................................................... 16

REFERENCES .................................................................................................................................................... 18

TABLES .............................................................................................................................................................. 22
The impact of the digital and green transitions on
investment inefficiency1

Francesco Cimini# and Fotios Kalantzis∆

Abstract – This study examines the impact of green and digital investments on the investment inefficiency level of
European firms. We define investment inefficiency as the deviation from the optimal investment level, which depends
on both the net present value (NPV) of the projects and the marginal benefit and cost of investment. Leveraging
matched data from the European Investment Survey (EIBIS) and ORBIS, which results in a sample of 4,892 firm-
year observations from 27 European countries surveyed over the period 2021-2023, we employed a panel data
regression model to estimate the effect of green and digital investments on investment inefficiency. Our analysis shows
that both types of investments reduce investment inefficiency, particularly for under-investing firms. We also find
evidence of a statistically significant interaction effect between green and digital investments for over-investing firms,
suggesting that digital technologies can enhance the efficiency gains from green investments. Our results have important
implications for policy makers and business managers who aim to foster the twin digital and green transition in Europe
and improve their investment efficiency and competitiveness.

J.E.L. Classification Numbers: M41, G31, Q53, O33

Keywords: European Investment Bank Investment Survey, Investment Inefficiency, Green investment,
Digital investment, Twin transition.

# ∆
francesco.cimini@aol.com; EIB, f.kalantzis@eib.org

1
We are very grateful to Sofia Dominguez and Christoph Weiss (EIB) for the useful comments and suggestions.
The views expressed in this paper do not necessarily reflect those of the European Investment Bank. The usual
disclaimers apply.

The impact of the digital and green transitions on investment inefficiency | 3

1 – Introduction
Digitalisation and the green transition are two of the most important trends characterising the European
economy nowadays. In 2022, 53% of European firms invested in measures aimed at adapting to weather
events and reducing GHG emissions; at the same time, 46% of them invested in digitalisation, a trend
which was reinforced by the COVID-19 pandemic (European Investment Bank, 2022). These
transformations are likely to further reinforce under the impulse of strong policy support in the European
Union. Specifically, with the aim of achieving the goals set forth in the European Green Deal, the
European Commission aims at mobilising around €1tn over the 2021-2030 period to finance green
investments; at the same time, around a third of the €750bn Next Generation EU Funds will be devoted
to digitalisation. Furthermore, the development of Industry 4.0-related technologies has gained
considerable momentum since the introduction of the European Industry Initiative in 2016 (Teixeira
and Tavares-Lehmann, 2022).

The literature points to the existence of important mutual implications between investments in green and
digital technologies (see Ortega-Gras et al., 2021; Husain et al., 2022; Fouquet and Hippe, 2022) and goes
so far as to analyse them jointly as the “twin transition”. Whether, however, digitalisation can enhance
the green transition or slow it down is still controversial (Bianchini et al., 2023). Some scholars tend to
adopt an optimistic stance, arguing that digitalisation has a positive impact on the transition (Haller et
al., 2023; Lesecq et al., 2022). In their view, digitalisation results in a larger availability of data about
production processes, thereby allowing to optimise them, monitor their environmental impact, and
reduce industrial waste (Muench et al., 2022). Other scholars are instead more pessimistic. Andersen et
al. (2021) suggest that the digitalisation of the economy may encourage the adoption of environmentally
unsustainable practices. Also, in advanced countries, data centres consume significant amounts of energy
(Strubell, 2019). Needhidasan et al. (2014) outline the issue of disposing technological devices, which
contain many elements that cannot be recycled. Finally, Fouquet and Hippe (2022) suggest that
communication transitions tend to be faster than energy transitions, which may result in a high-carbon
structural transition of the economy.

This paper attempts to analyse the relationship between the twin digital and green transition and
investments inefficiency levels on a sample of European firms. The concept of investment efficiency is a
direct by-product of an intense scholarly effort which, starting from Franco Modigliani and Merton
Miller’s 1958 seminal paper on the irrelevance of firm’s capital structure, attempted to construct a theory
of optimal corporate investment. In particular, according to Modigliani and Miller (1958), in a world
without market frictions firms should invest in all projects showing a positive net present value (NPV).
Hayashi (1982) built on this theory by adding that firms should keep investing until the marginal benefit
of investment equals the marginal cost. Recognising that market frictions do exist, however, the literature
has also examined the sources of potential investment inefficiencies. Specifically, agency problems and
information asymmetries are understood as the main sources of deviation from optimal investment level
(Myers and Majluf, 1984; Khediri, 2021). Hubbard (1998) points to financing constraints as a source of
frictions, while Jensen and Meckling (1976) outline the role of agency problems as a source of
inefficiencies. In particular, firms may under-invest or over-invest. In a scenario with no market frictions,
efficient investment amounts to selecting all projects with a positive NPV; thus, over-investment amounts
to selecting projects with a negative NPV, while under-investment amounts to forgoing some projects that
nonetheless have a positive NPV (Biddle et al., 2009).

In the first place, our study attempts to assess whether corporate digital and green investments can,
separately, affect a firm’s investment efficiency level; then, it focuses on firms embarking on the twin
transition, that is to say, on an investment strategy comprising green and digital investments at the same
time. The existing literature documents a negative relationship between green investments and

4 | The impact of the digital and green transitions on investment inefficiency

investment inefficiency, on one side, and between digitalisation and investment efficiency on the other.
Arguably, however, it does not address thoroughly the interaction between investment inefficiency and a
“twin” investment strategy involving both digitalisation and climate-related investments. Should these
latter engage in a mutually reinforcing relationship, there might be positive spillovers on the overall
investment strategy, resulting in a lower level of investment inefficiency; conversely, a mutually destructive
relationship between the two may lead to a suboptimal investment plan, signalled by a higher investment
inefficiency. In order to assess the effect of a combination of green and digital investments, this paper
combines the ORBIS dataset with European Investment Bank Investment Survey (EIBIS) dataset, which
provides insight into the investment behaviour of a large set of European companies.

Our analysis yields three key results. First, we show that green investments, aimed at mitigating climate
change or at enhancing firms’ resilience to climate change-related adverse events, reduce investment
inefficiency, and that this effect is largely driven by under-investing firms. We also show that investments
in digital technologies reduce firms’ investment inefficiency level. These results are robust to alternative
tests. Finally, we find evidence that a moderating effect of digital investments on green investments exists
for firms that over-invest, whereby digital investments enhance the positive effect of green investments on
investment efficiency.

The rest of the paper is organised as follows. Section 2 reviews the existing literature and presents the
testing hypotheses. Section 3 presents the data and the methodology. Section 4 and 5 present and discuss
the results, while section 6 concludes.

2 – Literature review and testing hypotheses

The purpose of this paper is to explore the relationship between investment inefficiency, green
investments and investments in digital technologies, with a particular focus on whether digital
investments can modify the effect, if any, of green investments on investment inefficiency. Several papers
have addressed separately the effects of green investments on investment inefficiency and of digital
investments on investment inefficiency. However, to the best of our knowledge, no study to date has dealt
expressly with whether a joint effect of digital and green investments on investment efficiency is verified.
To shed some further light on the topic, we start by reviewing the literature. The three research streams -
green and digital investments on investment efficiency, as well as the twin transition - are analysed
separately.

2.1 – Green investments and investment inefficiency

Climate change poses a severe threat to the stability of the biosphere and prompt action is needed in
order to slow it down and reverse its pace (IPCC, 2014). To this purpose, many firms have already adopted
measures to limit their climate footprint (European Investment Bank, 2022). Whereas few research papers
have investigated whether these efforts can impact a firm’s investment efficiency level, several studies have
addressed the relationship between corporate social responsibility and investment inefficiency (see, for
instance, Lee, 2020; Khediri, 2021; Benlemlih and Bitar, 2015; Samet and Jarboui, 2017; Cook et al.,
2018; Erawati et al., 2021). Since green investments are a type of Corporate Social Responsibility (CSR)
engagement, it is worth addressing the main results of these papers.

Two conflicting views exist on the effects of corporate CSR engagement. According to the first stream of
thought, which can be traced back to Friedman (1970), managers adopting CSR initiatives are arbitrarily
channelling funds towards negative-NPV projects in what constitutes an agency problem. Krüger (2015)
builds on this view and argues that CSR practices create conflicts between stakeholders. Some early
empirical results (see Aupperle et al., 1985) support this argument, showing that CSR investments have
a depressing effect on firm performance. Instead, according to the Stakeholder Theory (Freeman et al.,

2 – Literature review and testing hypotheses | 5

2010), CSR can enhance firm value. Following this view, Porter and Kramer (2006) argue that, if carefully
embedded in corporate strategy, CSR engagement can foster firms’ competitiveness.

The scholarly debate on the effects of CSR is far from resolved and, since Friedman and Freeman’s
attempts, it has extended to various aspects of business performance, including investment inefficiency.
Using a sample of Western European firms observed from 2004 to 2011, Khediri (2021) documents a
negative correlation between CSR engagement and investment inefficiency. The correlation is stronger
for those CSR practices that address primary stakeholders. He argues that the reduction of information
asymmetries is the main channel through which this effect materialises. Lee (2020) shows that CSR
“significantly mitigates investment inefficiency among Taiwanese firms”. Using a sample of Indonesian
companies, Erawati et al. (2021) find that higher CSR involvement mitigates the suboptimal investment
behaviour which typically characterises family businesses. Cook et al. (2018) use a large dataset of publicly
traded companies and find that firms with high CSR engagement invest more efficiently. Likewise, Samet
and Jarboui (2017) use a panel of around 400 European listed companies and find that better CSR
performance helps bring corporate investment levels closer to the optimum. They also point to the
reduction of information asymmetries as the main driver of this effect. Finally, in an influential paper
involving US firms, Benlemlih and Bitar (2015) use a sample of 21,030 firm-year observations and find a
strong negative correlation between firms’ CSR involvement and investment inefficiency, and that once
again such effect is more pronounced when considering CSR practices that address firms’ main
stakeholders.

Conversely, the correlation between green investments and investment inefficiency has received less
scholarly attention. Kim and Kim (2023) find that investment inefficiency correlates positively with firm-
level greenhouse gas emissions in a sample of Korean firms. Using a panel of Chinese listed companies,
Zeng et al. (2019) also find that environmental commitment has a strong negative effect on investment
inefficiency, which however needs some time to manifest itself.

Notwithstanding the scarcity of empirical results, we are then able to predict that higher environmental
commitment lowers firms’ investment inefficiency, and we postulate the following hypothesis:

H1: Climate action reduces the investment inefficiency level of EU firms.

2.2 – Digitalisation and investment efficiency

Corporate digitalisation can be understood as the progressive incorporation of digital technologies into
an organisation’s processes. Nowadays, many firms rely on digital technologies for conducting their
business operations; not coincidentally, digitalisation as a research topic in the field of social sciences has
been drawing significant interest over the last years.

Digitalisation is generally thought of as improving firm performance (Jung and Gómez-Bengoechea,

2022). Thanks to digitalisation, firms can analyse large amounts of data and identify new business
opportunities (Benitez et al., 2022). Also, digitalisation can potentially lower communication costs
(Jorgenson, 2001), both between the firm and external stakeholders (Yu et al., 2010) and at an internal
firm level (Zhai et al., 2022). Finally, according to Ribeiro-Navarrete et al. (2021), digitalisation, when
applied to production processes, has the power to generate competitive advantages.

Empirically, many studies confirm the existence of a positive effect of digitalisation on company
performance (Jung and Gomez-Bengoechea, 2022). These studies outline that such effect materialises
through the channels that were just outlined. However, not all studies agree with these general findings.
For instance, Aral and Weill (2007) find that digitalisation does not significantly affect performance,
while Cappa et al. (2021) find a negative relationship. Instead, the relationship between a firm’s
digitalisation effort and investment inefficiency has drawn less academic attention. Huo and Wang (2022)

6 | The impact of the digital and green transitions on investment inefficiency

analyse a panel of Chinese listed companies observed from 2007 to 2019 and find that the development
of the digital economy has a negative effect on firm-level investment inefficiency by reducing over-
investment. Xu et al. (2023) also use a panel of Chinese listed companies to examine the effect of firm-
level inefficient investment on digital transformation; they find that corporate investment inefficiency is
detrimental to firm-level digitalisation, and that such negative relationship is exacerbated by the presence
of financing constraints.

Based on the findings of these studies, we can predict that higher levels of corporate investment in digital
technologies will be associated with lower levels of investment inefficiency. Thus, we postulate the
following:

H2: Digital investments reduce the investment inefficiency level of EU firms.

2.3 – Digitalisation, green investments and investment efficiency

At a theoretical level, digitalisation has the potential to affect firms’ green transition to a significant extent
(Mondejar et al., 2021). Muench et al. (2022) attempt to classify the various channels by which
digitalisation impacts the transition, showing that the interplay between the two revolves essentially
around the opportunity offered by digital technologies to analyse large amounts of data. By acquiring
valuable and detailed information, firms may monitor their activities and reduce production waste; they
could also identify new and more efficient production processes employing forecasting techniques.
Additionally, virtualising commercial practices reduces the need for people to relocate, which lowers
emissions and increases efficiency. Other channels may exist in some specific industries. For instance, in
the transport sector, the availability of data can help optimise traffic flows, thus reducing emissions and
reducing inefficiency (Muench et al., 2022). That digitalisation affects the green transition in firms is also
supported by theoretical evidence. Using a panel of Chinese listed companies observed from 2007 to
2020, Liu et al. (2023) find that digitalisation improves corporate green innovation. Li et al. (2023) also
use a panel of Chinese companies to show that digitalisation fosters green innovation at multiple levels.

To the best of our knowledge, the relationship between digitalisation, green investments and investment
efficiency has never been addressed directly by the academic literature. This paper, among other things,
hypothesises the following:

H3: Green investments reduce investment inefficiency more effectively for firms that also invest in digital
technologies.

3 – Data and Methodology

3.1 – Sample Selection

To examine empirically the relationship between green investments, investments in digital technologies
and investment inefficiency, we use data from two sources: the 2024 edition of the annual European
Investment Bank Group Survey on Investment and Investment Finance (EIBIS) and the Bureau van Dijk
ORBIS database. EIBIS is an EU-wide survey that gathers information on firms' investment activities and
financing requirements since 2016. It uses a stratified sampling methodology and is designed to be
representative at the EU, country, sectoral and firm size levels. The Bureau van Dijk’s ORBIS database
provides the balance sheets and income statements of the surveyed firms. The main advantage of the
dataset is that it provides unique information on firms' investments to tackle climate change-related risks
and digital adoption, as well as other variables that describe their profiles and financial positions.

To construct our sample, we consider EIBIS firms with non-missing financial information observed from
2021 to 2023 in the 27 EU Member states. EIBIS does not include financial firms. This is convenient, in

3 – Data and Methodology | 7

that their investment behavior is affected by government regulation; thus, including them would weaken
the validity of our results. Not coincidentally, several papers (such as Biddle et al., 2009, Chen et al., 2017
and Benlemlih and Bitar, 2015) exclude financial firms from their analyses. Our final sample includes
4,892 firm-year observations derived from 3,628 different firms. Table 1 presents the sample distribution
by year, country and sector. The sectorial classification is based on four broad categories: manufacturing,
infrastructure, services and construction 2.

Most of the observations were taken in 2021 and 2022, with 2023 accounting for as little as 10% of the
overall sample. The reason is that only a fractional portion of the firms that were surveyed in 2023 had
published their annual reports at the time the survey was taken. The sectorial distribution is more
balanced, although manufacturing is slightly overrepresented as it accounts for almost 34% of
observations, while construction is underrepresented (19%). Finally, we consider the regional
breakdown 3. Firms in Central and Eastern Europe are overrepresented with 42% of observations, while
firms in Southern Europe account for around 29% of the observations each.

3.2 – Explanatory variables

3.2.1 - The GREEN variable

We selected two variables from the EIBIS questionnaire to quantify firms’ investment level in green and
digital technologies. First, we focus on green investments. Variable GREEN is a dummy variable that
equals one if a firm reported investing in measures to tackle the impact of weather events or to reduce
carbon emissions in a given year. These measures may include both adaptation and mitigation actions.

Table 1 also presents the distribution of GREEN by year, sector and country group. Over the three years,
an increasing number of firms reported investing or having invested in mitigation or adaptation. Such
increase was particularly pronounced between 2021 and 2022, arguably a positive rebounding after the
economic shock caused by the COVID-19 pandemic. When considering the regional distribution, the
highest average value for GREEN (around 50%) is observed in Western and Northern Europe, while the
lowest value (around 31%) is observed in Central and Eastern Europe. Manufacturing and transportation
have the highest average value for this variable.

3.2.2 - The DIGITAL variable

Variable DIGITAL instead is a dummy equal to one if the firm in a given year implemented multiple
digital technologies. A first striking difference with GREEN is that average values tend to be higher with
DIGITAL. In economic terms, this means that firms are on average more prone to invest in digital
technologies than to invest in technologies aimed at reducing emissions or making the firm more resilient
to climate-change-related events.

Table 1 also shows the distribution of DIGITAL by year, sector and country group. Once again, we observe
a strong positive rebounding between 2021 and 2022, although figures decline between 2022 and 2023.
Manufacturing and transportation still retain the two top positions, while construction is again the one
with lowest values. The regional distribution is fairly balanced, with Western and Northern Europe still
displaying the highest value.

2
Based on the NACE classification of economic activities, (a) “Manufacturing” includes firms in group C, namely manufacturing
companies; (b) “Construction” includes firms in group F (construction companies); (c) “Services” includes firms in group G
(wholesale and retail trade) and I (accommodation and food services activities) while (d) “Infrastructure” includes firms in group
E (utilities), H (transportation and storage) and J (information and communication).
3
Western and Northern Europe includes the following countries: Austria, France, Germany, Luxembourg, Netherlands,
Sweden, Finland, Ireland, Denmark, Belgium; Southern Europe includes Cyprus, Italy, Malta, Spain, Greece, Portugal; Eastern
Europe includes Bulgaria, Romania, Slovenia, Czech Republic, Slovakia, Poland, Latvia, Lithuania, Estonia, Hungary, Croatia.

8 | The impact of the digital and green transitions on investment inefficiency

3.3 – The dependent variable: investment inefficiency

Investment inefficiency can be thought of as the extent to which firms fail to undertake all projects with
positive net present value. Obtaining a measure of a firm’s level of investment inefficiency is difficult,
however, because the full array of projects that are available to a firm cannot be observed. In order to
quantify the investment inefficiency level associated with a company i in year t, we follow prior research
(e.g., Benlemlih and Bitar, 2015; Biddle et al., 2009; Gomariz and Ballesta, 2014; Mazboudi and Hasan,
2018) to estimate the firm-specific model of investment (measured by capital expenditures over one-year
lagged total assets) as a function of growth opportunities (measured by annual sales growth), and then use
residuals as a firm-specific proxy for deviation from expected investment, representing the level of
investment inefficiency. Specifically, this paper uses the residuals from the following equation as the
investment inefficiency variable:
𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 = 𝜇𝜇0 + 𝜇𝜇1 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡−1 + 𝛾𝛾𝑖𝑖,𝑡𝑡
(1)

where INVi,t is the yearly variation in fixed assets plus depreciation, all scaled by lagged total assets of firm
i in year t, while SALESGROWTHi,t-1 is the rate of change in sales of firm i from t – 2 to t – 1. Because
measurement errors in most datasets affect more than 1% of observations in each tail (Hampel et al.,
1986), we winsorise these variables – INVi,t and SALESGROWTHi,t – at the 10% level. The equation is
then estimated cross-sectionally for each year and sector for all industry-years with at least 15 observations.
The sectorial classification is based on NACE two-digit codes. The residuals γi,t from the regression model
reflect the deviation from the expected investment level, and they are used as a firm-specific proxy for
investment inefficiency. The positive residual indicates over-investment, while the negative residual
indicates under-investment (Jensen, 1986; Biddle et al., 2009). Since both under-investment and over-
investment are considered inefficient investments, we take the absolute value of the residuals and use it
as a firm-specific proxy for investment inefficiency.

Similar to the variables GREEN and DIGITAL, we look once again at the distribution by year, sector and
region (see Table 1). Although the absolute level of inefficiency has little economic significance, relative
values of inefficiency can provide useful insight. Investment inefficiency has remained stable over the
years; also, the level of investment inefficiency in Central and Eastern European countries is significantly
higher than that of other regions. Likewise, the construction sector shows a higher level of investment
inefficiency than the other sectors, which are fairly aligned.

3.4 – Other explanatory variables

To control for country-, sector- and year- specific effects, we include three categorical variables, namely,
COUNTRY, SECTOR and WAVE. COUNTRY includes all the 27 EU Member states, while SECTOR is
based on a broad classification which categorises firms as belonging to either, manufacturing, services,
construction and infrastructure.

Moreover, in order to identify the effect of green investments and investments in digital on investment
inefficiency, and following a large body of previous research (among others, Benlemlih and Bitar, 2015;
Lee, 2020; Khediri, 2021; Jensen, 1986), we include several control variables. These variables may have
an impact on a firm’s investment inefficiency level, and including them reduces the risk of omitted
variable bias.

Managers may be inclined to exercise caution if the firm is experiencing lower profitability. They may
become particularly careful if the firm is incurring losses. For this reason, following Benlemlih and Bitar
(2015), we include variables PROF and LOSS. The former is defined as EBITA/Total assets as it appears
in the financial statement of the year prior to the survey, while the latter is a dummy equal to one if the

3 – Data and Methodology | 9

firm experienced a loss during the financial year prior to the survey. We then predict that the sign of the
relationship between LOSS and investment inefficiency be negative and, similarly, we predict a positive
correlation between INVEFF and PROF.

AGE is the natural logarithm of the difference between survey year and the year in which company i was
founded. According to Benlemlih and Bitar (2015), older companies may have more financial experience,
which reduces their investment inefficiency. On the other hand, however, they may also have more
financial resources, which may be used in unproductive ways (Lee, 2020). For this reason, we do not
predict the sign of the coefficient associated with this variable.

Following Benlemlih and Bitar (2015) we also include LEV, a proxy for firm’s leverage, computed as the
ratio between the sum of loans and long-term debts over total assets. According to Jensen (1986), a higher
leverage ratio generates agency problems when obtaining additional funds, which raises investment
inefficiency; at the same time, however, debt holders play a monitoring role on the firm, reducing
investment inefficiency (Lee, 2020; Benlemlih and Bitar, 2015; Jensen, 1986). Thus, this paper does not
predict the sign of the relationship between a firm’s financial leverage level and its level of investment
inefficiency.

Following Chen et al. (2011) and Benlemlih and Bitar (2015), we include TANG, a proxy for the collateral
value of assets, computed as the ratio between fixed assets and total assets. The higher the collateral value
of assets, the easier it is for managers to raise funds, which may lead to investment inefficiency (Lee, 2020).
Therefore, this paper predicts a positive relationship between the level of collateral value of assets and
investment inefficiency.

As a measure of financial slack, we include CASH, a variable defined as cash reserves to total assets for
firm i in year t. Higher cash reserves may lead to greater agency problems and thus to higher investment
inefficiency (Benlemlih and Bitar, 2015; Khediri, 2021). At the same time, however, managers are more
likely to engage in projects with positive NPV if there is more cash available, which increases investment
efficiency (Lee, 2020). For these reasons, once again, we do not predict the sign of the correlation between
cash level and the level of investment inefficiency.

Table 1 – Descriptive Statistics

10 | The impact of the digital and green transitions on investment inefficiency

Table 1 presents the dataset. In the first section (upper-left corner), observations are classified based on wave, sector and region.
All other sections present mean and standard deviations for the variables that are used in the model, be they the variables of the
investment efficiency equation (INV and SALESGROWTH) or the variables of the main models (INVEFF, GREEN, DIGITAL,
LOSS, PROF, AGE, LEV, TANG, CASH). For each variable, means and standard deviations by wave, sector and region are also
shown.

3.5 – Models

To test our hypotheses, and following the approach taken by Kim and Kim (2023), we run three separate
regressions aimed at testing, respectively, the impact of green investments on investment inefficiency, the
impact of investments in digital on investment inefficiency, and the moderating effect of investments in
digital on the relationship, if any, between green investments and investment inefficiency. The first
regression is the following:
𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 = 𝛽𝛽0 + 𝛽𝛽1 𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡−1 + 𝛽𝛽2 𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝑖𝑖,𝑡𝑡 + 𝛽𝛽3 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖,𝑡𝑡 + 𝛽𝛽4 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖,𝑡𝑡 + 𝛽𝛽5 𝐿𝐿𝐿𝐿𝐿𝐿𝑖𝑖,𝑡𝑡 +
𝛽𝛽6 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑖𝑖,𝑡𝑡 + 𝛽𝛽7 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖,𝑡𝑡 + ∑ 𝛽𝛽𝑗𝑗 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 + ∑ 𝛽𝛽𝑘𝑘 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 + ∑ 𝛽𝛽𝑤𝑤 𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊 + 𝜀𝜀𝑖𝑖,𝑡𝑡
(2)

Where INVEFFi,t is investment inefficiency; ß0 is a constant; GREENi,t-1 is our explanatory variable for this
model, namely green investments; LOSSi,t is a dummy equal to one if the firm experienced a loss in the
financial year prior to the survey; PROFi,t is the company’s profitability level defined as EBITA/Total
assets as of the financial statement prior to the survey; AGEi,t is the age of the company; LEVi,t is the
leverage ratio, defined as loans and long-term debt over total assets; TANGi,t is a proxy for the collateral
value of assets and CASHi,t is a proxy for financial slack. Finally, ∑ßjCOUNTRY, ∑ßkSECTOR and
∑ßwWAVE are the country, sector and time fixed effects. The former, ∑ßjCOUNTRY, is a set of dummy
variables mapping the 27 different EU Member states; ∑ßkSECTOR maps instead the four sectors firms
may belong to – manufacturing, transportation, services and transportation. Similarly, ∑ßwWAVE maps
the waves. Finally, εi,t is the error term. The equation is estimated using the random effect generalised
least squares method. The use of this methodology instead of the fixed effects approach is warranted
whenever many entities are observed during a short period of time and in presence of a large number of
time-invariant variables. In these cases, the fixed-effects method would lead to multicollinearity issues and
inconsistency of the estimators (Baltagi, 2005; Pathan, 2009). Indeed, in our sample the N is very large
compared to the number of survey waves; moreover, fixed effects estimation would inevitably lead to the
exclusion of all country and sector fixed effects as they are time-invariant.

To mitigate the risk of omitted variable bias, we set temporal precedence and regress INVEFFi,t on the
one-year lag of variable GREENi,t. Since the control variables are mainly taken from financial statements,
they refer to the financial year prior to the survey, thus both GREENi,t-1 and the control variables refer to
the same period. Because we predict that green investments reduce a company’s investment inefficiency,
we expect ß1 to be negative and significant.

As far as our second hypothesis is concerned, we run a slightly modified version of the previous model,
where GREEN is replaced by DIGITAL as the main explanatory variable. This latter is a dummy equal to
one if, as of the survey year, the company incorporates multiple digital technologies in its business. The
model will be as follows:
𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 = 𝛽𝛽0 + 𝛽𝛽1 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡−1 + 𝛽𝛽2 𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝑖𝑖,𝑡𝑡 + 𝛽𝛽3 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖,𝑡𝑡 + 𝛽𝛽4 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖,𝑡𝑡 + 𝛽𝛽5 𝐿𝐿𝐿𝐿𝐿𝐿𝑖𝑖,𝑡𝑡 +
𝛽𝛽6 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑖𝑖,𝑡𝑡 + 𝛽𝛽7 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖,𝑡𝑡 + ∑ 𝛽𝛽𝑗𝑗 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 + ∑ 𝛽𝛽𝑘𝑘 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 + ∑ 𝛽𝛽𝑤𝑤 𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊 + 𝜀𝜀𝑖𝑖,𝑡𝑡
(3)

Once again, we set temporal precedence and regress INVEFFi,t on the one-year lag of DIGITALi,t. This
latter is, again, temporally aligned with the control variables. Since our second hypothesis predicts that

3 – Data and Methodology | 11

investments in digital technologies reduce investment inefficiency, we expect once again ß1 to be negative
and significant.

Finally, to test our third hypothesis, we run a new version of the model that combines the previous two.
In this final version, we add an interaction term between the one-year lags of GREENi,t and DIGITALi,t,
which aims at capturing the moderating effect of digital investments on the relationship, if any, between
green investments and investment inefficiency. The mediator role of DIGITALi,t is tested based on the
intuition proposed by Baron and Kenny (1986) and frequently used in the subsequent literature (see
Hicks and Tingley, 2011; Samet and Jarboui, 2017; Cook et al., 2018), whereby a moderating effect is
verified to be in place if the coefficient associated with the interaction term is statistically significant.
Notably, in line with Baron and Kenny (1986), whether the coefficients associated with the individual
dummies GREEN and DIGITAL are statistically significant is, instead, not relevant to testing the
moderator hypothesis. We thus run the following model:
𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 = 𝛽𝛽0 + 𝛽𝛽1 𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡−1 + 𝛽𝛽2 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡−1 + 𝛽𝛽3 𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡−1 × 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡−1 +
𝛽𝛽4 𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝑖𝑖,𝑡𝑡 + 𝛽𝛽5 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖,𝑡𝑡 + 𝛽𝛽6 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖,𝑡𝑡 + 𝛽𝛽7 𝐿𝐿𝐸𝐸𝑉𝑉𝑖𝑖,𝑡𝑡 + 𝛽𝛽8 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑖𝑖,𝑡𝑡 + 𝛽𝛽9 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖,𝑡𝑡 + ∑ 𝛽𝛽𝑗𝑗 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 +
∑ 𝛽𝛽𝑘𝑘 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 + ∑ 𝛽𝛽𝑤𝑤 𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊 + 𝜀𝜀𝑖𝑖,𝑡𝑡
(4)

Because of hypothesis 3, which postulates a moderating effect of digital investments, whereby these latter
enhance the negative effect of green investments on investment inefficiency, we expect ß3 to be negative
and significant. Given the above, we also disregard ß1 and ß2.

4 – Results
Table 2 presents the results of the three models outlined in the previous section, namely, the regression
of INVEFF on GREEN, on DIGITAL as well as on GREEN×DIGITAL. We run the three models both
without and with controls. Moreover, not only do we run them on the whole sample; we also run separate
regressions using firms that over-invest exclusively (that is, firms with positive values of the residuals from
the investment inefficiency equation) and using firms that under-invest exclusively (namely, firms with
negative values of residuals). The two subsets, namely firms that over-invest and that under-invest, are not
equal in size: most firms (3,024 observations, around 64% of the sample) under-invest, while only 1,724
over-invest (around 36%).
Generally speaking, we find that green investments are negatively associated with investment inefficiency
in a statistically significant way. Indeed, the coefficient associated with GREEN is negative and statistically
significant when adding controls, both when considering the overall sample as well as when considering
under-investment only. Statistical significance, however, is not verified in the over-investment case. This
confirms our hypothesis 1, namely, that green investments reduce the level of investment inefficiency.
These results are in line with the empirical findings that we addressed in the literature review section (see
Kim and Kim, 2023; Zeng et al., 2019).
We also find that investments in digital technologies are negatively associated with investment inefficiency
and that such relationship is strongly statistically significant: the coefficient that we obtain from running
regression 2 has a very low p-value both without and with controls. The same results emerge when
considering the over-investment and under-investment subsets of our sample, as the coefficients
associated with DIGITAL are negative and statistically significant at the 1% level when adding controls.
This shows that our hypothesis 2 is also confirmed, which is to say, digital investments reduce the level
of investment inefficiency. These findings are also in line with those of other authors, namely Huo and
Wang (2022) and Xu et al. (2023).

12 | The impact of the digital and green transitions on investment inefficiency

Table 2 – Main model results

Table 2 presents the results obtained when running the main models, consisting of regressions of INVEFF on GREEN and
DIGITAL as well as on their interaction. The coefficients reported in the table are accompanied by either three, two, one or zero
stars, representing, respectively, a statistical significance level of 1%, 5%, 10% and higher than 10%. Below each coefficient the
corresponding standard error is also present. Each of the three sections, referred to regressions of INVEFF on GREEN, DIGITAL
and GREEN×DIGITAL, is further divided into three parts: regressions run using the whole sample; regression run over the
sample of firms that over-invest only; regressions run over the sample of firms that under-invest only. For each of these parts,
both the univariate and the multivariate analyses are shown. In particular, LOSS, PROF, AGE, LEV, TANG and CASH are the
control variables. Year and sector fixed effects results are also shown in the table, while country fixed effects, although present,
are not displayed.

Turning to the moderating role of digital investments on the relationship between green investments and
investment inefficiency, we find that such effect is present when considering the whole set of firms as well
as over-investing firms. Recalling that a moderating effect exists when the coefficient of the interaction
term is statistically different from zero (Baron and Kenny, 1986), we note that the coefficients in these
two cases are negative and statistically significant at the 5% level. Conversely, the coefficients are not
statistically significant in the under-investment case. This constitutes evidence that a moderator role of
digital investments exists whereby these latter enhance the positive effect of green investments on
investment efficiency; our third hypothesis is then also verified.
Next, we consider control variables. A number of studies such as Cook et al. (2018) and Lee (2020)
document a negative relationship between variable LOSS and investment inefficiency, perhaps reflecting
the fact that managers may exercise extra care when choosing investments when their firm is under-
performing. In our study, the coefficient associated with LOSS is generally not statistically significant.
However, notably, PROF correlates positively and significantly with investment inefficiency, arguably
reflecting that managers may be more inclined to dissipate financial resources when their firm is
performing well, which confirm our predictions. Notwithstanding this, the sign of the coefficient reverses
when we consider firms that under-invest, where indeed we observe a negative and statistically significant
correlation. In line with the majority of studies (see Samet and Jarboui, 2017; Cook et al., 2018; Lee,
2020), AGE correlates negatively with investment inefficiency in a statistically significant way, indicating
that older firms can deploy superior investment capabilities due to higher managerial experience. We also
find strong evidence that higher financial leverage correlates positively with investment inefficiency for
firms that over-invest, perhaps because highly leveraged firms have extra cash which enhances their agency
problems, as suggested by Lee (2020). However, when considering the overall sample as well as firms that
under-invest, we find no such statistical correlation, which perhaps reflects the contrasting effects about
LEV that were outlined in section 3.4. When considering the overall sample, TANG correlates positively

4 – Results | 13
with investment inefficiency in a statistically significant way, and this effect is mainly driven by the sub-
sample of over-investing firms. This is in line with the Lee’s (2020) predictions and with our hypothesis.
We also find that higher CASH has no statistically significant correlation with investment inefficiency.
Finally, we address briefly fixed effects. Wave 2022 is associated with a higher inefficiency level than wave
2021 for under-investing firms; inefficiency is even higher when considering wave 2023. Moreover, firms
in the construction and transportation industry have a higher inefficiency level than manufacturing firms.

5 – Robustness checks
5.1 – First alternative specification of the investment efficiency equation

In line with previous research (see Biddle et al., 2009), we run additional tests to check the validity of our
results. As a first robustness check, we design a different investment efficiency equation which attempts
to fix an issue that emerges when using the standard equation described in the previous pages and used
in many studies. Indeed, when running equation (1) cross-sectionally by each year and each industry,
some of the estimated coefficients associated with SALESGROWTH turn out to be negative. In other
words, when using equation (1), some of the values of the INVEFF variable, which constitute one of the
pillars of the research design, result from equations which model a negative relationship between firms’
growth opportunities and investment growth. This is clearly counterintuitive; to fix this issue, we run a
modified version of regression (1) which is not estimated cross-sectionally by year and industry, but rather
across the whole sample of observations, and which includes country, year and industry fixed effects:
𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 = 𝜇𝜇0 + 𝜇𝜇1 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡−1 + ∑ 𝛽𝛽𝑗𝑗 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 + ∑ 𝛽𝛽𝑘𝑘 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 + ∑ 𝛽𝛽ℎ 𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊 + 𝛾𝛾𝑖𝑖,𝑡𝑡
(5)

Again, we re-run the equations. Results are displayed in Table 3. Because we are no longer dropping
observations when their industry-year is too small, we observe a slight increase in sample size. The first set
of regressions, without the interaction term, yields similar results to those of the main model. When
adding controls, the coefficients associated with the lagged GREEN variable are negative and statistically
significant both when using the whole sample and when considering firms that under-invest only.
Notably, in this latter case, the coefficient is statistically significant at the 1% level. When adding controls,
coefficients associated with the lagged DIGITAL variable are also negative and statistically significant when
considering the whole sample as well as firms that under-invest, while they are not statistically significant
when considering over-investing firms. As usual, we then consider the regressions with the interaction
term. Notably, the coefficient associated with the interaction is negative and statistically significant at the
1% level when running the regressions on the whole sample of firms, while it is negative and significant
at the 5% level when considering firms that under-invest and at the 10% level when considering firms
that over-invest. As a whole, these results provide strong support for our three hypotheses.

5.2 – Second alternative specification of the investment efficiency equation II

As a second robustness check, we attempt to modify the investment inefficiency equation by accounting
for the fact that firms’ investment behaviour may change in case of negative revenue growth. We then
introduce in equation (1) variable NEG, a dummy equal to one if the firm experienced negative sales
growth during the financial year prior to the survey, and zero otherwise (see Chen et al., 2011; Gomariz
and Ballesta, 2014; Samet and Jarboui, 2017). The modified investment inefficiency equation looks as
follows:
𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 = 𝜇𝜇0 + 𝜇𝜇1 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡−1 + 𝜇𝜇2 𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑡𝑡−1 + 𝛾𝛾𝑖𝑖,𝑡𝑡
(6)
After running again the main models, we confirm the previous results (see Table 4). When considering
the first set of regressions, we observe that green investments affect negatively the newly defined proxies

14 | The impact of the digital and green transitions on investment inefficiency

for investment inefficiency. Coefficients are statistically significant at the 10% level when considering the
overall sample, while they are statistically significant at the 5% level in the under-investment case.
Table 3 - Regression results (one single investment efficiency regression)

Table 3 presents the results of the first robustness test, consisting of regressions of INVEFF on GREEN and DIGITAL as well as
on their interaction. In this second set of regressions, the tests use a modified investment efficiency equation which is not
estimated cross-sectionally by year and industry, but rather across the whole sample of observations, and which includes country,
year and industry fixed effects. The coefficients reported in the table are accompanied by either three, two, one or zero stars,
representing, respectively, a statistical significance level of 1%, 5%, 10% and higher than 10%. Below each coefficient the
corresponding standard error is also present. Each of the three sections, referred to regressions of INVEFF on GREEN, DIGITAL
and GREEN×DIGITAL, is further divided into three parts: regressions run using the whole sample; regression run over the
sample of firms that over-invest only; regressions run over the sample of firms that under-invest only. For each of these parts,
both the univariate and the multivariate analyses are shown. Controls as well as country, year and sector fixed effects, although
present, are not shown in the table. The complete table can be found in the appendix.

Table 4 - Regression results (investment efficiency equation with NEG dummy)

Table 4 presents the results of the first robustness test, consisting of regressions of INVEFF on GREEN and DIGITAL as well as
on their interaction. In this second set of regressions, the tests use a modified investment efficiency equation which includes
NEG, a dummy variable equal to one if the firm experienced negative revenue growth during the financial year prior to the
survey. The coefficients reported in the table are accompanied by either three, two, one or zero stars, representing, respectively,
a statistical significance level of 1%, 5%, 10% and higher than 10%. Below each coefficient the corresponding standard error is
also present. Each of the three sections, referred to regressions of INVEFF on GREEN, DIGITAL and GREEN×DIGITAL, is
further divided into three parts: regressions run using the whole sample; regression run over the sample of firms that over-invest
only; regressions run over the sample of firms that under-invest only. For each of these parts, both the univariate and the
multivariate analyses are shown. Controls as well as country, year and sector fixed effects, although present, are not shown in the
table. The complete table can be found in the appendix.

Digital investments also seem to mitigate investment inefficiency in this last robustness check: when
adding controls, the depressing effect on investment inefficiency is verified at the 5% level in the over-
investment case and at the 1% level in the whole sample and under-investment case. Finally, we turn to
the models with the interaction term. Again, we observe that a moderating effect exists when running the
regressions on the whole sample of firms, in that the coefficient is negative and statistically significant at
the 5% level when adding controls. This effect is mostly driven by over-investing firms. No such effect is
verified for firms that under-invest, instead. When considered in their entirety, these results confirm
hypotheses one, two as well as three. Finally, it should be noted that because we are no longer dropping

5 – Robustness checks | 15
observations when industry-year groups are too small, we observe a slight increase in sample size from
previous tests.
6 – Conclusions
In this study, we investigate the influence of investments in green, digital and both areas simultaneously
on corporate investment efficiency. To the best of our knowledge, no study to date has directly addressed
the effect of simultaneous corporate investment in green and digital on investment efficiency. We report
the following findings: (a) corporate green investments aimed at either mitigating a company’s
environmental footprint or adapting to adverse weather events due to climate change reduce a firm’s
investment inefficiency level mainly by reducing under-investment; (b) corporate investments in digital
technologies also reduce a firm’s investment inefficiency level; the results are particularly robust for under-
investing firms; (c) a moderating effect of corporate digital investments on corporate green investments,
whereby digital technologies enhance the positive effect of green investments on investment efficiency, is
also verified; (d) for firms that under-invest, wave 2022 is associated with a higher inefficiency level than
wave 2021; inefficiency is even higher when considering wave 2023; (e) firms in the construction and
transportation industry are generally associated with a higher inefficiency level than manufacturing firms
(namely, the base case); (f) higher company age is strongly and significantly associated with higher
efficiency levels for firms that over-invest as well as for firms that under-invest; (g) there is evidence that
higher leverage increases inefficiency on over-investing firms; (h) overall, there is evidence that higher
collateral value of assets increases investment inefficiency; (i) finally, across all tests the number of firms
that under-invest is significantly larger than that of firms that over-invest.

Our results provide useful information at multiple levels. At a corporate level, they encourage the
adoption of green and digital technologies and, upon further scrutiny, they may also encourage their
contemporaneous implementation. At a policy level, they show that public authorities should actively
support firms’ efforts to mitigate their environmental footprint and to enhance their digital infrastructure.
Arguably, the enhanced level of investment efficiency generated by these types of investments is desirable
for two reasons: first, more effective investment plans are directly associated with a better allocation of
factors of production to their most valuable uses, resulting in higher economic efficiency and higher
values of aggregate welfare; second, higher aggregate investment efficiency can be understood as a lower
statistical dispersion of investment levels, which may allow for a deeper understanding of the comovement
between aggregate investments and other macroeconomic variables. This in turn would help policymakers
design more effective monetary policy, thereby further enhancing overall economic efficiency. Aside from
investment efficiency considerations, investments in green and digitals should arguably be supported for
a variety of additional reasons, perhaps the most important being the necessity to face up to climate
change and its adverse economic and social implications.

Despite these contributions, our study has some limitations. By imposing temporal precedence between
the explanatory and explained variables we showed that the latter may be a product of the former;
however, the causal mechanism linking investments in green and digital on one side and investment
efficiency on the other deservers further scrutiny. Similarly, a potential explanation for why digital
investments may enhance the effect of green investments has been proposed, but not tested. Third, the
survey was conducted on advanced economies, characterised by high levels of capital accumulation. It
remains to be tested whether these effects materialise in emerging economies as well. Finally, and most
importantly, our dataset did not include any continuous variable measuring the amount of financial
resources invested in either green or digital technologies; indeed, GREEN and DIGITAL were dummies.
Further research involving continuous variables may confirm or disprove our results, although we believe
there is enough evidence by now to predict that investments in green and digital technologies reduce
corporate investment inefficiency. Notwithstanding these limitations, we believe that our results are
noteworthy, and we hope they will be useful for researchers as well as policymakers.

16 | The impact of the digital and green transitions on investment inefficiency

6 – Conclusions | 17
References
Andersen, Allan Dahl, Koen Frenken, Victor Galaz, Florian Kern, Laurens Klerkx, Matthijs Mouthaan,
Laura Piscicelli, Juliet B. Schor, Taneli Vaskelainen (2021). “On digitalization and sustainability
transitions.” Environmental Innovation and Societal Transitions No. 41, 96–98.
Aral, Sinan and Peter Weill (2007). “I.T. Assets, Organizational Capabilities and Firm Performance: How
Resource Allocations and Organizational Differences Explain Performance Variation.” Organization
Science No. 18(5), 763-780.
Aupperle, Kenneth E., Archie B. Carroll and John D. Hatfield (1985). “An Empirical Examination of the
Relationship between Corporate Social Responsibility and Profitability.” Academy of Management
Journal No. 28(2), 446-463.
Baltagi, Badi H. (2005). “Econometric Analysis of Panel Data.” John Wiley and Sons.
Baron, Reuben M., and David A. Kenny (1986). “The moderator–mediator variable distinction in social
psychological research: Conceptual, strategic, and statistical considerations.” Journal of Personality and
Social Psychology No. 51(6), 1173–1182.
Benitez, Jose, Alvaro Arenas, Ana Castillo and Jose Esteves (2022). “Impact of digital leadership capability
on innovation performance: The role of platform digitization capability.” Information & Management
No. 59(2).
Benlemlih, Mohammed and Mohammad Bitar (2015). “Corporate Social Responsibility and Investment
Efficiency.” Academy of Management Proceedings No. 1.
Bianchini, Stefano, Giacomo Damioli and Claudia Ghisetti (2023). “The environmental effects of the
“twin” green and digital transition in European regions.” Envinronmental and Resource Economics No.
84(4), 877-918.
Biddle, Gary C., Gilles Hilary and Rodrigo S. Verdi (2009). “How does financial reporting quality relate
to investment efficiency?” Journal of Accounting and Economics No. 48(2-3), 112-131.
Cappa, Francesco, Raffaele Oriani, Enzo Peruffo and Ian McCarthy (2021). “Big data for creating and
capturing value in the digitalized environment: Unpacking the effects of volume, variety, and veracity on
firm performance.” Journal of Product Innovation Management No. 38(1), 49–67.
Chen, Feng, Ole-Kristian Hope, Qingyuan Li and Xin Wang (2011). “Financial Reporting Quality and
Investment Efficiency of Private Firms in Emerging Markets.” The Accounting Review No. 86(4), 1255-
1288.
Chen, Ruiyuan, Sadok El Ghoul, Omrane Guedhami, He Wang (2017). “Do state and foreign ownership
affect investment efficiency? Evidence from privatizations.” Journal of Corporate Finance No. 42, 408-
421.
Cook, Kirsten A., Andrea M. Romi, Daniela Sánchez and Juan Manuel Sánchez (2018). “The influence
of corporate social responsibility on investment efficiency and innovation.” Journal of Business Finance
and Accounting No. 46(3-4), 494-537.
Erawati, Ni Made Adi, Sutrisno T., Bambang Hariadi and Erwin Saraswati (2021). “The Role of
Corporate Social Responsibility in the Investment Efficiency: Is It Important?” Journal of Asian Finance,
Economics and Business No. 8(1), 169-178.
Fouquet, Roger and Ralph Hippe (2022). “Twin transitions of decarbonisation and digitalisation: A
historical perspective on energy and information in European economies.” Energy Research and Social
Science No. 91.
Freeman, R. Edward, Jeffrey S. Harrison, Andrew C. Wicks, Bidhan L. Parmar, Simone de Colle.
Stakeholder Theory: the State of the Art. (Cambridge University Press, 2010).

18 | The impact of the digital and green transitions on investment inefficiency

Friedman, Milton (1970). “The Social responsibility of Business is to Increase its Profits.” The New York
Times.
Gomariz, Ma Fuensanta Cutillas and Juan Pedro Sánchez Ballesta (2014). “Financial reporting quality,
debt maturity and investment efficiency.” Journal of Banking & Finance No. 40, 494-506.
Haller, Alina-Petronela, Mirela Ștefănică, Gina Ionela Butnaru and Rodica Cristina Butnaru (2023).
“Climate neutrality through economic growth, digitalisation, eco-innovation and renewable energy in
European countries.” Kybernetes.
Hampel, Frank, Elvezio Ronchetti, Peter Rousseeuw and Werner Stahel (1986). “Robust Statistics: The
Approach Based on Influence Functions.” John Wiley & Sons, New York.
Hayashi, Fumio (1982). “Tobin’s Marginal q and Average q: A Neoclassical
Interpretation.” Econometrica No. 50(1), 213–224.
Hicks, Raymond and Dustin Tingley (2011). “Casual mediation analysis.” The Stata Journal No. 11(4),
1–15.
Hubbard, R. Glenn (1998). “Capital-Market Imperfections and Investment.” Journal of Economic
Literature No. 36(1), 193–225.
Huo, Peng and Luxin Wang (2022). “Digital economy and business investment efficiency: Inhibiting or
facilitating?” Research in International Business and Finance No. 63.
Husain, Shaiara, Kazi Sohag and Yanrui Wu (2022). “The response of green energy and technology
investment to climate policy uncertainty: An application of twin transitions strategy.” Technology in
Society No. 71.
IPCC (2014). “Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and III
to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change.” IPCC, Geneva,
Switzerland.
Jensen, Michael C. (1986). “Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers.” The
American Economic Review 76(2), 323-329.
Jensen, Michael C. and William H. Meckling (1976). “Theory of the firm: Managerial behavior, agency
costs and ownership structure.” Journal of Financial Economics No. 3(4), 305-360.
Jorgenson, Dale W. (2001). “Information Technology and the US Economy.” The American Economic
Review No. 91(1), 1-32.
Jung, Juan and Gonzalo Gómez-Bengoechea (2022). “A literature review on firm digitalization: drivers
and impacts.” Estudios sobre la Economía Española No. 20.
Khediri, Karim Ben (2021). “CSR and investment efficiency in Western European countries.” Corporate
Social Responsibility and Environmental Management No. 28(6), 1769-1784.
Kim, Su-In and Yujin Kim (2023). “Analysis of the relationship between investment inefficiency and
climate risk and the moderating effects of managerial ownership.” Environment, Development and
Sustainability No. 25, 9337-9358.
Krüger, Philipp (2015). “Corporate goodness and shareholder wealth.” Journal of Financial Economics
No. 115(2).
Lee, Ming-Te (2020). “Corporate Social Responsibility and Investment Efficiency: Evidence from an
Emerging Asian Market.” Business & Economics Review No. 29(2), 1-16.
Lesecq, Suzanne, Gareth Keeley, Gabriela Dudnik and Alan O’Riordan (2022). "Would digitalisation
support transition towards more sustainable agriculture to meet the Green Deal ambition?" 8th
International Conference on Energy Efficiency and Agricultural Engineering (EE&AE), Ruse, Bulgaria,
2022, 1-5.

References | 19
Li, Xufang, Dijun Fan, Zhuoxuan Li and Mingzhu Pan (2023). “The Impact Mechanism of Digitalization
on Green Innovation of Chinese Manufacturing Enterprises: An Empirical Study.” Sustainability No.
15(12), 1-22.
Liu, Xing, Fengzhong Liu, Xiaoyi Ren (2023). “Firms' digitalization in manufacturing and the structure
and direction of green innovation.” Journal of Environmental Management No. 335.
Massacesi, Ludovica, Désirée Rückert and Christoph Weiss (2022). “Digitalisation in Europe 2021-2022.
Evidence from the EIB Investment Survey.” European Investment Bank.
Mazboudi, Mohamed and Iftekhar Hasan (2018). “Secrecy, information shocks, and corpo- rate
investment: Evidence from European Union countries.” Journal of International Financial Markets,
Institutions and Money No. 54, 166-176.
Modigliani, Franco and Merton H. Miller (1958). “The Cost of Capital, Corporation Finance and the
Theory of Investment.” The American Economic Review No. 48(3), 261–297.
Mondejar, Maria E., Ram Avtar, Heyker Lellani Baños Diaz, Rama Kant Dubey, Jesús Esteban, Abigail
Gómez-Morales, Brett Hallam, Nsilulu Tresor Mbungu, Chukwuebuka Christopher Okolo, Kumar Arun
Prasad, Qianhong She and Sergi Garcia-Segura (2021). “Digitalization to achieve sustainable development
goals: Steps towards a Smart Green Planet.” Science of The Total Environment No. 794.
Muench, Stefan, Eckhard Stoermer, Kathrine Jensen, Tommi Asikainen, Maurizio Salvi and Fabiana
Scapolo (2022). “Towards a green and digital future. Key requirements for successful twin transitions in
the European Union.” European Commission, Joint Research Center (JRC).
Myers, Stewart C. and Nicholas S. Majluf (1984). “Corporate financing and investment decisions when
firms have information that investors do not have.” Journal of Financial Economics No. 13(2), 187-221.
Needhidasan, Santhanam, Samuel Melvin and Chidambaram Ramalingam (2014). “Electronic waste –
an emerging threat to the environment of urban India.” Journal of Environmental Health Science and
Engineering No. 12(36).
Ortega-Gras, Juan-José, Josefina Garrido-Lova, María-Victoria Bueno-Delgado and Gregorio Cañavate-
Cruzado (2021). “Twin Transition through the Implementation of Industry 4.0 Technologies: Desk-
Research Analysis and Practical Use Cases in Europe.” Sustainability No. 13(24).
Pathan, Shams (2009). “Strong Boards, CEO Power and Bank Risk-Taking.” Journal of Banking and
Finance No. 33(7), 1340-1350.
Porter, Michael E. and Mark R. Kramer (2006). “Strategy and Society: the Link between Competitive
Advantage and Corporate Social Responsibility.” Harvard Business Review.
Ribeiro-Navarrete, Samuel, Dolores Botella-Carrubi, Daniel Palacios-Marqués and Maria Orero-Blat
(2021). “The effect of digitalization on business performance: An applied study of KIBS.” Journal of
Business Research No. 126, 319-326.
Samet, Marwa and Anis Jarboui (2017). “How does corporate social responsibility contribute to
investment efficiency?” Journal of Multinational Financial Management No. 40, 33-46.
Strubell, Emma, Ananya Ganesh and Andrew McCallum (2019). “Energy and Policy Considerations for
Deep Learning in NLP.” Proceedings of the 57th Annual Meeting of the Association for Computational
Linguistics, Florence, Italy, July 28 - August 2, 2019, 3645-3650.
Teixeira, Josélia Elvira and Ana Teresa C.P. Tavares-Lehmann (2022). “Industry 4.0 in the European
union: Policies and national strategies.” Technological Forecasting & Social Change No. 180.
Xu, Guyiang, Guanggui Li, Peibo Sun and Dan Peng (2023). “Inefficient investment and digital
transformation: What is the role of financing constraints?” Finance Research Letters No. 51.
Yu Ming-Min, Shih-Chan Ting and Mu-Chen Chen (2010). “Evaluating the cross-efficiency of
information sharing in supply chains.” Expert Systems with Applications No. 37(4), 2891-2897.

20 | The impact of the digital and green transitions on investment inefficiency

Zeng, Shihong, Yujia Qin and Guowang Zeng (2019). “Impact of Corporate Environmental
Responsibility on Investment Efficiency: The Moderating Roles of the Institutional Environment and
Consumer Environmental Awareness.” Sustainability No. 11(17).
Zhai, Huayun, Min Yang and Kam C. Chan (2022). “Does digital transformation enhance a firm's
performance? Evidence from China.” Technology in Society No. 68.

References | 21
Tables
Tables A1 and A2 shown below are a more detailed version of in-text tables 3 and 4, respectively.

Table A1 - Regression results (one single investment efficiency regression) – Complete table

Table A1 presents the results of the first robustness test, consisting of regressions of INVEFF on GREEN and DIGITAL as well
as on their interaction. In this second set of regressions, the tests use a modified investment efficiency equation which is not
estimated cross-sectionally by year and industry, but rather across the whole sample of observations, and which includes country,
year and industry fixed effects. The coefficients reported in the table are accompanied by either three, two, one or zero stars,
representing, respectively, a statistical significance level of 1%, 5%, 10% and higher than 10%. Below each coefficient the
corresponding standard error is also present. Each of the three sections, referred to regressions of INVEFF on GREEN, DIGITAL
and GREEN×DIGITAL, is further divided into three parts: regressions run using the whole sample; regression run over the
sample of firms that over-invest only; regressions run over the sample of firms that under-invest only. For each of these parts,
both the univariate and the multivariate analyses are shown. In particular, LOSS, PROF, AGE, LEV, TANG and CASH are the
control variables. Year and sector fixed effects are shown, with wave “2021” and sector “Manufacturing” representing the base
cases. Country fixed effects are also present, but they are not shown.

22 | The impact of the digital and green transitions on investment inefficiency

Table A2 - Regression results (investment efficiency equation with NEG dummy) – Complete table

Table A2 presents the results of the first robustness test, consisting of regressions of INVEFF on GREEN and DIGITAL as well
as on their interaction. In this second set of regressions, the tests use a modified investment efficiency equation which includes
NEG, a dummy variable equal to one if the firm experienced negative revenue growth during the financial year prior to the
survey. The coefficients reported in the table are accompanied by either three, two, one or zero stars, representing, respectively,
a statistical significance level of 1%, 5%, 10% and higher than 10%. Below each coefficient the corresponding standard error is
also present. Each of the three sections, referred to regressions of INVEFF on GREEN, DIGITAL and GREEN×DIGITAL, is
further divided into three parts: regressions run using the whole sample; regression run over the sample of firms that over-invest
only; regressions run over the sample of firms that under-invest only. For each of these parts, both the univariate and the
multivariate analyses are shown. In particular, LOSS, PROF, AGE, LEV, TANG and CASH are the control variables. Year and
sector fixed effects are shown, with wave “2021” and sector “Manufacturing” representing the base cases. Country fixed effects
are also present, but they are not shown.

Tables | 23
The impact of the digital
and green transitions on
investment inefficiency
October 2024

© European Investment Bank, 10/2024 EN

2024-0244

pdf: ISBN 978-92-861-5824-7